Introduction: Previous meta-analyses point to positive associations of type 2 diabetes (T2D) history with risk of non-Hodgkin lymphoma (NHL). However, these pooled estimates included both prospective cohorts and case-control studies, inconsistent adjustment for covariates, and a relatively small number of NHL diagnoses. Furthermore, few studies have had sufficient statistical power to evaluate associations for specific NHL histologic subtypes. Additionally, T2D is a common diagnosis in patients presenting with multiple myeloma (MM). Pooled studies suggest a positive association of T2D history with MM risk, but the epidemiologic evidence on this association remains inconclusive.

Methods: We pooled data from 6 large prospective cohort studies including the Cancer Prevention Study-II Nutrition Cohort, the California Teachers' Study, the Health Professionals Follow-up Study, the Nurses' Health Study (NHS), the NHS II, and a sample of Kaiser Permanente Southern California members. We confirmed incident NHL and MM diagnoses by medical record review or cancer registry linkage and classified NHL histologic subtypes according to World Health Organization and InterLymph Consortium guidelines. We analyzed all NHL (in aggregate), the more common NHL histologic subtypes, and MM. T2D history was assessed by self-report or by clinical diagnosis. We did not observe significant heterogeneity by cohort using Cochran's Q test from a random-effects meta-analysis model; thus, we analyzed the pooled sample across all cohorts to maximize our sample size. We used multivariable Cox proportional hazard models stratified on sex-specific cohort, age, and calendar year of follow-up and adjusted for race, educational attainment, smoking status, and adult body mass index (BMI) to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the association of T2D (yes/no) with risk of each NHL endpoint or MM. We also analyzed T2D duration (e.g., <5, 5 - 9.9, ≥10 years from T2D diagnosis) using similar Cox models to explore relevant etiologic timing.

Results: During a pooled total of ~12 million person-years of follow-up, we confirmed 11,485 incident NHL and 2,783 MM diagnoses. We did not observe a significant association of T2D history with the risk of all NHL or most NHL subtypes. However, T2D history was positively associated with risk of diffuse large B-cell lymphoma (DLBCL; HR, 95% CI: 1.14, 1.03 - 1.27) and inversely associated with risk of lymphoplasmacytic lymphoma/Waldenström macroglobulinemia (LPL/WM; 0.45, 0.27 - 0.74), mycosis fungoides/Sezary syndrome (MF/SS; 0.67, 0.67 - 0.98), and T-cell NHL (0.78, 0.63 - 0.98). Additionally, T2D history was positively associated with risk of MM (1.19, 1.06 - 1.33). In analyses of T2D duration, we observed elevated risks of DLBCL after 5 to 9.9 and after ≥10 years from T2D diagnosis, whereas the risks for LPL/WM and T-NHL were the lowest ≥10 years after T2D diagnosis. The risk of MM was the highest within 5 years and 5 to 9.9 years after T2D diagnosis. Lastly, we did not observe significant effect modification by sex, race, young adult or adult BMI for most associations of T2D history with risk of NHL or MM endpoints.

Conclusions: Our results suggest an association of T2D history with elevated risk of MM and with risk of DLBCL but not for other NHL subtypes. These findings of positive associations provide evidence for an etiologic role for hyperglycemia for these aggressive cancers and suggest that T2D prevention, independent of BMI, may be important in strategies to reduce MM and DLBCL incidence. The unexpected associations for T-cell NHL, MF/SS, and LPL/WM need to be interpreted with caution given the relatively small number of cases and the lack of plausible biological mechanisms to explain them. These observations warrant confirmation in other epidemiological studies, and further work is needed to elucidate potential mechanisms.

Disclosures

No relevant conflicts of interest to declare.

This content is only available as a PDF.
2024
Sign in via your Institution