Background
Schizophrenia (SCZ) is one of the most disabling illnesses [
1], affecting approximately 1% of the general population worldwide. SCZ patients have a higher rate of premature mortality [
2], and their life expectancy is roughly 15 years less than healthy individuals [
3]. Though suicides and accidents account for a large portion of premature mortality in SCZ, the majority of morbidity is attributed to age-related diseases, such as diabetes [
4], cardiovascular disease [
5,
6], and cancer [
7].
Multiple lines of evidence suggest that patients with SCZ show premature aging characteristics, such as cognitive decline [
8,
9], dendritic spine loss [
10], cortical atrophy [
11], shorted telomere [
12], and increased levels of inflammatory factors and oxidative stress [
13]. The accelerated aging hypothesis of SCZ has thus been proposed as a cause for the excess mortality in SCZ. This hypothesis proposes SCZ as a syndrome of accelerated aging associated with premature physiological change that increases the risk of aging-related medical conditions and mortality [
14]. However, testing this hypothesis is difficult due to the lack of accurate and robust biomarkers for biological age.
The recent development of DNA methylation (DNAm) based epigenetic clocks (also called epigenetic ages) offers a promise for addressing this challenge. The epigenetic age was integrated with DNAm levels at a set of cytosine-phosphate-guanine (CpG) sites using mathematical algorithms, producing an accurate and well-validated measure of chronological age [
15]. Epigenetic age acceleration or deceleration is defined as an increase or decrease in epigenetic age compared to chronological age. Accumulating evidence has shown that epigenetic age acceleration is associated with numerous risk factors and outcomes for psychiatric disorders, including adversity exposure [
16‐
18], cognitive decline [
19‐
21], altered brain structures [
22‐
25], and depression [
26‐
28]. However, previous studies have shown inconsistent results in SCZ. Although no acceleration of epigenetic aging has consistently been reported in blood and brain tissues of patients with SCZ [
29,
30], some studies also have detected accelerated or delayed epigenetic aging in patients with SCZ [
31‐
35]. Considering the pathological and clinical heterogeneity of SCZ, confounding variables such as antipsychotic medications and illness duration may partially explain these inconsistent findings.
Here, we investigated the epigenetic age in a cohort of drug-naive FSCZ patients and healthy controls using three independent epigenetic clock approaches, including Horvath [
36], Hannum [
37] and Levine [
38] methods. All drug-naive FSCZ patients received treatment with risperidone monotherapy and were followed up for 8 weeks. Then, we assessed the effect of antipsychotic treatment on the epigenetic aging process in FSCZ and investigated the potential associations of epigenetic age acceleration with psychotic symptoms, cognitive function, as well all subcortical volumes.
Discussion
This study investigated the epigenetic age and the effect of antipsychotic treatment on the epigenetic aging process in drug-naive FSCZ patients. We measured the epigenetic age acceleration by using three independent algorithms, including Horvath, Hannum and Levine’s epigenetic clocks. Our findings demonstrated a significant epigenetic age deceleration in Horvath’s epigenetic clock among drug-naive FSCZ patients relative to controls. In addition, we found that epigenetic aging of Hannum and Levine clocks was significantly accelerated in patients with FSCZ after 8-week risperidone treatment.
This study showed that epigenetic age acceleration is delayed in patients with drug-naive FSCZ against the accelerated aging hypothesis of SCZ. Our finding is consistent with two recent studies [
33,
42]. Talarico et al. found longer telomere length and decreased epigenetic age in drug-naive FSCZ patients [
42]. Wu et al. revealed an epigenetic age deceleration in SCZ patients by using the largest size of methylation datasets from 1211 brain tissues and 2333 whole blood samples [
33]. These results support the hypothesis that SCZ may be a neurodevelopmental disorder [
43]. Intriguingly, Wu et al. also found that some CpG sites of epigenetic clocks are differentially methylated in SCZ patients [
33]. Among these differentially methylated CpG sites (DMPs), 70–80% were located within the promoter regions. Furthermore, genes regulated by these DMPs displayed differential expression in SCZ patients and involved in the SCZ-related biological processes, such as immune dysregulation and neurological dysplasia [
33]. These findings suggest that epigenetic clocks might be mediated by the dysregulation of pathophysiological processes in SCZ [
33], which may be a potential cause of the epigenetic age deceleration underlying the development of SCZ. However, some previous studies have found that patients with SCZ had an accelerated epigenetic age or no difference in epigenetic age compared with healthy controls [
29‐
32,
44]. Multiple clinical variables, such as trauma history, sex, and antipsychotic treatment have been associated with epigenetic aging processes [
18,
35,
44], which may account for these inconsistent results and therefore should be taken into account in the future studies.
The current study demonstrated a significant effect of antipsychotic treatment on the Hannum and Levine clock, but not on the Horvath clock. Since each epigenetic clock is developed with different algorithms and captures distinct features of biological aging [
15], these inconclusive findings among three epigenetic clocks are understandable. Horvath clock was developed with DNAm datasets from multiple tissues and development stages, measuring cellular aging independent of cell-type compositions [
36]. Hannum clock was trained on blood samples, capturing more cell-extrinsic aging with moderate correlation with cell compositions [
37]. Whereas the Levine clock was trained on the older adult population incorporating biological measures and capturing phenotypic age [
38]. As Horvath clock is independent of cell-type compositions and cell-extrinsic biological measures [
36], we supposed that the antipsychotic effect on accelerated epigenetic age in SCZ patients may be caused by the alterations of blood cell composition and other aging-related blood biomarkers. In line with our supposition, Talarico et al. recently found an accelerated epigenetic age in FSCZ patients after 10 weeks of treatment with risperidone. However, this result was not observed after adjusting for blood cell composition [
42]. Furthermore, amounting evidence also has shown that antipsychotic treatment may influence the blood cell type compositions or other blood-based biomarkers [
45,
46], which further support that the effects of antipsychotic treatment on epigenetic age may be biased by the blood cell composition. However, we cannot rule out the possibility that risperidone may accelerate epigenetic age through other biological processes. For example, risperidone treatment may change the methylation level of some CpG sites among epigenetic clocks, but this effect might be insufficient to be detected in the current study because of the short-term treatment.
Although SCZ-related genes were found to be regulated by the epigenetic clock and displayed abnormal expression in SCZ patients [
33], the mechanism of epigenetic aging process implicated in the SCZ pathogenesis remains unclear. A recent study demonstrated a significant cross-sectional association between epigenetic age acceleration and psychosis severity that was measured by the Symptom Checklist 90 (SCL-90) psychotic domain [
47]. Unlike this association finding, our analyses showed that epigenetic aging acceleration from all three epigenetic clocks was neither correlated with psychotic symptom severity (PANSS scores) nor symptom improvement (the percentage change on PANSS scores). In addition to the difference in symptom measures, the medicinal uses and substantially clinical heterogeneity in SCZ patients may partly account for this discrepancy.
We found no significant associations between epigenetic age acceleration and cognitive function in all participants at baseline. Our longitudinal analyses showed that baseline epigenetic age acceleration has an inverse association with the improvement of cognitive functions in some measures. However, this significant finding is not retained after multiple test corrections. This is not surprising given that the statistical power of our study is limited by the small sample size and a short-term follow-up. A recent large cross-sectional study investigated the association between two measures of epigenetic age acceleration (Horvath and Hannum) and three different neuropsychological tests in 4827 middle-aged participants from two independent cohorts, and only found a significant inverse association between Hannum AgeAccel and Word Fluency Test scores [
48]. Another twin cohort study examined both cross-sectional and longitudinal associations (11.5-year) between four epigenetic clocks using Horvath, Hannum, Levine, and Grim algorithms and cognitive function using Trail Making Test (TMT) [
20]. This twin study found that Horvath AgeAccel were correlated with cognitive decline longitudinally, but no association between epigenetic age acceleration and cognitive function at baseline [
20]. Differences in the measures of epigenetic clocks and cognitive function in previous studies may account for the inconsistent results.
Only a few studies have investigated the relationship between epigenetic age acceleration and subcortical volume. A previous longitudinal study of 46 adolescent girls found Horvath AgeAccel related to reduced left hippocampal volume (4 years follow up) [
23]. Similarly, two recent cross-sectional studies found a significant cross-sectional association between Hannum AgeAccel and reduced hippocampal volume, but this finding was not observed in Horvath’s clock [
22,
25]. In our study, we failed to find any significant relationships between epigenetic age acceleration and subcortical volumes. This may also be related to different measures of epigenetic clocks and our small sample size as the cause of the discrepancy in cognitive function. Large replicate data with well-developed epigenetic clocks will be needed to draw a conclusion.
This study has several strengths. First, we investigated the epigenetic age acceleration in drug-naive FSCZ patients with a similar illness duration to exclude the potential confounding effects of medication and the pathophysiological process of SCZ on our results. Second, we analyzed the effect of medication on the epigenetic aging process in FSCZ through a longitudinal design of the cohort. Third, we implemented multiple epigenetic clocks that are designed with different algorithms and capture distinct epigenetic aging features, enabling a deeper understanding of various aging processes in SCZ. However, a major limitation of the present study should be noted. Our study has a relatively small sample size, which may limit the detective power of our study, accounting for the inconsistent findings from distinct epigenetic clocks analyses.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.