Does Buccal Cancer Have Worse Prognosis Than Other Oral Cavity Cancers?

The Laryngoscope VC 2014 The American Laryngological, Rhinological and Otological Society, Inc. Does Buccal Cancer Have Worse Prognosis Than Other Oral Cavity Cancers? P. Ryan Camilon, BA; William A. Stokes, BS; Colin W. Fuller, MD, MS; Shaun A. Nguyen, MD, MA; Eric J. Lentsch, MD Objectives/Hypothesis: To determine whether buccal squamous cell carcinoma has worse overall survival (OS) and disease-specific survival (DSS) than cancers in the rest of the oral cavity. Study Design: Retrospective analysis of a large population database. Methods: We began with a Kaplan-Meier analysis of OS and DSS for buccal versus nonbuccal tumors with unmatched data, followed by an analysis of cases matched for race, age at diagnosis, stage at diagnosis, and treatment modality. This was supported by a univariate Cox regression comparing buccal cancer to nonbuccal cancer, followed by a multivariate Cox regression that included all significant variables studied. Results: With unmatched data, buccal cancer had significantly lesser OS and DSS values than cancers in the rest of the oral cavity (P <.001). After case matching, the differences between OS and DSS for buccal cancer versus nonbuccal oral cancer were no longer significant. Univariate Cox regression models with respect to OS and DSS showed a significant difference between buccal cancer and nonbuccal cancer. However, with multivariate analysis, buccal hazard ratios for OS and DSS were not significant. Conclusions: With the largest series of buccal carcinoma to date, our study concludes that the OS and DSS of buccal cancer are similar to those of cancers in other oral cavity sites once age at diagnosis, tumor stage, treatment, and race are taken into consideration. The previously perceived poor prognosis of buccal carcinoma may be due to variations in tumor presentation, such as later stage and older patient age. Key Words: Buccal cancer prognosis, survival analysis, survival comparison, oral cavity cancer, oral cancer. Level of Evidence: 2b Laryngoscope, 124:1386 1391, 2014 INTRODUCTION In the United States, over 13,000 new cases of cancer in the oral cavity were diagnosed in 2013, the majority being oral squamous cell carcinoma (OSCC). 1 In North America and Western Europe, buccal squamous cell carcinoma (BSCC) accounts for approximately 10% of oral cavity cancers. 2 BSCC is the most common form of oral cancer among the Indian population of Southeast Asia and can account for up to 37% of intraoral cancer cases in Taiwan. 3,4 The higher frequency of BSCC in Southeast Asia is due to the endemic practice of chewing betel quid (consisting of betel leaf, areca nut, and lime), exposing the buccal mucosa to carcinogens. In contrast, tobacco and alcohol use are considered as the major risk factors for BSCC in the United States. Given its relative From the Department of Otolaryngology Head and Neck Surgery (P.R.C., C.W.F., S.A.N., E.J.L.), College of Medicine (W.A.S.), and Hollings Cancer Center (E.J.L.), Medical University of South Carolina, Charleston, South Carolina, U.S.A. Editor s Note: This Manuscript was accepted for publication October 25, 2013. The authors have no funding, financial relationships, or conflicts of interest to disclose. Send correspondence to P. Ryan Camilon, Clinical Research Fellow, Department of Otolaryngology Head and Neck Surgery, Medical University of South Carolina, 135 Rutledge Ave. MSC 550, Charleston, SC 29425. E-mail: camilon@musc.edu DOI: 10.1002/lary.24496 1386 rarity in the United States, all studies of BSCC are limited to small numbers of cases from single institutions. In general, most studies evaluate how various tumor characteristics and variations in treatment affect the prognosis of BSCC. 2 Almost all studies agree that buccal carcinoma, due to its invasive tumor behavior and high incidence of locoregional recurrence, has a poor prognosis; however, there is a paucity of studies that thoroughly assess buccal cancer prognosis in comparison to that of other oral cavity cancers. 2,5 7 A recent British study attempted to address this topic with data that showed no significant difference between the survival of BSCC and other oral cavity cancers. 8 Unfortunately, the authors had a small sample size and only evaluated survival secondarily; they compared buccal cancer survival to nonbuccal oral cancer survival without controlling for variations in tumor stage and other confounding prognostic factors. To date, there are no studies that primarily focus on a survival comparison. Our purpose is to determine whether BSCC has worse overall survival (OS) and disease-specific survival (DSS) than cancers in the rest of the oral cavity. MATERIALS AND METHODS Data Sources The Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute is a collection of

TABLE I. Patient Demographics. Characteristic Oral Cavity* Buccal Total P Sex Female 40.00% 46.50% 4,505 <.05 Male 60.00% 53.50% 6,629 <.05 Total 10,309 825 11,134 Stage I 36.30% 27.20% 3,231 <.05 II 17.50% 21.20% 1,612 III 15.40% 19.40% 1,421 <.05 IV 30.70% 32.20% 2,790 Total 8,411 643 9,054 UNK 1,898 182 2,080 Treatment None 8.20% 9.60% 45 Radiation 13.60% 14.90% 779 Surgery 49.00% 43.30% 799 <.05 Combined 29.20% 32.20% 9,314 Total 10,137 800 10,937 UNK 172 25 197 Race AI/AK 0.40% 0.30% 914 Asian 6.80% 11.40% 1,510 <.05 Black 7.30% 7.10% 5,363 White 85.50% 81.30% 3,247 <.05 Total 10,214 820 11,034 Other/UNK 95 5 100 Mean age, yr (95% CI) 62.64 (62.34 62.94) 66.87 (67.84-69.90) <.001 *Nonbuccal oral cavity cancers. Interaction test of difference between buccal and oral cavity groups is significant at P <.05. AI/AK and Asian represents American Indian/Alaskan Natives and Asian and Pacific Islander respectively; CI 5 confidence interval; UNK 5 unknown. population-based cancer registries across the United States that publishes cancer incidence and survival data covering approximately 28% of the population. 9 Patient demographics, primary tumor site, tumor morphology and stage, course of treatment, and other statistics are routinely collected. Data are subject to rigorous quality control studies and various assessments. This study was exempt from review by the institutional review board at our institution due to deidentification of patient data. Patient Selection Our study included patients in the SEER database who were diagnosed with OSCC from 2004 to 2009. Any cases with a history of previous malignancy or with multiple primary tumors were excluded from analysis. Primary location and histological status were obtained from the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) variables in both databases. They were then recoded into new variables as demonstrated in Table I. Demographics Race, age at diagnosis, primary location, stage at diagnosis, radiological treatment status, and surgical treatment status were obtained from the SEER database variables RacerecodeWBAIAPI, Agerecodewith1yearolds, ICD-O-3 site recode, American Joint Committee on Cancer (AJCC) stage (20041), Radiation,SumSurgeryStatus, and RXSummSurgPrimSite1998, respectively. Location of the primary tumor was categorized by subsite with 52.0% tongue, 19.9% floor of mouth, 11.3% gingival, 7.4% buccal, 6.7% retromolar trigone, and 2.8% hard palate primary tumors. Tongue, floor of mouth, gingiva, retromolar trigone, and hard palate tumors were grouped together under the oral cavity heading, whereas buccal primary tumors were separated for our analysis. Patients were separated into categories by site of primary tumor and then further stratified according to the individual variables mentioned in Table I. For this study, oral cavity cancer is synonymous with nonbuccal oral cavity subsite cancers unless otherwise specified. Age refers to age at diagnosis unless otherwise specified. Survival Analysis Survival time was obtained through the survivaltimerecode#ofmonths function of the SEER database, whereas OS status and DSS status were obtained from the vitalstatusrecodestudycutoffused and SEERcausespecificdeathclassification functions, respectively. Kaplan-Meier survival models that evaluated buccal verses nonbuccal primary tumors for the oral cavity with respect to OS and DSS for unmatched data were created. Then we performed matching between buccal and nonbuccal primary tumors by race, stage at diagnosis, treatment modality, and age at diagnosis. The treatment categories are exclusive: for example, the patients categorized as having received surgery only received surgery and this category does not include patients who received both surgery and radiation. (combined). Race, stage, and treatment were matched exactly on a case by case basis, whereas age was matched to patients within the same decade of life (e.g., age 40 49 and 50 59 years). After matching, Kaplan-Meier survival models of the oral cavity with respect to OS and DSS were created once again. Patients were divided into groups, with buccal primary tumors as the experimental group and nonbuccal primary tumors as the control. We calculated 2-year and 5-year survival for both groups. Statistical comparisons were performed using a pairwise v 2 test covering the entire stratum. In addition, unmatched cases were analyzed by Pearson correlation to determine the relationship among variables and survival. Sex was found not to correlate with survival in the oral cavity and was excluded from analysis. The correlation was followed by univariate and multivariate Cox proportional hazard regression models to compare buccal cancer to nonbuccal cancer with respect to OS and DSS. The multivariate Cox proportional hazard regression analysis included confounding variables such as stage at diagnosis, treatment, race, and age at diagnosis. P <.05 was considered indicative of a statistically significant difference for all tests. Statistical Analysis All variables were treated as categorical inputs except age, which was continuous. Comparisons were performed with a Pearson v 2 test for categorical variables and Student t test for the continuous variable. All data analyses and matching were performed with SPSS 21.0 (SPSS/IBM, Chicago, IL). RESULTS Demographics A total of 11,134 patients in the SEER database with oral cavity SCC were identified. The subsite 1387

TABLE II. Unmatched Kaplan-Meier Survival Models for the Oral Cavity. Site OS Cumulative Survival Lower 95% CI Upper 95% CI Oral 2-year 66.80% 65.82% 67.78% 5-year 51.10% 49.73% 52.47% Buccal 2-year 60.90% 57.18% 64.62% 5-year 44.10% 39.40% 48.80% Site DSS Cumulative Survival Lower 95% CI Upper 95% CI Oral 2-year 73.40% 72.42% 74.38% 5-year 63.80% 62.43% 65.17% Buccal 2-year 68.00% 64.28% 71.72% 5-year 57.30% 52.60% 62.00% Oral signifies nonbuccal oral cavity cancers; 95% CI represents CI for percentage cumulative survival. P <.001 for both OS and DSS. CI 5 confidence interval; DSS 5 disease-specific survival; OS 5 overall survival. distribution was 7.41% buccal (825 cases) and 92.59% rest of oral cavity (10,309 cases). We found several significant differences in the distribution of variables between our two groups. Namely, buccal cancer patients presented at an advanced age (P <.001), with significantly fewer stage I tumors (P <.05) more stage III tumors (P <.05), had a higher proportion of female patients (P <.001), a higher proportion of patients who defined themselves as Asian (P <.05), and a smaller proportion of patients who received solely surgical treatment (P <.05). See Table I. regression models with respect to OS and DSS showed a significant difference between the buccal site and the rest of the oral cavity. For example, the buccal hazard ratio for the OS model was 1.25 (P <.001). However, in a multivariate analysis with the inclusion of age at diagnosis, stage at diagnosis, treatment, and race, the buccal hazard ratio decreased for both OS and DSS, and was no longer significantly different from that of the rest of the oral cavity, as evident from the CIs which included a value of 0 and p-values that were greater than.05 (Table IV). Survival Analysis Kaplan-Meier survival models of unmatched patients showed a significant decrease in the OS and DSS of buccal patients compared to the rest of the oral cavity. Buccal patients had a 2- and 5-year OS of 60.49% and 44.10%, and 2- and 5-year DSS of 68.00% and 57.30%, respectively. This was significantly lower than the oral cavity 2- and 5-year OS of 66.80% and 51.10%, and DSS of 73.40% and 63.80%, respectively (P <.001; Table II, Fig. 1). After case matching our buccal population to the oral cavity population, we were left with 817 matched pairs. Kaplan-Meier survival models of the matched patients did not show a significant difference between the buccal and oral cavity group OS (P 5.11) and DSS (P 5.18), as shown by overlapping confidence intervals (CIs). The Oral cavity OS fell from 66.80% to 64.20% at 2 years and from 51.10% to 48.10% at 5 years, whereas the buccal OS remained relatively unchanged. The oral cavity DSS fell to 71.50% at 2 years and 61.90% at 5 years (Table III, Fig. 2). Again, the buccal DSS remained relatively unchanged. DISCUSSION Current literature suggests that buccal cancer is aggressive and has poor prognosis. Buccal cancer in the United States was thoroughly described by Diaz and colleagues at MD Anderson Cancer Center. With 119 cases of buccal cancer, their seminal study had the largest sample size in North America and Europe. Diaz et al. Regression Analysis Our Cox proportional hazard regression models mirrored our Kaplan-Meier survival results. Univariate Cox Fig. 1. Unmatched nonbuccal oral cancer versus buccal cancer: Kaplan-Meier overall survival (OS). 1388

TABLE III. Matched Kaplan-Meier Survival Models for the Oral Cavity. Site OS Cumulative Survival Lower 95% CI Upper 95% CI Oral 2-year 64.20% 60.67% 67.73% 5-year 48.10% 43.20% 53.00% Buccal 2-year 60.90% 57.18% 64.62% 5-year 44.10% 39.40% 48.80% Site DSS Cumulative Survival Lower 95% CI Upper 95% CI Oral 2-year 71.50% 67.97% 75.03% 5-year 61.90% 57.39% 66.41% Buccal 2-year 67.90% 64.18% 71.62% 5-year 57.40% 52.70% 62.10% Oral signifies nonbuccal oral cavity cancers; 95% CI represents CI for percentage cumulative survival. P 5.113 for OS; P 5.184 for DSS. CI 5 confidence interval; DSS 5 disease-specific survival; OS 5 overall survival. calculated a 5-year DSS of 63% for all patients and found significantly worse DSS values for increasing stage (P 5.05), node-positive neck disease (P 5.012), and extracapsular spread (P 5.005). 5 In addition, the authors found that 45% of cases had recurrence and added that of 86 patients with N0 necks at presentation, those who did not receive elective neck treatment had a higher percentage of regional recurrence than patients who received an elective neck dissection or radiation (25% vs. 10% respectively, P <.05). With these results, one of their study s conclusions was that patients with buccal mucosa SCC have a worse stage-for-stage survival rate than do patients with other oral cavity sites. Despite their comprehensive evaluation of BSCC, their analyses did not include a comparison of buccal DSS to nonbuccal oral cavity DSS. Many subsequent studies regarding BSCC continued to conduct a detailed assessment of buccal cancer prognostic factors and management strategies, all contributing to the trend that cancer of the buccal mucosa is worse than other oral cavity sites without actually comparing survival rates in their analyses. 2,6,10 Recently, a British study contradicted this trend. Shaw and colleagues reviewed 81 patients with BSCC and compared them to 401 patients with other oral site SCC in search of significant differences in prognostic factors. They found that buccal cancer presented less often as early stage tumors and was more likely to have involved margins after primary resection (16% of buccal tumors were T1 vs. 28% of nonbuccal tumors, P 5.02; 26% of buccal tumors had involved margins vs. 15% nonbuccal tumors, P 5.02). 8 In their tumor characteristic analysis, they conducted a calculation of DSS for both buccal and nonbuccal oral tumors and found that the difference between DSS values was insignificant (70% for buccal vs. 75% for nonbuccal, P 5.34). The authors concluded that their results offered no evidence of the poor outcome of buccal cancers previously reported by other studies. Despite their novel stance on buccal cancer, their study had severe limitations in addition to being low-powered and of a single institution, as all of these studies are. All of their patients had OSCC that had previously been resected primarily; therefore, their characteristics of recurrent tumors are not generalizable to primary cancers. In addition, when calculating DSS, the authors did not control for tumor stage, patient age, and treatment (cases varied in use of free flap, neck dissection, and postoperative radiotherapy), all of which affect DSS. Our study is the first to compare the OS and DSS of primary buccal cancer to other oral site cancers. Initially, the stance that buccal cancer has worse prognosis than other oral cancers appears true, with BSCC having significantly worse 2-year and 5-year OS (60.9% and 44.1% vs. 66.8% and 51.1% for nonbuccal, respectively). However, after matching for age, race, tumor stage, and treatment modalities, this difference loses statistical significance (P 5.11 for OS, P 5.18 for DSS). The same pattern of having a significant difference before case matching, then loss of significance after matching occurs Fig. 2. Matched nonbuccal oral cancer versus buccal cancer: Kaplan-Meier overall survival (OS). 1389

TABLE IV. Comparison of Univariate and Multivariate Analysis of Buccal and Nonbuccal Oral Cancer. OS P HR Lower 95% CI Upper 95% CI Univariate Buccal.003* 1.252 1.080 1.452 Multivariate Buccal.375 1.070 0.921 1.243 DSS P HR Lower 95% CI Upper 95% CI Univariate Buccal <.001* 1.265 1.114 1.437 Multivariate Buccal.527 1.042 0.916 1.186 Oral signifies nonbuccal oral cavity cancers; 95% CI represents CI for hazard ratio; multivariate signifies the inclusion of patient age, race, tumor stage, and treatment modality in the analysis. *Signifies a statistically significant difference between the HRs of Buccal cancer and Oral (nonbuccal) cancer. CI 5 confidence interval; DSS 5 disease-specific survival; HR 5 hazard ratio; OS 5 overall survival. for DSS. Further analysis with multivariate Cox regressions supported these results. Univariate analysis (buccal location vs. nonbuccal oral cavity location only) shows a significant difference in hazard ratio, 1.25 OS and 1.26 DSS for buccal compared to reference nonbuccal, 95% CIs [1.080 to 1.452] and [1.114 to 1.437] respectively.). Multivariate analysis which included age, race, tumor stage, and treatment in addition to buccal versus nonbuccal primary site, resulted in the loss of statistical significance (95% CIs [0.921 to 1.243] for Buccal OS and [0.916 to 1.186] for Buccal DSS). Therefore, as with our Kaplan-Meier models, it appears that the notion of the poorer survival of patients with buccal primary tumors may be attributed to differences in tumor presentation. Differences in stage and age may account for why buccal cancer appears to have worse prognosis than cancers in the rest of the oral cavity. Fewer buccal cancers present as stage I tumors (27.2% vs. 36.3% nonbuccal, P 5.05), and more buccal cancers present as stage III (19.4% vs. 15.4% nonbuccal, P 5.05), which agrees with studies that demonstrate a high rate of regional lymph node disease and conclude that elective neck dissection may improve survival. 5,11 Because buccal cancer more likely presents at a later stage, we hypothesize that there must be some problem in identifying potentially cancerous lesions. To date, there are no studies that explore whether buccal cancer may be more difficult to diagnose than other oral cavity cancers. DeConde et al. found that increasing age trended toward having an increased risk of cancer recurrence (P 5.062), and although the age difference between buccal cancer presentation and that of other oral cavities is small (approximately 4 years), the statistically significant older age of buccal cancer patients may be causing worse prognosis by means of recurrence. 11 1390 We acknowledge several limitations in this study. Our analyses are restricted to the data collected by SEER. Many studies regarding buccal cancer address recurrence rate, stating that BSCC has a high rate of recurrence, potentially greater than that of other oral cavity cancers. 2,5,6 Unfortunately, SEER does not include data on recurrence, and we are unable to comment on whether it is more frequent in BSCC. Concerning treatment modalities, neck dissection for early stage disease is a relatively recent suggestion in the literature, and data on surgical treatment were not stratified into patients who received an elective neck dissection and those who did not. 12 Adjuvant therapy with radiation, chemotherapy, or both is reserved for advanced-stage disease and cases with high-risk histopathologic findings such as lymphovascular invasion, extracapsular spread, bone invasion, perineural invasion, or close margins. 4,5,11 The exact effect of radiation on survival was not explored, and chemotherapy data are not available in SEER. Assessment of either treatment would be beyond the goals of this study. Nevertheless, the use of adjuvant therapy in early stage tumors should be evaluated as a means to decrease recurrence. Finally, the SEER database does not include information regarding specific histopathological features (aside from grade), margin status, or risk factors such as molecular markers and socioeconomic data. Because tumor grade is attainable from SEER, studies using the SEER database have shown that high histologic grade in early stage OSCC is independently associated with worse survival, although differences in tumor grade at presentation between buccal and nonbuccal SCC have not been explored. 13 Concerning surgical margins, positive or negative status as a predictor of recurrence has been debated, as some buccal cancer studies show a high rate of local recurrence despite negative margins. 14 Regarding various risk factors, very little information exists evaluating the involvement of molecular markers with buccal cancer. Lack of research on this topic may be due to the small sample sizes available at most institutions. Future research should explore how these factors vary between buccal and nonbuccal oral cancers. Despite these limitations, this study reports the largest series of buccal carcinoma to date. Furthermore, the SEER database represents the standard of care of buccal carcinoma across the nation in various community hospitals, private institutions, and academic medical centers. Several other strengths were achieved by using such a recently diagnosed population (2004 2009). The patients in this cohort were diagnosed under one edition of the AJCC staging criteria, and the SEER database began recording TNM staging for patients in 2004, allowing us to stratify patients by cancer stage. Use of case matching, followed by multivariate analysis to double-check survival results, ensured that age, race, stage, and treatment variations between buccal cancer and other oral cancers were being controlled in our survival calculations. CONCLUSION Our study reports that the OS and DSS of buccal cancer are similar to those of cancers in other oral cavity

sites once age at diagnosis, tumor stage, treatment, and race are taken into consideration. Despite the majority of studies suggesting that buccal cancer has worse prognosis than other oral cavity cancers, none of these studies actually compared buccal and nonbuccal cancer survivals in their methods. The loss of significant differences between the survival of buccal and nonbuccal oral cancer after controlling for confounding variables suggests that the poor prognosis of buccal carcinoma may be due to variations in presentation, such as later stage and older age, rather than tumor biology. Future research should explore reasons why buccal cancer tends to present more often in later stages, because early diagnosis of BSCC may lead to improved survival in patients. BIBLIOGRAPHY 1. Cancer Facts & Figures 2013. Atlanta, GA: American Cancer Society; 2013. 2. Lubek JE, Dyalram D, Perera EH, Liu X, Ord RA. A retrospective analysis of squamous carcinoma of the buccal mucosa: an aggressive subsite within the oral cavity. J Oral Maxillofac Surg 2013;71:1126 1131. 3. Chhetri DK, Rawnsley JD, Calcaterra TC. Carcinoma of the buccal mucosa. Otolaryngol Head Neck Surg 2000;123:566 571. 4. Jan JC, Hsu WH, Liu SA, et al. Prognostic factors in patients with buccal squamous cell carcinoma: 10-year experience. J Oral Maxillofac Surg 2011;69:396 404. 5. Diaz EM Jr, Holsinger FC, Zuniga ER, Roberts DB, Sorensen DM. Squamous cell carcinoma of the buccal mucosa: one institution s experience with 119 previously untreated patients. Head Neck 2003;25:267 273. 6. Bachar G, Goldstein DP, Barker E, et al. Squamous cell carcinoma of the buccal mucosa: outcomes of treatment in the modern era. Laryngoscope 2012;122:1552 1557. 7. Lin CS, Jen YM, Cheng MF, et al. Squamous cell carcinoma of the buccal mucosa: an aggressive cancer requiring multimodality treatment. Head Neck 2006;28:150 157. 8. Shaw RJ, McGlashan G, Woolgar JA, et al. Prognostic importance of site in squamous cell carcinoma of the buccal mucosa. Br J Oral Maxillofac Surg 2009;47:356 359. 9. Surveillance, Epidemiology, and End Results Program. Public use data (2000 2009). November 2011 submission. Available at: http://seer.cancer.gov. Accessed on July 7, 2013. 10. Strome SE, To W, Strawderman M, et al. Squamous cell carcinoma of the buccal mucosa. Otolaryngol Head Neck Surg 1999;120:375 379. 11. DeConde A, Miller ME, Palla B, et al. Squamous cell carcinoma of buccal mucosa: a 40-year review. Am J Otolaryngol 2012;33:673 677. 12. Hakeem AH, Pradhan SA, Tubachi J, Kannan R. Outcome of per oral wide excision of T1-2 N0 localized squamous cell cancer of the buccal mucosa analysis of 156 cases. Laryngoscope 2013;123:177 180. 13. Thomas B, Stedman M, Davies L. Grade as a prognostic factor in oral squamous cell carcinoma: a population-based analysis of the data. Laryngoscope. Oct 4 2013. 14. Sieczka E, Datta R, Singh A, et al. Cancer of the buccal mucosa: are margins and T-stage accurate predictors of local control? Am J Otolaryngol 2001;22:395 399. 1391