Frequent mutation of the PI3K pathway in head and neck cancer defines predictive biomarkers

Abstract

Genomic findings underscore the heterogeneity of head and neck squamous cell carcinoma (HNSCC). Identification of mutations that predict therapeutic response would be a major advance. We determined the mutationally altered, targetable mitogenic pathways in a large HNSCC cohort. Analysis of whole-exome sequencing data from 151 tumors revealed the phosphoinositide 3-kinase (PI3K) pathway to be the most frequently mutated oncogenic pathway (30.5%). PI3K pathway-mutated HNSCC tumors harbored a significantly higher rate of mutations in known cancer genes. In a subset of human papillomavirus-positive tumors, PIK3CA or PIK3R1 was the only mutated cancer gene. Strikingly, all tumors with concurrent mutation of multiple PI3K pathway genes were advanced (stage IV), implicating concerted PI3K pathway aberrations in HNSCC progression. Patient-derived tumorgrafts with canonical and noncanonical PIK3CA mutations were sensitive to an mTOR/PI3K inhibitor (BEZ-235), in contrast to PIK3CA-wild-type tumorgrafts. These results suggest that PI3K pathway mutations may serve as predictive biomarkers for treatment selection.

Figure 1

Mutations in oncogenic signaling pathways in HNSCC. (A) Mutation rates of the major mitogenic pathways (the PI3K pathway, the MAPK pathway and the JAK/STAT pathway) in 151 HNSCC patient tumors determined by whole exome sequencing. Components of each pathway examined are displayed underneath each pie chart. (B) Bar graph detailing the number of mutations (dark bars) of each particular component of the PI3K pathway as well as the number of HNSCC tumors harboring these mutations (grey bars). (C) PI3K pathway-mutated HNSCC tumors have higher of non-synonymous mutations and (D) cancer gene mutations when compared to HNSCC tumors without any PI3K pathway mutations. Bar graph representing the average number of non-synonymous mutations per tumor (C) and the average number of cancer gene mutations (D) in 151 HNSCC tumors. Statistical significance was calculated by Fisher’s Exact test, P<0.0001 (N=151). (E) Graphical representation of the number of HNSCC tumors with mutation of multiple components of the PI3K, MAPK and the JAK/STAT pathways, respectively.

Figure 2

PIK3CA mutations in HNSCC tumors. A) Schematic of all PIK3CA mutations found in 151 HNSCC tumors by whole exome sequencing. The amino acid (a.a.) positions of each domain is shown in grey below each domain. The number of mutational events at each site is indicated by a filled triangle (▲) in the graph above. Blue triangles indicate mutations found in HPV/HNSCC tumors. Keys: ABD: p85 binding domain; RBD: Ras binding domain; C2 Superfamily; Helical: PIK domain; Kinase: Kinase domain of PIK3CA. (B) Effects of PIK3CA mutations on PI3K signaling in HNSCC cells. WT PIK3CA, hotspot mutant H1047R, and novel mutants: R115L, G363A, C971R, and R975S were stably expressed in an HNSCC cell line harboring no endogenous mutations in the PI3K pathway, PE/CA-PJ34 (clone C12) cells, by retroviral infection. Shown here is a representative Western blot with densitometry values normalized to beta-tubulin loading controls for each engineered cell line. Increased phosphorylation of AKT at the T308 and/or S473 residue was generally observed in HNSCC cells stably expressing WT or mutant PIK3CA constructs relative to the EGFP expressing HNSCC cells, indicating enhanced activation of the PI3K signaling pathway. (C) Effects of PIK3CA mutations on HNSCC cell growth. HNSCC cells stably expressing WT or mutant PIK3CA constructs demonstrated enhanced growth at 72 hours in media with 2% FBS by MTT assay compared to cells expressing EGFP vector control (p<0.0001***). PIK3CA(H1047R) expressing cells further demonstrated enhanced growth when compared to simulated WT PIK3CA amplification (p = 0.001). Data shown here represents growth studies from three sets of independent replicate cell lines (separate infections, n=18 for each group).

Figure 3

PIK3CA mutation enhances sensitivity to PI3K pathway inhibition. (A) HNSCC cells containing endogenous PIK3CA mutation (H1047R) (CAL-33; Detroit 562) and cells containing WT PIK3CA [SCC-9, PE/CA-PJ34 (clone C12)] were treated with a PI3K/mTOR inhibitor, BEZ-235, followed by growth determinations at 48 hrs (n=4). Experiments were repeated 3 times with similar results. (B) BEZ-235 inhibited growth of CAL-33 xenografts [with endogenous PIK3CA(H1047R) mutation]. CAL-33 cells (0.5 ×10⁶ cells) were inoculated into the flanks of nude mice. Treatment was started when the tumors became palpable 8 days after tumor cell inoculation. BEZ-235 (25mg/kg/day, n=9) or vehicle (n=10) was given by oral gavage. (C) Sanger sequencing results showing PIK3CA(E542K) mutation in the HPV-HNSCC patient tumor that were implanted the flanks of NOD SCIDγ mice. (D) PIK3CA-mutated tumorgrafts are sensitive to BEZ-235 treatment. HNSCC patient tumorgrafts were implanted into the flanks of NOD SCIDγ mice and treatment was started when tumors became palpable. BEZ-235 (25mg/kg) or vehicle control was given daily by oral gavage. Mice were given vehicle (n=7) or BEZ-235 (n=5) when tumors became palpable. Treatment with BEZ-235 significantly reduced the tumor size when compared to vehicle control (P<0.0001). (E) Western blots showing the effects of BEZ-235 (vs. vehicle control) on expression of PI3K signaling components in the PIK3CA-mutated tumorgrafts. Tumors were harvested for Western blotting at the end of the experiment on day 22. Densitometry values of band intensity are shown below each band (normalized to total beta-tubulin level). Phospho-AKT (S473) and phospho-S6 (S235/236) levels were significantly reduced upon BEZ-235 treatment when compared to the vehicle-treated tumors (P=0.0124 and P<0.0001, respectively). (F) HNSCC patient tumorgrafts from a WT PIK3CA tumor expressing low levels of pAKT were implanted into the flanks of NOD SCIDγ mice and treatment was started when tumors became palpable. BEZ-235 (25mg/kg) (n=6) or vehicle control (n=6) was given daily by oral gavage. Treatment with BEZ-235 failed to significantly reduce the tumor size when compared to vehicle control (P=0.300).