Deciphering the causal relationship between atopic dermatitis and lymphoma via multi-omics Mendelian randomization and colocalization analyses

To the Editor: Atopic dermatitis (AD) is a chronic inflammatory skin condition marked by recurrent eczematous lesions and intense pruritus, affecting approximately 15–20% of children and up to 10% of adults worldwide.[1] The presence of AD significantly impairs patient quality of life and imposes substantial economic burdens on society. In recent years, the potential link between AD and lymphoma has attracted considerable attention from researchers and clinicians. Lymphoma, a malignancy of the lymphatic system, poses significant health risks due to its high mortality potential.[2] However, observational studies investigating the potential correlation between AD and lymphoma have yielded inconsistent results. Consequently, the association between AD and lymphoma remains controversial and warrants further investigation. To assess the causal links between AD and lymphoma, we initially employed two-sample bidirectional Mendelian randomization (MR). Subsequently, metabolome-wide MR analyses were conducted to evaluate the roles of blood and urine metabolites in mediating these causal relationships. Additionally, we integrated proteome-wide MR, summary-data-based MR (SMR), and colocalization analyses to identify the causal relationships between plasma proteins and the risk of AD and lymphoma, aiming to discover new druggable targets. This study utilized publicly available summary-level data [Supplementary Table 1, https://links.lww.com/CM9/C273]. All studies received ethical approval from the relevant institutional review boards, with informed consent obtained from all participants. Detailed materials and methods are provided in Section “Material and Methods” in Supplementary Material, https://links.lww.com/CM9/C271. The two-sample MR analyses indicated that AD was associated with an increased risk of NK/T cell lymphoma (NKTL) in both the UK Biobank (UKB) and FinnGen cohorts. This association remained statistically significant following meta-analysis and Bonferroni correction [Figure 1A]. A positive association between AD and the risk of mantle cell lymphoma (MCL) was suggested in the FinnGen cohort, but not in UKB cohort. However, this association reached statistical significance after meta-analysis and Bonferroni correction. No evidence was found supporting a correlation between AD and the risk of developing Hodgkin lymphoma, follicular lymphoma, diffuse large B cell lymphoma, or marginal zone B cell lymphoma [Supplementary Tables 2 and 3, https://links.lww.com/CM9/C273]. The reverse MR analyses suggested that lymphoma was not associated with an elevated risk of AD.Figure 1: Research diagram and summary of study. (A) Meta-analyses reveal the associations between AD and the risk of NKTL and MCL. (B). Metabolome-wide MR analyses reveal blood and urine metabolites associated with the risk of AD, NKTL, and MCL. (C). Workflow of conducting proteome-wide MR analysis using pQTL data from deCODE and UKB-PPP studies, supplemented by SMR and colocalization analysis, to identify associations between plasma proteins, AD, NKTL, and MCL. (D). Manhattan plot displaying PheWAS analysis results for ERN1, FAS, PRDX6, and PSMD5 in the UKB-SAIGE study. The Y-axis represents the logarithm of the P-values. The dotted line indicates the threshold for suggestive statistical significance (P <0.05). The top six associations are labeled by name. The 783 phenotypes from the UKB-SAIGE study are clustered into 17 groups shown in X-axis: 1 circulatory system; 2 congenital anormalies; 3 dermatologic; 4 digestive; 5 endocrine/metabolic; 6 genitourinary; 7 hematopoietic; 8 infectious diseases; 9 injuries & poisonings; 10 mental disorders; 11 musculoskeletal; 12 neoplasms; 13 neurological; 14 pregnancy complications; 15 respiratory; 16 sense organs; 17 other symptoms. An interactive 3D illustration is provided in Supplementary Figure 1, https://links.lww.com/CM9/C272 (electronic version only). AD: Atopic dermatitis; AST: Aspartate aminotransferase; CEs: Cholesterol esters; CHO: Cholesterol; coloc: colocalization; ERN1: Endoplasmic reticulum to nucleus signaling 1; Finn: FinnGen study; MCL: Mantle cell lymphoma; MR: Mendelian randomization; NKTL: NK/T cell lymphoma; PheWAS: Phenome-wide association studies; pQTL: Protein quantitative trait locus; PRDX6: Peroxiredoxin 6; PSMD5: Proteasome 26S Subunit Non-ATPase 5; SMR: Summary-data-based Mendelian randomization; UKB: UK Biobank study.The metabolome-wide MR analyses identified 9 metabolites associated with the risk of developing AD in the FinnGen cohort and 11 metabolites in the UKB cohort [Supplementary Tables 4 and 5, https://links.lww.com/CM9/C273]. Meta-analysis of the FinnGen and UKB datasets identified six metabolites that may influence the risk of developing AD [Figure 1B]. The MR analyses also suggested that AD was associated with a slight decrease in albumin and insulin-like growth factor 1 (IGF-1), and a moderate increase in cholesteryl ester levels in chylomicrons and extremely large very low-density lipoprotein (VLDL) particles after meta-analysis [Supplementary Tables 4 and 5, https://links.lww.com/CM9/C273]. Furthermore, the MR analyses indicated that elevated levels of apolipoprotein A (ApoA) were associated with a decreased risk of developing NKTL, while sphingomyelins, cholesterol, and its derivatives were associated with an increased risk of developing MCL [Figure 1B and Supplementary Tables 6–9, https://links.lww.com/CM9/C273]. Next, we conducted proteome-wide MR, SMR, and colocalization analyses to elucidate the association between circulating proteins and the risk of AD, NKTL, and MCL. These analyses identified 11 proteins associated with the risk of AD, 5 proteins associated with the risk of NKTL, and 10 proteins associated with the risk of MCL [Figure 1C and Supplementary Tables 10–13, https://links.lww.com/CM9/C274]. Both proteome-wide MR and SMR analyses indicated that Endoplasmic Reticulum to Nucleus Signaling 1 (ERN1) and Proteasome 26S Subunit Non-ATPase 5 (PSMD5) were associated with the progression of both AD and NKTL. Additionally, proteome-wide MR analyses from both the deCODE and UKB-PPP studies, as well as the SMR analysis, implicated Peroxiredoxin 6 (PRDX6) in the progression of AD and Fas Cell Surface Death Receptor (FAS) in the pathogenesis of MCL. Notably, the association between ERN1, PSMD5, PRDX6, and FAS with AD and/or lymphoma were corroborated by multiple analyses, highlighting them as potential targets for drug repurposing. Given that the associations of these proteins with their respective diseases were corroborated by multiple screening methods and their relevance as drug targets, these four proteins were selected for phenome-wide association studies (PheWAS) to evaluate potential safety implications. MR analyses revealed that ERN1 was primarily associated with digestive diseases and neoplasms, FAS was predominantly associated with neoplasms and melanomas of the skin, PRDX6 had a strong association with epilepsy, and PSMD5 demonstrated significant associations with respiratory and skeletal diseases [Figure 1D and Supplementary Table 14, https://links.lww.com/CM9/C274]. An interactive 3D illustration is provided [Supplementary Figure 1, https://links.lww.com/CM9/C272]. Aligned with previous observational studies,[3,4] our study suggests that AD is associated with an increased overall risk of lymphoma, specifically NKTL and MCL. Metabolome-wide MR analyses indicated that docosahexaenoic acid (DHA) and histidine are associated with a reduced risk of AD, consistent with their known pharmacological and nutritional benefits. AD led to alterations in cholesterol accumulation, which elevated the risk of lymphomas.[5] Our analysis further identified an association between sphingomyelins and an increased risk of developing MCL, aligning with the recognized role of sphingomyelins in cancer initiation, growth, and immune evasion.[6] Overall, alterations in lipid metabolism, particularly cholesterol, may bridge the causal links between AD and lymphoma. Proteome-wide MR using pQTL data from the deCODE and UKB-PPP studies, aided by SMR and colocalization analyses, identified distinct plasma proteins associated with AD, NKTL, and MCL. However, shared proteins between these diseases were also identified. Furthermore, JAK-STAT signaling pathway and cytokine dysregulation may drive both conditions.[7] Studies have also demonstrated that inflammatory and immune-related genetic loci exhibit pleiotropy, affecting multiple diseases.[8] These findings, together with our study, further corroborate the genetic correlations between AD and lymphoma, underscoring the importance of conducting lymphoma screening in patients with AD to potentially improve early detection and intervention strategies. By leveraging multi-omics data and employing various analytical tools, our research provides insights into the intricate relationship between AD and lymphomas, ultimately guiding the development of targeted interventions and improving risk stratification and personalized patient management. Funding This study was supported by grants from the National Key R&D Program of China (Nos. 2022YFC3602002 and 2024YFF0507404), and National High Level Hospital Clinical Research Funding (Nos. 2022-NHLHCRF-LX-02-03 and 2023-NHLHCRF-YXHZ-ZRZD-06).