PIK3CA, HRAS and PTEN in human papillomavirus positive oropharyngeal squamous cell carcinoma

Background Recent genomic evidence suggests frequent phosphatidylinositide 3-kinase (PI3K) pathway activation in human papillomavirus (HPV) positive oropharyngeal squamous cell carcinoma. Mutations/amplification of the gene encoding p110α catalytic subunit of phosphoinositide 3-kinase (PIK3CA), loss of phosphatase and tensin homolog (PTEN) and HRAS mutations are known to activate PI3K pathway. Methods and results PIK3CA mutations were identified by Sanger sequencing in 23 of 75 (31%) HPV-positive oropharyngeal carcinomas, including exon 9 (p.E545K [n = 10] and p.E542K [n = 5]) or exon 20 (p.H1047Y, n = 2) mutations. Five rare and one novel (p.R537Q) PIK3CA mutations were identified. HRAS mutation (p.Q61L) was detected in 1 of 62 tested cases. PIK3CA amplification by fluorescence in situ hybridization (FISH) was identified in 4 cases (4/21, 20%), while PTEN loss was seen in 7 (7/21, 33%) cases (chromosome 10 monosomy [n = 4], homozygous deletion [n = 3]). Conclusions Overall, genetic alterations that likely lead to PI3K pathway activation were identified in 34 of 75 cases (45%) and did not correlate with disease specific survival. These findings offer a molecular rationale for therapeutic targeting of PI3K pathway in patients with HPV-positive oropharyngeal carcinoma.


Background
The phosphatidylinositide 3-kinase (PI3K) pathway is activated in about half of head and neck squamous cell carcinomas (SCC) by a number of mechanisms, including mutation or amplification of the gene encoding p110α catalytic subunit of phosphoinositide 3-kinase (PIK3CA) [1][2][3][4]. The higher incidence of PI3K pathway activation in oropharyngeal SCC was previously reported [5]. Oropharyngeal SCC are increasingly associated with human papillomavirus (HPV) infection [6,7] and the higher prevalence of PI3K pathway abnormalities in these tumors was eventually linked to HPV [8,9].
Most recent characterization of the mutational landscape of head and neck SCC showed that the genetic profile of HPV-positive SCC is distinct from that of HPV-negative SCC. For instance, HPV-positive oropharyngeal SCC harbor fewer mutations overall (e.g., no TP53 mutations) and more PIK3CA mutations. Specifically, of the 15 HPV-positive SCC with known PIK3CA status reported in the literature, 4 tumors harbored PIK3CA mutation (4/15, 27%) [10,11]. In contrast, PIK3CA mutations are present in about 5% (5/91) of HPV-negative head and neck SCC. The higher incidence of PIK3CA mutations in HPV-positive SCC suggests a new therapeutic option, as PI3K pathway is targeted by multiple drugs in development: PX-866 [12], and MK-2066 [13], and RAD001 [14]. Indeed, our most recent findings demonstrated that HPV-positive SCC tumorgrafts with activating PIK3CA mutation were highly responsive to PI3K-targeted therapy [15].
Increased PI3K signaling can also result from mutations in other genes in the PI3K pathway such as HRAS [16,17]. In addition to PIK3CA mutations and/or amplification, PI3K pathway may also be activated due to phosphatase and tensin homolog (PTEN) deletion, a known negative regulator of the PI3K signaling pathway [18].
The aim of the present study was to elucidate the molecular basis for therapeutic targeting of PI3K pathway in HPV-positive oropharyngeal SCC by characterizing the prevalence and prognostic significance of PIK3CA and HRAS mutations, PIK3CA amplification, and PTEN loss in 75 patients with HPV-positive oropharyngeal SCC.

Patients
This study was approved by the Institutional Review Board of the University of Pittsburgh Medical Center (IRB# PRO11010195). Seventy five cases of HPV-positive oropharyngeal SCC were identified from 1983 to 2007 and satisfied the following inclusion criteria: availability of formalin fixed paraffin embedded tissue, p16 immunohistochemistry and HPV in situ hybridization positivity, presence of tumor areas with >50% represented by cancer cells, and extraction of adequate DNA.
For p16 analysis, five-micrometer sections were deparaffinized. Heat-induced epitope retrieval was then performed in a citrate buffer. Immunohistochemistry for p16 (G175-405; BD Pharmingen, San Diego, CA) was performed as per the manufacturer's protocol. Cases were considered positive if >70% of tumor cells showed diffuse and strong cytoplasmic and nuclear staining [19].

PIK3CA and HRAS mutation analysis
Tissue cores from tumor targets were obtained as previously described [20]. DNA was isolated from tissue cores using the DNeasy tissue kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. For the detection of mutations, DNA was amplified with primers flanking exon 3 of the HRAS gene (forward primer 5′-GTC CTC CTG CAG GAT TCC TA -3′ and reverse primer 5′-CGG GGT TCA CCT GTA CT -3′), exon 9 of the PIK3CA gene (forward primer 5′-TGA AAA TGT ATT TGC TTT TTC TGT -3′ and reverse primer 5′-TGT AAA TTC TGC TTT ATT TAT TCC -3′) and exon 20 of the PIK3CA gene (forward primer 5′-TTT GCT CCA AAC TGA CCA A -3′ and reverse primer 5′-GCA TGC TGT TTA ATT GTG TGG -3′). PCR products were sequenced in both sense and antisense directions using the BigDye Terminator v3.1 cycle sequencing kit on ABI 3730 (Applied Biosystems, Inc., Foster City, CA) according to the manufacturer's instructions (Additional file 1: Figure S1 and Additional file 2: Figure S2). The sequences were analyzed using Mutation Surveyor software (SoftGenetics, LLC., State College, PA).

PTEN and PIK3CA fluorescence in situ hybridization (FISH)
Cases with known wild type PIK3CA and HRAS and available tissue were tested for PIK3CA and PTEN copy number changes by FISH (n = 22) (Additional file 4: Figure S4). Sixty to 130 cells were analyzed. PTEN (SpectrumOrange) and chromosomal enumeration probe 10 (CEP10, Spectrum Green) FISH was performed as per manufacturer's recommendations (Abbot Molecular, Des Plaines, IL, USA) and as previously described [22]. Results were interpreted using previously established thresholds [23,24]: PTEN homozygous deletion was defined as >20% of cells without PTEN locus signal and the presence of ≥2 CEP10 signals. Hemizygous PTEN deletion was defined as >30% of cells with only one PTEN signal and the presence of ≥2 CEP10 signals. As previously suggested, cases with >50% of cells with a single CEP10 signal were categorized as "Chromosome 10 monosomy", an additional mechanism of PTEN loss [22]. Despite repeated hybridization attempts, no data were obtained in one case.
PIK3CA (Spectrum Green) and CEP3 (Spectrum Orange) (Abbot Molecular, Des Plaines, IL, USA) FISH was performed as per manufacture's recommendation. PIK3CA amplification was defined as PIK3CA/CEP3 >2. Despite repeated hybridization attempts, no data were obtained in one case.

Statistical analysis
The primary endpoint was disease-specific survival (DSS) defined as time elapsed from the date of diagnosis until death from cancer. DSS was chosen as a primary endpoint over the overall survival due to the predominance of patients deceased from causes unrelated to the oropharyngeal SCC (n = 21). DSS was assessed only for patients with > 2 months of follow-up (n = 72). Patients who were alive at last followup or had died from other causes were censored. Survival data were presented as Kaplan-Meier plots. The log rank test was used to test survival equality. Covariates examined for association with survival included age, gender, smoking (ever versus never), site (tonsil and base of tongue), T and N classification, AJCC clinical stage, adjuvant chemotherapy or radiotherapy. Cross-tabulated categorical data were tested for independence with Fisher's exact test.

Results
The clinicopathologic characteristics of 75 HPV-positive oropharyngeal SCC patients are summarized in Table 1. The mean follow-up was 122 months (minimummaximum, 4-322). While 21 patients deceased of causes unrelated to oropharyngeal SCC, 14 patients died of disease. The 3-year disease specific survival (DSS) was 85%. In this clinicopathologically uniform group of patients only pathologic tumor stage (pT) correlated with DSS ( Figure 1A). For instance, the DSS at 5 years was 87% for pT1 and 40% for pT4 (p = 0.039). There was a trend toward a worse DSS for patients with clinical stage IV disease; however, it did not reach statistical significance, most likely due to the low number of events in stage III patients ( Figure 1B). It is noteworthy that patients' gender, age, smoking history, oropharyngeal sub-site, pN, and treatment modality was not associated with survival in these individuals.
PIK3CA, HRAS, and PTEN alterations PIK3CA mutations were identified in 23 of 75 patients with oropharyngeal SCC (31%), including exon 9 (p.E545K [n = 10] and p.E542K [n = 5]) or exon 20 (p.H1047Y, n = 2) mutations. Five cases with rare mutations and one case with novel mutation are presented in Table 2. Patients' gender, age, smoking history, oropharyngeal sub-site, pT, pN, clinical stage, and treatment modality were similar Smoking history was not available for 13 patients. 3 The oropharyngeal sub-site for one case was unknown. 4 pT was unknown for 4 cases with biopsies only. 5 Margin status was not assessed in 11 cases (biopsies, fragmented specimens). 6 pN and clinical stage were unknown for 1 patient. 7 Seven patients received no postoperative therapy; for 4 patients these data were not available.
between cases with wild type and mutated PIK3CA. Disease specific survival (DSS) of the patients in these two groups was not significantly different ( Figure 1C). HRAS mutation (c.182A > T, p.Q61L) was identified in 1 of 62 tested cases (or in 1 of 58 successfully tested cases; in 4 cases the status of HRAS was indeterminate). In the only case with HRAS mutation, the mutational status of PIK3CA was indeterminate.
Assuming that PIK3CA mutation or amplification, HRAS mutation, or loss of PTEN lead to PI3K pathway activation, patients with tumors harboring one of these  events were combined into a "PI3K-activated" group and compared to patients whose tumors did not harbor any of the above genetic alterations. PI3K pathway activation did not correlate with DSS ( Figure 1D).

Discussion
The clinical and pathologic characteristics of our HPVpositive oropharyngeal SCC population and the performance of conventional pathologic prognosticators (e.g., pT, pN) are consistent with prior reports [31]. To our knowledge, this is the largest HPV-positive oropharyngeal SCC cohort to undergo evaluation for PIK3CA and HRAS mutation and PIK3CA and PTEN amplification/loss. Our findings suggest that mutation or amplification of PIK3CA may represent the most common alteration in HPV-positive oropharyngeal SCC. It is noteworthy that recent mutational analyses of head and neck SCC also found PIK3CA alterations, albeit at lower rates [10,11,15]. The variation in PIK3CA mutation incidence is most likely due to the relative underrepresentation of HPV-positive oropharyngeal SCC in other cohorts, use of oropharyngeal site as a surrogate marker for HPV status, and the use of different techniques to assess for PIK3CA mutations. The recently published data [11,15] highlighted an interesting phenomenon that even though HPV-positive SCC harbored fewer mutations on average, as high as 20% of HPV-positive SCC (3/15 cases [15]) harbored PIK3CA mutation as the only cancer gene mutation, indicating that PI3K pathway mutations are enriched in HPV-positive tumors despite the lower rate of gene mutations in general. The higher prevalence of PI3K pathway abnormalities in oropharyngeal SCC was previously linked to HPV [8,9].
All mutations found in the samples of HPV-positive oropharyngeal SCC were heterozygous with mutant allelic frequency that appeared to range from 20% to 50% of alleles (corresponding to 40%-100% of cancer cells with a heterozygous mutation). None of the cases showed mutant allelic frequency of more than 50% suggesting that loss of the wild type PIK3CA allele or amplification of the mutant PIK3CA allele in cancer cells is exceedingly rare.
Although HRAS mutations have been reported to modulate signaling through the PI3K pathway [32,33], the role of the mutation found in a single HPV-positive oropharyngeal SCC in this study remains unclear.
PTEN is generally understood to function as a tumor suppressor gene and to negatively regulate PI3K pathway. Therefore, loss of PTEN should lead to PI3K pathway activation. The incidence of PTEN alterations in head and neck SCC varies in the literature and there is little indication that PTEN loss has an independent prognostic value [34,35]. We found that PTEN loss (as assessed by FISH) was relatively common in HPVpositive oropharyngeal SCC.
Activation of the PI3K pathway, generally by virtue of PIK3CA gene amplification, has been previously reported to represent a poor prognostic biomarker in head and neck SCC [36]. Others have reported that phosphorylation of AKT, a downstream target of PIK3CA, is associated with poor clinical outcome in oropharyngeal SCC, specifically [37]. Although HPV status was not specifically assessed in this cohort of oropharyngeal SCC, it is reasonable to presume that it was enriched for HPV-positive SCC. Our analysis showed no association between the genetic alterations we assessed for (combined into a "PI3K activated" group) and clinical outcome. Prior reports have generally focused on a single alteration or biomarker assessment. It is possible that some of the alterations we detected in HPV-positive oropharyngeal SCC do not activate the pathway as predicted. Or, more likely, each alteration modulates PI3K oncogenic signaling. Further functional studies in relevant preclinical models are needed to decipher the precise contribution of each mutation, amplification and/or loss to PI3K pathway status in HPV-positive oropharyngeal SCC.
One of the technical limitations of this study is that we restricted our assessment to exons 9 and 20 of PIK3CA gene and we have likely underestimated the frequency of PIK3CA mutation in this cohort. Similarly, we only assessed codon 61 of HRAS and did not perform codon 12/13 testing. Therefore, the actual mutation frequency of both PIK3CA and HRAS could be higher than reported here.
The variety of potential mechanisms leading to PI3K pathway activation underscores the complexity of the potential implications of our findings. It is possible, as reported by others and us, that head and neck SCC harboring "driver" PIK3CA mutations demonstrate enhanced response to PI3K pathway inhibitors [15,38,39]. Similar findings have been reported in clinical trials of patients with breast or gynecologic malignancies [40]. PI3K pathway inhibitors are under early investigation in head and neck SCC and clinical results are not yet available.
The EGFR monoclonal antibody cetuximab is FDAapproved in both newly diagnosed head and neck SCC as well as in the recurrent or metastatic setting [41]. We previously reported that PI3K pathway activation correlates with clinical resistance to cetuximab in head and neck SCC patients and targeting the PI3K pathway enhanced the antitumor effects of EGFR inhibitors in head and neck SCC preclinical models [42][43][44]. Therefore, molecular determinants of PI3K activation may identify individuals who may benefit from co-targeting of EGFR in conjunction with PI3K pathway inhibition.

Conclusion
In conclusion, we report an analysis of a large HPV-positive oropharyngeal SCC cohort and demonstrate distinct, but perhaps functionally homologous, mechanisms of PI3K pathway activation: PIK3CA mutations/amplification, HRAS mutation, or PTEN loss. We provide evidence, for the first time, of potentially activating genetic alterations of the PI3K signaling pathway in about 45% (34/75) of HPV-positive oropharyngeal SCC. The significance of the affected PIK3CA exon or specific PIK3CA mutation types, mechanism of PTEN loss, and the association with alternative mechanisms of PI3K signaling remain incompletely understood. Our findings offer a molecular basis for future studies of therapeutic targeting of PI3K pathway in HPVpositive oropharyngeal SCC.

Additional files
Additional file 1: Figure S1. Additional file 2: Figure S2. Representative sequencing electropherograms (SE) illustrating PIK3CA mutations. The entirely scanned SE are presented to illustrate low background.
Additional file 3: Figure S3. SNaPshot detection of common hotspot mutations in the exon 9 of the PIK3CA gene. SNaPshot result illustrating c. 1633G>A, p.E545K electropherogram pattern (mutant peak is indicated by a grey arrow). The bases are color-coded: "A"green, while "G"blue. The red trigon-shaped peaks represent internal size standards.
Competing interests P50CA97190 and the American Cancer Society (to JRG) and the Patricia L. Knebel Fund of the Pittsburgh Foundation (to VWYL).
Authors' contributions MM and AL carried out molecular genetic studies. SC, JRG, VWYL drafted the manuscript. SC, MNN conceived, designed the study, and acquired the data. SC performed the statistical analysis. All authors were involved in analysis and interpretation of data and critical revision of the manuscript. All authors read and approved the final manuscript.