The prognostic value of TP53 mutations in hypopharyngeal squamous cell carcinoma

In OPSCC, p16 immunopositivity is commonly used as a surrogate marker for HPV determination [16]. Therefore, the p16 status was evaluated in surgically excised specimens using immunohistochemistry (IHC) according to the standard techniques as previously described [17]. A mouse p16 monoclonal antibody (1:100 dilution; Santa Cruz Biotechnology, CA, USA) was used as the primary antibody, and immunostained samples were blindly reviewed and scored independently by two investigators (M. A and Y. S). In accordance with previous studies, p16 positivity by IHC was defined as strong and diffuse nuclear and cytoplasmic staining in ≥70% of the tumor cells [16, 17].

However, p16 expression does not always indicate the presence of HPV DNA, and the combination of p16 expression determined by IHC with HPV DNA determination by polymerase chain reaction (PCR) or in situ hybridization (ISH) is considered to provide the almost perfect sensitivity and specificity [18, 19]. Therefore, p16-immunopositive specimens were also tested for HPV DNA by HPV-ISH, as previously described [19, 20]. Briefly, HPV DNA was detected using an ISH method with catalyzed signal amplification (GenPoint signal amplification system; Dako, Kyoto, Japan), in accordance with the manufacturer’s instructions. Slides were hybridized using a biotinylated GenPoint HPV probe (This probe has been found to react with HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68 on FFPE tissues and/or cells by ISH, Dako). Slides were scored as positive for HPV if a punctate signal pattern was observed in almost all tumor nuclei.

Tumor tissue specimens were collected during surgery, and snap-frozen in liquid nitrogen and stored at −80 °C. Genomic DNA was extracted using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany), in accordance with the manufacturer’s protocol. In specimens where the harvest of frozen sections appeared to interfere with the pathological margins, DNA was isolated from formalin-fixed, paraffin-embedded (FFPE) tissue blocks. Briefly, the tumor lesions on hematoxylin and eosin-stained slides were marked, and the corresponding areas were identified on unstained tissue sections. Each selected area was carefully dissected under microscopic observation. Genomic DNA was then extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen).

PCR amplification and Sanger sequencing were performed to detect TP53 mutations in exons 2–9, containing 98% of all mutations described in HNSCC cases [21]. A total of 20 ng/μl genomic DNA per sample was used for PCR amplification using PrimeSTAR HS DNA Polymerase(Takara Bio, Shiga, Japan). Amplification conditions included two-step cycle of 98 °C for 15 s and 68 °C for 90 s, for a total of 44 cycles, except for the amplification of exon 2–3 fragments harvested from frozen and FFPE specimens and exon 6 fragments harvested from FFPE specimens, which were amplified by nested PCR (25 cycles each) using two primer pairs. Subsequently, PCR products harvested from FFPE tissue were purified using the QIAquick PCR Purification Kit (Qiagen), in accordance with the manufacturer’s protocol. Mutations were confirmed by Sanger sequencing using the Big Dye Terminator v3.1 Cycle Sequencing Kit and 3130xL Genetic Analyzer (Applied Biosystems, CA, USA). In this study, nonsense mutations, splice variants, and frameshifts were defined as truncating mutations, that lead to nonfunctional p53, based on previous studies [13, 22]. All samples were sequenced twice with independent PCR using forward and reverse primers.

IHC for p53 expression was performed according to standard IHC techniques. A mouse p53 monoclonal antibody clone DO-7 (1:100 dilution; Leica Biosystems, Nussloch, Germany) was used as the primary antibody. In accordance with a previous study, a sample was determined as p53-immunopositive when ≥10% of tumor nuclei were immunostained [23].

Primary endpoint was disease-specific survival (DSS) and secondary endpoint was overall survival (OS). Potential correlations between the treatment method and several clinical features were evaluated using the chi-square test; for analyses in which there were <4 patients, the Fisher’s exact test was used. Survival was analyzed using the Kaplan–Meier method and the log-rank test. Variables were also analyzed by multivariate survival analysis using the Cox proportional hazards model. Hazard ratios (HR) and 95% confidence intervals (CI) were calculated to determine the effect of each variable on outcomes. P values <0.05 were considered statistically significant. GraphPad Prism software version 5 (GraphPad Software, CA, USA) was used for the chi-square, Fisher’s exact, Mann-Whitney’s U, and log-rank tests. Mac Tahenryo-Kaiseki version 2.0 (ESUMI, Tokyo, Japan) was used for multivariate Cox regression models.

In this cohort of 57 HPSCC patients, 3 (5%) patients were immunopositive for p16; however, none of these three patients had detectable HPV DNA by HPV-ISH. Therefore, HPSCC was confirmed to be unrelated to HPV in all patients in this study (Fig. 1).

TP53 mutations were detected in 39 (68%) patients. Missense mutations, nonsense mutations, splicing variants, and frameshift mutations were found in 24 (42%), 9 (16%), 4 (7%), and 2 (3%) patients, respectively. TP53 mutations in exon 2, 3, 4, 5, 6, 7, 8, and 9 were found in 1, 0, 3, 11, 9, 4, 8, and 3 patients, respectively (Fig. 2).

Table 1 summarizes the clinical data and TP53 status. Table 2 shows the clinicopathological features according to TP53 mutation status. Histopathological analysis revealed positive surgical margins in 11 (19%) patients, ≥4 metastatic LNs in 14 (25%) patients, and ENE in 15 (26%) patients. Postoperative RT/CRT was administered to 16 (28%) patients. Of note, all of stage I/II patients (5 patients) had wild-type TP53, and patients with a past history of HNSCC or ESCC were significantly greater in the TP53 mutation groups than in the wild-type groups (P = 0.02). Administration of postoperative RT/CRT did not correlate with TP53 mutation status (P = 0.25).

Representative images of specimens exhibiting p53 immunopositivity are presented in Fig. 3. p53 immunopositivity was detected in 22 patients, including 5 (28%) and 17 (71%) patients in the wild-type and missense mutation groups, whereas there were no patients with p53 immunopositivity in the truncating mutation group (P = 0.0001, chi-square test).

Eighteen (32%) patients died from HPSCC, whereas 8 (14%) patients died from other causes. The remaining 31 (54%) patients were alive and disease-free on last follow-up date. The median follow-up period for the entire cohort was 29 months (range: 3.5–101 months), whereas 45 months (range: 24–101 months) for patients who survived (n = 31) and 16 months (range: 3.5–85 months) for those who died (n = 26). The 3-year DSS of the wild-type group was significantly longer than that of the TP53 mutation group (94% vs 55%; P = 0.01, Fig. 4). Furthermore, patients with wild-type/missense mutations had significantly better 3-year DSS than those with truncating mutations (76% vs 43%; P = 0.03). The 3-year DSS rate of wild-type, missense mutation, and truncating mutation groups were 94%, 61%, and 43%, respectively (Fig. 5). The 3-year OS of the wild-type group was significantly longer than that of the TP53 mutation group (89% vs 42%; P = 0.007). The 3-year OS rate of wild-type/missense mutation group was not significantly different than that of the truncating mutation group (66% vs 40%; P = 0.14). The 3-year OS rate of the wild-type, missense mutation, and truncating mutation group were 89%, 43%, and 40%, respectively. In contrast, p53 immunopositivity did not correlate with DSS (P = 0.77). In the subgroup analyses of 52 stage III/IV patients, the 3-year DSS of the wild-type group was significantly longer than that of the TP53 mutation group (92% vs 55%; P = 0.02). The 3-year OS of the wild-type group was significantly longer than that of the TP53 mutation group (92% vs 42%; P = 0.006).

Table 3 shows the associations between the clinicopathological factors and DSS in univariate analysis. The presence of ≥4 metastatic LNs (P = 0.04) and ENE (P = 0.03) were poor prognostic factors. In contrast, tumor differentiation grade, T classification, stage, surgical margin, and postoperative RT/CRT did not correlate with DSS.

Multivariate Cox proportional hazard analysis using variables based on univariate analyses was conducted to determine independent prognostic factors for DSS and OS. The presence of TP53 mutations (P = 0.04; HR, 4.96; 95% CI, 1.08–22.8, and P = 0.02; HR, 4.75; 95% CI, 1.35–16.7, respectively) and ≥4 metastatic LNs (P = 0.03; HR, 3.00; 95% CI, 1.12–8.04, and P = 0.02; HR, 2.89; 95% CI, 1.22–6.86, respectively) have significant adverse effects on both DSS and OS. In the subgroup analyses of 52 stage III/IV patients, the presence of TP53 mutations was a significant adverse prognostic factor on OS, and nearly reached significance on DSS. (Table 4).