BMI and Its Changes As a Predictor of Survival in DLBCL Patients

Introduction

The association between body mass index (BMI) at the time of diagnosis (dg) and the outcomes of patients with diffuse large B-cell lymphoma (DLBCL) remains controversial. As a systemic symptom, BMI loss presumably reflects the disease aggressivity, while during therapy, therapy failure, treatment-related complications, nutritional interventions failure and other factors might contribute to BMI changes. We aimed to investigate the impact of BMI at the time of dg as well as BMI changes from their pre-illness values to dg (pre-to-dg) and from dg to the end of first-line therapy (dg-to-EoT) on prognosis of DLBCL patients.

Methods

Altogether 726 consecutive patients with newly diagnosed systemic DLBCL between 2010–2021 have received first-line R-CHOP regimen at single university hospital center in the Czech Republic, Of themwith anthropometric data available at the dg and during the first-line therapy entered the analysis. Medical records of each patient were reviewed in detail including height, weight (i.e., BMI) and its changes (pre-illness BMI before dg; at dg; EoT), baseline characteristics, treatment-related complications, nutritional interventions, response rates (overall response rate, ORR; complete remission rate, CR) and outcomes (PFS, OS). Patients were divided into 3 groups based on the BMI at the dg (< 25, 25–29.9, ≥ 30 kg/m²), and based on the BMI changes pre-to-dg as well as dg-to-EoT. Univariate analysis was performed using log-rank test, multivariate analysis was performed using Cox regression model.

Results

BMI at diagnosis. Median age at dg of the 634 analyzed patients was 64 years (IQR 52–70 years), PS ECOG ≥ 2 was observed in 23%, clinical stage III–IV in 66%, elevated LDH in 62%, EN involvement ≥ 2 in 41%, IPI 3–5 in 53%, and decreased serum albumin levels in 20%. Median BMI at dg was 27.8 kg/m². BMI value < 25 kg/m² at dg was observed in 247 (39%), 25–29.9 kg/m² in 243 (38%), and BMI ≥ 30 kg/m² in 144 (23%) of patients. EoT response rates, as well as in PFS (P = 0.236) and OS (P = 0.233) between the BMI subgroups didn’t significantly differ between the subgroups.

BMI trends. ROC analyses revealed the optimal threshold of BMI loss pre-to-dg of 10%, while dg-to-EoT od 5%. A total of 169 (27%) patients presented with ≥ 10% pre-to-dg BMI loss were associated inferior CR rate (72% vs 84%, P = 0.019), and survival (PFS HR 1.5, P = 0.003; OS HR 1.4, P = 0.033). Consequently, 195 (31%) patients with dg-to-EoT BMI decrease of ≥ 5% had inferior CR rate (70% vs 84%, P = 0.001), and survival (PFS HR 2.0, P < 0.001; OS HR 2.0, P < 0.001). These results are presumably given by correlation with inferior baseline characteristics of both BMI losing cohorts.

Combination of BMI at diagnosis with BMI trends. The significantly inferior survival with pre-to-dg BMI loss of ≥ 10% was observed among all patients, and was mainly driven by the BMI < 25 kg/m² group (n = 89 vs 158; PFS HR 2.2, P < 0.001; OS HR 2.3, P < 0.001). Multivariate analysis of systemic BMI trend together with the IPI factors revealed BMI decrease, age, and clinical stage to be independent prognostic factors of PFS and OS in the BMI < 25 kg/m² subgroup. The significance of dg-to-EoT BMI loss of ≥ 5% was retained among all 3 BMI groups. These BMI losing patients had significantly inferior response rates to therapy (CR 69% vs 85%, P < 0.001), more frequent chemotherapy delays/reductions (25% vs 15%, P = 0.008), nutritional interventions during therapy (37% vs 17%, P < 0.001), but the incidence of treatment-related complications didn’t differ in comparison to those with BMI decrease < 5% (52% vs 52%, P = 0.985). Multivariate analysis revealed the dg-to-EoT BMI trend, response rates, and chemotherapy delays/reductions to be independent prognostic factors of PFS and OS.

Conclusion

BMI as a single value at the time of DLBCL dg does not significantly impact outcomes of DLBCL patients. However, the BMI decrease both as a systemic symptom and during the first-line therapy identifies higher-risk patients with worse response rates to the first-line treatment, as well as inferior survival. These patients might require closer observation and more personalized approach during therapy.

First two authors contributed equally. This work was supported by the Charles University Hematology-Oncology Cooperatio Program and grant NU21-03-00411.