INTERNATIONAL PATENT APPLICATION

UNDER THE PATENT COOPERATION TREATY

To all whom it may concern:

Be it known that Gloria Huei-Ting Su and Wanglong Qiu

has invented certain new and useful improvements in

DETECTING MUTATED GENE SEQUENCES BY MUTANT-ENRICHED SEQUENCING

of which the following is a full, clear and exact description.

DETECTING MUTATED GENE SEQUENCES BY MUTANT-ENRICHED

SEQUENCING

[0001] This application claims the priority of U.S. Serial No. 60/940,904, filed May 30, 2007. The entire contents and disclosures of the preceding application is incorporated by reference into this application.

[0002] Throughout this application, various references or publications are cited. Disclosures of these references or publications in their entireties are hereby incorporated by reference into this application in order to more fully describe the state of the art to which this invention pertains.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH

[0003] The present invention is supported in part by fund from the U.S. government (NCI Temin Award CA95434 and the NCI ROl CA109525); therefore, the U.S. government has certain right in this invention.

BACKGROUND AND SIGNIFICANCE

[0004] HNSCC and its genetic profile. Nine-five percents of head and neck tumors are squamous cell carcinomas (HNSCCs). HNSCCs are defined as cancer that originates from the cuboidal cells along the basement membrane of the mucosa. The disease is characterized by local tumor aggressiveness, early recurrence, and a high frequency of a second primary tumor. Patients with HNSCC often face a poor prognosis secondary to a late presentation and ineffective therapies. The 5-year survival rate is 40%. The annual incidence of HNSCC is estimated at 41,000 cases and 12,300 associated deaths in the United States (16), and an estimated 270,500 deaths annually worldwide (17). Despite advances in the management of HNSCC, overall survival has hardly improved for the last 25 years.

[0005] The genetic profile of HNSCC is a work in progress (Fig 1). We know that pi 6 is inactivated in -80%, p53 is inactivated in -50%, and Cyclin Dl is amplified in -30-60% of HNSCC (1, 18). Although some candidate genes have been named for the frequent LOH or amplification observed in HNSCC, such as Rb (Chr 13q LOH), p21, TNF (Chr 6p amplification), MIN (Chr 4q LOH) ₁ and PTEN (Chr 1Oq LOH), their low mutation frequencies do not seem to account for all the loss or gain on the chromosomes. In addition, we are still not sure what contributes to the selective LOH on Chr3p (67%), 8p (40%), 14q (44%)(1, 18). In search of novel oncogenes and tumor-suppressor genes responsible for the tumorigenesis of

HNSCC, we recently added PIK3CA, STKlULKBl, SMAD2, and SMAD4 to the growing list

(13, 19, 20).

[0006] Oncogene PIK3CA and its Hot-spot Mutations. The phosphatidylinositol 3-kinase (PBK) signaling pathway regulates many normal cellular processes, such as cell proliferation, survival, and apoptosis (21-23). Dysregulation of this pathway and aberrant changes to the genetic components include AKT ₉ PTEN, and PIK3CA transformations which have been associated with cancer development (4-6, 24-29) (Fig 2).

[0007] PIKSCA is located on chromosome 3q26.32 and encodes for the catalytic subunit pi 10a of class IA PI3 -kinase. It has been implicated to function as an oncogene in human cancer both because increased kinase levels and genomic amplifications have been appreciated in tumor samples (4-6, 27-29). Recently high frequencies of somatic mutations in the PIK3CA gene have been reported in several human cancer types, including colon, brain, stomach, breast, and ovary (3, 9-11, 30, 31). More than 80% of these mutations are clustered in the helical (exon 9) and kinase domains (exon 20) of the gene (3) (Fig 3). The three most frequently reported mutation hot spots in PIK3CA, named E542K, E545K and Hl 047R, have been shown to elevate the lipid kinase activity of PIK3CA and lead to the activation of its downstream Akt signaling pathway (3, 7). Theses hot-spot mutations account for 78.6% of all reported PIK3CA mutations and are completely oncogenic in vivo (2, 14). Interestingly, PIK3CA mutations and PTEN loss are nearly mutually exclusive, suggesting that the homeostasis of phosphatidylinositoI-3,4,5- triphosphate is regulated by both PIK3CA and PTEN, both of which are critical to carcinogenesis (32). This conclusion further supports the importance of the PI3K pathway in the tumorigenesis of many cancer types. Of note, no mutation has ever been detected in other members of the PI3K gene family (3).

SUMMARY OF THE INVENTION

[0008] The present invention provides a method of identifying mutated gene sequences, comprising the steps of: obtaining a sample comprising mutated and wild-type gene sequences; conducting a first round of PCR; treating the sample with a restriction enzyme that would digest the wild-type gene sequences but not the mutated gene sequences; conducting a second round of PCR to amplify the mutated gene sequences but not the wild-type gene sequences; and performing sequencing to identify the mutated gene sequences.

[0009] The present invention also provides a method of identifying mutated gene sequences in the PIK3CA gene encoding catalytic subunit pi 10a of phosphatidylinositol 3-kinase, comprising the steps of: obtaining a sample comprising mutated and wild-type gene sequences; conducting a first round of PCR; treating the sample with a restriction enzyme that would digest the wild-type gene sequences but not the mutated gene sequences; conducting a second round of PCR to amplify the mutated gene sequences but not the wild-type gene sequences; and performing sequencing to identify the mutated gene sequences.

[0010] The present invention also provides a method of identifying mutated gene sequence in the PIKiCA gene encoding catalytic subunit pi 10a of phosphatidylinositol 3-kinase, comprising the steps of: obtaining a sample comprising mutated and wild-type gene sequences; conducting a first round of PCR, wherein primers used in the PCR introduce one or more unique restriction enzyme site into the wild-type gene sequence; treating the sample with a restriction enzyme that would digest the wild-type gene sequence but not the mutated gene sequence; conducting a second round of PCR to amplify the mutated gene sequence but not the wild-type gene sequence; and performing sequencing to identify the mutated gene sequence.

BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 shows the genetic profile of HNSCC tumorigenesis (1).

[0012] Figure 2 shows the signaling pathway of PI3K (2). PI3K signaling involves PTEN, Akt, and mTOR.

[0013] Figure 3 shows E545K and H1047R are the two most frequent hot-spot mutations of PlKiCA. Figure 3A shows more than 80% of the PIK3CA mutations have been detected in its exons 9 and 20 (3). Figure 3B shows the three hot-spot mutations, E542K, E545K and H1047R, account for 78.6% of all PIK3CA mutations reported (2).

[0014] Figure 4 shows target therapies in development for PIK3CA and its signaling pathway. Many target therapies are in clinical trials that target EGFR, Akt, or mTOR. These are potential therapeutic drugs, whose treatment outcome can be impacted by the activities and genetic status of PlKiCA. Evidence suggests that it would be beneficial for a patients' tumor to be tested for PlKiCA and PTEN mutations, before enrolling the patient in a target therapy for EGFR, Akt, or mTOR.

[0015] Figure 5 shows the schematic of the mutant-enriched sequencing methods for detecting

PIK3CA mutations, H1047R, E545K and E542K. Figure 5A shows the protocol for detecting PIK3CA mutation H1047R. Enzyme BsaBI (GATNNNNATC) specifically cuts the wild-type sequences of the exon 20, but not the mutant copies with A3140G nucleotide alteration. After digesting the first PCR product with the enzyme BsaBI, the second PCR selectively amplifies the mutant copies. Figure 5B shows a unique restriction enzyme site Hpy 1881 (TCNGA) was introduced by mismatch PCR for detecting PIKiCA mutation E545K. The mismatch primer (PIK-E9MF: TCTACACGAGATCCTCTCTCTGTAATCTC) has two A→ T nucleotide substitutions in the forward primer to create a unique enzyme site Hpy 1881 (TCNGA) in the wild-type sequences of the PIKiCA exon 9, but not the mutant sequences. Figure 5C shows the mismatch primer PIK-2E9MR was designed to create a unique restriction enzyme site EcoRI for enriching PIK3CA mutation E542K (G1624A) and E542G (A1625G) with the similar strategy for the other hot-spot mutation E545K.

[0016] Figure 6 shows high-risk patients will be consented to contribute saliva and sputum specimens at the time of clinical examinations. The results of the PIKiCA mutation analyses will not affect patient care decisions. However, we will investigate whether there is a correlation between PIK3CA mutation status and the needs for CT scan and/or biopsy.

[0017] Figure 7 shows the detection of PIKiCA hot-spot mutations by mutant-enriched sequencing. Figure 7al-a2 show the sensitivity of the mutant-enriched sequencing protocol for the exon 20 H1047R mutation was investigated in head and neck cell line Detroit 562, whose genome had been reported to harbor a H1047R mutation. Figure 7al shows both wild-type A and mutant G peaks were detected at 1 : 1 ratio as expected by conventional genomic sequencing of the cell line Detroit 562 DNA (nondiluted). Figure 7a2 shows when the ratio of mutant and wild-type DNA reached 1:360, the mutant G peak was still the only peak detected in the cell line Detroit 562 DNA by mutant-enriched sequencing.

[0018] Figure 7bl-b2 show a patient sample with a known E545K mutation was used to test the mutant-enriched sequencing protocol for the exon 9 E545K mutation. Figure 7bl shows antisense sequencing of the mutated site by the conventional genomic sequencing (marked by the black arrow) detected both the wild-type and the mutant alleles. Figure 7b2 shows the peak representing the wild-type allele in the same sample disappeared when the mutant-enriched sequencing method was applied (indicated by the black arrow). The red arrow marks the

nucleotide Al 63 OT change that was introduced by mismatch primer (PIK-E9MF) to generate the restriction enzyme Hpyl88I (TCNGA) site.

[0019] Figure 7cl-c4 show the mutant-enriched sequencing method identified an undetected PIK3CA hotspot mutation E542K (Gl 624A) by the conventional sequencing. Figure 7cl shows forward sequencing of a clinical sample with a known E542K mutation by the conventional genomic sequencing method (marked by the black arrow) displayed a dominant presence of the wild-type allele over the mutant allele. Figure 7c2 shows the wild-type allele in the same sample disappeared when the mutant-enriched sequencing method was applied (indicated by the black arrow). The red arrow marks the nucleotide A1627T change that was introduced by mismatch primer (PIK-2E9MF) to generate a unique restriction enzyme EcoRI (GAATTC) site. The case of pharyngeal cancer that had been tested negative for a mutation by the conventional genomic sequencing method (Figure 7c3), but was subsequently identified with a PIK3CA E542K mutation using the mutant-enriched sequencing method (Figure 7c4).

DETAILED DESCRIPTION OF THE INVENTION

[0020] The present invention provides a method of identifying mutated gene sequences, comprising the steps of: obtaining a sample comprising mutated and wild-type gene sequences; conducting a first round of PCR; treating the sample with a restriction enzyme that would digest the wild-type gene sequences but not the mutated gene sequences; conducting a second round of PCR to amplify the mutated gene sequences but not the wild-type gene sequences; and performing sequencing to identify the mutated gene sequences. In general, samples comprising mutated and wild-type gene sequences can be, but are not limited to, tissue swab sample, saliva sample, biopsy sample, pancreatic duct juice, blood, bodily fluid, and surgically resected samples. In one embodiment, the samples can be obtained from a subject having head and neck cancer, breast cancer, liver cancer, lung cancer, colorectal cancer, brain cancer, gastric cancer, or ovarian cancer. In another embodiment, the samples are obtained from a subject having a disease or abnormality comprising mutated sequences in the PIKSCA gene.

[0021] In one embodiment, the first and second rounds of PCR in the above method are performed with one or more primers comprising DNA sequences that anneal to a desired gene locus with an annealing temperature between 45-85 ⁰ C and aim to amplify exonic sequences of the desired gene. In another embodiment, the primers that bind to coding or non-coding sequences inside the desired locus are from about 10-bp long to about 100-bp long, and can amplify regions from about 20-bp to 20-kb inside the desired locus. In yet another embodiment,

the first round of PCR is performed with primers that introduce one or more unique restriction enzyme site into the wild-type gene sequences so that only the wild-type gene sequences will be digested by the one or more restriction enzymes. This method can detect at least 1 mutated gene sequence among 360 copies of wild-type gene sequence. The sensitivity of this method can be enhanced and amplified by optimizing the restriction enzyme digestion and PCR amplification conditions. For example, the above method can detect 1 mutated gene sequence among 200 copies of wild-type gene sequence, or 1 in 500, or 1 in 750, or 1 in 1000, or 1 in 2000, or 1 in 3000 copies of wild-type gene sequence.

[0022] The present invention also provides a method of identifying mutated gene sequences in the PIK3CA gene encoding catalytic subunit pi 10a of phosphatidylinositol 3 -kinase, comprising the steps of: obtaining a sample comprising mutated and wild-type gene sequences; conducting a first round of PCR; treating the sample with a restriction enzyme that would digest the wild-type gene sequences but not the mutated gene sequences; conducting a second round of PCR to amplify the mutated gene sequences but not the wild-type gene sequences; and performing sequencing to identify the mutated gene sequences. In general, samples comprising mutated and wild-type gene sequences can be, but are not limited to, tissue swab sample, saliva sample, biopsy sample, pancreatic duct juice, blood, bodily fluid, and surgically resected samples. In one embodiment, the samples can be obtained from a subject having head and neck cancer, breast cancer, liver cancer, lung cancer, colorectal cancer, brain cancer, gastric cancer, or ovarian cancer. In another embodiment, the samples are obtained from a subject having a disease or abnormality comprising mutated sequences in the PIK3CA gene.

[0023] In general, primers that bind to coding or non-coding sequences inside the PIK3CA locus are from about 10-bp long to about 100-bp long, and can amplify regions from about 20-bp to

20-kb inside the PIK3CA locus. In one embodiment, the first and second rounds of PCR are performed with one or more primers comprising DNA sequences that anneal to the PIK3CA locus with an annealing temperature between 45-85 ⁰ C and aim to amplify sequence in exon 20 of the PIK3CA gene. In one embodiment, the first round of PCR is performed with primers GACATTTGAGCAAAGACCTGAA and ATCAAACCCTGTTTGCGTTT, whereas the second round of PCR is performed with primers CATTTGCTCCAAACTGACCA and

TGAGCTTTCATTTTCTCAGTTATCTTTTC.

[0024] In one embodiment, the restriction enzyme employed in the above method recognizes wild-type but not mutated sequences of exon 20 of the PIK3CA gene. For example, the

restriction enzyme is BsaBI or any enzyme that is specific for the A3140 and does not digest the

G3140. In one embodiment, the mutated gene sequence detected by the above method is hot- spot mutation A3140G (H 1047R). This method can detect at least 1 mutated gene sequence among 360 copies of wild-type gene sequence. In another embodiment, the sensitivity of this method can be enhanced and amplified by optimizing the restriction enzyme digestion and PCR amplification conditions. For example, the above method can detect 1 mutated gene sequence among 200 copies of wild-type gene sequence, or 1 in 500, or 1 in 750, or 1 in 1000, or 1 in 2000, or 1 in 3000 copies of wild-type gene sequence.

[0025] The present invention also provides a method of identifying mutated gene sequence in the PIK3CA gene encoding catalytic subunit pi 10a of phosphatidylinositol 3-kinase, comprising the steps of: obtaining a sample comprising mutated and wild-type gene sequences; conducting a first round of PCR, wherein primers used in the PCR introduce one or more unique restriction enzyme site into the wild-type gene sequence; treating the sample with a restriction enzyme that would digest the wild-type gene sequence but not the mutated gene sequence; conducting a second round of PCR to amplify the mutated gene sequence but not the wild-type gene sequence; and performing sequencing to identify the mutated gene sequence. Samples comprising mutated and wild-type gene sequences have been described above.

[0026] In one embodiment, the first and second rounds of PCR are performed with one or more primers comprising DNA sequences that anneal to the PIK3CA locus with an annealing temperature between 45-85 ⁰ C and aim to amplify sequence in exon 9 of the PIK3CA gene. In one embodiment, the mutated gene sequence detected by this method is hot-spot mutation G1633A (E545K) or G1624A (E542K). A non hot-spot mutation, E542G (A1625G), can also be detected by this method. In one embodiment, the first round of PCR for detecting mutation E545K is performed with primers TCTACACGAGATCCTCTCTCTGTAATCTC and GCATTTAATGTGCCAACTACCA, wherein the first round of PCR for detecting mutation E542K is performed with primers GATTGGTTCTTTCCTGTCTCTG and CATAGAAAATCTTTCTCCTGCTCAGTGAAT. In another embodiment, the second round of PCR for detecting mutation E545K is performed with primers TCTACACGAGATCCTCTCTCTGTAATCTC and CTGAGATCAGCCAAATTCAGT TATTTTTTC, wherein the second round of PCR for detecting mutation E542K is performed with primers TTGCTTTTTCTGTAAATCATCTGTG and CATAGAAAAT CTTTCTCCTGCTCAGTGAAT.

[0027] In one embodiment, the restriction enzyme employed in the above method recognizes wild-type but not mutated sequences of exon 9 of the PIK3CA gene. For example, the restriction enzyme is Hpy 1881, EcoRl, or any enzymes that can distinguish between wild-type and mutant sequences at E542K and E545K. The sensitivity of this method is the same as that described above.

[0028] In another embodiment, the restriction enzymes and primers described above can be used together so that mutated sequences in both exons 9 and 20 of the PIK3CA gene can be detected.

[0029] The present invention also provides uses of the methods described above for the detection or diagnosis of cancers, diseases or abnormalities comprising mutated sequences in the PIK3CA gene. Cancers or diseases comprising mutated PIK3CA gene have been described herein.

[0030] The present invention also provides kits comprising reagents for the practice of the methods described above. In one embodiment, the kits comprise the primers and restriction enzymes described above for detecting mutated gene sequences in the PIK3CA gene.

[0031] In vivo and in vitro studies have demonstrated that PIK3CA (phosphatidylinositol 3- kinase, catalytic subunit pllOα) functions as an oncogene in various human cancer types (3-8). Moreover, in several of these cancers, including head and neck squamous cell carcinoma (HNSCC), studies have revealed that there is a high frequency of somatic mutations that occur within the PIK3CA gene (3, 9-12) (13). The three most frequent mutations are focused in exons 9 and 20 of the PIK3CA gene. These mutations (H1047R, E545K, and E542K) account for -80% of those within PIK3CA and have subsequently been shown to be oncogenic both in vivo and in vitro (3, 7, 14). Recently, we have developed mutant-enriched sequencing methods that specifically detect the three hot-spot changes in PIK3CA. These protocols are unique in that they enhance the sensitivity with which hot-spot changes are detected (15). The sensitivity of sequencing protocols is especially relevant because traditional mutation analyses require that tissue samples contain a predominance of tumor cells with only a minor contribution from normal tissues. Unfortunately, in the clinical setting, these traditional analyses often prove inadequate because patient specimens often contain a predominance of normal cells with only a minor contribution from tumor tissues. Our novel methods should mitigate this problem by increasing the sensitivity with which frequent hot-spot changes are detected. In this proposal we

attempt to investigate the clinical applicability of these new mutant-enriched sequencing protocols.

Detection of PIK3CA mutation in patients with known HNSCC pathology. [0032] We have shown that our mutant-enriched sequencing methods have a sensitivity up to 0.0028 (1 mutant: 360 wild-type DNA copies) (15). We propose to investigate the sensitivity and specificity of our methods by focusing on patients with pathologically confirmed HNSCC. We will ask affected patients to contribute a salivary specimen, a sputum specimen, and a cotton swab off the tumor mucosa all taken during the course of a routine examination. Additionally, we will ask patients for access to their biopsy and/or surgically resected tissues. Using our mutant-enriched PCR method we will compare mutation results in the samples taken non- invasively with those taken on biopsy (or in surgery). This experiment will provide us with information on the sensitivity and specificity of our methods. If successful, our methods could potentially be used to better diagnose patients who have P1K3CA mutations and eventually to direct those patients with mutations towards a designer therapy that targets the PIK3CA pathway.

Detection of PIK3CA mutation in patients with unknown HNSCC status.

[0033] To investigate whether our mutant-enriched sequencing methods can be a tool for early detection, we will test our methods on clinical samples collected from patients with unknown

HNSCC status. This will include patients who come to clinic a) because of a suspicious clinical symptoms (often primary symptom is a neck mass), with a diagnosis that has not been confirmed by biopsy, b) for follow-up after tumor resection, and c) for follow-up after remission. This high-risk population is chosen, rather than a general healthy population (for whom our early detection assay is ultimately meant for), so we can evaluate the sensitivity and specificity of our methods in a better controlled, short-term experiment. These patients will be asked to contribute a salivary specimen and a sputum specimen during a routine clinical examination. We will correlate the mutant-enrich mutation results from the saliva and sputum with whether a patient has undergone a CT scan or a biopsy. This experiment will inform us as to the utility with which our assay predicts whether a patient will require a CT scan or a biopsy.

Optimization of the mutant-enriched sequencing methods for high-throughput analysis.

[0034] We have shown that our mutant-enriched sequencing method for each hot-spot mutation has a sensitivity up to 0.0028 (1 mutant: 360 wild-type DNA copies) (15), however, currently each method is optimized for an individual hot-spot mutation. Since there are three hot-spot

mutations for PIK3CA, it would be optimal to analyze all three hot-spots in one assay instead of three separate ones. We are proposing to combine the three protocols to create one optimal protocol for high-throughput analysis. The ability to perform high throughput analysis will be critical for clinical applications where time and patient materials are limited. This is a logical next-step and yet a novel approach, since it has not been reported on other mutant-enriched sequencing methods. Optimization will focus on the specificity and sensitivity of the assay to detect mutant DNA in pooled DNA samples with various wild-type: mutant ratios.

Detection of PIK3Cλ mutations in the biopsy and resected tumor samples of breast cancer patients with enhanced sensitivity.

[0035] We propose to investigate the sensitivity and specificity of our optimized assay on clinical samples. We have access to a bank of 131 primary breast tumors at the Columbia University Medical Center (32). It has been reported that 24% of this study cohort (32/131) harbored mutated PIK3Cλ (32). From this study cohort, we will select 50-60 cases that have corresponding resected tumors. Ideally, of the 50-60 cases, 50% will have wild-type PlKiCA and the other 50% will have a known PIK3CA mutation. The PIK3CA genetic status will be examined by both the conventional genomic sequencing methods and our optimized assay in a blinded fashion (without knowing the previous published PIK3CA results of each sample). We hypothesize that our assay will be more sensitive than the conventional sequencing method in the biopsy samples but not in the resected tumors, because the contribution of tumor DNA should be relatively high in resected tumors. Under this premise, the data will be investigated as following: a) investigate the abilities of our assay to detect PIK3CA mutation in samples with little tumor DNA by asking if PIK3CA mutation is more frequently detected by our assay than the conventional sequencing method in biopsy samples; b) investigate the abilities of our assay to detect tumor subclone with PIK3CA mutation by asking if PIK3CA mutation is detected in any resected tumor samples that are not detected by conventional method.

[0036] The present invention provides novel assays that allow for enhanced sensitivity in detecting the most frequent PIK3CA hot-spot mutations (E542K, E545K and H1047R) (15). The three hot-spot mutations account for 78.6% of all reported cancer-causing PIK3CA mutations (2). The assay we have developed for H1047R allows detection at a sensitivity of 1 mutant DNA copy per 360 wild-type DNA copies in a mixed population of wild-type and mutant DNA. These assays can impact cancer patients in several potential aspects: 1) patients diagnosed with PIK3CA mutation can undergo pathway-specific or target therapy 2) PIK3CA

mutation can be utilized as a prognosis and recurrent marker 3) PIK3CA mutation can be used as an early detection marker.

Potential Therapy Targeting PIK3CA [0037] Several studies have shown that PIK3CA and its pathway are potential targets for chemotherapy or radiation therapy (33, 34). As depicted in Fig 1, PI3K can be activated by EGFR ₅ Her-2, and Ras, while Akt and mTOR are downstream mediators of PI3-K. PTEN can reverse the action of PI3K by removing the 3 '-phosphate from the inositol ring of PI3,4P2 and PIP3, thus preventing the activation of downstream molecules like Akt. Emerging evidence demonstrates, that targeted cancer therapies against EGFR (such as Gefitinib/Iressa by AstraZeneca, Erlotinib/Tarceva by OSI Pharmaceuticals, Cetuximab/Erbitux by Imclone, and Lapatinib/Tykerb by GlaxoSmithKline) or Her-2 (CM 033 by Pfizer and Lapatinib/Tykerb by GlaxoSmithKline ) may have had mixed results possibly due to the upregulation or activation of the PI3K pathway. If we can better determine the mutational status of PIK3 CA in a patient's tumor, we can make immediate impacts in the EGFR and Her-2 target therapies by excluding patients with PIK3CA mutations.

[0038] Inhibitors of Akt and mTOR are also available; some are in clinical use, while the others are at various phases of clinical trials (33). Rapamycin (an anti-proliferative inhibitor of mTOR) has been shown to be effective treatment for glioma tumors that are EGFR-dependent and PTEN-deficient (35, 36). Similarly, in clinical trials we can expect to treat patients with PIKiCA mutations with an Akt inhibitor (Perifosine by Keryx Biopharmaceuticals) or an mTOR inhibitor (Rapamycin, RADOO 1/Everolimus by Novartis, CCI-779/Temsirolimus by Wyeth, and AP23573 by Ariad). Ultimately, the genetic status of PIK3CA has tremendous impacts on the success of target therapies for EGFR, Her-2, mTOR, and Akt. Furthermore, it is imperative that we determine the activity of the PI3K pathway (by the combination of mutational analysis of PIK3CA and immunohistochemistry analysis of PTEN) in order for affected patients to reap the full benefits of these targeted therapies.

PIK3CA as a tumor, prognosis, or recurrent marker

[0039] Amplification of chromosome 3q26 is observed in up to 40% of HNSCC and is linked to tumor progression and negatively correlated with clinical outcome (37-39). Gene amplification and overexpression of PJK3CA may be a critical early event of HNSCC tumorigenesis because they are observed in low to moderate dysplasic cases, and their increased frequencies are associated with transition to invasive cancer (40, 41). Mutations of PIK3CA have been reported

in early lesions of breast cancer carcinoma, hepatocellular carcinoma, and gastric carcinoma, also supporting the notion that PIKSCA mutation is an early event in tumorigenesis (32, 42). Given the involvement of PIK3CA in many aspects of tumorigenesis, our ability to detect PIKSCA mutation with enhanced sensitivity will impact patient care in many cancer fronts.

EXAMPLE 1 PIK3Cλ mutation in HNSCC

[0040] Amplification of chromosome 3q26 is frequently observed in HNSCC and is linked to tumor progression and negatively correlated with clinical outcome (37-39). Gene amplification and overexpression of PIKSCA are observed in low to moderate dysplasic cases, but their increased frequencies are associated with transition to invasive cancer (40, 41). To investigate whether PIK3CA activating mutation is a common mechanism involved in the tumorigenesis of HNSCC, we analyzed for genetic alterations of the PIKSCA gene in 38 HNSCC specimens including eight cell lines by direct genomic DNA sequencing. Eight HNSCC cell lines, RPMI 2650, A-253, SW579, Detroit 562, FADU, CAL 27, SCC-15 and SCC-25, were purchased from American Type Culture Collection (Rockville, MD). Thirty frozen primary tumor samples and their corresponding match normal muscle specimens were obtained from the Tumor Bank facility of the Herbert Irving Comprehensive Cancer Center and Department of Otolaryngology/ Head and Neck Surgery of the Columbia University Medical Center. Acquisition of the tissue specimens was approved by the Institutional Review Board of Columbia University Medical Center and performed in accordance with Health Insurance Portability and Accountability Act (HIPAA) regulations. Fresh-frozen tumor samples were dissected to ensure that the specimen contained at least 75% cancer cells. The cancer sites were nasal cavity (2), pharynx (6), larynx (10), oral cavity (8) and other sites (4). The patients' ages ranged from 40 to 85 years, average 64.0 ± 14.5. The grades of the tumors were moderately to poorly differentiated.

[0041] Only exons 1, 4, 5, 6, 7, 9, and 20 of the gene were sequenced because they covered the most common PIKSCA mutations previously observed in human cancer (3, 9-11, 30, 43-45). All samples found to have a genetic alteration in the target were subsequently sequenced in the reverse direction to confirm the mutation using the reverse PCR primers (3). The mutation was then further verified by sequencing of a second PCR product derived independently from the original template.

[0042] Four missense mutations of the PIK3CA gene were identified in the 38 HNSCC specimens (4/38, 11%). Two of the mutations were in the exon 9 (E545K, E542K), one was in the exon 20 (H1047Y) and one was in the exon 4 (Y343C). None of these mutations were

detected in the corresponding normal tissues except for the H 1047 Y mutation, which was identified in HNSCC cell line Detroit 562. Three of the four PIK3CA missense mutations are hot-spot mutations (3). The mutation in the exon 4 nucleotide 1028 A →G, which leads to alteration at codon 343 TAC (Y) -→TGC(C), has not been described before.

[0043] Interestingly, three out of the four cases with mutations are from the same organ site, pharynx (Table 1). Cancer of the pharynx is the 9 ^th most common cancer worldwide (46). It is characterized as the following subsites: posterior pharynx, hypopharynx and lateral pharyngeal walls. A total of six pharyngeal squamous cell carcinoma cases were examined in this study, suggesting that as high as 50% (3/6) of pharyngeal tumor samples may harbor PIK3CA mutations.

TABLE l

Nucleotide Alterations Within The Coding Exons of PIK3CA Identified In 38 HNSCC Specimens

Cases Exon Nucleotide Amino acid Present in Tumor site (number) normal tissue

Detroit 562 20 A3140G H1047R N/A pharynx (1)

102T 9 G1624A E542K no oropharynx (1)

109T 9 G1633A E545K no hypopharynx (1)

182T 4 A1028G Y343C no tongue (1) The nucleotide alterations are described according to the cDNA sequence with GenBank accession number NM 006218.

EXAMPLE 2 PIK3CA Mutation in Pharyngeal Cancer and Mutant-enriched sequencing analysis for PIK3Cλ

[0044] To determine the mutation frequency of PIK3CA in pharyngeal cancer, we studied 24 additional cases of pharyngeal squamous cell carcinoma in a subsequent study. Twenty-four paraffin-embedded blocks of pharyngeal cancer were obtained from the Departments of Otolaryngology/Head and Neck Surgery and Pathology at Columbia University Medical Center. The 24 specimens came from five female and nineteen male patients, with ages ranged from 38 to 78 years-old (average 57.9 ± 12.2 years-old). Of the 24 patients, eleven were heavy smokers (more than 40 packs per year), two of the eleven were also with heavy alcohol drinking, and one of the heavy smokers with cocaine abuse; two were moderate smokers, two had no smoking and had only occasional alcohol use, and the remain nine patients' history were not available. One patient was a HIV carrier. All patients were diagnosed as squamous cell carcinomas of

pharyngeal cancer. The grades of the patients were four well-, 18 moderately-, and two poorly- differentiation.

[0045] Initially we investigated the three hot-spot mutations of PIKiCA (located in exons 9 and 20) using direct sequencing of genomic DNA. We detected two mutations of PIK3CA in 24 laryngeal cancers, consisting of a hotspot mutation E545K and a missense mutation in exon 20 (nucleotide 3127 A→ G, which leads to amino acid code 1043 ATG (met)→GTG (val) substitution). Because most of pharyngeal cancers are not resected, 19 of our specimens were biopsy tissues, which contain high concentrations of normal cells with little tumor tissues. Conventional DNA sequencing method only recognizes the mutant DNA if it is present in more than 10 percent of a mutant/wild-type mixed population in primary tumor tissue (47, 48). Therefore we hypothesized that the low tumor to normal cell ratio in our biopsy specimens may have caused some false negative results and contributed to the lower frequency of PIK3CA mutation observed in the current study (2/24, 8.3%) than our previous report (3/6, 50%) (13). We set out to develop mutant-enriched sequencing methods to selectively amplify and sequence the three frequent hotspot mutations (E542K, E545K and Hl 047R) of PIKBCA in specimens with low tumor to normal cell ratio, such as biopsy samples.

[0046] Mutant-enriched sequencing methods generally involve a first-round PCR, restriction enzyme digest, and a second-round PCR. In contrast to conventional DNA sequencing, mutant- enriched sequencing methods selectively amplifies the mutant copy of a targeting gene by reducing the wild-type copy via a restriction enzyme digestion that is specific for the wild-type DNA after the first round PCR. In the second PCR amplification, only the mutant strands but not the wild-type would be further amplified. Thus, mutant-enriched DNA sequencing is particularly helpful to screen gene mutation in samples with high concentrations of normal cells. In our method of mutant-enriched sequencing for the detection of PIK3CA exon 20 hotspot mutation A3140G (H 1047R), we took advantage of a natural restriction enzyme site on wild-type DNA that can be recognized and digested by enzyme BsaBI (cuts G ATNNNNATC). This enzyme site is destroyed when the DNA is mutated, thus enzyme BsaBI exclusively digests the wild-type PIK3CA exon 20 DNA but not the mutant with the A3140G (H1047R) (Fig.SA).

[0047] The feasibility of our method was first investigated in a head and neck cancer cell line,

Detroit 562, which harbors H1047R mutation. Both mutant and wild-type peaks were observed at 1:1 ratio as expected using conventional genomic sequencing, and the wild-type peak was entirely eliminated when our mutant-enriched sequencing method was applied. To determine the

sensitivity of our assay, we made a series of dilutions of the genomic DNA from mutant cell line

Detroit 562 with DNA from another cell line that has wild-type PIK3CA alleles. When the ratio of mutant and wild-type DNA reach beyond 1: 360, the peak of mutant allele still can be recognized by mutant-enrich sequencing (data not shown). In contrast, with conventional PCR sequencing, the maximum ratio of mutant and wild-type DNA was 1:18. This indicates that mutant-enrich sequencing al least twenty fold more sensitive than traditional direct DNA sequencing.

[0048] Using this mutant-enrich sequencing method, two additional mutations of PIKiCA gene were identified in these 24 pharyngeal cancer (data not shown). This result supported our hypothesis that the low frequency of PIKSCA mutation detected in clinical pharyngeal cancer biopsy samples by conventional DNA sequencing was partially caused by the contamination of normal cells.

[0049] For mutant-enrich sequencing of PIK3CA exon 9 hotspot mutation, G1633A (E545K), mismatch primer (PIK-E9MF: TCTACACGAGATCCTCTCTCTGTAATCTC) was designed to introduce two A→ T nucleotide mismatches in the forward primer to introduce a unique restriction enzyme site Hpy 1881 (TCNGA) in the exon 9 region (Fig. 5B), because there is no natural unique restriction enzyme site to digest the wild-type but not the mutant DNA sequences at this hot-spot. Using a patient's tumor DNA with known E545K mutation (patient No. 24, Table 1), we showed that using our mutant-enrich sequencing method, only the mutant peak remained while the more prominent wild-type peak observed in the conventional sequencing assay disappeared. Applying this powerful method, we screened the 24 pharyngeal cancer samples. We did not uncover additional mutation of PIK3CA at this hot-spot.

[0050] We applied the same mismatch PCR strategy to enrich the mutant allele of PIK3CA hotspot mutation H542K (G1624A). A restriction enzyme EcoRI site was introduced by the mismatch primer PIK-2E9R. EcoRI digestion disrupted the wild-type PIK3CA DNA, but not the mutant of PJK3CA E542K sequences.

[0051] This mutant-enriched sequencing method was tested in a previously reported patient sample with a known PIK3CA E542K mutation (13). We showed that only the mutant peak remained while the corresponding wild-type peak completely vanished when the mutant- enriched sequencing was applied (data not shown). Interestingly, this method can detect not only the E542K (Gl 624A) mutation, but also a previously described non-hot-spot E542G (A1625G)

mutation (3)(data not shown). We screened the 24 biopsy specimens of pharyngeal cancer and an additional case of PlKiCA E542K (G1624A) mutation was identified (data not shown).

[0052] In summary, using the direct genomic DNA sequencing approach, we detected two mutations of PIK3CA in 24 laryngeal cancers. Three additional PIK3CA mutations were detected in the 19 pharyngeal cancer biopsy specimens by mutation-enriched sequencing methods. Mutant-enriched sequencing can identify the H1047R mutant DNA in a mixed population with wild-type DNA at sensitivity of 0.0028 (1 mutant: 360 wild-type DNA copies). Totally five mutations of PIK3CA gene were identified in these 24 pharyngeal cancer specimens (20.8%). Four of the five mutations could have been identified by the mutant-enriched sequencing methods (4/5, 80%), but only two were detectable by the conventional sequencing method (2/5, 40%). The data further confirmed that oncogenic PIKSCA may play a critical role in the pharyngeal carcinogenesis.

EXAMPLE 3

Detection of PIK3Cλ Mutation In Patients With Known HNSCC Pathology

[0053] We have established previously that our mutant-enriched sequencing methods are applicable to biopsy tissues. To investigate the sensitivity and specificity of our mutant- enriched sequencing methods on clinical samples such as saliva and swab, we will collect saliva and swab samples from patients with biopsy-confirmed HNSCC. This will allow us to compare results of PlKiCA mutation analyses of DNA from saliva, swab, and biopsy. The biopsy samples will be our controls.

[0054] We realize that the frequency of P1K3CA in HNSCC is not the highest among all cancers (e.g. colorectal and breast cancers are -30%), however, saliva and sputum collections are least invasive to patients. The ease in collection and the reproducibility of saliva production therefore make HNSCC patients logical persons with whom to test our assay. We have working protocols that allow extraction of DNA from saliva. We have preliminary data that shows that the amount the DNA extracted from saliva is sufficient for our mutant-enriched sequencing analyses.

[0055] Patients with newly diagnosed HNSCC or unresectable HNSCC will be asked to enroll. Consented patients will be asked to contribute a salivary specimen, a sputum specimen, and for a cotton swab of the tumor mucosa at the time of routine examination. In addition, permission to access previously taken biopsy and/or surgical specimens will also be procured from the patients. Patients will be admitted and attended to by the medical residents of the Department of Otolaryngology and Head and Neck Surgery (OTO/HNS) under the supervision of Drs. Lanny

Close and Spiros Manolidis. Specimens can be collected at all OTO/HNS clinical locations, including HP 7 ^th , VC 10 ^th , East 60 ^th St. (Columbia Eastside Practice), BH 5 ^th , Melstein 3 ^rd , and BH OR. Our residents also rotate through the Cornell New York Hospital and have privilege to consent patients there as well. Mr. Tom Karnezis, who is going to be a ^-year medical student of Columbia this July, will serve as the study coordinator during the first year of the study. Mr. Karnezis is currently doing a research year in my lab with external funding provided by the Doris Duke Clinical Research Foundation.

[0056] The participants will be asked not to eat, drink, smoke, or brush their teeth for at least 60 minutes prior to sample collection. The subjects will be swabbed first (a scope is only necessary for cancers outside the oral cavity, oropharynx, hypopharyx, and pharynx. For sites that can not be visualized, a laryngoscope, nasopharyngoscope, or esophagoscope will be used), then asked to swish and gargle with 25 ml of sterile NaCI solution (0.9%) for 3 minutes for saliva collection (49). Swab samples and Saliva samples will be stored at 4°C before transferring to the research lab (unprocessed samples can be stored up to 2 weeks at 4°C). In the lab, samples will be treated in Hank's solution with 1% DTT and centrifuged, and cell pellets will be frozen at -80 ⁰ C before analysis (49). Genomic DNA from exfoliated cells will be digested with proteinase K and extracted with ChargeSwitch gDNA Buccal Cell Kit (Invitrogen) according to the manufacturer's direction. Up to 6μg of genomic DNA can be extracted per sample. The participants will then be asked to provide an induced sputum sample using a variation of the ultrasonic nebulization technique described by Saccomanno et al.(50, 51). Subjects will use water or saline to brush tongue, buccal surfaces, teeth, and gingiva gently to remove superficial epithelial cells and bacteria, followed by gargling and rinsing with tap water. Participants then inhale a nebulized 3% saline solution from an ultrasonic nebulizer for 20 to 30 minutes. Sputum will be collected in a sterile specimen cup and an equal volume of Saccomanno solution was added immediately. Sputum samples will be defined as adequate by the presence of deep lung macrophages or Curschmann's spiral (50) and, irrespective of adequacy, processed for DNA extraction by extensive vortex mixing, washing once with Saccomanno solution, and storage at room temperature until analysis. In addition, at least two Papanicolaou-stained slides underwent morphologic examination by a certified cytopathologist. Genomic DNA was isolated from sputum as by digestion with Pronase in SDS (1%), followed by standard phenol-chloroform extraction and ethanol precipitation (52). Six slides (10 μm thick each) of biopsy or resected tumors will be microdissected and processed into genomic DNA using QIAmp DNA Mini Kit (Qiagen) according to manufacturer's instruction for paraffin-embedded tissues.

[0057] We typically have 11 new cases of HNSCC with confirmed biopsies per month or 132 cases per year. That is approximately 400 cases for the duration of three award years. As described above, saliva, sputum, tumor swab, biopsy, and/or resected tumors will be collected from the participants and processed into genomic DNA for mutation analyses. All the samples will be subject to mutant-enriched sequencing for H1047R and E545K as described previously. Assuming 10% of the 400 cases harbor a PIK3CA mutation, and 67.8% (two hot-spots) of the mutations is a hot-spot mutation, we estimate that we will detect a PIK3CA hot-spot mutation in the biopsy and/or resected tumor specimens of ~27 cases. The number of cases may be higher if pharyngeal cancer predominates the study cohort. If our mutant-enriched sequencing analysis methods are sensitive enough for PIK3CA mutation detection in saliva, sputum, or tumor swab, the same PIK3CA mutation that is identified in the corresponding biopsy and/or resected tumor specimen should be observed in saliva, sputum, or tumor swab. Since biopsy and/or resected tumor will serve as the control, pair-wise comparison will be made between saliva/control, sputum/control, and tumor swab/control. If there is a discrepancy between biopsy and resected tumor when both specimens are available from a same patient, we will uphold the result from the resected tumor as the more reliable control of the two.

[0058] Using one-sample chi-square analysis, we estimate that if 67.8% of our participants have PIK3CA mutation in their biopsies, to detect 50% sensitivity using our mutant-enriched sequencing methods at a power of 80%, we will need 358 participants. We will have approximately that number of patients over three years to conduct a meaningful study.

[0059] For statistical analysis after sample analyses, we will use mean and standard deviation for descriptive analysis of continuous variables and frequencies and percentages for categorical variables. We will use Chi-square test for comparison of categorical variables and t-test for comparison of continuous variables. We will also create a logistic regression model with the dependent variable being the specimen challenge and the independent variable being age, gender, tumor site (oral, pharyngeal, laryngeal, etc), tumor stage (TO-4, N0-3), presence of metastasis (MO-I), and clinical history (smoking and drinking). A "p" value < 0.05 will be conceded statistically significant.

EXAMPLE 4 Detection of PIK3CA Mutation In Patients With Unknown HNSCC Status

[0060] Since PIK3CA mutation is often detected in early precancerous lesions, we would like to investigate whether our mutant-enriched sequencing methods can be a tool for early detection.

For many cancer types, including HNSCC, the lack of early detection often means missed

opportunity for cure and prolonged survival. Any early detection, even at low frequency, as long as it is specific and accurate, will save lives. If a non-invasive early detection method can be developed, such as our methods, it can be prevalently used for HNSCC by enlisting the help of dental profession. The logical step is for us to first find out if our mutation analysis methods are sensitive enough to detect PIKSCA mutation in a high-risk population before we test its sensitivity in a low-risk general population. We intend to test our methods on clinical samples collected from patients who suspect of but not yet unconfirmed with HNSCC. If successful here, we will test our methods on high-risk populations of other cancer types and low-risk general population.

[0061] Recruitment will include patients who come to clinic a) because of suspicious clinical symptoms (whose primary symptom is neck mass), but not yet confirmed by biopsy, b) for follow-up after tumor resection, and c) for follow-up after remission. When a high-risk patient comes to the clinic, he/ she will receive a clinical examination and possibly a CT scan. A biopsy is subsequently taken only if the CT scan is positive (larger than lcm). All high-risk patients will be asked to contribute a salivary specimen and a sputum specimen regardless whether a CT scan or a biopsy is ordered at the time of clinical examination. This will be a double-blind experiment where the clinicians will not know the results of the mutation analyses of the specimens while providing patient care. The scientists will not know the treatment plan of the patient when performing the mutational analyses on the saliva and sputum samples. Only after the patient has been treated, we will correlate the mutation analyses results from saliva and sputum with whether a patient has undergone a CT scan or biopsy (Fig. 6). This experiment will inform us whether our assay is useful in predicting whether a patient requires a CT scan or biopsy. The rationale is to investigate if our methods are sensitive enough to substitute for a clinical examination.

[0062] Our clinics typically see 50 high-risk patients per month, who would fall into one of the three categories described above. Usually when a high-risk patient comes in a clinic, 60% of them may require a CT scan after clinical examination. Of those who receive CT scans, about 50% will need to be followed up with a biopsy. We estimate that we will have a cohort of 1 ,800 high-risk patients over the three-year award period, of those, 1,080 will likely to need a CT scan and 540 will need a biopsy.

[0063] We estimate that about 80% of the patients who undergo biopsy will be diagnosed with HNSCC. Again, about 10% of HNSCC patients have PIK3CA mutation and 67.8% of the

mutations occur in the two hot-spots that we are investigating. That would mean that about

1.63% of all high-risk patients are expected to be positive for our two hot-spot mutations. Using one-sample chi-square test, we estimate that we will need 1,500 participants to detect 50% sensitivity with our methods at the power of 80%. We have enough potential participants over the study period of three years.

[0064] For data analysis after sample collection and mutation analyses, we will determine the statistical significance for the high-risk population of specimen and clinical outcomes. We will also evaluate the sensitivity and specificity for the same groups. Receiver Operating Characteristics (ROC) curves wiil also be used to compare the diagnostic performance of specimen and clinical outcome. The ROC curve will show the true positive rate against the false positive rate for the six different confidence ratings for each case of CT / biopsy. For each specimen a positive or negative rating will be assigned. In addition, the biopsy or CT scan for each subject will be classified according to an ordinal rating scale: No Stenosis =1, Minimal (<20%)=2, Mild (20-50%)=3, Moderate (50-70%)=4 ₅ Severe (>70%)=5, Occlusion=6. We will also create a logistic regression model with the dependent variable being the specimen challenge and the independent variable being age, gender, tumor site (oral, pharyngeal, laryngeal, etc), tumor stage (TO-4, N0-3), presence of metastasis (MO-I), and clinical history (smoking and drinking). A "p" value < 0.05 will be conceded statistically significant.

EXAMPLE 5 Optimization of The Mutant-Enriched Sequencing Methods For High-Throughput

Analysis

[0065] We have developed mutant-enriched sequencing methods to selectively amplify and sequence three most frequent mutations (H1047R, E545K, and E542K) of PIK3CA in specimens with low tumor DNA contribution, such as biopsy samples. Here we propose to combine the three separate protocols for the three hot-spot mutations into one: Multiplex PCR with three primer sets will be performed in the first-round PCR, restriction enzyme digest involving three enzymes, and multiplex PCR with three primer sets will be performed in the second-round PCR. The PCR product from the second-round PCR will be divided and submitted for sequencing with three sequencing primers in three separate reactions. Optimization will involve primer designs, enzyme selection, digestion conditions (e.g. DNA amount, enzyme amount, buffer, temperature, and duration), and PCR protocols (e.g. annealing temperature, number of cycles, and buffer). The sensitivity will be determined again by testing on a series of dilutions with various mutant to wild-type DNA. The combination assay for high-throughput analysis is a novel proposal, because it has not been reported on other mutant-enriched sequencing methods.

[0066] The optimization can also be designed to complement high-throughput sequencers available on the market. For example, primers can be re-designed to include "fusion primers" necessary for amplicon sequencing on the Roche's 454 sequencing (Genome Sequencing FLX System). Or the primers can be re-designed to allow the annealing of "adaptors" required for the Illumina's sequencing technology.

EXAMPLE 6 Detection of PIK3CA mutations In The Biopsy And Resected Tumor Samples of Breast Cancer Patients With Enhanced Sensitivity

[0067] We have previously tested our methods on a set of 5 surgically resected and 19 biopsy pharyngeal tumor samples (15). Using the conventional genomic sequencing method, PIK3CA hot-spot mutations were found in 2/5 resected and 0/19 biopsy tumor samples. Our mutant- enriched PCR sequencing methods identified additional PIK3CA mutations in 4/19 biopsy tumor samples. No additional mutation was identified among the resected tumor samples. This led to the hypothesis that our methods are more sensitive than the conventional sequencing method on biopsy samples but not resected tumors, because the contribution of tumor DNA should be relatively high in resected tumors. In collaborations with Drs. Ramon Parsons and Hanina Hibshoosh, we will select 50-60 cases that have both biopsy and resected tumor paraffin blocks at the Columbia University Medical Center from a previously published 131 breast cancer database (32). Of the 50-60 cases, 50% will have wild-type PIK3CA and the other 50% will have a known PIK3CA mutation. The PIK3CA genetic status will be examined by both the conventional genomic sequencing methods and our optimized assay in a blinded fashion (without knowing the previous published PIK3CA results of each sample).

[0068] Based on our choice of the study cohort, we expect that among the resected tumors, -25 cases (or 50%) will be identified with one of the PIK3CA hot-spot mutations using either method. If the abilities of our assay to detect PIK3CA mutation in samples with little tumor DNA are more superior, then we hypothesize that in biopsy samples, the PIK3CA mutation frequency detected by our assay should be higher than the conventional sequencing method in biopsy samples. Using the pair-wise chi-square test, if we hypothesize that the one of the two methods will also detect 50% mutation rate in the biopsy samples as in their corresponding resected tumors, we will have 80% power to detect mutation rate >78% or <22% with the other method. It's possible that the difference between the two assays is smaller. In that scenario, more cases will be included in a follow-up study to achieve a significant difference.

[0069] If our mutant-enriched assay identifies mutations undetectable by the conventional method in the resected tumors, we will then suspect that multiple tumor clones co-exist in those resected tumors and our assay is sensitive enough to detect presences of tumor subclones. To further investigate the presence of subclones, the DNA of those suspected tumor samples will be PCR-amplified, subcloned into bacterial vectors, and transfected into bacteria to generate bacterial clones. Approximately 100 bacterial clones will be analyzed per tumor sample to confirm the presence of the subclones and determine the percentage of mutant PlKSCA DNA copies.

REFERNECES

L Chin, D., Boyle, G. M., Theile, D. R., Parsons, P. G., and Coman, W. B. Molecular introduction to head and neck cancer (HNSCC) carcinogenesis. British Journal of Plastic Surgery, 57: 595-602, 2004. 2. Bader, A. G., Kang, S., Zhao, L., and Vogt, P. K. Oncogenic PBK deregulates transcription and translation. Nat Rev Cancer, J: 921-929, 2005.

3. Samuels, Y., Wang, Z., Bardelli, A. _s Silliman, N., Ptak, J., Szabo, S., Yan, H., Gazdar, A., Powell, S. M., Riggins, G. J., Willson, J. K., Markowitz, S., Kinzler, K. W. _} Vogelstein, B., and Velculescu, V. E. High frequency of mutations of the PIK3CA gene in human cancers. Science, 304: 554, 2004.

4. Andrew, S. PIK3CA: determining its role in cellular proliferation and ovarian cancer. Clin Genet, 56: 190-191, 1999.

5. Ma, Y. Y., Wei, S. J., Lin, Y. C, Lung, J. C, Chang, T. C, Whang-Peng, J., Liu, J. M., Yang, D. M., Yang, W. K., and Shen, C. Y. PIK3CA as an oncogene in cervical cancer. Oncogene, 19: 2739-2744, 2000.

6. Shayesteh, L., Lu, Y., Kuo, W. L., Baldocchi, R., Godfrey, T., Collins, C, Pinkel, D., Powell, B ₁ , Mills, G. B., and Gray, J. W. PIK3CA is implicated as an oncogene in ovarian cancer. Nat Genet, 21: 99-102, 1999.

7. Kang, S., Bader, A. G., and Vogt, P. K. Phosphatidylinositol 3-kinase mutations identified in human cancer are oncogenic. Proc Natl Acad Sci U S A, 102: 802-807,

2005.

8. Buttitta, F., Felicioni, L., Barassi, F., Martella, C, Paolizzi, D., Fresu, G., Salvatore, S., Cuccurullo, F., Mezzetti, A., Campani, D., and Marchetti, A. PIK3CA mutation and histological type in breast carcinoma: high frequency of mutations in lobular carcinoma. J Pathol, 208: 350-355, 2006.

9. Lee, J. W., Soung, Y. H., Kim, S. Y., Lee, H. W., Park, W. S., Nam, S. W., Kim, S. H., Lee, J. Y., Yoo, N. J., and Lee, S. H. PIK3CA gene is frequently mutated in breast carcinomas and hepatocellular carcinomas. Oncogene, 2004.

10. Campbell, I. G., Russell, S. E., Choong, D. Y., Montgomery, K. G., Ciavarella, M. L., Hooi, C. S., Cristiano, B. E., Pearson, R. B., and Phillips, W. A. Mutation of the P1K3CA gene in ovarian and breast cancer. Cancer Res, 64: 7678-7681, 2004.

11. Broderick, D. K., Di, C, Parrett, T. J., Samuels, Y. R., Cummins, J. M., McLendon, R. E., Fults, D. W., Velculescu, V. E., Bigner, D. D., and Yan, H. Mutations of PIK3CA in anaplastic oligodendrogliomas, high-grade astrocytomas, and medulloblastomas. Cancer Res, 64: 5048-5050, 2004.

12. Schδnleben, F., Qiu, W., Ciau, N. T., Ho, D. J., Li, X., Allendorf, J. D., Remotti, H. E., and Su, G. H. PIK3CA mutations in intraductal papillary mucinous neoplasm/carcinoma of the pancreas. Clin Cancer Res, 12: 3851-3855, 2006.

13. Qiu, W., Schonleben, F., Li, X., Ho, D. J., Close, L. G., Manolidis, S., Bennett, B. P., and Su, G. H. PIK3CA mutations in head and neck squamous cell carcinoma. Clin Cancer Res, 12: 1441-1446, 2006.

14. Bader, A. G., Kang, S., and Vogt, P. K. Cancer-specific mutations in PIK3CA are oncogenic in vivo. Proc Natl Acad Sci U S A, 103: 1475-1479, 2006.

15. Qiu, W., Tong, G.-X., Manolidis, S., Close, L. G., Assaad, A., and Su, G. H. Mutant- enrich sequencing identified high frequency of PIK3CA mutations in pharyngeal cancer., Manuscript in Preparation.

16. Landis, S. H., Murray, T., Bolden, S., and Wingo, P. A. Cancer statistics, 1999. CA Cancer J CHn, 49: 8-31, 3L, 1999.

17. Pisani, P., Parkin, D. M., Bray, F., and Ferlay, J. Estimates of the worldwide mortality from 25 cancers in 1990. Int J Cancer, 83: 18-29., 1999.

18. Forastiere, A., Koch, W., Trotti, A., and Sidransky, D. Head and neck cancer. N Engl J Med, 345: 1890-1900., 2001. 19. Qiu, W., Schonleben, F., Li, X., and Su, G. H. Disruption of transforming growth factor beta-Smad signaling pathway in head and neck squamous cell carcinoma as evidenced by mutations of SMAD2 and SMAD4. Cancer Lett, 245: 163-170, 2007.

20. Qiu, W., Schonleben, F., Thaker, H. M., Goggins, M., and Su, G. H. A novel mutation of STKl 1/LKBl gene leads to the loss of cell growth inhibition in head and neck squamous cell carcinoma. Oncogene, 25: 2937-2942, 2006.

21. Klippel, A., Escobedo, M. A., Wachowicz, M. S., Apell, G., Brown, T. W., Giedlin, M. A., Kavanaugh, W. M., and Williams, L. T. Activation of phosphatidylinositol 3-kinase is sufficient for cell cycle entry and promotes cellular changes characteristic of oncogenic transformation. MoI Cell Biol, 18: 5699-5711, 1998. 22. Chang, H. W., Aoki, M., Fruman, D., Auger, K. R., Bellacosa, A., Tsichlis, P. N.,

Cantley, L. C, Roberts, T. M., and Vogt, P. K. Transformation of chicken cells by the gene encoding the catalytic subunit of PI 3-kinase. Science, 276: 1848-1850, 1997.

23. Kennedy, S. G., Wagner, A. J., Conzen, S. D., Jordan, J., Bellacosa, A., Tsichlis, P. N., and Hay, N. The PI 3-kinase/Akt signaling pathway delivers an anti-apoptotic signal. Genes Dev, 11: 701-713, 1997.

24. Yuan, Z. Q., Sun, M., Feldman, R. I., Wang, G., Ma, X., Jiang, C, Coppola, D., Nicosia, S. V., and Cheng, J. Q. Frequent activation of AKT2 and induction of apoptosis by inhibition of phosphoinositide-3-OH kinase/Akt pathway in human ovarian cancer. Oncogene, 19: 2324-2330, 2000. 25. Bellacosa, A., de Feo, D., Godwin, A. K., Bell, D. W., Cheng, J. Q., Altomare, D. A.,

Wan, M., Dubeau, L., Scambia, G., Masciullo, V., and et al. Molecular alterations of the AKT2 oncogene in ovarian and breast carcinomas. Int J Cancer, 64: 280-285, 1995.

26. Cantley, L. C. and Neel, B. G. New insights into tumor suppression: PTEN suppresses tumor formation by restraining the phosphoinositide 3-kinase/AKT pathway. Proc Natl Acad Sci U S A, 96: 4240-4245, 1999.

27. Phillips, W. A., St Clair, F., Munday, A. D., Thomas, R. J., and Mitchell, C. A. Increased levels of phosphatidylinositol 3-kinase activity in colorectal tumors. Cancer, 83: 41-47, 1998.

28. Benistant, C, Chapuis, H., and Roche, S. A specific function for phosphatidylinositol 3- kinase alpha (p85alpha-pl lOalpha) in cell survival and for phosphatidylinositol 3-kinase beta (p85alpha-pl lObeta) in de novo DNA synthesis of human colon carcinoma cells. Oncogene, 19: 5083-5090, 2000.

29. Pedrero, J. M., Carracedo, D. G., Pinto, C. M., Zapatero, A. H., Rodrigo, J. P., Nieto, C. S., and Gonzalez, M. V. Frequent genetic and biochemical alterations of the PI 3- K/AKT/PTEN pathway in head and neck squamous cell carcinoma. Int J Cancer, 114:

242-248, 2005.

30. Bachman, K. E., Argani, P., Samuels, Y., Silliman, N., Ptak, J., Szabo, S., Konishi, H., Karakas, B., Blair, B. G., Lin, C, Peters, B. A., Velculescu, V. E. _; and Park, B. H. The PIK3CA Gene is Mutated with High Frequency in Human Breast Cancers. Cancer Biol Ther, 3: 772-775, 2004. 31. Wang, Y., Helland, A., Holm, R., Kristensen, G. B., and Borresen-Dale, A. L. PIK3CA mutations in advanced ovarian carcinomas. Hum Mutat, 25: 322, 2005.

32. Saal, L. H., Holm, K., Maurer, M., Memeo, L., Su, T., Wang, X., Yu, J. S., Malmstrom, P. O., Mansukhani, M., Enoksson, J., Hibshoosh, H., Borg, A., and Parsons, R. PIK3CA mutations correlate with hormone receptors, node metastasis, and ERBB2, and are mutually exclusive with PTEN loss in human breast carcinoma. Cancer Res, 65: 2554-

2559, 2005.

33. Rogers, S. J., Box, C, Harrington, K. J., Nutting, C, Rhys-Evans, P., and Eccles, S. A. The phosphoinositide 3 -kinase signalling pathway as a therapeutic target in squamous cell carcinoma of the head and neck. Expert Opin Ther Targets, 9: 769-790., 2005. 34. Kim, D. W., Huamani, J., Fu, A., and Hallahan, D. E. Molecular strategies targeting the host component of cancer to enhance tumor response to radiation therapy. Int J Radiat Oncol Biol Phys, 64: 38-46, 2006.

35. Wang, M., Lu, K., Zhu, S., Dia, E., Vivanco, L, Shackleford, G. ₅ Cavenee, W., Mellinghoff, L, Cloughesy, T., Sawyers, C, and Mischel, P. Mammalian Target of Rapamycin Inhibition Promotes Response to Epidermal Growth Factor Receptor Kinase

Inhibitors in PTEN-Deficient and PTEN-Intact Glioblastoma Cells. Cancer Res, 66: 7864-7869., 2006.

36. Mellinghoff, I. K, Wang, M. Y., Vivanco, I., Haas-Kogan, D. A., Zhu, S., Dia, E. Q., Lu, K. V., Yoshimoto, K., Huang, J. H., Chute, D. J., Riggs, B. L., Horvath, S., Liau, L. M., Cavenee, W. K., Rao, P. N., Beroukhim, R., Peck, T. C, Lee, J. C, Sellers, W. R.,

Stokoe, D., Prados, M., Cloughesy, T. F., Sawyers, C. L., and Mischel, P. S. Molecular determinants of the response of glioblastomas to EGFR kinase inhibitors, [see comment] [erratum appears in N Engl J Med. 2006 Feb 23;354(8):884]. New England Journal of Medicine, 353: 2012-2024, 2005. 37. Hashimoto, Y., Oga, A., Kawauchi, S., Furuya, T., Shimizu, R, Nakano, T., Imate, Y., Yamashita, H., and Sasaki, K. Amplification of 3q26 approximately qter correlates with tumor progression in head and neck squamous cell carcinomas. Cancer Genet Cytogenet, 129: 52-56., 2001.

38. Liehr, T., Ries, J., Wolff, E., Fiedler, W., Dahse, R., Ernst, G., Steininger, H., Koscielny, S., Girod, S., and Gebhart, E. Gain of DNA copy number on chromosomes 3q26-qter and

5pl4-pter is a frequent finding in head and neck squamous cell carcinomas. Int J MoI Med, 2: 173-179., 1998.

39. Singh, B., Stoffel, A., Gogineni, S., Poluri, A., Pfister, D. G., Shaha, A. R., Pathak, A., Bosi, G., Cordon-Cardo, C, Shah, J. P., and Rao, P. H. Amplification of the 3q26.3 locus is associated with progression to invasive cancer and is a negative prognostic factor in head and neck squamous cell carcinomas. Am J Pathol, 161: 365-371., 2002.

40. Woenckhaus, J., Steger, K., Werner, E., Fenic, I., Gamerdinger, U., Dreyer, T., and Stahl, U. Genomic gain of PIK3CA and increased expression of pi lOalpha are associated with progression of dysplasia into invasive squamous cell carcinoma. J Pathol, 198: 335-342, 2002.

41. Estilo, C. L., P, O. C, Ngai, L, Patel, S. G., Reddy, P. G., Dao, S., Shaha, A. R., Kraus, D. H., Boyle, J. O., Wong, R. J., Pfister, D. G., Huryn, J. M., Zlotolow, I. M., Shah, J. P., and Singh, B. The role of novel oncogenes squamous cell carcinoma-related oncogene and phosphatidylinositol 3 -kinase pi lOalpha in squamous cell carcinoma of the oral tongue. Clin Cancer Res, 9: 2300-2306, 2003.

42. Lee, J. W., Soung, Y. H., Kim, S. Y., Lee, H. W., Park, W. S., Nam, S. W., Kim, S. H., Lee, J. Y., Yoo, N. J,, and Lee, S. H. PIK3CA gene is frequently mutated in breast carcinomas and hepatocellular carcinomas. Oncogene, 24: 1477-1480, 2005.

43. Levine, D. A., Bogomolniy, F., Yee, C. J., Lash, A., Barakat, R. R., Borgen, P. L, and Boyd, J. Frequent mutation of the PIK3CA gene in ovarian and breast cancers. CHn

Cancer Res, 11: 2875-2878, 2005.

44. Li, V. S., Wong, C. W., Chan, T. L., Chan, A. S., Zhao, W., Chu, K. M., So, S., Chen, X., Yuen, S. T., and Leung, S. Y. Mutations of PIK3CA in gastric adenocarcinoma. BMC Cancer, 5: 29, 2005. 45. Samuels, Y. and Velculescu, V. E. Oncogenic mutations of PIK3CA in human cancers.

Cell Cycle, 3: 1221-1224, 2004. 46. Blanchaert, R. H., Jr. Oral and oral pharyngeal cancer: an update on incidence and epidemiology, identification, advances in treatment, and outcomes. Compend Contin

Educ Dent, 23: 25-29., 2002. 47. Levi, S., Urbano-Ispizua, A., Gill, R., Thomas, D. M., Gilbertson, J., Foster, C, and

Marshall, C. J. Multiple K-ras codon 12 mutations in cholangiocarcinomas demonstrated with a sensitive polymerase chain reaction technique. Cancer Res, 51: 3497-3502, 1991.

48. Nakahori, S., Yokosuka, O., Ehata, T., Chuang, W. L., Imazeki, F., Ito, Y., and Ohto, M. Detection of hepatitis B virus precore stop codon mutants by selective amplification method: frequent detection of precore mutants in hepatitis B e antigen positive healthy carriers. J Gastroenterol Hepatol, 10: 419-425, 1995.

49. Righini, C, de Fraipont, F., Timsit, J., Faure, C, Brambilla, E., Reyt, E., and Favrot, M. Tumor-specific methylation in saliva: a promising biomarker for early detection of head and neck cancer recurrence. Clin Cancer Res, 13: 1179-1185., 2007. 50. Saccomanno, G., Archer, V. E., Auerbach, O., Saunders, R. P., and Brennan, L. M.

Development of carcinoma of the lung as reflected in exfoliated cells. Cancer, 33: 256- 270, 1974.

51. Machida, E. O., Brock, M. V., Hooker, C. M., Nakayama, J., Ishida, A., Amano, J., Picchi, M. A., Belinsky, S. A., Herman, J. G., Taniguchi, S., and Baylin, S. B. Hypermethylation of ASC/TMS1 is a sputum marker for late-stage lung cancer. Cancer

Research, 66: 6210-6218, 2006.

52. Palmisano, W. A., Divine, K. K., Saccomanno, G., Gilliland, F. D., Bayiin, S. B., Herman, J. G., and Belinsky, S. A. Predicting lung cancer by detecting aberrant promoter methylation in sputum. Cancer Research, 60: 5954-5958, 2000.

EXAMPLE 7 Mutant-Enriched Sequencing Identified High Frequency of PIK3Cλ Mutations in

Pharyngeal Cancer [0070] Frequent somatic mutation of PIK3CA has been identified in many human cancer types. We previously reported four PIK3CA mutations in 38 head and neck cancer samples; three of which were identified in six pharyngeal cancer samples. To determine the mutation frequency of PIK3CA in pharyngeal cancer, we studied 24 additional cases of pharyngeal squamous cell carcinoma in this study. Using both direct genomic DNA sequencing and novel mutant- enriched sequencing methods developed specifically for the three hot-spot mutations (H1047R, E545K and E452K) of PIK3CA, we detected five mutations of PIK3CA in the 24 pharyngeal cancers (20.8%). Three of the five mutations had been missed by the conventional sequencing

method and were subsequently detected by novel mutant-enriched sequencing methods. We showed that the mutant-enriched sequencing method for the H1047R hot-spot mutation can identify the mutation in a mixed population with wild-type DNA sequences at minimum sensitivity of 0.0028 (1 mutant: 360 wild-type DNA copies). These novel mutant-enriched sequencing methods allow the detection of the PlKiCA hot-spot mutations in clinical specimens which often contain limited tumor tissues (i.e. biopsy specimens). The mutant-enriched sequencing methods also allow the detection of the hot-spot mutations existing in tumor subclones that usually go undetected by the conventional sequencing method because of their minor cellular populations.

[0071] The data further support that oncogenic PIK3CA may play a critical role in pharyngeal carcinogenesis, and the mutant-enriched sequencing methods for PlKiCA are sensitive and reliable ways to detect PIKiCA mutations in clinical samples. Because PIKiCA and its pathway are potential targets for chemotherapy and radiation therapy, and frequent somatic mutation of PIKiCA has been identified in many human cancer types (e.g. breast cancer, colorectal cancer), the abilities to detect PIKiCA mutations with enhanced sensitivities have great potential impacts on target therapies for many cancer types.

[0072] Specimens of tumor tissues. Twenty-four cases of paraffin-embedded pharyngeal cancer blocks were obtained from the Department of Pathology. The acquisition of the tissue specimens was approved by the Columbia University Medical Center Institutional Review

Board and performed in accordance with Health Insurance Portability and Accountability Act

(HIPAA) regulations. The 24 specimens came from five female and nineteen male patients, with ages ranging from 38 to 78 years-old (average 57.9 ± 12.2 years-old). Of the 24 patients, eleven were heavy smokers (more than 40 packs per year), two of the 11 were also with heavy alcohol consumption, and one of the heavy smokers abused cocaine; two were moderate smokers, two had no smoking and had only occasional alcohol use, and the remaining nine patients' history was not available. One patient was a HTV carrier. All patients were diagnosed as squamous cell carcinomas of the pharynx. The grade of cancer in these patients was four well-, 18 moderately-, and two poorly-differentiated.

[0073] DNA isolations from paraffin embedded tissue. The cases were reviewed by two pathologists and the diagnosis confirmed. The paraffin embedded blocks containing tumor tissues were selected and five 10 μm thickness sections were cut for each case. The 24 cases studied included five surgical resection specimens and nineteen cases of small biopsy

specimens. Genomic DNA were extracted from the tumor tissues using QIAmp DNA Kit

(QIAGEN Inc., Valencia, CA). The procedures were performed according to the manufacturer's instructions for purification of genomic DNA from paraffin-embedded tissue.

[0074] Conventional genomic sequencing. Exons 9 and 20 of PIK3CA gene were analyzed by PCR amplification of genomic DNA (40ng each) and direct sequencing of the PCR products. New PCR primers were designed for this study to allow more efficient amplifications of genomic DNA from paraffin-embedded tissues. Primers for exon 9 were also designed to avoid interference from a homologous pseudogene located on chromosome 22ql 1.2 cat eye syndrome region (18). The primers for the PIK3CA exons 9 and 20 are PIK-E9F: CCAGAGGGGAAAAATATGACA; PIK-E9R: CATTTTAGC ACTTACCTGTGAC; PIK- E20F: CATTTGCTCCAAACTGACCA; PIK-E20R: TGAGCTTTCATT

TTCTCAGTTATCTTTTC. Before sequencing, PCR products were purified using the Geneclean Turbo Nucleic Acid purification Kit (Qbiogene, Irvine, CA). Finally, purified DNA fragments were sequenced using the corresponding forward PCR primers. Samples found to have a genetic alteration in the target gene were subsequently sequenced in the reverse direction to confirm the mutation using the reverse PCR primers. The mutation was then further verified by sequencing of a second PCR product derived independently from the original template. All sequencings were performed with ABI' s 3100 capillary automated sequencers at the DNA facility of Columbia University Medical Center in New York (Qiu et al., Clin. Cancer Res. 12:1441-1446 (2006)).

[0075] Mutant-enriched sequencing for detecting PIK3CA mutations, H1047R, E545K, and

E542K. To detect the PIK3CA hot-spot mutation A3140G (Hl 047R), each sample (40ng of genomic DNA) was first amplified using outer primers PIK-E20OF (GACATTTGAGCAAAGACCTGAA) and PIK-E20OR (ATCAAACCCTGTTTGCGTTT) for 30 cycles. After this first round of PCR, 2μl from each PCR product were digested with 2μl of restriction enzyme BsaBI (lOU/ul, New England BioLabs, Ipswich, MA) in a total of 50 μl volume at 60 °C overnight. Then 2μl of the digest were used for the second round of PCR for 40 cycles. The primers for the second PCR are PIK-E20IF (CATTTGCTCCAAACTGACCA) and PIK-E20IR (TGAGCTTTCATTTTCTCAGTTATCTTTTC). Each PCR product with the correct size was purified for DNA sequencing using the same primers as for the second PCR (PIK-E20IF or PIK-E20IR) (Fig. 5A).

[0076] For mutant-enriched sequencing of PIK3CA exon 9 hot-spot mutation, G 1633 A (E545K), the procedure is similar to the one described above for the hot-spot mutation A3140G, except for the enzyme and primers used. A mismatch primer PIK-E9MF (TCTACACGAGATCCTCTCTCTGTAATCTC) was used as the forward primer for both rounds of PCR. The reverse primers for the first and second PCR were respectively PIK-E9OR (GCATTTAATGTGCCAACTACCA) and PIK-E9IR

(CTGAGATCAGCCAAATTCAGTTATTTTTTC). The restriction enzyme digestion was performed with Hpy 1881 at 37 °C overnight. The reverse PCR primer PIK-E9R was also used as the DNA sequencing primer (Fig. 5B).

[0077] For the hot-spot mutation at exon 9, G1624A (E542K), the PCR strategy of mutant- enriched sequencing is the same as the one described above for the hot-spot mutation G 1633 A (E545K), in which a mismatch primer is designed to create a unique restriction enzyme site EcoRI in the PIK3CA exon 9 region. The mismatch primer PIK-2E9MR (CATAGAAAATCTTTCTCCTGCTCAGTGAAT) was used as the reverse primer for both rounds of PCR. The forward primers for the first and second PCR were respectively PIK- 2E9OF (GATTGGTTCTTTCCTGTCTCTG) and PIK-2E9IF

(TTGCTTTTTCTGTAAATCATCTGTG). The restriction enzyme EcoRI digestion was performed at 37 ⁰ C overnight. The forward PCR primer PIK-2E9IF was also used as the DNA sequencing primer (Fig. 5C).

[0078] The PCR condition for all the PCR reactions is 94°C, 2 minutes; (94 ^α C, 30 seconds; 60 ⁰ C, 30 seconds; 72°C, 30 seconds) x 40 cycles; 72°C, 5 minutes.

RESULTS

[0079] Conventional DNA sequencing detected PIK3Cλ mutations only in surgically resected but not in biopsy specimens of pharyngeal cancer. We initially screened for PIK3CA mutation in 24 cases of pharyngeal squamous cell carcinoma samples using the direct genomic sequencing method that we had applied to our previous study on HNSCC (Qiu et al., Clin. Cancer Res. 12:1441-1446 (2006)). Because most pharyngeal cancers are treated non- surgically (i.e. concurrent chemoradiation), 19 of our specimens were biopsy tissues. Two mutations of PIK3CA were identified in the five resection samples and none in the biopsy specimens. One was PIK3CA hot-spot mutation G1633A (E545K), which resulted in the replacement of codon 545 glutamic acid (GAG) by lysine (AAG). The other mutation, was a missense mutation in exon 20 nucleotide 3127 A→ G, led to a codon 1043 ATG (Met)→GTG

(VaI) substitution. This missense mutation of PIK3CA has been reported previously. Both mutations were not detected in the surrounding normal tissues, thus both were somatic mutations.

[0080] Mutant-enriched sequencing identified PIK3CA hotspot mutation H1047R in samples screened negative by conventional DNA sequencing approach. Conventional DNA sequencing method only recognizes the mutant DNA if it is present in more than 10 percent of a mutant/wild-type mixed population in primary tumor tissue (Levi et al., Cancer Res. 51:3497 (1991); Nakahori et al., J. Gastroenterol. Hepatol. 10:419 (1995)). Therefore we hypothesized that the low tumor to normal cell ratio in our biopsy specimens may have caused some false negative results and contributed to the lower frequency of PIKiCA mutation observed in the current study (2/24, 8.3%) than our previous report (3/6, 50%) (Qiu et al., Clin. Cancer Res. 12:1441-1446 (2006)). The unexpectedly low mutation frequency of PIK3CA in the biopsy samples might also have been caused by the overall low or poor DNA contents available in these tissues. We set out to develop a mutant-enriched sequencing method to detect hot-spot mutations of PIK3CA in specimens with low tumor DNA contribution, such as biopsy samples.

[0081] Mutant-enriched sequencing methods generally involve a first-round PCR, restriction enzyme digest, and a second-round PCR. In contrast to conventional DNA sequencing, mutant- enriched sequencing methods selectively amplify the mutant copy of a targeted gene by reducing the wild-type copy number via a restriction enzyme digestion that is specific for the wild-type DNA after the first round PCR. In the second PCR amplification, only the mutant strands but not the wild-type ones would be further amplified. Thus, mutant-enriched DNA sequencing is particularly valuable when the ratio of mutant DNA is expected to be low. In our method of mutant-enriched sequencing for the detection of PIK3C 'A exon 20 hotspot mutation A3140G (H 1047R), we took advantage of a natural restriction enzyme site on the wild-type sequence that can be recognized and digested by enzyme BsaBI (cuts GATNNNNATC). This enzyme site is destroyed when the DNA is mutated, thus enzyme BsaBI exclusively digests the wild-type PIK3CA exon 20 DNA, but not the mutant A3140G (H1047R) sequence (Fig. 5A).

[0082] The feasibility of our method was first investigated in a head and neck cancer cell line, Detroit 562, which harbors the H1047R mutation.11 As shown in Figure 7al-a2, both the mutant and wild-type peaks were observed at 1:1 ratio as expected using conventional genomic sequencing (Fig. 7al), and the wild-type peak was entirely eliminated when our mutant-enriched sequencing method was applied (Fig. 7a2). To determine the sensitivity of our assay, we made a

series of dilutions of the genomic DNA from mutant cell line Detroit 562 with DNA from another cell line that has two wild-type PIK3CA alleles. When the ratio of mutant and wild-type DNA copies reached beyond 1:360, the mutant peak could still be recognized by mutant- enriched sequencing (Fig.7a2). In contrast, with conventional PCR-sequencing, the mutant peak disappeared when the ratio of mutant and wild-type DNA reached 1:18. This indicated that mutant-enriched sequencing was at least 20-fold more sensitive than direct genomic sequencing.

[0083] Using this mutant-enriched sequencing method, 2 additional mutations of the PIK3CA gene were identified in these 24 pharyngeal cancer samples (data not shown and Table 2). This result supported our hypothesis that the low frequency of PIK3CA mutation detected in clinical pharyngeal cancer biopsy samples by the conventional DNA sequencing method was partially caused by the contamination of normal cells.

[0084] Mutant-enrich sequencing for PIK3CA exon 9 hotspot mutation E545K. For mutant- enriched sequencing of PIK3 CA exon 9 hotspot mutation, G1633A (E545K), mismatch primer (PIK-E9MF) was designed to introduce 2 A→T nucleotide mismatches in the forward primer to create a unique restriction enzyme site Hpyl88I (TCNGA), because there is no natural unique restriction enzyme site specific for the wild-type but not the mutant DNA sequences at this hot- spot. Using a patient's tumor DNA with known PIK3CA E545K mutation (patient No. 4, Table 2), we showed that using our mutant-enriched sequencing method, only the mutant peak remained while the more prominent wild-type peak observed in the conventional sequencing assay disappeared (Fig. 7bl-b2). Applying this powerful method, we screened the 24 pharyngeal cancer samples. We did not uncover any additional cases at this hot-spot mutation.

[0085] Mutant-enrich sequencing for PIK3CA exon 9 hotspot mutation E542K (G1624A) and a nonhot-spot mutation E542G (A1625G). We applied the same mismatch PCR strategy to enrich the mutant allele of PIK3CA hotspot mutation E542K. A restriction enzyme EcoRI site was introduced by the mismatch primer PIK-2E9R. EcoRϊ digestion disrupted the wild-type PIK3CA DNA, but not the mutant of P1K3CA E542K sequences (Fig. 5c). This mutant-enriched sequencing method was tested in a previously reported patient sample with a known PIK3CA E542K mutation. We showed that only the mutant peak remained while the corresponding wild- type peak completely vanished when the mutant-enriched sequencing was applied (Fig. 7cl-c2). Interestingly, this method can detect not only the E542K (G 1624A) mutation, but also a previously described nonhot-spot E542G (Al 625G) mutation (data not shown). We screened the

24 specimens of pharyngeal cancer and an additional case of PIK3CA E542K (G 1624A) mutation was identified (Fig. 7c3-c4).

[0086] Five mutations of the PIK3CA gene were identified in the 24 pharyngeal cancer samples. Five mutations of PIK3CA were identified in the 24 cases of pharyngeal cancer by the combination of conventional sequencing and mutant-enriched sequencing methods (Table 2). Four of the 5 mutations could have been identified by the mutant-enriched sequencing methods alone because they were hot-spot mutations (4/5, 80%), but only 2 were detectable by the conventional sequencing method (2/5, 40%). Two patients with PlKiCA mutations were not smokers and only drank alcohol occasionally, while one was a heavy smoker and drank alcohol occasionally (Table 2). There is no apparent association between PIKiCA mutation and smoking or alcohol consumption. There is also no apparent association between PIK3CA mutation and the degree of tumor cell differentiation (Table 2).

TABLE 2 Clinical Features of Five Pharyngeal Cancers With PIK3CA Mutation.

No. Gender Age History Histology Mutation

(year) (differentiated)

1 Male 59 Not Available Poorly A3127G

2 Female 70 No smoking, Well A3140G occasional alcohol

3 Male 66 No smoking, Well A3140G occasional alcohol

4 Male 43 Not Available Moderate G1633A

5 Male 64 Heavy smoking, Moderate G1624A occasional alcohol

DISCUSSION

[0087] Previously, we reported PIK3CA mutations in head and neck cancers (4/38, 10%), 3 of the 4 were identified in pharyngeal cancer samples (3/6, 50%). However, whether PIK3CA can be used as a potential biomarker for diagnosis and molecular target therapy in pharyngeal cancer was unclear, due to the small sample number available at the time. Therefore, we decided to investigate the frequency of PIK3 'CA mutation in pharyngeal cancer using 24 additional samples.

[0088] Initially, only 2 mutations of PIK3CA, including a hotspot mutation E545K and a missense mutation in exon 20 nucleotide 3127 A→G, were found in 5 surgically resection specimens and none in 19 biopsy specimens by the conventional genomic sequencing method. We attributed this low mutation frequency to the quality of clinical biopsy samples. Clinical biopsy samples often contain small numbers of tumor cells mixed with a large population of

normal cells. The mutant DNA is often missed by the conventional PCR-sequencing method. To increase the sensitivity of detecting PIKSCA gene mutation, we developed novel mutant- enriched DNA sequencing methods for its 3 hot-spot mutations, H1047R, E545K and E542K. The 3 hot-spot mutations account for 78.6% of all PIK3CA mutations reported. We were able to show that mutant-enriched sequencing can identify the H1047R mutant DNA in a mixed population with wild-type DNA at sensitivity of 0.0028 (1 mutant: 360 wild-type DNA copies). Using this mutant-enriched sequencing method for H 1047, we found 2 additional mutations in these pharyngeal cancer samples. An additional mutation was identified by the mutant-enriched sequencing protocol for hotspot mutation H542K (G 1624A). Thus, a total of 5 PlKiCA mutations were identified in 24 cases of pharyngeal cancer in combination of regular DNA sequencing and mutant-enriched sequencing.

[0089] It is important to note that 4 of the 5 mutations were hot-spot mutations and could have been identified by the mutant-enriched sequencing methods alone (4/5, 80%), but only 2 were detectable by the conventional sequencing method (2/5, 40%). This means that the ability to detect PIK3CA mutations increased by 200% when the mutant-enriched sequencing methods are utilized. The 80% detection rate is within the expectation, because the 3 hot-spot mutations account for -80% of total PIK3CA mutations. Two patients with PlKSCA mutations did not have a history of smoking or alcohol-abuse. This suggests that PIKSCA mutation might be a critical cause for those pharyngeal cancer patients without history of smoking and alcohol consumption. PIK3CA mutation was not found associated with the degree of differentiation in the current study.

[0090] The primers for the mutant-enriched sequencing analyses of the PIK3CA exon 9 hotspot mutations were designed to avoid interference from a homologous pseudogene located on chromosome 22ql l.2 cat eye syndrome region. At least one of the 2 primers for each PCR amplification contains mismatched base pairs to the pseudogene. Those mismatched nucleotides in either the forward and/or reverse primers seemed sufficient to prevent PCR amplification of the pseudogene in our mutant-enriched sequencing analyses. This was supported by the fact that we never observed the A1634C change in our samples. The A1634C change was mistaken as a mutation in previous literatures, when in fact it is only one of the dissimilarities between the pseudogene and PlKSCA at Exon 9, nucleotide 1634 (it's A for PIKSCA and C for the pseudogene). We also did not observe the frameshift change that had also been associated with the amplification of the pseudogene.

[0091] Comparing to mutant-selective PCR and restriction fragment length polymorphism

(RFLP) analysis (Levi et al., Cancer Res. 51:3497 (1991); Nakahori et al., J. Gastroenterol. Hepatol. 10:419 (1995)), we believe that our mutant-enriched sequencing method is more sensitive and specific because of the following (i) there is the possibility of the nonspecific digestion by the restriction enzyme when confirming the mutation by PCR-RFLP; (ii) small amount of mutant DNA after digestion may not be enough to be visualized on an agarose gel in the PCR-RFLP analysis. The mutant-enriched sequencing method directly displays the exact nucleotide sequence; (iii) the mutant-enrich sequencing method is not limited by available restriction enzyme sites. A unique enzyme site could be introduced by mismatch PCR, as we have done for the detection of PlKSCA exon 9 hot-spot mutations E545K and E542K. Thus, the mutant-enriched sequencing method is more superior to both the PCR-RFLP analysis and the conventional genomic sequencing assay for detecting hot-spot mutations. This method is particularly valuable in clinical applications where tumor samples are often mixed with a large population of normal cells.

[0092] Our current study concluded that PIK3CA mutation frequency in pharyngeal cancer is -21% (5/24). A recent study of PIKSCA mutation in nasopharyngeal carcinoma reported 2 mutations in 6 cell lines and but none was identified in 40 clinical samples (4.3% or 2/46). It is possible that the low tumor to normal DNA ratio in their clinical samples might have contributed to the negative finding in that study. Thus, a more sensitive assay could have improved the detection of PIKSCA mutations in such clinical samples where limited amounts of tumor DNA were available.

[0093] In conclusion, sensitive mutant-enriched sequencing methods were developed to detect the 3 hotspot mutations of the PIKSCA gene in clinical tumor samples. This novel detection method can detect -80% of all PlKSCA mutations and is valuable in clinical applications particularly in samples with little tumor cell contribution (i.e., biopsy samples) or with tumor subclones that usually go undetected by the conventional sequencing method because of their minor cellular populations. Several studies have shown that PIKSCA and its pathway are potential targets for chemotherapy or radiation therapy, including target therapies for EGFR, Her-2, mTOR, and Akt. Therefore, the clinical applications of the mutant-enriched sequencing methods would have great potential impacts on early detection and target therapy for many cancer types harboring frequent PIKSCA mutations (e.g., breast cancer, colorectal cancer).