|Home | About | Journals | Submit | Contact Us | Français|
The initial report of an interaction between a serotonin transporter promoter polymorphism (5-HTTLPR) and stress in the development of depression is perhaps the best-known and most cited finding in psychiatric genetics. Two recent meta-analyses explored the studies seeking to replicate this initial report and concluded that the evidence did not support the presence of the interaction. However, even the larger of the meta-analyses included only 14 of the 56 studies that have explored the relationship between 5-HTTLPR, stress and depression.
We sought to perform a meta-analysis including all relevant studies assessing whether 5-HTTLPR moderates the relationship between stress and depression.
We identified relevant articles from previous meta-analyses and reviews and a PubMed database search.
We excluded two studies presenting data that were included in other, larger, studies already included in our meta-analysis to avoid duplicate counting of subjects.
In order to perform a more inclusive meta-analysis, we used the Liptak-Stouffer Z-score method to combine findings of primary studies at the significance test level rather than raw data level.
We included 54 studies and found strong evidence that 5-HTTLPR moderates the relationship between stress and depression, with the 5-HTTLPR s allele associated with an increased risk of developing depression under stress (p<0.0001). When restricting our analysis to the studies included in the previous meta-analyses, we found no evidence of association (Munafo studies p=0.16; Risch studies p=0.11). This suggests that the difference in results between previous meta-analyses and ours was not due to the difference in meta-analytic technique but instead to the expanded set of studies included in this analysis.
Contrary to the results of the smaller earlier meta-analyses, we find strong evidence that 5-HTTLPR moderates the relationship between stress and depression in the studies published to date.
The principal function of the serotonin transporter is to remove serotonin from the synapse, returning it to the presynaptic neuron where the neurotransmitter can be degraded or re-released at a later time. A polymorphism in the promoter region of the serotonin transporter gene (5-HTTLPR) has been found to affect the transcription rate of the gene, with the short (s) allele transcriptionally less efficient that the alternate long (l) allele. In 2003, Caspi and colleagues examined the relationship between 5-HTTLPR, stress and depression using a prospective, longitudinal birth cohort and found that subjects carrying the less functional 5-HTTLPR s allele reported greater sensitivity to stress1.
This study has been cited over 2000 times in the scientific literature and generated a great deal of excitement and controversy around the potential of gene × environment interaction studies2. To date, there have been 55 follow-up studies, exploring whether 5-HTTLPR moderates the relationship between stress and depression, with some studies supporting the association between the 5-HTTLPR s allele and greater stress sensitivity and others not. Two recent meta-analyses have assessed a subset of these studies and concluded that there is no evidence supporting the presence of genetic moderation3, 4.
Since their publication, these meta-analyses have been criticized for only including a subset of the studies investigating the relationship between 5-HTTLPR stress and depression5–9. In fact, while 56 primary data studies have assessed whether 5-HTTLPR moderates the relationship between stress and depression, the Munafo and Risch meta-analyses included only 5 and 14 of those studies respectively10–48. Further, Uher and McGuffin have demonstrated that the larger, Risch meta-analysis included a significantly greater proportion of negative replication studies than positive replication studies8.
There are multiple reasons that the studies included in the meta-analyses were limited. First, the primary study data needed for traditional meta-analysis was often not available, either in the original publications or in follow-up email inquiries to study authors. For instance, Munafo and colleagues reported that 15 studies met criteria for inclusion in their meta-analysis. However, they were only able to obtain the primary study data needed for inclusion for five of those studies. There is no evidence that the studies that were able to be included in the meta-analyses were of higher “quality” than those not included.
Another reason why many studies were not included in the Risch and Munafo meta-analysis is that both meta-analyses focused exclusively on studies that explored an interaction between 5-HTTLPR and stressful life events (SLEs) in the development of depression. The original Caspi article, however, not only reported an interaction between 5-HTTLPR and SLEs, but also an interaction between 5-HTTLPR and childhood maltreatment stress. Nine studies have attempted to replicate this interaction with childhood maltreatment, but these studies were not included in the meta-analyses.
Some observers have noted that the SLE study design may have limited power to detect genetic moderation effects because they are susceptible to biases introduced by impaired recall of stressors by subjects and highly variable stressors between subjects9, 45. A newer class of studies has attempted to bypass these potential problems by focusing on specific populations that have experienced a substantial, specific stressor. These studies test whether 5-HTTLPR moderates the relationship between a specific stressor and depression. Eighteen studies have employed such specific stressor designs, but like the childhood maltreatment studies, these studies were excluded from the previous meta-analysis.
In this study, rather than focus on a limited of studies, we sought to perform a meta-analysis on the entire body of work assessing the relationship between 5-HTTLPR, stress and depression. Unfortunately, different types of studies have generally used different study designs to explore this question, rendering it very difficult to combine the studies into a single traditional meta-analysis. An approach useful in situations where equivalent raw data are not available across all studies, is to combine the studies at the level of significance tests 49. The Liptak-Stouffer Z-score method is a well-validated method for combining p-values across studies that has been utilized widely across genomics and biostatistics 50–56. In this study, we utilize the Liptak-Stouffer Z-score method to combine the results from studies investigating whether the 5-HTTLPR variant moderates the relationship between stress and depression.
Potential studies were identified from previous meta-analyses and review articles and through PubMed at the National Library of Medicine, using the search terms (depression OR depressed) AND (“serotonin transporter” OR 5-HTTLPR) AND (stress OR stressful OR maltreatment)3, 4, 9. We subsequently checked the reference sections of the identified publications and reviews found and contacted authors through email to identify additional studies in press or review. We considered all English language studies published by March 2010 assessing whether 5-HTTLPR moderates the relationship between either stressful life events, childhood maltreatment or specific stressors and depression. Two studies were excluded because their data was part of another, larger study included in the analysis14, 57. In total, data from 54 publications met inclusion criteria and were included in the analysis.
In addition to investigating all studies together, we also utilized a grouping method proposed in an earlier review to stratify studies by the type of stressor studied (Childhood Maltreatment, Specific Medical Conditions and Stressful Life Events) and assessed the presence of the association within each group9. When publications reported results for multiple types of stressors that matched different groups, we included the study in each relevant group1, 46, 58–61.
We evaluated the methodological quality of the included studies by applying an 11-item quality checklist, derived from the STREGA and STROBE checklists 62, 63. We extracted information relevant to methodological quality criteria (items 2–7, 10), and basic reporting standards (items 1, 8, 9, 11) from the Introduction, Methods, Results and Discussion sections of all included studies. Consistent with current guidelines, we did not weigh studies by quality scores or exclude studies with low quality studies 64. Instead we report the quality data extracted, so that it is available for readers to evaluate (Supplemental Table 1) 64, 65. Further, to assess whether our results were influenced by studies rated as lower quality through this measure, we repeated our overall meta-analysis with only studies with a quality score above the median 66.
Two investigators (KK and SS) independently extracted the relevant p value from each study. There were no cases of disagreement between the two investigators. When several p values were provided (due to the use of several depression scales or separate p values for different subsets of samples) we used a weighted mean p value for our analyses. For studies with non-significant results that did not provide exact probabilities, a p value of 1 (no association in either direction) was assumed. When an article reported analyses that matched different groups of our study, we incorporated the mean of the p value of each group into the overall analysis.
The Liptak-Stouffer Z-score method was utilized to combine studies at the level of significance tests, weighted by study sample size. First, all extracted p-values were converted to one-tailed p-values, with p-values below 0.5 corresponding to greater s allele stress sensitivity and p-values above 0.5 corresponding to greater l allele stress sensitivity.
Next, these p-values were converted to Z-scores using a standard normal curve such that p-values below 0.5 were assigned positive Z-scores and p-values above 0.5 were assigned negative Z-scores. Subsequently these Z-scores were combined by calculating , where the weighting factors wi corresponds to the individual sample sizes, k corresponds to the number of total studies and Zi corresponds to the individual study Z-scores. The outcome of this test, Zw, follows a standard normal distribution and the corresponding probability can be obtained from a standard normal distribution table. We applied this procedure to the overall sample as well as to each of the three study groups.
To assess whether our results were substantially influenced by the presence of any individual study, we conducted a sensitivity analysis by systematically removing each study and recalculating the significance of the result. Further, to compare our method of combining studies at the significance test level with the method of combining studies at the raw data level utilized in the previous meta-analyses, we performed an analysis with only the studies included in the previous meta-analyses4.
In order to account for the possibility that results of the meta-analysis were affected by publication bias, we calculated the number of unpublished studies that would have to exist to change the outcome of the Liptak-Stouffer test from significant to non-significant (Fail-safe N)67. The ratio between the Fail-safe N and the number of studies actually published gives an estimate for the likelihood that the significant meta-analytical result is due to publication bias.
Our initial search identified 148 publications. Out of these studies, we identified 54 studies that included 40,749 subjects meeting criteria for inclusion (Table 1). We found strong evidence that 5-HTTLPR moderates the relationship between stress and depression, with the s allele associated with an increased risk of developing depression under stress (p=0.00002). The significance of the result was robust to sensitivity analysis, with the overall p values remaining significant when each study was individually removed from the analysis (1.0E-6<p<0.00016). In addition, when we restricted our analysis to those studies with a study “quality” score above the median, the p value remained highly significant (3.2E-10, N=14).
In examining the three groups of stress studies separately, we found strong evidence for an association between the s allele and increased stress sensitivity in studies of childhood maltreatment (p=0.00007), in studies of specific medical conditions (p=0.0004), but only marginal evidence in the studies of stressful life events (p=0.033) (Table 2, ,33 and and44 respectively). The removal of individual studies did not lead to changes in the significance of the outcome in studies of childhood maltreatment (7.4E-6<p<0.00014) or specific medical conditions (0.00017<p<0.0068). However, because the genetic effect in the set of stressful life events was barely below the significance threshold (p=0.033), the result was no longer significant after the exclusion of any one of several studies 1, 32, 35, 37, 68 (0.013<p<0.62).
When we restricted our analysis to the studies included in the two previous meta-analyses, we found no evidence of an association between 5-HTTLPR and stress sensitivity (Munafo studies p=0.16; Risch studies p=0.11).
One criticism of meta-analyses is that positive studies may be more likely to be published than negative studies and this sort of publication bias can create false positive results. We thus determine how many unpublished studies would need to exist to make the result of our overall analysis non-significant (p=0.05). We found that 729 unpublished or undiscovered studies with an average sample size (N = 755) and a non-significant result (p = 0.5) would need to exist. This corresponds to a fail-safe ratio of 14 studies not included in this meta-analysis for every included study.
We found strong evidence that a serotonin transporter promoter polymorphism (5-HTTLPR) moderates the relationship between stress and depression, with the less functional short (s) allele associated with increased stress sensitivity. Our results differ from the results of the two other meta-analyses that have explored this specific association. To test whether this difference in results was due to the expanded set of studies that we included or the different meta-analytic technique utilized, we applied our meta-analytic technique to the sets of studies used in the previous meta-analyses. With these limited set of studies, our meta-analytic technique produced the same non-significant results as the previous meta-analyses, suggesting that the difference in results between meta-analyses was due to the different set of included studies.
The results of our secondary meta-analyses, where we stratified studies by stressor type, provide insight into how the inclusion of studies missing from previous studies resulted in an overall highly significant result in our meta-analysis. Both previous meta-analyses focused exclusively on stressful life events and reported no evidence that 5-HTTLPR moderates the relationship between SLEs and depression. Here, we were able to include 11 additional SLE studies, most of which were published too recently for inclusion in the previous meta-analyses. Still, we found only marginal evidence that 5-HTTLPR moderates the relationship between stressful life events and depression6. In contrast, we found robust evidence that 5-HTTLPR moderates the relationship between both childhood maltreatment and specific stressors and depression.
One important variable that may help to account for the different results in the different stressor groups is the variation in methods between the studies within each group69. Within the childhood maltreatment and specific stressors groups, the methodological details of the primary studies were generally similar. In contrast, there is marked variation in methods between SLE studies. First the studies vary substantially in what was considered a stressful life event. In addition, the method through which stressful life events were measured varied substantially between studies in this group. Some studies measured stress through one-time self-report life events checklists while others employed repeated in-person interviews and life history calendars1, 70. It is noteworthy that almost all of the studies that failed to identify an effect of genetic moderation used self-report checklists (Table 1). In addition, some studies asked subjects about SLEs and depressive episodes that occurred decades earlier while others assessed SLEs and depressive episodes soon after they occurred30, 71. As a result, the extent to which recall bias affected the findings of these studies varied substantially between the studies. Given this marked variation in methods between SLE studies, it is not surprising that the results between studies have also varied. In contrast, the methods of childhood maltreatment and specific stressors have been more uniform and the results of the studies have been more consistent.
An additional reason for the difference between the meta-analyses of the different stressor subgroups may be the nature of the stressors studied. Most of the specific stressor studies focused on chronic stressors while the SLE studies focused on acute stressful life events. Interestingly, three studies have explicitly looked at both acute and chronic stressors in their cohorts and all three have found that the 5-HTTLPR moderating effects were stronger for chronic stressors 10, 16, 45. Future primary studies that are able to systematically test genetic moderation effects on different types of stressors will be valuable in furthering our understanding of the specific characteristics of stressors that are moderated by 5-HTTLPR and other genetic loci important in stress response.
One criticism of the 5-HTTLPR-stress studies published to date is that investigators often performed multiple tests, using different subsets of their population or different stress or depression measures, but focus their paper on the tests that produced the most significant results and present their overall findings as a confirmation of the original hypothesis4, 72. For instance, different studies have found evidence of genetic moderation only in the female subset of their sample, only in the subset of their sample that was evaluated through an in-depth clinical interview or only when the analysis was restricted to chronic stressors10, 73, 74. As we discuss above, some of the variation in results with different population sub-samples or depression and stress measures may represent true and important heterogeneity in the 5-HTTLPR moderation effect. However another possible explanation for the variation in results within the same study is that some of these secondary findings are actually false positive results that resulted from uncorrected multiple testing. To guard against false positive from the primary studies causing a false positive in our meta-analysis, we did not rely on the statistical tests highlighted by authors. Instead, we calculated a weighted average of p values of the tests that were performed in a given study. When authors only reported the significance results for a subset of these tests, we assumed that p = 1 for the unreported tests. The fact that we confirmed the previous non-significant results when we applied our meta-analytic technique to the sets of studies included in the previous meta-analyses suggests that statistical bias from primary studies did not unduly affect our results.
While this meta-analysis focused specifically on observational studies specifically assessing whether 5-HTTLPR moderates the relationship between stress and depression, the results we found are consistent with a broad range of studies exploring the relationship between functional serotonin transporter genetic variation and stress in different ways. Experimental neuroscience studies have found consistent evidence that 5-HTTLPR s allele carriers demonstrate a more pronounced amygdala and HPA axis response to affective or threatening stimuli 75–77. In addition, non-human primates studies that have found increased stress sensitivity among individuals with a low functioning serotonin transporter allele78, 79. Together, these lines of evidence provide clear and converging evidence that 5-HTTLPR plays a role in moderating the response to stress. It is also clear from these studies however, that this variant explains only a small proportion of the genetic variance relevant to stress response. The successes and failures of the studies exploring the 5-HTTLPR variant included in this analysis should guide our future work as we try to develop a broader understanding of the genetic architecture that moderates the relationship between stress and depression.
Funding/Support: This work was supported by NIH KL2 grant number: UL1RR024986 (SS), the University of Michigan Depression Center (SS) and Studienstiftung des deutschen Volkes (KK).
Role of the Sponsor: The funding agencies played no role in the design and conduct of the study; collection management, analysis, or interpretation of the data; and preparation, review, or approval of the manuscript.
Author Contributions: Dr. Sen and Ms. Karg had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.Study concept and design: Sen, Karg, Burmeister
Acquisition of data: Karg, Sen
Analysis and interpretation of data: Sen, Karg, Shedden.
Critical revision of the manuscript for important intellectual content: Sen, Burmeister, Karg, Shedden.
Statistical analysis: Shedden, Karg, Sen.
Obtained funding: Sen, Burmeister.
Administrative, technical, or material support: Sen
Study supervision: Sen, Burmeister.
Financial Disclosures: None reported.
Additional Contributions: Brady West, MA, Center for Statistical Consultation and Research, University of Michigan, Ann Arbor, MI, helped with the statistical analyses for this article.
Competing Interests: Karg, Burmeister, Shedden and Sen declare no competing interests and therefore have nothing to declare.