Characterization of head and neck squamous cell carcinoma arising in young patients: Particular focus on molecular alteration and tumor immunity

The purpose of this study was to comprehensively characterize head and neck squamous cell carcinoma (HNSCC) arising in young patients (<45 years old).


| INTRODUCTION
Head and neck squamous cell carcinoma (HNSCC) is a common cancer worldwide and occurs in various anatomical subsites, including the nasal cavity/paranasal sinuses, oral cavity, nasopharynx, oropharynx, hypopharynx, and larynx. HNSCC usually occurs in late adulthood, with a mean age of 57-62 years old. [1][2][3][4] HNSCCs of the oral (mobile) tongue and oropharynx, however, are increasingly prevalent in patients younger than 45 years old, although these young patients lack of significant exposure to tobacco and alcohol. [4][5][6] However, factors underlying the increasing incidence of oral tongue and/or oropharynx SCC in young patients are not yet well known.
Conventionally, several toxins from tobacco smoking and alcohol consumption are known to be involved in the carcinogenesis of HNSCC by inducing genetic mutations and immune system impairment. Recently, high-risk human papilloma virus (HPV) infection was also identified to be involved in the carcinogenesis of HNSCC through deregulation of p16/INK4a-related cell cycle checkpoints and mutations of several molecules. [7][8][9] Recent comprehensive genetic Hyang Joo Ryu and Eun Kyung Kim contributed as co-first authors. analyses by The Cancer Genome Atlas Network show several known targets and driver oncogenes such as FGFR1 and CCND1 in HNSCC. 10 Furthermore, recent studies show the implications of an impaired immune system for the growth and progression of HNSCC. Expression of immunemodulatory signals such as PD-1, ICOS, LAG3, and CTLA-4 deregulates tumor-infiltrating CD8+ cytotoxic T cells, FOXP3+ regulatory T cells (Tregs), and antigen-presenting dendritic cells/macrophages in HNSCC. HPV infection has been suggested to alter immune surveillance of HNSCC. 11,12 Although genomic and immunogenic characteristics have been studied, further validation studies with large scale clinical samples using immunohistochemistry (IHC), in situ hybridization, and quantification assays of tumor-infiltrating immune cells are rare. 13 Furthermore, validation studies are very rare focusing on young age HNSCCs, which are lack of significant exposure to tobacco and alcohol.
In the present study, we characterize HNSCC arising in young patients with particular focus on molecular alteration and tumor immunity.
For further analyses of clinical and histologic factors, genetic alteration, protein expression, immune cell infiltration, and immune checkpoint markers, 552 consecutive cases that underwent curative surgical resection without preoperative treatment were retrieved from the overall 1787 HNSCC cases. These patients underwent standard adjuvant chemotherapy and/or radiation therapy according to tumor stage after pathologic evaluation of the surgical specimen. Demographic features of the 552 cases were summarized in Supporting Information Table S1 and Supporting Information Figures S2  and S3. Cases with distant metastasis and unclear clinical information were excluded. For this cohort group, clinical and pathological parameters were characterized, including sex, age, smoking history, alcohol consumption, lymphovascular invasion (LVI), perineural invasion (PNI), pathologic T (tumor) classification, and pathologic N (lymph node) classification. Tumors were classified according to the Seventh American Joint Committee on Cancer TNM cancer classification system and the World Health Organization system. [1][2][3] Neversmokers were defined as those with a lifetime smoking dose of fewer than 100 cigarettes, while former smokers were those who had not smoked for more than 1 year, and current smokers were those who were current smokers or who had quit smoking less than 1 year prior. 14 Alcohol consumption status was determined by the amount of pure alcohol consumed which was calculated as grams per day, according to the average amount, frequency, and type. "Heavy drinkers" referred to people who consumed more than 30 g/day, and those who drank less were defined as "social drinkers." 15,16 For IHC evaluation, dual-color silver in situ hybridization (SISH) and fluorescence in situ hybridization (FISH) were performed on archived formalin-fixed paraffin-embedded specimens obtained from HNSCC patients. However, samples that underwent decalcification or those with too few tumor cells were excluded. Cases of the nasal cavity/paranasal sinuses were also excluded because there were too few cases to perform appropriate statistical analyses. Finally, 396 cases of the oral tongue (N = 172), other oral cavity (N = 32), oropharynx (N = 122), hypopharynx (N = 28), and larynx (N = 42) were selected. The overall flow of case selection is summarized in Supporting Information Figure S4. The study was approved by the Institutional Review Board of Severance Hospital (protocol No.: 4-2015-0954). Histology as well as IHC, SISH, and FISH tests were interpreted by 4 experienced pathologists (S.O.Y, H.J.R, and E.K.K) blinded to clinical data.

| IHC and microscopic analysis
Two or 3 different representative tumor areas per sample were selected for tissue microarray (TMA) construction. Core tissues 3 mm in diameter were obtained from donor tissue blocks and arranged in recipient TMA blocks using a trephine apparatus.
IHC was performed on 4-μm TMA sections with a Ventana Bench Mark XT Autostainer (Ventana Medical Systems, Tucson, Arizona), as described previously. 17  c-Met, TrkA, TrkB, and panTrk protein expression was analyzed according to the semiquantitative H-score method; this method yields a total score range of 0-300 by multiplying the dominant staining intensity score (0, no staining; 1, weak or barely detectable staining; 2, distinct brown staining; 3, strong dark brown staining) by the percentage (0%-100%) of positive cells. NUT and cyclin D1 protein expression was analyzed by semiquantitatively measuring the percentage (0%-100%) of tumor cells showing positive nuclear expression; this method yields a total H-score range of 0-100. 18,19 IHC for p16 overexpression has emerged as a robust surrogate biomarker for HPV-mediated carcinogenesis 20 and it is conventionally considered positive when strong, dark, diffuse nuclear, and cytoplasmic expression is observed in >70% of tumor cells. All other staining patterns are scored as negative. [21][22][23] Densities of tumor infiltrating T lymphocytes and macrophages were semiquantitatively scored by counting CD3+ T cells, CD8+ T cells, CD163+ macrophages, and FOXP3+ regulatory T cells (Tregs) as well as immune checkpoint markers of ICOS, LAG3, PD-L1, and PD-1+ immune cells within tumor cell nests. We additionally scored tumor cells (TC) expressing PD-L1 as a percentage of tumor area according to previously described (TC3 ≥ 50%, TC2 ≥ 5% and <50%, TC1 ≥ 1% and <5%, and TC0 < 1%). 24 The 5 most representative ×400 magnification high-power fields were selected. Preserved intact lymphocytes and macrophages expressing CD3, CD8, CD163, FOXP3, ICOS, LAG3, PD-L1, and PD-1 were counted manually, and counts were averaged according to a previously described method. [25][26][27] Palatine tonsil tissues obtained from cancer-free individuals (N = 10) were used as the normal control of nontumorous squamous epithelium and lymphoid, immune cell tissues. Tonsillar SCC (N = 3) cases, which have been confirmed high-risk HPV-type 16-positive through HPV genotyping test (HPV DNA chip), were used as a positive control for p16 IHC.

| Dual-color SISH and interpretation
Using the MET DNA (Roche-Ventana) and Chromosome 7 enumeration (CEP7) probes (Roche-Ventana), dualcolor SISH was performed on a Ventana BenchMark XT apparatus (Ventana Medical Systems). Signals were enumerated in 100 tumor cells with intact nonoverlapping nuclei per core under a light microscope with magnification of ×600. According to the interpretive guide provided for INFORM HER2 DNA probe staining of breast carcinoma (Ventana Medical Systems), tumor nuclei with clearly distinct color signals for MET (black signal) and CEP7 (red signal) were evaluated. Normal tissues including vessels, fibroblasts, lymphocytes or adjacent normal squamous epithelial cells were considered as internal negative control. MET copy number gain (CNG) was defined as 2-4 MET signals per nucleus in ≥20% of tumor cells. MET amplification was defined as ≥5 MET signals per nucleus in ≥20% of tumor cells. 19 2.4 | FGFR1 FISH and interpretation FGFR1 probes labeled with the fluorophore Spectrum Orange (red) that hybridizes to the 8p12-8p11.23 region and probes labeled with the fluorophore Spectrum green that recognizes the centromere region of chromosome 8 (CEP 8) were used for FISH following the manufacturer's instructions (Abbott Molecular, Abbott Park, Illinois). Tumor tissue was scanned for amplification hot spots using an ×40 or ×63 objective. If FGFR1 signals were homogeneously distributed, random areas were used to count the signals. Twenty contiguous tumor cell nuclei from 3 hot spots or random areas, resulting in a total of 60 nuclei, were individually evaluated with the ×100 objective by counting red FGFR1 and green CEP8 signals. Normal tissues including vessels, fibroblasts, lymphocytes, or adjacent normal squamous epithelial cells were considered as internal negative control. High amplification of FGFR1 was defined based on modification of a previous protocol 28,29 as follows: FGFR1/CEP8 ratio is ≥2.0; or average number of FGFR1 signals per tumor cell nucleus ≥6 or; percentage of tumor cells containing ≥6 FGFR1 signals or large clusters is ≥10%. Representative figures of each marker of protein expression, gene alteration, and infiltrated immune cells are presented in Figure 1A,B.

| Statistical analysis
Chi-square test, Fisher exact test, two-sample t test, or one-way analysis of variance tests were used to analyze differences between evaluated variables. Overall survival was measured from date of initial diagnosis to date of death or last follow-up. Progression-free survival was measured from the date of initial diagnosis to that of disease progression, defined as cancer recurrence, progressive disease without complete remission, or cancer-related death during the study period. The Kaplan-Meier method was used to analyze survival rates, and differences were compared using the log-rank test. The Cox proportional hazards model was implemented for multivariate analysis. Two-sided P values <.05 were considered statistically significant. Statistical analyses were conducted using IBM SPSS 23 software for Windows (IBM Corp, Armonk, New York).

| Demographic pictures
The median (mean) age of overall 1787 HNSCC patients was 60.2 (61.0) years old. When dividing the age group into young age (age < 40, and 40 ≤ age < 45 years old) and late adulthood (age ≥ 45 years old), the majority (89.1%, 1593/1787) of HNSCC cases occurred in late adulthood, and only 10.9% (194/1787) of HNSCC cases occurred in young age (age < 45 years old). Among the 194 young age HNSCC cases, SCC occurred mostly in the oral tongue (50.5%, 68/194), followed by the oropharynx (17.5%,  (Figure 2A). In univariate survival analysis according to anatomical sites, hypopharynx SCC revealed the worst overall survival rate while oral tongue SCC revealed the best overall survival rate (P < .001; Figure 2B). In univariate survival analysis according to age, young age group showed superior overall survival rate compared to that of late adulthood group (P < .001; Figure 2C).

| Young age versus late adulthood: Clinicopathological factors, molecular alteration, and tumor immunity
When comparing the clinicopathological features of HNSCC between young age (<45 years old) and late adulthood (≥45 years old) groups (Table 1) in 552 radically resected HNSCC, young age HNSCC (81 cases; 14.7%) was more associated with female sex, never smoker, oral tongue site, p16 (HPV) negativity, absence of LVI, and presence of PNI compared to late adulthood SCC. Other factors, especially pathologic T or N classification, were not different between these 2 groups.

| Analysis of survival outcomes
In univariate Cox analysis for overall survival in young age HNSCC patients (Table 2), pT and pN classification types were significantly related to inferior overall survival rates. High pN classification types (pN2-3 vs pN0-1) were determined to be independent factors for poor overall survival rate in multivariate Cox analysis. High expression of cyclin D1 protein showed a tendency related to inferior overall survival rates in the univariate Cox analysis ( Table 2).
In univariate Cox analysis for progression-free survival in young age HNSCC patients (Table 3), PNI, PD-L1 FIGURE 3 Heat maps for mean values of each marker, which were evaluated with the percent or H-score of gene expression level and alteration frequency between young age (<45 years) and late adulthood (≥45 years) groups [Color figure can be viewed at wileyonlinelibrary.com] positivity in tumor cells (TC2 + TC3), and higher ratio of CD163+ TIMs to CD8+ TILs were significantly related to progression-free survival rates; and these three variables were determined to be independent factors for poor progression-free survival rate in multivariate analysis. High expression of cyclin D1 and pan-Trk protein showed a tendency related to inferior overall survival rates in the univariate Cox analysis (Table 3).

| DISCUSSION
In this study, we comprehensively characterize HNSCC arising in the young age (<45 years old) patients compared to the late adulthoods (≥45 years old), with focus on molecular alteration and tumor immunity. This may be the first comprehensive validation study using diverse molecular and immune markers in a large number of clinical samples that compares patient age groups.
Young age HNSCCs occurred in 10.9% (194/1787) and showed different clinicopathological features compared to late adulthood HNSCCs. Young age SCC was more associated with female sex, never-smoker status, p16 (HPV) negativity, and oral tongue site. The exact incidence of young age HNSCC still remains unclear, as most of the previous studies on young age HNSCCs have focused on the oral cavity or oropharyngeal SCC, which show relatively higher incidence of young age, rather than SCC of various anatomical locations of head and TABLE 2 Cox analysis for overall survival in young age head and neck squamous cell carcinoma (HNSCC) patients. In univariate Cox analysis, pT and pN classification types were significantly related to inferior overall survival rates. High pN classification types (pN2-3 vs pN0-1) were determined to be independent factors for poor overall survival rate in multivariate Cox analysis neck region. [4][5][6] A systemic review of oral and oropharynx cancer in young age patients reported a significantly different incidence between North America (5.5%) and both Africa (17.2%) and Asia (12.1%). 5 With respect to the finding that young age HNSCC showed lower p16 positive rate compared to late adulthood HNSCC, it is considered that young age HNSCC is composed of mostly oral tongue cancer which is usually p16 negative. In fact, most p16 positive young age HNSCCs were located in the oropharynx (14/15; 93.3%) in the present study. Similarly, in young age HNSCC, oropharyngeal (tonsil) cancer is associated with HPV and males (smokers and nonsmokers), while oral (tongue) cancer is associated with women (nonsmokers), but HPV is not. 30,31 Although pathologic T or N classification types were not different, less frequent LVI and more frequent PNI were observed in young age SCC. More frequent PNI in young age HNSCC seems to be related to anatomical characteristics of oral tongue where young age HNSCC occurs prevalently. Oral tongue is highly innervated with peripheral nerve systems, and PNI correlates with recurrence and survival. [1][2][3] In fact, we also noted that PNI was significantly related to tumor progression in young age HNSCC. These findings should be considered for the management of young age HNSCC patients.
Regarding molecular alteration, higher expression of c-MET protein and higher frequency of MET gene CNG was more characteristic in young age SCC. Although the alterations TABLE 3 Cox analysis for progression-free survival in young age head and neck squamous cell carcinoma (HNSCC) patients. In univariate Cox analysis, perineural invasion, PD-L1 positivity in tumor cells (TC2 + TC3), and higher ratio of CD163+ tumor infiltrating macrophages (TIMs) to CD8+ TILs were significantly related to progression-free survival rates; and these 3 variables were determined to be independent factors for poor progression-free survival rate in multivariate analysis were not exclusive in young ages when compared to late adulthoods, high cyclin D1 expression and FGFR1 amplification were noted with similar frequencies as previously reported in smoking-related HPV-negative HNSCC. 10 Considering young age SCC is more related to both HPV-negativity and neversmoker, alterations of these oncogenes might involve in the tumorigenesis of young age SCC through somewhat different manners from the usual HNSCCs occurring in late adulthoods. Furthermore, high cyclin D1 expression showed a tendency related to poor prognosis of young ages in the present study. From these findings, targeting therapy to MET and CCND1 may be useful in the management of young age HNSCCs.
In the present study, we investigated trophomyosin-related kinases (Trk), TrkA, TrkB, and pan-Trk(A+B+C). 32,33 The expression level of TrkA and pan-Trk were not low in both young age and late adulthood patients. Furthermore, young age cases showing high pan-Trk expression showed a tendency related to inferior progression-free survival rate. These findings suggest that Trk inhibitors such as entrectinib (panTrk inhibitors) could be applied to disease control for young age HNSCC patients. 32,33 In the previous study, the prognostic implications of PD-L1 positivity in tumor cells remain controversial in overall HNSCC. 34 When limited to young age HNSCC patients, however, PD-L1 positivity in tumor cells (TC2 + TC3) was not uncommon and was determined to be a poor prognostic factor for tumor progression. In addition to higher PD-L1 (SP142) positivity in tumor cells (TC2 + TC3), higher infiltration of ICOS+ TILs, and higher ratio of FOXP3+ Tregs and ICOS+ TILs relative to CD8+ T cells were more distinct in young age HNSCC. Furthermore, higher ratios of CD163+ TIMs relative to CD8+ TILs were determined to be independent factors for poor progression-free survival rate.
Expression of immune blockade signals such as PD-1, PD-L1, ICOS, LAG3, and CTLA-4 deregulates tumorinfiltrating CD8+ cytotoxic T cells, and upregulation of these signals alters immune surveillance of cancers. 11,12 FOXP3+ Tregs and CD163+ M2 type macrophages are also kwon to be generally protumorigenic and immunosuppressive. The prognostic value of the expression of immune blockade signals on tumor-infiltrating lymphocytes, FOXP3+ Tregs, or CD163+ M2 type macrophages is controversial; they are associated with poor or good prognosis or not associated with prognosis in various cancer types as well as HNSCC. However, most of such studies consider the absolute cell number of immune cells. 26,[35][36][37][38] When considering relative densities to effector CD8+ T cells, characteristic differences were noted, and prognostic implications were identified. Especially, higher infiltration of CD163+ M2 type protumorigenic macrophages relative to effector CD8+ T cells seems to be related to characteristic immune impairment of young age HNSCC. Tobacco smoking, alcohol consumption, and HPV infection are known to be involved in inducing impaired immune surveillance HNSCC. 7,11,12 Considering majority of young age HNSCC are lack of significant exposure to tobacco and alcohol, and HPV infection, the immune surveillance could be somewhat different from that of the usual HNSCC occurring in late adulthoods.
In summary, young age HNSCC occurred mostly in the oral tongue and revealed lower frequency of p16 positivity, higher c-MET expression and MET gene copy gain as well as lower pan-Trk expression. Higher PL-L1 positivity in tumor cells, higher numbers of ICOS-positive TILs, and higher ratio of FOXP3+ Tregs and ICOS+ TILs relative to effector CD8+ T cells were also more associated with the young age patients. PD-L1 positivity in tumor cells and higher ratio of CD163+ M2 type macrophages relative to effector CD8+ T cells revealed a prognostic impact in young age HNSCC patients. Although these clinical, molecular, and immunologic pictures are described in overall HNSCC regardless of age group in our previous studies. 19,34,38,39 the present results might help to understand the tumorigenesis of young age HNSCC, especially in oral tongue sites and to improve patient outcome through different treatment strategies.