FIELD OF THE INVENTION

The present invention relates to the field of classification, prognostics and treatment of cancer, in particular colorectal cancer.

BACKGROUND

Colorectal cancer (CRC) is the second leading cause of mortality among cancer patients in the world and the third most diagnosed form of cancer globally. It represents a huge burden on healthcare systems. The most important prognostic characteristic of CRC is presence or absence of lymph node metastasis (Chang G. J. et al. J. Natl. Cancer Inst, vol.99, p. 433-441 (2007); Iddings D. and Bilchik A. J. Surg. Oncol., vol. 96, p. 671-677 (2007); Nicastri D. G. et al. J. Mol. Diagn., vol. 9, p. 563-571 (2007)). Relevant lymph nodes are accessible for investigation only when patients are resected for cure.

Therefore, thorough determination of the lymph node status in the resected tumor specimen is crucial. Currently, approximately 50 % of patients with tumor-cell-positive lymph nodes, i.e. stage III CRC (anyTNl-2M0) and about 25% with no detected tumor- cell positive lymph nodes, i.e. stage I fJl-2 N0M0) and stage II (T3-4N0M0) patients will recur (Bockelman C. et al. Acta Oncol., vol. 54, p. 5-16 (2015)). These results strongly suggest that tumor cells in lymph nodes may vary in aggressiveness, and that presence of tumor cells in the node in many cases is missed by the present standard method. Therefore it is of utmost importance to 1) accurately detect presence of tumor cells in lymph nodes and 2) determine their metastatic potential i.e. their aggressiveness. By improving determination of lymph node status, N-staging, and introducing the aggressiveness parameter for the spread tumor cells, improved staging will be achieved thereby avoiding undertreatment of stage I and II patients and overtreatment of stage III patients. Moreover, if patients with tumor cells in their lymph nodes can be classified into subgroups according to differences in risk of recurrence and cancer death this information may be used not only for treatment with the current arsenal of drugs but also in the development of new drugs, new treatment schedules as well as for follow-up schedules adjusted to the risk of recurrence, etc.

In clinical practice, presence or absence of lymph node metastasis is currently determined by histopathological examination of hematoxylin & eosin (H&E) stained tissue sections of resected regional lymph nodes. Present guidelines require that at least 12 lymph nodes should be examined (Tsai H. L. et al. BMC Surg., vol. 16, p. 17 (2016)) In the TNM classification, Nl signifies that 1 to 3 examined nodes were positive for presence of tumor cells and N2 that 4 to 6 nodes were positive. N2 patients have poorer prognosis than Nl patients. Moreover, the lymph node ratio, i.e., number of positive lymph nodes over total number of examined lymph nodes, is an important prognostic factor - the higher ratio the worse prognosis (Parnaby C. N. et al, Br. J. Cancer, vol. 113, p. 212-219 (2015)). The main reasons why tumor cells are missed by the routine method are twofold: too small sample size and inadequate sensitivity. At best, only a few % of the volume of the lymph node is examined by H&E staining of tissue sections. An alternative method is to determine the mRNA level of one or several biomarkers that is expressed in all tumor cells of this type, and to extract RNA from the entire lymph node or, as for ethical reasons is the current option, half the node. It has been shown that real time quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) analysis with copy standard is a most useful method for mRNA analysis of biomarkers. It is highly sensitive, objective, quantitative, and amendable for automation. It was found that mRNA analysis of the biomarker carcinoembryonic antigen (CEA, CEACAM5) is very useful for detection of tumor cells originating from the large intestine. This biomarker allowed the identification of stage I and stage II patients with tumor cells in their lymph nodes that were not detected by the present gold standard, i.e. histopathology of H&E stained sections (Ohlsson L. et al. Br. J. Cancer, vol. 95, p. 218-225 (2006); Ohlsson L. et al. Int J. Cancer, vol. 130, p. 1833-1843 (2012)). Some of these patients have

succumbed from recurrent disease (Ohlsson L. et al. Br. J. Cancer, vol. 95, p. 218-225 (2006); Ohlsson L. et al. Int. J. Cancer, vol. 130, p. 1833-1843 (2012)). Thus, a more sophisticated stratification was obtained by using this marker compared to the gold standard only. The biomarker cytokeratin 20 (CK20) is also useful for this purpose, albeit somewhat less sensitive (Ohlsson L. et al. Br. J. Cancer, vol. 95, p. 218-225 (2006). Currently there is only one published biomarker for CRC that displays the properties of an aggression marker, namely kallikrein related peptidase 6 (KLK6) (Ohlsson L. et al. Br _* J. Cancer, vol. 107, p. 150-157 (2012)]. It is ectopically expressed in CRC tumor cells and appears to be expressed at increasing levels with increasing aggressiveness.

A generally accepted pathway for the development of distant metastases in CRC is that tumor cells leave the primary site in colon or rectum via lymphatic vessels, first settle in a regional lymph node, and thereafter, spread to distant sites like the liver. It is the distant metastasis that eventually kills the patient Evidence for this pathway is the fact that presence or absence of tumor cells in a regional lymph node is the best prognostic marker for CRC death or survival.

The present invention concerns: 1) the identification of two new aggression biomarkers for CRC; one expressed in the CRC tumor cells themselves and the other in supporting cells in the microenvironment of the lymph node. 2) A method for determination of lymph node status, which accurately detects presence or absence of tumor cells in the lymph nodes, and in addition provides information on the aggressiveness of these cells. In the proposed method, quantitative mRNA levels of the 2 new biomarkers and 3 previously described biomarkers are determined. If applied for CRC lymph node analysis, it will accurately determine lymph node involvement and allow classification of CRC patients into different risk groups with respect to risk for recurrence and cancer death after the primary treatment, i.e. surgical resection of the tumor. This goal has hitherto not been possible to achieve.

OBJECT OF THE INVENTION

It is an object of the present invention to provide a group of molecular biomarkers, which are useful for classification, for prediction of prognosis and for guiding treatment decisions of a subject with colorectal cancer. It is another object of the present invention to provide objective and quantitative methods for classifying colorectal cancer in a subject, as well as for using the classification for predicting prognosis of the subject and for making a treatment decisions for the subject

DESCRIPTION OF THE INVENTION

The present inventors have identified the expression levels of the genes solute carrier family 35 member D3 (SLC35D3) (GenBank N.M_001008783) and periostin, osteoblast specific factor (POSTN) (GenBank NMJJ06475) as molecular biomarkers that can be used for determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject

Expression levels of the genes SLC35D3 and POSTN can preferable be used together with the expression levels of the gene kallikrein related peptidase 6 (KLK6) ( GenBank NM_002774), and even more preferably also together with the expression level of the gen mucin 2, oligomeric mucus/gel-forming (MUC2) (GenBank NM_002457) for determining the metastatic potential and/or tumor aggressiveness.

The method can be applied to determine gene expression levels in regional lymph node samples obtained from the subject, or in primary intestinal tumor, blood, and/or feces samples obtained from the subject

The expression levels of these genes can be related to the expression level of the gene carcinoembryonic antigen related cell adhesion molecule 5 (CEACAM5) (GenBank NM_004363), which is a known tumor marker, and/or related to the level of 18S rRNA.

Accordingly, one aspect of the present invention provides methods for determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject comprising the steps:

a] determining the gene expression levels of genes SLC35D3, and POSTN in a regional lymph node sample obtained from the subject; and b) comparing the gene expression levels determined in step a) with reference gene expression levels of the same genes in a reference patient population; wherein higher expression levels of the genes SLC35D3 and POSTN compared to the reference are associated with an increased metastatic potential and /or tumor aggressiveness.

Preferably the method can further comprise determining the gene expression level of the gene KLK6 in said sample.

Accordingly, in one embodiment the first aspect of the present invention provides methods for determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject, comprising the steps:

a) determining the gene expression levels of genes SLC35D3, POSTN and KLK6 in a regional lymph node sample obtained from the subject; and

b) comparing the gene expression levels determined in step a) with reference gene expression levels of the same genes in a reference patient population; wherein higher expression levels of the genes SLC35D3, POSTN and KLK6 compared to the reference are associated with an increased metastatic potential and /or tumor aggressiveness.

Preferably the method can further comprise determining the gene expression level of the gene MUC2 in said sample.

Accordingly, in another embodiment the first aspect of the present invention provides methods for determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject, comprising the steps:

a) determining the gene expression levels of genes SLC35D3, POSTN, KLK6 and MUC2 in a regional lymph node sample obtained from the subject; and b) comparing the gene expression levels determined in step a) with reference gene expression levels of the same genes in a reference patient population; wherein higher expression levels of the genes SLC35D3, POSTN, KLK6 and MUC2 compared to the reference are associated with an increased metastatic potential and /or tumor aggressiveness. Preferably the method can further comprise the steps

c) determining the gene expression level of the gene CEACAM5 and the level of 18S rRNA in said sample ;

d) based on the results obtained in steps a) and c) calculating the ratios

SLC35D3/CEACAM5, KLK6/CEACAM5, POSTN/18S rRNA, and MUC2/CEACAM5; e) giving the ratios obtained in step d) the values of (+1) or (0) depending on

whether said ratio is larger than a cut-off value based on the same ratio in said reference patient population, and where ratios higher than the cut-off value obtain a value of (+1) and values lower than the cut-off level obtain a value of (0); and

f) calculating an index using the ratios obtained in step e) using the formula [A = SLC35D3/CEACAM5 + KLK6/CEACAM5 + POSTN/18S rRNA - MUC2/CEACAM5]; wherein the index (+3) is associated with very high metastatic potential and /or tumor aggressiveness, the index (+2) and (+1) with high metastatic potential and /or tumor aggressiveness, and the index (0) and (-1) with low metastatic potential and /or tumor aggressiveness.

Said cut-off values can be the ratios of the 7th decile of said reference patient

population, the ratios of the 3 ^rd quartile of said reference patient population, or the ratios of the 8th decile of said reference patient population.

The methods can be performed in vitro and/or ex vivo.

The methods can further comprise the additional step of treating colorectal cancer in a subject in need thereof.

In another aspect the present invention provides methods of determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject, comprising: a) determining the gene expression levels of the genes SLC35D3, KLK6, MUC2

and CEACAM5 in a primary intestinal tumor, blood, or feces sample obtained from the subject;

b) based on the result obtained in step a) calculating the ratios

SLC35D3/CEACAM5, KLK6/CEACAM5 and MUC2/CEACAM5; and c) comparing the ratios determined in step b) with reference ratios calculated from expression levels of the same genes in a reference patient population; wherein higher ratios SLC35D3/CEACAM5 and KLK6/CEACAM5 compared to reference are associated with an increased metastatic potential and/or tumor aggressiveness and a higher ratio MUC2/CEACAM5 compared to reference is associated with decreased metastatic potential and/or tumor aggressiveness.

The methods can be performed in vitro and/or ex vivo.

The methods can further comprise the additional step of treating colorectal cancer in a subject in need thereof.

According to the invention the gene expression levels can be determined by quantifying the amount of mRNA expressed from said genes.

The amount of mRNA can be determined by hybridization, sequencing or quantitative RT-PCR.

More specifically the amount of mRNA can be determined by use of a method selected from microarray and bead array technologies, transcriptome sequencing, real time quantitative RT-PCR, multiplex quantitative RT-PCR.

According to the methods the gene expression levels can be determined using RNA or DNA copy standard and/or the 18S rRNA level can be determined using 18S rRNA standard.

Another aspect of the present invention provides methods for determining the prognosis of a subject diagnosed with colorectal cancer. Said method can comprise determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject using a method according to the present invention.

Said methods can comprise determining that the subject has a good prognosis if the metastatic potential and/or tumor aggressiveness is low, or

determining that the subject has a poor prognosis if the metastatic potential and/or tumor aggressiveness is high. Poor prognosis can be a decrease in the likelihood of survival compared to the good prognosis.

The methods can be performed in vitro and/or ex viva.

The methods can further comprise the additional step of treating colorectal cancer in a subject in need thereof.

Another aspect of the present invention provides methods for determining the treatment for a subject diagnosed with colorectal cancer and having a tumor. Said method can comprise determining the metastatic potential and/or tumor

aggressiveness of a colorectal cancer in a subject using a method according to the present invention and determining the treatment for said subject dependent on the metastatic potential and/or tumor aggressiveness determined.

The methods can be performed in vitro and/or ex vivo.

The methods can further comprise the additional step of treating colorectal cancer in a subject in need thereof.

The treatment can be to give postoperative treatment, e.g. chemotherapy, to a patient determined to have a high metastatic potential and/or tumor aggressiveness.

The treatment can be to abstain from postoperative treatment to a patient with a low metastatic potential and/or tumor aggressiveness.

Another aspect of the present invention provides a kit for determining metastatic potential and/or tumor aggressiveness of a subject diagnosed with colorectal cancer. The kit can comprise nucleic acid primers and probes for determination of the gene expression levels, of one or more of the genes CEACAM5, KLK6, POSTN, SLC35D3, and MUC2 and optionally nucleic acid primers and probes for determination of the level of 18S rRNA. In one embodiment the invention provides a kit for determining metastatic potential and/or tumor aggressiveness of a subject diagnosed with colorectal cancer, comprising nucleic acid primers and probes for determination of the gene expression levels of the genes SLC35D3 and POSTN.

In another embodiment the invention provides a kit for determining metastatic potential and/or tumor aggressiveness of a subject diagnosed with colorectal cancer, comprising nucleic acid primers and probes for determination of the gene expression levels of the genes SLC35D3, POSTN and KLK6.

In another embodiment the invention provides a kit for determining metastatic potential and/or tumor aggressiveness of a subject diagnosed with colorectal cancer, comprising nucleic acid primers and probes for determination of the gene expression levels of the genes SLC35D3, POSTN and MUC2.

In another embodiment the invention provides a kit for determining metastatic potential and/or tumor aggressiveness of a subject diagnosed with colorectal cancer, comprising nucleic acid primers and probes for determination of the gene expression levels of the genes SLC35D3, POSTN, KLK6 and MUC2.

In another embodiment the invention provides a kit for determining metastatic potential and/or tumor aggressiveness of a subject diagnosed with colorectal cancer, comprising nucleic acid primers and probes for determination of the gene expression levels of the genes SLC35D3, POSTN, KLK6, MUC2, and CEACAM5.

The nucleic acid primers and probes can be selected from those given in Table 1. More specifically the nucleic acid primers and probes can be selected from SEQ ID NO:l, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13 ,14, or IS.

The kit can further comprise mRNA, RNA and/or DNA copy standards.

Another aspect of the present invention provides methods for treatment of colorectal cancer. Said method can comprise determining the metastatic potential and/or tumor aggressiveness of a colorectal cancer in a subject using a method according to the present invention, and treating said subject dependent on the metastatic potential and/or tumor aggressiveness determined.

The treatment can be to give postoperative treatment, e.g. chemotherapy, to a patient determined to have a high metastatic potential and/or tumor aggressiveness.

The treatment can be to abstain from postoperative treatment to a patient with a low metastatic potential and/or tumor aggressiveness.

FIGURE LEGENDS

Figure 1. Expression levels of CEACAM5, MUC2, KLK6, POSTN and SLC35D3 mRNA in

Figure 2. (A) SLC35D3 and (B) POSTN mRNA expression levels in lymph nodes of patients with stage I to IV CRC and control patients (Ctr). Each of the 166 CRC patients and 23 control patients is represented by the lymph node with the highest mRNA value.

Figure 3. Ratios of biomarker mRNA over CEACAM5 mRNA in lymph nodes from patients with stage I to IV CRC. Each of the 166 patients is represented by the lymph node with the highest mRNA value and indicated by a filled circle.

Figure 4. Cumulative survival curves according to Kaplan-Meier for CRC patients (n = 166). Patients are classified in groups (-1, 0, +1, +2 and +3) based on the mRNA value of the biomarkers SLC35D3, POSTN, KLK6, MUC2 and CEACAM5 and calculated according

decile of the mRNA values for each marker was used to

classify the marker value as positive or negative, giving the former a value of CI) and the latter a value of [0). The lymph node with the highest CEACAM5 mRNA value was chosen to represent the patient For further details see text. Figure 5. Cumulative survival curves according to Kaplan-Meier for CRC patients (n = 166). Patients are classified in groups (-1, 0, +1, and +2) based on the mRNA value of the biomarkers SLC35D3, KLK6, MUC2 and CEACAM5 and calculated according to formula: [Formula E = SLC35D3/CEACAM5 + KLK6/CEACAM5 - MUC2/CEACAM5). For further details see legend to Figure 4 and text

EXAMPLES

Identification of genes and gene signatures that are significantly correlated to risk of cancer death in colorectal cancer subjects

A gene that is of importance for tumor progression is most likely expressed both in the primary tumor tissue and in secondary tumors present in a regional lymph node draining the intestine. A microarray-search for progression markers was performed by analyzing RNA from 4 different H&E positive lymph nodes (i.e. tumor cell containing lymph nodes) of 4 patients with stage III CRC plus 3 primary tumors from 3 of these patients. RNA from 7 control patients Pymph nodes from 2 ulcerative colitis patients, 1 Crohns' colitis patient, 1 colon lipoma patient and 3 normal colon epithelial cells samples) were also analyzed. CRC samples were compared individually relative to all control samples as one group. The microarray data were filtered by setting statistical significance to P < 0.05, fold change to≥ 5, and minimum intensity to 15. In this way a number of genes that were expressed in most of the CRC samples (≥ 5/7) with a fold change≥ 5 times were identified. Among these were SLC35D3, POSTN and KLK6.

Commercially available real-time qRT-PCR assays were used to verify the microarray results (TaqMan Gene Expression Assays) for POSTN, SLC35D3 and KLK6. In the latter case 3 assays for different splice forms. All three genes were expressed in a panel of primary CRC tumors samples (n = 8) while SLC35D3 and KLK6 but not POSTN were expressed in all CRC cell lines (n = 5).

Real-time qRT-PCR assays with RNA copy standards using the Taqman EZ RT-PCR technology as described (Fahlgren A. et al. Clin. Exp. Immunol., vol. 131, p. 90-101 (2003); Ohlsson L. Thesis, ISBN 978-91-7459-318-1 (2011)) were constructed. Primer and probe sequences for real-time qRT-PCR assays for SLC35D3, POSTN, KLK6, MUC2, and CEA, mRNA are shown in Table 1 and primers for construction of RNA copy standards in Table 2. Using these assays a panel of RNA samples including primary CRC tumors, normal colon tissue and purified colon epithelial cells, CRC cell lines, peripheral blood mononuclear cells (PBMCs), different immune cell lines and a fibroblast cell line were analyzed (Table 3). The individual values of 56 primary CRC tumors and 5 normal colon samples are shown in Figure 1. For comparison the result for CEACAM5 and MUC2 is included (Ohlsson L. Thesis, ISBN 978-91-7459-318-1 (2011)).

It is apparent that all five biomarker mRNAs are expressed in primary CRC tumors, although at highly different copy number levels normalized to the 18S rRNA content in the sample, from a median of 164 for CEACAM5 to 0.17 for SLC35D3 reflecting the abundance of the protein molecule that the particular mRNA is coding for. Secondly, that the CRC cell lines express all marker mRNAs except POSTN, which instead is expressed at high levels in the fibroblast cell line.

Thirdly, that none of the markers is expressed to a significant degree in immune cell lines.

Fourthly, that CEACAM5 is expressed at similar levels in primary CRC tumors and normal colon epithelial cells. Based on the latter finding and previous knowledge (Ohlsson L. et al. Br. J. Cancer, vol. 95, p. 218-225 (2006)), CEACAM5 was considered to be the preferable marker for cells originating from the large intestine. Moreover, its high expression level makes it a very sensitive marker for detection of colorectal cancer cells in lymph nodes.

Fifthly, MUC2 measures to what extent a CRC tumor is mucinous or not, MUC2 being the dominant mucin in colon and rectum. Patients with mucinous tumors have a better prognosis than those with non-mucinous tumors (Byrd ]. C. and Bresalier R. S. Cancer Metastasis Review, vol. 23, p. 77-99 (204); Ohlsson L. et al. Int. J. Cancer, vol. 130, p. 1833-1843 (2012)).

Finally, in contrast to CEACAM5 and MUC2, both KLK6 and SLC35D3 were expressed in CRC tumors and most CRC lines but not in normal colon epithelial cells, i.e. they are ectopically expressed in CRC tumors.

Application of combined biomarker mRNA analysis for predicting probability of CRC-death

A clinical material of lymph nodes from 166 surgically treated patients with CRC representing all four TNM stages and with known CEACAM5 mRNA, KLK6 mRNA, MUC2 mRNA and 18S rRNA expression levels was analyzed for expression levels of SLC35D3 mRNA and POSTN mRNA. In total mRNA from more than 600 lymph nodes were analyzed. The mRNA values were normalized against 18S rRNA and expressed as mRNA copies/18S rRNA unit Previous studies by our group have demonstrated that 18S rRNA is an excellent RNA species for normalization (Bas A. et al, Scand. ]. Immunol., vol. 59, p. 566-573 (2004); Ohlsson L. et al. Int. J. Cancer, vol. 130, p. 1833-1843 [2012)). The node with the highest mRNA expression level was used to represent the patient in further analysis. This is in analogy with the present clinical practice that H&E positive nodes are considered informative, while H&E negative nodes are considered non-informative except in the case when all nodes are negative. Figure 2 shows the result. The figure also shows the mRNA values for lymph nodes from non-CRC control patients and the dashed line indicates the highest value of this control group. Lymph nodes from stage III and IV patients displayed a larger fraction of nodes with mRNA values above the cut-off level than nodes from stage I or II patients [SLC35D3: stage I = 18%, stage II = 9%, stage III = 25%, and stage IV = 79%; POSTN: stage I =25%, stage II = 13%, stage III = 32%, and stage IV = 69%].

The results from analysis of SLC35D3- and POSTN mRNA expression levels were used in combination with the known expression levels for CEACAM5-, KLK6- and MUC2 mRNA in the same nodes of the CRC patients (Ohlsson L. et al. Br. J. Cancer, vol. 95, p. 218-225 (2006); Ohlsson L. Thesis, ISBN 978-91-7459-318-1 (2011); Ohlsson L. et al. Br. J.

Cancer, vol. 107, p. 150-157 (2012)). Cut-off levels were determined for the 5

biomarkers as follows: The patients were ranked according to the biomarker expression level in the highest lymph node and then divided into five groups of equal number of patients. The groups were compared with respect to disease-free survival using Cox regression analysis. From this analysis, the cut-off level was defined as the mRNA expression level at the 8 ^th decile because, for all five markers, the groups below the 8 ^th decile did not differ significantly in disease-free survival. Patients who died from causes other than CRC were considered as disease-free. Patients were divided into two groups, mRNA expression value above and mRNA expression value below the cut-off and for each group the mean survival time after surgery was calculated by cumulative survival analysis according to Kaplan-Meier and risk for recurrence of CRC estimated according to univariate Cox regression analysis. The result for the five biomarkers is shown in Table 4. As can be seen, mRNA values above the cut-off levels for all of them were correlated with poorer prognosis with highly significant P-values.

Table 4. Comparative analysis of average survival time and risk for recurrence of disease of CRC patients with biomarker (+) or biomarker (-) lymph nodes

Determining the levels for all five biomarkers and combining the different

measurements achieves further differentiation of the patient groups with respect to survival. In one embodiment of the invention the combined information derived from the biomarker analysis to predict survival after surgery is used as follows: For each highest lymph node the values for the biomarkers, SLC35D3, KLK6 and MUC2 is first divided by their corresponding CEACAM5 value. For SLC35D3/CEACAM5 and

KLK6/CEACAM5 the ratios were then referred to one of two groups >0.00001 or < 0.00001 (Figure 3). The former group was assigned a value of 1 and the latter a value of 0. For MUC2/CEACAM5 the division was achieved at a ratio of 3.0 assigning nodes with values above 3.0 a value of 1 and below a value of 0 (Figure 3). For POSTN the

POSTN/18S rRNA ratio and the clinical cut-off (8.0 copies/18S rRNA unit; Figure 2) were used to achieve the two groups, assigning values of 1 above the clinical cut-off and 0 for below. A formula, (Formula A: SLC35D3/CEACAM5 + KLK6/CEACAM5 +

POSTN/18S rRNA -MUC2/CEACAM5) was used to classify each patient into one of five groups (formula result: -1, 0, +1, +2, +3) and cumulative survival analysis according to Kaplan-Meier was performed on these groups. The result is shown in Figure 4. Five different curves are obtained. Group (-1) and (0) show good 3 and 5 years survival, group (+1) and (+2), relatively poor survival and group (+3) very poor survival

(Table 5). The risk ratios calculated according to univariate Cox regression analysis, for groups (0), (+1), (+2) and (+3) in comparison to group (-1), is shown in Table 6.

In other embodiments of the invention the biomarker mRNA measurements were calculated in the same way as in Formula A except that in these formulas, e.g. Formula B to Formula E, one of the terms was systematically excluded. Figure 5 shows the cumulative survival according to Kaplan-Meier calculated according to formula E.

Table 5 summarizes 3 and 5 years survival for biomarker mRNA measurements as determined by the 5 formulas (Formula A to Formula E) and Table 6 summarizes the hazards ratios for biomarker mRNA measurements as determined by Formula A, B and C. Although useful information with respect to survival after surgery is generated by biomarker data treated according to formula B to formula E it is clear that treating the biomarker mRNA data according to Formula A is the most informative demonstrating that all of these biomarkers contribute to the result

A kit for determination of biomarker mRNAs

The invention also includes a kit for analysis of biomarker mRNA and 18S rRNA and transformation of raw data to clinically useful information as illustrated by formulas Formula A to Formula E.

In one embodiment of the invention the particular forward and reverse primers as well as probe sequences given in Table 1 are used in real-time quantitative RT-PCR.

Quantitation is achieved by using specific copy standards [RNA) and 3' primers for reverse transcription with biomarker mRNA values normalized to content of 18S rRNA and /or content of CEACAM5 mRNA in the sample. Normalized values are allocated to one of two groups, either (1 =high risk for recurrence) and (0 = low risk for recurrence) according to the biomarker level with cut-off levels determined from analysis of a clinical material of lymph nodes from surgically treated CRC patients. Using a

specifically designed algorithm the (1) and (0) values for each biomarker is transformed to an estimate of relative risk of cancer death, with a range -1, 0, +1, +2, +3, where -1 stands for the lowest risk and +3 for the highest risk, based on the formula:

SLC35D3/CEACAM5 + KLK6/CEACAM5 + P0STN/18S rRNA -MUC2/CEACAM5.

In the embodiments of the invention as exemplified in Figures 4 and 5 and Tables 5 and 6 preferably only the information from the lymph node with the highest biomarker mRNA is of value. However, a number of patients have more than one lymph node harboring tumor cells. The methods according to the invention can also be used in this case, i.e. for differentiation between Nl and N2 stage patients, adding prognostic value.

EXPERIMENTAL METHODS

General methods

Bioinformatics Analysis - Results from microarray gene expression analysis were analyzed by using Illumina Beadstudio software [version 3.3) for direct hybridization assays. Intensity data were normalized by Beadstudios cubic spline algorithm with subtracted background. Significant difference in expression was calculated using Beadstudio software Error Model Illumina Custom with multiple testing corrections using Benjamini and Hochberg False Discovery Rate (Reiner A. et al. Bioinformatics, vol. 19, p. 368-375 (2003)). Difference in gene expression was calculated as fold change, dividing the signal in the CRC samples of interest over average signal of controls.

Cell lines and peripheral blood mononuclear cells - The following established human cell lines were utilized: LS174T, HT29, T84, HCT8 and CaCo2 (all colon carcinomas), Jurkat and Molt-4 (T-cell lymphomas), B6 and KR4 (EBV-transformed B cell lines), U266 (plasmacytoma), U937 (monocyte-like cell line), K562 (erythroblastoid cell line), HL60 (granulocyte cell line), FSU (fibroblast cell line). Peripheral blood mononuclear cells (PBMCs) were isolated from peripheral blood of healthy adults by Ficoll-Isopaque gradient centrifugation. PBMCs were activated in vitro by incubation with the OKT3 monoclonal antibody (50 ng/ml) in HEPES-buffered RPMI 1640 supplemented with 0.4% human serum albumin. PBMCs from seven individuals were incubated with the stimulus in parallel cultures for 4, 7, 20, 48 and 72 hours, washed, pooled and RNA extracted.

Clinical characteristics of the CRC patients and Controls - Surgery for treatment of CRC was carried out in 166 patients [81 men, 85 women, median age 72, (range 42-90) years]. Thirteen of the tumors were located in rectum and 153 in the colon. Seven of the rectal cancer patients received 25 Gy of preoperative radiotherapy. A locally radical tumor resection was carried out in all patients. The tumor differentiation grade was poor, moderate and high in 11, 145 and 10 tumors, respectively. Routine hematoxylin and eosin (H&E) staining was performed on 2,351 lymph nodes, giving a median of 13 (range 1-51) nodes per patient According to the TNM classification, 30 patients were in stage I (T1-2N0M0), 74 in stage II (T3-4N0M0), 46 in stage III (anyTNl-2M0) and 16 in stage IV (anyTanyNMl). Thirty-four patients (4 in stage II, 19 in stage III and 11 in stage IV) received chemotherapy after surgery. The median follow-up time was 75 (range 33- 147) months and no patient was lost at follow-up.

Controls included 18 men and 5 women [median age 25 years, (range 10-61)] undergoing surgery for ulcerative colitis (n=18), Crohn's colitis (n=3), rectal prolapse (n=l), and colon lipoma (n=l). Informed consent was obtained from the patients and in one case his parents. The Research Ethics Committee of the Medical Faculty, Umea University, Sweden approved the study.

Primary and distant CRC tumor and normal colon tissue - One hundred and thirteen samples from 85 primary CRC tumors were analyzed for biomarker mRNA levels (22 samples from 16 stage I patients, 44 samples from 35 stage II patients, 41 samples from 25 stage II patients, and 8 samples from 8 stage IV patients). Primary tumor stage distribution (pTl-pT4) was 2, 14, 55 and 13 respectively. The differentiation grade was poor in 11 tumors, moderate in 71 tumors and high in 3 tumors. One to four samples, approximately 0.5 x 0.5 x 0.5 cm in size, were collected from primary tumor specimens immediately after resection, snap-frozen, and stored at -70°C until RNA extraction. Six normal colon samples, retrieved from the proximal or distal resection margin and two distant liver metastasis samples were collected and treated in the same way as the primary CRC tumors.

Epithelial cells from colon tissue - Colonic epithelial cells (ECs) were isolated from the normal colon mucosa at the resection margins as described (Fahlgren A. et al. Clin. Exp. Immunol., vol. 131, p. 90-101 (2003)).

Lymph nodes - Lymph nodes were retrieved from the resected specimens and bisected with separate, sterile knives. One half of each node was fixed in 10% buffered formalin for routine H&E-staining. The other half was snap frozen in liquid nitrogen and stored at -70°C until RNA extraction. From CRC patients, 503 lymph nodes (91, 253, 107 and 52 nodes from stage I-IV patients, respectively) were collected. A median of 2 (range 1-15) lymph nodes was obtained per patient

From control patients, 108 lymph nodes (82, 9, 13 and 4 nodes from ulcerative colitis, Crohn's colitis, colon lipoma and rectal prolapse patients, respectively) were collected.

RNA isolation - Total RNA was extracted from lymph nodes, normal and tumor colon tissues, colon epithelial cells, PBMCs and cell lines using the acid guanidine phenol chloroform method (Chomczynski P and Sacchi N. Analyt Biochem., vol. 162, p. 156-159 (1987)] by adding 0.5 ml of a solution containing 4 M guanidinium thiocyanate, 25 mM sodium citrate (pH 7.0), 0.5% sarcosyl and 0.1 M 2-mercaptoethanol per 25 mg tissue and up to 2.5 x 10 ⁶ cells in the first homogenization step. Extracted RNA was dissolved in RNAse-free water containing the RNAse inhibitor RNAsin (lU/μΙ; Promega, Madison, WI). The RNA concentration was measured in a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies) and for bead microarray analysis the integrity of the RNA was analyzed in a 2100 Bioanalyzer using an RNA nano assay (Agilent Technologies).

Preparation of RNA copy standards - Total RNA from a primary CRC tumor, two lymph nodes from two patients with CRC and the colon carcinoma cell lines LS174T and T84 were used as starting material for copy standard preparations. The primers used for RT- PCR are given in Table 2. The PCR products, which include the respective sequences amplified in quantitative RT-PCR, were cloned, sequenced and used as template for in vitro transcription with T7 polymerase/RiboProbe In Vitro Transcription Systems (Promega). Linearized DNA, 3-7μg was used in large-scale synthesis reactions carried out at 37°C for 2-3 hr. The reaction products were then treated with 1 U/μg of RNase- free DNase (Promega) for 30-40 min at 37°C followed by extraction with phenol:

chloroform: isoamylalcohol (25:24:1) and chloroform: isoamylalcohol (24:1). RNA was precipitated with 2.5 volumes of 99.5% ethanol and 0.5 volumes of 7.5 M ammonium acetate at -70°C for at least one hr. DNase treatment was repeated at least twice. Finally the copy standards were checked by RT-PCR and PCR to evaluate the content of DNA, which proved to be less than 0.2% for all of them. Concentration of the transcripts was calculated on the basis of the OD260 value, the molecular weight of the transcript and Avogadro's number. The standards were finally diluted to 10 ⁸ copies/ ul.

Real-time qRT-PCR - Real-time qRT-PCR assays with RNA copy standards were constructed for SLC35D3, POSTN, KLK6, CEACAM5, and MUC2 using the Tacman EZ RT- PCR technology (Applied Biosystems Foster City Ca). Primer and probe sequences are shown in Table 1. The RT-PCR profile was 49° C for 2 min, 59° C for 30 min, 94° C for 5 min, followed by 45 cycles of 93° C for 20 sec and 61° C for 1 min. Serial dilutions of the respective RNA copy standard at concentrations from 10 ³ to 10 ⁸ copies/ul were included in each analysis. All qRT-PCR analyses were carried out in triplicates. Emission from released reporter dye was monitored by the ABI Prism 7900 Sequence Detection System (Perkin-Elmer, Wellesley, Ma). For normalization of mRNA levels, the concentration of 18S rRNA was determined in each sample by real-time qRT-PCR according to the manufacturer's protocol (Applied Biosystems) or by use of primers and probe given in Table 1 (SEQ ID NO: 16-18) and copy standard prepared by using the primers given in Table 2 (SEQ ID NO: 26,17). Results were expressed as mRNA copies per unit of 18S rRNA or as RNA copies per copy of 18S rRNA in both cases yielding directly comparable levels of biomarkers.

Statistical Analysis - Differences in disease free survival and risk for recurrent disease after surgery between patients groups were calculated according to Kaplan-Meier survival model in combination with the log rank test and univariate Cox regression analysis. Differences in survival time and hazards ratios with a P value <0.05 were considered to be statistically significant. The software utilized was SPSS (version 18).