|
|
||
 |
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2005.04.5005 on April 17 2006 © 2006 American Society of Clinical Oncology. Duration of Adjuvant Chemotherapy for Colon Cancer and Survival Among the Elderly
From the Department of Medicine and the Herbert Irving Comprehensive Cancer Center, College of Physicians and Surgeons, and the Departments of Epidemiology and Biostatistics, Mailman School of Public Health, Columbia University; and the New York Presbyterian Hospital, New York, NY Address reprint requests to Alfred I. Neugut, MD, PhD, Division of Medical Oncology, Columbia University Medical Center, 722 W 168th St, Rm 725, New York, NY 10032; e-mail: ain1{at}columbia.edu
PURPOSE: In randomized trials, patients with stage III colon cancer who received 6 months of fluorouracil (FU)-based adjuvant chemotherapy had better survival than patients who did not. However, little is known about the predictors of, or the survival associated with, duration of chemotherapy in the community.
PATIENTS AND METHODS: The linked Surveillance, Epidemiology, and End Results-Medicare database was used to identify individuals
RESULTS: Among 1,722 patients who received 1 to 7 months of FU-based chemotherapy, older age, being unmarried, and having comorbid conditions were associated with receiving less than 5 months of treatment. Among the 1,579 patients who survived CONCLUSION: More than 30% of elderly patients who initiated FU-based chemotherapy for stage III colon cancer and survived for at least 8 months discontinued treatment early. Mortality rates among such patients were nearly twice as high as among patients who completed 5 to 7 months of treatment. If the association we observed between duration of treatment and survival is confirmed, additional investigation is warranted to determine whether dose-intensity, cumulative dose, or other factors related to receipt of full adjuvant treatment are responsible.
The use of adjuvant chemotherapy for localized breast cancer and colon cancer has been among the most important advances in medical oncology in terms of lives saved.1-3 For breast cancer adjuvant chemotherapy, studies have shown that patients who receive dose-intense treatment fare better than those who do not. Bonadonna et al4 demonstrated that the benefit of adjuvant chemotherapy was limited to patients who received  85% of the optimal dose, and that patients who received less treatment fared no better than those who received none. Large-scale clinical trials have established that dose reductions, delays, or interruptions of chemotherapy for breast cancer can reduce its benefits.5-7 However, as many as 50% of patients either fail to initiate or do not receive all their recommended cycles of chemotherapy in a timely fashion.8,9 Our group recently found that among women with breast cancer who initiated adjuvant chemotherapy, 28% terminated treatment early and that the earlier they terminated treatment, the higher was their mortality hazard ratio compared to those who completed treatment.10 Colon cancer is a significant public health problem in the United States. Among individuals age older than 75 years, it is the most frequently diagnosed malignancy11,12 and the second leading cause of cancer death.13-15 In the 1980s, the combination of fluorouracil (FU) with biochemically modulating agents, such as levamisole or folinic acid, was shown to extend survival for locally advanced stage III colon cancer.16 A meta-analysis confirmed the benefits of such chemotherapy for the elderly.17 Even before the North Central Cancer Treatment Group–National Cancer Institute of Canada trial reported that patients treated for 12 months do not fare better than those who received 6 months of adjuvant chemotherapy, 6 months of treatment was becoming accepted as the standard of care.3 No consensus has emerged at this time regarding the optimal dose-intensity of bolus FU. We, and others, have found that elderly patients benefit as much from FU-based adjuvant chemotherapy as do younger node-positive colon cancer patients treated during clinical trials, but that 50% or more of elderly patients with stage III colon cancer do not receive treatment.18-20 Since 1995, the standard of care for the adjuvant treatment of node-positive colon cancer has been FU-based chemotherapy administered for about 6 months,16,21 but no studies have explored the prevalence and effects of early termination. Therefore, we have analyzed the frequency, determinants, and outcomes of early discontinuation of FU-based chemotherapy in a nationwide sample of elderly colon cancer patients.
Data Source Our analysis used a database developed by Potosky et al22 in 1993, in which the files of patients with cancer from the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) cancer registry were linked with their Medicare claims files. The SEER database provides information on tumor histology, location, stage of disease, and individual demographics, as well as primary surgical and radiation treatment and survival data for cancers occurring in a representative sample (approximately 14% during the period of our study) of the US population. The Medicare database, maintained by the Health Care Finance Administration (HCFA) complements the SEER data with extensive diagnostic, treatment, and cost information.
Sample Selection The remaining patients were categorized by duration of treatment, age group at diagnosis, year of diagnosis, race/ethnicity, sex, marital status, number of positive lymph nodes, tumor grade (well/moderately differentiated or poorly differentiated), comorbidity score, residence (metropolitan or nonmetropolitan area), the type of hospital in which they were diagnosed (teaching or nonteaching), and the socioeconomic status (SES) of their census tract/zip code. We obtained data on age, race, sex, marital status, surgery, stage, tumor grade, lymph nodes, type of hospital, and area of residence from the SEER database, and data on comorbid conditions and treatment from Medicare.
Treatment With FU
SES
Comorbid Disease
Survival
Statistical Analysis Propensity scoring is a technique used to measure the likelihood that a patient will be treated solely based on that patient’s independent predictors of treatment.29 We used the coefficients for the predictors of 5 to 7 months (v 1 to 4 months) of treatment to assign to each patient a composite propensity score that could be used in lieu of the covariates included in the multivariable proportional hazards analysis. We added any covariates that remained statistically significant to the propensity score-adjusted model. The hazard ratio for the association of treatment duration with mortality derived from the propensity score–adjusted model did not differ appreciably from that derived from the multivariate analysis. We decided to present the multivariable analysis results because they include information about the association of mortality with the covariates. Unadjusted Kaplan-Meier analyses were used to compare the two treatment groups with respect to colon cancer–specific and overall survival.
Baseline Characteristics of the Analysis Group Of the 3,733 patients in the SEER-Medicare database who received a pathologically confirmed diagnosis of stage III colon cancer and were not members of an HMO during 1995 to 1999, most were non-Hispanic whites (86.5%), resided in a metropolitan area (87.6%), and were treated in nonteaching hospitals (79.1%; Table 1).
In a preliminary analysis of month-by-month duration of treatment, we found that the mortality rates of those who received 1, 2, 3, or 4 months of FU-based chemotherapy were similar to the mortality rates of those who received no treatment and markedly higher than those of patients who received 5 to 7 months of treatment (roughly consistent with the 1995 recommendation of 6 months of treatment3; Fig 1). In our subsequent analyses, therefore, we compared patients who received 1 to 4 months with patients who received 5 to 7 months of FU treatment.
Predictors of Receiving 5 to 7 Months of FU Chemotherapy Table 2 lists the results of the logistic regression analysis of determinants of duration of treatment. In our multivariable model, treatment for 5 to 7 months was associated with younger age, more recent year of diagnosis, being married, having a tumor of well/moderately differentiated grade, and having no comorbid conditions.
Early Mortality Some patients who discontinued treatment early may have done so because of deteriorating performance status and prognosis. Moreover, patients who died within 8 months after diagnosis did not have the opportunity to benefit from FU chemotherapy even if they initiated it. Table 3 summarizes the results of the logistic regression analysis of mortality within 8 months after diagnosis among patients who received 1 to 7 months of FU treatment. The predictors of early mortality were older age, earlier year of diagnosis, having more than one positive lymph node, having a tumor of poorly differentiated grade, and having a comorbidity score of more than 1.
Mortality After 8 Months Table 4 lists the results of the Cox proportional hazards regression analyses of factors associated with disease-specific and overall mortality among patients who had the opportunity to benefit from FU treatment. Among the 488 patients who received 1 to 4 months of therapy and survived 8 months, 154 were recorded as subsequently dying as a result of colon cancer, whereas 55 died as a result of other causes. For the 1,091 patients treated with 5 to 7 months of chemotherapy, 188 died as a result of colon cancer, whereas 79 died as a result of other causes. Disease-specific mortality among the 1,091 patients who received 5 to 7 months of treatment was half that of the 488 patients treated for 1 to 4 months (OR, 0.53; 95% CI, 0.43 to 0.66). Overall mortality was approximately 44% lower among patients who were treated longer (OR, 0.59; 95% CI, 0.49 to 0.71). More than one positive lymph node, a tumor of poorly differentiated grade, and a higher comorbidity score were also associated with worse disease-specific and overall mortality.
Survival In unadjusted Kaplan-Meier survival analyses, both colon cancer–specific (Fig 2A) and overall survival (Fig 2B) were better among those stage III colon cancer patients receiving 5 to 7 months of FU therapy than among those receiving 1 to 4 months of therapy.
In a population-based sample of colon cancer patients older than 65 years who survived at least 8 months after diagnosis and initiated FU-based chemotherapy, more than 30% discontinued their treatment early. Factors associated with early discontinuation included increasing age, white race, worse tumor grade, and having a comorbidity score higher than 1. Mortality among patients who discontinued treatment early was nearly twice as high as among those who did not. In a prior study by our group,19 mortality among patients who received any adjuvant FU was 37% lower than among those who received surgery alone. Our current analysis suggests that the benefit we observed previously was driven by patients who completed 6 months of treatment, consistent with the results of randomized trials.3 Some prior studies have found that black patients are less likely than white patients to initiate adjuvant therapy in the treatment of both breast cancer9 and colorectal cancer.18,19 Our group has also observed an association between race and early termination of breast cancer treatment.10 However, other investigators have found inconsistent associations between treatment adherence and race or SES.30,31 Although there was a trend toward more early discontinuation of chemotherapy among black patients (OR, 1.56; 95% CI, 0.98 to 2.48), race was not significantly associated with mortality. The literature on the quality of colon cancer treatment in relation to survival is sparse. One study of treatment compliance in 139 patients who were to receive 12 cycles of adjuvant chemotherapy for stage III colon cancer found that, among patients who received six or more cycles, the recurrence rate was 7%; among those who received five or fewer cycles, the recurrence rate was 36%, suggesting that survival was associated with the cumulative dose of FU.32 In randomized clinical trials, in which adherence is known to be higher than in the community, 15% to 30% of patients also fail to complete all of the prescribed cycles of FU-based chemotherapy33-35; dose reductions occurred frequently after the first two cycles.34 Six months of chemotherapy appears to be as effective as 12 months of therapy.3 In a recent trial, 3 months of continuous-infusion FU appeared to be as effective as 6 months of bolus FU36; however, the participants included both patients with stage II disease and patients who were treated with radiation therapy for rectal cancer. No other studies have analyzed the effects of limited bolus FU treatment with mortality. Our findings in this population-based sample of patients with colon cancer are strikingly similar to our findings and those of others among patients with breast cancer.4,10,37 We found that, like many breast cancer patients, many colon cancer patients failed to complete their adjuvant chemotherapy. Others have reported poor adherence to chemotherapy for lymphoma38 and poor adherence to hormonal therapy for breast cancer.39,40 Although we also found that early discontinuation of treatment was associated with poor survival, we cannot assume that that association is causal. Some patients may be at high risk for adverse outcomes because of the same factors that cause them to discontinue treatment: poor physical condition, performance status, psychological outlook, or health behaviors. In addition, some patients may have discontinued treatment early in response to toxicity. However, FU toxicity is unlikely to account for much of the mortality in our study sample; recent clinical trials have reported only a 0.5% therapy-related death rate among patients receiving FU therapy.41 Furthermore, we found that duration of treatment was more strongly associated with colon cancer–specific survival than with overall survival. This pattern reinforces the view that the duration of treatment is indeed contributing to a reduction in colon cancer mortality and does not represent merely selection bias. Most studies of treatment discontinuation have focused on oral medications, such as those for the treatment of HIV,42 heart disease,43,44 and psychiatric disorders.45 In these settings, poor adherence accounts for worsening disease, death, and increased health care costs.46,47 A decision to discontinue treatment may be based on such factors as the adverse effects of medications,48 patient lack of knowledge about the illness or the potential benefits of the treatment,49,50 financial barriers,30 poor patient-physician communication,49,51 and the presence of comorbid conditions.43,52,53 Personal belief in the efficacy of the therapy may also play a significant role in adherence.54 Our analysis could take few of these factors into account and could not determine what roles patients, physicians, families, and the health care system played in determining the duration of treatment. However, in a Medicare population, at least some financial barriers are absent, and we used Medicare data to control for comorbid conditions, which were indeed statistically significant predictors of treatment and early and later mortality. Another limitation of our study was a lack of detailed information on chemotherapy dose and dose-intensity. Although dose reductions may contribute to suboptimal outcomes, especially among elderly patients, adjustment for dose is unlikely to account for the difference in survival between patients treated for 1 to 4 months and patients treated for 5 to 7 months. Our inability to take dose into account seems more likely to have biased our results toward the null than to account for the strong association we observed. New treatments for colorectal cancer are superseding the FU-based regimens, but although these agents appear to improve survival, they are also more toxic than the FU-based regimens.55,56 Patients who receive these new chemotherapeutic agents may therefore be even more inclined than the patients we studied to discontinue treatment. For example, in a clinical trial, adding oxaliplatin to FU resulted in a lower likelihood of completing all cycles of chemotherapy (75% v 86%) and a reduced total dose of FU administered (84% v 98%).41 Likewise, patients in a randomized trial of continuous-infusion FU were more likely to discontinue adjuvant therapy than those in the arm with bolus FU and leucovorin, due to the increased difficulties associated with administration.35 Research is needed on ways to minimize the adverse effects of chemotherapeutic drugs and to facilitate adherence so that patients can receive in full whatever survival benefits the treatments can provide. Our study is the first to explore factors associated with the duration of adjuvant FU chemotherapy in patients with colon cancer. We found that more than 30% of patients were treated for only 1 to 4 months, and that such patients had poorer survival than those who received 5 to 7 months of treatment. We do not know whether these differences in duration of treatment accounted for the observed differences in survival or reflected factors for which we were unable to control. Additional investigation is warranted to determine whether dose-intensity, cumulative dose, or other factors related to receipt of full adjuvant treatment for colon cancer are associated with survival.
The authors indicated no potential conflicts of interest.
Conception and design: Alfred I. Neugut, Matthew Matasar, Russell McBride, Judith S. Jacobson, Victor R. Grann, Dawn L. Hershman Financial support: Alfred I. Neugut, Dawn L. Hershman Collection and assembly of data: Matthew Matasar, Xiaoyan Wang, Russell McBride, Dawn L. Hershman Data analysis and interpretation: Alfred I. Neugut, Matthew Matasar, Xiaoyan Wang, Russell McBride, Judith S. Jacobson, Wei-Yann Tsai, Victor R. Grann, Dawn L. Hershman Manuscript writing: Alfred I. Neugut, Judith S. Jacobson, Dawn L. Hershman Final approval of manuscript: Alfred I. Neugut, Xiaoyan Wang, Russell McBride, Judith S. Jacobson, Wei-Yann Tsai, Victor R. Grann, Dawn L. Hershman
We thank the Applied Research Branch, Division of Cancer Prevention and Population Science, National Cancer Institute; the Office of Information Services, the Office of Strategic Planning, HCFA; Information Management Services; and the Surveillance, Epidemiology, and End Results tumor registries in the creation of the SEER-Medicare database.
Supported by a K05 Award from the National Cancer Institute (NCI; Grant No. CA89155), a grant from the American Cancer Society (Grant No. RSGT-01-024-04-CPHPS), and a Department of Defense Breast Cancer Center of Excellence Award (Grant No. BC043120) to A.I.N; an R25 fellowship from the NCI (Grant No. CA94061) to M.M.; a T32 fellowship from NCI (Grant No. CA09529) to R.M.; and an American Society of Clinical Oncology Career Development Award and a K07 Award from the NCI (Grant No. CA95597) to D.L.H. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2006 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 tests to compare patients who received 1 to 4 months and 5 to 7 months of treatment, with respect to the variables listed above. We then developed proportional hazards regression models to analyze the mortality associated with each incremental month of treatment, compared with no treatment. Among patients who received 1 to 7 months of treatment, we used logistic regression to estimate odds ratios (ORs) and 95% CIs for factors associated with receiving 5 to 7 versus 1 to 4 months of FU chemotherapy and with dying in the first 8 months given the baseline risk factors. We then excluded patients who died within the first 8 months from our analysis and developed Cox proportional hazards models to compare colon cancer–specific and overall mortality rates among patients who received 5 to 7 months of treatment with those for patients treated for 1 to 4 months, controlling for the baseline risk factors. 
