Our study investigated the genetic association and causality between frailty and schizophrenia from the genetic perspective based on the large-scale GWASs summary data. First of all, the global genetic correlation analyses presented they were positively associated, and the local genetic correlation demonstrated they were locally correlated in three genomes. Furthermore, the condFDR/conjFDR method indicated there were 111 genomic loci in total which are jointly associated with both frailty and schizophrenia. In addition, the tissue enrichment and SMR analyses demonstrated the genetic variants related to frailty and schizophrenia have overlapped tissue enrichments and functional genes in the brain. Lastly, the MR results implied there was a bidirectional causal relationship between frailty and schizophrenia.
Consistent with the previous study that reported a moderate genetic correlation between frailty and depression7, our LDSC and HDL analyses indicated that genetically predicted frailty was meanwhile positively related to schizophrenia. This genetic positive correlation was observed in several epidemiology studies demonstrating the prevalence of frailty in patients with schizophrenia was significantly higher than in the controls16,40. Additionally, we examined their local genetic correlation and discovered they were correlated in 202 genomic regions, while only three of these remained significant after correction for multiple tests. The gene NFIL341 in the first region (chr9: 94167203–96671698) and five genes in the second region (chr11: 112459488–114257728) including NCAM142, DRD243, HTR3A44, HTR3B44, and ZBTB1645, were involved in the pathobiology of schizophrenia. While the relevant studies on frailty were relatively limited, we did not find any risk genes that had been studied for frailty in these three regions.
We further identified 111 genomic loci that were shared by frailty and schizophrenia via the pleioFDR software, among which the most significant locus is located in chromosome 4 (Lead SNP: rs13107325 in the SLC39A8 gene). Currently, multiple studies have not only validated the association between rs13107325 and schizophrenia in Europeans46 but also in Chinese47. Besides, among the 228 independent SNPs in the shared genomic loci, an appropriate 10% were located in the three local genomic regions identified by the local genetic correlation analyses: 5 SNPs (rs1351117, rs10992733, rs10821168, rs10761245, rs10761247) in chr9: 94167203–96671698, 13 SNPs (rs17529477, rs10891564, rs7107293, rs78169211, rs12420205, rs10891570, rs12222458, rs10891571, rs4373974, rs6589386, rs75059851, rs733856, rs12277680) in chr11: 112459488–114257728, and 6 SNPs (rs55905661, rs12457876, rs586275, rs71367544, rs516890, rs62103240) in chr18: 77149991–78017158. These results further emphasized the 3 locally correlated loci play an important role for both traits.
The results of tissue enrichments indicated the tissues in which genetic variants related to frailty enriched were all brain tissues and completely overlapped by schizophrenia associated genetic variants. In addition, their top 2 significant enriched tissues were the Brain Frontal Cortex BA9 and Brain Cortex. These findings were consistent with the existing results that frailty was associated with reduced cortex volume48, and increased cortical brain infarcts49. Meanwhile, schizophrenia patients also have widespread cortical thinning and smaller cortical surface area, especially in frontal and temporal lobe regions50.
Based on the findings from shared enrichment of tissue types, we further utilized the brain cis-QTL and SMR method to explore their shared risk genes in brain, and we identified 425 candidate causal risk genes for schizophrenia and 101 for frailty. Interestingly, 11 were found to be shared between these two traits. Many of shared genes are reported to play important roles in central nervous systems. For instance, Deleted in Colorectal Cancer (DCC) has shown a significant association with various psychiatric diseases, including schizophrenia, bipolar disorder, and depression, according to the PGC cross-trait GWAS studies51. Functional studies have suggested that DCC may participate in axonal guidance52 and regulate synaptic function and plasticity53, indicating its potential contribution to these diseases by affecting synapse functions. Another gene, Neuronal growth regulator 1 (NEGR1), has been reported to be significantly associated with depression and schizophrenia46,54. Dysregulation of NEGR1 protein and gene transcript has been observed in the prefrontal cortex (PFC) region of individuals with schizophrenia55,56. NEGR1 is known to participate in cell adhesion and belongs to the IgLON superfamily57, which is associated with various central nervous functions such as intelligence58, learning, and behavior59,60. Additionally, NEGR1 has been reported to regulate synapse formation in hippocampal neurons61 and hippocampal neurogenesis60. Both DCC and NEGR1 have functions in synapses, suggesting that they may play a role in mediating the connection between schizophrenia and frailty.
We used the bidirectional MR to assess the causality between frailty and schizophrenia, and significant causality was discovered and supported by the series of sensitivity analyses. Although current studies on frailty and schizophrenia were very limited, our findings were consistent with some results of research targeting the relationship between schizophrenia and frailty components. Firstly, for physical activity, the meta-analysis of randomized trials found that physical activity interventions could obviously reduce the symptoms of schizophrenia62, and reversely the schizophrenia patients were found to engage in significantly less moderate and vigorous physical activity versus controls63. For the walking speed and grip strength, a reduced walking capacity and significant lower handgrip strength were discovered in schizophrenia patients due to impaired muscular fitness64,65, thus strength and endurance training were recommended for schizophrenia66. In addition, patients with schizophrenia spectrum disorders reported significantly higher fatigue scores than the general population67. However, opposite to weight loss, the prevalence of obesity and schizophrenia was positively correlated68,69, and plenty of genetic overlap between body mass index and schizophrenia was discovered by the recent study70.
Our study has several strengths worth pointing out. Firstly, to the best of our knowledge, this is the first study to investigate the relationship including the genetic association and causality between frailty and schizophrenia. Secondly, multiple statistical methods were performed to explore their genetic association, which improved the stability and reliability of our findings. Thirdly, our study implemented the two-sample MR analyses based on the genetic data from the large-scale GWASs, making the causality inference robust and unlikely to suffer from the bias from conventional observational studies, such as reverse causality, confounding factors, etc.
Despite these strengths, our findings should be interpreted in the context of some limitations. Firstly, the current study lacked ethnic diversity as the GWAS summary statistics were largely derived from individuals of European ancestry. Therefore, it is necessary to validate the findings in other ethnicities and conduct trans-ethnic GWAS to obtain more comprehensive results71. Secondly, it is important to note that GWAS only captures the association between common genetic variations and phenotypes. However, a complete understanding of the genetic components underlying disease risk requires consideration of other types of genetic variants, such as structural variations72. Therefore, further studies should be conducted if data on these variants are available for both schizophrenia and frailty. Thirdly, the frailty index was another useful scale to assess the frailty status based on 44 or 49 self-reported items on symptoms, disabilities, and diagnosed diseases. We adopted the fried frailty score as the measurement of frailty in our study as it was more concise and maneuverable than the frailty index. This discrepancy may induce different findings and need further investigation.