Epidemiologic studies suggest that obese women are more likely to die of ovarian cancer than those of ideal body weight, but it is not known whether increased incidence, comorbidities common to obese women, or altered tumor biology is responsible for this difference. The current study attempted to determine the influence of excess body weight on ovarian cancer survival, disease progression, and clinicopathologic factors.
The records of patients undergoing surgery for epithelial ovarian cancer at Cedars Sinai Medical Center between January 1, 1996 and June 30, 2003 were reviewed for height, weight, age, comorbidities, and treatment-specific details. Statistical analyses included the Fisher exact test, Kaplan-Meier survival, and Cox regression analyses.
In all, 216 patients were identified. Eight percent were underweight (body mass index [BMI] < 18.5), 50% were ideal body weight (18.5 ≤ BMI < 25), 25% were overweight (25 ≤ BMI < 30), and 16% were obese (BMI ≥ 30). Age, comorbidities including coronary artery disease and venous thromboembolism, and rates of optimal surgical cytoreduction were similar among BMI strata. Diabetes and hypertension were more common in obese women. Ten (29%) of the obese patients had International Federation of Gynecology and Obstetrics (FIGO) Stage I disease, compared with 19 (10%) of the patients with BMI < 30 (P = .01). In a subcohort of 149 patients with Stage III or IV disease, a significant trend was identified favoring increased BMI as an independent negative factor for disease-free (P = .02) and overall (P = .02) survival.
Obesity is a significant health concern in the US, with a dramatically increasing incidence over the last decade. Over 30% of US adults 20 years of age and older are obese, and 65% of adults are considered overweight based on their body mass index (BMI).1 In addition to its well-recognized role in diabetes, hypertension, and heart disease, obesity is also both a risk factor and a poor prognostic factor for several human malignancies, including endometrial, renal, esophageal, breast, and colon cancers.2 The impact of obesity on epithelial ovarian cancer risk is less clear, with inconsistent findings reported from several population-based cohort studies.3–7 Whereas the Iowa Women's Health Study8 found a positive association between BMI > 30 at age 18 and the risk of premenopausal serous ovarian cancer, the study failed to show a relation between current obesity and ovarian cancer incidence in postmenopausal women, a pattern echoed a year later in a prospective cohort study of 1.1 million Norwegian women.3
Despite conflicting data regarding ovarian cancer incidence, there are several studies that suggest that obesity is a poor prognostic factor for ovarian cancer survival. In 2 cohort studies of over 495,000 prospectively followed healthy women, there was a significantly increased ovarian cancer mortality among obese women when compared with ideal body weight study participants (relative risk = 1.26 for BMI > 30).9, 10 These studies, however, were not designed to differentiate as to whether the increased mortality seen among obese women with ovarian cancer was due to an increased incidence, more aggressive tumor biology, or increased mortality due to comorbid conditions.
Given an apparent association between ovarian cancer mortality and obesity, we chose to examine the relative risks of mortality and disease progression attributable to obesity in our own patient population. We hypothesized that the increased ovarian cancer mortality in obese women is due to more aggressive tumor biology. Our objectives were to compare obese and normal weight controls with regard to overall and disease-free survival (DFS), to evaluate and compare comorbidities and pathological features that may be associated with obesity and could impact survival, and to compare surgical and chemotherapeutic treatment between obese and nonobese individuals.
MATERIALS AND METHODS
After Institutional Review Board approval, we performed a retrospective review of all patients undergoing primary cytoreductive surgery by 1 of 5 gynecologic oncologists at Cedars-Sinai Medical Center for epithelial ovarian or primary peritoneal carcinomas between January 1 1996 and June 30 2003. Patients with nonepithelial tumor histologies, borderline tumors, synchronous uterine and ovarian malignancies, and patients who underwent neoadjuvant chemotherapy were excluded from review. Patients for whom accurate measurements of height and weight were lacking were similarly excluded.
Medical records for all eligible patients were reviewed and the abstracted data included patient height, weight, age, and comorbidities. Height and weight were abstracted from the first postoperative visit to reduce any potentially confounding effect from ascites. Comorbidities included a history of coronary artery disease (CAD), hypertension (HTN), diabetes (DM), previous malignancy, pulmonary disease (COPD), and thromboembolic disease (VTE). A patient was considered positive for any of these comorbidities if they were noted before or at any time during her treatment. Treatment-related data, such as chemotherapy regimen and doses, date of recurrence, and date of last contact were also noted. Cause of death was presumed to be cancer-related if the patient had advanced recurrent disease at the time of death.
A subcohort analysis of advanced stage individuals was performed for patients with International Federation of Gynecology and Obstetrics (FIGO) Stage III/IV disease who had primary cytoreductive surgery followed by platinum- and taxane-based chemotherapy with intent to treat for a minimum of 6 cycles. Patients refusing chemotherapy, those treated primarily with other chemotherapeutic regimens, or for whom requisite data were lacking were excluded from this analysis.
Patients were stratified by BMI according to guidelines set forth by the World Health Organization (WHO): underweight was defined as < 18.5 kg/m2, ideal weight as BMI 18.5–24.9 kg/m2, overweight as BMI 25–29.9 kg/m2, and obesity as BMI ≥ 30 kg/m2.11 Survival data and disease progression were compared across the strata using a Fisher exact test, Wilcoxon-Mann-Whitney U-test, Kaplan-Meier survival, and Cox regression analysis. A P-value < 0.05 was considered statistically significant.
In all, 216 patients were eligible for review. Seventeen (8%) were underweight, 108 (50%) were of ideal weight, 56 (26%) were overweight, and 35 (16%) were obese. The median BMI of all patients was 23.6 kg/m2 (range, 15.9–47.7). Demographic factors and comorbidities were not significantly different among different BMI strata, with the exceptions of HTN (34% vs. 11%, P = .002) and DM (23% vs. 2%, P = .0002), which were more common in patients with BMI ≥ 30 than their ideal body weight counterparts (Table 1).
BMI indicates body mass index; N/A, not applicable; HTN, hypertension; CAD, coronary artery disease; VTE, venous thromboembolism; DM, diabetes mellitus; COPD, chronic obstructive pulmonary disease.
Comparisons are between obese and ideal body weight cohorts.
With regard to operative factors, estimated blood loss was similar across BMI strata (mean 578 cc for underweight, 495 cc for ideal weight, 423 cc for overweight, and 404 cc for obese patients). The rate of optimal cytoreduction for Stage III/IV patients was 92%; the ability to optimally cytoreduce patients was not impacted by patient BMI in this series. Eleven (85%) underweight, 80 (90%) ideal body weight, 41 (91%) overweight, and 24 (100%) obese patients had residual disease implants of <1 cm greatest diameter. Stage of disease, however, correlated with BMI. Ten (29%) of the obese patients had Stage I disease, compared with 12 (10%) of the patients with ideal body weight. Histology also varied according to BMI (Table 2). Obese patients were more likely to have a mucinous histology (P = .028) than ideal body weight patients. There was also an overall trend toward nonserous histology in obese women when compared with those of ideal body weight, although this did not reach statistical significance (P = .051). A total of 5 patients died of treatment-related causes: 2 had postoperative complications and 3 died during primary chemotherapy. Evaluating all 216 patients, overall survival (OS) did not differ significantly between the obese and ideal body weight cohorts (Fig. 1, median 62 vs. 80 months, respectively, P = .28).
Due to the disparity in stage of disease, we performed a secondary subgroup analysis for advanced-stage patients only. In order to gauge the effect of obesity on tumor progression, we selected all patients with Stage III or IV disease and stratified them by BMI. For these 146 patients, BMI did not negatively influence rates of optimal surgical cytoreduction (Table 3), as 70 (91%) patients with ideal body weight and 22 (100%) patients with BMI ≥ 30 were optimally cytoreduced. On average, 6 to 7 cycles of chemotherapy were given in each group. Only 6/146 patients were unable to complete prescribed chemotherapy due to toxicity and these did not correlate with BMI grouping. There was also no significant difference in the time interval between surgery and chemotherapy when comparing the obese and ideal body weight groups (mean 22 vs. 17 days, P = .28). Chemotherapy dose data was available for 83 of the 146 patients; a significant difference was found in the median dose of paclitaxel per square meter of body surface area between the ideal body weight and obese groups (average 167 mg/m2 vs. 155 mg/m2, respectively, P = .01). Given the multiple variables affecting carboplatin doses, this comparison was not undertaken.
In this population with advanced-stage disease, BMI > 25 was associated with decreased DFS (17 vs. 25 months for BMI < 25, P = .04). A significant trend was also noted favoring increasing BMI classification as an independent negative predictor of DFS (Fig. 2, log-rank test for trend, P = .02) and overall survival (Fig. 3, log-rank test for trend, P = .02). Using a Cox proportional hazards model including BMI as a continuous variable, we further confirmed the negative prognostic impact of each 1-unit increase in BMI on disease free (hazard ratio [HR], 1.042; 95% confidence interval [CI]: 1.009–1.076; P = .01) and overall (HR, 1.050; 95% CI: 1.005–1.097; P = .03) survival.
A relation between ovarian cancer mortality and obesity has been previously established, although the exact nature of this relation remains unclear. This study supports the hypothesis that obesity impacts ovarian cancer mortality by influencing tumor biology. In advanced-stage patients, we found significant differences in the risk of cancer progression and cancer-related mortality associated with increasing BMI in a fairly “dose-dependent” fashion. There are several possible explanations for our observation that obesity seems to confer an increased risk of cancer recurrence. A slightly lower dose of paclitaxel relative to body surface area (BSA) was found in obese patients than in ideal weight controls. Whereas these data were not available for all patients, it is possible that underdosing chemotherapy may play a role in the increased mortality for obese patients with advanced ovarian cancer. Additionally, some bias may occur in the measurement of patient body weight due to the presence of ascites, with those individuals with the greatest volume of ascites and worse prognosis having artificially increased BMI. We attempted to minimize this effect by using the body weight recorded at the first postoperative visit to calculate BMI, rather than that obtained at the initial patient visit.
Despite the above potential confounders, our study population was fairly homogeneous across BMI strata with regard to most treatment factors and comorbidities. A notable and anticipated exception is the increased incidence in hypertension and diabetes among obese women. Given, however, that rates of more serious comorbidities such as VTE and CAD were similar across BMI strata and nearly all mortalities were in the setting of advanced, recurrent disease, it seems unlikely that comorbidities alone can account for the decreased overall survival seen in obese women. From these data we hypothesize that the most likely explanation for the correlation of increased patient BMI to decreased DFS and OS is a biologic effect of obesity itself in tumor growth, apoptosis, or chemotherapy resistance pathways. There are some data in the literature supporting this relation in other cancers. For example, leptin, a 16-kDa protein encoded by the ob gene, is primarily produced by adipose tissue and has been shown in a colon cancer cell line to activate mitogen-activated protein kinase (MAP-kinase) and nuclear factor kappa B (NF-κB) pathways. This activity led to apoptosis resistance after the addition of sodium butyrate.12 Unbound (free) insulin-like growth factor 1 (IGF1) is more highly expressed in obese individuals than their slimmer counterparts,13 and has itself been shown to correlate with increased breast cancer risk in women. IGFBP-2, a binding protein for IGF, also seems to promote invasion in ovarian cancer.14
This study is limited by its retrospective nature, relatively small sample populations, and potential treatment bias. Absent associations, such as the lack of correlation between BMI and the incidence of VTE or CAD, may be due to insufficient power rather than true similarity of the populations. However, we report the first data examining the potential impact of obesity on disease progression and survival in epithelial ovarian cancer. Our findings, noting average increases in the risk of recurrence of 4% (HR 1.042), and the risk of death 5% (HR 1.050), for each 1 kg/m2 of additional BMI are notable. Whereas further molecular studies are warranted, a correlation between obesity and adverse tumor biology seems likely. Studies are in progress to elucidate the mechanisms of this association.