Search

Individuals with schizophrenia (SCZ) have a 2–3-fold higher risk of mortality than the general population. Heritability of mortality in psychiatric disorders has been proposed; however, few have investigated SCZ family history and genetic variation, with all-cause and specific causes of death. We aimed to identify correlates of SCZ mortality using genetic epidemiological and genetic modelling in two samples: a Swedish national population sample and a genotyped subsample. In the Swedish national population sample followed from the first SCZ treatment contact until emigration, death or end of the follow-up, we investigated a standardised measure of SCZ family history. In a subgroup with comprehensive genetic data, we investigated the impact of common and rare genetic variation. Cox proportional hazards regression was used to estimate the association between various factors and mortality (all and specific causes). A total of 13727 SCZ cases fulfilled criteria for the population-based analyses (1268 deaths, 9.2%). The genomic subset contained 4991 cases (1353 deaths, 27.1%). Somatic mutations associated with clonal hematopoiesis with unknown drivers were associated with all-cause mortality (HR 1.77, 95% CI: 1.26–2.49). No other heritable measures were associated with all-cause mortality nor with any specific causes of death. Future studies in larger, comparable cohorts are warranted to further understand the association between hereditary measures and mortality in SCZ.

Schizophrenia (SCZ) is associated with high mortality. DNA methylation levels vary over the life course, and pre-selected combinations of methylation array probes can be used to estimate “methylation age” (mAge). mAge correlates highly with chronological age but when it differs, termed mAge acceleration, it has been previously associated with all-cause mortality. We tested the association between mAge acceleration and mortality in SCZ and controls. We selected 190 SCZ cases and 190 controls from the Sweden Schizophrenia Study. Cases were identified from the Swedish Hospital Discharge Register with ≥5 specialist treatment contacts and ≥5 antipsychotic prescriptions. Controls had no psychotic disorder or antipsychotics. Subjects were selected if they had died or survived during follow-up (2:1 oversampling). Extracted DNA was assayed on the Illumina MethylationEPIC array. mAge was regressed on age at sampling to obtain mAge acceleration. Using Cox proportional hazards regression, the association between mAge acceleration and mortality was tested. After quality control, the following were available: n = 126 SCZ died, 63 SCZ alive, 127 controls died, 62 controls alive. In the primary analyses, we did not find a significant association between mAge acceleration and SCZ mortality (adjusted p > 0.005). Sensitivity analyses excluding SCZ cases with pre-existing cancer demonstrated a significant association between the Hannum mAge acceleration and mortality (hazard ratio = 1.13, 95% confidence interval = 1.04–1.22, p = 0.005). Per our pre-specified criteria, we did not confirm our primary hypothesis that mAge acceleration would predict subsequent mortality in people with SCZ, but we cannot rule out smaller effects or effects in patient subsets.