Survival implications of context-dependent oncogenic <i>PIK3CA</i> mutations in HNSCC.

Abstract

6114 Background: Head and neck squamous cell carcinoma (HNSCC) is a heterogenous disease that generally lacks clinically useful stratification methods beyond anatomy and HPV status. PIK3CA is the most frequently mutated oncogene in HNSCC, and previous studies have linked PIK3CA mutations to survival in HNSCC. Thus, we aimed to investigate overall survival (OS) in a large multi-center cohort of HNSCC as a function of specific PIK3CA mutation hotspots to better understand the molecular drivers of risk and prognosis in HNSCC. Methods: A subset of oral cavity (OC, n=606) and oropharyngeal (OP, n=480) tumors from the VOYAGER consortium underwent exon sequencing in 1,109 cancer related genes and whole HPV16/18 genomes. Mutations in PIK3CA were analyzed using a somatic variant calling and filtering pipeline, and driver alterations were annotated based on gain-of-function variants at recurrent hotspots in specific functional protein domains. HPV status was defined using serology, p16 IHC, RNA ISH, and viral genomics. Multivariable Cox proportional hazards models were used to assess the relationship between PIK3CA mutations and five-year OS adjusted for age, sex, HPV status, tumor type, and TNM staging. Results: In OP tumors, 289 were HPV(+) and 174 were HPV(-), while only 13 of OC tumors were HPV(+). The overall PIK3CA mutation rate was 18% (13% of OC vs 24% of OPs), and among OP cases, PIK3CA mutations occurred in 28% of HPV(+) vs 18% of HPV(-) tumors. Of the nine recurrent hotspots in PIK3CA, the most common were the APOBEC-associated ones in the helical domain, p.E545K (3% OC vs 12% OP) and p.E542K (2% OC vs 5% OP), as well as smoking-associated kinase domain mutations at p.H1047R (3% OC vs 2% OP). In OC and OP cases combined, tumors harboring PIK3CA driver alterations in one of the nine hotspots showed a statistically significant improvement in OS that was independent of age, sex, staging, tumor type, and HPV status (HR 0.72, P=0.048). Among all HPV(+) tumors, driver mutations in PIK3CA conferred significantly better OS in the adjusted model (HR 0.42, P=0.02). Likewise, in all OP tumors, PIK3CA driver mutations were significantly associated with improved OS in the multivariate models (HR 0.60, P=0.04) and when stratified by HPV status this effect was particularly strong for HPV(+) OP tumors only (HR 0.45, P=0.04). Among the hotspots, p.E545K was the only one with improved survival in a univariable analysis of the full cohort (Log-rank P=0.01), but this was not significant in the adjusted model (HR 0.62, P=0.07). Conclusions: Our findings validate previously described associations between PIK3CA mutation and OS in HNSCC that are dependent on anatomic site and HPV status. Moreover, we provide a new perspective by analyzing OS as a function of recurrent mutation hotspots in PIK3CA , which may signify distinct pathologic processes among HNSCCs and thus may be clinically useful for patient stratification with prognostic or therapeutic intent.