Emerging observational reports have suggested an association between oral cavity cancer (OCC) and various forms of lymphoma, but observational data can be confounded. The authors used a two-sample bidirectional Mendelian randomization (MR) approach to investigate whether the observed relationship reflects a causal effect in either direction: from genetic liability to OCC on lymphoma risk, and from genetic liability to lymphoma on OCC risk.
The stated aim was to screen genetic instrumental variables (IVs) from genome-wide association studies (GWAS) for OCC and lymphoma and to apply MR methods to assess causality while testing the robustness of results through sensitivity, heterogeneity, and pleiotropy analyses.
The analysis employed a two-sample bidirectional MR framework. Genetic instrumental variables were obtained from GWAS datasets for OCC and for lymphoma phenotypes. The design tests causality in both directions separately: (1) genetic liability to OCC as the exposure and lymphoma (overall and subtype) as the outcome, and (2) genetic liability to lymphoma as the exposure and OCC as the outcome.
The abstract reports that various MR methods were used and that sensitivity, heterogeneity, and pleiotropic tests were conducted to evaluate the reliability of causal inferences. Specific GWAS sources, sample sizes, or lists of instrumental single-nucleotide polymorphisms (SNPs) were not detailed in the abstract and therefore are not reported here.
The inverse variance weighted (IVW) method was specified as the main MR estimator. The authors also applied additional MR approaches (not specified in the abstract) to corroborate the IVW findings. To assess possible violations of core MR assumptions, they performed sensitivity analyses, tests for heterogeneity across instruments, and assessments for horizontal pleiotropy. Details about the exact sensitivity methods (for example, MR-Egger, weighted median, or leave-one-out analyses), heterogeneity metrics, thresholds for instrument selection, and pleiotropy test statistics were not reported in the abstract.
The principal result reported is that there was no evidence of a bidirectional causal association between OCC and Hodgkin lymphoma (HL) or overall non-Hodgkin lymphoma (NHL).
Subgroup analyses of NHL subtypes likewise did not reveal any noteworthy bidirectional causal relationships between OCC and the examined NHL subtypes according to the abstract.
An exception emerged in one subtype analysis: the authors identified a statistically significant inverse correlation between mature T/NK-cell lymphoma and OCC when using the IVW method. The reported IVW estimate for this association was an odds ratio (OR) of 0.985 with a 95% confidence interval (CI) of 0.820–0.978 and p = 0.013. The abstract interprets this finding to mean that patients initially diagnosed with mature T/NK-cell lymphoma would have a decreased risk of OCC in this MR analysis.
No other statistically significant bidirectional causal associations were reported in the abstract.
Based on the MR analyses reported in the abstract, the authors conclude that a causal relationship between OCC and lymphoma in general was not supported. The solitary inverse association observed with mature T/NK-cell lymphoma suggests a potential subtype-specific relationship, but the abstract does not provide mechanistic explanations or confirmatory analyses beyond the MR estimate.
The overall interpretation emphasizes that, within the limits of the datasets and MR methods applied, broad causal links between OCC and lymphoma were not found.
The abstract notes a need for replication: future multiethnic studies with substantially larger sample sizes are required to verify the findings. The abstract did not provide detailed information about ancestral composition of GWAS datasets, the number of instruments, instrument strength metrics, or full sensitivity-analysis results; therefore, those methodological details are unavailable from the source abstract.
Clinicians and researchers should interpret the reported inverse association with mature T/NK-cell lymphoma cautiously, given the need for replication and the absence of additional methodological detail in the abstract. The study highlights the utility of Mendelian randomization for probing causality between cancers but also underscores the requirement for well-powered, diverse GWAS datasets to draw robust conclusions.