A study was undertaken to determine the survival benefit of postoperative chemoradiation therapy for elderly patients with resected gastric adenocarcinoma.
A study was undertaken to determine the survival benefit of postoperative chemoradiation therapy for elderly patients with resected gastric adenocarcinoma.
The authors identified 1023 individuals aged 65 years and older (median = 76) who underwent gastrectomy for nonmetastatic stage IB-IV gastric adenocarcinoma diagnosed between 2000 and 2002 in the linked Surveillance, Epidemiology, and End Results-Medicare database. They examined factors associated with receiving postoperative chemoradiation and analyzed the survival benefit associated with receiving postoperative chemoradiation.
Thirty percent of patients received adjuvant chemoradiation. On multivariate analysis, younger age (P < .0001), lymph node involvement (P < .0001), and more recent diagnosis (P = .0284) were associated with receiving chemoradiation. There was a trend toward increased use among patients with less comorbidity (P = .0515). The median follow-up was 25.5 months, and 62% died. On multivariate survival analysis, older patients (P < .0001) and those with lymph node involvement (P < .0001), T3 or T4 disease (P = .0472), higher grade disease (P = .0355), and more comorbidity (P = .0411) were more likely to die. After adjustment for other factors, receipt of adjuvant chemoradiation therapy did not significantly increase survival (hazard ratio, 0.90; 95% confidence interval, 0.72-1.12; P = .3453) and did not increase survival in a multivariate analysis that included propensity scores (P = .2090).
The authors did not detect a survival benefit, suggesting that some elderly patients with resected gastric adenocarcinoma may not gain a survival benefit from the administration of adjuvant chemoradiation. The analysis had limitations, and the results are hypothesis generating. Future gastric cancer trials should enroll more elderly patients and stratify patients by age to better understand the impact of treatment regimens on older patients. Cancer 2012;. © 2011 American Cancer Society.
In a landmark trial, Intergroup (INT)-0116, postoperative chemotherapy and radiation therapy improved survival for patients with resected gastric adenocarcimona.1 However, the effectiveness of postoperative chemoradiation for elderly individuals treated outside of a controlled clinical trial setting is not known. It is important to understand the potential benefit for older patients, because gastric cancer mostly affects older individuals; the average age at diagnosis is 71 years, and almost ⅔ of those diagnosed with gastric cancer are older than 65 years.2
Older patients diagnosed with cancer are less likely to receive standard treatment compared with younger patients, even when such treatments are potentially curative.3-5 Indeed, older patients with gastric cancer are less likely to receive any type of treatment for their cancer compared with younger patients,6 and older patients who undergo gastrectomy for gastric adenocarcinoma are less likely to receive postoperative radiation therapy.7
The determination of treatment effectiveness in the elderly may substantially impact postoperative chemoradiation utilization among elderly patients with locally advanced gastric adenocarcinoma. Therefore, this study used the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database to examine clinical and sociodemographic factors associated with receiving postoperative chemoradiation for resected gastric adenocarcinoma among individuals aged 65 years and older and to evaluate the survival benefit associated with receiving postoperative chemoradiation. We hypothesized that postoperative chemoradiation would confer a survival benefit, but this survival benefit would be less than the benefit demonstrated in the INT-0116 trial.
Patients were identified from the linked SEER-Medicare database. The catchments for the 17 SEER tumor registries comprise 25% of the US population,8 and the registries participating in the SEER program capture approximately 97% of incident cancers.9 The registries collect data on patient demographics, primary tumor site, tumor morphology and stage at diagnosis, first course of treatment, and date and cause of death. The linked Medicare data include inpatient and outpatient medical claims and physician billings,8 and were used to determine radiation and chemotherapy treatment, comorbid illnesses, and treatment for metastatic disease. The SEER-Medicare database used for this analysis contained SEER diagnoses through 2002, Medicare claims through 2005, gastric cancer-specific mortality through 2003, and vital status follow-up through February 2005.
The cohort contained individuals aged 65 years and older diagnosed with nonmetastatic invasive American Joint Committee on Cancer (AJCC, sixth edition) stage IB to IV gastric adenocarcinoma in a SEER region between January 1, 2000 and December 31, 2002 and who underwent gastrectomy as initial therapy. We selected this date range because the results of the INT-0116 trial were disseminated in 2000,10 and the results of the next landmark gastric cancer trial, the MAGIC trial, were initially presented in early 2003.11 Individuals within our study cohort met basic criteria for enrollment in the INT-0116 trial (resected tumor [T]2 to T4 disease or lymph node involvement).1 We excluded subjects who died within 1 month of gastrectomy, subjects with a prior cancer diagnosis because prior treatment can impact adjuvant therapy recommendations, and subjects who developed metastatic disease within 6 months of gastrectomy. Subjects who did not have continuous Medicare enrollment (both Part A and Part B) and those who were enrolled in a health maintenance organization (HMO) any time from 13 months before diagnosis (for use in comorbidity assessment) through 6 months after gastrectomy were excluded because they did not have complete claims data. In total, 1023 subjects met our inclusion criteria.
The primary outcomes were receipt of adjuvant chemoradiation within 6 months postgastrectomy and overall survival. Medicare claims identified adjuvant chemotherapy administration and have previously been shown to correlate well with chemotherapy receipt.12 Medicare claims used to capture gastrectomy, general chemotherapy administration, and general radiation therapy administration are detailed in Table 1. Both SEER and Medicare were used to identify radiation therapy receipt within 6 months after the first gastrectomy claim to ensure a comprehensive assessment of radiation treatment.13 In the SEER database, patients are coded as receiving radiation therapy or recommendation for radiation therapy as a component of the first course of treatment.
|Gastrectomy||ICD-9-CM procedure 43.5x-43.99|
|CPT 43620-43622, 43631-43634|
|General chemotherapy administration||ICD-9-CM diagnosis V58.1|
|ICD-9-CM procedure 99.25|
|HCPCS C1166, C1167, C1178, C9110, C9205, C9207, C9213-C9216, C9411, C9414-C9419, C942x, C9430-C9438, G0355, G0356, G0359-G0362, J7150, J85xx-J87xx, J8999, J9xxx, Q0083-Q0085, S9325-S9329, S933x-S937x, S9494-S9497|
|CPT 9651x-9654x, 964xx|
|Revenue center 0331, 0332, 0335|
|Radiation therapy administration||ICD-9-CM diagnosis V58.0|
|ICD-9-CM procedure 92.2x|
|CPT 77xxx, 79xxx; revenue center 0330, 0333, 0339|
|Inpatient (MEDPAR) indicator for receipt of radiology oncology services, and MEDPAR indicator for receipt of radiology therapeutic services|
|Diagnosis of metastatic disease||ICD-9-CM diagnosis 196-199|
Explanatory variables evaluated for association with receiving adjuvant chemoradiation therapy and with survival included: diagnosis year, tumor characteristics (size, number of lymph nodes involved), clinical characteristics (age at diagnosis, comorbidities), sociodemographic factors (ethnicity, socioeconomic status, region of the country), type of treating institution (academic vs community hospital), and distance to nearest radiation treatment facility. Diagnosis year was categorized into 6-month blocks. Tumor stage was categorized by AJCC sixth edition T stage, which is based on the depth of penetration (T1, T2, T3, and T4). The number of involved regional lymph nodes were grouped according to AJCC sixth edition regional lymph node (N) staging as none (N0), 1 to 6 (N1), 7 to 15 (N2), and more than 15 nodes (N3). We identified comorbidities by collecting diagnostic billing codes for specific health conditions during the year before diagnosis of gastric cancer using the Deyo implementation of the Charlson score applied to both inpatient and outpatient claims.14-16 Subjects were categorized as receiving their gastrectomy in a teaching hospital if there was a bill for indirect medical education during their stay. Distance to the nearest radiation treatment facility was determined by an established algorithm that calculated the distance from the Zip Code of the patient's residence to that of the closest radiation therapy facility.17
Descriptive statistics were generated for the study cohort. The study subjects were stratified by adjuvant treatment received: 1) no adjuvant chemotherapy or radiation therapy; 2) either radiation therapy or chemotherapy; and 3) both chemotherapy and radiation therapy. Chi-square tests and Kruskal-Wallis test were used to compare categorical and continuous variables across treatment groups, respectively. The crude association of each potential explanatory variable with the outcome of receiving chemoradiation therapy was examined using univariate logistic regression. The independent association of an explanatory variable was examined using a multivariate logistic regression model constructed using forward and backward elimination. Subjects with missing data for T stage (n = 10), N stage (n = 74), grade (n = 17), or distance from radiation facility (n = 1) were excluded from univariate and multivariate models including these variables. The survival of the subjects who received both adjuvant chemotherapy and radiation therapy was compared with the survival of the subjects who received no adjuvant therapy. Individuals who received either only chemotherapy or only radiation were removed from the survival analysis (n = 131). Survival, calculated from date of gastrectomy, was examined using multivariate and propensity-based Cox-proportional hazard regression models that included all explanatory variables. Propensity scores were created to account for unmeasured factors associated with receiving chemoradiation that may also influence overall survival. A multivariate logistic regression model with receipt of chemoradiation as the outcome was used to generate the propensity scores. Subjects were stratified into quintiles based on their scores, which were then added as covariates to the multivariate Cox proportional hazards model. In an exploratory analysis, survival outcome among those with stage III or IV disease was examined using a Cox-proportional hazard regression analysis. Interaction terms to test a priori hypotheses regarding the receipt of adjuvant chemoradiation therapy (chemoradiation therapy and age, and chemoradiation therapy and nodal involvement) were studied. For illustrative purposes, unadjusted Kaplan-Meier survival curves were constructed comparing patients who received adjuvant combined chemoradiation therapy to those who received neither adjuvant chemotherapy nor adjuvant radiation therapy (Fig. 1). In a sensitivity analysis, gastric-cancer specific survival rather than overall survival was examined using multivariate and propensity-based Cox-proportional hazard regression models.
Statistical analyses were conducted using SAS software, version 9.1.3 for Windows (SAS Institute, Cary, NC).
This study was reviewed by the institutional review board at the Dana-Farber Cancer Institute and determined to be exempt.
Among the 1023 elderly patients with nonmetastatic resected stage IB to IV gastric adenocarcinoma, 5 % had T1, 65% had T2, and 30% had T3 or T4 disease. Sixty-nine percent of the patients had lymph node involvement (Table 2). The median diagnosis age was 76 years (interquartile range, 72-81), and the majority of patients were white (72%) or Asian (17%). Adjuvant chemoradiation was administered to 30% (n = 309) of subjects during the study period, 57% (n = 583) of subjects received no adjuvant therapy, and the remainder (n = 131) received either adjuvant chemotherapy or adjuvant radiation therapy. Thirty-two percent of subjects diagnosed between July and December 2002 received adjuvant chemoradiation therapy, compared with 21% of subjects diagnosed between January and June 2000.
|Characteristic||All Subjects, N=1023||No Adjuvant Therapy, n=583||Either Chemotherapy or Radiation Therapy, n=131||Chemotherapy and Radiation Therapy, n=309||Pa|
|Age at diagnosis, median [IQR]||76.3 [71.6-81.0]||78.8 [73.7-83.6]||75.1 [69.9-79.0]||73.0 [70.0-76.7]||<.0001b|
|Sex, No. (%)||.0503|
|Male||561 (54.8)||303 (52.0)||71 (54.2)||187 (60.5)|
|Female||462 (45.2)||280 (48.0)||60 (45.8)||122 (39.5)|
|Race, No. (%)||.0811|
|White||737 (72.0)||419 (71.9)||96 (73.3)||222 (71.8)|
|Asian||173 (16.9)||93 (16.0)||17 (13.0)||63 (20.4)|
|Other (including black)||113 (11.1)||71 (12.2)||18 (13.7)||24 (7.8)|
|SES quintile, No. (%)||.0185b|
|0||220 (21.5)||145 (24.9)||23 (17.6)||52 (16.8)|
|1||214 (20.9)||118 (20.2)||33 (25.2)||63 (20.4)|
|2||197 (19.3)||119 (20.4)||17 (13.0)||61 (19.7)|
|3||182 (17.8)||91 (15.6)||31 (23.7)||60 (19.4)|
|4||210 (20.5)||110 (18.9)||27 (20.6)||73 (23.6)|
|6-month diagnosis block, No. (%)||.1393|
|Early 2000||189 (18.5)||118 (20.2)||32 (24.4)||39 (12.6)|
|Late 2000||160 (15.6)||87 (14.9)||19 (14.5)||54 (17.5)|
|Early 2001||179 (17.5)||105 (18.0)||21 (16.0)||53 (17.2)|
|Late 2001||148 (14.5)||85 (14.6)||19 (14.5)||44 (14.3)|
|Early 2002||176 (17.2)||95 (16.3)||17 (13.0)||64 (20.7)|
|Late 2002||171 (16.7)||93 (16.0)||23 (17.6)||55 (17.8)|
|T category,c No. (%)||.0002b|
|T1 or T2||705 (69.6)||432 (74.6)||77 (59.7)||196 (64.3)|
|T3 or T4||308 (30.4)||147 (25.4)||52 (40.3)||109 (35.7)|
|N category,c No. (%)||<.0001b|
|N0||294 (31.0)||234 (43.5)||21 (17.7)||39 (13.4)|
|N1||453 (47.7)||231 (42.9)||60 (50.4)||162 (55.5)|
|N2 or N3||202 (21.3)||73 (13.6)||38 (31.9)||91 (31.2)|
|Grade,c No. (%)||.0046b|
|1 or 2||329 (32.7)||211 (36.8)||32 (24.8)||86 (28.3)|
|3 or 4||677 (67.3)||362 (63.2)||97 (75.2)||218 (71.7)|
|Charlson score, No. (%)||<.0001b|
|0||575 (56.2)||301 (51.6)||90 (68.7)||184 (59.6)|
|1||273 (26.7)||154 (26.4)||23 (17.6)||96 (31.1)|
|2+||175 (17.1)||128 (22.0)||18 (13.7)||29 (9.4)|
|Gastrectomy at teaching hospital, No. (%)||.4688|
|Yes||552 (54.0)||306 (52.5)||76 (58.0)||170 (55.0)|
|No||471 (46.0)||277 (47.5)||55 (42.0)||139 (45.0)|
|Distance to RT facility,a miles, median [IQR]||5.1 [2.5-12.2]||5.2 [2.5-12.0]||5.0 [2.5-13.5]||5.3 [2.9-13.1]||.7486|
|SEER region, No. (%)||.3798|
|Northeast||277 (27.1)||160 (27.4)||35 (26.7)||82 (26.5)|
|South||134 (13.1)||85 (14.6)||15 (11.5)||34 (11.0)|
|Midwest||106 (10.4)||63 (10.8)||17 (13.0)||26 (8.4)|
|West||506 (49.5)||275 (47.2)||64 (48.9)||167 (54.1)|
On univariate analysis (Table 3), male sex (P = .0165), younger age (P < .0001), lymph node involvement (P < .0001), less comorbidity (P = .0019), higher socioeconomic status (P = .0097), and more recent diagnosis (P = .0161) were associated with increased receipt of adjuvant chemoradiation after gastrectomy (Table 3). Distance to radiation therapy facility was not associated with receiving chemoradiation (P = .1032).
|Covariate||Chemoradiation Therapy, %||Odds Ratio (95% CI)||P|
|Age at diagnosis, per year increase||30.2||0.88 (0.85-0.90)||<.0001a|
|75-85 years||22.1||0.34 (0.25-0.45)||<.0001a|
|85 years and older||3.3||0.04 (0.02-0.11)||<.0001a|
|Other, including black||21.2||0.63 (0.39-1.01)||.0542|
|SES quintile, per quintile increase||30.2||1.13 (1.03. 1.24)||.0097a|
|6-month diagnosis block, per 6-month increase||1.10 (1.02-1.19)||.0161a|
|Late 2000||33.8||1.96 (1.21-3.17)||.0061a|
|Early 2001||29.6||1.62 (1.01-2.61)||.0478a|
|Late 2001||29.7||1.63 (0.99-2.68)||.0555a|
|Early 2002||36.4||2.20 (1.38-3.51)||.0010a|
|Late 2002||32.2||1.82 (1.13-2.94)||.0135a|
|Charlson comorbidity score, per score increase||30.2||0.81 (0.71-0.93)||.0019a|
|Gastrectomy at teaching hospital|
|Distance to RT facility, per additional mile||1.00 (1.00-1.00)||.1032|
On multivariate analysis, younger age (P < .0001), lymph node involvement (P < .0001), and more recent diagnosis (P = .0284) were associated with chemoradiation after gastrectomy (Table 4). There was a trend toward the use of chemoradiation among patients with less comorbidity (P = .0515).
|Covariate||Adjusted Odds Ratio (95% CI)||P|
|Age at diagnosis, per year increase||0.87 (0.85-0.90)||<.0001b|
|Charlson comorbidity score, per score increase||0.87 (0.75-1.00)||.0515|
|6-month diagnosis block, per 6-month block increase||1.10 (1.01-1.21)||.0284b|
The survival of the subjects who received both adjuvant chemotherapy and radiation therapy was compared with the survival of the subjects who received no adjuvant therapy. In total, 62% (554 of 892) of these subjects died during the follow-up period. The median follow-up was 25.5 months after gastrectomy (range, 4-62). On Cox proportional hazards multivariate analysis, older patients (P < .0001) and patients with lymph node involvement (P < .0001), T3 or T4 disease (P = .0472), higher grade disease (P = .0355), and comorbidity (P = .0411) were more likely to die (Table 5). In addition, Asian patients were less likely to die than white patients (hazard ratio [HR], 0.73; 95% confidence interval [CI], 0.54-0.97; P = .0327). After adjustment for other factors, receipt of adjuvant chemoradiation therapy did not significantly increase survival among this elderly population (HR, 0.90; 95% CI, 0.72-1.12; P = .3453). The median survival among individuals who received chemoradiation therapy was 25.4 months versus 25.5 months for those who did not. Receipt of adjuvant chemoradiation therapy also did not significantly increase survival in a multivariate analysis that included propensity scores (HR, 0.87; 95% CI, 0.69-1.09; P = .2090), or in an analysis limited to patients with stage III or IV disease (HR, 0.98; 95% CI, 0.70-1.38; P = .9115). A sensitivity analysis evaluating gastric cancer survival rather than overall survival determined that adjuvant chemoradiation therapy was not associated with improved gastric cancer survival (P = .665).
|Covariate||Hazard Ratio (95% CI)||P|
|No adjuvant therapy||1.0||—|
|Chemotherapy and radiation therapy||0.90 (0.72-1.12)||.3453|
|Age at diagnosis, per year increase||1.03 (1.01-1.04)||.0002b|
|Other, including black||1.13 (0.84-1.53)||.4243|
|SES quintile, per quintile increase||1.00 (0.93-1.06)||.9277|
|6-month diagnosis block, per block increase||0.97 (0.92-1.03)||.3190|
|T3 or T4||1.60 (1.01-2.55)||.0472b|
|3-4||1.98 (1.05-3.75)||.0355 b|
|Charlson comorbidity score, per score increase||1.08 (1.00-1.16)||.0411 b|
|Gastrectomy at teaching hospital|
|Distance to RT facility, per additional mile||1.00 (1.00-1.00)||.9661|
No significant interaction between age and receipt of adjuvant chemoradiation (P = .4903) or between nodal involvement and receipt of adjuvant chemoradiation (P = .2724) was noted on survival analysis.
Although a landmark trial demonstrated that postoperative chemoradiation improved survival for patients with locally advanced resected gastric adenocarcimona,1 our population-based analysis of 1023 patients aged 65 years and older with resected gastric adenocarcinoma found no significant survival benefit from postoperative chemoradiation therapy.
In our study, elderly patients most likely to clinically benefit from adjuvant therapy—those with fewer comorbidities, those with more advanced disease, and the younger elderly—more likely to receive adjuvant chemoradiation. Similar treatment patterns have been demonstrated in elderly patients undergoing cancer treatment for prostate, breast, colon, and ovarian cancer.18-21 Patients diagnosed with gastric adenocarcinoma during the later months of the study period were more likely to receive adjuvant chemoradiation therapy, reflecting the dissemination of trial results and adoption of adjuvant chemoradiation into clinical practice. An initial report of the INT-0116 trial was presented in May 2000,10 and the findings were published in September 2001.1 Prior studies, using receipt of radiation therapy as a proxy for receipt of both radiation therapy and chemotherapy, demonstrated an increase in adjuvant radiation therapy administration among patients of all ages after the May 2000 presentation of study results.7, 22 We postulated that patients undergoing gastrectomy at teaching institutions would be more likely to receive adjuvant therapy, potentially reflecting early adoption of the study results at teaching institutions, and that patients living closer to radiation therapy facilities would be more likely to receive adjuvant therapy, because daily travel for radiation treatments would be easier for them. However, neither surgery at a teaching institution nor living closer to a radiation therapy facility were associated with receiving adjuvant therapy.
Studies of patients diagnosed with other types of cancer have demonstrated that older patients are less likely to receive standard cancer treatment than younger patients, even when such treatments are potentially curative.3-5 During our study period, adjuvant chemoradiation was administered to <⅓ of the patients. Cancer stage at diagnosis and the perception of the elderly as frail are potential explanations for why such a small proportion of the patients in our study received adjuvant chemoradiation. More than half of the patients in our study had a Charlson comorbidity score of zero, because patients must be healthy to tolerate a gastrectomy. However, if these elderly patients had a decrease in functional status because of surgery, they may no longer be good candidates for adjuvant therapy. Patients may not have received adjuvant therapy because of physician or patient concern about possible treatment toxicity in this elderly population, as the acute toxicity of adjuvant chemoradiation reported in the INT-0116 trial was considerable. Fifty-four percent of patients experienced grade 3 or worse National Cancer Institute-Common Toxicity Criteria (NCI-CTC) hematological toxicity, and ⅓ of patients experienced grade 3 or worse NCI-CTC gastrointestinal toxicity.1 Early cancer stage at diagnosis is another potential explanation for why such a small proportion of the elderly patients in our study received adjuvant chemoradiation. Patients in our study had earlier stage disease than the patients in INT-0116; 71% had T1 or T2, and 32% had N0 compared with only 32% with T1 or T2 and 15% with N0 disease in INT-0116. The applicability of the INT-0116 results to patients with early stage disease has been questioned because of the small number of study patients with early stage disease and because of the relatively good prognosis among those with early stage disease.6, 23 The smaller percentage of patients with early stage disease (T1, T2, or N0) enrolled in the INT-0116 trial compared with our population-based cohort suggests that only a small proportion of such patients were considered for enrollment in INT-0116. This enrollment bias may reflect a preconceived belief that adjuvant therapy is not necessary for early stage patients. Alternatively, the stage distribution in our study may reflect a tendency to forgo surgery in older patients with locally advanced disease.
In our population-based multivariate analysis that adjusted for clinical and demographic differences between treatment groups, the addition of postoperative chemoradiation did not improve survival for elderly patients with resected gastric adenocarcinoma. The median survival was 25.4 months among those who received adjuvant chemoradiation and 25.5 months among those who did not receive combined adjuvant chemoradiation. This is in stark contrast to the 9-month survival benefit, from 26 to 35 months, seen with the addition of chemoradiotherapy in the INT-0116 trial.1, 24 The median follow-up in our study was not as long as the follow-up in INT-0116, but it was well beyond the time when the survival curves in INT-0116 diverged. Prior institutional series and population-based studies that suggested improved survival with the addition of adjuvant chemoradiation studied younger patients,25, 26 did not have information on medical comorbidities that can influence the likelihood of receiving adjuvant chemoradiation and truncate survival,25-27 and/or did not have information on chemotherapy administration.27
There are several potential explanations for why our study did not find a survival benefit. First, inadequate radiation therapy may have been administered to patients in our population-based cohort. The radiation field design and treatment planning for gastric cancer is technically challenging. Upon central review of treatment plans in the INT-0116 study, 34% of the radiation treatment plans required a change before radiation administration. If these plans were not changed, ⅔ of the deviations would have resulted in undertreatment of patients, whereas ⅓ had the potential to deliver extremely toxic radiation.28 If radiation treatment caused serious toxicity that aborted treatment before completion or if the radiation treatment target was missed, the patients would not benefit from radiation therapy. We did not have information regarding radiation treatment plans, radiation treatment dose, or whether chemotherapy and radiation therapy courses were completed after initiation of therapy. A consensus statement on appropriate radiation treatment fields was released in 2002, but was not available during the first half of our study period.29 Second, as mentioned previously, the patients in our study had earlier stage disease than INT-0116 patients and may not have benefited from chemoradiation because of the relatively good prognosis among those with early disease. Third, margin status may have impacted survival outcomes. Enrollment in INT-0116 required complete resection with negative margins, but margin status is not available in SEER-Medicare. However, data suggest that the impact of surgical margin involvement on survival among patients with resected gastric adenocarcinoma who receive adjuvant chemoradiation therapy is minimal.25, 30 The median survival of patients who received no adjuvant treatment in our study was similar to the median survival of patients in the observation arm of INT-0116. Moreover, although margin status cannot be examined in our dataset, our results reflect treatment patterns in the United States. Finally, the demographics of our population-based cohort were different from the INT-0116 trial (median age, 60 years); our cohort was older (median age, 76 years) and included a higher proportion of individuals who were Asian. We may not have found a survival benefit in our study because of competing causes of mortality in these elderly patients.
This study has additional limitations common to observational studies using administrative data. The data source only captures Medicare patients and has incomplete data on the roughly 15% of patients in managed care. Previous studies suggest that HMO patients tend to have fewer comorbidities than patients in the general Medicare population31 and that practice patterns in HMOs can differ significantly from those in a fee-for-service setting.32 However, other studies have found few significant differences in cancer diagnosis and treatment between managed care and fee-for-service patients.33 Methods for comorbidity adjustment are still undergoing development and revision.16 Although the SEER-Medicare database is large, gastric cancer is a relatively rare cancer, and only 30% of patients received adjuvant chemoradiation therapy; thus, our ability to detect significant associations was limited by the size of our study cohort.
We did not detect a survival benefit from the administration of adjuvant chemoradiation therapy in our population-based study. These results suggest that some elderly patients with resected gastric adenocarcinoma may not gain a survival benefit from adjuvant chemoradiation. These findings should be considered hypothesis generating, and further investigation is necessary. Randomized trials should enroll more elderly patients with gastric cancer and should stratify patients by age to permit subgroup analysis of the elderly to better understand the impact of treatment regimens on older patients.
This work was supported by a National Institutes of Health grant (1K07CA118629, R.S.P.) and a J C R T Foundation grant (K.E.H.).
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.