Treatment effects, disease recurrence, and survival in obese women with early endometrial carcinoma

A Gynecologic Oncology Group study

Authors

  • Vivian E. von Gruenigen MD,

    Corresponding author
    1. Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, University Hospitals of Cleveland, MacDonald Women's Hospital, and the Ireland Cancer Center, Cleveland, Ohio
    2. Department of Reproductive Biology, Case Western Reserve University, Cleveland, Ohio
    • Division of Gynecologic Oncology, University MacDonald Women's Hospital, 11100 Euclid Avenue, Room 7128, Cleveland, Ohio 44106
    Search for more papers by this author
    • Fax: (216) 844-8772

  • Chunqiao Tian MS,

    1. Gynecologic Oncology Group Statistical and Data Center, Roswell Park Cancer Institute, Buffalo, New York
    Search for more papers by this author
  • Heidi Frasure MS,

    1. Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, University Hospitals of Cleveland, MacDonald Women's Hospital, and the Ireland Cancer Center, Cleveland, Ohio
    2. Department of Reproductive Biology, Case Western Reserve University, Cleveland, Ohio
    Search for more papers by this author
  • Steven Waggoner MD,

    1. Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, University Hospitals of Cleveland, MacDonald Women's Hospital, and the Ireland Cancer Center, Cleveland, Ohio
    2. Department of Reproductive Biology, Case Western Reserve University, Cleveland, Ohio
    Search for more papers by this author
  • Henry Keys MD,

    1. Department of Radiation Oncology, Albany Medical College, Albany, New York
    Current affiliation:
    1. 1 Foxcare Drive, Suite 310, Oneonta, New York 13820
    Search for more papers by this author
  • Richard R. Barakat MD

    1. Gynecology Service, Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, New York
    Search for more papers by this author

  • The following Gynecologic Oncology Group member institutions participated in the related treatment study: University of Alabama at Birmingham, Oregon Health Sciences University, Duke University Medical Center, Abington Memorial Hospital, University of Rochester Medical Center, Walter Reed Army Medical Center, University of Minnesota Medical School, University of Southern California Medical Center at Los Angeles, University of Mississippi Medical Center, Colorado Foundation for Medical Care, University of California Medical Center at Los Angeles, University of Washington Medical Center, University of Miami School of Medicine, Milton S. Hershey School of Medicine of the Pennsylvania State University, Georgetown University Hospital, University of Cincinnati College of Medicine, University of North Carolina School of Medicine, University of Iowa Hospitals and Clinics, University of Texas Health Science Center at Dallas, Indiana University School of Medicine, Bowman Gray School of Medicine of Wake Forest University, Albany Medical College of Union University, University of California Medical Center at Irvine, Tufts-New England Medical Center, Rush Presbyterian-St. Luke's Medical Center, State University of New York Downstate Medical Center, University of Kentucky, Eastern Virginia Medical School, Cleveland Clinic Foundation, Johns Hopkins Oncology Center, Pennsylvania Hospital, Washington University School of Medicine, Cooper Hospital/University Medical Center, Columbus Cancer Council, Fox Chase Cancer Center, Medical University of South Carolina, Women's Cancer Center, University of Oklahoma Health Science Center, University of Virginia Health Science Center, University of Chicago, and Tacoma General Hospital.

Abstract

BACKGROUND.

The objective was to examine whether rates of disease recurrence, treatment-related adverse effects, and survival differed between obese or morbidly obese and nonobese patients.

METHODS.

Data from patients who participated in a randomized trial of surgery with or without adjuvant radiation therapy were retrospectively reviewed.

RESULTS.

Body mass index (BMI) data were available for 380 patients, of whom 24% were overweight (BMI, 25–29.9), 41% were obese (BMI, 30–39.9), and 12% were morbidly obese (BMI, ≥40). BMI did not significantly differ based on age, performance status, histology, tumor grade, myometrial invasion, or lymphovascular-space involvement. BMI > 30 was more common in African Americans (73%) than non-African Americans (50%). Patients with a BMI ≥ 40 compared with BMI < 30 (hazards ratio [HR], 0.42; 95% confidence interval [CI], 0.09–1.84; P = .246) did not have lower recurrence rates. Compared with BMI < 30, there was no significant difference in survival in patients with BMI 30–39.9 (HR, 1.48; 95% CI, 0.82–2.70; P = .196); however, there was evidence for decreased survival in patients with BMI ≥ 40 (HR, 2.77; 95% CI, 1.21–6.36; P = .016). Unadjusted and adjusted BMI hazards ratios for African Americans versus non-African Americans in the current study differed, thus suggesting a confounding effect of BMI on race. Eight (67%) of 12 deaths among 45 morbidly obese patients were from noncancerous causes. For patients who received adjuvant radiation therapy, increased BMI was significantly associated with less gastrointestinal (R, −0.22; P = .003) and more cutaneous (R, 0.17; P = .019) toxicities.

RESULTS.

In the current study, obesity was associated with higher mortality from causes other than endometrial cancer but not disease recurrence. Increased BMI was also associated with more cutaneous and less gastrointestinal toxicity in patients who received adjuvant radiation therapy. Future recommendations include lifestyle intervention trials to improve survival in obese endometrial cancer patients. Cancer 2006. © 2006 American Cancer Society.

Endometrial carcinoma is the most common gynecologic malignancy in the United States. Although an annual estimated 40,880 new diagnoses and 7310 deaths are attributed to this disease, the majority has a favorable prognosis.1 However, the American Cancer Society's publications on cancer statistics suggest an increasing trend in new cases and estimated deaths. Risk factors for endometrial cancer include unopposed estrogen therapy, increasing age, diabetes, nulliparity, and obesity.2–5 Higher body mass index (BMI) is associated with an increased risk for developing endometrial cancer.6 Retrospective reviews written before the 1970s found that approximately 30% of patients with endometrial cancer were obese.3, 4, 7 A recent prospective study reported that 68% of women with early endometrial cancer are obese.8

In the United States, approximately 65% of women are overweight,9 31% are obese,9 and these numbers are expected to rise.10 Obesity places women at risk not only for endometrial cancer but also for other cancers, as well as cardiovascular disease, type 2 diabetes, asthma, and death.11 A recent prospective study by Calle et al. assessed the relationship between body mass index and risk of death from all cancers.12 The relative risk (RR) of death for morbidly obese women (BMI ≥ 40) with endometrial cancer was 6.25 compared with those who were of normal weight. Furthermore, obese endometrial cancer patients had the highest fatality rate among all cancers. Patients with endometrial cancer are unique given that the majority are obese, obesity is the primary risk factor for the disease, and these women are not making lifestyle changes after diagnosis.8

Obesity in cancer patients may effect overall survival, recurrence rates, and other clinical outcomes. For example, obese patients with breast or prostate cancer have higher recurrence and death rates compared with patients of normal weight.13–16 In addition, obese patients with breast cancer have a higher incidence of radiation complications.17

We retrospectively reviewed and compared data from patients who participated in a Gynecologic Oncology Group (GOG) randomized trial of surgery with or without adjuvant radiotherapy (GOG 99) as treatment for early stage endometrial cancer. The objective was to test the hypothesis that rates of disease recurrence, treatment-related adverse effects, and survival would differ between obese to morbidly obese patients and non-obese patients.

MATERIALS AND METHODS

Patient data for this analysis represent both arms of a randomized trial of surgery with or without adjuvant radiation therapy for early stage endometrial cancer (GOG 99) that was conducted between June 1987 and July 1995. Eligibility criteria for GOG 99 included histologically confirmed primary endometrial adenocarcinoma, surgery with total abdominal hysterectomy, and bilateral salpingo-oophorectomy, selective pelvic and para-aortic node sampling, and pelvic washings, and had been found to be surgical stage IB, IC, IIA (occult) and IIB (occult) disease. Participating institutions obtained institutional review board approval of the protocol before enrolling any patients; and all patients provided written informed consent consistent with all federal, state, and local requirements before they received any protocol therapy. Three hundred ninety-two eligible and consenting patients were enrolled onto GOG 99 and randomly allocated to receive surgery followed by either no additional treatment (NAT) or external-beam whole pelvic radiation (RT). Treatment information and outcome has been previously reported.18 For the current report, BMI data were available for 380 patients. BMI was calculated as a patient's baseline weight in kilograms divided by her height per meters squared, and BMI was defined as: underweight (BMI < 18.5), normal weight (BMI 18.5 to 24.9), overweight (BMI 25.0 to 29.9), obese (BMI 30.0 to 39.9), and morbidly obese (BMI ≥ 40.0) consistent with national guidelines.9

The current study endpoint was to explore the relation between BMI and recurrence-free interval (RFI), overall survival (OS), and treatment-related toxicity. RFI was defined as the time from study entry to clinical, pathological, or radiographic evidence of disease recurrence. OS was defined as the time from study entry to death from any cause. The association of BMI with patient characteristics was assessed by an analysis of variance (ANOVA) using BMI as a continuous variable. The impact of BMI on RFI or OS was assessed using a Cox model adjusted for other important prognostic factors (age, race, performance status, stage, histology, tumor grade, myometrial invasion, lymphovascular space involvement (LVSI), and RT status). For the survival analysis, data were categorized as: BMI < 30, BMI 30–39, and BMI ≥ 40. Survival distribution by BMI category was also estimated using the Kaplan-Meier method.

Treatment-related toxicity was graded according to the 1985 GOG Adverse Event Criteria. The association of BMI with toxicity was assessed by using rank correlation stratified by treatment arm (NAT vs RT). Although clinical significance is often determined by only grade 3 or 4 toxicities, some grade 2 categories are clinically important; thus, we analyzed all (grade 1–4) toxicities in this report. All statistical analyses were conducted on SAS Software 9.1 (SAS Institute, Cary, NC), and reported P-values were 2-sided.

RESULTS

BMI data for 380 patients were analyzed and distributed as follows: 1% underweight (BMI < 18.5), 23% normal weight (BMI 18.5 to 24.9), 24% overweight (BMI 25 to 29.9), 41% obese (BMI 30 to 39.9), and 12% morbidly obese (BMI ≥ 40) (Table 1). The median BMI was 30 (range, 16–63). There were no differences in BMI by age group, performance status, histologic type, tumor grade, myometrial invasion, or lymphovascular space involvement (LVSI). However, increasing BMI was more frequent in African-American patients than other groups (P = .002), with 73% of African Americans having a BMI > 30 compared with 50% of non-African Americans. Comorbidity data was not collected in the prior protocol. Among 380 patients, 44 experienced disease recurrence, and 61 had died (25 because of endometrial cancer and 36 because of other causes) at the writing of this report. The median follow-up time was 65 months.

Table 1. Average BMI by Patients Characteristics (N = 380)
CharacteristicNo. (%)Mean ± SDP
  1. BMI indicates body mass index. Analysis of variance (ANOVA) was used to compare the difference in mean BMI across groups.

Age, y  .080
 <5046 (12.1)32.4 ± 9.7
 50–5993 (24.5)32.4 ± 8.3
 60–69147 (38.7)31.1 ± 7.5
 ≥7094 (24.7)29.7 ± 6.2
Race  .002
 Caucasian315 (82.9)30.7 ± 7.8
 African American38 (10.0)35.3 ± 7.3
 Other27 (7.1)31.8 ± 5.9
GOG performance status  .742
 0228 (60.0)31.1 ± 7.9
 1 or 2152 (40.0)31.4 ± 7.5
FIGO stage  .056
 IB225 (59.2)31.9 ± 7.9
 IC120 (31.6)29.8 ± 6.8
 II, occult35 (9.2)31.4 ± 8.7
Myometrial invasion  .296
 Inner 1/3160 (42.1)31.6 ± 7.8
 Middle 1/3156 (41.1)31.4 ± 7.8
 Outer 1/364 (16.8)29.9 ± 7.3
Histology  .199
 Endometrioid286 (75.3)30.9 ± 7.9
 Other94 (24.7)21.1 ± 7.0
Tumor grade  .485
 Well-differentiated161 (42.4)31.7 ± 7.6
 Moderately differentiated150 (39.5)31.2 ± 8.2
 Poorly differentiated69 (18.2)30.3 ± 7.1
Lymphovascular space involvement  .632
 Yes85 (22.4)31.6 ± 8.0
 No295 (77.6)31.1 ± 7.7
Assigned treatment regimen  .231
 Surgery alone (NRT)193 (50.8)30.7 ± 7.9
 Surgery + RT (RT)187 (49.2)31.7 ± 7.6
All patients380 (100.0)31.2 ± 7.7 

In multivariate analysis, significant factors associated with increased risk of recurrence included increasing age, myometrial invasion >⅓, presence of LVSI, poorly differentiated tumor grade, and no adjuvant RT treatment (Table 2). Similarly, multivariate analysis of prognostic factors revealed that older age, myometrial invasion >⅓, presence of LVSI, and poorly differentiated tumor grade, but not adjuvant RT status were associated with worse overall survival. There was no significant difference toward decreased risk for disease recurrence in patients whose BMI ≥ 40 compared with those whose BMI < 30 (hazard ratio [HR], 0.42, 95% confidence interval [CI], 0.09–1.84, P = .246) (Table 2).

Table 2. Multivariate Analysis of Association Between BMI and Other Factors With Endometrial Cancer Recurrence
CharacteristicHazard ratio95% CIP
  1. Cox proportional hazards model used to estimate the relative risk.

Age, increase each 10 y1.441.03–2.01.035
Race
 Non-African AmericanReferent  
 African American0.890.26–3.05.853
GOG performance status
 0Referent  
 1 or 21.340.73–22.48.349
FIGO stage
 IBReferent  
 IC1.080.49–2.37.847
 II, occult1.890.71–5.63.223
Myometrial invasion
 Inner 1/3Referent  
 Middle 1/34.131.46–11.69.008
 Outer 1/34.111.15–14.67.030
Histology
 EndometrioidReferent  
 Other0.580.24–1.38.217
Tumor grade
 Well-differentiatedReferent  
 Moderately differentiated1.000.45–2.25.995
 Poorly differentiated2.160.98–4.74.056
Lymphovascular space involvement
 NoReferent  
 Yes2.991.50–5.97.002
Assigned treatment regimen
 Surgery alone, NRTReferent  
 Surgery + RT, RT0.480.25–0.95.034
BMI, kg/m2
<30Referent  
30 to 390.980.51–1.86.942
≥400.410.09–1.83.245

Compared with patients with a BMI < 30, survival time was not shorter in obese (BMI 30 to 39.9) patients (HR, 1.48, 95% CI, 0.82–2.70, P = .196), but survival time was significantly shorter in morbidly obese (BMI ≥ 40) patients (HR, 2.77, 95% CI, 1.21–6.36, P = .016). There was no interaction between BMI and treatment in multivariate analysis on RFI or OS, suggesting that the impact of BMI on prognosis was consistent by 2 treatment groups (Table 3).

Table 3. Multivariate Analysis of Association Between BMI and Other Factors With Overall Survival
CharacteristicHazard ratio95% CIP
  1. GOG indicates Gynecologic Oncology Group.

  2. Cox proportional hazards model used to estimate the relative risk.

Age, increase each 10 y2.091.49–2.93<.001
Race
 Non-African AmericanReferent  
 African American1.630.77–3.43.200
GOG performance status
 0Referent  
 1 or 21.090.64–1.86.745
FIGO stage
 IBReferent  
 IC1.140.55–2.34.724
 II, occult1.950.84–4.55.122
Myometrial invasion
 Inner 1/3Referent  
 Middle 1/32.581.22–5.46.013
 Outer 1/32.610.94–7.27.067
Histology
 EndometrioidReferent  
 Other0.760.39–1.48.422
Tumor grade
 Well-differentiatedReferent  
 Moderately differentiated0.840.44–1.63.612
 Poorly differentiated22.301.20–4.42.013
Lymphovascular space involvement
 NoReferent  
 Yes2.281.26–4.12.007
Assigned treatment regimen
 Surgery alone, NRTReferent  
 Surgery + RT, RT0.850.49–1.47.565
BMI, kg/m2
 <30Referent  
 30–391.470.81–2.67.208
 ≥402.761.20–6.33.017

Because increased BMI was more common among African-American patients, and these patients may have a poorer prognosis, the effect of race on OS was evaluated with and without controlling for BMI. Without adjusting for BMI, the hazard ratio between African-American patients and non–African-American patients was 2.24 (95% CI, 1.12–4.49, P = .023); after adjusting for BMI, the hazard ratio was 1.61 (95% CI, 0.77–3.40, P = 0.208). These results suggest that the confounding effect of BMI should be taken into account when evaluating race as a prognostic factor for endometrial cancer.

Figure 1 shows the estimates of OS by BMI. Morbidly obese patients (BMI ≥ 40) had a significantly shorter survival compared with those with BMI < 40. This difference appears more evident in later stages of follow-up. Furthermore, morbidly obese patients experienced a greater percentage of deaths unrelated to endometrial cancer compared with patients with BMI < 40 (Table 4). Among 45 morbidly obese patients, there were 12 deaths, 8 (67%) of which were from causes other than endometrial cancer. In 335 patients with BMI < 40, there were 49 deaths, of which 21 (43%) were unrelated to endometrial cancer. Causes of death of patients with BMI 30 to 40 were essentially the same as those of patients whose BMI < 30.

Figure 1.

Overall survival estimate by body mass index (BMI) group.

Table 4. Cause of Death by BMI Group
Cause of deathBMI < 30BMI 30 to 39.9BMI ≥≥ 40
No. (%)No. (%)No. (%)
Endometrial cancer12 (48.0)11 (45.8)2 (16.7)
Other cancer2 (8.0)3 (12.5)2 (16.7)
Other11 (44.0)10 (41.7)8 (66.7)

The potential impact of BMI on treatment-related adverse effects was also assessed for each treatment arm (Table 5). Evaluated treatment effects included fever and grade 1–4 hematologic, gastrointestinal (GI), GI obstructive, genitourinary, cutaneous, pulmonary, cardiovascular, and lymphatic effects. Although grade 3 or 4 toxicity is usually the concern in clinical application, all toxicities (grades 1 through 4) were evaluated because of a few severe adverse events identified in the protocol. No significant correlation between BMI and reported adverse effects was found in patients treated with surgery alone (NAT). In patients treated with RT, increased BMI was significantly associated with less GI (rank correlation coefficient [R], −0.21, P = .004) and more cutaneous (R, 0.18, P = .016) toxicity.

Table 5. Correlation of BMI With Reported Adverse Events by Treatment Regimen
Adverse effectSurgery alone, NRTSurgery + RT, RT
CorrelationPCorrelationP
  1. Rank correlation coefficient reported between BMI and toxicity grade.

Hematologic−0.007.921−0.133.069
Gastrointestinal−0.092.204−0.211.004
Genitourinary0.035.630−0.033.651
Cutaneous0.065.3670.176.016
Pulmonary−0.123.0870.035.639
Cardiovascular−0.036.621−0.019.792
Lymphatic−0.019.7970.006.934
Fever0.062.388−0.103.161

DISCUSSION

It is generally accepted that obesity is a risk factor for endometrial cancer. However, the relationship between obesity and clinical outcomes, such as recurrence and mortality, in endometrial cancer patients has not been adequately assessed. Nevertheless, in the current study, there was not a significant association between obesity and risk of endometrial cancer recurrence. It has been suspected that obesity is a risk factor for recurrence in cancer patients. Evidence from studies of breast cancer,13 colon cancer in women,19 and prostate cancer14 support this hypothesis.

Obesity is associated with decreased life expectancy10 in the general population and in cancer patients specifically. Obese breast, prostate, colon, and ovarian cancer patients have experienced increased death rates compared with patients of normal weight.15, 16, 20, 21 However, studies suggest that women with endometrial cancer have a greater potential to die from obesity-related diseases.3, 5, 18 In a retrospective study of 492 endometrial cancer patients, Anderson et al reported that 15% of patients died from their disease, and 7% died of other causes.3 In addition, BMI was a factor influencing survival (P = .06). Similarly, Everett et al reported that in patients with a BMI > 40, 2% died without evidence of disease, although nearly 5% died of unknown causes.7 Excess weight has been linked to cancer deaths.22 The relative risk of death for morbidly obese (BMI ≥ 40) women with endometrial cancer was 6.25 compared with normal-weight patients; this was the highest rate of death in men or women.12 Unlike obese breast cancer patients who have higher recurrence and mortality rates secondary to their cancer, our analysis demonstrates that, in endometrial cancer patients, an increased BMI is associated with shorter OS, which may be explained as death related to obesity and not cancer progression.

Although the focus of the current study was to explore associations between obesity and treatment outcomes, secondarily we also examined relations between other demographic and clinical factors and prognosis. These analyses were undertaken because they were not fully reported in the original article for GOG 99 and are of interest. Identified risk factors for endometrial cancer recurrence included age, myometrial invasion, tumor grade, LVSI, and regimen, confirming prior GOG research.5, 18

Racial variation in endometrial cancer mortality may be attributable in part to lifestyle factors including obesity. Without adjusting for BMI, the hazard ratio between African-American patients and non–African American-patients was 2.24; after adjusting for BMI, the hazard ratio was 1.61. Race as an independent risk factor for recurrence or survival may be confounded by BMI, suggesting that weight should be taken into account when evaluating race as a prognostic factor for endometrial cancer. Because the majority (75%) of patients on this study had an endometrioid cell type, we cannot delineate potential disparities among BMI, race, and unfavorable histologic types. Prior studies that have analyzed racial differences in incidence, treatment, and survival of endometrial cancer have not included BMI as a sociodemographic characteristic.23–25

Obese patients with breast cancer have higher incidences of RT complications.16 Our analysis found that increasing BMI was associated with less GI and more cutaneous adverse effects after RT treatment. Although the clinical significance of these findings is yet to be determined, the influence of obesity on RT tolerability may be limited, given that reported effects do not rise to the level of serious (grade 3) or life-threatening (grade 4). Others who have assessed RT morbidity in the treatment of women with endometrial cancer have not included obesity in their analyses.26, 27

There are several limitations to the current study. The fact that patient height was possibly self-reported might have introduced bias secondary to measuring techniques. More importantly, these findings cannot be generalized to obese endometrial cancer patients with advanced disease, although exploratory analyses in that population are in progress within the GOG.

The strength of this report can be attributed to the study design and data from GOG 99, a randomized phase III clinical trial. All patients met stringent eligibility criteria and received uniform treatment; central pathology review provided histologic confirmation of disease; study objectives, including analyses of toxicity, response, recurrence, and survival, were achieved through close monitoring and adherence to standard methods of assessment; and various internal checks provided quality assurance. These conditions minimized confounding and bias.

As previously reported, endometrial cancer survivors are not making positive lifestyle changes despite obesity being the greatest risk factor for development of the disease and potentially affecting their survival. At 6 months postsurgery, these patients are not losing weight, increasing their exercise, or changing their diets.8 Future recommendations include lifestyle intervention trials to improve survival in obese endometrial cancer patients.

Acknowledgements

We thank Caron Modeas for critical review of the current article.

Ancillary