Molecular drivers of oral cavity squamous cell carcinoma in non-smoking and non-drinking patients: what do we know so far?

Oral cavity squamous cell carcinoma (OCSCC) is one of the most common head and neck cancers worldwide. It is well known that risk factors for OCSCC include tobacco and excess alcohol consumption. However, in recent years, OCSCC incidence has been increasing in patients without these traditional risk factors. The cause of this increase is unclear and various genetic, environmental, and infectious factors have been hypothesized to play a role. Additionally, there are expert opinions that oral cancer in non-smoking, non-drinking (NSND) patients have a distinct phenotype resulting in more aggressive disease presentation and poorer prognosis. In this review, we summarize the current state of knowledge for oral cavity cancer in patients without traditional risk factors. ©Copyright: the Author(s).

Introduction

Oral cavity squamous cell carcinoma (OCSCC) is the most common head and neck malignancy.[1] With annual domestic projected incidence of 35,310 new cases and 7,110 deaths in 2020 alone,[2] OCSCC has a significant impact on populations in the United States and around the world. According to data collected by the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program between 2010 and 2016, the 5- year survival rates are estimated at 85.1%, 66.8%, and 40.1% for localized, regional, and distant metastatic OCSCC, respectively.[3]

Current treatment regimens for OCSCC include surgical resection, followed by adjuvant radiation, chemotherapy, or chemo-radiotherapy, depending on the disease stage.[4] Cetuximab (anti-EGFR monoclonal antibody), is the only approved standard of care targeted therapy for OCSCC.[5] However, the response rate to cetuximab is low, and majority of patients develop resistance or relapse even after an initial response. Given the limited success of targeted treatment and cytotoxic chemotherapy, the current clinical focus has turned to immunotherapy with antibodies targeting T cell inhibitory receptors that function as immune checkpoints, such as programmed death 1 (PD-1).[6,7] However, despite recent therapeutic advancements and numerous clinical trials underway, the overall survival of OCSCC is still hovering around 50%.[8]

Consumption of alcohol and tobacco products, poor dental hygiene, and chewing betel quid and areca nut increase risk of OCSCC.[9-14] Although human papillomavirus (HPV) is a prominent risk factor in oropharyngeal cancer, the rate of HPV infection in OCSCC is low, and its significance remains debatable.[12,13,15,16] According to the Centers for Disease Control and Prevention (CDC), cigarette smoking among US adults has reached an alltime low of 13.7% in 2018, a decline of approximately two-thirds over the last 50 years.[17] While alcohol consumption has not experienced such a drastic decline, it has also decreased sharply over the past couple of decades. Surprisingly, however, epidemiological studies have shown a steady rise in the incidence of head and neck squamous cell carcinomas (HNSCCs) in non-smoking, nondrinking (NSND) patients, especially young adults.[9,18,19]

While several large sequencing studies provided a comprehensive landscape of molecular alterations that may serve as prognostic and therapeutic biomarkers in OCSCC,[20-22] most of these studies included heterogeneous patients’ population with different risk factors. As such, molecular changes that drive the tumorigenesis in NSND patients remain scarce. Although more recently several studies have attempted to evaluate the clinical characteristics of OCSCC in non-smoking (NS) patients,[10,23-25] the clinicopathologic data regarding the frequency of such tumors and age of tumor onset are conflicting,[10,23,25,26] and relatively few molecular drivers of the progression of OCSCC in NSND patients are currently recognized. In this overview, we summarize the current state of knowledge regarding the clinicopathologic, survival, and molecular characteristics of OCSCC patients without traditional risk factors. Better understanding of complex biological processes underlying this unique subset of patients would allow for improved diagnosis, risk assessment, management, and ultimately prevention.

Methods

Relevant literature was identified using the following PubMed search strings:

“oral squamous cell carcinoma” OR “oral cancer” OR “tongue cancer” OR “gum cancer” OR “oral cavity cancer” OR “floor of mouth cancer” AND nonsmok* AND nondrink* returned 41 results (including hyphenated terms “non-smok*” and “nondrink*). Asterisks indicate wildcard endings (e.g., nonsmok *includes “nonsmoking” and “nonsmoke”).

“HNSCC” AND nonsmok* AND nondrink* returned 69 results, 28 of which were non-overlapping with search #1, for a total of 97 publications. HNSCC denotes “head and neck squamous cell cancer” or “head and neck squamous cell carcinoma”.

Articles were excluded if only tangentially related to the subject matter, not available in English, dealt primarily with anatomic sites outside the oral cavity, or were primarily studies about tobacco and alcohol as risk factors for disease (e.g., using healthy NSND controls). After exclusion, 48 articles remained. To provide necessary context and background information, additional literature was included at the discretion of the authors.

Results

Epidemiology of oral cavity squamous cell carcinoma in non-smoking, non-drinking populations

Common OCSCC risk factors include tobacco, excess alcohol consumption, and in many regions of the world, betel quid chewing. [27] Prior US data indicate that most patients are male, between 55 and 64 years of age, with higher incidence among Black and Hispanic populations.[3] Over the past few decades, however, the incidence of OCSCC is rising globally among NSND patients.[18] Evidence suggests a bimodal age distribution, with most NSND OCSCC found in distinct ‘younger’ and ‘older’ age cohorts when compared to traditional smoking/drinking (SD) patients.[28-31] For example, in a study of 128 SD and 41 NSND patients with new or recurrent OCSCC, Koo et al. noted a bimodal age distribution with peaks at 50-59 and 70-79 years in NSND, but a single peak at 60-69 years in SD patients.[24] Further, in a retrospective cohort study of 172 NSND and 1131 ever-smoking ever-drinking (ESED) patients newly diagnosed with HNSCC, Dahlstrom et al.[28] found that 41% of NSND were under 50 years old, compared with only 22% of ESED. For those over 70, these numbers were 18% and 15% respectively. In another study that reported perineural invasion (PNI) as an independent prognostic factor for disease specific survival in young patients with OCSCC, presence of PNI was not associated with either tobacco or alcohol consumption.[32] Additionally, OCSCC incidence is rising among females.[19,24,29,33] Patel et al. examined SEER data for OCSCC from 1975 to 2007, and found that disease incidence decreased across this time period except in patients aged 18-44.[19] Among these patients, white females experienced the steepest rise in OCSCC, with an annual percentage increase of 2.2% during this time period. Notably, among young white women within the 18-44 age group, the annual percentage increase was even more prominent at 4.0%.[19] Evidence for a similar pattern in elderly white women comes from a crosssectional study of 1633 HNSCC patients: elderly white NSND women had higher rates of OCSCC than their SD counterparts, who tended to have cancers at other sites.[34] These observations were further supported by a study of 195 NSND HNSCC patients in the Netherlands.[35] In summary, in the past several decades, OCSCC incidence has been rising among individuals who do not fit the expected demographic or risk factor profiles.

Genetic aberrations

Various studies have explored potential genetic contributors to OCSCC in atypical patients. While the conclusions of these reports have been varied, some studies reported unique genetic or genomic characteristics of OCSCC in the NSND population. Unsurprisingly, some of the mutated genes are related to apoptosis and/or cell cycle progression signaling networks. In a study of 505 OCSCC patients (including 201 NS 230 ND patients), Tang et al.[36] identified a polymorphism in caspase-8 (a gene regulating apoptosis) that appears to be a risk factor for OCSCC in NS and ND patients. Specifically, a CASP8 genotype containing a single nucleotide polymorphism (SNP) rs1306963 conferred an increased risk of OCSCC in NS (odds ratio (OR) 1.6) and ND (OR 2.2) patients.

Another study of 282 patients found an association between a CCND1 gene polymorphism at codon 242 (exon 4) in ND patients only (OR 7.5).[37] While mutations in CCND1 (which encodes the cyclin D1 protein) were reported to play role in oncogenesis of several types of cancer including HNSCC,[38] this study examined all squamous cell cancers of the upper aerodigestive tract, so the applicability of this polymorphism to OCSCC specifically is yet to be determined.

DEC1, a gene of unknown function that is often down-regulated in esophageal cancer,[39] has also been implicated in OCSCC. In a case-control study of 1111 HNSCC patients, Huang et al. (2010)[40] examined four SNPs in DEC1, with a particular focus on a c.606 T>C mutation. They found a protective effect of the c.606CC homozygous genotype in NS, ND, and younger patients, although notably the study included large numbers of both OCSCC and OPSCC. While DEC1 aberrations may pose as potential OCSCC biomarkers for NSND individuals, the molecular mechanism of this protective effect remains to be elucidated.[40]

A recent study of 448 HNSCCs from The Cancer Genome Atlas (TCGA) dataset found acquired uniparental disomy (aUPD) regions that were associated with cancer in NSND patients.[41] Specifically, aUPD regions on chromosome 5q occurred more frequently in ND than alcohol-using patients, and in NS than smoking patients. This association was observed in the entire HNSCC cohort as well as in the subset of oral cavity cancers.[41] 5q deletions have been associated with myelodysplastic syndromes and are thought to involve deletions of multiple tumor suppressor genes.[42,43]

Adding another twist to this already complicated picture, a large study of over 2000 HNSCC cases identified 10 E2F transcription factors 1 and 2 (E2F1 and E2F2) “risk genotypes” that did not correlate with cancer risk alone, but when 5 or more were combined, increased the risk of cancer among NSND and young patients.[44] While these observations suggest that E2F1 and E2F2 genetic variants may jointly play roles in NSND head and neck carcinogenesis, it should be noted that this study included tumors from different histological sites, and the role of these “risk genotypes” in OCSCC requires further clarification.

In contrast, Pickering et al. found no meaningful differences between OCSCC in SD and NSND populations.[45] The authors performed an analysis of whole-exome sequencing profiles of patients from TCGA-HNSC dataset, as well as oral tongue cancers collected from young (<46) and older patients treated at the MD Anderson cancer center. While in both datasets more mutations were seen in older patients, known cancer driver genes (e.g., TP53, FAT1, and CASP8) were proportionately mutated in both groups, and no specific aberrations defining the NSND population were identified. Similarly, whole-genome copy number analyses did not identify any differences between the two cohorts.[45]

In summary, genetic changes potentially associated with OCSCC in NSND include aberrations in CCND1, CASP8, DEC1, and 5q region. Notably, tumor suppressor genes such as TP53 are mutated at similar rates in both NSND and SD cancer patients. While there is no single predominant pattern of cancer progression in NSND, it may be necessary for multiple aberrations to be present concurrently to promote carcinogenesis.

Transcriptomic changes

In addition to the aberrations on DNA level, differences in gene expression patterns have also been reported in association with NSND cases. Soares et al. examined HNSCC tumor tissue from 47 NSND and 37 SD patients matched for stage, grade, and site.[46] They assessed the expression of glutathione S-transferase π (GTSP1), a detoxification enzyme upregulated in the presence of carcinogens. They found that while GTSP1 was expressed in tumors from both NSND and SD patients, its expression level was substantially higher in tumor margins from SD individuals. As GTSP1 plays a protective role in patients exposed to tobacco and alcohol carcinogens, the authors suggest that low GTSP1 expression in NSND individuals might make them susceptible to other carcinogens. This may be a mechanism of carcinogenesis in the absence of tobacco/alcohol. Nevertheless, studies in larger cohorts are required to support or refute this observation.

A different study that analyzed 55 cancer tissue samples showed that positive staining for PD-L1 protein was more prominent in OCSCC samples collected from NSND patients.[47] Although sample sizes were small, this observation suggests that NSND may represent a promising candidate group for immunotherapy with PD-1/PD-L1 axis blockade.

Another group analyzed gene expression profiles derived from a mixed cohort of 89 SD and 15 NSND patients, and found 49 genes that were differentially dysregulated (28 overexpressed and 21 downregulated) between the two groups of patients.[48] The genes involved had a wide range of functions; notable examples include upregulation of IFN-g related genes and downregulation of genes involved in the NFKB pathway in NSND patients.

Interestingly, though dysregulation of the tumor suppressor gene p53 is associated with many types of cancer, a study including 33 NSND and 29 SD HNSCC patients reported no differences in p53 expression (assessed by immunohistochemical staining) between the two cohorts,[49] with another small study of 11 NSND,[50] further supporting the suggestion that p53 does not play a major role in NSND tumorigenesis.

While elevated expression of a gene does not necessarily correspond with an increase in the activity of its product, transcriptomic changes play crucial role in driving OCSCC evolution.[51,52] Due to the limited number of currently available NSND datasets it is challenging to dissect the molecular pathways underlying cancer progression in NSND patients, but as more data sets become available, a comprehensive transcriptomic analysis may aid in identifying functional categories or pathways that may be relevant as therapeutic targets.

Epigenetic changes

As mutations alone are insufficient to explain pervasive transcriptomic changes seen in OCSCC malignancies, it was suggested that epigenetic modifications may be central to gene expression dysregulation during oral carcinogenesis.[53] One study, utilizing a cohort of HNSCC tumors collected from 136 patients, found 8 microRNAs differentially regulated between drinking and ND patients.[54] However, other studies have failed to find such differences. For example, a study of 90 SD and NSND participants (in both cancer and control groups) found higher levels of p15 methylation in cancer patients, regardless of tobacco/alcohol use. Notably, a higher prevalence of p15 methylation was found in histologically- normal surgical margin epithelia of SD compared with NSND, suggesting that p15 methylation is not a major driver of tumorigenesis in the NSND population.[55] While epigenetic alterations are common in tobacco-associated oral carcinogenesis, epigenetic changes specifically associated with NSND OCSCC patients are less understood. Further studies with larger sample sizes are needed to define specific epigenetic alterations that have vital importance in NSND patients.

Infectious factors

HPV infection is hypothesized to be a contributing factor to the increasing rates of OCSCC in NSND patients, analogous to HPVassociated OPSCC.[56-58] It is acknowledged that patients with HPVassociated OPSCC have a distinct epidemiologic profile, with younger age at presentation, absence of strong smoking history, and higher socioeconomic status. The etiologic contribution of HPV in OCSCC is less understood.[12,13,15,16,56,59] One study of 53 patients with oral tongue cancer showed an association between HPV positivity and NSND status, however sample sizes were small as only 4 of the 53 tumors tested positive for high-risk HPV.[60]

In an international study of 3680 samples, Castellsague et al. estimated HPV positivity rates to be 22.4% for OPSCC but only 4.4% for OCSCC, suggesting that HPV plays a more important role in OPSCC than OCSCC.[61] Another study examined samples from 45 HNSCC patients younger than 40. While the association between HPV and cancer was found when the entire cohort was analyzed together, only 2 of 18 OCSCCs were positive for HPV.[62] In a larger study, Belobrov et al. performed immunohistochemical staining for biomarkers such as p53, p16, cyclin D1, and EGFR using samples from 129 OCSCC patients.[58] While overexpression of these biomarkers was seen in many of the analyzed specimens, p16 overexpression was significantly stronger in NSND cases under the age of 70. Expression of p16 has been used as a proxy for HPV-associated oncogenesis, as inactivation of Rb by the HPV viral protein E7 leads to p16 upregulation.[63] In a more recent study, Dediol et al. (2016) found that p16 overexpression was more frequent in NSND than in SD with OCSCC.[11] Although these findings support the role of HPV in a subset of younger NSND patients, the accuracy of p16 as a marker for HPV-associated oncogenesis in OCSCC remains debatable. In a study of 409 OCSCCs, Lingen et al. found a positive predictive value of only 41.3% for p16 expression when compared with PCR detection of HPV E6/E7 mRNA, with only 5.9% of OCSCCs found to be related to HPV infections.[13] Mirghani et al. further supported the notion that p16 immunohistochemistry is an unreliable surrogate marker for HPV oncogenesis in OCSCC, given its low specificity,[12] whereas Tomo et al. demonstrated that p16 levels may be high even in the absence of HPV infection.[64] Moreover, presence of HPV DNA fails to accurately represent a transcriptionally active viral process in OCSCC.[14,65] Taken together, while evidence suggests that HPV may play a significant etiological role in NSND for OPSCC, HPV does not appear to be a main driver of OCSCC.

Foy et al. have recently postulated that other viruses, such as HSV-2, may cause OCSCC via changes in chronic inflammatory gene expression (e.g., JAK2) and epigenetic deregulation.[66] Theoretically, such changes could persist after clearance of the virus. While another study detected no significant viral RNA transcripts in a set of 68 oral cancers from NSND patients,[67] this may be a fruitful avenue for investigation into mechanisms of OCSCC in NSND populations.

In summary, although the role of HPV in OPSCC is welldemonstrated, evidence for HPV-dependent carcinogenesis in OCSCC is equivocal. In the future, other oncogenic viruses might be explored as contributors to OCSCC oncogenesis.

Other factors

Environmental factors have also been explored as potential contributors to OCSCC in NSND patients. One such factor is metallic dental hardware. In a retrospective analysis of 54 NSND patients with OCSCC, Yesensky et al. administered a dental health questionnaire to identify a history of metallic hardware and orthodontic procedures.[30] Forty of the 54 patients (74%) had a history of metal-containing dental hardware, including braces in younger patients, and crowns, implants, or dentures in older ones. A retrospective analysis of 31 OCSCC patients found that lesions occurring next to dental implants were more common in female and NSND patients.[68] The mechanism for hardware-associated OCSCC is unclear, but DNA damage from cytotoxic metal ions,[69] and chronic mucosal irritation from adjacent hardware[30] have been proposed. Poor oral hygiene and occupational exposures may also be involved.[70]

Yan et al. conducted a case-control study of 319 NSND OCSCC patients and 994 NSND controls using an “environmental exposure index” consisting of nine variables, including aspects of diet, dental hygiene, hardware, and exposure to other potential carcinogens. [71] They found a linear relationship between the index score and the risk of OCSCC, in particular identifying “recurrent oral ulceration” as the strongest risk factor. Interestingly, these authors noted that the effects of these exposures appeared to be more prominent in patients with a family history of cancer, suggesting that genetic predisposition may also contribute to OCSCC in NSND cases.[71] Several other studies have found compelling evidence for dietary effects. A case-control study of 421 NSND OCSCC patients and 1398 NSND controls reported protective effects of tea and milk consumption.[72] Another case-control study of 236 oral cancer patients and 300 controls reported a higher level of erythrocyte membrane fatty acids (a proxy for dietary fatty acid intake) in the control group, and found that the association between fatty acid levels and cancer was stronger in NSND patients, further supporting that dietary factors can affect cancer risk in NSND populations. [73]

In summary, while environmental factors certainly play a role in the development of OCSCC in NSND, more work is needed to determine the contributions of these various factors to oral tumorigenesis and their interactions with smoking and alcohol.

Clinical behavior

It has been suggested that oral cancers in NSND patients are more aggressive, but this has not been conclusively determined. Evidence in favor of this suggestion includes a study of 76 oral leukoplakia patients, which showed faster and more frequent progression in NSND and female patients.[10] A meta-analysis of 26 articles with a total of 2532 oral cancer patients (not limited to the oral cavity) found that those with PD-L1 overexpression had lower disease-specific and disease-free survival; these patients also tended to be females, NS, and/or ND.[74] Others have found a higher rate of recurrent/persistent disease and poorer 5-year survival in elderly NSND women vs SD male and female age-matched controls,[23] and proposed that elderly NSND females,[24] or young NSND patients[75] constitute a clinically distinct subset of OCSCC. Interestingly, NSND patients tend to develop malignancies at different histological sites than SD individuals, with more tumors located in the cheek mucosa and alveolar ridge, which also suggests different clinical behavior.[76,77]

The true prognostic impact of OCSCC in NSND patients is likely more complicated. Bachar et al. studied a group of 175 SD and 116 NSND patients with oral tongue cancer, and showed similar rates of overall, local, and regional recurrence in both cohorts. Disease-specific and disease-free survival were also similar. When the cases in the cohort were limited to patients under 40, however, NSND patients had significantly worse disease-free and overall survival.[25] The authors suggest that, because of these differences in prognosis, there may be a distinct mechanism of carcinogenesis in this younger subset of NSND patients with more aggressive disease. On the other hand, some studies have found no association between tobacco/alcohol use and prognosis.[26,78] For example, a historical cohort study of 120 patients (66 ever-smoking (ES)) by Durr et al. found no significant differences in overall survival or recurrence-free survival between NS and ES patients with oral tongue cancer, although worse overall survival in NSND patients trended towards significance.[79] Finally, a series of studies suggest that NSND patients with OCSCC may have better survival than their SD counterparts. A study of 218 HNSCC patients (24 NS, 50 ND) found worse survival among smoking and/or drinking patients after chemoradiation, although it should be noted that this study included cancers from various histological locations.[80] In study of 1165 oral cancer patients, NSND had better overall and disease-specific survival than SD patients.[81] Similarly, a retrospective comparative study of patients with oral lichen planus associated OCSCC (OLP-OCSCC) and non-OLP-OCSCC reported that patients with OLP-OCSCC were more likely to be NSND and female, and had better overall and disease-specific survival. Interestingly, however, these OLP-OCSCC patients had an increased risk of recurrence and second primary tumors.[82]

Assuming true differences in prognosis and disease progression between these patient subsets, it is reasonable to suggest that distinct mechanisms could underlie carcinogenesis in these subsets. Indeed, it was reported that changes in copy number variation burden predicted prognosis in young oral cancer patients but not older ones, suggesting that copy number variation may uniquely contribute to carcinogenesis and/or disease progression in NSND cases.[83] Another hypothesis is that loss of heterozygosity (LOH) at tumor suppressor loci in premalignant lesions may contribute to malignant transformation. This is supported by the finding that particular “high risk” LOH profiles may predict the risk of progression in a subset of premalignant oral lesions.[84] While LOH of 3p and/or 9p (containing p16INK4a and p14ARF) was reported to predict progression of oral leukoplakia in both NSND and SD patients, there is some evidence that the contribution of LOH to malignant transformation is greater in NSND subset. For example, Rock et al. (2018) found that NS patients with oral epithelial dysplasia had a higher risk of progression to cancer than those who smoked (38-fold higher for floor-of-mouth lesions).[84] Based on these observations, the authors conclude that LOH may be a more important factor for progression in NS than in smoking patients. While these studies provide early evidence for different molecular drivers in NSND OCSCC patients, the mechanisms underlying cancer initiation and progression in this group of patients remain elusive.

Comparing the aggressiveness of cancers in NSND and SD patients may be difficult due to the heterogeneity of the NSND population. The study that reported worse survival in NSND patients indicated the greatest effects in patients under 40 years old.[25] In contrast, both studies describing better survival among NSND patients had a NSND cohort that was older, on average, than the SD cohort. Consequently, it is tempting to rationalize that there are distinct populations of NSND OCSCC patients. A younger population may be genetically predisposed to OCSCC, with a distinct mechanism of carcinogenesis and more aggressive disease. In contrast, an older group might develop cancer via mechanisms similar to the SD patients, in which case their NSND status may be protective, as it confers a lower cumulative exposure to carcinogens that would worsen the disease. In this group, factors such as dental hardware and environmental exposures may play a more prominent role. This hypothesis would be in accordance with data showing a bimodal age distribution among NSND OCSCC patients (see “Epidemiology of OCSCC in NSND populations”).

Conclusions

In this overview, we have discussed the recent rising incidence of OCSCC in patients who lack traditional risk factors such as smoking and drinking, and who skew towards a younger, female demographic. Attempts to explain this phenomenon have not been conclusive, in part due to the relatively small number of NSND cases and high heterogeneity of the disease (Table 1).

Table 1.

Clinical/preclinical studies highlighting molecular profile and treatment outcomes of patients with non-smoking, non-drinking oral cavity squamous cell carcinoma

Authors (Year)	Study design/objective	Treatment modality	Sample size (n)	NSND cases (n)	NSND prevalence (%)	Clinical significance	Ref
Yang et al. 2021	Retrospective cohort; treatment response and disease-specific prognosis	Surgery and adjuvant CRT	353	86	24.4	* NSND OCSCC unrelated to HPV status. * NSND patients showed better locoregional control and disease-specific survival than SD patients. * Higher expression of p53, p63, Ki-67 in SD compared to NSND population.	(85)
Adeoy et al. 2021	Retrospective cohort; treatment response and disease-specific prognosis	Surgery and adjuvant CRT	313	171	54.6	* No significant difference between recurrence free survival and disease-specific survival in NSND compared to SD patients. * Better overall survival in NSND patients compared to SD.	(86)
Koo et al. 2021	Molecular profiling	Not specifically indicated	176	59	33.5	* Higher prevalence of somatic CDKN2A mutations, EGFR amplifications and BRCA2 deletion in NSND OCSCC patients. * Unrelated to HPV infection. * No significant difference in overall survival between NSND and SD patients.	(87)
Bao et al. 2020	Clinicopathological characteristics and disease-specific prognosis	Surgery and adjuvant CRT	1165	646	55.5	* NSND patients associated with better overall survival and disease-specific survival than SD patients	(81)
Lenouvel et al. 2020	Retrospective cohort; molecular profiling	Not specifically indicated	55	8	14.5	* PD-L1 overexpression in NSND patients, potentially contributing to early tumorigenesis.	(74)
Brennan et al. 2017	Retrospective cohort; clinicopathological characteristics and molecular profiling	Not specifically indicated	528	184	34.8	* NSND OCSCC associated with distinct molecular profile, low overall somatic mutation rates in these patients compared to SD patients. * NSND group presents CIMP high and low phenotypes. * Significant differential hypermethylation profile in NSND group compared to SD. * Overexpression of PD-L1, PD-L2 and CD8+ T cells in NSND group.	(88)
De Angelis et al. 2018	Retrospective cohort; clinicopathological characteristics and disease-specific prognosis	Surgery and adjuvant CRT	287	70	24.4	* Elderly NSND females are associated with poor disease-specific survival compared to elderly SD patients. * No significant difference between disease-specific survival in NSND compared to SD patients.	(23)
Soares et al. 2017	Retrospective cohort; molecular profiling	Not specifically indicated	1633	47	2.87	* No significant difference in expression of GTSP1 in tumor of NSND and SD patients. * Lack of carcinogen detoxification could be associated with carcinogenesis in NSND patients.	(46)
Foy et al. 2017	Retrospective cohort; molecular profiling	Not specifically indicated	213	93	43.7	* Increased expression of PD-L1, IDO-1, CD8+ T cells in NSND patients. * Associated with low somatic mutation, copy number variation and incidence of chr 11q13 amplification.	(89)
Dediol et al. 2016	Retrospective cohort; clinicopathological characteristics and disease-specific prognosis	Surgery	1074	103	9.6	* No significant difference in disease specific survival, locoregional recurrence and metastasis between NSND and SD patients.	(11)
Fan et al. 2014	Retrospective cohort; clinicopathological characteristics and disease-specific prognosis	Surgery and adjunctive CRT	100	54	54	* No significant difference in recurrence free survival or overall survival between NSND and SD patients.	(26)
Farshadpour et al. 2012	Retrospective cohort; clinicopathological characteristics and molecular profiling	Not specifically indicated	104	15	14.4	* Differential gene expression signatures in NSND patients. * Increased expression Major histocompatibility complex (MHC) and downregulation of NF B related genes in NSND patients.	(48)
Bachar et al. 2011	Retrospective cohort; clinicopathological characteristics and disease-specific prognosis	Surgery and adjuvant CRT	291	116	39.9	* No significant difference in disease-free survival or overall survival between NSND and SD patients. * Young patients associated with poor overall survival.	(25)
Albuquerque et al. 2011	Retrospective cohort; clinicopathological characteristics	Not specifically indicated	354	146	41.2	* NSND patients associated with a distinct clinicopathological profile compared to SD patients.	(90)

NSND: non-smoking, non-drinking; OCSCC: oral cavity squamous cell carcinoma; CRT: chemoradiotherapy; PD-L1: programmed death-ligand 1; EGFR: epidermal growth factor receptor; MHC: major histocompatibility complex; HPV: human papillomavirus; PD-L2: programmed death-ligand 2; IDO-1: indoleamine 2,3-dioxygenase.

Among the factors explored as potential contributors to OCSCC in atypical patients are: genetic factors, infectious factors (particularly HPV), and environmental factors (e.g., chemical exposure and dental hardware). Studies examining genetic underpinnings of OCSCC in atypical patients have identified mutations, gene expression changes, and epigenetic factors. Due to variation in the study populations, these studies may not be directly comparable, and more work in larger cohorts is needed to elucidate genetic contributions to OCSCC in atypical patients. Additionally, while there is strong evidence that HPV is an etiologic factor for a large proportion of OPSCC in NSND patients, this is unlikely to be the case for OCSCC (Table 1).

Finally, it has been suggested that oral cancers in atypical and typical patients differ by clinical course, with more aggressive disease being prevalent among atypical cases. While reports are controversial and this evidence is not yet conclusive, cumulative data suggest that OCSCC in atypical patients may indeed be more aggressive, and that this may be associated with genetic mechanisms (e.g., copy number variation, LOH, and other genomic aberrations) that set OCSCC in atypical patients apart. Nevertheless, a comprehensive analysis of genetic alterations in a large cohort of NSND cases is required to identify the genetic drivers of cancer progression in this unique group of patients.