Study cohorts
A retrospective search (January 2010 to December 2016) of all the cases of OPSCC, oral SCC (OSCC) and laryngeal SCC (LSCC) was performed in the electronic database of the Department of Pathology and HN Cancer Tumor Board database, King Chulalongkorn Memorial Hospital, Bangkok. The inclusion criteria were pathologically confirmed SCC and availability of the paraffin-embedded tissue blocks. Non-invasive tumors (carcinoma in situ), cancers other than pure SCC (e.g., verrucous, adenosquamous and undifferentiated carcinoma, etc.), and undoubtedly metastatic SCCs from other regions were all excluded. Searchable clinical parameters were recorded, including the age, gender, sampling procedure, TNM category, clinical stage, treatment modality, smoking and alcohol consumption status, follow up duration, oncological outcome and survival status. In addition, OPSCC patients were classified into low-, intermediate- and high-risk groups based on the stratification algorithm by Ang et al. [8]
This study was approved by the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University (Certificate of Approval CoA 531/2016). The need for consent was waived by the Institutional Review Board.
Histopathological evaluation
All histological slides were reviewed to validate the original diagnosis and render the morphologic type of OPSCC, as keratinizing (K), non-keratinizing (NK) or non-keratinizing with maturation (NK-M), or grade of OSCC/LSCC, as well-, moderately- and poorly-differentiated. The NK morphology was defined as sheets, nests or trabeculae of tumor cells with oval-to-spindle-shaped, hyperchromatic cells, high nuclear-cytoplasmic ratio, indistinct cell borders and inconspicuous nucleoli, added by frequent comedo-type necrosis and a high mitotic index (Fig. 1a) [21]. The K type was assigned to tumors without areas of NK morphology, with totally maturing squamous differentiation or squamous maturation present throughout the tumor [21]. Squamous maturation was characterized by polygonal-shaped cells with abundant eosinophilic cytoplasm, distinct cell borders, intercellular bridges and keratin pearls (Fig. 1f). The NK-M type consisted of features of both types, where a definite NK morphology had to be noted along with squamous maturation in more than 10% of the total tumor surface area (Fig. 1d) [21]. Squamous maturation in less than 10% of the total tumor surface area was allowed for inclusion to the NK type [21]. Tumor grade was defined as per the 2017 WHO Classification of HN Tumors [22].
Immunohistochemistry and in situ hybridization
Whole mount sections of all the enrolled specimens (n = 504) were stained for p16 immunohistochemistry as a surrogate marker of HPV infection using a monoclonal antibody to p16 (CINtec® Histology, Ventana, Tucson, AZ) on a VENTANA BenchMark ULTRA instrument (Ventana, Tucson, AZ). A recently described “multiple sections per slide” technique was applied [23]. Immunostaining of p16 was scored as positive (Fig. 1b, e), equivocal or negative (Fig. 1h), if nuclear and cytoplasmic staining was observed in > 70%, 30–70% or < 30% of the cancer cells, respectively. Initially, only p16-positive and equivocal cases were tested by DNA ISH HPV III Family 16 Probe (Ventana Medical Systems, Tucson, AZ) specific for high-risk types HPV, including types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 66. Second run of HPV ISH was performed on all p16-negative OPSCC and 125 random p16-negative OSCC samples. The HPV ISH was considered positive when episomal and/or integrative patterns were noted within the cancer cells [24]. The episomal pattern was defined as large homogeneous round navy-blue probes occupying the nucleus. The integrative pattern was represented by smaller discrete navy-blue probes imprinting the cancer nuclei (Fig. 1c, inset).
HPV DNA genotype detection
According to the standard protocol, 10 sections of 10 μm thick paraffin-embedded tissue were microdissected from selected tissue blocks after matching with histological slides. DNA was extracted using QIAamp FFPE tissue extraction kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol and stored at − 20 °C until used. Only specimens with adequate amount of DNA were further analyzed using quantitative real-time PCR-based detection testing (qPCR) on Qubit 2.0 fluorometer (Thermo Fisher Scientific, CA, USA) and Qubit® dsDNA HS Assay Kit (Life Technologies, Thermo Fisher Scientific, CA, USA). Hundred nanograms of extracted DNA was used as an input for HPV genotype testing using AmoyDx® Human Papillomavirus Genotyping Detection Kit (Amoy Diagnostics, China) following the manufacturer’s protocol. This kit covered 19 high-risk HPV types (HPV16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 68, 70, 73, and 82) and 2 low-risk HPV types (HPV6 and 11). The kit was composed of 22 HPV DNA positive reference controls, and 5 negative reference controls which allowed 100% accuracy for concordance detection rate with minimal detection of 100 copies HPV DNA per reaction. Multiple blanks were incorporated in PCR reactions to exclude cross-contamination. All qPCRs were conducted in duplicate fashion using ABI 7500 sequence detection analyzer (Thermo Fisher Scientific, CA, USA).
Statistical analysis
Statistical analysis was performed using the t-test, U-test and Chi-square. Survival analysis was conducted for the outcome of overall survival, with time to the outcome calculated from the date of diagnosis. Patients without events were censored at the date of last known follow-up. Unadjusted survival curves were obtained using Kaplan-Meier estimates and compared with the log rank test. Two-sided P value less than 0.05 was considered statistically significant. Graphical representation, statistical and survival analysis were performed in GraphPad Prism 6 software (GraphPad, La Jolla, USA).