VARLITINIB FOR USE IN TREATING CANCER IN A PATIENT IDENTIFIED AS HAVING ACTIVATED HER1 , HER2 AND/OR HER3 RECEPTOR CANCEROUS CELLS

The present disclosure relates to a therapy, for example a monotherapy or combination therapy comprising a type I tyrosine kinase inhibitor for the treatment of cancer, for example liver cancer, such as hepatocellular carcinoma, in patients who are identified as having HER1 activated, HER2 activated and/or HER3 activated cancer (such as HER2 activated and/or HER3 activated cancer]. BACKGROUND

There are many cancers that are difficult to treat and although therapy is available, there appears to exist or to come into existence, a degree of resistance to the therapy. Primary resistance may occur in that cancer does not respond to treatment from the outset Secondary or acquired resistance also occurs quite frequently, which means that a therapy to which the patient seems to respond, at a certain time, loses its efficacy.

There are numerous reasons for resistance, for example some cancers are discovered at a late stage and/or a simply not responsive to treatment

Mechanisms by which cancers avoid the therapeutic effect of therapy include but are not limited to:

i] mutations which render the cancer less vulnerable to the treatment (eg mutation of the site of action of the therapy],

ii] active transportation of the drug out of the tumor, for example by p-glycoprotein,

iii] building up physical defences, for example stroma which inhibit certain immune responses, and

iv] certain cancers develop paths to repair damage caused by some anti-cancer therapies.

Tumor heterogeneity may also contribute to resistance, where small subpopulations of cells may acquire or stochastically already possess some of the features enabling them to emerge under selective drug pressure. This is a problem that many patients with cancer encounter, and it obviously limits the therapeutic alternatives that are effective and worsens the prognosis.

Cancer therapy guidelines describe the sequence of therapies, which are recommended and in which sequence, so that if a patient shows disease progression on the first therapy ("first line"], then a next therapy ("second line"] is recommended, and so on. These therapy recommendations are based on available scientific data and experience and illustrate that resistance to one therapy does not exclude that another therapy may be effective and prolong life or shrink a tumor. At late stages cancers do not respond, are completely therapy refractory, and no more avenues of therapy exist Thus, unless new therapies can be found, which are effective, these cancers cannot be treated.

Varlitinib is in clinical trials for the treatment of cancer. However, the present inventors have data that suggest the molecule is particularly effective in the treatment of a subcategory of cancers, namely those where cancer cells have activated HER1, HER2 and/or activated HER3 receptors (HER2 and/or activated HER3 receptors], in particular where the receptors are phosphorylated. These cancers seem to be sensitive to treatment with Varlitinib.

Thus, it is not simply a matter of the amount of the expression of the receptors, such as HER2 and/or HER3, but rather the target patient population has increased levels of activation of the relevant receptors. Although, patients within the group may also have increased levels of HER2 and/or HER3.

Generally, cancer therapies are not directed at patients which have activated HER3 cancer. However, as discussed in "Understanding the biology og HER3 receptor as a therapeutic target in human cancer (Lyu, H et al Acta Pharmaceutica Sinica B 2018; 8(4); 503-510) HER3 has only weak activity it frequently co-expresses and interacts with other receptor tyrosine kinases. It is well suited to activation of the PI-3K/Akt pathway, which is associated with tumorgenesis.

The C-terminal tail of HER3 has multiple tyrosine residues, which when phosphorylated is able to bind the p85 subunit of PI-3K.

In HER2 amplified breast cancer there is preferential phosphorylation of HER3. HER3 serves as a critical co-receptor of HER2 and its expression is essential for HER2-mediated breast cancer cell survival and proliferation.

HER3 is overexpressed in n50 to 70% of breast cancers. It may also have a role in many human cancers, for example colorectal cancer, gastric cancer, breast cancer, melanoma, ovarian cancer, head and neck cancer, pancreatic cancer and cervical cancer.

HER3 is implicated in resistance, distant metastasis, tumor size and risk of local reoccurrence.

Most of the therapies to target HER3 to date have been antibodies.

Thus, the present therapy may fill an unmet need because at the present time such patient populations have no therapies that particularly focused on their treatment

It now appears that the previously unidentified patient population (such as HER2 and/or HER3 activated) responds particularly well to Varlitinib therapy.

SUMMARY OF THE DISCLOSURE

The following numbered paragraphs summarise the present invention:

1. A method of treating a patient for cancer (for example a solid tumour) by administering a compound of formula (I :

an enantiomer thereof or a pharmaceutically acceptable salt of any one of the same, for example wherein the compound of formula (I) is Varlitinib or a pharmaceutically acceptable salt thereof, wherein the patient has been identified as having activated HER1 receptor cancerous cells, HER2 receptor cancerous cells, HER3 receptor cancerous cells, or any combination thereof, for example where the cancer is a solid tumor.

2. The method according to paragraph 1, wherein the patient has been identified as having activated HER2 receptor and/or activated HER3 receptor cancerous cells (for example HER1 is not activated. The method according to paragraphs 1 or 2, wherein the HER2 is activated.

The method according to any one of paragraphs 1 to 3, wherein the HER3 is activated.

The method according to paragraph 4, wherein the HER3 is activated by NRG1, for example wherein the expression of NRG1 is higher in said cancerous cells than in normal cells.

The method according to any one of paragraphs 1 to 5, wherein the HER1 is activated.

The method according to any one of paragraphs 1 to 6, wherein the activated receptor (such as HER3) is dimerised to another receptor (such as HER2).

The method according to any one of paragraphs 1 to 7, wherein the activated HER1 receptor, HER2 receptor and/or HER3 receptor is phosphorylated.

The method according to paragraph 8, wherein the levels of phosphorylation in respect of said receptor(s) in cancer cells is/are higher than levels of phosphorylation in normal cells.

The method according to any one of paragraphs 1 to 9, wherein the levels of HER2 receptor on said cancerous cells is normal.

The method according to any one of paragraphs 1 to 10, wherein the levels of HER3 receptor on said cancerous cells are normal.

The method according to any one of paragraphs 1 to 11, wherein the levels of HER1 receptor on said cancerous cells are normal.

The method according to any one of paragraphs 1 to 9, 11 or 12, wherein there are increased expression levels of total HER2 receptor on said cancerous cells compared to a non-cancerous cell.

The method according to any one of paragraphs 1 to 13, wherein the activated HER2 is phosphorylated at Tyrl221/1222.

The method according to any one of paragraphs 1 to 10, or 12 to 14, wherein there are increased expression levels of total HER3 receptor compared to a non-cancerous cell.

The method according to any one of paragraphs 1 to 15, wherein the activated HER3 is phosphorylated on a tyrosine in the C-terminal of the receptor.

The method according to any one of paragraphs 1 to 16, wherein the activated HER3 is phosphorylated at Tyrl289.

The method according to any one of paragraphs 1 to 11, or 13 to 16, wherein there are increased expression levels of total HER1 receptor on said cancerous cells compared to a noncancerous cell.

The method according to any one of paragraphs 1 to 18, wherein the activated HER1 is phosphorylated at Tyrl068.

The method according to any one of paragraphs 1 to 19, wherein the HER2 receptor has a P1170A mutation, an I655V mutation, or both, for example both P1170A and I665V mutations. The method according to any one of paragraphs 1 to 20, wherein the HER1 receptor has a R521K mutation.

The method according to any one of paragraphs 1 to 21, wherein the patient has a cancer selected from: liver cancer (such as hepatocellular carcinoma], biliary tract cancer, gall bladder cancer, breast cancer (such as none ER+ breast cancer], prostate cancer, colorectal cancer, ovarian cancer, cervical cancer, lung cancer, gastric cancer, pancreatic, bone cancer, bladder cancer, head and neck cancer, thyroid cancer, skin cancer, renal cancer, oesophagus cancer and combinations of two or more of the same.

The method of paragraph 22, wherein the cancer is gastric cancer.

The method of paragraphs 22 or 23, wherein the cancer is selected from liver cancer, such as hepatocellular carcinoma, gallbladder cancer, cholangiocarcinoma and a combination of two or more the same, such as hepatocellular carcinoma.

The method according to any one of paragraphs 22 or 24, wherein the cancer is cholangiocarcioma and is located in an intrahepatic bile duct, left hepatic duct, right hepatic duct, common hepatic duct, cystic duct, common bile duct, Ampulla of Vater or a combination thereof.

The method according to any one of paragraphs 22 to 25, wherein the cholangiocarcinoma is intrahepatic.

The method according to any one of paragraphs 22 to 25, wherein the cholangiocarcinoma is extrahepatic.

The method according to paragraph 22, wherein the cancer is breast cancer.

The method according to any one of paragraphs 1 to 28, wherein the compound of formula (I) is administered to treat or prevent metastasis, for example brain metastasis such as from a primary breast cancer tumor, or liver metastasis from a primary cancer, such as colorectal cancer.

The method according to any one of paragraphs 1 to 29, wherein the compound of formula (I) is Varlitinib:

or a pharmaceutically acceptable salt thereof.

A method according to any one of paragraphs 1 to 30, wherein the compound of formula (I) is provided as the free base.

A method according to any one of paragraphs 1 to 31, wherein the compound of formula (I) is administered as a pharmaceutical formulation, for example comprising at least one excipient, diluent or carrier.

A method according to any one of paragraphs 1 to 32, wherein the compound of formula (I] or a pharmaceutical formulation comprising same is administered orally.

A method according to any one of paragraphs 1 to 33, wherein the compound of formula (I) or a pharmaceutical formulation comprising the same is administered bi-daily. 35. A method according to any one of paragraphs 1 to 34, wherein each dose of the compound of formula [I] is in the range 100 to 900mg, for 100, 200, 300, 400, 500, 600, 700, 800 or 900mg.

36. A method according to paragraph 35, wherein each dose of the compound of formula (I) is in the range 100 to 500mg or 200 to 500mg, such as 250 mg or 490 mg.

37. A method according to paragraph 34, wherein each dose is lOOmg, 200mg, 300mg or 400mg (such as 300mg or 400mg).

38. A method according to any one of paragraphs 1 to 37, wherein the compound of formula (I) or formulation comprising the same is employed as a monotherapy.

39. A method according to any one of paragraphs 1 to 37, wherein the compound of formula (I) is employed in a combination with another anti-cancer agent

40. A method according to paragraph 39, wherein the combination therapy comprises a chemotherapeutic agent

41. A method according to paragraph 40, wherein the chemotherapeutic agent is independently selected from the group comprising a platin (such as cisplatin or oxaliplatin], gemcitabine, capecitabine, 5-FU, FOLFOX, FOLFIRI and FOLFIRINOX.

42. A method according to any one of paragraphs 39 to 41, wherein the anti-cancer agent is selected from: patritumab (U3-1287), seribantumab (mm-121], elgemtumab (LJM716), AV- 203, KTN-3379 or GSK2849330.

43. A compound of formula (I] as disclosed above an enantiomer thereof or a pharmaceutically acceptable salt of any one of the same, for example wherein the compound of formula (I] is Variitinib or a pharmaceutically acceptable salt thereof, for use in the treatment of a cancer patient identified as having activated HER1 receptor cancerous cells, HER2 receptor cancerous cells, HER3 receptor cancerous cells, or any combination thereof, for example wherein the cancer is a solid tumour.

Paragraph 43 is written in the style of an EPC2000 claim. The present disclosure explicitly includes EPC2000 style disclosure corresponding to paragraphs 2 to 41. These EPC 2000 style paragraphs will nominally be paragraphs 44 to 84.

85. Use of a compound of formula (I) disclosed above an enantiomer thereof or a pharmaceutically acceptable salt of any one of the same, for example wherein the compound of formula (I) is Variitinib or a pharmaceutically acceptable salt thereof, for the manufacture of a medicament for the treatment of a cancer patient identified as having activated HER1 receptor cancerous cells, HER2 receptor cancerous cells, HER3 receptor cancerous cells, or any combination thereof, for example where the cancer is a solid tumour.

Paragraph 85 is written in the style of a Swiss claim. The present disclosure explicitly includes

Swiss style disclosure corresponding to paragraphs 2 to 41. These Swiss style paragraphs will nominally be paragraphs 86 to 126.

The two-state receptor model suggests that the state of the receptor (such as activation) may affect the affinity of a chemical entity to the receptor. This theory does not usually apply to inhibitors but it may be that the compounds of formula (I), such as Variitinib, have greater affinity for receptors, such as HER3, in an activated state. In one embodiment, the cancer is both HER2 activated and HER3 activated. For example, the cancer has at least one of the following:

-higher expression levels of total ErbB2 and total ErbB3 compared to a non-cancerous cell; - higher expression levels of phospho-ErbB2 and total ErbB3 compared to a non-cancerous cell; -higher expression levels of total ErbB2 and phospho-ErbB3 compared to a noncancerous cell; -higher expression levels of phospho-ErbB2 and phospho-ErbB3 compared to a non-cancerous cell; -higher expression levels of total ErbB2, phospho-ErbB2 and total ErbB3 compared to a non-cancerous cell; -higher expression levels of total ErbB2, phospho- ErbB2 and phospho-ErbB3 compared to a non-cancerous cell; -higher expression levels of phospho-ErbB2, total ErbB3 and phospho-ErbB3 compared to a non-cancerous cell; -higher expression levels of total-ErbB2, total ErbB3 and phospho-ErbB3 compared to a noncancerous cell; or higher expression levels of total ErbB2, phospho-ErbB2, total ErbB3 and phospho-ErbB3 compared to a non-cancerous cell.

In one embodiment, the cancer is HER1 activated. In one embodiment, the cancer is HER1 activated and HER2 activated. In one embodiment, the cancer is HER1 activated and HER3 activated. In one embodiment, the cancer is HER1 activated, HER2 activated and HER3 activated.

In one embodiment, the HER3 is activated by NRG1 (neuregulin 1 also known as HRG -the ligand for HER3], for example wherein the expression of NRG1 is higher in the cancerous cells compared to normal cells.

In one embodiment, the cancer patient has been identified as having increased levels of

NRG1 compared to a normal control.

In one embodiment, the HER1 receptor has a R521K mutation.

In one embodiment, the HER2 receptor has a P1170A mutation, an 1655 V mutation or both. In one embodiment, the HER2 receptor has a P1170A mutation. In one embodiment, the HER2 receptor has an I665V mutation. In one embodiment, the HER2 receptor has a P1170A mutation and an I665V mutation.

In one embodiment the biliary duct cancer is in a location selected from intrahepatic bile ducts, left hepatic duct, right hepatic duct, common hepatic duct, cystic duct, common bile duct, Ampulla of Vater and combinations thereof.

In one embodiment the biliary duct cancer is in an intrahepatic bile duct In one embodiment the biliary duct cancer is in a left hepatic duct. In one embodiment the biliary duct cancer is in a right hepatic duct. In one embodiment the biliary duct cancer is in a common hepatic duct In one embodiment the biliary duct cancer is in a cystic duct In one embodiment the biliary duct cancer is in a common bile duct. In one embodiment the biliary duct cancer is in an Ampulla of Vater. In one embodiment the biliary duct cancer is a cancer of the Papilla of Vater.

In one embodiment the cancer is a metastatic form of a cancer, in the patient population disclosed herein disclosed herein.

In one embodiment the cancer according to the present disclosure has not metastasized in said patient population.

In one embodiment the compound of formula (I) is (i?)-N4-[3-Chloro-4-(thiazol-2- ylmethoxy]-phenyl]-N6-(4-methyl-4, 5,-dihydro-oxazol-2-yl]-quinazoline-4,6-diamine:

(Varlitinib) or a pharmaceutically acceptable salt thereof or a pro-drug thereof.

In one embodiment (fi)-N4- [3-Chloro-4-(thiazol-2-ylmethoxy)-phenyl]-N6-(4-methyl-4, 5,- dihydro-oxazol-2-yl)-quinazohne-4,6-diamine is employed/administered as the free base (also referred to herein as Varlitinib).

In one embodiment the patient population is not activated for all of HER1, HER2, HER3 and HER4, for example the population is at least negative for HER1 and/or HER4. Thus, in one embodiment the patient population is HER2 and/or HER3 and HER1 activated but not HER4 activated. In one embodiment the patient population is HER2 and/or HER3 activated and HER4 activated but not HER1 activated. In one embodiment the patient population is only HER2 and/or HER3 activated. In one embodiment the patient population is only HER2 activated. In one embodiment the patient population is only HER3 activated. In one embodiment the patient population is both HER3 and HER4 activated.

In one embodiment the compound of formula (I) an enantiomer thereof or a pharmaceutically acceptable salt thereof is employed as a monotherapy, for example first line therapy or second line therapy, such as a first line monotherapy, in said patient population.

In one embodiment the compound of formula (I) an enantiomer thereof or a pharmaceutically acceptable salt thereof is employed in a combination therapy, for example in combination with a chemotherapy and/or a biological therapeutic, in particular as a first line therapy or a second line therapy in said patient population.

In one embodiment the compound of formula (I) an enantiomer thereof or a pharmaceutically acceptable salt thereof is employed as a second line monotherapy, in said patient population.

In, for example a second line monotherapy employing Varlitinib in the treatment of cholangiocarcinoma in said patient patients may show a significant reduction in CA19-9 levels. CA19-9 is a marker employed in the management of cholangiocarcinoma. Thus in one embodiment a HER3 and HER4 positive or amplified patient has a 10, 20, 30, 40, 50, 60, 70, 80 or 90% decrease in CA19-9 level whilst on the therapy according to the present disclosure, wherein the level is decreased relative to the level of CA19-9 before initiation of said therapy. This decrease in CA19-9 may be observed 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 or weeks after initiating therapy according to the present disclosure.

In one embodiment the compound of formula (I), according to the present disclosure such as Varlitinib, is employed in a second line therapy together with a chemotherapy agent or chemotherapy regimen, for example gemcitabine, capecitabine, 5-FU, FOLFOX, a platin, such as cisplatin or oxaliplatin, and a combination thereof, in said patient population. In one embodiment the compound of formula [I], according to the present disclosure such as Varlitinib, is administered orally.

In one embodiment the compound of formula [I], according to the present disclosure such as Varlitinib, is administered at a dose in the range lOOmg to 900mg on each occasion, in particular 200 mg, 300mg, 400mg or 500mg each dose, such as 400mg, for example, 250 mg or 490 mg, for example administered once or twice daily, such as twice daily (BID).

In one embodiment, the compound of formula (I), according to the present disclosure such as Varlitinib is administered twice daily.

In one embodiment the compound of formula (I), according to the present disclosure such as Varlitinib, is administered for a 28 days, referred to herein as a 28 day treatment cycle.

In one embodiment the compound of formula (I), according to the present disclosure such as Varlitinib, is administered as pharmaceutical formulation comprising one or more pharmaceutically acceptable excipients.

In one embodiment the compound of formula (I), according to the present disclosure such as Varlitinib, or a formulation comprising the same is administered orally, for example as tablet or capsule.

In one embodiment the treatment is adjuvant therapy, for example after surgery or after chemotherapy, in said patient population.

In one embodiment the treatment is neoadjuvant therapy, for example before surgery, in particular to shrink the tumour or tumours, in said patient population.

In one embodiment the treatment according to the present disclosure is suitable for the treatment of secondary tumours, in said patient population. In one embodiment the cancer is metastatic cancer, in said patient population. In one embodiment the treatment according to the present disclosure is suitable for the treatment of primary cancer and metastases, in said patient population. In one embodiment the treatment according to the present disclosure is suitable for the treatment of secondary cancer and metastases. In one embodiment the treatment according to the present disclosure is suitable for the treatment of primary cancer, secondary cancer and metastases in said patient population.

In one embodiment the treatment according to the present disclosure is suitable for the treatment of cancerous cells in a lymph node, for a cancer of the present disclosure in said patient population.

In one embodiment the patient is a human.

DETAILED DISCLOSURE

The following disclosure is given in the context of treatment of the requisite patient population and intended to complement the disclosure in the summary of the invention section above,

The terms EGFR, HER1 and ErbBl are used interchangeably in this disclosure and are intended to refer to the same receptor.

The terms HER2 and ErbB2 are used interchangeably in this disclosure and are intended to refer to the same receptor.

Activation of a receptor as employed herein refers to one or more of the following: relay of a signal (i.e. where the receptor sends the signal onwards); amplification (where the receptor increases the "in-coming" signal]; and integration [where the receptor allows or facilitates the signal being incorporated into another biochemical pathway).

In the case of some receptor, such as HER3, dimerization is a pre-requisite to activation.

The terms HER3 and ErbB3 are used interchangeably in this disclosure and are intended to refer to the same receptor.

In one embodiment HER1, HER2 and/or HER3 activated as employed herein refers to where the cancer expresses an increased level of said receptor, such as total levels of the receptor and/or levels of the activated form of the receptor, such as the phosphorylated form of the receptor, in comparison a corresponding non-cancerous cell and includes but is not limited any of the following:

-increased levels of total EGFR only; -increased levels of phospho-EGFR only; -increased levels of total EGFR and phospho-EGFR only; -increased levels of total HER2 only; -increased levels of phospho-HER2 only; -increased levels of total HER2 and phospho-HER2 only; -increased levels of total HER3 only; -increased levels of phospho-HER3 only; - increased levels of total HER3 and phospho-HER3 only; -increased levels of total EGFR and total ErbB2 only; -increased levels of phospho-EGFR and total ErbB2 only; -increased levels of total EGFR and phospho-ErbbB2 only; -increased levels of phospho-EGFR and phospho-ErbB2 only; -increased levels of total EGFR and total ErbB3 only; -increased levels of phospho-EGFR and total ErbB3 only; -increased levels of total EGFR and phospho- ErbB3 only; -increased levels of phospho-EGFR and phospho- ErbB3only; -increased levels of total ErbB2 and total ErbB3 only; -increased levels of phospho-ErbB2 and total ErbB3 only; -increased levels of total ErbB2 and phospho-ErbB3 only; -increased levels of phospho-ErbB2 and phospho-ErbB3 only; -increased levels of total EGFR, phospho-EGFR and total ErbB2 only; -increased levels of total EGFR, phospho- EGFR and phospho-ErbB2 only; -increased levels of total EGFR, phospho-EGFR and total ErbB3 only; -increased levels of total EGFR, phospho-EGFR and phospho-ErbB3 only; - increased levels of total ErbB2, phospho-ErbB2 and total ErbB3 only; -increased levels of total ErbB2, phospho-ErbB2 and phospho-ErbB3 only; -increased levels of phospho-ErbB2, total ErbB3 and phospho-ErbB3 only; -increased levels of total-ErbB2, total ErbB3 and phospho-ErbB3 only; -increased levels of total ErbB2, phospho-ErbB2, total ErbB3 and phospho-ErbB3; -increased levels of total EGFR, total ErbB2, and total ErbB3 only; - increased levels of phospho-EGFR, phospho-ErbB2, and phospho-ErbB3 only; -increased levels of total EGFR, phospho-EGFR, total ErbB2, phospho-ErbB2, total ErbB3 and phospho- ErbB3; and Any combination of the above.

In one embodiment the activation is independent of increased expression, for example is dimerization and/or phosphorylation.

Methods of measuring said expression and amplification are known in the art, for example

FRET, BRET assays or chimeric report genes. Antibodies specific to phosphorylated forms or protein are also available, for example antibodies specific to Tryl289 are available. Total HER3 cellular assay kits can be purchase from Cisbio. Phospho-EGFR/p-EGFR, Phospho-ErbB2/p-ErbB2, or phospho-ErbB3/p-ErbB3 as used herein refers to the phosphorylated form of EGFR, ErbB2 or ErbB3 respectively. It includes phosphorylation of the receptor at any suitable phosphorylation site.

Known phosphorylation sites for EGFR (HER1) include but are not limited to: Tyr 1068, Tyr 845, Ser 991 and Tyr 998.

Known phosphorylation sites for ErbB2 (HER2)include but are not limited to: Tyr 877, Tyrl023, Tyrll39, Tyrl l96, Tyrl221, Tyrl222 and Tyrl248.

Known phosphorylation sites for ErbB3 (HER3) include but are not limited to Tyrl054, Tyrll97, Tyrll99, Tyrl222, Tyrl260, Tyrl262, Tyrl289 and Tyrl328.

In one embodiment, phospho-EGFR refers to EGFR [HER1] which has been phosphorylated at Tyr 1068.

In one embodiment, phospho-ErbB2 refers to ErbB2 (HER2) which has been phosphorylated at Tyrl221 or Tryl222.

In one embodiment, phospho-ErbB3 refers to ErbB3 (HER3) which has been phosphorylated at Tyrl289,

In one embodiment, increased levels as employed herein refers to a level that is at least 1, 2 or 3, such as 2 standard deviations higher than the mean level in a non-cancerous cell.

In another embodiment, increased levels as employed herein refers to a level that is above the level of the 95 ^th percentile of a population of non-cancerous cells.

"In a patient population" refers to administration of the therapy to a patient characterised as HER1, HER2 and/or HER3 activated.

In one embodiment, the cancer has a mutation in the HER2 receptor, such as a P1170A mutation, an 1655 V mutation, or both, for example both P1170A and 1665 V mutations. The present inventors have established that cancers, such as a hepatocellular carcinoma (HCC] that have one or more of these mutations were significantly more responsive to treatment with a pan-HER inhibitor such as Varlitinib compared to cancers that did not have any of these mutations. Accordingly, patients with cancers having the P1170A mutation, an I655V mutation, or both P1170A and I655V mutations may represent a specific patient population subset that is particularly responsive to treatment with a pan-HER inhibitor, such as Varlitinib. Thus, in one embodiment, the target patient population has a P1170A mutation, an I655V mutation, or both, for example both P1170A and I665V mutations.

In one embodiment, the cancer has a mutation in the HER1 receptor, such as R521K. The present inventors have established that cancers, such as a hepatocellular carcinoma [HCC) that have the R521K mutation are more responsive to treatment with a pan-HER inhibitor such as Varlitinib compared to cancers that did not have any mutations in HER1. Accordingly, patients with cancers having the R521K mutation may represent a specific patient population subset that is particularly responsive to treatment with a pan-HER inhibitor, such as Varlitinib. Thus, in one embodiment, the target patient population has a R521K mutation in HER1.

In one embodiment, the HER3 is activated by NRG1, for example wherein the expression of NRG1 is higher in the cancerous cells compared to normal cells. Surprisingly, the present inventors found that cancers that have higher expression levels of NRG1 compared to normal cells are more responsive to treatment with a pan-HER inhibitor, such as Varlitinib. Accordingly, patients with cancers having raised levels of NRGl may mutations may represent a specific patient population subset that is particularly responsive to treatment with a pan-HER inhibitor, such as Varlitinib. Thus, in one embodiment, the cancer patient has been identified as having increased levels of NRGl compared to a normal control.

Downregulation of miR-203 and/or miR-542-3p may also be an indication of activation of a receptor according to the present disclosure.

Unless the context indicates otherwise, "identified" in the context of the present disclosure refer to the fact the patient is known to be in the requisite patient population before the administration of the therapy according to the present disclosure.

Liver cancer as employed herein refers to cancer which starts in the liver, including starting from structures located within the liver, such as blood vessels, including hepatocellular carcinoma.

In one embodiment the liver cancer is, for example selected from the group hepatocellular carcinoma, cholangiocarcinoma, angiosarcoma, and hepatoblastoma, in particular hepatocellular carcinoma. In one embodiment the primary liver cancer is stage 1, 2, 3 or 4. In one embodiment the liver cancer is secondary or metastasized liver cancer.

In one embodiment liver cancer does not include biliary tract cancer, such as cholangiocarcinoma.

In one embodiment the cancer is liver cancer, for example a liver metastasis from a primary cancer, for example colon cancer, which has spread to the liver. In one embodiment the liver cancer is HCC hepatocellular carcinoma.

Biliary duct cancer (also referred to as biliary cancer) as employed herein refers to cancer which starts in the bile ducts and includes cholangiocarcinoma and gallbladder cancer.

In one embodiment Biliary tract cancer as employed herein refers to cholangiocarcinoma

(intrahepatic, extrahepatic), gall bladder cancer and ampullary carcinoma.

Cholangiocarcinoma as referred to herein is a form of cancer that is composed of mutated epithelial cells (or cells showing characteristics of epithelial differentiation) that originate in the bile ducts which drain bile from the liver into the small intestine, but not including gallbladder cancer.

General guidelines for operability of biliary duct cancer include: Absence of lymph node or liver metastases; Absence of involvement of the portal vein; Absence of direct invasion of adjacent organs; and Absence of widespread metastatic disease.

In one embodiment, the cancer is gall bladder cancer. Gallbladder cancer as employed herein cancer which starts in the gallbladder. The following stages are used for gallbladder cancer:

Stage 0 (carcinoma in situ): Abnormal cells are found in the inner (mucosal) layer of the gallbladder; these abnormal cells may become cancer and spread into nearby normal tissue; Stage I Cancer has formed and has spread beyond the inner (mucosal) layer to a layer of tissue with blood vessels or to the muscle layer; Stage II Cancer has spread beyond the muscle layer to the connective tissue around the muscle; Stage IIIA Cancer has spread through the thin layers of tissue that cover the gallbladder and/or to the liver and/or to one nearby organ (e.g., stomach, small intestine, colon, pancreas, or bile ducts outside the liver); Stage IIIB Cancer has spread to nearby lymph nodes and beyond the inner layer of the gallbladder to a layer of tissue with blood vessels or to the muscle layer; or beyond the muscle layer to the connective tissue around the muscle; or through the thin layers of tissue that cover the gallbladder and/or to the liver and/or to one nearby organ; Stage IVA Cancer has spread to a main blood vessel of the liver or to 2 or more nearby organs or areas other than the liver. Cancer may have spread to nearby lymph nodes; and Stage IVB Cancer has spread to either lymph nodes along large arteries in the abdomen and/or near the lower part of the backbone or to organs or areas far away from the gallbladder.

In one embodiment the gastric cancer is selected from the group comprising adenocarcinoma of the stomach, squamous cell carcinomas, lymphoma of the stomach, gastric stromal tumor, and neuroendocrine tumors.

Prostate cancer as employed herein refers to cancer of the prostate, for example ductal adenocarcinoma, transitional cell (urothelial cancer), squamous cell cancer, carcinoid of the prostate, small cell cancer or sarcoma and sarcomatoid cancer. In one embodiment the prostate cancer is any one of the same.

Pancreatic cancer as employed herein includes exocrine cancers (including rare forms thereof such as cystitic tumours, and cancer of the acinar cells), endocrine pancreatic tumours (including gastrinomas, insulinomas, somatostatinomas, VIPomas, glucagonomas), pancreatoblastoma, sarcomas of the pancreas and lymphoma.

Colorectal cancer as employed herein refers to cancer or the colon and/or rectum and includes squamous cell cancers, carcinoid tumours, sarcomas and lymphomas.

Breast cancer as employed herein refers to cancer of the breast and includes ductal cardinoma in situ, lobular carcinoma in situ, invasive ductal breast cancer, invasive lobular breast cancer, invasive breast cancer, Paget's disease, angiosarcoma of the breast and rare types of breast cancer such as medullary breast cancer, mucinous breast cancer, tubular breast cancer, adenoid cystic carcinoma of the breast metaplastic breast cancer, basal type breast cancer and papillary breast cancer. In one embodiment the breast cancer is one selected from any one of the same. In one embodiment the breast cancer is phyllodes or cystosarcoma phyllodes.

Lung cancers are classified according to histological type and are categorized by the size and appearance of the malignant cells seen by a histopathologist under a microscope. For therapeutic purpose, two broad classes are distinguished: non-small cell lung carcinoma and small cell lung carcinoma.

In one embodiment the epithelial cancer is lung cancer, for example selected from small- cell lung cancer (SCLC) and non-small-cell lung cancer (NSCLC), such as NSCLC.

Non-small-cell lung carcinoma-The three main subtypes of NSCLC are adenocarcinoma, squamous-cell carcinoma and large-cell carcinoma.

Nearly 40% of lung cancers are adenocarcinoma, which usually originates in peripheral lung tissue. A subtype of adenocarcinoma, the bronchioloalveolar carcinoma, is more common in female never-smokers, and may have a better long term survival.

Squamous-cell carcinoma accounts for about 30% of lung cancers. They typically occur close to large airways. A hollow cavity and associated cell death are commonly found at the center of the tumor. About 9% of lung cancers are large-cell carcinoma. These are so named because the cancer cells are large, with excess cytoplasm, large nuclei and conspicuous nucleoli.

Small-cell lung carcinoma-In small-cell lung carcinoma (SCLC), the cells contain dense neurosecretory granules (vesicles containing neuroendocrine hormones], which give this tumor an endocrine/paraneoplastic syndrome association. Most cases arise in the larger airways [primary and secondary bronchi]. These cancers grow quickly and spread early in the course of the disease. Sixty to seventy percent have metastatic disease at presentation.

In one embodiment the cancer is non-small lung carcinoma.

In one embodiment there is provided treatment of renal cancer, for example renal cell carcinoma and/or urothelial cell carcinoma. Other examples of renal cancer include squamous cell carcinoma, juxtaglomerular cell tumor (reninoma], angiomyolipoma, renal oncocytoma, Bellini duct carcinoma, clear-cell sarcoma of the kidney, mesoblastic nephroma, Wilms' tumor, mixed epithelial stromal tumor, clear cell adenocarcinoma, transitional cell carcinoma, inverted papilloma, renal lymphoma, teratoma, carcinosarcoma, and carcinoid tumor of the renal pelvis. Renal cancer as employed herein refers to cancer of the kidney

In one embodiment the cancer is bladder cancer, for example is any of several types of malignancy arising from the epithelial lining (i.e., the urothelium] of the urinary bladder. About 90% of bladder cancers are transitional cell carcinoma. The other 10% are squamous cell carcinoma, adenocarcinoma, sarcoma, small cell carcinoma, and secondary deposits from cancers elsewhere in the body. The staging of is given below.

T (Primary tumour)

TX Primary tumour cannot be assessed; TO No evidence of primary tumour; Ta Non-invasive papillary carcinoma; Tis Carcinoma in situ ['flat tumour']; Tl Tumour invades subepithelial connective tissue; T2a Tumour invades superficial muscle (inner half); T2b Tumour invades deep muscle (outer half]; T3 Tumour invades perivesical tissue; T3a Microscopically; T3b

Macroscopically (extravesical mass]; T4a Tumour invades prostate, uterus or vagina; T4b Tumour invades pelvic wall or abdominal wall;

N (Lymph nodes)

NX Regional lymph nodes cannot be assessed; NO No regional lymph node metastasis; Nl Metastasis in a single lymph node 2 cm or less in greatest dimension; N2 Metastasis in a single lymph node more than 2 cm but not more than 5 cm in greatest dimension, or multiple lymph nodes, none more than 5 cm in greatest dimension; N3 Metastasis in a lymph node more than 5 cm in greatest dimension

M (Distant metastasis)

MX Distant metastasis cannot be assessed; M0 No distant metastasis; Ml Distant metastasis.

The current disclosure extends to any stage of bladder cancer.

There are more than 30 different types of ovarian cancer which are classified according to the type of cell from which they start. Cancerous ovarian tumors can start from three common cell types: Surface Epithelium - cells covering the lining of the ovaries; Germ Cells - cells that are destined to form eggs; and Stromal Cells - Cells that release hormones and connect the different structures of the ovaries. The present disclosure relates to treatment of ovarian cancer from any source, for example as described herein, in particular epithelium cells. Epithelial ovarian carcinomas [EOCs] account for 85 to 90 percent of all cancers of the ovaries.

Common Epithelial Tumors - Epithelial ovarian tumors develop from the cells that cover the outer surface of the ovary. Most epithelial ovarian tumors are benign (noncancerous). There are several types of benign epithelial tumors, including serous adenomas, mucinous adenomas, and Brenner tumors. Cancerous epithelial tumors are carcinomas - meaning they begin in the tissue that lines the ovaries. These are the most common and most dangerous of all types of ovarian cancers. Unfortunately, almost 70 percent of women with the common epithelial ovarian cancer are not diagnosed until the disease is advanced in stage.

There are some ovarian epithelial tumors whose appearance under the microscope does not clearly identify them as cancerous. These are called borderline tumors or tumors of low malignant potential (LMP tumors). The present disclosure includes treatment of the latter.

Germ Cell Tumors - Ovarian germ cell tumors develop from the cells that produce the ova or eggs. Most germ cell tumors are benign (non-cancerous), although some are cancerous and may be life threatening. The most common germ cell malignancies are maturing teratomas, dysgerminomas, and endodermal sinus tumors. Germ cell malignancies occur most often in teenagers and women in their twenties. Today, 90 percent of patients with ovarian germ cell malignancies can be cured and their fertility preserved.

Stromal Tumors - Ovarian stromal tumors are a rare class of tumors that develop from connective tissue cells that hold the ovary together and those that produce the female hormones, estrogen and progesterone. The most common types are granulosa-theca tumors and Sertoli- Leydig cell tumors. These tumors are quite rare and are usually considered low-grade cancers, with approximately 70 percent presenting as Stage I disease (cancer is limited to one or both ovaries).

Primary Peritoneal Carcinoma-The removal of one's ovaries eliminates the risk for ovarian cancer, but not the risk for a less common cancer called Primary Peritoneal Carcinoma. Primary Peritoneal Carcinoma is closely rated to epithelial ovarian cancer (most common type). It develops in cells from the peritoneum (abdominal lining) and looks the same under a microscope. It is similar in symptoms, spread and treatment

Stages of Ovarian Cancer

Once diagnosed with ovarian cancer, the stage of a tumor can be determined during surgery, when the doctor can tell if the cancer has spread outside the ovaries. There are four stages of ovarian cancer - Stage I (early disease) to Stage IV (advanced disease). The treatment plan and prognosis (the probable course and outcome of your disease) will be determined by the stage of cancer you have.

Following is a description of the various stages of ovarian cancer:

Stage I - Growth of the cancer is limited to the ovary or ovaries.

Stage IA - Growth is limited to one ovary and the tumor is confined to the inside of the ovary.

There is no cancer on the outer surface of the ovary. There are no ascites present containing malignant cells. The capsule is intact. Stage IB - Growth is limited to both ovaries without any tumor on their outer surfaces. There are no ascites present containing malignant cells. The capsule is intact.

Stage IC - The tumor is classified as either Stage IA or IB and one or more of the following are present: (1) tumor is present on the outer surface of one or both ovaries; (2) the capsule has ruptured; and (3) there are ascites containing malignant cells or with positive peritoneal washings.

Stage II - Growth of the cancer involves one or both ovaries with pelvic extension.

Stage HA - The cancer has extended to and/or involves the uterus or the fallopian tubes, or both. Stage IIB - The cancer has extended to other pelvic organs.

Stage IIC - The tumor is classified as either Stage IIA or IIB and one or more of the following are present: (1) tumor is present on the outer surface of one or both ovaries; (2) the capsule has ruptured; and (3) there are ascites containing malignant cells or with positive peritoneal washings.

Stage III - Growth of the cancer involves one or both ovaries, and one or both of the following are present: (1) the cancer has spread beyond the pelvis to the lining of the abdomen; and

(2) the cancer has spread to lymph nodes. The tumor is limited to the true pelvis but with histologically proven malignant extension to the small bowel or omentum.

Stage IIIA - During the staging operation, the practitioner can see cancer involving one or both of the ovaries, but no cancer is grossly visible in the abdomen and it has not spread to lymph nodes. However, when biopsies are checked under a microscope, very small deposits of cancer are found in the abdominal peritoneal surfaces.

Stage IIIB - The tumor is in one or both ovaries, and deposits of cancer are present in the abdomen that are large enough for the surgeon to see but not exceeding 2 cm in diameter. The cancer has not spread to the lymph nodes.

Stage IIIC - The tumor is in one or both ovaries, and one or both of the following is present: (1) the cancer has spread to lymph nodes; and/or (2) the deposits of cancer exceed 2 cm in diameter and are found in the abdomen.

Stage IV - This is the most advanced stage of ovarian cancer. Growth of the cancer involves one or both ovaries and distant metastases (spread of the cancer to organs located outside of the peritoneal cavity] have occurred. Finding ovarian cancer cells in pleural fluid

(from the cavity which surrounds the lungs] is also evidence of stage IV disease.

In one embodiment the ovarian cancer is: type I, for example IA, IB or IC; type II, for example IIA, IIB or IIC; type III, for example IIIA, IIIB or IIIC; or type IV.

Thyroid cancer as employed herein refers to cancer of the thyroid originating from follicular or parafollicular thyroid cells and includes papillary thyroid cancer (75% to 85% of cases]; follicular thyroid cancer (10% to 20% of cases]; medullary thyroid cancer (5% to 8% of cases]- cancer of the parafollicular cells, often part of multiple endocrine neoplasia type 2; poorly differentiated thyroid cancer; anaplastic thyroid cancer (less than 5% of cases] is not responsive to treatment and can cause pressure symptoms, thyroid lymphoma, squamous cell thyroid carcinoma, sarcoma of thyroid. The present disclosure extends to treatment of the same.

Bladder cancer as employed herein refers to cancer of the bladder including transitional cell bladder cancer, carcinoma in situ, papillary cancer and rarer types of bladder cancer such as squamous cell cancer and adenocarcinoma. The present disclosure relates to treatment of the same.

Esophageal cancer as employed herein refers to cancer of the oesphagus including esophageal squamous-cell carcinomas, esophageal adenocarcinomas, and variants of squamous- cell carcinoma, and non-epithelial tumors, such as leiomyosarcoma, malignant melanoma, rhabdomyosarcoma, lymphoma, among others. The present disclosure extends to treatment of the same.

Head and neck cancer as employed herein refers to cancer of the neck and/or head, including mouth cancer, nasopharyngeal cancer, oropharyngeal cancer, paranasal sinus cancer and salivary gland cancer. The present disclosure relates to treatment of the same.

In one embodiment the treatment of the present disclosure is neo-adjuvant therapy, for example to shrink the tumour/carcinoma before surgery to remove the cancerous tissue or before chemotherapy to improve the chances of success of the latter or to reduce the severity of the treatment required.

In one embodiment the treatment of the present disclosure is adjuvant therapy, for example following surgery to remove the cancerous tissue.

In one embodiment the treatment of the present disclosure is adjuvant therapy, for example following chemotherapy.

In patients where not all the cancerous tissue is removed by surgery then the patient may benefit from adjuvant therapy which is monotherapy employing a compound of formula (I), such as Varlitinib.

In patients where not all the cancerous tissue is removed by surgery then the patient may benefit from combination adjuvant therapy comprising a compound of formula [I] and chemotherapy or radiotherapy.

First line therapy as employed herein is the first therapy employed for the treatment of the cancer and in some instances the first line therapy may be neo-adjuvant therapy, in this context surgery will generally be considered a treatment

Second line therapy as employed herein is treatment following first line therapy and may be adjuvant therapy. Thus, in the context of the present specification second line therapy is simply therapy other than first line therapy and includes, third line therapy, fourth line therapy etc.

Monotherapy as employed herein is wherein the compound of formula (I) an enantiomer thereof and/or a pharmaceutically acceptable salt thereof, is the only active agent being administered to the patient for the treatment of cancer.

In one embodiment there is provided treatment of the patient population according to the present disclosure with a combination therapy comprising Varlitinib, for example Varlitinib and an anti-cancer therapy, such as chemotherapy.

Combination therapy as employed herein refers to wherein the compound of formula (I) an enantiomer thereof or a pharmaceutically acceptable salt thereof is employed for the treatment of the cancer in conjunction with one or more further anticancer treatments, for example where the treatment regimens for the two or more active anticancer agents overlap or where the two or more anticancer agents are administered concomitantly. In one embodiment the combination therapy according to the present disclosure comprises a RON inhibitor, for example as disclosed WO2008/058229, incorporated herein by reference.

In one embodiment the combination therapy comprises a checkpoint inhibitor, such as a CTLA4 inhibitor, a PD-1 inhibitor or a PD-Ll inhibitor, in particular an antibody or binding fragment thereof.

Examples of pharmaceutically acceptable salts include but are not limited to acid addition salts of strong mineral acids such as HC1 and HBr salts and addition salts of strong organic acids, such as a methansulfonic acid salt, tosylates, furoates and the like, including di, tri salts thereof, such as ditosylates.

Analysis of patients to profile their cancer, for example to establish if their cancer is EGFR and HER2 positive is known and is routine in the art. Establishing if a cancer is HER2 amplified is also routine in the art

Chemotherapeutic Agents

Chemotherapeutic agent and chemotherapy or cytotoxic agent are employed interchangeably herein unless the context indicates otherwise.

Chemotherapy as employed herein is intended to refer to specific antineoplastic chemical agents or drugs that are "selectively" destructive to malignant cells and tissues, for example alkylating agents, antimetabolites including thymidylate synthase inhibitors, anthracyclines, anti- microtubule agents including plant alkaloids, taxanes, topoisomerase inhibitors, parp inhibitors and other antitumour agents. Selectively in this context is used loosely because of course many of these agents have serious side effects.

The preferred dose may be chosen by the practitioner, based on the nature of the cancer being treated.

Examples of alkylating agents, which may be employed in the method of the present disclosure include a platinum alkylating agent, nitrogen mustards, nitrosoureas, tetrazines, aziridines, platins and derivatives, and non-classical alkylating agents (such as procarbazine, altretamine and dacarbazine).

Examples of platinum containing chemotherapeutic agents (also referred to as platins], include cisplatin, carboplatin, oxaliplatin, satraplatin, picoplatin, nedaplatin, triplatin and lipoplatin (a liposomal version of cisplatin], in particular cisplatin, carboplatin and oxaliplatin. Others include eptaplatin, lobaplatin, miriplatin and dicycloplatin.

The dose for cisplatin ranges from about 20 to about 270 mg/m ² depending on the exact cancer. Often the dose is in the range about 70 to about 100mg/m ² .

Nitrogen mustards include mechlorethamine, cyclophosphamide, melphalan, chlorambucil, ifosfamide and busulfan.

Nitrosoureas include iV-Nitroso-N-methylurea (MNU), carmustine (BCNU], lomustine (CCNU] and semustine (MeCCNU], fotemustine and streptozotocin. Tetrazines include dacarbazine, mitozolomide and temozolomide.

Aziridines include thiotepa, mytomycin and diaziquone (AZQJ.

Examples of antimetabolites, which may be employed in the method of the present disclosure, include anti-folates (for example methotrexate and pemetrexed], purine analogues (for example thiopurines, such as azathiopurine, mercaptopurine, thiopurine, fludarabine (including the phosphate form], pentostatin and cladribine], pyrimidine analogues [for example fluoropyrimidines, such as 5-fluorouracil (5-FU] and prodrugs thereof such as capecitabine [Xeloda®]], floxuridine, gemcitabine, cytarabine, decitabine, raltitrexed (tomudex) hydrochloride, cladribine and 6-azauracil.

Examples of anthracyclines, which may be employed in the method of the present disclosure, include daunorubicin (Daunomycin], daunorubicin (liposomal], doxorubicin (Adriamycin], doxorubicin (liposomal], epirubicin, idarubicin, valrubicin currently are used only to treat bladder cancer and mitoxantrone an anthracycline analog, in particular doxorubicin.

Examples of anti-microtubule agents, which may be employed in the method of the present disclosure, include include vinca alkaloids and taxanes.

Vinca alkaloids include completely natural chemicals for example vincristine and vinblastine and also semi-synthetic vinca alkaloids, for example vinorelbine, vindesine, and vinflunine

Taxanes, which may be employed in the present disclosure, include paclitaxel, docetaxel, abraxane, carbazitaxel and derivatives of thereof. Derivatives of taxanes as employed herein includes reformulations of taxanes like taxol, for example in a micelluar formulations, derivatives also include chemical derivatives wherein synthetic chemistry is employed to modify a starting material which is a taxane.

Topoisomerase inhibitors, which may be employed in a method of the present disclosure, include type I topoisomerase inhibitors, type II topoisomerase inhibitors and type II topoisomerase poisons, Type I inhibitors include topotecan, irinotecan, indotecan and indimitecan. Type II inhibitors incl which has the following structure:

Type II poisons include amsacrine, etoposide, etoposide phosphate, teniposide and doxorubicin and fluoroquinolones.

In one embodiment the chemotherapeutic is a PARP inhibitor.

In one embodiment a combination of chemotherapeutic agents employed is, for example a platin and 5-FU or a prodrug thereof, for example cisplatin or oxaplatin and capecitabine or gemcitabine, such as FOLFOX.

In one embodiment the chemotherapy comprises a combination of chemotherapy agents, in particular cytotoxic chemotherapeutic agents.

In one embodiment the chemotherapy combination comprises a platin, such as cisplatin and fluorouracil or capecitabine.

In one embodiment the chemotherapy combination is capecitabine and oxaliplatin (XELOX].

In one embodiment the chemotherapy is a combination of folinic acid and 5-FU, optionally in combination with oxaliplatin (FOLFOX]. In one embodiment the chemotherapy is a combination of folinic acid, 5-FU and irinotecan (FOLFIRI), optionally in combination with oxaliplatin (FOLFIRINOX). The regimen, for example includes: irinotecan [180 rag/m ² IV over 90 minutes) concurrently with folinic acid (400 mg/ra ² [or 2 x 250 mg/m ² ] IV over 120 minutes); followed by fluorouracil (400-500 mg/m ² IV bolus) then fluorouracil (2400-3000 mg/m ² intravenous infusion over 46 hours). This cycle is typically repeated every two weeks. The dosages shown above may vary from cycle to cycle.

In one embodiment the combination therapy employs a microtubule inhibitor, for example vincristine sulphate, epothilone A, N- [2- [(4-Hydroxyphenyl)amino]-3-pyridinyl]-4- methoxybenzenesulfonamide (ABT-751), a taxol derived chemotherapeutic agent, for example paclitaxel, abraxane, or docetaxel or a combination thereof.

In one embodiment the combination therapy employs an mTor inhibitor. Examples of mTor inhibitors include: everolimus (RADOOl), WYE-354, KU-0063794, papamycin (Sirolimus), Temsirolimus, Deforolimus(MK-8669), AZD8055 and BEZ235 (NVP-BEZ235).

In one embodiment the combination therapy employs a MEK inhibitor. Examples of MEK inhibitors include: AS703026, CI- 1040 (PD 184352), AZD6244 (Selumetinib), PD318088, PD0325901, AZD8330, PD98059, U0126-EtOH, BIX 02189 or BIX 02188.

In one embodiment the combination therapy employs an AKT inhibitor. Examples of AKT inhibitors include: MK-2206 and AT7867.

In one embodiment the combination therapy employs an aurora kinase inhibitor. Examples of aurora kinase inhibitors include: Aurora A Inhibitor I, VX-680, AZD1152-HQPA (Barasertib), SNS-314 Mesylate, PHA-680632, ZM-447439, CCT129202 and Hesperadin.

In one embodiment the combination therapy employs a p38 inhibitor, for example as disclosed in WO2010/038086, such as iV-[4-({4- [3-(3-tert-Butyl-l-p-tolyl-l//-pyrazol-5- yl)ureido]naphthalen-l-yloxy}methyl)pyridin-2-yl]-2-methoxya cetamide.

In one embodiment the combination therapy employs a Bcl-2 inhibitor. Examples of Bcl-2 inhibitors include: obatoclax mesylate, ABT-737, ABT-263(navitoclax) and TW-37.

In one embodiment the chemotherapy combination comprises an antimetabolite such as capecitabine (xeloda), fludarabine phosphate, fludarabine (fludara), decitabine, raltitrexed (tomudex), gemcitabine hydrochloride and/or cladribine.

In one embodiment the combination therapy comprises ganciclovir, which may assist in controlling immune responses and/or tumour vasculation.

In one embodiment the combination therapy comprises a checkpoint inhibitor, for example selected from the group comprising:

PF-477736 is a CHK1 & 2 inhibitor (ATP-

AZD-7762 a CHK1 and CHK2 inhibitor (with an competitive CHK1 inhibitor with a Kj of 0.49 nM IC50 of about 5 nM in a cell-free assay, with less in a cell free assay. The molecule also VEGFR2, potent activity against CAM, Yes, Fyn, Lyn, Hck Aurora-A, FGFR3, Flt3, Fms (CSF-1R), Ret and an Lck) Yes. It shows approximately 100 fold selectivit for CHKl over CHK2)

V155422, V155991 and V158411 are CHKl and CHK2 inhibitors,

CCT244747 a CHKl inhibitor

GDC-0575 [structure not given];

CP-466722 is a potent ATM inhibitor, which does not inhibit ATR. The molecule also inhibits Pi3K and PIKK

US FDA approval]

PARP- 1 &-2 inhibitor

INO-1001 (3-aminobenzamide) a potent inhibitor of PARP [with IC50 of <50 nM in CHO cells and a mediator of oxidant-induced myocyte dysfunction during reperfusion]

BGB-290 a PARP-1 and PARP-2 inhibitor (structure not shown); MP-124 a PARP-1 inhibitor (structure not shown); or a pharmaceutically acceptable salt or solvate of any one of the same. The following articles disclose certain checkpoint inhibitors: Drugs Fut 2003, 28(9): 881 Cell cycle inhibitors for the treatment of cancer Kong, N., Fotouhi, N., Wovkulich, P.M., Roberts, J; Novel pyrrole derivatives as selective CHKl inhibitors: design, regioselective synthesis and molecular modelling Med. Chem. Commun., 2015,6, 852-859; PLoS One 2010; 5(8): el2214. Binding of protein kinase inhibitors to synapsin I inferred from pair-wise binding site similarity measurements. In one embodiment the checkpoint inhibitor is a PARP inhibitor. Examples of PARP inhibitors are disclosed in US7,449,464 and US8,071,623. The compounds disclosed in this paragraph are incorporated herein by reference, In one embodiment the checkpoint inhibitor is an antibody or binding fragment specific to a checkpoint protein, in particular one disclosed herein.

In one embodiment the checkpoint kinase inhibitor is independently selected from: 3- [(Aminocarbonyl)amino]-5-(3-fluorophenyl)-N-(3S)-3-piperidin yl-2-thiophenecarboxamide hydrochloride; (3R,4S)-4- [[2-(5-Fluoro-2-hydroxyphenyl)-6,7-dimethoxy-4-quinazolinyl] amino]- a, x-dimethyl-3-pyrrolidinemethanol dihydrochloride; 4,4'-diacetyldiphenylurea bis(guanylhydrazone) ditosylate; 9-Hydroxy-4-phenyl-pyrrolo [3,4-c]carbazole-l,3(2H,6H)-dione; (R)- -Amino-N- [5,6-dihydro-2-(l-methyl-lH-pyrazol-4-yl)-6-oxo-lH-pyrrolo[4 ,3,2-ef] [2,3]benzo diazepin-8-yl]-cyclohexaneacetamide; 9,10,11,12-Tetrahydro- 9,12-epoxy-lH-diindolo[l,2,3-fg: 3',2 ',l'-kl]pyrrolo[3,4-i] [l,6]benzodiazocine-l,3 (2H)-dione; 4'- [5- [[3-[(Cyclo propylamino) methyl]phenyl]amino]-lH-pyrazol-3-yl]- [l,l'-biphenyl]-2,4-diol; and (R)-5-((4-((Morpholin-2- ylmethyl)amino)-5-(trifluoromethyl)pyridin-2-yl)amino)pyrazi ne-2-carbonitrile (CCT245737).

In one embodiment the check point inhibitor is prexasertib.

Checkpoint kinase inhibitor as employed herein refers to an inhibitor that reduces or eliminates the biological activity of a cell regulatory checkpoint kinase 1 and/or 2. Cells that suffer DNA damage activate the checkpoint kinases CHKl and CHK2, which signal to initiate the DNA repair processes, limit cell-cycle progression and prevent cell replication, until the damaged DNA is repaired.

In one embodiment after combination therapy a monotherapy comprising a compound of formula (I), such as Varlitinib (as defined herein including doses described above) is employed, for example a maintenance therapy.

"Comprising" in the context of the present specification is intended to mean "including". Where technically appropriate, embodiments of the invention ma be combined.

Embodiments are described herein as comprising certain features/elements. The disclosure also extends to separate embodiments consisting or consisting essentially of said features/elements.

Technical references such as patents and applications are incorporated herein by reference.

Any embodiments specifically and explicitly recited herein may form the basis of a disclaimer either alone or in combination with one or more further embodiments.

The present application claims priority from SG10201708780Q filed 25 October 2017 and SG10201808394V filed 26 September 2018, the contents of which are incorporated herein by reference. The priority document documents may also be employed as basis to make correction.

The invention will now be described with reference to the following examples, which are merely illustrative and should not in any way be construed as limiting the scope of the present invention.

BRIEF SUMMARY OF THE FIGURES

Figure 1 ErbB family expression and its phosphorylation in representative HCC PDXs and the anti-tumour effects of Varlitinib on 3 PDXs with activated ErbB2/3. (A and B) Western blot shows the expression of EGFR, ErbB2, ErbB3, and its phosphorylation in representative HCC PDXs. (C-E) Growth curves and tumour weights of three high p-ErbB2/p-ErbB3 expressing PDXs, (C) HCCOl-0708, (D) HCC07-0409, and (E) HCC29-0909A, treated with three different doses of Varlitinib, 25, 50, and lOOmg/kg BID. Treatment was started when the tumours reached the size of approximately 120-150 mm ³ . The tumour volumes were measured every 2-3 days and the tumour weights were measured at the end of the experiments. Shown are mean and SD (n=8-10).

Figure 2 Dose- and time-dependent ErbB family pathway inhibition by Varlitinib in

HCC29-0909A PDX model. Mice bearing HCC29-0909A PDX were treated with vehicle control, 25, 50, or lOOmg/kg BID. Tumours were collected at day 2 (A) and day 14 (B) post- Varlitinib treatment Two tumours from each condition were lysed and equal amount of the protein lysates were used for Western blot analysis. Blots were incubated with indicated antibodies. Representative blots were shown.

Figure 3 Effects of Varlitinib on tumour cell proliferation, tumour cell death, and vessel normalisation in HCC29-0909A PDX model. Mice bearing HCC29-0909A PDX were treated with vehicle control and lOOmg/kg BID. Tumours collected at day 14 were processed for paraffin [for cleaved PARP and p-histone 3 SerlO staining] or Tissue-Tek embedding (for CD 31 staining). (A) Representative images of tumour sections from vehicle-treated and Varlitinib-treated mice stained for p- Histone H3 (SerlO), cleaved PARP, and CD31. (B) Representative images of tumour sections from vehicle-treated [top) and Varlitinib-treated (bottom) mice perfused with biotinylated Lycopersicon Esculentum (Tomato) lectin and pimonidazole hydrochloride, followed by immunohistochemical staining. Representative photographs were shown.

Figure 4 Transcriptome analysis of Varlitinib-treated HCC29-0909A and HCCOl-0708

HCC PDX models. (A) The area-proportional Venn diagram analysis showing the common activated genes (left) and the common repressed genes (right in lOOmg/kg BID and 50mg/kg BID Varlitinib-treated HCC29-0909A model for 14 days. (B) The heatmap showing the common Varlitinib-dysregulated 2331 genes in HCC29-0909A model. (C) The area-proportional Venn diagram analysis showing the common activated genes (left) and the common repressed genes (right) in lOOmg/kg BID and 50mg/kg BID Varlitinib-treated HCCOl-0708 model. (D) The heatmap showing the common Varlitinib-dysregulated 457 genes in HCCOl-0708 model. (E) The 3-way area-proportional Venn diagram analysis showing the common activated genes (left) and the common repressed genes (right) in lOOmg/kg BID and 50mg/kg BID Varlitinib-treated HCC29-0909A model as well as lOOmg/kg BID Varlitinib-treated HCCOl-0708 model.

Figure 5 Dose-dependent β-catenin pathway inhibition and membrane translocation of β-catenin by Varlitinib in the tested HCC29-0909A and HCCOl-0708 PDX models. Mice bearing HCC29-0909A (A) or HCCOl-0708 (B) with three different doses of Varlitinib, 25, 50, and lOOmg/kg BID for indicated time. Treatment was started when the tumours reached the size of approximately 120-150 mm ³ . Two tumours from each condition were lysed and equal amount of the protein lysates were used for Western blot analysis for the β-catenin, its related signalling molecules, and downstream targets of β-catenin. (C) Representative images of tumour sections from vehicle-treated (top) and Varlitinib-treated (bottom) HCC29- 0909A, HCC07-0409, and HCCO l-0708 PDXs stained for β-catenin.

Figure 6 Hypothetic model of the Varlitinib-mediated tumour regression and vessel normalisation in HCC.

Figure 7 ErbB family expression in additional 28 hepatocellular carcinoma (HCC) patient-derived xenografts (PDXs) and the anti-tumour effects of Varlitinib on the high ErbB2/3-expressing PDXs. (A and B) Western blot shows the expression of EGFR, EbB2, ErbB3, and its phosphorylation in 28 HCC PDXs. (C-D) Growth curves and tumour weights of HCC21-0208 (C) and HCC16-1014 (D) treated with Varlitinib at lOOmg/kg BID. Treatment was started when the tumours reached the size of approximately 120-150 mm ³ . The tumour volumes were measured every 2-3 days and the tumour weights were measured at the end of the experiments. Shown are mean and SD (n=5).

Figure 8 Dose- and time-dependent ErbB family pathway inhibition by Varlitinib in

HCCOl-0708 PDX model. Mice bearing HCCOl-0708 PDX were treated with vehicle control, 25, 50, or lOOmg/kg BID. Tumours were collected at day 2 (A) and day 11 (B) post-Varlitinib treatment Two tumours from each condition were lysed and equal amount of the protein lysates were used for Western blot analysis. Blots were incubated with indicated antibodies. Representative blots were shown. Dose-dependent inhibition of tumour cell proliferation, induction of apoptosis and formation of capillary-like blood vessels by Varlitinib in HCC29-0909A PDX model. Mice bearing HCC29-0909A PDX were treated with vehicle control, 25, 50, or l OOmg/kg BID. Tumours collected at day 14 were processed for paraffin [for cleaved PARP and p-histone 3 SerlO staining] or Tissue- Tek embedding [for CD31 staining]. Representative images shown of tumour sections from vehicle-treated and Varlitinib-treated (25, 50, or lOOmg/kg BID] mice stained for p-Histone H3 [SerlO] [upper], cleaved PARP (middle], and CD31 [bottom].

Tumour regression by Varlitinib treatment in three other PDX models, HCCOl-0708, HCC07-0409, and HCC21-0208. Mice bearing indicated PDX were treated with vehicle control or lOOmg/kg BID. Tumours collected at indicated day were processed for paraffin [for cleaved PARP and p-histone 3 SerlO staining] or Tissue-Tek embedding (for CD31 staining]. Representative images of tumour sections from vehicle-treated and Varlitinib-treated mice stained for p-Histone H3 [SerlO] [upper], cleaved PARP (middle], and CD31 (bottom].

Effect of Varlitinib on vessel normalisation in HCC29-0909A PDX model. Mice bearing HCC29-0909A PDX were treated with vehicle control and lOOmg/kg BID. Tumours collected at day 14 were processed for immunohistochemistry. Representative images of tumour sections from vehicle-treated and Varlitinib- treated mice perfused with biotinylated Lycopersicon Esculentum (Tomato] followed by immunohistochemical staining. Representative photographs were shown. Magnification 100X [Top], 200X (Middle] and 400X [Bottom].

Identification of Varlitinib-resistant gene signature by global gene expression analysis. (A) Heat-map showing the identified differentially expressed genes from the quantile normalised data of four HCC PDXs, HCC21-0208, HCCOl-0708, HCC07- 0409, and HCC29-0909A with high stringent cut-off threshold of >2 and <-2 fold- change and FDR adjusted p- value < 0.0001. (B) The expression of NRG1 in the four analysed PDXs indicated by normalised probe intensity. Order of data points from left to right are HCC21-0208, HCCOl-0708, HCC07-0409 and HCC29-0909A. See also Table 12. (C) The expression of ErbB family in the four analysed PDXs indicated by normalised probe intensity. For each gene, the order of the data points from left to right are HCC21-0208, HCCOl-0708, HCC07-0409 and HCC29-0909A. Identification of β-catenin mutation and the constitutively active β-catenin pathways in the Variitinib-sensitive PDXs. (A) The expression of β-catenin and its downstream targets in the four analysed PDXs indicated by normalised probe intensity. (B) The expression of Wnt/TGF -related targets in the four analysed PDXs indicated by normalised probe intensity. For each gene, the order of the data points from left to right are HCC21 -0208, HCCOl-0708, HCC07-0409 and HCC29- 0909A. ** indicates p-value < 0.01, *** indicates p-value < 0.001, **** indicates p- value < 0.0001. See also Table 13.

Identification of Varlitinib-potency gene signature by transcriptome analysis.

Heat-map showing the identified differentially expressed genes from the normalised RNA-Seq data of HCCOl-0708 and HCC29-0909A with cut-off threshold of >2 and <-2 fold-change and FDR adjusted p-value < 0,05.

Genetic alterations of CTNNB1, EGFR, ERBB2, ERBB3, and ERBB4 in TCGA HCC data. Oncoprint displaying the genetic alterations of the indicated genes in the TCGA HCC dataset [n=360] (A) and the subsets of HCC which harbour any types of β-catenin mutation (n-96) (B). Colour coding indicates types of genetic alterations: dark grey, homozygous amplification; light grey, heterozygous amplification; grey, heterozygous deletion; and black, mutation. Left, genetic alteration percentage, shows dose-dependent tumour volume growth inhibition in HCC patient derived xenograft model of HCC29-0909A after administration of 25mg/kg BID, 50 mg/kg BID or lOOmg/k BID ofvarlitinib.

shows dose-dependent tumour volume growth inhibition in HCC patient derived xenograft model of HCCOl-0708 after administration of 25mg/kg BID, 50 mg/kg BID or lOOmg/kg BID ofvarlitinib.

shows the results of an in vitro experiment to investigate the induction of apoptosis by varlitinib in HCC cell lines, including sorafenib-resistant cell lines,

shows the apoptosis (AnnexinV) profile for PLC/PRF/5 cells after 48 hour culture in the presence of varlitinib.

Frequent upregulation of ErbB3 in HCC and in vitro effect of Varlitinib in high ErbB3-expressing liver cancer cell lines. (A) ERBB3 expression in various publicly available HCC dataset T: HCC tumours, NT: non-cancerous tissues; *p < 0.05, ***p < 0.001, ****p < 0.0001, paired or unpaired t test was used if data passed normality test, Wilcoxon matched paired signed rank test or unpaired Mann- Whitney test was used if data did not pass normality test. (B) Dose-dependent inhibition of Varlitinib in representatives of ERBB3Vi\gh and ERBB3Low liver cancer cell lines. PLC/PRF/5, Hep3B, HepG2/C3A, and Huh7 are ERBB3 {\ \ cell lines and SNU182, SNU423, SNU449, and SNU475 are ERBB3Low cell lines. Different doses of Varlitinib in normal culture medium with 1% DMSO were added into the 6-well plates seeded with different cell lines. The cell lines were cultured for 2-4 weeks, followed by fixation and Giemsa staining. (C) Inhibition of ErbB receptor family and their downstream signalling molecules by Varlitinib. The ERBB3H\g PLC/PRF/5 cell line was treated with 3μΜ of Varlitinib for indicated time. Blots were incubated with indicated antibodies, a- Tubulin was used as equal loading marker. See also Figure 25.

Identification of Varlitinib responder and non-responder PDXs and the importance of ErbB2/3 dependence in HCC. (A) In vivo effect of Varlitinib in treated PDXs. The 3 Varlitinib responders, HCCOl-0708, HCC07-0409, and HCC29- 0909A models, were treated with vehicle control, 25, 50, and l OOmg/kg BID [in the red box], whereas the 3 Varlitinib non-responders, HCC21-0208, HCC16-1014, and HCC26-0808A, were treated with vehicle control and lOOmg/kg BID fin the yellow box). Tumour volume was measured every 2-3 days. Data represent mean ± SEM (n = 5-10). **p < 0.01, ****p < 0.0001, ns, not significant. Comparisons with vehicle control- treated PDXs [unpaired t test). (B) Western blot analysis of 3 responders and 2 non-responders. oc-Tubulin was used as equal loading marker. (C) Gene expression analysis of NRGl, EGFR, ERBB2, and ERBB3 between responders and non-responders. Data represent mean ±SEM [responders n=9 and non-responders n=27), *p < 0.05, ****p < 0.0001 [unpaired t test). (D) Survival analysis of ERBB2, ERBB3, and NRGl in HCC. Cutoff finder [Budczies et al., 2012) was used to identify the best cutoff gene expression value for ERBB2 [left), ERBB3 (middle), and NRGl [right) in the Singapore HCC dataset[44] [n=48). HR represents hazard ratio. See also Figure 26.

Figure 22 (A) Enrichment plots of the two highly enriched and significant gene sets in

Varlitinib responders and one in Varlitinib non-responders. (B) Expression of the reported G6 signature-related genes in Varlitinib responders and nonresponders. Data represent mean ± SEM (responders n=9 and non-responders n=27), **p <0.01, ****p < 0.0001 (unpaired t test), (C) Correlation analyses between ERBB3 and G6- related genes (LEF1 and AXIN2) and between ERBB3 and IGF1R. Pearson correlation analysis is used when both data passed normality test, whereas Spearman correlation analysis is used when one of the data did not pass normality test See also Figure 27.

Figure 23 Transcriptomic analysis identifies dose-dependant responses and the potential inhibition mechanisms in responder PDXs. (A) Three-way area proportional Venn diagram analysis of differential expressed gene (DEG) analysis from RNA-Seq analysed Varlitinib-treated PDXs. (B) The normalised read counts were analysed based on gene sets of glycolysis and gluconeogenesis (left) and hypoxia (right). Data from control (indicated in blue box), 50mg/kg (indicated in red box), and lOOmg/kg BID (indicated in green box) of Varlitinib-treated HCC29- 0909A are shown. Data from control (indicated in blue box) and lOOmg/kg BID (indicated in green box) of Varlitinib-treated HCC01-0708 are shown. (C) Ingenuity Pathway Analysis (IP A) predicted HIF1A as upstream regulator in the treated PDXs.

(D) Repressed expression of HIF1A in treated PDXs. Data represent mean ± SEM (n = 3). **p < 0.01. Comparisons with vehicle control-treated PDXs (unpaired t test).

(E) Repressed expression of LGR5 and upregulated CDH1 in treated PDXs. Data represent mean ± SEM (n = 3), **p < 0.01, ****p < 0.0001, ns, not significant Comparisons with vehicle control-treated PDXs (unpaired t test). See also Figure 28.

Figure 24 Varlitinib-mediated β-catenin pathway inhibition in the treated responder

PDXs. (A-B) Dose-dependent β-catenin pathway inhibition by Varlitinib. Western blot analysis of the protein lysates collected on day 14 post-treatment of Varlitinib in HCC29-0909A (A) and on day 11 post-treatment of Varlitinib in HCCOl-0708 (B). Two tumours from each condition were lysed and equal amount of the protein lysates were used for Western blot analysis. Blots were incubated with indicated antibodies. a-Tubulin was used as equal loading marker. Representative blots were shown. See also Figure 30.

Analysis of ERBB3 ^m and ERBB3 ^ov/ liver cancer cell lines. (A) Identification of ERBB3 ^m and ERBB3 ^ow liver cancer cell lines. Expression of ERBB3 was compared among 27 liver cancer cell lines in Cancer Cell Line Encyclopedia (CCLE) dataset. Median expression of ERBB3 in these 27 cell lines was used as as cut-off for ERBB3 ^Hi and ERBB3 ^l0V1 cell lines. **"p < 0.0001, unpaired t test. (B) Expression of phospho- and total ErbES family receptors in selected ERBB3 ^m and £7?f¾?.3 ^Low cell lines. Blots were incubated with indicated antibodies. α-Tubulin was used as equal loading marker. (C) 72-hour cytotoxicity test of Varlitinib in ERBB3 ^Hi cell lines. Different concentrations of Varlitinib in 1%DMS0 was added into the cells for 72 hours. CyQUANT NF Cell Proliferation Assay Kit was used to measure the absorbance and GraphPad Prism version 7.00 was used to determine the cytotoxicity.

Varlitinib treatment in HCC PDXs. (A) ErbB family expression in 56 HCC (PDXs. Western blot shows the expression of EGFR, ErbB2, ErbB3, and its phosphorylation in 56 HCC PDXs. Blots were incubated with indicated antibodies. α-Tubulin was used as equal loading marker. (B) In vivo effect of Varlitinib in treated non- responder PDXs. Additional 11 HCC PDXs, HCCOl-0909, HCC06-0606, HCC09-0913, HCC13-0109, HCC13-0212, HCC 5-0114, HCC17-0211, HCC25-0705A, HCC26- 0808B, HCC29-1104, and HCC30-0805B models, were treated with vehicle control and l OOmg/kg BID Tumour volume was measured every 2-3 days. Data represent mean ± SEM (n = 5-10). ns, not significant. Comparisons with vehicle control- treated PDXs (unpaired t test]. (C) Principal component analysis of quantile normalised gene expression dataset. Smaller circle indicates Varlitinib responders, larger circle circle indicates non-responders. Triplicate samples of 3 responder PDXs and 9 non-responder PDXs were analysed by Affymetrix Human Genome U133 Plus 2.0 microarray. (D) Heatmap of the DEG.

Analysis of responders and non-responders. Gene expression analysis of TGFBR1, TGFBR2, TEAD1, IGF2, IGF1R, IRS2, N0TCH1, NOTCH2, and JAG1 between responders and non-responders. Triplicate samples of 3 responder PDXs and 9 non- responder PDXs were analysed by Affymetrix Human Genome U133 Plus 2.0 microarray. Probe intensities in each sample were quantile normalised. Data represent mean ± SEM (responders n=9 and non-responders n=27), *p < 0.05, **p < 0.01, *"p < 0.001, ""p < 0.0001 (unpaired t test).

Transcriptome analysis of Varlitinib treatment in HCC29-0909A, HCCOl-0708, and HCC16-1014. (A-B) Area proportional Venn diagram analysis of differential expressed gene (DEG) analysis from RNA-Seq analysed Varlitinib-treated responder PDXs, HCC29-0909A (A) and HCCOl-0708 (B). (C-E) Downregulation of pro- angiogeneic factors by Varlitinib in treated responder PDXs. (C) Inhibition of EPO, PDGFA, NRP2, BMP2 in both Varlitinib-treated HCC29-0909A and HCCOl-0708. (D) Inhibition of ANG, PGF, VEGFB in Varlitinib-treated HCC29-0909A. (E) Inhibition of PDGFC in Varlitinib-treated HCCOl-0708. Data represent mean ± SEM [n = 3). *p < 0.05, **p < 0.01, "*p < 0.001, **"p < 0.0001. Comparisons with vehicle control- treated PDXs [unpaired t test]. (F-H) Downregulation of β-catenin- and YAP1- related genes by Varlitinib in treated responder PDXs. (F) Inhibition of YAP 1, IRX3, and MYC in both Varlitinib-treated HCC29-0909A and HCCOl-0708. (G] Inhibition of LEF1 in Varlitinib-treated HCC29-0909A. (H) Inhibition of SOX9, HEY1, CTGF, CYR61 in Varlitinib-treated HCCOl-0708. Data represent mean ± SEM [n = 3). *p < 0.05, "p < 0.01, *"p < 0.001. Comparisons with vehicle control-treated PDXs [unpaired t test). (I) Pathways enriched in HCCO l-0708 when compared to HCC29- 0909A by Ingenuity Pathway Analysis [IPA]. (J) CTNNB1 expression in control of HCC29-0909A, control of HCCOl-0708, and lOOmg/kg Varlitinib-treated HCCO l- 0708. ^« *p < 0.001. Comparison with vehicle control-treated HCCOl-0708 PDX [unpaired t test]. (K) Heatmap of the inhibition of embryonic stem cell core gene set in Varlitinib-treated HCCOl-0708. Control of HCC29-0909A, Control of HCC16- 1014, and Varlitinib-treated HCC16-1014 are included for the comparison. Data from control [indicated in dark grey box] of HCC29-0909A, from control [indicated in dark grey box) and lOOmg/kg BID [indicated in light grey box) of Varlitinib- treated HCCOl-0708, and from control [indicated in dark grey box) and 75mg/kg BID (indicated in light grey box) of Varlitinib-treated HCC16-1014 are shown. (L & M) Heatmap of the inhibition of YAP signalling in Varlitinib-treated PDXs. YAP conserved signature (L) and GO Hippo signalling (M) gene sets were used. Data from control (indicated in dark grey box), 50mg/kg (indicated in black box), and lOOmg/kg BID (indicated in light grey box) of Varlitinib-treated HCC29-0909A, from control (indicated in dark grey box) and lOOmg/kg BID (indicated in light grey box) of Varlitinib-treated HCCOl-0708, and from control (indicated in dark grey box) and 75mg/kg BID (indicated in light grey box) of Varlitinib-treated HCC16- 1014 are shown.

Figure 29 Dose-dependent ErbB family pathway inhibition by Varlitinib in responder

PDX models. Western blot analysis of the protein lysates collected on day 14 post- treatment of l OOmg/kg QD Varlitinib in HCC07-0409 PDX model. Two tumours from each condition were lysed and equal amount of the protein lysates were used. Blots were incubated with indicated antibodies. a-Tubulin was used as equal loading marker. Representative blots were shown.

Figure 30 Varlitinib-mediated β-catenin pathway inhibition in the treated responder

PDXs. (A-B) β-catenin pathway inhibition by Varlitinib. Western blot analysis of the protein lysates collected on day 14 post-treatment of Varlitinib in HCC29-0909A (A) and on day 11 post-treatment of Varlitinib in HCC07-0409 (B). Two tumours from each condition were lysed and equal amount of the protein lysates were used for Western blot analysis. Blots were incubated with indicated antibodies, a- Tubulin was used as equal loading marker. Representative blots were shown. Figure 31 shows hypothetical model of the Varlitinib-mediated tumour growth inhibition and vessel normalisation in HCC.

EXAMPLES

Example 1

This study demonstrates Varlitinib-mediated tumour regression and vascular normalisation in ErbB-dependent and β-catenin mutated hepatocellular carcinoma.

Materials and Methods

Reagents

Antibodies against EGFR, ErbB2, ErbB3, Akt, β-catenin, Axin2, Axinl, c-met, survivin, N- cadherin, E-cadherin, Dvl3, Dvl2, Cdc25C, p27, p21, E2F1, Rb, Cleaved caspase 3, cleaved caspase 7, cleaved PARP, and phosphorylarJon-specific antibodies against Akt Ser473, LRP6 Serl490, RanBP3 Ser58, β-catenin Tyrl42, β-catenin Tyr654, Non-p (Active) β-catenin Ser33/Ser37/Thr41, EGFRTyrl068, ErbB3 Tyrl289, ErbB2 Tyrl221/1222, Rb Ser780, p90RSK Thr359/363, mTOR Ser2448, p70S6K Thr421/Ser424, S6R Ser235/236, 4EBP1 Thr70, Cdc2 Tyrl5, c-Jun Thr73, Histone 3 SerlO, Cdk Thrl4/Serl5, and ERK1/2 were obtained from Cell Signalling Technology, Beverly, MA. The antibodies against ERK1/2 and a-tubulin were from Santa Cruz Biotechnology Inc. Santa Cruz, CA, USA. Anti-mouse CD31 antibody was from BioLegend, San Diego, CA, USA. Varlitinib was obtained from ASLAN Pharmaceuticals Ltd, Singapore.

Xenograft models

This study received ethics board approval at the SingHealth. All mice were maintained according to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health, USA.

HCC tumours have previously been used to create patient-derived xenograft models(Huynh et al., 2006) of which the 56 models were used to screen for the expression of EGFR, ErbB2, ErbB3, p-EGFR, p-ErbB2 and p-ErbB3 by Western blot analysis.

Three phospho-ErbB2/ErbB3-positive lines [HCC29-0909A, and HCC07-0409, HCCOl-0708 and two phospho-ErbB2/ErbB3-negative models (HCC21-0208 and HCC16-1014) as determined by Western blot analysis were used to establish tumours in male SCID mice (In Vivo, Singapore) aged 9 to 10 weeks.

For dose-response experiment, mice bearing the HCC29-0909A, HCC07-0409, and HCCOl- 0708 xenografts (8-10 mice per group) were orally given vehicle [3 parts Polyethylene glycol) average Mn 300 (PEG, Aldrich Cat#202371) and 7 parts 30% w/v Research Grade Captisol (Ligand Pharmaceuticals, San Diego, CA) or 3 doses of Varlitinib (25 or 50, 75 and 100 mg/kg BID) for indicated days. Treatment started when the tumours reached the size of approximately 120-150 mm ³ .

To investigate the antitumor effects of Varlitinib, mice bearing HCC21-0208 and HCC16- 1014 tumours were orally administered either vehicle (n=5) or 100 mg/kg BID Varlitinib (n=5) for indicated days. Treatment started when the tumours reached the size of approximately 120-150 mm ³ . Tumour growth was monitored and tumour volume was calculated as described (Huynh et al., 2006, 2010). At the end of the study, the mice were sacrificed with body and tumour weights recorded and the tumours harvested for later analyses.

Vessel perfusion study Each mouse bearing tumour xenografts received intravenously with 100 mg of Biotinylated Lycopersicon Esculentum (Tomato) Lectin (VectorLabs #B-1175) prepared in 100 μ] of 0.9% NaCl. The tumours were harvested 10 minutes after lectin perfusion, fixed in 10 % formalin buffer solution, embedded in paraffin. Five μιτι sections were prepared. After blocking endogenous peroxidase activity and nonspecific staining, the sections were incubated 1 hour at room temperature with Streptavidin Peroxidase (Lab Vision Corporation, Fremont, CA]. To visualize productive microvessels, immunohistochemistry was performed using the streptavidin-biotin peroxidase complex method, according to the manufacturer's instructions (Lab Vision, Fremont, Calif). For the quantification of mean microvessel density in sections, 10 random 0.159 mm ² fields at lOOx magnification were captured for each tumour.

To determine the extent of hypoxia in tumour tissues, mice bearing indicated tumours were treated with vehicle, or 100 mg/kg Varlitinib BID for indicated days. Mice were i.p. injected with pimonidazole hydrochloride (60 mg/kg, 2.5 μΐ/g of mouse body weight) 1 hour before tumours harvested. Hypoxic regions of tumour were identified by staining the sections with H poxyprobe plus Kit HP2 (Chemicon) as described by the manufacturer.

Western blot analysis

3-4 independent tumours from vehicle and drug-treated mice were homogenized separately in lysis buffer and 80 μg of proteins per sample were analysed by Western blot analysis as described (Huynh et al., 2006, 2010).

Immunohistochemistry

5-μηι sections were stained with CD31, p-Histone 3 SerlO and cleaved PARP antibodies to assess microvessel density, cell proliferation, and apoptosis respectively, as described previously (Huynh et al., 2006, 2010).

Transcriptome sequencing analysis, global gene expression analysis, and bioinformatic analysis

A total of 200 ng of QC-qualified total RNA was used for TruSeq mRNA library prep, followed by 150bp paired-end sequencing in HiSeq4000 platform by BGI HK, Ltd. Since the RNA samples would be contaminated with mouse cells due to the growth of PDXs in immunocompromised SCID mouse, an extra filtering step to remove mouse component was introduced. The raw sequencing reads were aligned to hgl9_mml 0 mixed reference by Burrows- Wheeler Aligner (BWA) and the read pairs were removed as long as any of the paired reads mapped to mmlO chromosome and/or rRNA sequence to filter mouse contamination and rRNA reads, respectively. The detected mouse read rates at genome and gene levels ranged from 0.68% to 11.2% and 0.21% to 5.12%, respectively. In average, 74 million to 90 million clean reads per sample were obtained. The BAM files were then uploaded to Partek Flow for further analysis. Aligned reads were quantify to Partek E/M annotation model in Partek Flow (Partek Inc. St Louis, MO, USA), followed by total count and add 0,0001 normalisation of gene counts and GSA differential expression detection from the comparisons of 50mg/kg varlitinib treatment vs control and l OOmg/kg varlitinib treatment vs control in both models using the cut-off threshold of >2 and -2 fold-change and FDR adjusted p-value < 0.05, followed by heatmap generation. The global gene expression analysis was done by quantile normalized data from Affymetrix GeneChip Human Genome U133 Plus 2.0 Array with high stringent cut-off threshold of >2 and -2 fold-change and FDR adjusted p-value < 0.0001. The identified gene lists were then compared using BioVenn online tool for the area-proportional Venn diagram analysis (Hulsen et al., 2008). The KEGG Pathway Enrichment analysis was carried out in Partek Genomics Suite (Partek Inc. St. Louis, MO, USA) with the cut-off threshold as FDR adjusted p-value <0.05. The differentially expressed gene lists were analysed through the use of Ingenuity Pathway Analysis (IPA) (Kramer et al., 2014) (QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuitypathw ay-analysis). TCGA data access and Oncoprint analysis ware carried out with the use of cBioPortal (Cerami et al., 2012; Gao et al., 2013).

Results

Varlitinib suppresses tumour growth through dose-dependent inhibition of ErbB family pathways in HCC PDXs

Previous clinical trials have shown that targeting one or two ErbB members concurrently had subtle clinical benefit. To have a meaningful clinical benefit, inhibition of all ErbB family simultaneously may be needed. Varlitinib is a best-in-class reversible pan-HER inhibitor with a good clinical safety profile.

To investigate the anti-tumour activity of Varlitinib, we first determine the protein expression and activation of EGFR, ErbB2, and ErbB3 in 56 HCC PDXs (Huynh et al., 2006) by Western blot analysis. As shown in Figure 1A and IB and Figure 7 A and 7B, 91% (51/56), 82% (46/56) and 93% (52/56) of the tested PDX expressed high levels of EGFR, ErbB2, and ErbB3, respectively. ErbB4 was undetectable in all PDX models tested. Low but detectable levels of phosphorylated EGFR at Tyrl068 were detected in 89% (50/56) of the PDX models. Phospho- ErbB2 at Tyrl221/1222 and p-ErbB3 at Tyrl289 were detected in 10.7% (6/56) and 37.5% (21/56) of PDX models tested, respectively. The data suggests that ErbB2 and ErbB3 were activated in small subset of HCC. Since HCCOl-0708, HCC07-0409, and HCC29-0909A PDX models expressed EGFR, ErbB2, and ErbB3 with the highest levels of activated ErbB2 and ErbB3 among all the tested PDXs, as indicated by the levels of phosphorylation, they were chosen for further Varlitinib testing (Figure 1A). HCC21-0208 was selected to serve as negative control because it has the very low level of ErbB2 and undetectable level of ErbB3 (Figure IB). Mice bearing HCCOl- 0708, HCC07-0409, and HCC29-0909A tumours were treated with three different doses of Varlitinib (25, 50, and lOOmg/kg BID). Figure 1C-1E showed that Varlitinib inhibited tumour growth in a dose- dependent manner. This correlated well with tumour weight at harvest. HCCOl- 0708 is a fast-growing PDX that tolerated Varlitinib at low dosing, but the tumour growth and weight of that were significantly suppressed at the highest tested dosing, suggesting that higher dosing of Varlitinib is able to overcome the drug tolerance (Figure 1C). Comparatively, HCC07- 0409 and HCC29-0909A were more sensitive to Varlitinib and showed better dose-dependent inhibition effect (Figure ID and IE). Statistically significant growth inhibition is also observed among all different dosing in these two models, meaning that these models are sensitive and respond to Varlitinib very well. The latter one was dramatically suppressed even at 50mg/kg BID, indicating that Varlitinib is highly potent on HCC29-0909A model, which has the highest level of p- ErbB3 among all PDXs (Figure IE). On the other hand, Varlitinib at the dose of 100 mg/kg BID had no significant anti-tumour activity in HCC21-0208, which expresses undetectable level of ErbB3. Similarly, the growth of HCC16-1014 was unaffected by Varlitinib at the dose of 75mg/kg BID. Correlative study reveals that the levels of phosphorylated [activated] ErbB2 and ErbB3 but not total ErbB are indicative of Varlitinib sensitivity [Figure 7C and 7D).

To study the effects of Varlitinib on the ErbB family and their downstream targets. Western blot was carried out The protein lysate was prepared at equal concentrations of the treated tumour samples collected on day 2-post treatment and/or the last day of the experiments for Western blot with the use of a panel of antibodies detecting the ErbB members and their downstream targets. In HCC29-0909A model, Varlitinib effectively inhibited p-EGFR and p-ErbB2, and p-ErbB3 on day 2 and 14 post-drug treatment (Figure 2). The suppression of phosphorylated and total ErbB3 became more significant on day 14. While p27 level was elevated, p-AKT, p-mTOR, p-p90RSK, p-S6R, p-Cdc2, and p-Rb were decreased dose-dependently. At the later time point, p- ERK1/2, p-p70S6K, p-4EBPl, CDC25C, and E2F1 were significantly decreased, while p21 was increased. These findings suggested the dose- and time-dependent inhibition of ErbB downstream targets in this model. Furthermore, in the highest dosing of Varlitinib-treated HCC07-0409 model (Figure 29), inhibition of p-EGFR, p-ErbB2, p-ErbB3, p-ERKl/2, and p-Rb were detected. Interestingly, the Varlitinib-mediated ErbB pathway blocking was different in the HCCO l-0708 model (Figure 8). There was mild inhibition found in p-EGFR, p-ErbB2, p-ErbB3, p-AKT, p-mTOR, p-p90RSK, and p-S6R on day 2 post-treatment (Figure 8A). However, significant suppression of p- Cdc2 and p-Rb as well as activation of p27 was found in the highest dosing of Varlitinib-treated HCCOl-0708 model. On day 12 post-treatment, marked decrease of p-ERKl/2 and p-p70S6K were observed (Figure 8B]. The data of Varlitinib-treated HCCOl-0708 model provided molecular insights to explain the tolerance ability of HCCOl-0708 at low dosing treatment Better treatment outcome was achieved only at the dose, where Varlitinib effectively inhibits its target activation.

Subsequently, immunohistochemical staining of formalin-fixed paraffin-embedded (FFPE] samples was carried out to determine the changes of proliferation and apoptosis in the Varlitinib- treated PDXs. It clearly demonstrates that Varlitinib effectively inhibited tumour cell proliferation, induction of apoptosis in HCC29-0909A dose-dependently as indicated by reduced p-Histone H3 (SerlO] staining as well as enhanced cleaved PARP staining (Figure 3 A and Figure 9]. Similar phenotypic changes were observed in other two Varlitinib-sensitive PDXs (HCCOl-0708 and HCC07-0409] as well, but not in the Varlitinib-resistant PDX, HCC21-0208. It demonstrated that Varlitinib effectively and molecularly suppressed cell cycle progression and induced apoptosis in the ErbB-dependent PDXs, as a result tumour shrinkage is observed (Figure 10).

Varlitinib promotes vessel normalisation and tumour perfusion in HCC PDX

More than a decent ago, vessel normalisation concept which recalculates the imbalance between pro- and anti-angiogenesis was introduced (Goel et al., 2012; Jain, 2001, Cerniglia et al., 2009; Izumi et al., 2002). To determine whether Varlitinib is able to normalize the blood vessels in HCC, we performed the IHC on vehicle- and Varlitinib-treated tumours using CD31 antibody. Surprisingly, the density of blood vessels was significantly increased following Varlitinib treatment as determined by CD31 immunostaining (Figure 3A and Figure 9). The blood vessel diameter was significantly reduced and more capillary-like blood vessels were found in the tumours treated with Varlitinib.

To determine whether the capillary-like blood vessels induced by Varlitinib were functional, the experiment of biotinylated tomato lectin perfusion was performed to label the vascular endothelium and to detect perfused vasculature structure, followed by pimonidazole HC1 infusion to measure the hypoxic microenvironment in HCC29-0909A PDX tumours as direct evidence of vessel normalisation. Figure 3B and Figure 11 clearly showed that very little or no lectin was detected in blood vessels of vehicle-treated tumours suggesting that most of the blood vessels in vehicle-treated tumours are not functional. Large regions of the tumour section were stained positively with hypoxyprobe suggesting the hypoxic regions. In contrast, majority of capillary-like blood vessels induced by Varlitinib were lectin immunostaining suggesting that they were well perfused and functional. In addition, no hypoxyprobe were detected across the large section of the tumour indicating the region was well oxygenated. These data suggest that inhibition of HER family members by Varlitinib results in blood vessel normalisation. These capillary-like blood vessels are well-perfused leading to reduction of hypoxia in the tumour microenvironment Transcriptome analysis reveals the molecular basis of Varlitinib-mediated tumour suppression and vessel normalisation

Although both HCCOl-0708 and HCC29-0909A PDX models are ErbB3-highly expressed and activated, their responses to Varlitinib are quite distinct that highest tested dosing of Varlitinib is needed to suppress the growth of HCCOl-0708, whereas lower dosing is sufficient to suppress that of HCC29-0909A (Figure 1C and IE]. Therefore, to further investigate the similarities and differences between two models and between Varlitinib treatments at different doses, total mRNA sequencing (RNA-Seq] was used to analyse the triplicate samples of vehicle-, 50mg/kg Varlitinib, and lOOmg/kg Varlitinib-treated PDX samples which were collected on Day 11 and Day 14 post- treatment for HCCOl-0708 and HCC29-0909A, respectively.

There were 3422 differentially expressed genes identified (1720 downregulated and 1702 upregulated] in lOOmg/kg Varlitinib- treatment and 2778 differentially expressed genes identified (1377 downregulated and 1401 upregulated) in 50mg/kg Varlitinib-treatment in HCC29-0909A (Figure 4A]. In total, 1122 and 1209 commonly downregulated and upregulated genes, respectively, were identified in Varlitinib-treated HCC29-0909A (Figure 4B]. The top 20 activated and repressed genes in HCC29-0909A are shown in Tables 1A and IB, respectively. According to pathway enrichment analysis, the top 6 significantly suppressed KEGG pathways are ribosome pathway, RNA transport, steroid biosynthesis, central carbon metabolism in cancer, and HIF1 signalling pathway (Table 1C]. The Ingenuity Pathway Analysis (IPA] demonstrated the suppression of EIF2, eIF4, p70S6K, and mTOR signalling with predicted HIFIA as the inhibited upstream regulator as well as activation of ILl-mediated inhibition of RXR function, FXR/RXR and LXR/RXR pathway with predicted HNF1A and PPARA activated upstream regulators (Table ID], showing that the repressed ErbB downstream pathways correlates with hepatic lipid differentiation. In the contrast, there are 2151 differentially expressed genes identified (1167 repressed and 984 activated] in lOOmg/kg Varlitinib-treatment and 1842 differentially expressed genes identified (931 repressed and 911 activated] in 50mg/kg Varlitinib-treatment in HCCOl- 0708 (Figure 4C].

The top 20 activated and repressed genes in high dosing treated HCCOl-0708 are shown in Tables 2A and 2B, respectively. Interestingly, the number of common differentially expressed genes reduced significantly. In total, only 265 and 192 commonly downregulated and upregulated genes, respectively, are identified in Varlitinib-treated HCCOl-0708 (Figure 4D]. Two of the significantly repressed KEGG pathways in Varlitinib-treated HCCOl-0708 PDXs are central carbon metabolism in cancer [p-value = 2.40 x 10 ³ ) and HIF-1 signalling pathway (p-value = 1.11 x 10 ² ). IPA analysis suggested that high dosing of Varlitinib suppressed Wnt/ -catenin pathway (p-value = 2.41 x 10-3) with predicted NUPR1, TGFB1, and Raf as the inhibited upstream regulators (p-value = 1.90 x 10-13, 3.10 x 10 ¹² , and 1.21 x 10 ¹¹ , respectively). The numbers of differentially expressed genes identified from different dosing in both models indicate that higher dosing of Varlitinib used, more genes will be affected. As the area-proportional Venn diagrams showed (Figure 4A and 4C), there are much more common genes identified in Varlitinib-treated HCC29-0909A than those in treated HCCOl-0708 (2331 genes vs 457 genes), suggesting that similar genes and pathways were modulated by different dosing of Varlitinib in the former model. However, different dosing of Varlitinib has distinct effects on the latter model. These phenotypes molecularly align with the dose-dependent Varlitinib-mediated growth suppression in PDXs (Figure 1C and IE).

Furthermore, to interrogate the Varlitinib inhibition gene signature from the treated PDX, three-way area-proportional Venn diagram analysis was used to analyse the gene lists generated from l OOmg/kg BID and 50mg/kg BID treatment in HCC29-0909A and l OOmg/kg BID treatment in HCCOl-0708 (Figure 4E]. The top 20 activated and repressed genes are shown in Table 3A, respectively. The gene list and heatmap with 433 genes (195 downregulated and 238 upregulated) was identified and generated, followed by IPA and KEGG pathway enrichment analysis. Top 2 KEGG pathways, central carbon metabolism (p-value = 1.46 x 10 ⁸ ) and HIFl-related pathways (p- value = 6.44 x 10 ⁷ ), overrepresented in both treated PDXs reveals the core pathways that are effectively targeted by Varlitinib in HCC (Table 3B). IPA analysis further supported that glycolysis I pathway (p-value = 1.28 x 10 ³ ) with HIF1A and PKM predicted as upstream targets (p-value = 2.50 x 10 ¹⁵ and 1.44 x 10 ⁷ , respectively) were suppressed, whereas LPS/ILl mediated inhibition of RXR function (p-value = 7.74 x 10 ⁷ ) and FXR/RXR activation (p-value = 1.30 x 10 ⁵ ) were elevated in the Varlitinib-treated PDXs (Table 3C).

Jain's group pointed out that imbalance of pro- and anti-angiogenic factors in the tumour microenvironment contributes to angiogenesis and tumour progression (Goel et al., 2012; Jain, 2005), As the strong evidence from Figure 3B proved the effective vessel normalisation was induced by Varlitinib, we further hypothesise that there is significant repression of pro-angiogenic factors and other factors in the tested PDXs, contributing to the normalisation. HIF1A expression is found to be significantly downregulated in the high dosing of Varlitinib-treated group in HCCOl- 0708 model (-2.05 fold-change, FDR-adjusted p-value = 3.05 x 10 ), which demonstrates the consistency of the cellular phenotype and gene expression level. We then analysed the angiogenesis-related genes in the treated PDXs. The data showed that Varlitinib significantly repressed gene expression of the key pro-angiogenetic factors, PDGFA (-299.46 fold-change, FDR- adjusted p-value = 6.12 x 10 ¹¹ ), EPO (-157.64 fold-change, FDR-adjusted p-value = 8.78 x 10 ⁹ ), VEGFB (- 15.68 fold-change, FDR-adjusted p-value = 2.31 x 10 ¹⁰ ), and PGF (-4.33 fold-change, FDR- adjusted p-value = 4.18 x 10 ⁸ ), and BMP2 (-2.99 fold-change, FDR-adjusted p-value = 4.48 x 10 ^s ), NRP2 (-2.23 fold-change, FDR-adjusted p-value = 0.030) in the high dosing of Varlitinib-treated HCC29-0909A model. For the Varlitinib-treated HCCOl-0708 model, pro-angiogenetic factors, TM4SF1 (-216.53 fold-change, FDR-adjusted p-value = 8.68 x 10 ¹¹ ), NRP2 (- 11.04 fold-change, FDR-adjusted p-value = 1.44 x lO ⁶ ), EPO (-5.19 fold-change, FDR-adjusted p-value = 3.9 x 10 ³ ), BMP2 (-4.34 fold-change, FDR-adjusted p-value = 2.15 x 10 ⁶ ], BMP2-target gene FST (-4.16 fold- change, FDR-adjusted p-value = 0.010] and PDGFC [-2.19 fold-change, FDR-adjusted p-value = 1.47 x 10" ⁵ ] were identified.

The gene expression analysis demonstrated here that Varlitinib could inhibit similar EPO- related pathway and distinct VEGF-dependent [PDGFA, VEGFB, PGF, and NRP2) and VEGF- independent [PDGFC and BMP2/FST) anti-angiogenic pathways in two different PDX models to facilitate the vessel normalisation (Andrae et al., 2008; Kertesz et al,, 2004; Krneta et al,, 2006; Li et al., 2010; Lin et al., 2014; Zuo et al., 2016]. Furthermore, 48.82-fold activation (FDR-adjusted p- value = 4.25 x 10 "] of HRG expression is found in Varlitinib-treated HCCOl-0708. HRG which encodes histidine-rich glycoprotein was found to suppress tumour growth by inducing vessel normalisation and macrophage polarisation in the immunocompetent tumour mouse model (Rolny et al., 2011]. Recently, HRG has been further proved to function as a tumour suppressor in HCC through inhibiting FGF to Erkl/2 pathway and diminishing FGFR activation (Zhang et al., 2015]. Therefore, HRG gene activation is accounted for another mean of vessel normalisation. Taken together, Varlitinib potently and effectively inhibits expression of anti-angiogenic factors as well as elevates expression of vessel normalisation factor, resulting in vessel normalisation and potentially improving drug delivery and efficacy.

Varlitinib enhances immune infiltration in the treated tumours

Recently, a study has suggested the mutual regulation of vessel normalisation and immune infiltration in multiple human cancers (Tian et al., 2017]. Therefore, we sought to understand if immune infiltration would be affected in the Varlitinib-treated PDXs, following vessel normalisation. The RNA-Seq reads that align to mouse genome reference, mmlO were used to analyse the stromal components in the PDX models. Myeloid cell-related markers were specifically analysed. Interestingly, much higher expression of Cd68, Itgax (Cdllc], Itgam (Cdllb], Adgrel (F4/80), Ly6cl (Ly6C], II4ra (IL-4Ra], Lgals3 (MAC2], Csfl r (Cdl l5], and major histocompatibility complex (MHC]-class II related genes, H2-Aa, H2- Abl, H2-DMa, H2-DMbl, H2-DMb2, H2-Ea-ps, and m-Ebl have been detected in the high dosing Varlitinib-treated HCCOl-0708 PDX (Table 11 A], suggesting the presence of higher infiltration of monocytes, tumour-associated macrophages (TAMs], and Cdllb+MHC-class 11+ dendritic cells [DCs] in the treated tumours. Undetectable Ly6G (Grl] suggests the absence of tumour-associated neutrophils [TANs] in tumours. TAMs could indicate a wide spectrum of activations towards anti-tumoural Ml or pro-tumoural M2 states (Murray et al., 2014]. Accordingly, a much higher M(LPS+IFNgamma]-associated gene expression, including elevated expression of Statl, Tnf, Nos2, Irf5, and Nfkbiz, has been detected in Varlitinib- treated PDX, revealing the infiltration of highly activated Ml macrophage (Table 11B], Separately, the high expression of Socs3, Tgfbl, Sbno2, Stat6, Ccl24, II4ra, and Id3 observed suggested that the presence of M2-like macrophage without defined activation status was found in the treated PDX (Table 11B]. Also, the undetectable expression 1110 and Argl suggested that the M2-like macrophages might not be highly activated. Taken together, significantly higher myeloid immune gene signature was found in the Varlitinib-treated HCC PDX, possibly facilitated by vessel normalisation.

Varlitinib effectively inhibits mutated β-catenin pathways and mediates membrane translocation of mutated β-catenin Due to the difference of Varlitinib efficacy in the tested PDXs, we have further investigated the gene signature that represents Varlitinib efficacy (Figure 12). Whole exome sequencing (WES] was further performed for all five PDXs to determine if these PDXs contain any mutation on β- catenin and ErbB family members (Table 9). Interestingly, the WES data demonstrated that all three Varlitinib-sensitive PDXs contain β-catenin T41A mutation, whereas ErbB family members exhibit different mutational patterns. Lower CTNNB1 expression but very high expression of β- catenin-interacting co-activators LEF1 and TCF4 as well as β-catenin-target genes, CCND1, MMP2, TBX3, EPHB2, and SPARCL1 were found in Varlitinib-sensitive PDXs (Figure 13A and Table 13A). In the contrast, the Varlitinib-resistant PDX, HCC21-0208, contains wild-type CTNNB1 but with much higher TGFBRl /2, IGF2/IGFR/IRS1, NOTCHl/JAGl, TEADl/2, and CYR61 expression and lower MST1 expression comparatively from the gene expression microarray data (Figure 13B and Table 13B). The data aligns with pathway analysis and suggests that Varlitinib-resistant PDX exhibited highly activated TGF-β, NOTCH, and Hippo pathways. Compared to the proposed molecular classes of HCC (Lachenmayer et al., 2012; Sia et al., 2017], the Varlitinib-resistant PDX, HCC21-0208 is related to Wnt/TGF-β class, whereas the Varlitinib-sensitive PDXs, HCCOl-0708, HCC07-0409, and HCC29-0909 are related to CTNNB1 class or more specifically CTNNB1 mutation class. Furthermore, the gene signature of Varlitinib-potency of the two RNA-sequenced PDXs was determined and shown in Figure 14.

Given that HCCO l-0708, HCC07-0409 and HCC29-0909A harbour T41A β-catenin mutation and the Wnt/^-catenin-related genes/pathways are inhibited according to WES and RNA-Seq analysis separately, we sought to further investigate if the mutated β-catenin and its related pathway members are suppressed at protein level and whether the localisation of mutated β- catenin is affected by the Varlitinib treatment Figure 5 and Figure 30B displayed marked inhibition of β-catenin and its related pathways in the lOOmg/kg BID Varlitinib-sensitive PDXs, HCC29-0909A, HCCO l-0708, and HCC07-0409 models, β-catenin upstream regulators, p-LRP6 and DVL3, as well as downstream targets, Axin2, survivin, c-Jun, c-Met, and N-cadherin were suppressed and E-cadherin expression was elevated by Varlitinib in the treated HCC29-0909A (Figure 5A]. c-Jun was reported to physically interact with β-catenin and TCF4 and to stabilise β- catenin/TCF4 interaction for the transcription of β-catenin- target genes and cancer development (Gan et al., 2008; Nateri et al., 2005). The inhibition of p-c-Jun by Varlitinib suggested the transcription repression of β-catenin. In addition, the β-catenin-transcriptional marks, ρ- β-catenin (Tyrl42 and Tyr654) (Brembeck et al., 2004; van Veelen et al., 2011), active β-catenin mark, non- ρ-β-Catenin (S33/S37/T41), and total β-catenin were repressed (Figure 5B). Phosphorylated- LRP6, active β-catenin mark, survivin, and p=-c-Jun were found to be suppressed in HCCOl-0708 model and reduced transcriptional mark and active mark of β-catenin, Axin2, survivin, DVL3, and c-Met were also observed in HCC07-0409 model (Figure 30B). Interestingly, degradation mark, p- β-Catenin (S33/S37/T41) and p-RanBP3 Ser58were reduced in Varlitinib-treated HCC29-0909A (Figure 30A). The data suggested that reduction of β-catenin protein expression is not caused by phosphorylation of S33/S37/T41-mediated β-TrCP-ubiquitination and β-catenin may be exported from nucleus through phosphorylation of β-catenin at Tyrl42 (Krejci et al., 2012) and RanBP3- dependent mechanism (Hendriksen et al., 2005; Spiegelman et al., 2000; Yoon et al., 2008). Immunohlstochemistry clearly demonstrated that β-catenin in vehicle-treated tumours was mainly located in the cytoplasm and nucleus (Figure 5 C), However, expression of β-catenin in Varlitinib- treated tumours was indeed located in the membrane, indicated by the diminished nuclear staining of β-catenin and the enhanced membrane staining. Taken together, these results strongly support that Varlitinib targets mutated β-catenin and its related pathways in ErbB-dependent tumours by inhibition of p-LRP6, ρ-β-catenin at Tyrl42, and p-RanBP3 Ser58, resulting in β- catenin membrane translocation and inhibiting downstream targets of β-catenin.

NRG1/ERBB3 pathway dependence and differentiation status correlate with Varlitinib treatment efficacy in PDXs

To reveal the differences in gene expression and signalling pathways between varlitinib- sensitive and varlitinib- resistant PDX models, gene expression microarray was performed on four treatment-naive PDXs (HCCOl-0708, HCC07-0409, HCC29-0909A, and HCC21-0208), followed by gene set and pathway enrichment analyses (Figure 12]. The differentially expressed genes were identified from the quantile normalised data of Affymetrix GeneChip Human Genome U133 Plus 2.0 Array with cut-off threshold of 2 fold-change and high stringent FDR adjusted p-value < 0.0001, followed by heatmap generation (Figure 12 A), In total, 2591 differentially expressed genes were detected, in which 1035 genes were highly expressed and 1556 genes were lowly expressed in the varlitinib-sensitive group. The gene lists with top 20 highly and lowly expressed genes are summarised in Table 4A and Table 4B, respectively. The most abundantly expressed gene in varlitinib-sensitive PDXs was NRG1 (fold-change > 500; Figure 12B]. It encodes neuregulin 1, which is well-documented as ErbB family receptor ligand(Hsieh et al., 2011). Moreover, higher EGFR and ERBB3 expression was also detected in the varlitinib-sensitive group (Figure 12C and Table 12). The data is in line with the Western blot data shown in Figure IB that HCC21 -0208 has undetectable ErbB3 protein, suggesting that ErbB3 pathway dependence is important for varlitinib treatment. Subsequently, Ingenuity Pathway Analysis (IPA) was used to analyse the differentially expressed gene sets. The analysis revealed that the varlitinib-sensitive PDXs have much higher glutathione metabolism and FXR/RXR and PXR/RXR activation with higher HNF4A and HNF1A activity predicted (Table 4C). In the contrast, the varlitinib-resistant PDX has statistically significantly higher pathway activities of angiogenesis, epithelial adherent junction signalling, epithelial-mesenchymal transition signalling, embryonic stem cell pluripotency, RhoA signalling, and Hippo signalling with higher CD 24 signalling predicted. Studies have demonstrated that HNF4A is a key liver differentiation driver and CD24 is a tumour-initiating cell marker in HCC (Enane et al., 2017; Lee et al., 2011). In summary, varlitinib sensitivity correlates with ErbB3 expression and differentiation status in in HCC PDXs.

Sternness/differentiation status correlates with Varlitinib potency in high-ErbB3 expressing and β-catenin mutated PDXs

According to the Western blot analysis shown in Figure 2 and Figure 8, there was lesser degree of ErbB member suppression detected in Varlitinib-treated HCC01-0708 when compared to the treated HCC29-0909A. However, the ErbB downstream targets such as p-Erkl/2, p-p70S6K, p- Cdc2, and p-Rb were suppressed. Also, functionally, Varlitinib is able to enhance vessel normalisation in HCCO l-0708 model. As HRG was heavily upregulated in the treated HCCOl-0708 and a study reported that HRG is able to inhibit FGFR pathway (Zhang et al., 2015), we then postulated that other receptor tyrosine kinases (RTKs) including FGFR, which act upstream of Erkl/2 and Akt, could be also inhibited by Varlitinib in the HCCOl-0708 model. Subsequently, we re-analysed the RNA-Seq data with the focuses of RTKs. We discovered that IGF2, FGFRl, and FGFR2 were downregulated in 25.63-fold, 24.49-fold, and 4.46-fold, respectively. Conversely, this observation was not found in the HCC29-0909A model. Therefore, this led us to explore the global gene expression difference between the two Varlitinib-sensitive PDX models. RNA-Seq data of the vehicle-treated PDXs were then compared to elucidate the differences of Varlitinib potency in gene expression level. In total, 1615 differentially expressed genes are identified, in which 1069 genes and 546 genes are highly and lowly expressed in HCCOl-0708, respectively, when compared to HCC29-0909A (Figure 14). Among the top highly expressed genes, HEYl and SOX9 were showed significantly higher expression in HCCOl-0708 (48.88 and 28.90-fold, respectively) than those in HCC29-0909A (Table 5A). Furthermore, the pathway enrichment analysis also demonstrated (Table 5 C) that HCCOl-0708 has much higher Hippo, Wnt, IGFR, and FGFR pathway activities, whereas HCC29-0909A retains stronger metabolic activities, revealing the impaired differentiation in the former model (Table 5D). The RNA-Seq comparison further showed that β-catenin, Notch, and Hippo targets, MYC, CTGF, CYR61, and YAP1 were highly expressed in HCCOl-0708, but were significantly inhibited by high dosing of Varlitinib (Table 5E and Table 5F).

Activated Wnt, Hippo, Notch, IGF, and FGFR are related the tumour-initiating cells and chemoresistance in HCC (Lau et al., 2016; Liu et al., 2016; Martinez-Quetglas et al., 2016; Villanueva et al., 2012). Accordingly, our study demonstrated that Varlitinib potency correlates with a spectrum of sternness/differentiation level in ErbB3-expressing and β-catenin mutated HCC PDXs.

Discussion

We have found that Varlitinib treatment potently inhibited tumour growth of ErbB- dependent HCC and effectively promoted vascular normalisation. Better efficacy was identified in HCC29-0909A, HCC07-0409, and HCCOl-0708 HCC PDX models, which display high levels of p- ErbB2 and p-ErbB3. Among these PDXs, HCC29-0909A possesses the highest p-ErbB3. HCC21- 0208 model has undetectable p-ErbB2 and p-ErbB3 demonstrated poor efficacy in the Varlitinib treatment. This observation suggests that Varlitinib sensitivity is determined by the levels of p- ErbB3 and/or pErbB2 and raises the possibility that p-ErbB3 and p-ErbB2 are more reliable biomarkers for patient selection than total ErbB2 or ErbB3. Among the group of Varlitinib- sensitive PDXs, higher Wnt, Hippo, Notch, IGF, and FGFR pathway signalling were observed in the HCCOl-0708 PDX, which required high dosing of Varlitinib to achieve better tumour growth inhibition. Pathway analyses revealed that Varlitinib suppressed tumour cell proliferation and promoted apoptosis by inhibiting phosphorylation of ErbBl-3, RAS/RAF/MEK/MAPK, p70S6K, S6 ribosomal, 4EBP1, Cdk-2, Cdc-2, and retinoblastoma pathways in dose- and time-dependent manners. Transcriptome and gene expression analyses have further revealed the Varlitinib- efficacy, -potency, and -inhibition gene signatures. Our findings also suggested that NRGl/ErbB3- dependant HCC tumours, displaying low sternness properties and high hepatic differentiation markers correlates with better efficacy and potency in Varlitinib-treatment The findings are summarised in the hypothetic model shown in Figure 6.

In the present study, we also found that while angiogenic tumour vasculature from vehicle- treated tumours are hyperdilated and distorted, the majority of vessels from Varlitinib-treated tumours were slim, elongated, and regularly shaped resemble to blood vessels in normal liver. Varlitinib-treated tumours are well vascularized with a rather normal and more functional vasculature as determined by lectin perfusion than distorted and sparse tumour vasculature of vehicle-treated tumours. The decrease in tumour hypoxia in Varlitinib-treated tumours suggests that the dense capillary-like network of vessels was well perfused. Varlitinib induced vessel normalisation possibly through inhibiting HIF/VEGF-depending and -independent angiogenesis. Our hypothesis is supported by early studies showing that poor oxygen, concomitant with HIF-loc induction, is associated with a more aggressive tumour phenotype, genetic ablation of HIF-loc in various tumours results in reduced tumour mass and increased apoptosis. HIF-1 activity stimulates neovascularization by enabling tumour and host cells to produce a variety of proangiogenic factors like VEGF-A, PDGF-B, FGF-2, and angiopoietins that stimulate new blood vessel formation within hypoxic areas [Calvani et al., 2006; Okuyama et al., 2006). Also, support to our observation that Varlitinib induces blood vessel normalization by previous study showing that anti-HER2 therapies (trastuzumab) promote normalisation in HER2-positive breast cancer (Goel et al., 2011).

T41A mutations have been identified from 2% to 18.8% in analysed HCC and the β-catenin mutations are found to be associated with low-stage of HCC (Austinat et al., 2008; Boyault et al., 2007; Cleary et al., 2013; Enane et al., 2017; Guichard et al., 2012; Hoshida et al., 2009; Kan et al., 2013; Legoix et al., 1999; Nault et al., 2013; Nhieu et al., 1999; Waisberg and Saba, 2015; Wong et al., 2001). Molecularly, upon phosphorylation at serine 45, followed by serines 37/33 and threonine 41, β-catenin is ubiquitinated by β-TrCP ubiquitin ligases, followed by proteasomal degradation. However, mutations at T41 is found to prevent GSK3 -mediated p-phosphorylation at serines 37 and 33, further avoiding β-TrCP recognition (Aberle et al., 1997; Liu et al., 2002; MacDonald et al., 2009; Orford et al., 1997). T41A mutation is further shown to be constitutively active mutant, which enhances nuclear localisation of TCF4, resulting in elevated expression of β- catenin-targets [Hsu et al., 2006). Since phosphorylation of β-catenin at tyrosine 654 and at tyrosine 142 is essential for binding to E-cadherin [Roura et al., 1999) and oc-catenin [Aberle et al., 1996; Piedra et al., 2003; Pokutta and Weis, 2000) respectively, inhibition of tyrosine phosphorylation at these sites by Varlitinib would facilitate the interaction of E-cadherin and cc- catenin with β-catenin. This could lead to the assembly of the E-cadherin- oc-catenin^-catenin complex at the plasma membrane and to decreased β-catenin-dependent transcription.

In HCC, there are several molecular classifications proposed, describing thatWnt ^-catenin is one of the main classes. The first FDA-approved targeted therapy in HCC, Sorafenib, is found to modulate Wnt _/ ^-catenin signalling in in vitro and in vivo CTNNB1 -class liver cancer models, demonstrated by reduced TCF/LEF luciferase-reporter activity and β-catenin expression [Boyault et al., 2007; Hoshida et al., 2009; Lachenmayer et al., 2012; Sia et al., 2017). Targeting Wnt/β- catenin has been comprehensively discussed [Pez et al., 2013; Vilchez et al., 2016). Wnt _/ ^-catenin pathway inhibitor ICGOOl, FH535, and other small inhibitors have been tested in HCC in different in vitro and in vivo models [Delgado et al., 2014; Gedaly et al., 2014; Handeli and Simon, 2008). Cheng's group and Gedaly's group further proved that the combination of sorafenib and ICGOOl as well as the combination of sorafenib and FH535 had better treatment outcome in their experimental models [Galuppo et al., 2014; Lin et al., 2016). However, studies thus far displayed that the promoted downregulation or degradation of wild-type β-catenin is the mean of β-catenin pathway inhibition in HCC and there is no small inhibitor available and proven to be mutant β- catenin inhibitor. Furthermore, according to the latest HCC TCGA data (Ally et al., 2017], 46% of HCC patients contain CTNNB1 genetic alteration, including missense mutations and amplification, while 26%-38% of patients carry ErbB family genetic alteration (n=360. Figure 15A). Figure 15B specifically demonstrated the ErbB family genetic alterations (23%-36%) in the subset of HCC patients with CTNNB 1 mutation [n=96]. It suggests that CTNNB1 mutation and ErbB family genetic alteration are independent but not mutual exclusive events in HCC patients. However, our study suggests that Varlitinib would show high efficacy and potency in the specific subset of HCC which depends on both ErbB pathway and mutated β-catenin. Molecularly, Varlitinib effectively targeted β-catenin with T41A mutation in ErbB-dependent HCC by inhibiting ρ-β-catenin Tyrl42 and p- RanBP3 Ser58 as well as downregulated p-LRP6 and DVL3. As a result, the mutated β-catenin is transported from nucleus and cytoplasm to membrane. Several clinical trials of Varlitinib in different cancer types, including cholangiocarcinoma and gastric cancer have been initiated and presented its safety profile with greatly improved toxicity profile compared to the irreversible pan-Her inhibitors such as Neratinib and Dacomitinib. Therefore, Varlitinib is a clinically available pan-HER and mutated β-catenin inhibitor, that can be used to target p-ErbB2 and p-ErbB3-highly expressing and CTNNB1 subset of HCC. The robust and tolerate anti-tumour activity of Varlitinib in HCC PDX models and its safety profile warrants the development of clinical trial with trial enrichment for Varlitinib in HCC.

Example 2 - Varlitinib induces apoptosis in HCC cell lines

An in vitro experiment was conducted to investigate whether varlitinib induced apoptosis in HCC cell lines.

HCC cells were grown in cell culture medium with 10% FBS and with varying concentrations of varlitinib. Apoptosis profiles were analysed at 24 hour and 48 hour timepoints, using Muse Annexin V & Dead Cell Assay kit and Muse Cell Analyser. Early apoptosis cells (identified as Annexin V-PE positive and Dead Cell Marker negative] were measured and plotted.

Figure 18 shows the results of the experiment As can be seen, varlitinib was able to induce early apoptosis in all of the HCC cell lines tested after 48 hours of incubation. Varlitinib was particularly effective in sorafenib resistant cells (Huh7-SorR] where almost 70% early apoptosis was observed when high dose of varlitinib was used, suggesting that varlitinib could be effective in patients who progress on sorafenib.

Figure 19 shows the apoptosis profile for the PLC/PRF/5 [PLC] cells after 48 hour culture in the presence of varlitinib. Note the increase in percentage of apoptotic cells correlates with increasing varlitinib concentration.

In summary, the data presented demonstrates that Varlitinib can induce apoptosis in HCC cell lines, including sorafenib resistant cell lines.

Example 3 - Varlitinib-mediated tumour growth inhibition and vascular normalisation in activated ErB2/3-dependent and mutated β-catenin hepatocellular carcinoma

In this study, we evaluated the biological effects and efficacy of Varlitinib in HCC cell lines and patient-derived xenograft (PDX] models, and studied the molecular mechanisms of Varlitinib- mediated tumour suppression and anti-angiogenic potential. We sought to identify gene signatures associated with efficacy of Varlitinib, and the responsive subclass of HCC - by transcriptome and gene expression analyses.

Results:

Upregulation of ERBB3 is frequent in HCC and Varlitinib effectively inhibits high ErbB3- expressing liver cancer cell lines

To evaluate the gene expression profile of the ErbB family in HCC, we analysed 10 publicly available gene expression datasets. Interestingly, all 10 datasets displayed a statistically significant upregulation of ERBB3 in the HCC tumours versus the experimental controls, whereas expression of other ErbB family members showed no significant change (Figure 20A). Next, we sought to understand the gene expression profile of the ErbB family in a panel of 27 liver cancer cell lines from the Cancer Cell Line Encyclopedia (CCLE) gene expression dataset. Using the mean expression of ERBB3, 27 liver cancer cell lines were grouped in ERBB3 ^Hi s ^h and ERBB3 ^Low accordingly (Figure 25A and Table 7). We then tested the protein expression of EGFR, ErbB2, and ErbB3 in selected liver cancer cell lines by Western blot (Figure 25B). The Western blot data showed that ERBB3 ^Hl s ^h cell lines (Hep3B, HepG2/C3A, Huh7, and PLC/PRF/5] had strong ErbB3 expression, whereas ERBB3 ^Low cell lines (SNU398, SNU387, SNU182, SNU423, and SNU475) had undetectable ErbB3 protein expression, with the exception of SNU449 cell line. The Hep3B Sorafenib-resistant cell line (Hep3B-SorR), which we established by step-wise exposure to increasing amounts of Sorafenib until the maximal dose that the cell lines could tolerate, showed higher phosphorylation status of EGFR, ErbB2, and ErbB3. Upregulated ErbB pathway activity could therefore be an acquired Sorafenib resistance mechanism.

We tested the cytotoxicity of Varlitinib in representative ERBB3 ^Hi s ^h and ERBB3 ^Low cell lines (Figures 20B and 25C]. Colony formation assay in Figure 20B showed that Varlitinib-mediated growth inhibition was more potent in ERBB3 ^m & cell lines in a dose-dependent manner. Furthermore, Varlitinib effectively inhibited ErbB receptors and their downstream signalling in the treated PLC/PRF/5 cell line (Figure 20C), suggesting that Varlitinib could effectively and potently inhibit ERBB3 ^Hi &> HCC.

We next determined the protein expression of EGFR, ErbB2, and ErbB3 in 56 previously established HCC PDXs (Huynh et al 2006) by Western blot analysis. As shown in Figure 26A, 91% (51/56), 82% (46/56), and 93% (52/56) of the tested PDX expressed high levels of EGFR, ErbB2, and ErbB3, respectively. ErbB4 was undetectable in all PDX models.

Varlitinib effectively suppresses in vivo tumour growth in high p-ErbB2/3 and NRG1 expressing HCC patient-derived xenografts (PDXs)

To investigate the in vivo anti-tumour activity of Varlitinib, we then randomly selected 17 PDXs, with a range of expression of the ErbB family, to be used for Varlitinib treatment The mice bearing PDX tumours were treated with Varlitinib (lOOmg/kg BID). Additional dosing of 25 and 50mg/kg BID were adopted to treat three PDXs, HCCO l-0708, HCC07-0409, and HCC29-0909A (Figures 21A and 26B). Surprisingly, Varlitinib showed effective tumour growth inhibition in only three PDXs (HCCOl-0708, HCC07-0409, and HCC29-0909A), at the dose of lOOmg/kg BID. HCC07- 0409 and HCC29-0909A were more sensitive to Varlitinib and showed superior dose-dependent inhibition (Figure 21A). Statistically significant growth inhibition was also observed among all three doses in the two models (p-122 value <0.0001). The latter model exhibited dramatic tumour suppression even when treated with Varlitinib at 50mg/kg BID, signifying that Varlitinib is highly potent in the HCC29-0909A model. HCCOl-0708 was a fast-growing PDX mouse model that tolerated Varlitinib at low dosing. Tumour growth was significantly suppressed at the highest dose, suggesting that a higher dosing of Varlitinib may overcome intrinsic drug resistance. Conversely, Varlitinib at the dose of 100 mg/kg BID had no significant anti-tumour activity in the rest of the tested PDX models (Figures 21A and 26B].

The PDXs were categorised into responders and non-responders according to their treatment responses. Three responders and two non-responders were reassessed by Western blot (Figure 21B). As shown in Figure 21B, the three responders showed much higher p-ErbB2/3 and ErbB2/3 compared to the two non-responders. The expression of p-EGFR, total EGFR, p-Erkl/2, and p-Akt were similar in all tested PDXs, suggesting that responders are highly dependent on activated ErbB2/3 signalling. As Erkl/2 and Akt coud be activated by ErbB receptors, other growth receptors (GRs) and receptor tyrosine kinases (RTKs], simultaneously (Turke et al), our data suggests that the high expression of p-Erkl/2 and p-Akt in non-responders could be driven by other GRs or RTKs.

We further sought to understand the molecular differences between Varlitinib responders and non-responders. Three responders (HCCOl-0708, HCC07-0409, and HCC29-0909A) and 9 non- responders (HCCOl-0909, HCC06-0606, HCC13-0109, HCC21-0208, HCC25-0705A, HCC26-0808A, HCC26-0808B, HCC29-1104, and HCC30-0805B] were analysed by gene expression microarray, followed by gene set enrichment analysis. Interestingly, all 3 responders grouped together based on the principal component analysis (Figure 26C). 46 differentially expressed probes were identified by analysis between responders and non-responders with the threshold of fold- changed and FDR adjusted p-value=0.0001. The heatmap was generated accordingly (Table 8 and Figure 26D). NRGl ranked the top differentially expressed gene (Figures 21C and Table 8), NRGl is recognised as an ErbB3 ligand. The binding of NRGl to ErbB3 facilitates the heterodimersation of ErbB3 and ErbB2 to activate the downstream signalling pathways. The gene expression dataset also validated that responders possessed statistically higher expression of EGFR, ERBB2, and ERBB3 (Figure 21C). As illustrated in Figure 2 ID, poorer prognosis was correlated to high ERBB2 and ERBB3 expression in HCC tumours in the Singapore HCC dataset, using the cutoff finder for survival analysis. High NRGl expression also reflected a similar trend (insignificant p-value).

Responders belong to G6 subgroup and CTNNB1 subclass of HCC regardless of Hepatitis B status

Gene set enrichment analysis (GSEA) was performed to better understand the difference between responders and non-responders against all gene sets in the Molecular Signatures Database (MSigDB) v6.2 (Table 6 and Figure 22A). The GSEA identified that the gene sets which were related to HCC CTNNB1 subclass, HCC G6 subclass, lower interferon signalling, Wnt pathway, Myc signalling, and lower YAP1 signalling, were highly enriched in responders. In contrast, gene sets of CTNNB1 phosphorylation cascade (beta-catenin degradation), Notch signalling, and TGF- beta signalling pathways were highly enriched in non-responders (Table 6 and Figure 22 A}. As described by Boyault and co-workers (Boyault et al], G6 subgroup had much higher CTNNB1 pathway activities. This is similar to the CTNNB1 subclass described by Llovet's group (Lachenmayer et al, Chiang et al). We then focused on several G6- and CTNNB1 subclass-specific genes in responders and non-responders (Figure 22B]. Responders had significantly higher expression of FGF13, AXIN2, LEF1, EPHB2, GLUL, LAMA3, TBX3, SPARCL1, and LGR5. Conversely, non-responders had higher expression of TGFRB1, TGFRB2, TEAD1, IGF2, IGF1R, IRS2, NOTCH1, NOTCH2, and JAG1 (Figure 27]. In addition, expression of ERBB3 positively correlated with LEF1 and AXIN2 and negatively correlated with IGF1R (Figure 22C]. These suggest that non-responders could potentially belong to G1-G4 subgroups (Boyault et al]. Taken together, the gene expression analysis of responders and non-responders suggests that responders possess high ERBB2/3 and NRG1 expression, correlating to the G6 subgroup and CTNNB1 subclass of HCC. This correlation was not related to viral hepatitis status. Additionally, the mutations in EGFR, ERBB2, and CTNNB1 were mainly identified in responders (Table 9].

Transcriptome analysis reveals the molecular basis of Varlitinib-mediated tumour suppression

Although HCCOl-0708 and HCC29-0909A PDX models highly expressed activated ErbB3, they had distinct, differentiated responses to Variitinib. Maximal dosing (lOOmg/kg BID] of Variitinib was needed to suppress HCCOl-0708 tumour growth, whereas a lower dose (50mg/kg BID] was sufficient to suppress HCC29-0909A tumour growth (Figure 21A]. To further investigate the similarities and differences between sample variation and dose-dependent Variitinib treatments, total mRNA sequencing (RNA-Seq] was used to analyse samples of vehicle-, 50mg/kg BID Variitinib-, and lOOmg/kg BID Var tinib-treated PDX samples. The samples were collected on Day 11 and Day 14 post-treatment for HCCOl-0708 and HCC29-0909A, respectively (Figures 23 and 28A-B]. The samples of vehicle- and 75mg/kg BID Varlitinib-treated non-responder HCC16- 1014 were used as negative control of the treatment for RNA-Seq analysis (Table 10].

There were 3728 differentially expressed genes identified (1904 upregulated and 1824 downregulated] in lOOmg/kg Varlitinib-treatment and 3023 differentially expressed genes identified (1522 upregulated and 1501 downregulated] in 50mg/kg Varlitinib-treatment in HCC29-0909A (Figure 28A]. When comparing between the two treatments in Varlitinib-treated HCC29-0909A, there were 1337 and 1243 commonly upregulated and downregulated genes, respectively. In contrast, there were 2474 differentially expressed genes identified (836 activated and 1638 repressed] in l OOmg/kg Varlitinib-treatment and 1947 differentially expressed genes identified (1024 activated and 923 repressed] in 50mg/kg Varlitinib-treatment in HCCOl-0708 (Figure 28B]. Also, there were very few differentially expressed genes detected in the treated non- responder HCC16-1014 (Table 10]. The numbers of differentially expressed genes identified from the different dosing in both HCC29-0909A and HCCO l-0708 models indicated that the higher the dose of Variitinib used, the higher the number of changes in gene expression. The area- proportional Venn diagrams showed (Figures 28A and 28B] that there were more common genes identified in the Varlitinib-treated HCC29-0909A than those in the treated HCCOl-0708 (2580 genes vs 496 genes], suggesting that similar genes and pathways were modulated by different dosing of Variitinib in the farmer model. However, different dosing of Variitinib had distinct effects on the latter model. These phenotypes molecularly aligned with the dose-dependent Varlitinib- mediated growth suppression in PDXs (Figure 21 A].

The top 20 commonly activated and repressed genes in HCC29-0909A are shown in Tables 1A and IB, respectively. The top 6 significantly suppressed KEGG pathways were the ribosome pathway, RNA transport, steroid biosynthesis, central carbon metabolism in cancer, and HIF1 signalling pathway (Table 1C). The Ingenuity Pathway Analysis (IPA) demonstrated the suppression of EIF2, eIF4, p70S6K, and mTOR signalling with predicted HIF1A as the inhibited upstream regulator, as well as activation of ILl-mediated inhibition of RXR function, FXR/RXR and LXR/RXR pathway with predicted HNF1A and PPARA activated upstream regulators (Table ID], showing that the repressed ErbB downstream pathways correlate with hepatic lipid differentiation.

The top twenty commonly activated and repressed genes in the Varlitinib-treated HCC01- 0708 are shown in Tables 2A and 2B, respectively. In the Varlitinib-treated HCCOl-0708 model, there were 270 and 226 commonly upregulated and downregulated genes, respectively (Figure 28B). Interestingly, there were fewer number of common differentially expressed genes in HCCOl- 0708 than in HCC29-0909A. Two of the significantly repressed KEGG pathways in Varlitinib- treated HCCOl-0708 PDXs were involved in the central carbon metabolism in cancer (p-value = 2.40 x 10-3] and the HIF-1 signalling pathway (p-value = 1.11 x 10-2). IPA analysis suggested that high dosing of Varlitinib could suppress the Wnt/p-catenin pathway (p-value = 2.41 x 10-3] with predicted NUPR1, TGFB1, and Raf as the inhibited upstream regulators (p-value = 1.90 x 10- 13, 3.10 x 231 10-12, and 1.21 x 10-11, respectively].

To examine the Varlitinib inhibition gene signature from the treated PDXs, differentially expressed genes were compared between l OOmg/kg BID and 50mg/kg BID treatment in HCC29- 0909A, and lOOmg/kg BID treatment in HCCOl-0708 (Figure 23A]. The overlapped 546 genes (259 upregulated and 287 downregulated) were investigated using IPA and KEGG pathway enrichment analysis. The top 20 activated and repressed genes are shown in Table 1A. The top 2 KEGG pathways include the central carbon metabolism (p-value = 1.46 x 10-8) and the HIF1- related pathways (p-value = 6.44 x 10-7). The two overrepresented pathways reveal the core mechanisms that are effectively targeted by Varlitinib in HCC (Table 4A). IPA analysis demonstrated a suppressed glycolysis pathway (p-value = 1.28 x 10-3) with HIF1A (p-value = 2.50 x 10- 15) and PKM (p-value= 1.44 x 10-7) predicted as upstream targets. LPS/IL1 mediated inhibition of RXR function (p-value = 7.74 x 10-7) and FXR/RXR activation (p-value = 1.30 x 10-5) were elevated in the Varlitinib-treated PDXs (Table 3C).

Varlitinib inhibits glycolysis, HIF1A pathway, and angiogenesis in the treated PDXs

We then specifically analysed glycolysis and HIFlA-related pathways that were highly enriched and significantly inhibited in the treated PDXs. Figure 23B demonstrated that Varlitinib treatment was able to suppress selected glycolytic genes, as well as hypoxia- and angiogenesis- related genes. HIF1A was predicted as the key upstream regulator of Varlitinib inhibition (Figure 23C). The expression of HIF1A was significantly inhibited in both the treated HCC29-0909A and HCCOl-0708 models (Figure 23D). Figures 28C-E further revealed that Varlitinib could inhibit the EPO pathway, VEGF-dependent [PDGFA, VEGFB, PGF, and NRP2) and VEGF-independent [ANG, PDGFC and BMP2) pro- angiogenic pathways to facilitate vessel normalisation (Krneta et al, Kertesz et al and Andrae et al). Expression of LGR5 (β-catenin upstream regulator), YAPI, IRX3, and MYC were downregulated in both treated HCC29-0909A and HCCO l-0708 models. Conversely, expression of CDH1 was enhanced (Figures 23E and 28F). LEF1 was reduced in the treated HCC29- 0909A model (Figure 28G). On the contrary, higher expression of YAPl-related genes as well as repression of these genes were observed in the treated HCCOl-0708 [Figure 28H). By comparing HCC29-0909A and HCCOl-0708, we have further identified that stem cell related pathways are enriched in the HCCOl-0708, which displayed instrinsic resistance to Varlitinib when treated at lower dosing (Figure 281). Given that both HCC29-0909A and HCCOl-0708 contain mutated CTNNB1 (Table 9], it should also be noted that the expression of CTNNB1 in HCCOl-0708 was higher than in HCC29-0909A (Figure 28J), suggesting that CTNNB1 is a high-expressing mutant in HCCOl-0708. Heatmap analysis clearly indicated that HCCOl-0708 had significantly more embryonic stem cell core genes and Hippo/YAPl-related genes expressed than HCC29-0909A and the gene expression could be effectively repressed by high Varlitinib treatment. Comparatively, the non-responder HCC16-1014 did not show any inhibition in these pathways (Figures 28K, 28L and 28M]. As a result, these analyses suggest that higher dosing of Varlitinib is required for optimal effect against high ErbB2/3-expressing HCC with higher instrinsic sternness-related gene expression.

Time- and dose-dependent Varlitinib treatment inhibits ErbB receptor pathways and promotes vessel normalisation and tumour perfusion in HCC PDX

The effects of Varlitinib on the ErbB family and their downstream targets were studied using Western blot (Figures 2A, 2B, and 29]. Protein levels in PDXs were investigated, using a panel of antibodies detecting the ErbB members and their downstream targets. Protein lysates were prepared at equal concentrations from the treated tumour samples collected on day 2 post treatment and/or the last day of the experiments. In the HCC29-0909A model, Varlitinib effectively inhibited p-EGFR, p-ErbB2, and p-ErbB3 on days 2 and 14 post-drug treatment (Figures 2A and 2B). The suppression of phosphorylated and total ErbB3 became more prominent on day 14 post treatment (Figure 2B]. While p27 level was elevated, p-AKT, p-mTOR, p-p90RSK, p-S6R, p-Cdc2, and p-Rb levels were decreased in a dose-dependent manner. At a later time point, p-ERKl/2, p- p70S6K, p-4EBPl, CDC25C, and E2F1 expression were significantly decreased, while p21 expression was increased. These findings demonstrated dose- and time-dependent inhibition of ErbB downstream targets in this model. Furthermore, at the highest dose of Varlitinib-treated HCC07-0409 model (Figure 29), expression of p-EGFR, p-ErbB2, p-ErbB3, p-ERKl/2, and p-Rb were all inhibited.

Immunohistochemistry staining of formalin-fixed paraffin-embedded (FFPE) samples was carried out to determine changes in proliferation, apoptosis, and angiogenesis in the Varlitinib- treated PDXs (Figures 3A and 3B, 9, and 10]. The analysis demonstrated that dose-dependent Varlitinib effectively inhibited tumour cell proliferation and induced apoptosis in HCC29-0909A. This was indicated by reduced p-Histone H3 (SerlO] staining and enhanced cleaved PARP staining (Figures 3A and 3B and 9). Similar phenotypic changes were observed in the other two Varlitinib responder PDXs (HCCOl-0708 and HCC07-0409), but not in the Varlitinib non-responder PDX, HCC21-0208 (Figure 10). This demonstrated that Varlitinib effectively suppressed cell cycle progression and induced apoptosis in the ErbB-dependent PDXs, which corresponded to tumour shrinkage. In addition, the density of blood vessels was significantly increased following Varlitinib treatment, as demonstrated by positive CD31 immunostaining (Figure 3A and 3B). There was a significant reduction in the diameter of the blood vessels and more capillary-like blood vessels were found in Varlitinib responder PDXs. To determine whether Varlitinib-induced capillary-like blood vessels were functional, biotinylated tomato lectin was injected intravenously into vehicle- and Varlitinib-treated tumour-bearing mice for labelling of the vascular endothelium. This allowed the detection of perfused vasculature structure. Additionally, pimonidazole HC1 infusion was used to measure the hypoxic microenvironment in HCC29-0909A PDX. These provided direct evidence of vessel normalisation as a treatment effect Figure 3A and 3B clearly depicts that little or no lectin was detected in blood vessels of vehicle-treated PDXs, suggesting that most of the blood vessels in vehicle-treated PDXs were non-functional. Large regions of the tumour section stained positively with hypoxyprobe, indicating regions of hypoxia. In contrast, the majority of capillary- like blood vessels, in the HCC PDX exposed to Varlitinib treatment, stained positive for biotinylated lectin, suggesting that these blood vessels were well-perfused and functional. Furthermore, hypoxyprobe staining was negative across the large section of the Varlitinib-treated responder PDXs, showing that the region was well-oxygenated. The data suggests that the inhibition of ErbB family members by Varlitinib results in blood vessel normalisation. This was supported by the observation of well- perfused capillary-like blood vessels that reversed hypoxia in the tumour microenvironment. This IHC analysis was concordant to the gene set enrichment analysis findings that angiogenesis- and hypoxia-related gene sets were highly enriched in the Varlitinib-treated PDXs. (Figures 23B-D] Varlitinib effectively inhibits mutated β-catenin pathways and mediates membrane translocation of mutated β- catenin

According to the RNA-Seq analysis [Figures 23E and 28G], β-catenin- related genes, such as LGR5 and LEF1, were inhibited by Varlitinib. We then sought to further investigate if the β-catenin and its related pathway members would be suppressed at the protein level, and whether the localisation of mutated β-catenin was affected by the Varlitinib treatment

Figures 24A, 24B, 30A and 30B displayed marked inhibition of β-catenin and its related pathways in the Varlitinib responder PDXs, HCC29-0909A, HCCO l-0708, and HCC07-0409 models. β-catenin upstream regulators, p-LRP6 and DVL3, as well as downstream targets, Axin2, survivin, c-Jun, c-Met, and N-cadherin, were suppressed. In contrast, E-cadherin expression was elevated by Varlitinib in the treated PDXs (Figure 24A). c-[un was reported to physically interact with β- catenin and TCF4 to stabilise -cate n/lC¥ interaction for the transcription of β-catenin target genes and cancer development (Nateri et al). The inhibition of p-c-Jun by Varlitinib suggested the transcriptional repression of β-catenin. In addition, factors associated with active β-catenin export, ρ-β-catenin at Tyrl42 (van Veelen et al) and p-RanBP3 at Ser58 (Hendriksen et al), active β- catenin indicator (ηοη-ρ-β- Catenin at S33/S37/T41), β-catenin and E-Cadherin interrupted signal (ρ-β-catenin at Tyr654), and total β-catenin were repressed in the treated HCC29-0909A (Figures 24A and 30A). Interestingly, degradation indicator, ρ- β-Catenin at S33/S37/T41 was also reduced, even as there is a missense mutation at β-Catenin, i.e. Threonine to Alanine (T41A, Figure 24A and Table 9]. Phosphorylated-LRP6, active β-catenin, Survivin, and p-c-Jun (Ser73) were found to be suppressed in the HCCOl-0708 model (Figure 24B). The reduction of transcriptional and active sites of β-catenin, Axin2, Survivin, DVL3, and c-Met were also observed in the HCC07-0409 model (Figure 30B). The data suggests that reduction of β-catenin phosphorylation at Tyrl42 (Krejci et al) and the RanBP3-dependent mechanism are responsible for its nuclear export (Yoon et al). Also, suppression of β-catenin phosphorylation at Tyr654 might facilitate the binding of β-catenin to E- Cadherin. Immunohistochemistry clearly demonstrated that β-catenin in vehicle-treated tumours was mainly located in the cytoplasm and nucleus [Figure 5C]. However, expression of β-catenin in Varlitinib- treated PDXs localised to the membrane, indicated by the diminished nuclear staining of β-catenin and the enhanced membrane staining. The combined results strongly support our hypothesis that Varlitinib targets mutated β-catenin and its related pathways in ErbB-dependent tumours - by inhibiting β-catenin upstream regulator LGR5 and p-LRP6, ρ-β-catenin at Tyrl42 and Tyr654, nuclear export regulator p-RanBP3 at Ser58, as well as upregulating E-Cadherin expression. These, in turn, result in β-catenin membrane translocation, thereby inhibiting downstream targets of β-catenin. T41A β-catenin mutation is thought to be constituitively active. However, our study demonstrates that its nuclear localisation and β-catenin-driven genes could be inhibited by Varlitinib in high p-ErbB2/3 expressing HCC.

Discussion

We have demonstrated that among the ErbB receptor family, ERBB3 is upregulated in a subset of HCC. We established that Varlitinib, a small molecule pan-ErbB inhibitor, effectively inhibited high £7?B£.3-expressing HCC in vitro. Additionally, in vivo efficacy was demonstrated in the responder PDXs, HCC29-0909A, HCC07-0409, and HCCOl-0708 in immunodeficient NOD/SCID mice, which displayed high levels of p-ErbB2/3, total ErbB2/3, and NRG1 gene expression. The non-responders, HCC21-0208 and HCC16-1014 models expressed lower p-ErbB2 and p-ErbB3 levels, and were found to have poor efficacy to Varlitinib treatment. This observation suggests that Varlitinib sensitivity is determined by the quantitative expression of p-ErbB3 and/or pErbB2 and raises the possibility that the presence of p-ErbB2/3 could potentially be specific for patient treatment stratification. Furthermore, expression of ERBB2 and ERBB3 are known to have prognostic value, whereas that of NRG1 has the same trend in our Singapore HCC dataset (Figure 21D).

Gene expression analysis demonstrated that the three responders, with established higher ERBB2/3 expression, belong to both the G6 subgroup and CTNNB1 subclass in the non- proliferation class of HCC, whereas the 14 tested non-responder PDXs are related to the HCC subgroups with ΤΰΡβ and IGF/IGF1R pathways and S1/S2 tumours in the proliferation class of HCC [Llovet et al 2017,Boyault et al 2007,Lachenmayer et al ). It indicates the strong correlation of the ErbB family and β-catenin pathways in the specific subclass of HCC PDXs, that could be effectively inhibited by Varlitinib. It also reveals the diverse oncogenic pathway dependence in HCC and that alternative targeted treatments such as the ΤΟΡ , IGF1R, mTOR, and FGFR inhibitors are likely needed to treat Varlitinib non-responders. Among the group of high ErbB2/3 -expressing responder PDXs, Varlitinib was especially potent against HCC29-0909A. It may be due to the low activity of mutated β-catenin in the model, as indicated by the IPA pathway enrichment analysis (Figure 281). Conversely, a higher dose is required in the HCCOl-0708 model, which has higher mutated CTNNB1 expression, YAP1 signalling activity, and sternness, β-catenin-driven cancers require YAP1 for tumour progression (Rosenbluh et al]. YAP1 and its related sternness properties are also associated with drug resistance in multiple cancers (Zanconato et al). Lin and colleagues previously identified through genetic screen that YAP1 is the key resistance driver of RAF- and MEK-targeted therapies (Lin L et al). However, our pre-clinical in vivo data showed that the intrinsic resistance, possibly contributed by mutated β-catenin and YAP1, in the HCCOl-0708 PDX model could be overcome by a higher dose of Varlitinib, further supported by the significant suppression of β-catenin/YAPl-driven genes, SOX9, HEY1, CTGF, and CYR61 (Figure 28H).

Pathway analyses further revealed that Varlitinib suppressed tumour cell proliferation and promoted apoptosis in the responder PDXs by inhibiting phosphorylation of EGFR, ErbB2/3, MEK/MAPK, AKT/mTOR, p70S6K, S6R, 4EBP1, Cdk-2, Cdc-2, and retinoblastoma pathways in a dose- and time- dependent manner. Subsequently, hypoxia reduction, vessel normalisation, β- catenin membrane translocation, and β-catenin pathway inhibition were induced in the Varlitinib- treated PDXs. Taken together, our findings suggest that NRGl/ErbB2/3-dependent and β-catenin- mutated HCC tumours are most vulnerable to Varlitinib. These findings are summarised in the hypothetical model shown in Figure 31.

In this study, we found that while tumour vasculature from vehicle-treated tumours were hyper-dilated and distorted, the majority of blood vessels from Varlitinib-treated tumours were instead slim, elongated, and regularly-shaped, bearing resemblance to physiologically healthy blood vessels in a normal liver. The reduction in tumour hypoxia likely resulted from better perfusion of dense capillary-like blood vessels in the tumour microenvironment following Varlitinib exposure. Varlitinib possibly induced vessel normalisation through inhibiting HIF/VEGF- independent and HIF/VEGF-dependent angiogenesis. Our data also shows direct evidence that pan-ErbB inhibitor is able to repress HIF1A expression in a dose-dependent manner. Previous studies have reported that poor oxygen supply to the tumour bed, concomitant with HIF-loc induction, is associated with a more aggressive tumour phenotype. Genetic ablation of HIF-loc in various tumours results in reduced tumour mass and increased apoptosis. HIF-1 activity stimulates neovascularisation by enabling tumour and host cells to produce a variety of proangiogenic factors like VEGF-A, PDGF-B, FGF-2, and angiopoietins that stimulate new blood vessel formation within hypoxic areas (Okuyama et al, Calvani et al]. A previous study in other cancer models showed that Erlotinib and Gefitinib could inhibit pro-angiogenic factor VEGF expression by HIFlA-dependent and independent mechanisms (Pore et al]. Another study showed that anti-ErbB2 therapies (Trastuzumab) promote normalisation in ErbB2-positive breast cancer by inhibiting HIF-loc and pro-angiogenic factors (Goel et al]. Recently, the synergistic anti-tumour role of anti-angiogenesis in reprogramming the immune and metabolic tumour microenvironment is clinically proven in the July 2018 Federal Drug Agency (FDA] granting of breakthrough therapy designation for the combination of atezolizumab (anti-PDLl antibody] and bevacizumab (anti- angiogenesis inhibitor] in the first and second line treatment of advanced HCC following highly compelling Phase lb clinical trial results. A phase 3 trial using the same combination as first line treatment for advanced HCC patients has been started (ClinicalTrials.gov Identifier: NCT03434379). In our study, we have clearly shown that Varlitinib can induce vessel normalisation in HCC. Thus, it would be rational to combine Varlitinib and an immune checkpoint inhibitor in HCC, firstly tested in humanised HCC model (Zhao et al] and/or immunocompetent HCC model, then in human studies.

β-catenin mutations are found to be associated with the earlier stages of HCC and the T41A mutation has been identified in up to 18.8% of analysed HCC (Guichard et al]. Upon phosphorylation at serine 45, followed by serines 33/37 and threonine 41, wild-type β-catenin is ubiquitinated by β-TrCP ubiquitin ligases, destined for proteasomal degradation. However, mutations at T41 prevent GSK3 -mediated phosphorylation at serines 33/37, avoiding β-TrCP recognition [Liu et al). T41A mutation is also found to be a constitutively activated mutant, which enhances nuclear localisation of TCF4, resulting in elevated expression of β-catenin-targets (Hsu et al H-T et al). Since phosphorylation of β-catenin at tyrosines 142 and 654 is essential for binding to E-cadherin (Roura et al) and oc-catenin (Pokutta et al), inhibition of tyrosine phosphorylation at these sites by Varlitinib would facilitate the interaction of E-cadherin and a-catenin with β-catenin. This could lead to the assembly of the E-cadherin-<x-catenin^-catenin complex at the plasma membrane, and the decrease of β-catenin-dependent transcription. Our findings are consistent with early studies in other cancer models that inactivation of ErbB2 reduces phosphorylation of β- catenin at tyrosine 654, which faciliates β-catenin-E-cadherin interaction (Shibata et al). It was previously reported that EGFR is the direct target of Wnt/^-catenin in liver (Tan et al). Still, we are the first group to show that pan-ErbB inhibition effectively represses mutated β-catenin in in vivo models, linking ErbB receptor activities to mutated β-catenin localisation and functions in HCC. Further experiments will be needed to further confirm the proposed molecular mechanisms. Also, the effect of Varlitinib on other β-catenin mutations would need to be mechanistically explored.

The Wnt/ -catenin has been described as one of the main functional classes of molecular classification in HCC. Sorafenib - a BRAF, C-RAF and VEGF-R inhibitor - can modulate Wnt/β- catenin signalling in in vitro and in vivo CTNNB1 -class liver cancer models. This was demonstrated by showing reduced TCF/LEF luciferase-reporter activity and β-catenin expression following Sorafenib treatment (Boyault et al, Lachenmayer et al). There has been considerable exploration of targeting Wnty^-catenin against HCC. Wnty^-catenin pathway inhibitors ICGOOl, FH535, and other small inhibitors have been tested in HCC in different in vitro and in vivo models. Several groups showed that the combination of Sorafenib and ICGOO l, and the combination of Sorafenib and FH535, resulted in better treatment outcomes in experimental models respectively (Galuppo et al, Gedaly et al). However, studies have shown that inhibition of the β-catenin pathway in HCC is associated with the downregulation or degradation of wild-type β-catenin. Currently, no small molecule inhibitor that directly targets mutant β-catenin exists. Llovet and coauthors recently reviewed that immune exclusion and so resistance to immunotherapy is contributed by activation of the β-catenin pathway in HCC - consistent with findings in melanoma (Llovet et al 2018). Thus, Varlitinib could potentially reverse immunosuppression via vessel normalisation and targeting the activating β-catenin pathway in HCC. Combining Varlitinib with an immune checkpoint inhibitor in HCC would also be rational. To date, there are limited enrichment strategies for HCC clinical trials with any molecular targeted inhibitor (Llovet et al 2018).

Our present study demonstrates that Varlitinib shows good efficacy in the specifically defined subset of HCC that is dependent on both the ErbB pathway and β-catenin mutation. We also demonstrate that Varlitinib promotes apoptosis and vessal normalisation with reduction of tumour progression and hypoxia in this subset of HCC. Varlitinib effectively inhibits ErbB receptors and their downstream oncogenic signalling such as MEK/Erk and AKT/mTOR pathways. In addition, it also targets β-catenin with T41A mutation in ErbB-dependent HCC by inhibiting ρ-β- catenin Tyrl42/654 and p-RanBP3 Ser58, as well as inducing LGR5, p-LRP6, and DVL3 downregulation. As a result, mutated β-catenin is transported from the nucleus and cytoplasm to the cell membrane, reducing the expression of β-catenin-driven genes. Several clinical trials of Varlitinib in different cancer types, including cholangiocarcinoma and gastric cancer, have been initiated. Varlitinib has a comparably favorable safety profile compared to irreversible pan-ErbB inhibitors such as Neratinib and Dacomitinib. The recommended dose of Varlitinib for clinical trials is between 200mg BID to 500mg BID, which is equivalent to 41mg/kg BID to 102.5mg/kg BID in mouse (calculation is based on Nair and Jacob (Hulsen et al). Our study reveals that 50mg/kg BID in mice (~250mg BID in human) showed efficient growth supression in high p-ErbB2/3 expressing PDXs. Intrinsic resistance demonstrated by high β-catenin/YAPl-related gene signature in high p-ErbB2/3 expressing PDX could be overcome by increasing the dose to lOOmg/kg BID in mice (~490mg BID in human). In contrast, even with very high dose of Varlitinib given to the mice [lOOmg/kg BID), there is no anti-tumour benefit in the non-responders. Expression of p-ErbB2/3 and/or ErbB2/3 could be a predictive biomarker in the clinical treatment of HCC with Varlitinib.

In summary, Varlitinib is a pan-ErbB and, as we report here, an inhibitor of mutated β- catenin - that can selectively target the p-ErbB2/3 highly-expressing and CTNNB1 subset of HCC. Our data suggests that the selection of HCC patients with high p-ErbB2/3 expression could be a useful target and predictive biomarker for Varlitinib clinical efficacy. As the molecular and immune classification of HCC is further evolved and expanded, the role of tailoring systemic therapies, whether molecular or immune modulating, can potentially optimise treatment efficacy based on HCC subtypes.

Materials and Methods

Reagents, Xenograft models, Vessel perfusion study, Western blot analysis and Immunohistochemistry were the same as described in Example 1.

Transcriptome sequencing analysis, global gene expression analysis, and bioinformatic analysis

Three tumours from each condition (control, 50mg/kg, and lOOmg/kg Varlitinib treatment) of 95 HCC29-0909A and HCCOl-0708 and two tumours from each condition (control and l OOmg/kg treatment) of HCC16-1014 were used for transcriptome analysis. A total of 200 ng of total RNA was used for illumina TruSeq mRNA library prep, followed by 150bp paired-end sequencing on an illumina HiSeq4000 platform by BGI HK, Ltd. The raw sequencing reads were aligned to mouse mm 10 genome reference and human hg38 genome reference by Spliced Transcripts Alignment to a Reference (STAR) aligner version 2.5.3a, separately using Partek Flow (Partek Inc. St. Louis, M0). In average, 74 million to 90 million clean reads per sample were obtained. The total human alignments ranged from 70 million to 87 million per samples. Aligned reads were then quantify to hg38 - RefSeq Transcripts 85 -2018-05-02 annotation model using Partek E/M algorithm, followed by low expressed gene filtering and total count and add 0.0001 gene counts normalisation, GSA differential expression detection model was used to compare 50mg/kg Varlitinib treatment vs control and l OOmg/kg Varlitinib treatment vs control in PDX models using the cut-off threshold of >2 and -2 fold-change and FDR adjusted p-value < 0.05, followed by heatmap generation.

The gene expression analysis was done by quantile normalised data from Affymetrix

GeneChip Human Genome U133 Plus 2.0 Array with cut-off threshold of >2 and -2 fold-change and high stringent FDR adjusted p-value < 0.0001. The identified gene lists were compared using BioVenn online tool for the area-proportional Venn diagram analysis [Hulsen et al). The KEGG Pathway Enrichment analysis was carried out in Partek Genomics Suite [Partek Inc. St Louis, MO) with the cut-off threshold of FDR adjusted p-value <0.05. The differentially expressed gene lists were analysed through the use of Ingenuity Pathway Analysis (IPA) [Kramer et al) [QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuitypathw ay-analysis). Gene Set Enrichment Analysis [GSEA) was carried out in Partek Genomics Suite [Partek Inc. St Louis, MO) or javaGSEA with 100000 number of permutations. Molecular Signature Database (MSigDB) v6.2 was used for the analysis [Liberzon et al, Subramanian et al).

Statistical analysis

Statistical analysis was performed in GraphPad Prism version 7.00 for Mac OS X [GraphPad Software, La Jolla California USA, www.graphpad.com)

Tables

A B

Gene Sym ol Fold-change FDR adjusted p-value Gene Symbol Fold-change FDR adjusted p -value

FM03 30S.49 1.03E-11 PDGFA -299.455 6.12E-11

FOXG1 241.30 9.10E-14 CERS1 -298.654 7.96E-12

CLDIN4 225.57 7.12E-15 6DF1 -248.957 3.14E-12

TBC2DS 198.24 1.47E-Q9 LY6E -224.589 9.46E-15

ATP2B4 166.61 5.23E-10 CD81 -212.999 7.12E-15

ERVH4S-1 141.06 3.58E-04 CTD15 -209.852 9.76E-13

RORC 139,28 1.00E-11 ZNF853 -208.827 2.28E-06

I F6 109.17 2.SSE-12 TMEM132A -176.207 1.04E-10 1F26B 94.93 8.61E-09 EPO -157.635 8.78E-09

KCNK1 90.20 1.84E-10 COTL1 -127.493 2.21E-11

EVPL 86.83 1.12E-14 ARTS -125.S5 9.50E-13

CYP1A1 76.02 S.12E-12 ADGR01 -120.298 2.65E-12

ENPP4 71.01 2.72E-10 TUSC3 -95.4738 1.08E-10

GUCY2C 63.91 6.40E-11 GAA -91.7216 1.31E-13

NAGS 50.31 2.45E-10 FA 19A5 -81.2813 2.42E-05

MUC13 57.24 1.Q8E-09 TB BP1 -72.0157 4.34E-10

MAGEA6 56.06 6.11E-10 PN A6A -59.9852 7.12E-15

DSC3 52.33 4.75E-08 NAT8L -59.4931 3.99E-14

RNASEL 51.30 2.21E-10 DBN1 -58.5321 1.76E-04

S1∞A1 47.57 2.Q2E-14 ASMS -51.5197 4.37E-11

Pathway Name Enrichment Score Enrichment e -value

Ribssome 33.535S 6.00E-11

RNA transport 8.54517 G.G001944S2

Steroid biosynthesis 6.36656 0.000942895

Central carbon metabolism in cancer S.0SXS7 Q.Q02267S6

Ceil cycte 5.65022 0,00351575

HiF-l signaling at way 5.S632S 0.00383515

Biosynthesis of amino eciiis 5.2943S 0.00501967

Protein processing ifi erido ssms; reticulum 5.18743 0.0055S633

Fatty acid metabolism 4.82731 0.00800803

Biosynthesis of unsaturated f etty acids 4.48445 0.0112831 Table 1 - Transcriptome analysis of Varlitinib treatment in HCC29-0909A. (A) Top 20 common activated genes and (B) top 20 common repressed genes in l OOmg/kg and 50mg/kg BID Varlitinib-treated HCC29-0909A. FDR adjusted p-values are shown. (C) KEGG pathway analysis showing the enriched pathways generated from the gene lists of significant differentially expressed genes with the threshold of >2 and <-2 fold-change with FDR adjusted p-values < 0.05.

D Predicated repressed patinsays by iPA analysis Enrichment p-vaiue

B 2 Sig^aiing 3.34 x IO ^"!5

Regulation of siF4 ma p?OSSK Signaling 1.92 1 10 ^"¾

mTOR Signals 4.57 κ ΐΰ ^&

Purine Nudeotides De Novo Siosyrsitesis i) 3.04 x 10 ^' '

Superpath ay of Cholesterol Biosynthesis 5.25 yt 10 ^"' *

Predicted activated pat ways fey ΨΑ analysis Enrichment p -value

U V!H Mediated inhibition of RXR RmcttoR 2.7S s 10 ^'s

Acute Phase Response S¾naiir¾g 1.43 x 10 ^"4

FXR/FiXFS Activation 4.71 x 10

LXR/RXR Activation 9.31 x 10 ⁴

Table 1 - Transcriptome analysis of Varlitinib treatment in HCC29-0909A. (D) Ingenuity Pathway Analysis showing the predicted repressed and activated pathways with enrichment FDR adjusted p-values <0.05.

A B

Gene Symbol Fold-change FDR adjusted p -value

Gene Symbol Fold-change FDR adjusted p -value

CYP4A11 55059.40 1.98E-16

TM4SF1 -216.53 8.58E-11

GSTA1 90.87 5.16E-14

CPLX2 -44.9391 5.64E-07

HPD 57.74 S.72E-12

CDKN1C -43.3875 8.12E-07

NPW 48.32 4.55E-13

STRA6 -33.1811 3.77E-08

CFHR2 36.68 2.64E-04 DK 1 -31.3167 4.25E-11

5NMT 33.12 1.45E-0S

LYZ -28.5324 6.06E-1O

C!DEC 28.84 2.84E-11 NT5E -24.7924 4.30E-12

CD4 28.63 1.01E-13 FHL2 -22.5319 7.16E-10

CFHR1 2S.63 4.23E-12 SLC6A8 -21.7241 4.66E-07

1TIH4 27.51 1.45E-09 BGN -21.1068 2.Q7E-09

SLCD4C1 24.30 4.55E-13 CD8B -19.9052 1.26E-06

SERPfNGl 24.00 1.17E-12 FRZB -18.9016 2.77E-10

HMGCS2 23.49 8.72E-12 PLE HA2 -15.0221 2.40E-07

T P SS6 22.13 2.87E-12 UNGOl -13.8232 2.95E-11

ABUM3 21.92 1.77E-09 BNIP3L -13.5145 9.Q7E-12

AGPAT9 21.47 2.93E-G9 FA 19A5 -13.4038 3.03E-O2

AK 1C1 20.72 2.83E-09 TNFRSF25 -11.8909 4.60E-09

AKR1D1 20.56 1.71E-10 TFEB -11.5725 8.85E-09

CDH 5 18.36 5.64E-13 RNASET2 -11.3201 6.06E-10

LBP 18.31 1.96E-08 FA 43A -11.0796 1.79E-07

Table 2 - Transcriptome analysis of Varlitinib treatment in HCCOl-0708. (A) Top 20 common activated genes and (B) top 20 common repressed genes in lOOmg/kg and 50mg/kg BID

Varlitinib-treated HCCOl-0708. FDR adjusted p-values are shown. Predicated repressed pathways by I PA analysis Enrichment -vaiue

Axonai Guidance Signaling 6.00 x 10"

Adipogenesis pathway 1.09 x 10^

Osteoarthritis Pathway 2.08 x 10 ^";

Wnt/P-catentn Signaling 2.41 x ICf

Pyridoxa! 5' -phosphate Salvage Pathway 2.43 x lO '

Predicted activated pathways by IPA anaiysts Enrichment p -value

^■ 23

Superpathway of Cholesterol Biosynthesis 1 ,53 x 10

23

LXR/ X Activation 7.14 x 10

FXR/ XR Activation 2 £5 x 10 ^":

15

Choiesteroi Biosynthesis I 1.04 x 10

Cholesterol Biosynthesis II (via 24,25-dthydrolanosterol) 1.04 x 10 ^"

Table 2 - Transcriptome analysis of Varlitinib treatment in HCCOl-0708. (C) Ingenuity Pathway Analysis showing the predicted repressed and activated pathways with enrichment FDR adjusted p-values based on the gene lists of significant differentially expressed genes with the threshold of >2 and <-2 fold-change with FDR adjusted p-values < 0.05.

A

Activated Genes Repressed Genes

?MQ3 TNSEMI32A

RO C EPO MAGS FAfV lSAS

AO! F ASMS

cm ma

TTB l

51O0A14 nsm

S5JC16A14 HAL

AZSP1 CBSL:

PPi IOC1005071S4

LQC2GQ772 6PC4

TN !X CERC iyi B

SiJClA? A0RA2C

Pathway Name Enrichment Score Enrichment p-va

BCHE STONEl Central carbon metabolism in cancer 1.8OE+01 1.46E-08

CYP4A11 SPARC H!F-l signaling pathway 14.2551 6.44E-07

SLC6AS G!ycoiysis / Glucorteogenesis 7.61446 0.000493269

ENPEP SCARF2 Alanine, aspartate and glutamate metaboiism 5.33005 O.0O4843S2

HAOl HS6ST2 Biosynthesis of amino acids 5.11366 0.00601403

CYP3A5 TGF-beta signaling pathway 4.67044 0.00936819

SHMT Ribosome 4,58088 0.0102459

G!ycosaminogiycan biosynthesis - heparan su!f; 3.76538 0.0231589

Carbon metaboiism 3.64368 0.0259994

Giycosphmgoiipid biosynthesis - !acto and neo! 3.61461 0.0269275

Valine, leucine and isoleucine biosynthesis 3,2439 0.0390115

Fructose and tnannose metabolism 3.17348 0.0418576

Neomycin, kanamydn and gerttamidn biasyntr 3.02562 0.0485276 Table 3 - The 3-way Venn diagram analysis revealing the Varlitinib-inhibition gene signature. (A) The top 20 activated and repressed genes commonly identified in 50 and l OOmg/kg BID Varlitinib-treated HCC29-0909A as well as lOOmg/kg BID Varlitinib-treated HCCOl-0708. (B) KEGG pathway analysis showing the enriched pathways with FDR adjusted p-values <0.05 generated from the lists of commonly dysregulated genes.

c Predicated repressed pathways by iPA analysis Enrichment ρ -value

Glycolysis i 1,28 x 19

Histidine Degradation Hi 2.19 x 10 -3

Purine Nucleotides De Novo Biosynthesis Si 4.23 10

Asparagirte Biosynthesis i 9,03 x 10 ^""

E!F2 Signaling 1.27 x 1G

Predicted activated pathways by \?A analysis Enrichment p -value

cont

LPS/!L-1 Mediated inhibition of RXR Function 7.74 x 10 ^"7

Siyctfie Setairse Degradation 3,21 X 10

FXR/RXR Activation 1.30 x lO ^"5

Acute Phase Response Signaling 2.39 X 10 ^' ·

Tryptophan Degradation l!i {Eukaryottc} 1.21 x 10 ^" *

Table 3 - The 3-way Venn diagram analysis revealing the Varlitinib-inhibition gene signature. (C) Ingenuity Pathway Analysis showing the predicted repressed and activated pathways with enrichment FDR adjusted p-values < 0.05.

A Gene Symbol Foid-c ange FDR adjusted p tese Sym B CNTN3 -148.708 1.76E-12 bol P okf -change ?0f¾ adjusted p -v

ADH1B -138.106 5.74E-G9 mm. S51.22 1,3S£-11

2NF667-AS1 -92.1645 8.S6E-13

4¾,S2 S.54E.-0S

CYP26B1 -83.2785 4.45E-11

201.28 9.73E-H3

ADH1ES -66.5402 3.76E-07

DiOl 194.33 1.23.E-OS

PCDHB5 -62.4209 1.18E-11

FCER36 185.30 3.08E-13

nmffj m%&?s 2.25E 07 iGFBP2 -57.8371 4.S0E-09

DNAJC1S 147.41 7.7 -Ϊ3 TUBB2B -53.0594 8.98E-10

1J.2E 11 ZMF567 -52.277 2.23E-11 mm CADPS -47.7625 6.26E-Q7

101.27 S.¾K-09

8C034319 // Ϊ¾ Η-ΗΝ3. 33.70 3..0G&-1Q UMS3-LOC440895 -43.4773 1.24E-09 m 14

33.43 ZNF667 -42.Q6S2 4.63E-

9.7SE-0?

87,01 S.?5£-0S DiP2C -40.777 2.14E-13

78.39 6.7Q£-H3 PDGFA -38.5822 1.41E-10

74.41 9.856-07 PCDHB14 -37.4345 1.20E-11

CCOCtSS-SOHlH2 /// SOH1.H2 S8.48 GABRB2 -37.1071 2.87E-14

S8.3 1.93M4 NAP1L3 -34.2161 5.92E-08

S7.05 &78E-li TDRP -33.0318 9.52 E- 10

S3.52 1.27M! GLYR1 /// SEPT6 -32.8903 5.86E-12

Ps.A<5U 59,78 2.34E-li PRDM16 -31.3786 1.32E-12

C036 5S.OS 3.4SE-08 Predicated repressed pathways by IPA analysis Enrichment p-value c Inhibition of Angiogenesis by TSPl 1.44 x ID ^'5

Epithelial Adherens Junction Signaling 4.7? x ID ^'4

Regu!ation of the Epitheiiai Mesenchymal Transition Pathway 5.39 x 10 ^'4

Human Embryonic Stem Cell P!uripctency 1.14 x 10 ^¾

Predicted activated pathways by IPA analysis Enrichment p -value

Glutathione Redox Reactions I 1.10 x 10 ^'e

Giutathione-mediated Detoxification 2.10 x 1G ^"3

FXR/RXR Activation 6.62 x 10 ⁴

Tetrahydrofolate Saivage from 5,10-methenyitetrahydrofoiate 5.78 x 10 ^"4

PXR/RXR Activation 1.01 x lO ^"3

Table 4 - Identification of Varlitinib-resistant gene signature by global gene expression analysis. (A) The top 20 highly expressed genes and (B) the top 20 lowly expressed genes identified in Varlitinib-sensitive PDXs, HCC29-0909A, HCCOl-0708, and HCC07-0409 models. (C) Ingenuity Pathway Analysis showing the predicted repressed and activated pathways with enrichment FDR adjusted p-values < 0.05. ** indicates p-value < 0.01, *** indicates p-value < 0.001, **** indicates p-value < 0.0001.

Gene Symbol Fo!d-change F R adj tssted ρ -va i tte

Gene Symbol ^: old-change FD adjusted p-value

DUSS¾ 52.90 S.03E-11

A CYP4A11 -21309.6 3.05E-15

SEMASB 51,47 6.41 E- 11 B GSTA1 -43.6589 Z46E-12

PPP1R14C 47,44 2.5QE-9?

SAA4 -38.6979 5.03E-11

HEYl 45,08 2.72E-12

CFHR2 -2S.45S 6.48E-04

NFE2 40.87 1.67E-09

HRG -27.1965 2.08E-09

TM4SF1 38,23 8.17E-Q8 CiDEC -22.4431 2.52E-10

COL9A3 37.07 6.22E-09 NPW -21.3434 4.41E-11

P6C 35.26 3.6QE-Q3

UGT1A4 -20.3517 1.29E-I0

SOX9 29.47 3.9SE-11

CFHR1 -20.8533 5.96E-11

SGF2 24,55 7.12E-Q9

AGPAT9 -16.4033 2.36E-QS

FGFR1 24,15 1.09E-1Q AKR1C1 -15.85 2.86E-08

C8orf4 23,70 4.43E-06 POMl -15.5251 2.58E-D5

FR2B 21.88 3.S0E-1Q SERP1NG1 -15.1625 4.41E-11

TWRSF25 19,76 8.58E-19 HULC -13.6682 1.35E-07

6TO 19.70 1.18E-09 CD4 -13.1305 2.24E-11

17.61 1.35E-05 ALDOB -13.0788 6.22E-09

TFES 17.51 2,17£-08 WNT11 -12.6081 1.25E-07

CPLX2 17,23 2.28E-05 HPD -12.5572 2.04E-08

SPHKl 15.38 4.39E-OS A R1D1 -11.9887 7.12E-09

M JC13 15,13 9.81E-07 FETUB -11.8688 1.81E-10

Table 5 - Identification of Varlitinib-potency gene signature by transcriptome analysis. (A)

The top 20 highly expressed genes identified in higher Varlitinib-potent PDX HCC29-0909A when compared to lower Varlitinib-potent HCCOl-0708. (B) the top 20 lowly expressed genes identified in higher Varlitinib-potent PDX HCC29-0909A when compared to lower Varlitinib-potent HCCOl- 0708. Pathways activated in HCCOl-0708 nrichment Score Enrichment p-vaiue

c Proteoglycans in cancer 9.25E+00 9.65E-05

ECM-receptor interaction 7.62369 0.000488737

Hippo signaling pathway 6.49953 0.00150415

Centra! carbon metabolism in cancer 6.46148 0,00156248

Small ceil lung cancer 5.93642 0.00264148

Arrhythmogenic right ventricular cardiomyopa 5.76271 0.00314259

Focal adhesion 5.55414 0.003S714

Hippo signaling pathway -multiple species 5.27186 0.00513406

Wnt signaling pathway 5.20257 0.00550239

Hypertrophic cardiomyopathy (HCM) 4.99425 0.00677678

Pathways activated in HCC29-0909A nrichment Score Enrichment p-vaiue

D PPAR signaling pathway 37.5943 4.71Ε-Ϊ7

Metabolic pathways 30.2388 7.37E-14

Complement and coagulation cascades 19.2842 4.22E-09

Fatty acid metabolism 17.9968 1.53E-08

Steroid biosynthesis 17.8928 1.70E-08

Steroid hormone biosynthesis 15.5438 1.78E-07

Fatty acid degradation 14.5233 4.93E-07

Fatty acid biosynthesis 12.4488 3.92E-06

Terpenoid backbone biosynthesis 11.4075 1.11E-05

Pyruvate metaboli m 11.3895 1.13E-05

HCCOl-0708 lOOmg/kg Var!itinib BID

HCCOl-0708 HCC29-Q9Q9A

Gene ID Fold-change FDR p -value

Gene ID Fold-change FDR p -value

CT6F 13.54 CYR61 -49.52

9.17 x l0 ^'s 9.75 x 10 ^s

CYR61 12.53 CTGF -43.59

1.09 x 10 ^' * 1.81 x lO ^"11

MYC 3.68 2.68 x 10 ^"5 MYC -9.88 4.63 10 ^"9

YAPl 2.57 2.62 x 10 ^'6 YAPl -3.84 8.43 x 10 ^"3

Table 5 - Identification of Varlitinib-potency gene signature by transcriptome analysis. . (C)

Ingenuity Pathway Analysis showing the predicted activated pathways with enrichment FDR adjusted p-values < 0.05. (D) Ingenuity Pathway Analysis showing the predicted repressed pathways with enrichment FDR adjusted p-values < 0.05. (E) The expression of CTGF, CYR61, MYC, and YAPl in HCCOl-0708 when compared to those in HCC29-0909A. (F) The expression of CTGF, CYR61, MYC, and YAPl in lOOmg/kg BID Varlitinib-treated HCCOl-0708 when compared to those in vehicle control-treated HCCO l-0708. Gem set enriched in Rftsgwriersvs Nen-respomfers fjo. of probe: s £S NES p-vatus tf-value

143 0.6724 2.1126 < 0.0001 0.0Π9

122 0.8784 2.1088 0,0001 0.0t07

375 0.6835 2.0936 < 0.0001 o.otso

2,0333 < 0,0001 0.0245

YAP1J 0.5640 2.0263 « 0,0001 0.S23¾

SANSOM_APC„TAR ¾TS„UP 2 0.4834 1.δ45δ « 0.0001 0.0427

HALUWRKJ^YCJARQETSj/2 30 0.534? 1.78S4 0.0119 0.0427

Re CTOME .C NR8 _^ HOSPHORV TiON _^ CASCADe &1 0,7036 -2 0.0012 0.0468

HALL AR „ OTCH„Si¾NALSNS 86 0.5550 - S47S < 0.0001 0.0028 EGG-TGF-beia sSgnating patitway 2t9 0.4275 -t.S420 « 0-0001 0.0480

Table 6 - Gene enrichment analysis in Varlitinib responder and non-responder PDXs. GSEA of Varlitinib responders vs non-responders with the use of MSigDB v6.2 [Liberzon et al, Enane et al] . The enriched gene sets with q-value < 0.05 are shown. ES refers to enrichment scores, NES refers to normalised enrichment scores.

ERBB^ £ BB3 ^L ™ Table 7 - Analysis of ERBB3 ^m and ERBB3 ^Low

SNU878 JHH6

SNU761 SNU423 liver cancer cell lines. The list of ERBB3 ^Ui and

Hu 1 SNU475

ERBB3 ^ovi liver cancer cell lines.

Li7 SNU387

Huh7 SNU398

Alexander SNU182

JHH1 JHH2

PLC HLF

Huh6 HLE

JHH7 JHH4

Hsp3B SNU449

JHH5 SNU886

HepG2

MCIH684

C3A

Gene Symbol Fold-Chanqe FDR-adjusted p-value Table 8

NRG1 3.199 3.998E-19

BC034319///RP11-11N9.4 2.945 1.186E-17 Varlitinib

— 2.884 8.527E-22 treatment in HCC

CRISP2 2.536 7.433E-15

FGF13 2.511 2.891 E-30 PDXs. The

LINC01419 2.489 6.242E-11

CDH19 2.427 1.991E-12 differential

MME 2.393 1.156E-13 expressed gene

FCER1G 2.393 4.980E-10

SLC1A2 2.348 8.165E-14 (DEG) analysis of

OLR1 2.326 1.156E-11

FSTL5 2.318 2.313E-10 Varlitinib

CCND2 2.316 8.888E-09

COL1A2 2.300 1.634E-12 responders vs non-

ERVW-1 2.228 1.127E-26 responders. The

RP11-382F24.1 ///RP11-382F24.2 2.224 5.033E-09

— 2.217 1.752E-13 DEG was done

FMN2 2.200 2.281 E-14

PTN 2.173 2.777E-17 with microarray

RBM46 2.153 1.060E-12 data and threshold

FMN2 2.132 2.537E-12

CD36 2.127 1.227E-08 of fold-change > 2

DCAF4L2 2.113 6.228E-08

TMEM52B 2.113 1.149E-09 and FDR-adjusted

NFE4 2.078 1.934E-32 p-value < 0.0001 in

GPX8 2.069 3.985E-11

FBXL13 2.067 1.132E-29 Partek Genomics

APCDD1 2.065 1.062E-11

EDDM3A 2.049 2.538E-11 Suite.

CD36 2.048 1.108E-08

BAGE 2.039 3.454E-21

COLEC12 2.036 6.134E-21

ABCA8 2.027 3.800E-17

RNASE2 2.020 4.932E-17

PTN 2.016 1.463E-16

PAG1 -2.005 7.201E-12

GLYR17/1 SEPT6 -2.020 4.046E-18

SEPT6 -2.049 1.686E-18

SMARCA2 -2.098 4.456E-16

MBP -2.116 3.778E-22

RASEF -2.161 6.082E-09

LIMS3-LOC440895///LOC100288570

///LOC100507334///LOC440895 -2.222 1.178E-11

STOX2 -2.272 6.801E-11

DIP2C -2.320 9.043E-09

SMARCA2 -2.360 7.745E-20

SMARCA2 -2.602 1.588E-19

PDX moefcls fe>spo»s«si t» VwStmib CGfR mz £8883 £¾Βδ ann «ef»t&>s Statu*

$-£€81-0708 Res onds? ΡΙΠ0Α, SS55V T41A Hepatitis 8 positive

NCC07-04Q3 itespomief P-ilXfA, 56S5V m T41A MepaStis g positive

HCO9O90SA Sesporsctet RSSIK m70A, i$55V mi m T4 H^ st!tis 8 positte

BCOBi-0308 Sto esiWssier 8:i mi m m H«patW« 3 pos¾fo«

fiOfi-ress ifSief mi m w\ Hesjaiitis 8 positi e

HCC89-C913 mi mi m m fiepfivs

Hceis-iiios Neri-res&onder H\i i m mi H«p8tW« 8 positive mi m mi t® m Hepatitis 8 positive

Stef r sswsier mi m i m i Hepatitis C positive

H€Ci.S-1814 m mi me mi tegativfi

Hca?~02ii mi m m mi tegative

HCC21-¾208 mi MtHstixi m Hepatitis 8 positive

HCC2S-0TO5A mi m Mil m mi Hepatitis 8 positive

HCOS-OSOSA mi m mi i mi Hepatitis 8 positives

(«0.6-03088 itetv-resiWicte!' mi m mi un i

H€C2S 11G4 mi mi m mi Hepatitis 8 positive mi m mi mi mi He atsfe S positiva

Table 9 - Analysis of responders and non-responders. Mutations and Hepatitis statuses of the treated PDXs. Mutations at EGFR, ERBB2, ERBB3, and CTNNBl were determined by whole exome sequencing. Nil represents no mutations being detected.

HCC16-1014

Differentially Expressed Gene Fold-change FDR-adjusted p -value

SDS 3.83 6.64 x 10

CYP3A4 3.65 4.00 x 10

UGT1A1 2.65 2.00 10

FXYD2 -2.88 2.00 x 10

Table 10 - Transcriptome analysis of Varlitinib treatment in HCC29-0909A, HCCOl-0708, and HCC16-1014. DEG analysis from RNA-Seq analysed Varlitinib-treated non-responder PDX, HCC16- 1014. Four genes passed the threshold of fold change >2 and p-value < 0.05.

Coft¾fiKsr! name f 08 adjus ed vaiue

c s CD68 10-17 1.03 se 10 ^' ~ ^!

itgax emu 80,33 1,30 x 10 ^"s

itgam CDiib 52B 5 52 x 10 *

Aelgr&l F4/80 S.96 4.43 X 10 ^' *

ty€C &M 7.12 X 30 ^s

m∞ !i-4R S.09 8.43 x 10 ^" *

igais3 MSAC2 22,18 3,09 x 10*

Csflf C 115 18.48 1S7 x 10 ^* *

Η2Άϋ MHC ass Si 1 34 x IS *

H2-Abt MHC class « 40-00 7,18 x 10*

Η2-ΰΜύ MHC class 20-62 5,46x 10 ^' *

2~OMbl MHC class il 4¾-S« 1,64 X 10

MHC class il 2-40 2.28 X .10 ^'

H2~ebl MHC class » 38-84 1,08 10* Table 11 - Immune gene signature in high dosing of Varlitinib-treated HCCOl-0708 PDX model. (A) Differentially expressed myeloid-related genes. FOR adjusted p -value

B Steffi 5.16 3.48 x iff ⁷

T»/ 39,33 1,32 * Iff

No$2 29.52 4.20 * 10*

trfS 20.73 7.05* iff*

s a z.m 0.02

Sac$3 9.03 1.16 x 10*

Tfhl 7. 5.20)! Iff*

Sbno2 &M 1.74 x 10*

State 6.44 231 a 10*

CdM 5.18 S.llx ID ^'3

m 4.14 6.71x10*

Table 11 - Immune gene signature in high dosing of Varlitinib-treated HCCOl-0708 PDX model. (B) Differentially expressed Ml-like and M2-like tumour-associated macrophage (TAM]- related genes.

1.66

2.41 1.01 x 5 ·

Table 12 - Higher EGFRand ERBB3 expression detected in Varltinib sensitive group

CTNNBf -« 0? ι.*δχία· ⁸

L F1 33.00 β.52 1ΰ' ^!

TCF4 1486

cmt 3.21 6.01 x icH

MMP2: 2.10 4.04 χΐδ"*

T8H3 7.46 xiiH

EPN82 3.23

SPARCl 1 2J0

¾enfe¾yj¾-teal ^; Fpk*~c ¾ftge Fm 6fa <i p-vzUm

B T F R1 4,61 x10*

7&FBR2 -6.51 1,02x10*

IGP2 -42,64 .0 x10 ^* *

st ■ 33 1.83*10'*

mrcrn S,0§x 0«

JAG1 -7.10 2.SSxia*

ΤΈΑΒ1 1..24X1 - ⁸

TEAD2 1.46 κΐβ- ⁸

Table 13 - Identification of β-catenin mutation and the constitutively active β-catenin pathways in the Variitinib-sensitive PDXs. (A) Expression of β-catenin and its downstream targets and (B) expression of Wnt/TGF -related targets, in the four analysed PDXs indicated by normalised probe intensity.