TABLE 2.
Outcomes and methodology of studies included in the meta-analysis1
| Study (reference), country | Outcomes of interest | Statistical methodology | Adjustment variables2 | Duration or time points of follow-up |
| ALL (n = 7) | ||||
| Aldhafiri (10), United Kingdom | CIR of obese and overweight vs. healthy vs. underweight | Frequencies and P values reported3 | None | Censored at 7.2 y from study closure |
| Baillargeon (11), United States | EFS and OS of obese (≥95%) vs. nonobese (<95%) | Cox proportional hazards multivariate regression model of predictors and survival time; analyses further stratified by older vs. younger age (2–9 vs. 10–18 y) | Age (for overall cohort), sex, WBC, and ethnicity | 1- and 5-y survival analyses |
| Butturini (7), United States | EFS, OS, and CIR of obese (≥95%) vs. nonobese (<95%) | Cox proportional hazards multivariate regression model of predictors and survival/relapse time; subset analyses were performed in an older age strata (≥10 y) | Age, sex, race-ethnicity, WBC, and early bone marrow response | Median 7.8 y (range 0.1–13.3 y) |
| Ethier (12), Canada | EFS and OS of obese (≥95%) vs. nonobese (<95%) | Cox proportional hazards multivariate regression model of predictors and survival time; analyses additionally stratified into low-, standard-, or high-risk groups based on combination of prognostic predictors; secondary subset analyses performed by leukemia phenotype | Age, WBC, and cytogenetic risk group | 5-y survival analyses |
| Hijiya (9), United States | EFS, OS, CIR, and TRT comparing obese (≥95%), overweight (85–94%), normal (5–84%), and underweight (<5%) | Frequencies and P values reported3 | None | Median 10.5 y (range 2.4–16.9 y) |
| Orgel (13), United States | EFS and TRT of obese (≥95%) vs. normal/overweight (5–94%) vs. underweight (<5%) | Cox proportional hazards multivariate regression model of predictors and survival time, stratified by treatment regimen; logistic regression model of predictors and weight status at start of treatment phase | Sex, and CNS disease after stepwise selection of predictors (nonsignificant: age, race-ethnicity, WBC, phenotype, and disease response) | Median 8.5 y |
| Orgel (14), United States | EFS comparing lean (<85%), overweight (≥85%), and obese (≥95%) | Cox proportional hazards multivariate regression model of predictors and survival time, stratified by treatment regimen | NCI risk group and end-induction minimal residual disease after stepwise selection of predictors (nonsignificant: age, sex, ethnicity, trisomy 21, WBC, and cytogenetic risk group) | Median 1.9 y (maximum 5.4 y) |
| Niinimaki (15), Finland | TRT (osteonecrosis) comparing obese (≥95%), overweight (85–94%), normal (5–84%), and underweight (<5%) at diagnosis | Logistic regression model of BMI as a predictor | Age, sex, and dexamethasone | Post-therapy (NR) |
| AML (n = 3) | ||||
| Canner (16), United States | EFS and OS of overweight or obese (≥85%) vs. normal (5–84%) vs. underweight (<5%) | Cox proportional hazards multivariate regression model of predictors and survival time; analyses further stratified by age (<16 y vs. 16–20 y) | Race-ethnicity, WBC, and cytogenetic risk group | 5-y survival analyses |
| Inaba (17), United States | EFS and OS of overweight or obese (≥85%) vs. normal (5–84%) vs. underweight (<5%) | Cox proportional hazards multivariate regression model of predictors and survival time, stratified by treatment regimen | Age, WBC, and FAB | 5-y survival analyses |
| Lange (18), United States | EFS and OS comparing obese (≥95%), middle weight (10–94%), and underweight (<10%) | Univariate analysis of BMI on outcomes only | None | Median 4.2 y |
1
ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CIR, cumulative incidence of relapse; CNS, central nervous system; EFS, event-free survival; FAB, French–American–British classification of acute myeloid leukemia by morphology; NCI, National Cancer Institute; NR, not reported; OS, overall survival; TRT, treatment-related toxicity; WBC, white blood cell count.
2
With the use of ALL- or AML-specific cytogenetic risk groups from historical/validated favorable or poor prognostic mutations (fusions, deletions, or chromosome number).
3
Statistical methodology refers to estimates for BMI exposure; however, HRs and 95% CIs were not reported.