Objective
We sought to quantify the relationship of uterine malignancy with body mass index (BMI).
Study Design
The University HealthSystem Consortium database was queried to identify all women undergoing total hysterectomy with a recorded BMI in the overweight and obese categories. Least squares regression was applied to evaluate the association between increasing BMI and the proportion of women with a diagnosis of uterine malignancy. Multivariate binary logistic regression was performed to adjust for other known risk factors including age, race, and other comorbidities.
Results
There were 6905 women who met inclusion criteria; 1891 (27.4%) of these had uterine malignancy. There is a linear relationship (y = 0.015x – 0.23, R 2 = 0.92) of the probability of uterine malignancy vs BMI. After adjusting for other risk factors, we found that each 1-U increase in BMI was significantly, independently associated with an 11% increase in the proportion of patients diagnosed with uterine malignancy (odds ratio, 1.11; 95% confidence interval, 1.09–1.13; P < .001).
Conclusion
In a population of women undergoing hysterectomy, we observed a linear increase in the frequency of uterine cancer associated with increasing BMI. This finding suggests that even relatively modest weight gain may significantly raise cancer risk. In the United States, the mean BMI for women is 26.5 kg/m 2 and it is estimated that more than half of US women have a BMI within the study’s range. Our results could, therefore, be relevant to a majority of the population. The findings could increase popular acceptance of weight management as a key component of general health maintenance and, possibly, as an additional approach to cancer risk reduction.
Over two-thirds of the US population is currently overweight or obese and the prevalence continues to increase. As the second leading cause of preventable death, obesity is one of the most serious public health problems facing the United States. There is evidence of a widening gap in life expectancy among women in the United States as compared to women in developed countries that has been attributed, in part, to the rise in obesity in the United States. The increasing prevalence of obese and overweight persons in the United States is driving health care spending, with associated costs approaching $150 billion per year.
In addition to being associated with multiple cardiovascular comorbidities and other medical problems, obesity has been linked to several cancer types. Approximately 20% of cancers can be attributed to being overweight or obese. If the current trend in obesity continues, there will be an additional 500,000 estimated cases of obesity-associated cancer in the United States by 2030. However, if every US adult could reduce his or her body mass index (BMI) by 1%, this would prevent the increased incidence of cancer by eliminating approximately 100,000 cases. The public health and economic implications of this finding are great, and more research is needed to develop interventions to battle the obesity epidemic.
Of all the obesity-related cancers in women, endometrial cancer is most strongly associated with increasing BMI, with 39% of cases resulting from obesity. Approximately 47,000 women were diagnosed with endometrial cancer in 2012, making it the most common malignancy of the female reproductive tract. Over 80% of endometrial cancers are type I, hormonally-mediated lesions with obesity being one of the most significant risk factors. Multiple studies have implicated obesity and adiposity in endometrial cancer. Results from the Million Women Study demonstrated an increase in relative risk for endometrial cancer of 2.89 for every 10 points of increasing BMI. Another study found an 18% increased risk of endometrial cancer for every 1 kg/m 2 over normal BMI in a population of Korean women. Wolin et al has calculated an increased relative risk of endometrial cancer of 1.59 per 5 mg/m 2 increase in BMI using metaanalysis data obtained by Renehan et al, which included multinational data. However, we did not find any studies that quantified uterine cancer risk by BMI specifically in US women.
Given that most women in the United States are overweight and obese, it is important to quantify the risk of endometrial cancer according to BMI in this population. This study uses data from a large administrative database to quantify the increased risk of uterine cancer as BMI increases in a cohort of US women undergoing hysterectomy. The women in our cohort have BMIs 25-40 kg/m 2 , which is especially pertinent to the US population as the majority of women have BMIs within this range.
Materials and Methods
This is a retrospective cohort study utilizing the University HealthSystem Consortium (UHC) database to evaluate the relationship of BMI and endometrial cancer in women having undergone hysterectomy. The UHC maintains an administrative database with information contributed by 116 academic medical centers and 276 affiliate hospitals, representing >90% of US nonprofit academic medical centers. Clinical research is encouraged by the UHC and multiple observational studies have been published looking at a variety of clinical subjects.
To avoid biasing the sample with multiple admissions for the same patient, we restricted our analysis to all women undergoing hysterectomy (all International Classification of Diseases, Ninth Revision [ ICD-9 ] 68x.x) in all available fiscal years (2008 through 2012) and hospitals. We then selected patients with a recorded BMI in the overweight and obese categories (BMI 25-39.9 kg/m 2 ) that were coded as discrete values ( ICD-9 v6521-39); discrete values were not available for BMI <25 or >39.9 kg/m 2 and these women were excluded. Age at admission (in years) and race (combined into white, black, or other) were also evaluated. A single dichotomous variable (yes or no) was created to account for the presence of any of the following comorbidities: acquired immune deficiency syndrome, alcohol abuse, chronic blood loss anemia, chronic pulmonary disease, coagulopathy, congestive heart failure, deficiency anemias, depression, diabetes, drug abuse, fluid and electrolyte disorders, hypertension, hypothyroidism, liver disease, lymphoma, other neurological disorders, paralysis, peptic ulcer disease, peripheral vascular disease, psychoses, pulmonary circulation disease, renal failure, rheumatoid arthritis/collagen vascular disease, valvular disease and weight loss. This comorbidity variable was dichotomized to avoid double-counting subjects due to the format of the database. The outcome of interest for all analyses was the presence of uterine malignancy (all ICD-9 182.x and 179).
Least squares regression was applied to evaluate the association between increasing BMI and the proportion of women with a diagnosis of uterine cancer. To evaluate if there was inconsistency between the slopes, this analysis was also performed separately for women in the overweight category (BMI 25.0-29.9 kg/m 2 ) and for the obese category (BMI 30.0-39.9 kg/m 2 ).
Demographics and population characteristics were evaluated. Univariate analyses including Student t test and χ 2 were performed for each covariate by presence of uterine malignancy.
Multivariate binary logistic regression was then performed to adjust for other known risk factors for uterine malignancy. Age, race, and presence of any other comorbidities were included in the initial model, in addition to BMI. Backwards stepwise regression with a cutoff of P > .2 was performed, and comorbidities were removed from the model as being nonsignificant and having minimal effect on the main predictor variables. The final model included BMI, age, and race as predictors of uterine cancer.
Statistics were performed with software (PAWS, version 18; IBM Corp, Armonk, NY) as was graphing (Excel; Microsoft, Redmond, WA).
Results
A total of 165,876 women underwent hysterectomy during the study period. Of those, 6905 (4.2%) had discretely recorded BMI; 1891 (27.4%) of these had uterine cancer. Among all women with hysterectomy, 7.0% (1891) of those with uterine malignancy had their BMI recorded, and 3.8% (5014) of those without uterine malignancy had their BMI recorded. Characteristics of those women with BMI available are shown in Table 1 . Among women with BMIs 25-29.9 kg/m 2 , 17.5% had uterine malignancy, and among women with BMIs 30-39.9 kg/m 2 , 29.7% had uterine malignancy.
| Characteristic | With uterine malignancy | Without uterine malignancy | Total |
|---|---|---|---|
| No. | 1891 | 5014 | 6905 |
| Mean age, y (±SD) a | 62.3 (11.0) | 48.9 (11.9) | 52.6 (13.1) |
| Mean BMI, kg/m 2 (±SD) a | 34.6 (3.3) | 33.7 (3.7) | 34.0 (3.7) |
| Race, a n (%) | |||
| White | 1366 (72.2) | 2620 (52.3) | 3986 (57.7) |
| Black | 293 (15.5) | 1654 (33.0) | 1947 (28.2) |
| Other | 232 (12.3) | 740 (14.8) | 972 (14.1) |
| Comorbidities, a n (%) | |||
| Yes | 1606 (84.9) | 3809 (76.0) | 5418 (78.4) |
| No | 285 (15.1) | 1205 (24.0) | 1490 (21.6) |
In the overweight and obese cohorts separately, and when combined into a single study population, least squares fit of the probability of uterine malignancy vs BMI demonstrates a linear relationship ( Figure ). For the entire population, the line is described by the equation: y = 0.015x – 0.23, R 2 = 0.92, where risk of uterine cancer is the ordinate and BMI is the abscissa. Evaluating overweight and obesity separately, the line for the overweight cohort alone is described by y = 0.026x – 0.53 and the line for the obese cohort is described by y = 0.013x – 0.16.
On univariate analyses, women with uterine cancer were more likely to be older (62.3 vs 48.9 years, P < .001) and have a slightly higher BMI (34.6 v 33.7 kg/m 2 , P < .001). Significant differences were also seen in presence of comorbidity and distribution of race ( P < .001). Those with uterine cancer were more likely to be white and to have at least 1 comorbid condition.
In the multivariate model, the comorbidities variable was not significantly associated with uterine malignancy ( P = .32) and was removed. The final model is shown in Table 2 . After adjusting for age and race, we found that each 1-U increase in BMI was independently associated with an 11% increase in the proportion of patients diagnosed with uterine malignancy (odds ratio, 1.11; 95% confidence interval, 1.09–1.13; P < .001) among those with a BMI 25-40 kg/m 2 . Increasing age (9% increased risk per year) and white race were also independently associated with uterine cancer. The model indicates that 32% of the probability of being diagnosed with uterine malignancy is accounted for by the variables included, the remaining 68% is attributable to factors not tested in this study.
| Predictor | ß | SE ß | P value | Expected (ß) a | 95% CI for expected (ß) |
|---|---|---|---|---|---|
| Constant | –9.37 | 0.37 | < .001 | 0.00 | |
| BMI, kg/m 2 | 0.103 | 0.01 | < .001 | 1.11 | 1.09–1.13 |
| Age | 0.09 | 0.00 | < .001 | 1.09 | 1.09–1.10 |
| Race b | < .001 | ||||
| Black | –0.65 | 0.08 | < .001 | 0.52 | 0.45–0.61 |
| Other | –0.05 | 0.09 | .60 | 0.95 | 0.79–1.14 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
