The Association of Abdominal Adiposity With Mortality in Patients With Stage I–III Colorectal Cancer

Justin C. Brown; Bette J. Caan; Carla M. Prado; Elizabeth M. Cespedes Feliciano; Jingjie Xiao; Candyce H. Kroenke; Jeffrey A. Meyerhardt

Disclosures

J Natl Cancer Inst. 2020;112(4):377-383. 

In This Article

Results

Characteristics of the Study Cohort

The average (SD) age of the 3262 participants was 62.6 (11.4) years, and 49.9% were women (Table 1). Baseline CT images were obtained a median of 6 days (interquartile range = 0–12 days) after biopsy-proven diagnosis of colorectal cancer. During a median follow-up of 6.9 years (interquartile range = 5.3–8.4 years), 879 (26.9% of the cohort) deaths occurred, with 451 (51.3% of all deaths) attributable to colorectal cancer (Supplementary Table 1, available online).

The median BMI was 27.2 kg/m2 (range = 14.0–59.6 kg/m2), muscle area was 135.9 cm2 (range = 60.3–319.6 cm2), visceral adipose tissue was 138.6 cm2 (range = 0.1–676.7 cm2), and subcutaneous adipose tissue was 184.9 cm2 (range = 0.0–931.1 cm2). Visceral and subcutaneous adipose tissue were moderately and positively correlated (r = 0.37, 95% CI = 0.34 to 0.40, P < .001). Demographic, clinical, and behavioral characteristics were associated with visceral and subcutaneous adipose tissue area (Supplementary Table 2, available online).

Prognostic Effects of Adipose Tissue Distribution on Mortality

Visceral adipose tissue was prognostic of all-cause mortality in a reverse L-shaped pattern (nonlinear P = .02); risk was flat to a threshold (~260 cm2), then increased linearly (Figure 1; Supplementary Table 3, available online). Subcutaneous adipose tissue was prognostic of all-cause mortality in a J-shaped pattern (nonlinear P < .001); risk was higher at the extreme (<50 cm2) but lower at intermediate values (>50 to ≤560 cm2; Supplementary Table 4, available online). Subcutaneous adipose tissue (nonlinear P = .02) but not visceral adipose tissue (linear P = .24, nonlinear P = .32) was prognostic of colorectal cancer-specific mortality (Supplementary Figure 2, available online). Sensitivity analysis that adjusted for physical activity did not substantively alter the shape or magnitude of the above-described effect estimates (Supplementary Figure 3, available online). The exclusion of 61 (1.9%) patients with underweight BMI (<18.5 kg/m2) did not substantively alter the shape or magnitude of the above-described effect estimates.

Figure 1.

Risk of all-cause mortality on the relative hazard scale in 3262 patients with colorectal cancer. Shaded regions indicate 95% confidence bands for risk of mortality as a function of visceral (red) and subcutaneous (blue) adipose tissue area. Estimates are adjusted for age, sex, race and ethnicity, cancer site, cancer stage, chemotherapy, radiation therapy, smoking history, Charlson comorbidity index, height, muscle area, subcutaneous adipose tissue area (for visceral adipose tissue area models), and visceral adipose tissue area (for subcutaneous adipose tissue area models).

Supplementary Figure 2.

Risk of colorectal cancer-specific mortality on the relative hazard scale in 3,262 patients with colorectal cancer. Shaded regions indicate 95% confidence bands for risk of mortality as a function of visceral (red) and subcutaneous (blue) adipose tissue area. Estimates are adjusted for age, sex, race and ethnicity, cancer site, cancer stage, chemotherapy, radiation therapy, smoking history, Charlson comorbidity index, height, muscle area, subcutaneous adipose tissue area (for visceral adipose tissue area models), and visceral adipose tissue area (for subcutaneous adipose tissue area models).

Supplementary Figure 3.

Risk of all-cause mortality on the relative hazard scale, additionally adjusted for physical activity, as a sensitivity analysis. Shaded regions indicate 95% confidence bands for risk of mortality as a function of visceral (red) and subcutaneous (blue) adipose tissue area. Estimates are adjusted for age, sex, race and ethnicity, cancer site, cancer stage, chemotherapy, radiation therapy, smoking history, Charlson comorbidity index, height, muscle area, subcutaneous adipose tissue area (for visceral adipose tissue area models), and visceral adipose tissue area (for subcutaneous adipose tissue area models).

Patient Sex Modified the Prognostic Effects of Adipose Tissue Distribution on Mortality

Analyses stratified by patient sex demonstrated that men had more visceral adipose tissue (+91.3 cm2, 95% CI = 84.4 to 98.2 cm2, P < .001) and less subcutaneous adipose tissue (–50.4 cm2, 95% CI = –58.5 to –42.3 cm2, P < .001) than women. The correlation between visceral and subcutaneous adipose tissue was moderate among men (r = 0.47, 95% CI = 0.43 to 0.50, P < .001) and women (r = 0.61, 95% CI = 0.57 to 0.63, P < .001; Figure 2).

Figure 2.

Distribution and correlation between visceral adipose tissue area and subcutaneous adipose tissue area among 3262 men (navy blue) and women (magenta) with colorectal cancer.

Patient sex modified the prognostic association of visceral adipose tissue and all-cause mortality (P interaction = .049). Among men, visceral adiposity was associated with mortality in a J-shaped pattern (nonlinear P = .003), whereas among women, visceral adiposity was associated with mortality in a linear pattern (linear P = .008; Figure 3). Patient sex modified the prognostic association of subcutaneous adipose tissue and all-cause mortality (P interaction = .04). Among men, subcutaneous adiposity was associated with mortality in an L-shaped pattern (nonlinear P = .01), whereas among women, subcutaneous adiposity was associated with mortality in a J-shaped pattern (nonlinear P < .001). Patient sex did not modify the prognostic association between visceral adipose tissue (P interaction = .44) or subcutaneous adipose tissue (P interaction = .39) with colorectal cancer-specific mortality (Supplementary Figure 4, available online). Sensitivity analysis that adjusted for physical activity did not substantively alter the shape or magnitude of the above-described effect estimates (Supplementary Figure 5, available online). The exclusion of 61 (1.9%) patients with underweight BMI (<18.5 kg/m2) at colorectal cancer diagnosis did not substantively alter the shape or magnitude of the above-described effect estimates.

Figure 3.

Risk of all-cause mortality by sex and adipose tissue compartment on the relative hazard scale in 3262 patients with colorectal cancer. Shaded regions indicate 95% confidence bands for risk of mortality as a function of visceral (top) and subcutaneous (bottom) adipose tissue area among men (navy blue) and women (magenta). Estimates are adjusted for age, race and ethnicity, cancer site, cancer stage, chemotherapy, radiation therapy, smoking history, Charlson comorbidity index, height, muscle area, subcutaneous adipose tissue area (for visceral adipose tissue area models), and visceral adipose tissue area (for subcutaneous adipose tissue area models).

Supplementary Figure 4.

Risk of colorectal cancer-specific mortality by sex and adipose tissue compartment on the relative hazard scale in 3,262 patients with colorectal cancer. Shaded regions indicate 95% confidence bands for risk of mortality as a function of visceral (top panel) and subcutaneous (bottom panel) adipose tissue area, among men (navy blue) and women (magenta). Estimates are adjusted for age, race and ethnicity, cancer site, cancer stage, chemotherapy, radiation therapy, smoking history, Charlson comorbidity index, height, muscle area, subcutaneous adipose tissue area (for visceral adipose tissue area models), and visceral adipose tissue area (for subcutaneous adipose tissue area models).

Supplementary Figure 5.

Risk of all-cause mortality by sex and adipose tissue compartment on the relative hazard scale, additionally adjusted for physical activity, as a sensitivity analysis. Shaded regions indicate 95% confidence bands for risk of mortality as a function of visceral (top panel) and subcutaneous (bottom panel) adipose tissue area, among men (navy blue) and women (magenta). Estimates are adjusted for age, race and ethnicity, cancer site, cancer stage, chemotherapy, radiation therapy, smoking history, Charlson comorbidity index, height, muscle area, subcutaneous adipose tissue area (for visceral adipose tissue area models), and visceral adipose tissue area (for subcutaneous adipose tissue area models).

Age, cancer site, cancer stage, smoking history, and muscle area did not modify the prognostic associations of visceral and subcutaneous adipose tissue with mortality (results not shown).

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....