Financial strain and stressful social environment drive depressive symptoms, while FKBP5 variant intensifies the effect, in African Americans living in Tallahassee

The World Health Organization estimates that almost 300 million people suffer from depression worldwide. Depression is the most common mental health disorder and shows racial disparities in disease prevalence, age of onset, severity of symptoms, frequency of diagnosis, and treatment utilization across the United States. Since depression has both social and genetic risk factors, we propose a conceptual model wherein social stressors are primary risk factors for depression, but genetic variants increase or decrease individual susceptibility to the effects of the social stressors. Our research strategy incorporates both social and genetic data to investigate variation in symptoms of depression (CES-D scores). We collected data on financial strain (difficulty paying bills) and personal social networks (a model of an individual’s social environment), and we genotyped genetic variants in five genes involved in stress reactivity (HTR1a, BDNF, GNB3, SLC6A4, and FKBP5) in 135 African Americans residing in Tallahassee, Florida. We found that high financial strain and a high percentage of people in one’s social network who are a source of stress or worry were significantly associated with higher CES-D scores and explained more variation in CES-D scores than did genetic factors. Only one genetic variant (rs1360780 in FKBP5) was significantly associated with CES-D scores and only when the social stressors were included in the model. Interestingly, the effect of FKPB5 appeared to be strongest in individuals with high financial strain such that participants with a T allele at rs1360780 in FKBP5 and high financial strain had the highest mean CES-D scores in our study population. These results suggest that material disadvantage and a stressful social environment increases the risk of depression, but that individual-level genetic variation may increase susceptibility to the adverse health consequences of social stressors.

8 138 interview, to participate in the social network interview, and for their data to be used in other 139 health-related studies besides blood pressure and cardiovascular disease. Participants were able 140 to consent separately for each of these aspects of the study and only participants who consented 141 to all four aspects were included in the study. Copies of the signed informed consents forms were 142 scanned and uploaded to a secure online site accessible only by CCG and CJM. All data 143 analyzed in this study are available at https://data.mendeley.com/datasets/3f46kg3m55/1. 151 Postal addresses were then selected randomly from groups of census blocks. Participants were 152 able to participate if they were at least eighteen years old. The initial sample size of this study 153 was n=185. Twenty participants did not consent to give saliva samples and 19 participants did 154 not agree to participate in the second interview, which collected the social network data. After 155 accounting for DNA samples that did not type for all genetic variants and removing participants 156 with incomplete data, the final sample size for this study was n=135 participants. 9 161 tool that evaluates symptoms of depression in an individual [44]. The CES-D is a 20-item 162 questionnaire with a maximum score of 60. The American Psychological Association (APA) 163 classifies a score ≥ 16, the standard cutoff point, as an indication of the potential for clinical 164 depression [45]. The potential responses to the CES-D questions are "rarely or none of the time 165 (less than 1 day), "some or a little of the time (1-2 days)", "occasionally or a moderate amount of 166 time (3-4 days)", and "most or all of the time (5-7 days). In our study, the CES-D form was 167 modified, by separating one of the existing responses ("rarely or none of the time") into two 168 responses ("rarely" and "none of the time"), in order to capture a wider range of responses, 169 which then increased the maximum possible score from 60 to 80. When using the validated 170 cutoff of ≥ 16 for risk of clinical depression, the responses were translated back to their original 171 form, i.e. "rarely" and "none of the time" were grouped into the original response of "rarely or 172 none of the time".

173
In keeping with CBPR principles, interviews were conducted by members of the 238 The mean CES-D score for all study participants was 12.7. Average financial strain was 239 approximately 1.5, indicating there were relatively equal numbers of participants who said that 240 bill paying was "Not very difficult" or "Not at all difficult" (coded as 2) and those who said that 241 bill paying was "Very difficult" or "Somewhat difficult" (coded as 1). As a measure of social 242 network stress, study participants said that 5-6% of their social network members caused them 243 stress or worry "often" or "always". There was no significant difference in mean CES-D scores 244 between males and females (14.5 vs 11.8, Pr(>F) =0.09). The genotype counts for the tested 245 SNPs (rs140701, rs1360780, rs4922793, rs5443, and rs6295) are detailed in Table 2. 253 Four models were tested in order to identify the optimal model that explained the most variation 254 in CES-D score (Table 3). The model with the highest adjusted R 2 and lowest AIC score was 255 chosen as the optimal model, which was the fourth model that contained both genetic and social 256 data (Table 3). The optimal model accounted for 19% of variation (adjusted R 2 ) in CES-D scores 257 in comparison to the genetic model and social models, which explained 2% and 16% of CES-D 258 score variation, respectively. The variables included in the optimal model were financial strain, 259 social network stress, and SNPs in SLC6A4 and FKBP5. Of the variables included in the optimal 260 model, all except the SLC6A4 SNP were significantly associated with CES-D score (Table 3).
15 261 In the optimal model, high financial strain ("very difficult to pay bills" plus "somewhat 266 difficult to pay bills") was associated with higher CES-D scores (q-value = 0.05) and participants 267 who reported a larger percentage of alters in their social network who were "often" or "always" a 268 source or stress or worry also had higher CES-D scores than their peers (q-value ≤ 0.001). Both 269 of these variables were significantly associated with CES-D score in the social model and the 270 optimal model (Table 3). Social network stress was found to have the largest beta value and the 271 smallest p-value, and therefore explained the most variation in depressive symptoms in our 272 sample.

273
Of the genetic data, only one SNP (rs1360780 in FKBP5) was significantly associated  (Table 3).

319
Poverty is generally associated with increased prevalence of poor physical and mental 320 health, although the best measure of economic disadvantage and the mechanism by which such 321 disadvantage acts are not clear. In general, income appears to be less impactful than the ability to 322 pay bills, as is captured in Zimmerman and Katon's measure of financial strain that is used in 323 this study and has been causally related to symptoms of depression [25]. Specifically, we found 324 that participants who reported feeling that it was "very difficult" or "somewhat difficult" to pay 19 325 bills every month had significantly higher CES-D scores than those who responded otherwise.

386
There are several limitations to note in our study. We were able to detect important 387 associations between financial stress, social network stress, and the rs1360780 variant at FKBP5 388 with symptoms of depression; however, we were limited by a relatively small sample population 389 that could have impacted our ability to detect more complex associations or interactions.
390 Furthermore, we lacked data on participant history of being diagnosed with or treated for 391 depression, which meant that we could not account for history of depression in our analyses.
392 Future studies should test larger sample sizes for the impact of social and material risk factors on