A review of 61 cohort studies involving 46,572 cancer patients receiving immune checkpoint inhibitors found that higher BMI measured before treatment was associated with a lower average hazard of death and a lower average hazard of progression or death. The review made adjusted estimates its primary evidence. For overall survival, the pooled hazard ratio comparing higher with lower BMI was 0.74, with a 95% confidence interval from 0.67 to 0.82. For progression-free survival, it was 0.78, with a 95% confidence interval from 0.69 to 0.89. A hazard ratio below 1 indicates a lower average hazard in the higher-BMI group. These findings describe an association, not proof that changing BMI improves outcomes.
The headline result came with a wide warning about variation. The overall-survival analysis combined 42 cohort estimates from 38 publications and had I-squared of 70.6%, a measure of how much study results differed. Its 95% prediction interval ran from 0.48 to 1.13. For progression-free survival, 34 adjusted estimates from 31 publications had I-squared of 60.3%, and the prediction interval ran from 0.47 to 1.32. Both prediction intervals included the point representing no association, so the favorable average finding was not guaranteed to appear in every setting.
Obesity showed the clearest category-level association.
To make the comparisons easier to interpret, the review also used standard BMI categories. BMI is weight in kilograms divided by height in meters squared. Under the WHO reference categories used in the analysis, underweight was below 18.5, normal weight was 18.5 to below 25, overweight was 25 to below 30, and obesity was 30 or higher, all in kilograms per square metre. Compared with normal weight, obesity was associated with a lower mortality hazard, with a hazard ratio of 0.75 and a 95% confidence interval of 0.66 to 0.85. The corresponding progression-or-death hazard ratio was 0.74, with a 95% confidence interval of 0.59 to 0.93.
The pattern was not a simple step-by-step advantage for every higher category. Overweight alone was not significantly associated with overall survival or progression-free survival, and underweight was not significantly associated with overall survival. For progression-free survival, the underweight estimate was 1.85, but its 95% confidence interval stretched from 0.47 to 7.35, making it imprecise. When overweight and obesity were grouped together and compared with normal weight and underweight, the higher-BMI group had hazard ratios of 0.65 for overall survival and 0.73 for progression-free survival.
Looking at BMI as a continuous measure produced a similarly mixed picture. Each increase of 5 kilograms per square metre was associated with lower progression-free survival hazard, with a hazard ratio of 0.87 and a 95% confidence interval of 0.81 to 0.95. The corresponding overall-survival hazard ratio was 0.91, with a 95% confidence interval of 0.81 to 1.01, and was not statistically significant. The prediction intervals included no association for both outcomes, ranging from 0.67 to 1.23 for overall survival and from 0.72 to 1.05 for progression-free survival.
Changing the BMI threshold used to define the comparison did not significantly alter the pooled associations. For every 5-kilogram-per-square-metre increase in the cutoff, the ratio of hazard ratios was 1.10 for overall survival, with a 95% confidence interval of 0.80 to 1.51, and 1.00 for progression-free survival, with a 95% confidence interval of 0.63 to 1.57. Both moderator tests had P values above 0.05.
Response and side effects remained uncertain.
The review found less settled evidence for tumor response and immune-related side effects. The adjusted objective response estimate was an odds ratio of 1.50, with a 95% confidence interval of 0.94 to 2.41, and was not statistically significant. For any-grade immune-related adverse events, the adjusted odds ratio was 1.00, with a 95% confidence interval of 0.63 to 1.57. Only one adjusted estimate was available for grade 3 or higher events: an odds ratio of 4.08 with a 95% confidence interval of 0.95 to 17.56. That evidence was too sparse for a meta-analysis.
In the anti-PD-1 and anti-PD-L1 subgroup, higher BMI was associated with lower hazards for both overall survival and progression-free survival, with hazard ratios of 0.71 and 0.73. But the review detected no subgroup differences by cancer type, region, or ICI class for either survival outcome; all interaction tests had P values above 0.05. A nominal cancer-type difference in adjusted objective response was exploratory, because several subgroups contributed only one or two estimates.
The finding is prognostic context, not a treatment rule.
The evidence base was built from English-language prospective and retrospective cohort reports that included pretreatment BMI and usable outcome estimates or contingency-table data. Searches covered the Cochrane Library, Embase, PubMed and Web of Science from database inception through November 30, 2025, and the researchers also screened relevant reference lists. Adjusted estimates were the primary evidence, while unadjusted estimates were pooled separately, using inverse-variance random-effects models.
Unadjusted analyses pointed in the same direction for survival: the pooled hazard ratios were 0.72 for overall survival and 0.77 for progression-free survival. Unadjusted objective response and any-grade adverse-event estimates were higher, with odds ratios of 1.32 and 1.50, while the grade 3 or higher adverse-event estimate was 1.20. Leave-one-out checks were stable except for unadjusted objective response, and Egger's tests suggested possible small-study effects for unadjusted overall survival, progression-free survival and objective response.
Because these are associations from cohort reports, the review cannot establish that changing BMI causes longer survival. It also does not show that BMI identifies patients who will benefit more from an immune checkpoint inhibitor, or that it should be used to select or dose treatment. The authors' interpretation is narrower: pretreatment BMI may offer prognostic context, but it is not a treatment-selection or dosing biomarker.
Paper data and sources
Original title: Before-treatment body mass index as a prognostic factor in cancer patients receiving immune checkpoint inhibitors: a systematic review and dose-response meta-analysis
Authors: Sun S, Zhang J, Li Y et al.
Journal/Repository: Frontiers in immunology
Status: Peer-reviewed
First online: 2026-10-09
DOI: Not available
Original paper · Full text