General Obesity and Prostate Cancer in Relation to Abdominal Obesity and Ethnic Groups: A US Population-Based Cross-Sectional Study

Introduction

Prostate cancer remains a significant public health concern in the United States, representing the number one most common cancer diagnosed in men and the second leading cause of cancer-related mortality among men in 2022.¹ An estimated 268,490 new cases of prostate cancer were diagnosed and total prostate cancer-related deaths totaled 34,500 in the United States in 2022.^1,2 The introduction of prostate-specific antigen (PSA) testing in the early 1990s has led to increased detection of early-stage prostate cancer, which in some cases may not require surgical intervention or treatment.³ Although the PSA alone is not sufficient to detect aggressive prostate cancer, it remains the most important marker for screening of prostate cancer.⁴ Due to the potential harms of PSA tests such as overdiagnosis, overtreatment and anxiety, the population-based screening is currently not recommended and the decision to screen for prostate cancer relies mainly on two factors: age (55–69 years) and men’s choice.⁵ The identification of additional risk factors and associations to further differentiate which patients may benefit from screening is imperative, given the current US Preventive Services Task Force guidelines. For instance, race and ethnicity^6–8 and body fat indicators^9,10 are important factors to personalize prostate cancer risk and improve health outcomes.

The association between obesity (measured by body mass index (BMI) and prostate cancer in the literature remains widely inconsistent, with existing studies and meta-analyses presenting positive,^11,12 negative,^13–16 or null associations.¹⁷ Recently, obesity has garnered attention for its potential role in prostate cancer aggressiveness, as it is associated with high grade,¹⁸ higher rate of cancer recurrence,¹⁹ and increased mortality.^13,20,21

The relationship between obesity and prostate cancer risk might differ across ethnic groups. There is evidence that BMI is inversely associated with prostate cancer risk in non-Hispanic White men but positively associated in non-Hispanic Black men.²² This complex relationship in various ethnic groups can be further explored by measures of interrelation between body fat measurements such as obesity and abdominal obesity as defined by waist circumference (WC ≥ 102 cm). The association between abdominal obesity and prostate cancer also remains controversial, with some studies suggesting an elevated risk^15,23,24 and with others finding no apparent association.^25,26

Conflicting evidence also exists on whether the association between BMI and risk of prostate cancer is related to abdominal obesity. A Korean study reported a strong association between BMI and prostate cancer in patients with high WC (WC ≥ 95 cm).²⁷ Abdominal obesity alone was not associated with prostate cancer, while increased risk was observed when the analysis was adjusted for BMI.²⁸ However, in a cohort of 46,094 men from the Cancer Prevention Study II Nutrition Cohort, WC was inversely associated with the rate of low-grade prostate cancer, but this association was not significant after adjustment for BMI.²⁵ The associations of general and abdominal obesity with prostate cancer have not been investigated in the National Health and Nutrition Examination Survey (NHANES). Our analysis used NHANES data collected over nine cycles (2001–2020).

The interaction effect between general and abdominal obesity on prostate cancer has not been widely studied, particularly in the ethnic populations of non-Hispanic Black and White men. We hypothesized that the association between general obesity and prostate cancer might be modified by abdominal obesity and that this association might vary by the ethnic groups. We utilized the extensive data available from a large population-based cross-sectional study to provide a more comprehensive understanding of whether there is an interaction effect between general and abdominal obesity on the prevalence of prostate cancer.

Materials and Methods

Study Design

This study utilized data from the NHANES cross-sectional population-based surveys. A series of independent surveys have been conducted since 1999 by the National Center for Health Statistics of the Centers for Disease Control and Prevention (CDC). The CDC uses a stratified, multi-stage probability sampling design to routinely collect data from a non-institutionalized civilian population.²⁹

Study Population

We included data of adult men, who were 18 years and older at the time of screening, from the NHANES 2001–2020-March. We combined eight two-year samples (2001 to 2016) with a prepandemic 3.2-year sample (2017 to 2020-March) to increase the precision of our estimates. We applied survey weights to calculate nationally representative estimates. The total participants during this period was 97,657. We excluded 49,483 women and 20,297 respondents aged 17 and below. We also excluded 1,671 respondents who reported a diagnosis of cancer other than prostate cancer. We restricted our final analytic file to two non-Hispanic populations totaling 16,703 (10,683 White and 6,020 Black). Other ethnic groups (Pacific Islanders and Native Americans) were not included in the analysis due to the small sample sizes. Also, there is no known genetic link to prostate cancer in these minorities, whereas 1) Black men have a higher burden of obesity,³⁰ 2) Black men were diagnosed at younger ages,³¹ and 3) Black men were diagnosed with higher grades and stages.^6–8 This study received ethical approval from the Uniformed Services University Institutional Review Board (Protocol # DBS.2024.766).

Data Collection

In this study, we merged public files on demographics (DEMO), ages at 1st cancer diagnosis using the questionnaire for medical conditions (MCQ230), and body fat indicator measures (BMX). Age at screening (18 years and ≥ 75 years) and high school or less (yes/no) were included to control for their potential confounding effects. Non-Hispanic race/ethnicity (White and Black) was used as a stratifying covariate. The primary outcome was the first prostate cancer diagnosis and that was identified based on the response to two questions: “Have you ever been told by a doctor or other health professional that you had cancer or a malignancy of any kind?” and first cancer type “What kind of cancer was it?” Participants were categorized into prostate cancer status (yes/no). As part of the NHANES survey, the body fatness data were measured in Mobile Examination Centers by trained health technicians. Abdominal obesity was defined as a WC of 102 cm or greater.³² General obesity was defined as a BMI of 30 kg/m² or greater.³³ We also examined several mutually exclusive classes of obesity indicators, including four categories: 1) general and abdominal obesity, 2) abdominal obesity without general obesity (or abdominal obesity only), 3) general obesity without abdominal obesity (or general obesity only), and 4) none of the two (the reference group).

Statistical Analysis

The data analysis was conducted using SAS 9.4 SURVEYFREQ and SURVEYLOGISTIC (SAS Institute Inc., Cary, NC) which accounted for the NHANES 2001–2020-March survey weights in all estimations. However, because we combined several cycles, we adjusted survey weights to generate representative US population estimates by covering a 19.2-year period: 16-years (2001–2002 to 2015–2016) and 3.2-years (2017–2020-March pre-pandemic data).³⁴ We calculated the overall weighted prevalence estimates (%) and 95% confidence interval (CI) for prostate cancer status, general obesity, and abdominal obesity. Weighted prevalence estimates were compared using the Rao-Scott chi-square test. Bivariate analysis was performed to estimate the weighted prevalence of prostate cancer by obesity indicators and potential confounders (age and education). We assessed unadjusted relationships between potential confounders, obesity indicators, and prostate cancer. The moderator effect was tested with an interaction term between abdominal obesity (yes/no) and general obesity (yes/no) in predicting prostate cancer, using three weighted logistic regression models. In this moderator analysis, Model 1 controlled for no demographic characteristics, Model 2 controlled for age, and Model 3 controlled for age and education level. Odds ratio (OR) and 95% CI of first prostate cancer diagnosis were reported. We computed the area under the receiver operating characteristic curve (AUC-ROC) to evaluate the utility of obesity indicators and their combination for predicting prostate cancer. We stratified all the analyses by race/ethnicity, non-Hispanic Black and White.

Data Availability

The raw data generated in this study are publicly available: the National Health and Nutrition Evaluation Survey (NHANES) from 2001 to 2020-March.

Results

The analytic sample consisted of 16,703 men with weighted mean age at the time of examination of 45.2 years old (SE = 0.24) years and the weighted mean age at 1st prostate cancer diagnosis was 66.0 (SE = 0.47) years. Missingness rates in reporting BMI and WC were 5.53% and 7.93%, respectively, which were less than 10%. Table 1 showed the sample characteristics and the prevalence of prostate cancer by the sample characteristics. The overall prevalence estimates of general and abdominal obesity were 32.33% (95% CI: 31.21–33.45%) and 41.64% (95% CI: 40.43–42.85%), respectively. In our sample, a large percentage (29.32%) of men were generally and abdominally obese, 12.14% abdominally obese only, and 1.86% generally obese only.

Table 1 Sample Characteristics and Weighted Prevalence of Prostate Cancer (NHANES 2001–2020)

The overall prevalence estimate of prostate cancer was 2.25% (95% CI: 2.00–2.50%). No significant difference was identified in the overall prevalence of prostate cancer between obese (2.14%) and non-obese (2.25%), P = 0.678. Men with abdominal obesity without general obesity have significantly higher prevalence of prostate cancer (4.28%) compared to men with both general and abdominal obesity (2.23%), men with neither abdominal obesity nor general obesity (1.70%), and men with general obesity without abdominal obesity (0.18%), P < 0.0001. This association remains significant after stratification by race/ethnicity, where Black men with abdominal obesity without general obesity (7.21%) have the highest prevalence of prostate cancer, followed by White men with abdominal obesity without general obesity (4.09%) (Figure 1).

Figure 1 The effect of obesity indicators on the prevalence of prostate cancer by ethnicity (NHANES 2001–2020). Notes: Unadjusted weighted prevalence estimates (in percentage %) of prostate cancer by ethnicity and obesity indicators. Error bars represent ± SE.

The unadjusted relationships between the obesity indicators, the potential confounders, and prostate cancer by race/ethnicity are illustrated in Table 2. Men age 75 or older and men with abdominal obesity had higher odds of prostate cancer regardless of their race/ethnicity. Without adjustment for abdominal obesity, there was no association between general obesity and the prevalence of prostate cancer, neither in the overall sample nor by race/ethnicity. While abdominal obesity without general obesity had a significant association with the prevalence of prostate cancer, regardless of race/ethnicity, obesity without abdominal obesity was a protective factor against prostate cancer in White men but not Black men. Conversely, the presence of both general and abdominal obesity was a significant predictor of prostate cancer in Black men but not White men.

Table 2 Unadjusted Relationship Between Demographic, Body Fat Indicators, and Prostate Cancer by Ethnicity (NHANES 2001–2020)

The role of abdominal obesity as a moderator in the relationship between general obesity and prostate cancer by race/ethnicity was also assessed (Table 3). The associations remain consistent after adjusting for potential confounders. For instance, after adjustment for age and education, general and abdominal obesity category was associated with a significant increase in the odds of prostate cancer in Black men, but not in White men. In both race/ethnicity Black and White men, abdominal obesity was only strongly associated with significant increase in the odds of prostate cancer, whereas general obesity only was associated with significant decrease n the odds of prostate cancer in White men, but not Black men. Almost none of the generally obese White had prostate cancer.

Table 3 Moderator Effect of Abdominal Obesity on the Relationship Between General Obesity and Prostate Cancer by Ethnicity (NHANES 2001–2020)

The combined effect of BMI and WC improved the predictive accuracy for prostate cancer in both Black (Figure 2A) and White men (Figure 2B). The ROC curves in Black men were slightly higher than in White men (BMI: AUC = 0.54 vs 0.50, WC: AUC = 0.63 vs 0.58, and their combination: AUC = 0.74 vs 0.72, respectively).

Figure 2 (A) ROC curves of BMI, WC, and their combination in Non-Hispanic Black men. (B) ROC curves of BMI, WC, and their combination in Non-Hispanic White men.

Discussion

We sought to understand the interrelationship between BMI and WC and their contribution to prostate cancer. We analyzed data collected over the last two decades to evaluate obesity indicators (general and abdominal obesity) and their combined effects on the prevalence of prostate cancer in different race/ethnicity groups. This is the first study to assess the combined effect of BMI and WC in NHANES participants. Although obesity and abdominal obesity are lifestyle factors and preventable, they remain risk factors for increased aggressiveness of prostate cancer and mortality.^{8,15,20,24,35} Our findings suggest that WC might modify the association between BMI and prostate cancer. There is a need to evaluate data on obesity and prostate cancer by distinguishing separate effects from the combined effect of BMI and WC.^27,28

The current published data on obesity indicators and prostate cancer are not consistent. Some studies have found that obesity increases prostate cancer risk,^{11,12,15,23,24} while others have shown a decreased risk,^13–16 or no such association.^17,25,26 The present study shows that the overall prevalence estimates of prostate cancer were quite similar across general obesity categories (2.14% yes vs 2.25% no), but significantly vary across abdominal obesity categories (2.84% yes vs 1.65% no). When both obesity measures were combined, the prevalence estimates in men with neither general obesity nor abdominal obesity, obesity only, abdominal obesity only, and both were 1.70%, 0.18%, 4.28%, and 2.23%, respectively. This might explain the discrepancy between previous epidemiologic studies, as the majority of currently published data evaluated general obesity as a separate effect rather than the combined effect of BMI and WC.

Despite its use of various cut-point values for BMI and WC, a large study utilized data from Cancer Prevention Study II Nutrition Cohort revealing that WC as a categorical and continuous covariate was not associated with low-grade (Gleason score < 8) or high-grade (Gleason score ≥ 8) prostate cancer after adjustment for BMI.²⁵ In our study, we found that abdominal obesity only (WC ≥ 102 cm and BMI < 30 kg/m²) was associated with prostate cancer. However, their approach did not investigate the combined effect of BMI and WC on prostate cancer grades, which might explain the discrepancy between their findings and the present study.

Though many studies have investigated the contribution of BMI and WC as separate effects, few studies have documented the combined effect of BMI and WC on prostate cancer. Our regression analysis findings were consistent with an age-matched (±5 years) Canadian study that evaluated the relationship between obesity indicators and prostate cancer.²⁸ They found BMI to be associated with a decreased prostate cancer incidence, whereas abdominal obesity increased prostate cancer incidence when the analysis adjusted for BMI. A Korean study reported that the association between obesity and prostate cancer might depend on abdominal obesity.²⁷ Although this study used lower cut-point values for abdominal obesity (eg, WC ≥ 90 cm) and obesity (eg, BMI ≥ 25 kg/m²), our study remains consistent with their findings, and revealed a significant association between abdominal obesity without general obesity and the presence of both general and abdominal obesity with prostate cancer. Strategies to identify prostate cancer and its progression should include fat distribution as adipose tissues can cause long-term inflammation and insulin resistance, which could be potential mechanisms for initiation or progression of prostate cancer.³⁶

The link between the combined effect of BMI and WC and prostate cancer is not well described in race/ethnicity populations. We sought to determine race/ethnicity differences in the association between obesity indicators and prostate cancer. Our findings were consistent with Beebe-Dimmer et al as their study shows obesity provided a protective effect against prostate cancer among White men (OR 0.51, 95% CI 0.33–0.80), but no effect against prostate cancer in Black men (OR 1.15, 95% CI 0.70–1.89).³⁷ Another study by Beebe-Dimmer et al shows abdominal obesity was strongly associated with prostate cancer in Black men.³⁸ Our results suggest that abdominal obesity only (based on WC ≥ 102 and BMI < 30) might be problematic for both Black and White men, whereas general and abdominal obesity (based on BMI ≥ 30 and WC ≥ 102) might be problematic for Black men. The racial/ethnic differences in combined obesity indicators and prostate cancer might be explained by the tumor biology due to the underlying genetic factors.^39,40 The combined effect of BMI and WC can be assessed in future studies as modifiable factors to improve screening strategies and prevent prostate cancer–related mortality. Managing and mentoring WC and BMI might decrease the racial disparity in prostate cancer incidence.

Limitations

The results of this study must be viewed in the context of certain limitations. Despite the large sample, causal inferences could not be implied due to the nature of the study design. Self-reported prostate cancer history of participants might underestimate the true prevalence of prostate cancer in the US population. Gleason score and prostate cancer grades were not reported in NHANES. BMI and WC cut-point values might differ across ethnicities and/or countries (eg, Korea, WC ≥ 90). The NHANES did not incorporate additional indicators of fat distribution, such as waist-to-hip ratio and collective visceral and subcutaneous adipose tissues, when estimating abdominal obesity. The body fat measures were obtained at the time of screening which after the prostate cancer diagnosis and might not reflect measurements at the time of diagnosis. Future studies on combined obesity indicators and prostate cancer and poor outcomes must consider temporality and account for the race-specific and other fat distribution indicators. Race-specific prospective studies can be conducted on prostate cancer risk categories or Gleason grade to determine if patients with “abdominal obesity only” have more aggressive prostate cancer than patients with general obesity; however, this may not be technically feasible given the large number of patients required for adequate power.

Conclusion

The association between general obesity and prevalence of prostate cancer depends on abdominal obesity and ethnic groups. This study illustrates that WC is a potential classifier of prostate cancer as it shows variation in the prevalence of prostate cancer by general obesity within the same and different racial and ethnic groups. The results of the present investigation demonstrate the importance of incorporating obesity indicators to address prostate cancer racial disparity in screening and poor outcomes. These differences in the combined obesity effect of BMI and WC and the prevalence of prostate cancer might also be explained by underlying genetic factors.