We recently read Association between Alzheimer's disease and risk of cancer: A retrospective cohort study in Shanghai, China, and the results suggested that Alzheimer's disease (AD) significantly increases the risk of lymphoma.¹ However, several previous studies have consistently shown an inverse correlation between AD and cancer.^2-4 Currently, the association between AD and hematological diseases is predominantly studied through cohort investigations, with limited attention given to lymphoma specifically.^{4, 5} Hence, due to less research and conflicting results from these cohort studies, we opted for Mendelian randomization (MR), a method that offers a higher level of evidence, to explore the causal relationship between AD and lymphoma.⁶ We were surprised to discover that AD significantly decreased the risk of non-Hodgkin's lymphoma (NHL; about 90% of all lymphomas), but not Hodgkin's lymphoma (HL; about 10% of all lymphomas; see Table S1 and S2, detailed analyses of HL are shown in Table S3 and Figures S1–S3 in supporting information).⁷

We used genome-wide association studies (GWAS) data from two major publicly available databases, namely, IEU Open GWAS (https://gwas.mrcieu.ac.uk) and FinnGen (R10, https://www.finngen.fi/fi), to assemble multiple cohorts of relevant GWAS data pertaining to AD, family history of AD, NHL and HL (Table S4 and Figure S4 in supporting information). The analysis process involves five methods named MR-Egger, weighted median, inverse variance weighted (IVW), simple mode, and weighted mode, and we selected the IVW method as the primary reference index. Analysis of three cohorts of AD revealed a significant reduction in the risk of developing NHL (I: OR_IVW = 0.823, P_IVW = 0.046; II: OR_IVW = 1.45E-15, P_IVW = 0.023; III: OR_IVW = 0.905, and P_IVW = 0.016), indicating that AD served as a protective factor (Figure 1, Figure S5 in supporting information, and Table S1). Additionally, we examined the effects of a family history of AD and it also significantly decreased the risk of developing NHL (OR_IVW = 0.777, P_IVW = 0.004). In addition to the IVW method, we also observed varying degrees of statistical significance with the other four methods (Figure 1). Meanwhile, we conducted heterogeneity, pleiotropy, and sensitivity analyses of the aforementioned results using MR-Egger, MR-PRESSO (pleiotropy residual sum and outlier), and “leave-one-out” methods. The findings demonstrated the robustness and reliability of our conclusions (Table S5, Figure S6 and S7 in supporting information). Detailed methods of analysis and instrumental variable (single nucleotide polymorphisms) information used are available in the supplementary materials (Supplementary Methods and Table S6 in supporting information).

In fact, we are not the first to elucidate the relationship between AD and NHL. Previous cohort studies, systematic reviews, and meta-analyses on AD and NHL have indicated that AD may serve as a protective factor, reducing the incidence of NHL.^{3, 4, 8} Only one MR study investigated AD in relation to follicular lymphoma (FL), a subtype of NHL, but the IVW method did not reveal a significant difference.² It is noteworthy that FL represents only 20% to 25% of new NHL cases and cannot fully represent all NHL subtypes.⁹ In our study, we utilized three different cohorts to investigate the effect of AD on NHL, all of which consistently showed AD as a protective factor. Nevertheless, it is important to acknowledge that we utilized GWAS studies from European populations, whereas Ren et al. utilized data from Chinese populations. Moreover, NHL does not represent the entirety of lymphoma. The above two points may be important reasons for the difference between our results.

Furthermore, we are the first to report that a family history of AD reduces the risk of developing NHL. Considering the significant role of genetic factors in AD pathogenesis, this finding extends the study's scope to a broader population and a longer timeline for understanding the relationship between the two. For instance, Valentine et al., utilizing the Utah Population Database (UPDB), identified individuals with a family history of AD as high-risk groups for AD to investigate various common diseases and cancers.¹⁰ Surprisingly, their findings indicated that high-risk groups for AD significantly elevated the risk of NHL, contradicting the conclusions drawn from other studies.¹⁰

Certainly, there is a broad spectrum of opinions regarding the mechanisms underlying the interactions between AD and NHL. While researchers have proposed a range of related mechanisms such as inflammation, immune response, infectious agents, and oxidative stress, these can only partially account for this inverse correlation.⁸ Clarifying the relationship between the two and identifying common pathophysiological pathways will aid clinics in developing new, optimal programs for the prevention and treatment of both AD and NHL.

In conclusion, from the perspective of evidence-based medicine, the MR methodology used in our study can offer a higher level of evidence regarding the relationship between the two. This can further stimulate research on their association, aiming to elucidate the intricate mechanisms underlying the interactions between AD and NHL.