Fax: (617) 632-5370
Influence of body mass index on outcomes and treatment-related toxicity in patients with colon carcinoma
Article first published online: 30 JUN 2003
Copyright © 2003 American Cancer Society
Volume 98, Issue 3, pages 484–495, 1 August 2003
How to Cite
Meyerhardt, J. A., Catalano, P. J., Haller, D. G., Mayer, R. J., Benson, A. B., Macdonald, J. S. and Fuchs, C. S. (2003), Influence of body mass index on outcomes and treatment-related toxicity in patients with colon carcinoma. Cancer, 98: 484–495. doi: 10.1002/cncr.11544
- Issue published online: 18 JUL 2003
- Article first published online: 30 JUN 2003
- Manuscript Accepted: 23 APR 2003
- Manuscript Revised: 17 APR 2003
- Manuscript Received: 9 DEC 2002
- colon carcinoma;
- body mass index;
- adjuvant chemotherapy;
- treatment-related toxicity
Obesity is a risk factor for the development of colon carcinoma. The influence of body mass index (BMI) on long-term outcomes and treatment-related toxicity in patients with colon carcinoma has not been well characterized.
This cohort study was conducted within a large, randomized adjuvant chemotherapy trial of 3759 men and women with high-risk, Stage II and Stage III colon carcinoma who were treated between 1988 and 1992 throughout the United States. With a median follow-up of 9.4 years, the authors examined the influence of BMI on disease recurrence, overall survival, and treatment-related toxicity.
Compared with women of normal weight (BMI, 21.0–24.9 kg/m2), obese women with colon carcinoma (BMI ≥ 30.0 kg/m2) experienced significantly worse overall mortality (hazard ratio [HR], 1.34; 95% confidence interval [95% CI], 1.07–1.67) and a nonsignificant increase in the risk of disease recurrence (HR, 1.24; 95% CI, 0.98–1.59). The influence of BMI among women was not related to any differences in chemotherapy dose-intensity across categories of BMI. In contrast, BMI was not related significantly to long-term outcomes among male patients in this cohort. Among all study participants, obese patients had significantly lower rates of Grade 3–4 leukopenia and lower rates of any Grade ≥ 3 toxicity compared with patients of normal weight.
Among women with Stage II–III colon carcinoma, obesity was associated with a significant increase in overall mortality as well as a borderline significant increase in disease recurrence. Nonetheless, obesity was not associated with any increase in chemotherapy-related toxicity. Cancer 2003;98:484–95. © 2003 American Cancer Society.
Data from both retrospective studies1–4 and prospective studies5–11 suggest a compelling association between obesity and the risk of developing colon carcinoma. In a large prospective study of male health professionals, a body mass index (BMI) ≥ 29.0 kg/m2 (obese men) was associated with an 82% increase in risk compared with a BMI of < 22.0 kg/m2 (normal and underweight men).6 This relation also was observed for patients with adenomas, particularly large adenomas, suggesting that some aspect of obesity is associated with the promotion of colorectal neoplasia.6, 12, 13 The mechanism by which obesity influences the incidence of colorectal neoplasia remains uncertain. Elevated glucose and insulin levels appear to promote the growth of adenomatous and malignant lesions in the bowel.14–16 Both BMI and visceral adiposity are critical determinants of insulin resistance and hyperinsulinemia.17–19
In contrast, the influence of obesity in patients with established colon carcinoma largely is unknown.20, 21 Among women with breast carcinoma, obesity has been linked to inferior recurrence-free and long-term survival.22–25 Moreover, elevated fasting serum insulin levels predicted an increased risk of disease recurrence among nondiabetic women with early-stage breast carcinoma.24
Therefore, we used data from a large, randomized trial of adjuvant chemotherapy to examine the influence of BMI on long-term outcomes after primary surgical treatment in patients with high-risk, stage II–III colon carcinoma. By studying patients who were enrolled in a prospective clinical trial, we were able to minimize confounding by inconsistent use of postoperative adjuvant therapy; to control for other clinical predictors of outcome; and to examine directly the influence of body habitus on overall survival, disease recurrence, and chemotherapy-related complications.
MATERIALS AND METHODS
Patients were participants in a clinical trial comparing different adjuvant chemotherapy regimens for Stage II and III colon carcinoma. The trial was conducted between August 1988 and July 1992 (Intergroup Trial 0089 [INT-0089]).26, 27 The study accrued 3759 patients, with participation by institutions affiliated with one of the following cooperative groups: Eastern Cooperative Oncology Group (ECOG; the sponsoring group), Southwest Oncology Group, or Cancer and Leukemia Group B (CALGB). Among the 3759 patients enrolled in the trial, 198 (5.3%) were deemed ineligible, principally due to failure to meet predefined eligibility criteria,26, 27 and were excluded from the current analysis. Details of the treatment trial were published recently.28
During the enrollment of patients into INT-0089, treating clinicians were required to complete and return treatment flow sheets to the ECOG data center. On these forms, clinicians recorded patient's height, weight on the first day of treatment, and first dose of 5-fluorouracil (5-FU) in total milligrams. Weight and height were determined by the procedures in place at each clinical site. Study staff members, who were blinded to patient outcomes, reviewed the charts of all eligible patients (n = 3561 patients) to extract these data. Among the patients who were considered eligible for the treatment trial, we excluded 123 patients whose height or Day 1 weight was not recorded clearly. Thus, our study cohort included 3438 patients who were eligible for the analysis.
In the treatment trial, eligible patients were stratified by extent of tumor invasion, presence or absence of obstruction, presence or absence of regional peritoneal or mesenteric implants, and extent of regional lymph node metastases. Stratification was performed separately within each of the three national cooperative groups. Patients then were assigned by permuted block randomization to one of four 5-FU-based treatment arms (Fig. 1).
Body Mass Index and Body Surface Area
The level of adiposity was measured by BMI, which was calculated as the patient's weight on Day 1 of chemotherapy (in kilograms; kg) divided by the patient's height squared (in meters; m2). Patients were divided into 5 categories of BMI: < 21.0 kg/m2, 21.0–24.9 kg/m2, 25.0–27.49 kg/m2, 27.5–29.9 kg/m2, and ≥ 30.0 kg/m2, consistent with prior studies5, 10 and with the World Health Organization classification system.29 Body surface area (BSA; in m2), a unit of measure used in the calculation of chemotherapy dosage, was calculated as follows: the square root of (height in centimeters multiplied by Day 1 weight in kilograms, divided by 3600).30 The expected dose of 5-FU using the patient's treatment assignment and the calculated BSA was considered the patient's weight-based dose. The percentage of the weight-based doses each patient actually received is the ratio of the dose recorded in the patient's study chart by the weight-based dose.
The primary endpoint of the current analysis was overall survival (OS) in the study cohort, defined as the time from study entry to death from any cause. Disease free survival (DFS) was defined as the time from study entry to tumor recurrence, occurrence of a new primary colorectal tumor, or death due to any cause. Finally, recurrence-free survival (RFS) was defined as the time from study entry to tumor recurrence or occurrence of a new primary colorectal tumor. For RFS, patients who died without known tumor recurrence were censored at the last documented evaluation by the treating provider. This definition of recurrence is consistent with other analyses in this cohort (INT-0089), although restriction to tumor recurrences only without inclusion of new primary tumors did not change our findings appreciably (data not shown).
Treatment-related toxicity was recorded by grade according to the National Cancer Institute Common Toxicity Criteria (version 1). Toxicity was assessed and documented at each treatment administration by qualified medical personnel (physician or oncology nurse).
Patients were followed at least every 3 months from the time of study entry for 1 year, then every 6 months for 5 years, and then annually until death. After treatment, providers were required to perform a history and physical examination, a complete blood count, chemistry analyses (including liver function tests), and a chest X-ray at follow-up. Data on treatment outcomes and therapeutic efficacy from INT-0089 had reached maturity at the time of the current analysis.
The distribution of baseline characteristics across BMI categories was evaluated using chi-square tests for categoric variables and analyses of variance for continuous variables. OS, DFS, and RFS at 5 years were examined using the methods of Kaplan and Meier,31 and differences were assessed with the log rank test. We hypothesized that patients who were underweight may have worse outcomes and, thus, designated patients with BMI 21.0–24.9 kg/m2 the reference group. The entire cohort was analyzed using a Cox proportional hazards regression model,32 with a priori inclusion in the model of age (age groups: younger than 50 years, 50–60 years, 60–70 years, and older than 70 years), gender, race, clinical bowel obstruction at presentation, clinical bowel perforation at presentation, baseline ECOG performance status, stage of disease, presence of peritoneal implants, completion of assigned adjuvant chemotherapy, and BMI class; in addition, based on differences in baseline characteristics, the predominant macroscopic pathologic feature was included in the model.
Toxicity rates were calculated for severe toxicities. These rates were compared across BMI categories using chi-square tests. We also assessed maximum toxicity, which was defined as the highest grade of any toxicity experienced by each patient during the course of adjuvant chemotherapy. Logistic regression analysis was performed to adjust toxicity rates for age, race, gender, baseline ECOG performance status, stage of disease, and treatment arm assignment. Differences in the maximum toxicity experienced by patients were assessed and adjusted for the aforementioned confounders with an ordinal logistic regression model.
We used SAS software (version 8.2; SAS Institute, Cary, NC) for all statistical analyses. All P values are two-sided.
Baseline Characteristics by BMI Category
Of 3438 patients who were included in our analysis, the distribution of patients by BMI category is shown in Table 1. Patients with a BMI of 21.0–24.9 kg/m2 were in the normal weight category, patients with a BMI < 21.0 kg/m2 were in the underweight category, and patients with a BMI ≥ 30.0 kg/m2 were in the obese category.10, 29 We prospectively compared the collected baseline characteristics of the patients enrolled in the treatment trial by BMI class. Compared with individuals of normal weight, underweight patients were more likely to be female, to have a lower performance status at the initiation of adjuvant chemotherapy, to present with bowel obstruction, to have right-sided tumors, and to have a more favorable Duke stage. Underweight patients did not differ significantly from patients of normal weight with respect to race, bowel perforation at presentation, tumor location, adjuvant chemotherapy treatment assignment, or completion of adjuvant therapy.
|Characteristic||BMI class (kg/m2)||P valuea||P value (≥ 21.0 kg/m2)b|
|< 21.0||21.0–24.9||25.0–27.49||27.5–29.9||≥ 30.0|
|No. of patients||488||1166||712||472||600||—||—|
|Percent of cohort||14.2||33.9||20.7||13.7||17.5||—||—|
|Mean age (yrs)||61.2||62.6||62.3||61.4||61.2||0.0005c||0.0001|
|Males (%)||36.3||54.0||64.5||63.1||53.7||< 0.0001||0.52|
|Performance status (%)d|
|Bowel obstruction at presentation (%)||41.4||32.8||30.1||27.1||26.7||< 0.0001||0.003|
|Bowel perforation at presentation (%)||5.8||6.3||5.1||7.6||5.0||0.33||0.64|
|Peritoneal implants at surgery (%)||7.4||7.6||7.7||8.9||10.3||0.27||0.041|
|Tumor location (%)e|
|Predominant macroscopic pathologic feature (%)|
|Tumor status/depth of tumor penetration (%)|
|Lymph node status/no. of positive lymph nodes (%)|
|Duke stage (%)|
|B2 (Stage II)||15.0||12.7||11.1||10.6||10.5||< 0.0001||0.074|
|B3 (Stage II)||12.9||6.8||5.8||6.6||5.5||—||—|
|C (Stage III)||72.1||80.5||83.1||82.8||84.0||—||—|
|Treatment arm (%)|
|Completed adjuvant therapy regimen (%)|
Among normal-weight and overweight patients, increasing BMI was associated with a greater prevalence of peritoneal implants but a lower rate of bowel obstruction at presentation. Furthermore, the macroscopic pathologic feature differed significantly according to BMI class. In contrast, among normal-weight and overweight patients, the BMI was unrelated to baseline performance status, gender, race, bowel perforation, number of positive lymph nodes, extent of tumor penetration in the bowel wall, adjuvant chemotherapy treatment assignment, and completion of adjuvant therapy.
Survival and Disease Recurrence by Body Mass Index Class
No significant OS advantage was observed among any of the four treatment arms in this study of adjuvant chemotherapy, as reported previously.26, 27 Consequently, patients in all four treatment arms were analyzed jointly according to BMI category. The median follow-up at the time of the current analysis was 9.4 years, with a maximum follow-up of 12.7 years.
After excluding underweight patients (BMI < 21.0 kg/m2), we observed no significant differences in DFS, OS, or RFS according to BMI classes of normal-weight, overweight, and obese individuals (Table 2, Fig. 2). Similarly, when comparing underweight patients (BMI < 21.0 kg/m2) with patients of normal weight (BMI, 21.0–24.9 kg/m2), there was no significant difference in 5-year DFS (P = 0.15), OS (P = 0.24), or RFS (P = 0.70).
|Measure||BMI class (kg/m2)||P value|
|< 21.0||21.0–24.9||25.0–27.49||27.5–29.9||≥ 30.0|
|Five yr DFS (%)a||56.4||60.0||56.8||57.9||57.8||0.60b|
|Five yr OS (%)c||63.5||67.0||65.2||65.8||64.8||0.56b|
|Five yr RFS (%)d||63.4||65.1||61.9||62.3||61.9||0.56b|
|Unadjusted HR (95% CI)|
|Overall mortality||1.10 (0.94–1.28)||1.00e||1.10 (0.96–1.26)||1.03 (0.88–1.21)||1.06 (0.92–1.23)||0.43f|
|Disease recurrence||1.04 (0.87–1.24)||1.00e||1.07 (0.92–1.24)||1.10 (0.93–1.31)||1.10 (0.94–1.29)||0.20f|
|Adjusted HR (95% CI)g|
|Overall mortality||1.15 (0.98–1.35)||1.00e||1.10 (0.95–1.26)||1.05 (0.90–1.24)||1.11 (0.96–1.29)||0.20f|
|Disease recurrence||1.06 (0.88–1.27)||1.00e||1.06 (0.88–1.27)||1.12 (0.94–1.33)||1.11 (0.94–1.30)||0.17f|
Unadjusted and adjusted hazard ratios (HR) and confidence intervals (CIs) for overall mortality and disease recurrence for each BMI class are shown in Table 2. For this analysis, patients in the normal weight category (21.0–24.9 kg/m2) were considered the referent group. In Cox proportional hazards models, we adjusted for age, gender, race, initial ECOG performance status, predominant macroscopic pathologic feature, clinical bowel obstruction at presentation, clinical bowel perforation at presentation, stage of disease, presence of peritoneal implants, and whether the patient completed adjuvant therapy. Compared with patients in the normal weight group, there were no significant differences in overall mortality or disease recurrence for either underweight patients or obese patients (BMI ≥ 30.0 kg/m2).
The distribution of adiposity differs considerably between men and women.33 Moreover, the influence of obesity on the risk of developing colon carcinoma appears to vary according to gender.16, 34–36 We examined the impact of BMI on colon carcinoma survival according to gender (Fig. 3, Table 3). Among women, obese patients had a significantly worse OS (HR, 1.34; 95% CI, 1.07–1.67; P for trend = 0.007) compared with patients of normal weight. Furthermore, female obese patients were more likely to experience disease recurrence compared with female patients of normal weight (HR, 1.24; 95%CI, 0.98–1.59), although this association did not reach statistical significance (P = 0.08). Excluding underweight females, the linear trend for colon carcinoma recurrence with increasing BMI was at the border of statistical significance (P = 0.06).
|Survival||BMI class (kg/m2)||P value|
|< 21.0||21.0–24.9||25.0–27.49||27.5–29.9||≥ 30.0|
|Five yr OS (%)||68.3||71.6||66.9||64.4||64.7||0.066a|
|Five yr RFS (%)||65.1||68.1||63.5||62.2||61.4||0.19a|
|Adjusted HR (95% CI)b|
|Overall mortality||1.08 (0.87–1.35)||1.00c||1.18 (0.94–1.49)||1.23 (0.95–1.60)||1.34 (1.07–1.67)||0.007d|
|Disease recurrence||1.01 (0.79–1.28)||1.00c||1.14 (0.89–1.47)||1.20 (0.91–1.60)||1.24 (0.98–1.59)||0.061d|
|Five yr OS (%)||55.0||63.1||64.2||66.7||64.9||0.51a|
|Five yr RFS (%)||60.1||62.5||61.0||62.4||62.4||0.99a|
|Adjusted HR (95% CI)b|
|Overall mortality||1.33 (1.05–1.67)||1.00c||1.03 (0.87–1.22)||0.96 (0.78–1.17)||0.94 (0.77–1.15)||0.39d|
|Disease recurrence||1.22 (0.93–1.60)||1.00c||1.00 (0.82–1.22)||1.05 (0.85–1.32)||0.98 (0.79–1.23)||0.93d|
In contrast, among normal-weight, overweight, and obese male patients, BMI had no significant influence on either overall mortality or disease recurrence. However, underweight men (BMI < 21.0 kg/m2) experienced a significantly increased risk of mortality (HR, 1.33; 95%CI, 1.05–1.67) and a nonsignificant increased risk of disease recurrence (HR, 1.22; 95%CI, 0.93–1.60). To test whether there was a significant difference in the relative effect of BMI among men and women, we entered into the proportional hazards model the cross-product interaction term of BMI and gender. We observed a statistically significant interaction between BMI and gender in the analysis of overall mortality (P for interaction = 0.02) and disease recurrence (P = 0.05).
We hypothesized that this gender interaction may be related to the effects of estrogen. Although the menopausal status of women in the trial was not recorded, we tested whether these findings were different among women younger and older than 50 years. The findings were considerably stronger for women 50 years or younger (n = 253 women), with an HR for disease recurrence in obese women compared with normal-weight women of 2.58 (95% CI, 1.23–5.42; P trend for BMI classes ≥ 21.0 kg/m2 = 0.0067). In contrast, disease recurrences increased marginally without reaching statistical significance in obese women older than 50 years (HR, 1.17; 95% CI, 0.90–1.51; P value for trend = 0.29). Furthermore, stratification of men by an age cut-off of 50 years did not demonstrate any significant differences between men 50 years or younger and men older than 50 years in either overall mortality rates or disease recurrence rates.
Although the rates of completion of adjuvant chemotherapy did not differ by BMI class, we repeated our analysis after excluding patients who did not complete their assigned chemotherapy. Compared with women of normal weight, obese women who completed therapy had a more significant increase in overall mortality (HR, 1.34; 95% CI, 1.04–1.73) and disease recurrence (HR, 1.38; 95% CI, 1.05–1.82). Furthermore, for women with BMI ≥ 21.0 kg/m2, there was a statistically significant trend toward worse overall mortality (P = 0.01) and disease recurrence (P = 0.02) with increasing adiposity. In contrast, we observed no significant correlation between BMI and survival among men who completed adjuvant chemotherapy.
The inverse relation between BMI and survival among women may reflect chemotherapy dosing levels that were calculated according to ideal weight rather than actual weight among obese women. We examined dose-intensity according to BMI to determine whether clinicians were using a modified chemotherapy doses for overweight and obese patients. We calculated the expected first dose of 5-FU (in milligrams) based on the patient's body surface area30 and the patient's treatment assignment on the study. We then compared the actual first dose of 5-FU delivered with the expected first dose of 5-FU. For the purposes of this analysis, we considered patients underdosed if they received < 95% of the expected dose. Despite protocol specifications to dose patients based on their actual weight, there were modest differences in the percentages of women who were underdosed (2.9% of women with BMI < 21.0 kg/m2, 2.1% of women with BMI 21.0–24.9 kg/m2, 1.6% of women with BMI 25.0–27.49 kg/m2, 2.9% of women with BMI 27.5–29.9 kg/m2, and 4.9% of women with BMI ≥ 30.0 kg/m2; chi-square test, P = 0.18). Adjustment for underdosing in the proportional hazards model did not alter the relation between obesity and outcome among women. Furthermore, the administration of < 95% of the expected first 5-FU dose was not predictive of overall mortality or disease recurrence in our multivariate model. Finally, there were no differences between men and women with regard to the completion of therapy (78.1% men vs. 76.5% women; P = 0.30) and underdosing of chemotherapy (3.6% men vs. 2.6% women); similar results were observed when this comparison was restricted to obese patients.
In a prior analysis of this cohort, diabetes mellitus was associated with an increase in both overall mortality and disease recurrence.37 To examine the influence of obesity independent of this diabetes effect, we repeated our analyses in women after excluding women who had a history of diabetes. In a nondiabetic subset, obese women continued to experience worse overall mortality (HR, 1.36; 95% CI, 1.07–1.73) and disease recurrence (HR, 1.25; 95% CI, 0.96–1.63) compared with women of normal weight.
Weight as a Measure of Adiposity
Because some studies have used weight as the measure of adiposity,20 we analyzed the cohort based on weight at the time of initiation of chemotherapy. We used gender specified medians of weight and compared patients whose weight was equal to or less than the median with patients whose weight was greater than the median. Among all eligible patients, the median weight was 64 kg (141 pounds) for women and 79 kg (174 pounds) for men. Consistent with the analysis of BMI, there were no significant differences in overall mortality or disease recurrence for men based on the gender specific median weight. However, compared with women whose weight was equal to or less than the gender-specific median, women whose weight was greater than the median had a significantly greater risk of overall mortality (HR, 1.20; 95% CI, 1.03–1.41) and disease recurrence (HR, 1.21; 95% CI, 1.02–1.43).
Treatment-Related Toxicity by Body Mass Index Class
We examined the influence of BMI on the rates of major chemotherapy-related toxicity (Table 4). Compared with patients in the normal weight group, underweight patients had higher rates of nausea, emesis, and stomatitis. After adjustment for age, treatment arm assignment, race, gender, baseline ECOG performance status, and stage of disease, only nausea remained statistically significant when comparing underweight patients with patients of normal weight (P = 0.02). In addition, toxicity rates according to BMI were not changed significantly when the analyses were limited to patients who received ≥ 95% of adjuvant chemotherapy doses based on their actual body weight.
|Toxicity (%)||BMI class (kg/m2)||Unadjusted P across all BMIsa||Adjusted (kg/m2)b|
|< 21.0||21.0–24.9||25.0–27.49||27.5–29.9||≥ 30.0||BMI ≥ 21.0c||BMI < 25.0d|
|Any Grade 3–4 toxicity||53.4||53.2||51.3||51.8||45.8||0.044||0.020||0.66|
Patients in increasing BMI classes experienced lower rates of Grade 3 and 4 leukopenia. The rate of leukopenia with increasing BMI > 21.0 kg/m2 remained statistically significant after adjusting for potential confounders of toxicity (P = 0.01). It is noteworthy that adjusting for underdosing of adjuvant chemotherapy did not alter the significant relation with leukopenia. Other major toxicities, including nausea, emesis, diarrhea, and stomatitis, did not differ materially across categories of patients with BMI > 21.0 kg/m2.
We also assessed maximum chemotherapy-related toxicity, which was defined as the highest grade of any toxicity experienced by each patient during the course of adjuvant chemotherapy. Compared with patients of normal weight, underweight patients did not experience a significant difference in the rate of any toxicity ≥ Grade 3 (adjusted P = 0.66). However, compared with normal-weight patients, overweight and obese patients experienced significantly lower rates of any toxicity ≥ Grade 3 (adjusted P for trend = 0.02). Fifty-three percent of normal-weight patients had any Grade 3–4 toxicity, compared with only 46% of obese patients. However, there were no significant differences in treatment-related deaths according to BMI class (P = 0.99). Although reports of differences in 5-FU toxicity according to patient gender have been reported,38 the differences in treatment-related toxicity according to BMI class were similar for both men and women.
Using data from a large, adjuvant chemotherapy trial of patients with high-risk, Stage II–III colon carcinoma, we found that increasing baseline BMI was associated with a significant increase in overall mortality among female patients, but not among male patients. Furthermore, obese women experienced a 24% increase in the risk of disease recurrence compared with women of normal weight, although this trend did not reach statistical significance. The influence of BMI among women was not related to differences in the dose intensity of adjuvant treatment, suggestive of a possible biologic influence of obesity.
In a previous study, Tartter et al. reported a significant difference in recurrence free rates using the Quetelet index (defined in the study as weight × 100/height2) in female patients with Stage II–III colon carcinoma, but not in male patients.20 However, that study was performed before the standard initiation of adjuvant chemotherapy, and residual confounding by other known predictors of disease recurrence could not be excluded in that analysis. After excluding underweight patients, Slattery et al. found that increasing baseline BMI was associated with a nonsignificant increase in overall mortality in a cohort of patients with colorectal carcinoma who were identified through the Utah Cancer Registry.21 However, that study was limited by the heterogeneous patient population and stage of disease and by an inability to adjust for treatment of disease and other potential clinical predictors of outcome in patients with colorectal carcinoma.
Numerous studies have reported a positive relation between obesity and the risk of colorectal carcinoma and adenoma.5–13 Although the mechanism for this correlation has not been characterized fully, excess adiposity, particularly visceral adiposity, is associated with insulin resistance and higher levels of circulating insulin.17–19 It has been shown that insulin is a promoter of colorectal neoplasia in an animal models,39 and elevated circulating insulin and C-peptide levels have been related prospectively to the risk of colon carcinoma.40, 41 In addition, circulating levels of insulin-like growth factor I (IGF-I), which promotes cell proliferation and inhibits apoptosis, have been associated positively with the risk of colon carcinoma in several studies.40, 42–45 Insulin increases the bioactivity of IGF-I by inhibiting the synthesis of certain IGF-binding proteins46 and by enhancing growth hormone-stimulated IGF-I synthesis.15
In the current study of patients with colon carcinoma, obese patients did not experience a higher rate of chemotherapy-related toxicity compared with patients of normal weight. In fact, obese patients had a significantly lower rate of Grade 3–4 leukopenia as well as any toxicity ≥ Grade 3. Poikonen et al. reported a similar finding in women who received adjuvant chemotherapy for lymph node–positive breast carcinoma.47 Patients with the highest BMIs (even patients who were dosed according to their actual body weight) had higher leukocyte nadir values compared with other women, confirming the finding that obese patients can tolerate actual weight-based doses of chemotherapy. Similarly, Georgiadis et al. found no increased chemotherapy-related toxicity in obese patients with small cell lung carcinoma who were dosed according to their actual body weight.48
Using clinical trial data to examine the influence of BMI offers several advantages over the use of data from other sources. First, the stage of disease was comparable within the constraints of the protocol entry criteria. Second, treatment and follow-up care were conducted according to a uniform standard, and the date and nature of disease recurrence were recorded prospectively. Third, weight and height were recorded prospectively by treating clinicians. Finally, detailed information on prognostic variables, such as the number of positive lymph nodes and performance status, were collected routinely.
The distribution of BMI in our trial was very consistent with data from the 1988–1994 National Health and Nutrition Examination Survey III (NHANES III) report on BMI distribution in the United States.49 In that nationwide survey, 32.6% of patients were overweight (BMI, 25.0–29.9 kg/m2), and 22.3% of patients were obese (BMI, ≥ 30.0 kg/m2), consistent with our findings of 34.3% and 17.5%, respectively.
Several potential limitations for our study should be considered. Because adjuvant chemotherapy was initiated within 5 weeks after patients underwent surgery for colon carcinoma, their body weight recorded at the time they received their first dose of chemotherapy may not reflect the patient's presurgery level of adiposity. We were able to adjust for some important predictors of perioperative weight loss, particularly clinical obstruction and perforation.49 Residual confounding by other predictors of colon carcinoma survival may limit the findings of studies that examine the impact of BMI on long-term outcomes. However, because patients in our cohort were enrolled on a prospective clinical trial, we were able to adjust our analyses for many such confounders. Moreover, all patients in our cohort received a prospectively assigned regimen of adjuvant therapy, and information on completion of therapy was available. Nonetheless, the influence of obesity on mortality and disease recurrence may have been more reflective of other BMI-related factors, such as physical activity, that were not assessed in our cohort.
Obesity may influence the outcome of patients with colon carcinoma through alternative mechanisms. It has been shown that the presence of concurrent adverse health conditions, or comorbidities, increases overall mortality among patients with malignant disease.50–53 The presence of obesity, as such, may have a negative influence on prognosis through other associated comorbidities that increase the risk of non-disease-related mortality. However, in the current study, obese women had increases in both total mortality and disease recurrence, suggesting the possibility of a more specific impact of obesity on disease progression in patients with colon carcinoma. Patients who were overweight and who potentially had fewer obesity-related comorbid conditions also had an increased risk of both endpoints. Furthermore, among patients without diabetes mellitus (a comorbidity that was collected during an extensive chart review37), obesity still was predictive of increased mortality and disease recurrence.
Consistent with the findings of Tartter et al.,20 the adverse influence of BMI on the outcome of patients with colon carcinoma in our study was restricted to women. The effect in younger women was consistent with studies demonstrating that the impact of obesity on developing colon carcinoma is seen primarily in premenopausal women.54, 55 Several explanations for this gender disparity are plausible. First, women have higher levels of the obesity-association protein, leptin,56, 57 and increasing BMI is correlated with increased leptin levels in females, but not males.58 It has been shown that leptin is a growth factor for colonic epithelial cells59 and promotes invasiveness of colonic epithelial cells.60 Second, some studies have demonstrated that the propensity for insulin resistance and adult-onset diabetes among obese individuals is stronger for women compared with men.61, 62 Third, using data from NHANES III, Wu and colleagues demonstrated that C-reactive protein levels (as a measure of inflammatory processes) were higher and were correlated more strongly with BMI and fasting insulin levels in women compared with in men.63 Finally, higher levels of circulating estrogen (particularly ovarian-source estrogen) among women may augment the tumor-promoting effect of insulin and IGF-I, both of which may be elevated in obese patients. Estrogen enhances the mitogenic effect of IGF-I, induces the expression of IGF-I, and stimulates production of the IGF-I receptor.64 This hypothesis is supported by our observation of increased disease recurrences in women younger than 50 years. Furthermore, some postmenopausal women may have been protected by the use of hormone replacement therapy (HRT), which has been shown to reduce the risk of developing colorectal carcinoma.65 Our database did not have concurrent medications available; thus, confounding by HRT use may explain in part the differences in premenopausal women and postmenopausal women.
It is noteworthy that patients with colon carcinoma who have an increased BMI tolerate adjuvant chemotherapy at least as well as patients of normal weight, even when they are treated at weight-based doses. Nonetheless, additional studies are needed to confirm the influence of obesity on the outcome of women with Stage II–III colon carcinoma. Although the magnitude of risk for obese women, compared with women of normal weight, is modest, the difference is comparable to the risk benefit of adjuvant 5-FU and leucovorin over surgery alone in a similar population of patients.66 Future studies should examine the influence of weight change, physical activity, diet, or other mechanisms on the prognosis of patients with colon carcinoma. In an ongoing trial of adjuvant chemotherapy for patients with Stage III colon carcinoma (CALGB study 89803), we are collecting data on patients' weight at the time chemotherapy is initiated and 6 months after its completion to analyze further the impact of BMI, change in body weight, diet, and physical activity on long-term outcomes.
The authors thank Shandi Smith and Trong Ao for their assistance in data collection for this study.
- 26Fluorouracil (FU), leucovorin (LV) and levamisole (LEV) adjuvant therapy for colon cancer: five-year final report of INT-0089 [abstract]. Proc Am Soc Clin Oncol. 1998; 17: 256a., , , .
- 27Fluorouracil (FU), leucovorin (LV) and levamisole (LEV) adjuvant therapy for colon cancer: four-year results of INT-0089 [abstract]. Proc Am Soc Clin Oncol. 1997; 16: 940a., , , .
- 29World Health Organization. Physical status: the use of interpretation of anthropometry: report of a WHO expert committee. World Health Organ Tech Rep Ser. 1995; 854: 1–452.
- 32Regression models and life tables. J R Stat Soc B. 1972; 34: 187–220..