Associations among homocysteine, left cortical thickness, and working memory in patients with treatment-resistant schizophrenia: preliminary findings of a cohort study

Abstract Background Treatment-resistant schizophrenia (TRS) is characterized by severe cognitive impairments. However, the underlying mechanisms driving these deficits and strategies to mitigate them remain unclear. Elevated homocysteine levels have also been implicated in disease progression. We aimed to explore the relationship among serum homocysteine levels, working memory, and cortical thickness in patients with TRS. Methods Eighty-five patients with schizophrenia (42 with TRS and 43 with non-TRS [NTRS]) and 58 healthy controls (HCs) were recruited. Serum homocysteine levels were measured using enzymatic cycling. Cortical thickness was assessed using FreeSurfer version 5.3. Mediation analysis was further performed to preliminarily explore the potential interplay among serum homocysteine levels, regional cortical thickness, and working memory performance. Given that the bivariate correlations between homocysteine levels and cortical thickness did not retain significance after false discovery rate (FDR) correction and the cross-sectional design, this analysis aimed to identify potential mediating or suppressive pathways for hypothesis-generating purposes, rather than verifying definite compensatory mechanisms. Results Working memory was significantly impaired in patients with TRS and NTRS compared with HCs. Homocysteine levels were elevated in patients with schizophrenia, particularly in the TRS group. Among individuals with TRS, an inverse relationship was observed between serum homocysteine concentration and working memory ( r = − 0.411, p FDR = 0.02). Our exploratory suppression analyses revealed preliminary statistical links between elevated cortical thickness in the left precentral, postcentral, and precuneus gyri and homocysteine-associated working memory deficits in patients with TRS. These hypothesis-generating observations are constrained by non-significant FDR-corrected correlations and cross-sectional design. Conclusions Our exploratory findings preliminarily indicate that greater cortical thickness within left-hemispheric brain regions may show statistical links to homocysteine-associated working memory deficits in patients with TRS, rather than definitive evidence of partial compensatory mechanisms. Notably, given the cross-sectional study design and non-significant bivariate correlations observed between homocysteine levels and cortical thickness following FDR correction, these results are strictly hypothesis-generating and cannot support causal inferences. The mediation-suppression analysis, performed despite non-significant FDR-adjusted pairwise associations, offers tentative clues regarding potential neural correlation patterns. Further longitudinal or interventional studies are required to validate these preliminary observations.