Fax: (410) 328-6911
Race impacts outcome in stage III/IV squamous cell carcinomas of the head and neck after concurrent chemoradiation therapy†
Article first published online: 10 FEB 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 8, pages 1744–1752, 15 April 2009
How to Cite
Settle, K., Taylor, R., Wolf, J., Kwok, Y., Cullen, K., Carter, K., Ord, R., Zimrin, A., Strome, S. and Suntharalingam, M. (2009), Race impacts outcome in stage III/IV squamous cell carcinomas of the head and neck after concurrent chemoradiation therapy. Cancer, 115: 1744–1752. doi: 10.1002/cncr.24168
Presented at the American Society of Therapeutic Radiology and Oncology, Los Angeles, California, October 30, 2007.
- Issue published online: 6 APR 2009
- Article first published online: 10 FEB 2009
- Manuscript Accepted: 27 AUG 2008
- Manuscript Revised: 1 JUL 2008
- Manuscript Received: 8 MAY 2008
- squamous cell carcinoma;
- head and neck;
The purpose of this study was to determine the impact of race on outcome in patients with stage III/IV squamous cell carcinoma of the head and neck (SCCHN) who have completed concurrent chemoradiotherapy.
The authors performed a retrospective analysis of 202 patients with stage III/IV SCCHN who were treated at the University of Maryland. Patients received daily radiation to a total dose of 70.2 Gray (Gy) (1.8 Gy/day), concurrently with weekly carboplatin (area under the curve [AUC] = 2) and paclitaxel (45 mg/m2) chemotherapy.
There were 108 Caucasian (CA) and 94 African American (AA) patients. The median age was 56 years, and 81% were stage IV. The median follow-up was 33 months. The median overall survival (OS) and disease-free survival (DFS) were 33 months and 19 months, respectively. When analyzed by race, the median DFS was 33 months (CA) versus 12 months (AA) (P = .028). The median OS was 44 months (CA) versus 24 months (AA) (P = .071). The 3-year DFS for stage IV AA versus stage IV CA was 29% versus 50% (P = .031). The 3-year DFS for N2 disease in AA versus CA was 32% versus 51% (P = .046). The 3-year DFS for AA versus CA with oropharyngeal tumors was 30% versus 60% (P = .006).
This analysis documents the inferior outcome for AA patients. They had inferior DFS and a trend toward worse OS. When stratified by several prognostic variables, the mediocre DFS in the AA patients remains. These data suggest that further investigation into the genetic characteristics of SCCHN in AA patients is warranted. Cancer 2009. © 2009 American Cancer Society.
There are roughly 45,000 cases of squamous cell carcinomas of the head and neck (SCCHN) diagnosed per year, representing <5% of new cancer cases diagnosed in 2007.1 Patients presenting with locally advanced, stage III/IV disease are a challenging group to cure, with 3-year survival rates ranging from 25% to 55% with platinum-based concurrent chemoradiotherapy.2-5 Several factors increase the risk of developing SCCHN, with the most significant being tobacco use, heavy alcohol abuse,6 or human papillomavirus–16 infection.7 In addition, patients who continue to smoke or consume more than 14 alcoholic drinks per week have a 2-fold risk and a 50% increased risk of developing a second primary SCCHN, respectively.8
Cancer incidence and death rates differ widely among racial and ethnic groups. Data from the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute 2006 review have indicated that African American (AA) men have, for all cancer sites combined, a 15% higher incidence rate and 38% higher death rate compared with white men.9 Particularly, AAs have a higher incidence rate of SCCHN, are more likely to present with advanced disease, and demonstrate diminished survival compared with Caucasians (CAs).9 There are no definitive biologic data to explain the poor prognosis of AA with SCCHN. Several hypotheses have been presented to explain these differences in survival. These include socioeconomic status, lack of insurance, advanced stage at presentation, genetic predisposition to more aggressive disease, and increased exposure to known risk factors such as tobacco and alcohol.10, 11 The literature contains a limited number of single institution series from the United States evaluating the impact of race on treatment outcomes for patients with SCCHN.
To date, there are no data evaluating the impact of race on treatment outcomes in a homogeneous cohort of patients with SCCHN who have received concurrent chemoradiotherapy. The purpose of our retrospective analysis was to evaluate whether the AA race negatively impacts outcome in patients with stage III/IV SCCHN, treated at a single institution, who received uniform concurrent chemoradiotherapy.
MATERIALS AND METHODS
From November 1995 through July 2006, roughly 2000 patients with SCCHN were treated at the University of Maryland. Patients with AJCC stage III/IV squamous cell carcinoma of the oral cavity, oropharynx, hypopharynx, and larynx who had long-term follow-up were identified. For the purpose of this retrospective analysis, a homogeneous cohort of 202 patients treated with concurrent chemoradiotherapy for curative intent was selected. One hundred and eight (53%) patients were non-AA (98% of which were CA), and 94 (47%) patients were AA. All patients had previously untreated SCCHN with no evidence of distant metastases. The characteristics of the patients are shown in Table 1. The majority of patients were presented to a multidisciplinary conference and evaluated by a surgeon, radiation oncologist, and medical oncologist before therapy.
|Characteristic||CA Patients (n=108), No. (%)||AA Patients (n=94), No. (%)||Chi-square P|
|Average age, y||57||57||.686|
|≤60||74 (69)||60 (64)|
|>60||34 (31)||34 (36)|
|Men||93 (86)||84 (89)|
|Women||15 (14)||10 (11)|
|Oral cavity||9 (8)||5 (5)|
|Oropharynx||69 (64)||54 (57)|
|Larynx||24 (22)||15 (16)|
|Hypopharynx||6 (6)||20 (21)|
|T1-T2||22 (20)||11 (12)|
|T3-T4||86 (80)||83 (88)|
|N0-N1||30 (28)||36 (30)|
|N2-N3||78 (72)||58 (62)|
|Overall AJCC stage||.169|
|III||17 (16)||22 (23)|
|IV||91 (84)||72 (77)|
|WD/MD||72 (71)||64 (74)|
|PD||29 (29)||22 (26)|
|Mean overall treatment time, d||59||58||.785|
|<55 d||36 (33)||30 (32)|
|≥55 d||72 (67)||64 (68)|
|Mean weight loss, lbs.†||17||13||.418|
|<10% body weight||45 (42)||54 (59)|
|≥10% body weight||63 (58)||38 (41)|
|Prior smoker‡||88 (84)||91 (97)||.001|
|Continued smoking after treatment‡||21 (20)||16 (17)||.614|
|Prior ETOH abuse§||48 (47)||64 (68)||.001|
|Continued ETOH after treatment§||4 (4)||3 (3)||.827|
Baseline evaluation consisted of a history and physical examination, pretreatment panendoscopy with documentation of tumor extent, basic blood work, electrocardiogram, and computed tomography (CT) scan of the head, neck, and chest. After 2002, positron emission tomography (PET)/CT scans were routinely ordered. Pretherapy dental evaluation was required before the start of chemoradiation. Gastrostomy feeding tubes were placed endoscopically in all patients who were felt to be at significant risk for nutritional compromise as a result of therapy.
The majority of patients (96%) received concurrent chemoradiotherapy. Of these patients, 79% received weekly carboplatin (area under the curve [AUC] = 2) and paclitaxel (45 mg/m2). The remaining (21%) chemotherapy regimens were roughly divided between cisplatin 100 mg/m2 every 3 weeks or weekly paclitaxel (40 mg/m2) and capecitabine (Xeloda) (500 mg twice daily, Mondays-Fridays). Baseline absolute neutrophil counts and records of possible chemotherapy reductions were not complete and therefore, not included in this study.
Radiation was delivered at 1.8 Gray (Gy) per day, 5 days a week, to a total dose of 70.2 Gy to areas of gross disease. Initial treatment fields encompassed the primary tumor plus a 2-cm margin, all clinically involved nodal disease, and draining lymphatics felt to be at risk for harboring microscopic disease. Uninvolved nodal sites received 50 Gy. The primary tumor and palpable nodes were included to the final dose of 70.2 Gy. All patients were treated on linear accelerators with either 4- or 6-MV photon beams. The treatment technique used shaped parallel-opposed lateral fields with doses prescribed to midplane, with a matched anterior low-neck field (prescribed to maximum density [DMAX]). Appropriate shielding was used to protect the spinal cord after a dose of 45 Gy had been delivered.
Response Assessment and Follow-up
Weekly treatment evaluations consisted of history and physical examinations, including documentation of weight and performance status. Mean weight loss, percentage weight loss, and overall treatment time were used as surrogate markers of treatment tolerance. Routine endoscopic evaluation was performed by the surgeon and radiation oncologist during follow-up visits. CT scans of the head and neck (PET scan routinely ordered after 2002) were obtained 4 to 6 weeks after therapy. Response evaluations were based on physical examination, endoscopic findings, and radiographic assessment. Pathologic confirmation was required for patients suspected to have clinical evidence of residual disease at the primary site or neck 6 weeks after therapy.
Statistical analysis was carried out using SPSS software (version 13.0; Chicago, Ill). Survival data, including overall survival (OS) and disease-free survival (DFS), were calculated using the Kaplan-Meir product limit method. OS was defined as death from any cause, known or unknown. DFS was defined as death from any cause, known or unknown, and time at recurrence. Factors included in univariate and multivariate analysis were primary site location, T stage, N stage, overall American Joint Committee on Cancer (AJCC) 2003 stage, overall treatment time <55 days and ≥55 days, weight loss <10% body weight and ≥10% body weight, prior tobacco use, continued tobacco use after completion of therapy, prior alcohol abuse defined as >2 drinks/day, and continued alcohol abuse after the completion of therapy. The log-rank test was used to detect the statistical difference in estimates between strata of the selected factors above. Multivariate analysis was performed using the Cox proportional hazards model. P≤.05 was considered statistically significant.
Two hundred and two patients were evaluated, with a median active follow-up of 33 months. The median patient age was 56 years (range, 36-86 years). Fifty-three percent of patients were CA, and 47% were AA. Eighty-five percent of the patients were men. Nineteen percent of the patients were AJCC stage III, and 81% of the patients were AJCC stage IV, with no difference in stage at presentation between the 2 racial cohorts. Ninety-seven percent of patients were able to complete the prescribed therapy. There was no statistically significant difference in mean or percentage weight loss during treatment between CA and AA patients (P = .418). However, 58% of CA experienced ≥10% weight loss versus 41% of AA patients. The average overall treatment time was 59 days for CA patients and 58 days for AA patients. There was no statistically significant difference in average overall treatment time, treatment completed in <55 days, or treatment completed in ≥55 days between CA and AA patients (P = .785). The patient characteristics were well balanced between the 2 racial groups, with the exception of a statistically significant difference in site, prior tobacco use, and prior alcohol abuse. Specifically, there were more hypopharyngeal carcinomas, prior smokers, and prior alcohol abusers in the AA cohort. However, when controlling for anatomic subsite, there was no difference in DFS between the hypopharynx and the oral cavity or laryngeal subsites. There was a statistically significant difference in DFS between hypopharyngeal tumors and oropharyngeal tumors (median survival [MS] 9 months and 33 months, respectively; P = .003).
At the time of analysis, 86 patients were alive, and 116 were dead. Of the 116 deaths, 17 were of unknown cause, and 5 patients died during or within 1 week of completing therapy. Thirty-five percent of CA patients died of their disease versus 47% of AA patients (P = .062).
Patterns of Failure
Ninety-five of the 202 patients had recurrence of their disease after completion of therapy. Forty-six (43%) CA patients developed recurrent disease, and the median time to recurrence was 10 months. Forty-nine (52%) AA patients recurred, and the median time to recurrence was 8 months. Table 2 describes the type of initial failure. There was no significant difference in the type of initial failure between the CA and AA patients.
|Endpoint||CA Patients (n=108), No. (%)||AA Patients (n=94), No. (%)||P|
|Dead from disease||38 (35)||44 (47)||.062|
|Total recurrence||46 (43)||49 (52)||.176|
|LR||23 (21)||21 (22)||.383|
|DM||10 (9)||12 (13)||.214|
|LR+DM||4 (4)||9 (10)||.214|
|Unknown||9 (8)||7 (7)||.214|
The median DFS and 3-year DFS for all patients were 19 months and 39%, respectively. On univariate analysis, race (P = .028), site (hypopharynx, P = .008; oral cavity, P = .004), and T3/T4 disease (P = .001) were statistically significant for predicting a poor outcome. A prior smoking history approached statistical significance (P = .063). The variables AJCC stage (P = .255), differentiation (P = .101), weight loss (P = .391), nodal stage (P = .415), and overall treatment time (P = .179) were not statistically significant. When analyzed by race, the median DFS was 33 months in CA patients versus 12 months in AA patients (P = .028). The 3-year DFS was 47% in CA patients versus 29% in AA patients (Fig. 1). The 3-year DFS based on race, stratified by known prognostic variables, is listed in Table 3. When controlling for the listed 8 prognostic variables in Table 3, there is a strong correlation between the AA race and a significantly worse DFS.
|Prognostic Factor||CA Patients||AA Patients||P|
|AJCC stage IV||50%||29%||.031|
|Overall treatment time <55 d||46%||36%||.017|
|<10% loss of body weight||45%||21%||.050|
To control for the imbalance of primary site location and prior smoking history within the 2 racial cohorts, DFS was further analyzed in 106 patients with stage III/IV oropharyngeal carcinomas who were all previous smokers. Patient characteristics were well balanced between the CA and AA cohorts, with the exception of a statistically significant (P = .048) increase in prior alcohol abusers in the AA cohort. The median DFS was 61 months in CA patients versus 15 months in AA patients (P = .021). The 3-year DFS was 57% in CA patients versus 28% in AA patients (Fig. 2). Thirty-four percent of CA patients died from their disease versus 49% of AA patients (P = .084). When controlled for a prior alcohol abuse history, AA patients continued to display a worse 3-year DFS (20% vs 53%, P = .003), as seen in Figure 3.
Given the poor DFS in the AA cohort, its impact on OS was evaluated. The median OS and 3-year OS for all patients were 33 months and 48%, respectively. On univariate analysis, T3/T4 disease (P = .003) and site (hypopharynx and oral cavity, P = <.01) were predictors of worse OS. Race approached statistical significance (P = .071). The variables AJCC stage (P = .815), nodal stage (P = .300), differentiation (P = .185), prior smoking history (P = .136), weight loss (P = .295), and overall treatment time (P = .122) did not statistically influence outcome. The median OS based on race was 44 months in CA patients versus 24 months in AA patients (Fig. 4). The 3-year OS was 56% in CA patients versus 36% in AA patients. On univariate analysis, race was stratified by known prognostic variables, and the 3-year OS survival is listed in Table 4. There was a strong correlation and trend toward poor outcome in the AA patients when controlling for site, differentiation, AJCC stage, and overall treatment time. In patients with oropharyngeal tumors and controlling for a previous smoking and alcohol abuse history, 3-year OS was worse in the AA cohort (27% vs 67%, P = .003).
|Prognostic Factor||CA Patients||AA Patients||P|
|AJCC stage IV||57%||33%||.079|
|Overall treatment time <55 days||64%||45%||.066|
|<10% loss of body weight||53%||30%||.149|
In this retrospective analysis, the AA race was found to be a poor prognostic factor in patients with stage III/IV SCCHN who received uniform concurrent chemoradiotherapy. These individuals had a statistically significant decrease in DFS and a trend toward reduced OS. There have been numerous hypotheses published as to the etiology of this racial disparity.10-15 Specifically, factors such as socioeconomic status, unequal cancer directed therapy, and advanced stage of presentation have been sited as an explanation for the poor outcomes in AA patients.14-15
In 2006, Gourin and Podolsky published data from a nonrandomized retrospective cohort analysis to determine whether racial disparities exist between black and white patients with SCCHN treated at a single large institution in the South.10 Their AA patients were more likely to be younger, have an advanced stage of presentation, have a previous history of alcohol abuse, lack insurance, have a lower mean education level, and have a lower median income as compared with the CA patients. They found that only insurance status had a significant effect on survival in the AA patients after controlling for other variables, suggesting that racial differences in outcome were primarily related to differences in access to healthcare. Their paper suggested that equal access to healthcare is able to even out outcomes. Murdock and Gluckman16 sought to evaluate the characteristics of AA patients and CA patients with SCCHN at the University of Cincinnati Medical Center, which is an equal-access facility. They reported that twice as many AA patients presented with stage IV disease. The average death rate over time was about twice as great for AA patients. Death from disease-specific causes occurred in 75% of the AA patients versus 25% of CA patients.
In 2007, Nichols and Bhattacharyya performed a retrospective analysis on AA and CA patients with squamous cell carcinoma of the oral tongue and glottis obtained from the SEER database.11 For each AA patient, a randomly selected CA control was matched for age at diagnosis, sex, stage, surgical treatment, and radiation. They found that when controlling for stage and treatment, the AA patients demonstrated poorer overall and disease-specific survival, suggesting other intrinsic or extrinsic factors influencing survival. Al-Othman et al17 from The University of Florida reported their assessment of the impact of race on the outcome of patients with SCCHN who have received definitive radiotherapy. Their population characteristics were well balanced; 80% of the patients presented with stage III/IV disease, twice-daily radiotherapy was delivered (no difference between races in median dose delivered of roughly 76 Gy), and ⅓ of patients underwent a planned unilateral neck dissection. Their data revealed equal locoregional control rates but the risk of distant metastases (DM) was significantly higher in AA patients. This translated into statistically significantly worse cause-specific and absolute survival rates in AA patients.
In this analysis, we sought to evaluate the majority of extrinsic and intrinsic factors previously sited to influence survival in racial groups. Our study was centered on 2 homogenous racial cohorts that were balanced in terms of AJCC stage at presentation, average age and overall treatment time, mean weight loss during therapy, and tumor differentiation. Weight loss and overall treatment time (thus, accounting for treatment breaks) were provided as surrogate markers of treatment tolerance. There was no difference between AA and CA patients when accounting for either of these variables. Eighty-five percent of our patients were men, which is consistent with the University of Florida patient population.17 The numbers are too small to make a conjecture as to whether sex influenced survival in this particular series. However, Neville and Day18 published data on age-adjusted mortality rates for cancers of the oral cavity and pharynx from 1950 to 1995. AA and CA women had equivalent mortality rates, whereas the mortality rate for AA men has significantly worsened.
Although we did not account for insurance status and pretreatment comorbidities, the majority of patients (>95%) received and completed concurrent chemoradiotherapy at a single institution. In contrast to the study by Gourin and Podolsky,10 uniform therapy and equivalent access to cancer treatment did not equalize outcomes in this analysis. Our AA patients were more likely to be smokers and heavy drinkers as compared with our CA cohort. This has been substantiated in previous studies, including that of Day et al,13 who evaluated data from a large, population-based case-control study of oral cancer risk factors conducted in 4 areas of the United States. They found that AA patients were more likely to smoke and abuse alcohol compared with the CA patients in their population of SCCHN.
In addition to inferior DFS and a trend toward a worse OS, our AA patients had a higher recurrence rate (52% vs 43%) and were more likely to die from their disease (47% vs 35%). The small number of patients likely influenced these outcomes, as they only approached but did not reach statistical significance. There was no statistical difference in the type of initial failure; however, the AA patients had a higher rate of DM disease (43% vs 30%). Again, it is plausible that the small patient numbers prevented statistical significance from being achieved. When controlling for tobacco and heavy alcohol use, the AA patients continued to have a worse DFS. Our AA cohort had more hypopharyngeal and fewer laryngeal carcinomas that the CA patients. Carter et al19 recently presented data indicating the laryngeal/oropharyngeal and hypopharyngeal/oral cavity sites as independent favorable and unfavorable prognostic factors, respectively, with regards to DFS and OS. To control for the influence of site, prior smoking, and alcohol use, a matched cohort of AA and CA patients with stage III/IV oropharyngeal carcinoma was evaluated. The data were striking; the AA patients had significantly worse 3-year DFS (20% vs 53%, P = .003) and 3-year OS (27% vs 67%, P = .003).
Our data are highly suggestive of the AA race, particularly men, being an independent poor prognostic variable. At this time, there have been no definitive data published to support a more biologically aggressive tumor or a genetic predisposition to more aggressive disease as explanations for the poor outcomes in AA patients. Could individual biologic differences account for the variability in outcome? Although AA patients are identified as presenting with more locally advanced disease and having poorer outcomes than their non-AA counterparts, race may still be a crude prognostic marker. In an editorial published in the New England Journal of Medicine,20 Dr. M. Gregg Bloche cautions physicians against using race as a marker for genetic variation. Although his commentary is directed at the data presented in the African-American Heart Failure Trial published in the same issue, he makes several important points that can be applied to the oncology world. The most important argument being that within the AA race or any ethnic grouping such as “Caucasian,” “Latino,” or “Asian,” there is genetic heterogeneity. However, the ultimate answer lies within a well-designed, prospective study focused on evaluating molecular characteristics that may help differentiate between CA and AA patients with locally advanced SCCHN.
This retrospective analysis documents the inferior outcome for AA patients as compared with their CA counterparts. The AA patients had inferior DFS and a trend toward worse OS. The homogenous cohort selected reduces confounding variables such as disparate treatments received or socioeconomic status. When stratified by several established prognostic variable such as AJCC stage, T stage, N stage, site of disease, differentiation, overall treatment time, weight loss during therapy, and previous smoking history, the mediocre DFS in the AA patients remains.
Although phenotype alone cannot explain the biological nature of cancer, it may be a stepping stone to evaluating genetic distinctions within AA patients. Our data suggest that intrinsic differences between AA and CA may result in poor outcomes. Further investigation into the molecular and cytogenetic characteristics of SCCHN in the AA patient population is warranted.
Conflict of Interest Disclosures
There is no conflict of interest with any of the authors.
- 1American Cancer Society. Cancer Facts & Figures, 2007. Atlanta, GA: American Cancer Society; 2007.
- 4Long-term results of intergroup RTOG 91-11: a phase III trial to preserve the larynx-induction cisplatin/5FU and radiation therapy versus concurrent cisplatin and radiation therapy versus radiation therapy. J Clin Oncol. 2006; 24( 18 suppl): 5517., , , et al.
- 6GundersonLL,TepperJE, eds. Clinical Radiation Oncology. 2nd ed. Philadelphia, PA: Churchill Livingstone; 2007.
- 19Site as a prognostic factor in patients with stage III/IV squamous cell carcinoma of the head and neck following definitive chemoradiation. IJROBP 2007; 69(3 suppl): 2481., , , et al.