Abstract
Multiple myeloma (MM) incidence and mortality are higher among African Americans (AAs) than among other population groups. The prevalence of obesity is also elevated among AAs, but few studies have examined risk of this cancer in relation to body size among AAs. We combined data from seven prospective cohorts tracking mortality among 239 597 AA adults and used Cox proportional hazards regression to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for death because of MM according to body mass index (BMI) at cohort entry, adjusted for age (as time-scale) and sex. Relative to those with normal BMIs (18.5-25 kg/m 2 ), mortality increased monotonically as BMI increased, with hazard ratios reaching 1.43 (95% CI = 1.03 to 1.97) for BMIs of 35 kg/m 2 or greater. The findings suggest that obesity is a risk factor for MM and a contributor to the elevated rates and rising incidence trends of MM among AAs in the United States.
Multiple myeloma (MM), a cancer of the plasma cells with a median survival of approximately four years and limited treatment options, accounts for approximately 15% of hematologic malignancies ( 1 , 2 ) and approximately 1.4% of all cancers ( 3 ). Although nearly all cases begin as monoclonal gammopathy of undetermined significance (MGUS), little is currently known about the etiology of MM ( 2 , 4 , 5 ). Few risk factors have been definitively identified aside from age, race, sex, and family history of MM ( 2 ). While the highest MM incidence and mortality rates worldwide are observed in African Americans (AAs), with risks double those of whites, few studies have examined risk factors specifically for AAs ( 2 , 4 , 5 ). Limited data exist on MM occurrence in Africa, but several national registries indicate that incidence rates are not high compared with either black or white Americans ( 6 ).
Obesity has been identified as a possible modifiable risk factor for MM ( 2 , 4 ). While individual study results have not been entirely consistent, meta-analyses to date have found a modest positive association (10%-20% increase) for a 5 kg/m 2 increase in body mass index (BMI) among predominantly white populations ( 7–10 ). Herein, we present data from the AA BMI-Mortality Pooling Project, examining BMI as a risk factor for MM death among AAs.
Detailed characteristics of the AA BMI-Mortality Pooling Project have been presented previously ( 11 , 12 ). All participants gave written informed consent. Seven cohorts with at least 10 000 AA participants were eligible for inclusion: NIH-AARP Diet and Health Study (AARP); Adventist Health Study 2 (AHS2); Black Women’s Health Study (BWHS); Cancer Prevention Study II (CPSII); Multiethnic Cohort Study (MEC); Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial Study (PLCO); and Southern Community Cohort Study (SCCS). In total, baseline and vital status data were available for 256 369 participants recruited over a 28-year span from as early as 1982 (CPSII) through 2009 (SCCS). We assessed deaths from MM beginning one year after entry into the respective cohorts or age 30 years, whichever occurred later. Follow-up continued until death, loss to follow-up, or most recent mortality follow-up. Multiple myeloma mortality was defined according to the International Classification of Disease (ICD) as an underlying cause of death of C90 (ICD-10) or 203 (ICD-9).
Hazard ratios (HRs) and 95% confidence intervals (CIs) for MM mortality by BMI categories overall and by sex were calculated using the Cox proportional hazards model (SAS Institute, Cary, NC). Age was the underlying time metric, and all models were stratified by cohort. BMI was calculated from self-reported height and weight at cohort entry and categorized as 15 to less than 18.5, 18.5 to less than 25 (referent), 25 to less than 30, 30 to less than 35, and 35 to less than 60. Because of limited information on risk factors for MM, the relationship between BMI and MM was adjusted only for sex although sensitivity analyses individually and jointly adjusted the relationship between BMI and MM mortality for education, marital status, alcohol consumption, physical activity, and intensity of cigarette smoking. The proportional hazards assumption was tested using time-varying covariates ( 13 ). Two-sided Ptrend values were calculated by including the midpoint of each BMI category as a continuous covariate for BMIs of 18.5 kg/m 2 or greater. The METAN command in STATA (Stata Corp, College Station, TX) was used to calculate a forest plot, I 2 index, and Cochran’s Q test for a continuous five-unit increase in BMIs of 18.5 kg/m 2 or greater to examine heterogeneity across cohorts. Only five MM deaths were recorded among AHS2 participants, and thus the cohort was excluded from modeling with little change in results. A two-sided P value of less than .05 was considered statistically significant. Each cohort study was approved by the relevant institutional review board.
After exclusions for extreme (<15 or ≥ 60 kg/m 2 , n = 564) or missing BMI (n = 8899), insufficient follow-up (<1 year or ended before age 30 years, n = 7272), or missing age (n = 37) or sex (n = 7), 239 597 AA participants with 2 416 815 person-years of follow-up were available for analysis. The average age at cohort entry was 52 years (range = 20–104 years), and the average duration of follow-up was 11.6 years (range = 1–26.5 years). Among all participants, 36% were overweight (BMI = 25–<30 kg/m 2 ) and 33% obese (BMI ≥ 30 kg/m 2 ). Overall, 496 MM deaths were observed.
Table 1 presents hazard ratios and 95% confidence intervals for MM death by BMI overall and by sex. Relative to those with BMIs of 18.5 to less than 25 kg/m 2 , hazard ratios were 1.07 (95% CI = 0.86 to 1.33) for BMIs of 25 to less than 30 kg/m 2 , 1.14 (95% CI = 0.87 to 1.49) for BMIs of 30 to less than 35 kg/m 2 , and 1.43 (95% CI = 1.03 to 1.97) for BMIs of 35 to less than 60 kg/m 2 , a statistically significant ( P = .04) rising trend across the range of 18.5 to less than 60 kg/m 2 . The relationship was somewhat more pronounced among men than women ( Pinteraction = .39).
Body mass index in relation to multiple myeloma mortality, stratified by sex *
| BMI, kg/m 2 | No. deaths | Person-years | HR (95% CI) |
|---|---|---|---|
| All participants † | |||
| 15 – <18.5 | 5 | 28 323.1 | 1.18 (0.48 to 2.88) |
| 18.5 – <25 | 134 | 773 265.8 | Referent |
| 25 – <30 | 204 | 904 058.3 | 1.07 (0.86 to 1.33) |
| 30 – <35 | 94 | 438 114.7 | 1.14 (0.87 to 1.49) |
| 35 – <60 | 54 | 273 053.3 | 1.43 (1.03 to 1.97) |
| Ptrend‡ | .04 | ||
| Men§ | |||
| 15 – <18.5 | 0 | 4496.8 | Undefined |
| 18.5 – <25 | 54 | 183 632.3 | Referent |
| 25 – <30 | 95 | 285 936.7 | 1.10 (0.78 to 1.53) |
| 30 – <35 | 33 | 99 262.2 | 1.22 (0.79 to 1.89) |
| 35 – <60 | 13 | 31 390.9 | 1.80 (0.97 to 3.31) |
| Ptrend‡ | .09 | ||
| Women§ | |||
| 15 – <18.5 | 5 | 23 826.2 | 1.65 (0.67 to 4.06) |
| 18.5 – <25 | 80 | 589 633.5 | Referent |
| 25 – <30 | 109 | 618 121.6 | 1.05 (0.79 to 1.40) |
| 30 – <35 | 61 | 338 852.5 | 1.10 (0.79 to 1.55) |
| 35 – <60 | 41 | 241 662.4 | 1.32 (0.90 to 1.94 |
| Ptrend‡ | .17 |
| BMI, kg/m 2 | No. deaths | Person-years | HR (95% CI) |
|---|---|---|---|
| All participants † | |||
| 15 – <18.5 | 5 | 28 323.1 | 1.18 (0.48 to 2.88) |
| 18.5 – <25 | 134 | 773 265.8 | Referent |
| 25 – <30 | 204 | 904 058.3 | 1.07 (0.86 to 1.33) |
| 30 – <35 | 94 | 438 114.7 | 1.14 (0.87 to 1.49) |
| 35 – <60 | 54 | 273 053.3 | 1.43 (1.03 to 1.97) |
| Ptrend‡ | .04 | ||
| Men§ | |||
| 15 – <18.5 | 0 | 4496.8 | Undefined |
| 18.5 – <25 | 54 | 183 632.3 | Referent |
| 25 – <30 | 95 | 285 936.7 | 1.10 (0.78 to 1.53) |
| 30 – <35 | 33 | 99 262.2 | 1.22 (0.79 to 1.89) |
| 35 – <60 | 13 | 31 390.9 | 1.80 (0.97 to 3.31) |
| Ptrend‡ | .09 | ||
| Women§ | |||
| 15 – <18.5 | 5 | 23 826.2 | 1.65 (0.67 to 4.06) |
| 18.5 – <25 | 80 | 589 633.5 | Referent |
| 25 – <30 | 109 | 618 121.6 | 1.05 (0.79 to 1.40) |
| 30 – <35 | 61 | 338 852.5 | 1.10 (0.79 to 1.55) |
| 35 – <60 | 41 | 241 662.4 | 1.32 (0.90 to 1.94 |
| Ptrend‡ | .17 |
*Models exclude Adventist Health Study 2 participants (see text). BMI = body mass index; CI = confidence interval; HR = hazard ratio.
†Adjusted for age (as time-scale), sex, and stratified by cohort.
‡ Ptrend values are two-sided and calculated for BMIs of 18.5 kg/m 2 or greater.
§Adjusted for age (as time-scale) and stratified by cohort.
Body mass index in relation to multiple myeloma mortality, stratified by sex *
| BMI, kg/m 2 | No. deaths | Person-years | HR (95% CI) |
|---|---|---|---|
| All participants † | |||
| 15 – <18.5 | 5 | 28 323.1 | 1.18 (0.48 to 2.88) |
| 18.5 – <25 | 134 | 773 265.8 | Referent |
| 25 – <30 | 204 | 904 058.3 | 1.07 (0.86 to 1.33) |
| 30 – <35 | 94 | 438 114.7 | 1.14 (0.87 to 1.49) |
| 35 – <60 | 54 | 273 053.3 | 1.43 (1.03 to 1.97) |
| Ptrend‡ | .04 | ||
| Men§ | |||
| 15 – <18.5 | 0 | 4496.8 | Undefined |
| 18.5 – <25 | 54 | 183 632.3 | Referent |
| 25 – <30 | 95 | 285 936.7 | 1.10 (0.78 to 1.53) |
| 30 – <35 | 33 | 99 262.2 | 1.22 (0.79 to 1.89) |
| 35 – <60 | 13 | 31 390.9 | 1.80 (0.97 to 3.31) |
| Ptrend‡ | .09 | ||
| Women§ | |||
| 15 – <18.5 | 5 | 23 826.2 | 1.65 (0.67 to 4.06) |
| 18.5 – <25 | 80 | 589 633.5 | Referent |
| 25 – <30 | 109 | 618 121.6 | 1.05 (0.79 to 1.40) |
| 30 – <35 | 61 | 338 852.5 | 1.10 (0.79 to 1.55) |
| 35 – <60 | 41 | 241 662.4 | 1.32 (0.90 to 1.94 |
| Ptrend‡ | .17 |
| BMI, kg/m 2 | No. deaths | Person-years | HR (95% CI) |
|---|---|---|---|
| All participants † | |||
| 15 – <18.5 | 5 | 28 323.1 | 1.18 (0.48 to 2.88) |
| 18.5 – <25 | 134 | 773 265.8 | Referent |
| 25 – <30 | 204 | 904 058.3 | 1.07 (0.86 to 1.33) |
| 30 – <35 | 94 | 438 114.7 | 1.14 (0.87 to 1.49) |
| 35 – <60 | 54 | 273 053.3 | 1.43 (1.03 to 1.97) |
| Ptrend‡ | .04 | ||
| Men§ | |||
| 15 – <18.5 | 0 | 4496.8 | Undefined |
| 18.5 – <25 | 54 | 183 632.3 | Referent |
| 25 – <30 | 95 | 285 936.7 | 1.10 (0.78 to 1.53) |
| 30 – <35 | 33 | 99 262.2 | 1.22 (0.79 to 1.89) |
| 35 – <60 | 13 | 31 390.9 | 1.80 (0.97 to 3.31) |
| Ptrend‡ | .09 | ||
| Women§ | |||
| 15 – <18.5 | 5 | 23 826.2 | 1.65 (0.67 to 4.06) |
| 18.5 – <25 | 80 | 589 633.5 | Referent |
| 25 – <30 | 109 | 618 121.6 | 1.05 (0.79 to 1.40) |
| 30 – <35 | 61 | 338 852.5 | 1.10 (0.79 to 1.55) |
| 35 – <60 | 41 | 241 662.4 | 1.32 (0.90 to 1.94 |
| Ptrend‡ | .17 |
*Models exclude Adventist Health Study 2 participants (see text). BMI = body mass index; CI = confidence interval; HR = hazard ratio.
†Adjusted for age (as time-scale), sex, and stratified by cohort.
‡ Ptrend values are two-sided and calculated for BMIs of 18.5 kg/m 2 or greater.
§Adjusted for age (as time-scale) and stratified by cohort.
Figure 1 presents a forest plot for a continuous five-unit increase in BMI by cohort for participants with BMIs over 18.5 kg/m 2 . The overall I 2 (magnitude of heterogeneity) was moderate ( 13 ), at 49% ( P = .08). Cohort-specific estimates ranged from a statistically nonsignificant reduction in risk in the BWHS (HR = 0.94, 95% CI = 0.70 to 1.24) to a statistically significant increase in the AARP study (HR = 1.32, 95% CI = 1.07 to 1.63). Hazard ratios for four of the six cohorts were above 1.0, with confidence intervals excluding 1.0 in two of them.
Forest plot examining heterogeneity across cohorts in continuous hazard ratios (HRs) and 95% confidence intervals (CIs) for multiple myeloma mortality per five-unit increase in body mass index (BMI) among participants with BMIs of 18.5 kg/m 2 or greater. Models were adjusted for age (as time-scale) and sex. AARP = NIH-AARP Diet and Health Study; BWHS = Black Women’s Health Study; CPSII = Cancer Prevention Study II; MEC = Multiethnic Cohort Study; PLCO = Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial; SCCS = Southern Community Cohort Study.
Sensitivity analyses demonstrated that the relationship between BMI and MM mortality was not sensitive to adjustment for education, marital status, alcohol consumption, physical activity, and cigarette smoking intensity, or to use of alternative reference categories. The proportional hazards assumption was not violated , with no statistically significant variation in the hazard ratios between duration of follow up categories 1-4 and five or more years.
The findings from this pooled analysis from ongoing cohorts in the United States following large numbers of AAs demonstrate that high BMI is associated with increased MM mortality. Meta-analyses of BMI in relation to MM mortality or incidence among whites show a 10% to 20% higher risk among the obese vs normal weight ( 7–10 ), with similar results from cohorts in the United States and Europe ( 15 , 16 ). Our data suggest that the patterns and magnitudes of BMI-associated risk are similar among AAs and that strategies aimed at obesity prevention and reduction may have benefit with respect to MM mortality.
Nationally, MM incidence is increasing, with few cancers showing a sharper incline since 2002 ( 3 ). Furthermore, incidence and mortality rates are twice as high among AAs as whites among both men and women ( 3 ). The prevalence of obesity, which has also been rising, is estimated to be 48% among AAs and 33% among whites in the United States ( 17 ), so that obesity seems likely to contribute to the rising trends and excess risk and mortality of MM among AAs.
Limitations of the current study include the absence of information on MGUS status on the participants being followed, a reliance on mortality rather than incidence rates, and use of self-reported weight and height from a single time point, at entry into the cohorts. Despite these limitations, we were able to demonstrate a clear association between BMI and MM mortality among AAs, the population with the highest incidence and mortality from this disease. Elevated mortality could be because of increased incidence and/or poorer survival. We could not distinguish effects associated with incidence vs survival, but surveys of MM patients among US veterans ( 18 ) and in Asia ( 19 ) have noted that survival was higher among those with higher BMI at cancer diagnosis. An improved survival among those with higher BMI suggests that the adverse mortality effects we found with increasing BMI are predominantly related to adverse associations of obesity upon onset rather than survival of MM. While causal inferences cannot be made from observational studies like ours, evidence is becoming convincing that obesity is associated with MM among major population groups. The findings thus continue to suggest that additional research into potential biologic mechanisms, as well as into means of prevention, of BMI-associated MM is warranted.
Funding
This work was supported by the National Cancer Institute’s Intramural Research Program in the Division of Cancer Epidemiology and Genetics and the Epidemiology and Genomics Research Program in the Division of Cancer Control and Populations Sciences. The NIH-AARP study is supported by the Intramural Research Program of the National Cancer Institute, National Institutes of Health. The Adventist Health Study 2 is funded by grant R01 CA94594 from the National Cancer Institute. The Black Women's Health Study is funded by grants R01 CA058420 and UM1 CA164974 from the National Cancer Institute. The Cancer Prevention Study II Nutrition Cohort is supported by the American Cancer Society. The Multiethnic Cohort Study (MEC) is funded by grant R37 CA54281 from the National Cancer Institute. The PLCO is supported by contracts from the Division of Cancer Prevention, National Cancer Institute, National Institutes of Health. The Southern Community Cohort Study (SCCS) is funded by grant R01 CA092447 from the National Cancer Institute including special allocations from the American Recovery and Reinvestment Act (3R01 CA092447-08S1).
Notes
The funders had no role in design of the study; the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication.
The authors declare no conflicts of interest.
References
