Abstract
Objective:
This study aims to investigate whether race influences the relationship between obesity and cancer-related deaths, with a particular focus on understanding these dynamics within diverse adult populations.
Methods:
Data from the Pennington Center Longitudinal Study, encompassing 18,296 adults (35.0% male, 34.3% Black), were analyzed. Body mass index (BMI) was calculated using the standard formula (kg/m2). Cancer death served as the primary endpoint of the study.
Results:
Over a 14.3-year follow-up period, 346 cancer deaths were recorded. A significant interaction between race and BMI in men was observed (Pinteraction=0.045). Specifically, when compared to a BMI of 22 kg/m2, white men with a BMI of 35 kg/m2 showed a hazard ratio of 1.74 (95% CI: 1.38–2.21) for cancer death. In stark contrast, black men with the same BMI of 35 kg/m2 presented a hazard ratio of 0.64 (95% CI: 0.45–0.90), indicating a reduced risk. Among women, race did not significantly modify the BMI-cancer death association (Pinteraction=0.43). However, similar to men, white women with a BMI of 35 kg/m2 had a hazard ratio of 1.42 (95% CI: 1.18–1.70), while black women showed a non-significant hazard ratio of 0.99 (95% CI: 0.82–1.20) when compared to a BMI of 22 kg/m2.
Conclusions:
This study, conducted on a diverse adult cohort, reveals that obesity is associated with an increased risk of cancer death for white men and women. Interestingly, obesity appears to be linked to a decreased risk of cancer death in black men and has no significant impact on cancer death risk for black women. These findings highlight critical racial differences in the obesity-cancer mortality relationship, underscoring the need for targeted health interventions and adult cohort learning strategies, especially for black men.
Keywords: Adult Cohort Learning Black Men, Body Mass Index, Cohort Study, Disparities, Mortality, Waist Circumference
INTRODUCTION
Significant health disparities persist in the United States, particularly affecting black men and women. Individuals born in 1980 face considerably shorter life expectancies compared to their white counterparts (1). Obesity prevalence is also notably higher among black adults (2). Understanding how these factors intersect to influence health outcomes, especially cancer mortality, is crucial. While obesity is a known risk factor for increased all-cause mortality among white adults (4, 5, its impact on black adults is less clear and may differ by sex (8–11. This nuanced relationship necessitates further investigation, particularly concerning cancer, a leading cause of death (12).
Obesity is linked to increased mortality from various cancers in both men and women (13, 14). However, it remains uncertain whether this risk is consistent across racial groups. With projections indicating that over half of black adults in the U.S. will be obese by 2030 (15), understanding the specific association between obesity and cancer death within this population is paramount. Comparing this association with that in white individuals is essential for developing targeted, evidence-based interventions and policies. Such efforts are vital to improve health outcomes and address racial disparities in cancer mortality (16, 17. This knowledge gap is particularly relevant for adult cohort learning initiatives aimed at health education for black men, where tailored information is needed to address specific risk profiles and promote effective lifestyle modifications.
This study was designed to test the hypothesis that race is a significant factor modifying the relationship between anthropometric measures of obesity and cancer death within a diverse adult cohort. The findings are intended to inform more effective public health strategies and adult learning programs focused on black men’s health.
METHODS
Study Population and Design
The Pennington Center Longitudinal Study (PCLS), an ongoing prospective cohort study, provided the data for this research. Established in 1992, PCLS aims to explore the long-term effects of obesity, physical activity, and diet on chronic diseases and mortality. Participants include men and women aged 18 years and older who participated in clinical trials at Pennington Biomedical Research Center in Baton Rouge, Louisiana. Baseline data collected before any study-specific interventions were used for this analysis. All participants provided informed consent, and the Pennington Biomedical Research Center institutional review board approved all PCLS procedures and this analysis. The study cohort is registered at clinicaltrials.gov as NCT00959270 (registered in 2009). Statistical analyses were conducted between January and May 2021.
Anthropometric Measures of Obesity
Standardized procedures were used to obtain anthropometric measurements. Height was measured twice using a wall-mounted stadiometer, ensuring the Frankfort Horizontal Plane alignment (18). A third measurement was taken if the initial two differed by more than 0.5 cm. Body weight was measured twice using a calibrated scale after removing outerwear and shoes; a third measurement was obtained if the first two varied by more than 0.5 kg. BMI was calculated as weight (kg) divided by height squared (m2). Waist circumference (WC) was measured twice at the midpoint between the lower ribcage and the upper iliac crest using a flexible, spring-loaded measuring tape (19); a third measurement was taken if the difference exceeded 1.0 cm. Staff blinded to study endpoint status compiled the anthropometric measures database.
While BMI and WC are correlated at the population level [r = 0.93; (20)] they are not interchangeable at the individual level (21). To examine their combined effects, we used residuals from WC regressed on BMI (22). These residuals represent the difference between observed and BMI-predicted WC, indicating individuals with larger WC than expected for their BMI.
Covariates
Baseline covariate data, including age, self-reported race, and smoking status (never, former, current), were collected at enrollment. The median enrollment year was 2003 [IQR: 1998, 2007], and the median sample size of the clinical trials was n=25 [IQR: 9, 60].
Study Outcomes
Cancer death was the primary endpoint, defined as time from enrollment to cancer-related death. Obesity-related cancer death was the secondary endpoint, focusing on cancers linked to obesity by the International Agency for Research on Cancer (IARC) (23). These include esophageal adenocarcinoma, gastric cardia, colon and rectum, liver, gallbladder, pancreas, postmenopausal breast, uterine, ovarian, renal cell kidney, meningioma, thyroid, and multiple myeloma. Death data were obtained from the National Death Index using Social Security Administration records through December 31, 2017.
Statistical Analysis
The analysis included participants who: 1) reported no prior cancer (excluding nonmelanoma skin cancer); 2) were under 90 years at enrollment; and 3) had at least one year of follow-up. Multivariable Cox proportional hazards models were used to assess the hazard ratios for cancer death associated with baseline BMI, WC, and WC residuals. Cause-specific hazard ratios were calculated to quantify relative changes in cancer death rates among individuals not deceased from other causes (24). These findings are consistent with sub-distribution hazard models using Fine and Gray competing risk models (not shown).
BMI and WC were modeled as continuous variables using B-splines (26) to accommodate non-linearity and provide efficient and visual representations of prognostic associations (27). Models were stratified by sex due to known sex differences in obesity and cancer death risk (28, 29). Adjustments included age, smoking status, enrollment year, and study size to account for temporal trends and small pilot study effects. Model parsimony was assessed using the Akaike Information Criterion.
Graphical depictions of statistical associations are presented to facilitate interpretation. Point estimates and 95% confidence intervals are reported for BMI at 35 kg/m2 (class II obesity) relative to 22 kg/m2 (normal weight), and WC at 110 cm relative to 80 cm, corresponding to approximate BMI values (30). Linearity and non-linearity were evaluated visually and statistically. Proportional hazards assumptions were verified using log-log plots and Schoenfeld residuals (31).
Effect modification by race was tested using interaction terms in regression models and likelihood ratio statistics. Baseline characteristics were compared using χ2 tests for categorical variables and t-tests for continuous variables. All tests were two-sided and analyzed using SAS (version 9.4; Cary, NC) and R (version 4.0; R Core Team).
RESULTS
Characteristics of the Study Cohort
The analytic sample comprised 18,296 participants (Appendix Figure 1), with 6,405 (35.0%) men and 6,273 (34.3%) black individuals (Table 1). The median follow-up was 14.3 years (IQR: 10.1–19.2 years), during which 346 cancer deaths and 161 obesity-related cancer deaths occurred (Appendix Table 1). Age-adjusted hazard ratios for cancer death in black men and women compared to their white counterparts were 1.45 (95% CI: 0.99–2.13) and 1.16 (95% CI: 0.85–1.58), respectively.
Table 1.
Baseline characteristics of the study population
| Characteristic | Overall Cohort | Men | Women |
|---|---|---|---|
| White | Black | White | |
| N | 18,296 | 4,877 | 1,528 |
| Age (y) | 42.2 (14.8) | 41.1 (16.1) | 38.2 (13.7) |
| Body Mass Index (kg/m2) | 31.3 (7.8) | 29.8 (6.7) | 30.2 (7.3) |
| Body Mass Index, n (%) | |||
| Underweight (<18.5 kg/m2) | 106 (0.6) | 19 (0.4) | 12 (0.8) |
| Normal Weight (18.5–24.9 kg/m2) | 3809 (20.8) | 1,156 (23.7) | 365 (23.9) |
| Overweight (25.0–29.9 kg/m2) | 5017 (27.4) | 1,606 (32.9) | 448 (29.3) |
| Obese (≥30.0 kg/m2) | 9364 (51.2) | 2,096 (43.0) | 703 (46.0) |
| Waist Circumference* (cm) | 95.6 (18.5) | 98.9 (17.4) | 95.8 (17.4) |
| Waist Circumference*, n (%) | |||
| Low (<88 cm women; <102 cm men) | 6977 (49.0) | 2,354 (61.1) | 817 (67.6) |
| High (≥ 88 cm women; ≥102 men) | 7266 (51.0) | 1,497 (38.9) | 392 (32.4) |
| Smoking History, n (%) | |||
| Never | 11291 (61.7) | 2,935 (60.2) | 956 (62.6) |
| Former | 2807 (15.3) | 1,044 (21.4) | 265 (17.3) |
| Current | 1149 (6.3) | 199 (4.1) | 146 (9.6) |
| Unknown | 3049 (16.7) | 699 (14.3) | 161 (10.5) |
*n=14,243.
Anthropometric Measures of Obesity
The mean BMI was 31.3 kg/m2 (SD 7.8) and mean WC was 95.6 cm (SD 18.5) across all participants. Detailed descriptions of racial and sex-based differences in these measures have been previously published (30).
Effect of Body Mass Index and Race on Cancer Death
Race significantly modified the association between BMI and cancer death in men (Pinteraction = 0.045). Compared to a BMI of 22 kg/m2, white men with a BMI of 35 kg/m2 had a hazard ratio of 1.74 (95% CI: 1.38–2.21), while black men at the same BMI had a hazard ratio of 0.64 [(95% CI: 0.45–0.90); Figure 1A]. In women, race did not significantly modify this association (Pinteraction = 0.43). However, white women with a BMI of 35 kg/m2 showed a hazard ratio of 1.42 (95% CI: 1.18–1.70), whereas black women had a hazard ratio of 0.99 [(95% CI: 0.82–1.20); Figure 1B].
Figure 1.
Figure 1.
Risk of cancer death by anthropometric measure of obesity on the relative hazard scale. Shaded regions indicate 95% confidence bands for the risk of cancer death as a function of body mass index among men (Panel A) and body mass index among women (Panel B); waist circumference among men (Panel C) and waist circumference among women (Panel D); and waist circumference residuals among men (Panel E) and waist circumference residuals among women (Panel F). White participants are depicted in blue color and black participants are depicted in red color. Estimates are multivariable adjusted.
Effect of Waist Circumference and Race on Cancer Death
In men, race did not significantly modify the WC-cancer death association (Pinteraction = 0.41). Compared to a WC of 80 cm, white men with a WC of 110 cm had a hazard ratio of 1.52 (95% CI: 1.12–2.03), while black men had a hazard ratio of 0.82 [(95% CI: 0.54–1.25); Figure 1C]. Similarly, in women, race did not significantly modify this association (Pinteraction = 0.66). White women with a WC of 110 cm had a hazard ratio of 1.58 (95% CI: 1.22–2.05), and black women had a hazard ratio of 1.19 [(95% CI: 0.88–1.59); Figure 1D] compared to a WC of 80 cm.
Effect of Waist Circumference Residuals and Race on Cancer Risk
Race did not significantly modify the association between WC residuals and cancer death in men (Pinteraction = 0.41). Compared to a WC residual of 0 cm, white men with a residual of 10 cm had a hazard ratio of 1.09 (95% CI: 0.75–1.59), and black men had a hazard ratio of 1.80 [(95% CI: 0.96–3.35); Figure 1E]. In women, race also did not significantly modify this association (Pinteraction = 0.55). White women with a WC residual of 10 cm had a hazard ratio of 1.11 (95% CI: 0.86–1.44), while black women had a hazard ratio of 1.40 [(95% CI: 1.00–1.98); Figure 1F].
Secondary and Sensitivity Analyses
Analyses using obesity-related cancer death as the endpoint yielded similar conclusions to the primary endpoint of all-cancer death (Appendix Figure 2).
DISCUSSION
This prospective cohort study of 18,296 adults revealed significant racial and sex-based differences in the associations between obesity and cancer death. Obesity was linked to an increased risk of cancer death in white men and women. Conversely, obesity was associated with a reduced risk of cancer death in black men and had no significant effect on cancer death risk in black women. These findings challenge the expectation that higher BMI uniformly increases cancer death risk across all populations. The observed decreased risk in black men and null effect in black women compared to the increased risk in white individuals highlights the complexity of the obesity-cancer relationship across different racial groups.
Our previous findings from the PCLS cohort also indicated that BMI-defined obesity increased all-cause mortality risk in white men and women but not in black men and women (Pinteraction < 0.05); (32). Furthermore, obesity was found to increase cancer risk among white men and white and black women but not among black men (Pinteraction < 0.05); (30). The current analysis expands on these findings by specifically focusing on cancer death and reinforcing the differential impact of obesity across racial and sex subgroups. These insights are crucial for tailoring public health messages and adult learning initiatives, particularly for black men, to ensure interventions are culturally relevant and effectively address their specific health risks.
Studies like the Multiethnic Cohort study have reported no racial differences in the obesity-breast cancer death association in women (33). Research on racial differences in obesity and post-cancer diagnosis survival also offers relevant context (34, 35. While our study examines risk estimates from a different perspective (pre-diagnosis), it contributes valuable, complementary data to a field with limited research (36. For instance, while obesity is associated with increased prostate cancer-specific mortality, no racial differences were found in this association (34. In contrast, for breast cancer, obesity increases mortality risk in white women but not in black women (35. These inconsistent findings underscore the need for more research to fully understand the nuanced relationship between obesity and cancer across diverse populations (37. Despite these complexities, maintaining a healthy body weight remains a critical health goal for all population subgroups (38. Adult learning programs for black men should emphasize this overarching message while also addressing the specific nuances highlighted by this research.
The mechanisms underlying obesity’s influence on cancer progression and death are not fully understood. Visceral adiposity and insulin resistance are frequently hypothesized as key factors (39–41). However, BMI and WC, while common obesity measures (38, may not consistently reflect adiposity and metabolic abnormalities across racial groups (42–44. Black individuals at a given BMI or WC often have less visceral adiposity and lower insulin resistance compared to white individuals (45–48. We propose that anthropometric surrogates like BMI and WC may mask true associations with health outcomes across races. Direct adiposity measures, such as visceral adipose tissue assessed by dual-energy X-ray absorptiometry, might reveal more consistent prognostic associations across population subgroups (49, 50. This is an important consideration for future research and for refining health risk assessments within adult cohort learning settings for black men.
Limitations of this study include the inherent susceptibility of observational studies to residual confounding. Obesity measures were taken at a single baseline time point, and the impact of BMI or WC changes over time, such as weight loss, on cancer death is unknown. Race is a social construct, but it remains a useful variable for describing health patterns in the U.S. due to data reporting practices (2). Despite a large sample size, subgroup analyses for specific cancer sites were limited by the infrequency of these cancers. Sample size and cancer death event rates also limited statistical power to detect race-based effect modification for all associations, though point estimates and graphical trends support the hypothesis of effect modification by race. Data on family cancer history, diet, type 2 diabetes/prediabetes, or socioeconomic status were not systematically collected. Cancer stage and post-diagnosis treatment information were also unavailable. Participants were volunteers in lifestyle and metabolic studies, potentially limiting generalizability to the broader population. Unmeasured health behavior differences between racial subgroups might also contribute to the findings. Adult learning interventions should consider these limitations and focus on modifiable risk factors while acknowledging the complexities of race and health.
Strengths include the diverse study population from a high cancer death rate region (51), mirroring the racial demographics of the Baton Rouge area (52). This diversity enabled informative sex- and race-stratified analyses. Objective, standardized anthropometric measurements were used. The use of both BMI and WC allowed for contrasting overall and abdominal obesity measures. The cohort’s broad age and obesity range enhances generalizability. Longitudinal follow-up (≥10 years for ≥75% of participants) strengthens the study’s findings. These strengths enhance the relevance and applicability of the findings for informing adult cohort learning strategies aimed at improving health outcomes for black men.
CONCLUSION
In a large, diverse adult cohort, race and sex significantly modify the prognostic associations of anthropometric obesity measures and cancer death. These findings underscore the importance of considering race and sex in obesity-related cancer risk assessments and interventions. Specifically, the reduced cancer death risk associated with obesity in black men, contrasting with the increased risk in white men, necessitates further investigation and tailored approaches to health education and intervention within adult cohort learning settings for black men.
Supplementary Material
supinfo
NIHMS1736204-supplement-supinfo.docx (11MB, docx)
Study Importance.
What is already known?
- Cancer is a leading cause of death in the United States.
- Obesity increases the risk of death from several cancers in both men and women. However, it’s unclear if this risk is uniform across white and black individuals.
What does this study add?
- This study of 18,296 adults demonstrates that race and sex modify the relationship between obesity and cancer death.
- Obesity increases cancer death risk in white men but decreases it in black men.
- Obesity increases cancer death risk in white women but has no significant effect in black women.
How might these results change the focus of clinical practice?
- If validated by further research, these findings could reshape interventions and policies designed to reduce cancer death disparities, particularly by highlighting the need for race-specific and sex-specific approaches to obesity and cancer risk management. Furthermore, these results emphasize the critical need for tailored adult learning and health education initiatives for black men, ensuring that information and interventions are culturally competent and address their unique risk profiles.
Funding:
This work was supported by the National Institute of General Medicine Sciences of the National Institutes of Health under Award Number U54-GM104940, the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health under Award Number P30-DK072476, and the National Cancer Institute of the National Institutes of Health under Award Numbers R00-CA218603; and R25-CA203650.
Disclosure:
Dr. Brown reports receiving grants paid to his institution from the National Cancer Institute, American Institute for Cancer Research, and the Susan G. Komen Foundation during the study. All other authors disclosed no relevant relationships.
REFERENCES
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
supinfo
NIHMS1736204-supplement-supinfo.docx (11MB, docx)
