Investigation of Epistasis Between DAOA and 5HTR1A Variants on Clinical Outcomes in Patients with Schizophrenia

Introduction

Schizophrenia (SCZ) is a disabling psychiatric disorder affecting about 1% of the population worldwide (Freedman, 2003). Both formal and molecular genetic studies suggest that this disorder has a strong genetic etiology (McGuffin et al., 1995). Similarly, genetic factors seem to significantly influence the response to antipsychotic drugs commonly employed to treat SCZ (Arranz et al., 2011). However, consensus has not yet been achieved regarding the specific genetic variants involved in SCZ, and current evidence suggests that it is unlikely that a single genetic variant could be the only determinant of responses to antipsychotics (Zhang and Malhotra, 2011). Rather, different systems are thought to interact with one another in a complex web that modulates the response to these drugs (Arranz et al., 2011).

In two recent studies, we found that rs10042486, a single-nucleotide polymorphism (SNP) within the 5-hydroxytryptamine receptor 1A (5HTR1A) (Crisafulli et al., 2012), and rs7139958, a SNP within the d-amino acid oxidase activator (DAOA) (Chiesa et al., 2011), could be associated with clinical improvement, as detected by the positive symptom subscale of the Positive and Negative Symptoms Scale (PANSS) (Kay et al., 1987) in a sample of 221 Korean SCZ patients treated with various antipsychotics. More specifically, we observed that SCZ subjects carrying the rs10042486 TT genotype were significantly more likely than were those carrying the CT and the CC genotypes to improve on PANSS total and on the PANSS-positive and -negative scores (Crisafulli et al., 2012). Furthermore, we observed that SCZ patients carrying the rs7139958 TT genotype were more likely to improve on the PANSS-positive subscale scores than were those carrying the AA and AT genotypes (Chiesa et al., 2011). These associations were driven primarily by a significant difference in PANSS-positive subscale scores at baseline that was no longer present at endpoint (Chiesa et al., 2011).

Given that results deriving from single studies have been largely inconsistent or associated with very small effect sizes (Arranz et al., 2011), increasing emphasis is being placed on epistatic interactions between polymorphisms in candidate genes possibly influencing treatment outcomes in complex diseases such as psychiatric disorders (Carlborg and Haley, 2004; Segurado et al., 2011). Epistasis is commonly defined as the functional interaction between genes. It encompasses various events, including epigenetic control, chromatin remodeling, promoter activity control, and many other molecular reactions (Phillips, 2008). These events can impact cell lifecycles and complex traits and are orchestrated through genetically driven complex yet flexible activities (Phillips, 2008). Thus, research on the epistatic interactions between different candidate genes may represent a significant advance over single gene investigations as a way to understand the biological diversity that may influence, for instance, drug response.

The aim of the present work was to investigate the possible epistatic interactions between rs10042486 within 5HTR1A and rs7139958 within DAOA associated with improvement on the PANSS-positive subscale and other clinical variables in a sample of 221 Korean SCZ patients treated with different antipsychotics.

Methods

The sample in the present study was described in detail in previous studies (Chiesa et al., 2011; Crisafulli et al., 2012). Briefly, 221 in-patients suffering from SCZ according to DSM-IV criteria based on the Mini-International Neuropsychiatric Interview (MINI) (Sheehan et al., 1998) were consecutively recruited at the Department of Psychiatry of the Catholic University of Korea College of Medicine, Seoul, Korea. There was no particular restriction with regard to the specific antipsychotic treatment employed. However, patients were excluded if they were characterized by one or more of the following: one or more current severe or unstable medical and/or neurological conditions, current treatment with a long-acting antipsychotic, concomitant alcohol and/or substance abuse disorders, and/or non-Korean ethnicity. Each patient admitted to the hospital was assessed with the PANSS to assess the baseline severity of illness, and then reassessed to determine the symptom severity at the time of discharge. The study protocol was approved by the Institutional Review Board (approval number HC10TISI0031).

In accordance with previous studies, a therapeutic response was defined a priori as a ≥50% reduction in symptom severity from baseline to discharge (Leucht et al., 2007). Eleven clinical and sociodemographic variables were likewise recorded (Table 1).

The main outcome measure was the effect of epistatic interactions between rs10042486 within 5HTR1A and rs7139958 within DAOA on PANSS-positive subscale scores. Secondary outcome measures included the effect of epistatic interactions between the same two variants on other clinical and sociodemographic variables included in the present study. Statistical analyses were performed using the Statistica software package (StatSoft, 1995). A multiple regression model was employed to investigate possible epistatic interactions between the two genotypes and to select clinical and sociodemographic variables included in the present study. Clinical improvements on PANSS total scores, including positive, negative, and general scores, were calculated at baseline (admission) and endpoint (discharge).

All p-values were two-tailed and statistical significance was conservatively set at the 0.002 level (approximately corresponding to the 20 variables outlined in Table 1). These parameters provided sufficient power (0.80) to detect a medium (d=0.26) effect size for patients carrying the TT rs7139958 genotype as compared with those carrying the AT genotype (Cohen, 1988). The effect size corresponded to the possibility of detecting final differences in PANSS total scores of 4.9 points.

Results

The sociodemographic and clinical characteristics of our sample are reported in Table 1. No significant effects of epistatic interactions between rs10042486 and rs7139958 on PANSS-positive subscale scores were observed (β=−0.02, p=0.93). Furthermore, we found that the independent associations between rs10042486 and rs7139958, on one hand, and improvements on the positive subscale, on the other hand, were no longer significant when they were included in the multiple regression model (β=−0.20, p=0.25 and β=0.17, p=0.34). This may be attributable to the reduced power resulting from the combined analysis or to an undetected masked interaction. In terms of secondary outcome measures, no significant associations were observed (all p-values>0.002).

Discussion

The present work examined the effect of epistatic interactions between rs10042486 within 5HTR1A and rs7139958 within DAOA on improvement on the PANSS-positive subscale and other clinical measures in a sample of SCZ patients treated with antipsychotics. The findings revealed no significant effects of an epistatic interaction between the two SNPs on PANSS-positive subscale scores. One possible explanation for this negative finding may be that the moderately small sample size was insufficient for detecting subtle differences that are usually associated with the effects of single genes or gene interactions on complex disorders (Risch et al., 2009). However, the possibility that our negative finding reflects a lack of epistatic interaction between the two SNPs that were previously independently associated with PANSS-positive subscale scores (Chiesa et al., 2011; Crisafulli et al., 2012) cannot be completely ruled out. Additionally, the positive association found in our previous studies (Chiesa et al., 2011; Crisafulli et al., 2012) may not have been sufficiently robust to yield an interaction effect between the two SNPs. Similar findings have been presented in other studies as well (Andreasen et al., 2012; Moon et al., 2011).

It should be noted, however, that it was not only the epistatic interaction of these two SNPs that was negative in the present study. Indeed, the independent associations between rs10042486 and rs7139958 and improvement on the PANSS-positive subscale observed in earlier studies were no longer significant in our model. This finding suggests that the positive genetic associations observed in some samples cannot be replicated in different samples because of consistent differences in the genotype frequencies of other genetic polymorphisms that epistatically interact with the specific variants in a given study. As recently argued, such a possibility raises formidable computational and theoretical difficulties when considering all possible gene-gene interactions, not to mention other forms of genetic expression controls (Serretti and Chiesa, 2012). Finally, we observed no additional significant effects of epistatic interactions on any of the secondary outcome measures. These findings substantially confirm findings in our original studies (Chiesa et al., 2011; Crisafulli et al., 2012). Furthermore, they extend those findings by suggesting that no epistatic interaction exists between these SNPs.

The shortcomings of the present study should be considered. First, the lack of association in the present study may be due simply to a lack of statistical power that, in turn, may have obscured the small effects exerted by single SNPs. This issue is particularly concerning given that the results of recent studies, including thousands of patients tend to report conflicting results (Abou Jamra et al., 2006; Shi et al., 2008). A further concern is related to the use of different antipsychotics (that may operate via different action mechanisms), which precludes definitive conclusions regarding the influence of the SNPs on the effects of different specific antipsychotics or classes of antipsychotics. Our decision to include patients treated with different drugs may have had the advantage of generalizability by replicating circumstances that may more closely resemble the real world of clinical practice. Another concern regards the possible effects associated with duration of hospitalization. In the present study, a patient's length of hospital stay may have been insufficient for ascertaining a lack of response and remission, although the time frame afforded to patient treatment in the current study resembles that of common clinical practice (Zimmerman et al., 2002). Furthermore, differences in the duration of hospitalization may obscure results indicating that clinical improvement varies as a function of time rather than of genetic variants. However, we examined this factor in our statistical analyses and found no significant influence.

In conclusion our findings suggest that epistatic interactions between rs10042486 within 5HTR1A and rs7139958 within DAOA may not affect clinical outcomes in SCZ patients treated with different antipsychotics. Future research should investigate whether our results can be replicated using larger samples of SCZ patients treated with different antipsychotics.