The Interplay of Genetics, Behavior, and Pain with Depressive Symptoms in the Elderly

Abstract

Purpose of Study:

About 25% of older adults suffer from depressive symptoms. Commonly studied candidate genes associated with depression include those that influence serotonin (SLC6A4), dopamine (COMT), or neuroplasticity (BDNF, NTRK3). However, the majority of candidate gene studies do not consider the interplay of genetics, demographic, clinical, and behavioral factors and how they jointly contribute to depressive symptoms among older adults. The purpose of this study was to gain a more comprehensive understanding of depressive symptoms among older adults.

Design and methods:

In this descriptive study, demographic, behavioral, and clinical characteristics (age, gender, comorbidities, volunteering, physical activity, pain, and fear of falling) were obtained via interview of 114 residents in a continuing care retirement community. Peripheral whole blood was collected for DNA extraction. We examined common single nucleotide polymorphisms (SNPs) in the aforementioned genes using path analyses.

Results:

SNPs in the NTRK3 gene, pain, physical activity, and fear of falling were directly associated with depressive symptoms in older adults. Those who had polymorphisms in the NTRK3 gene, pain, fear of falling, and were less physically active were more likely to exhibit depressive symptoms. None of the SNPs in SLC6A4, COMT, or BDNF genes were significantly associated with depressive symptoms.

Implications:

Our use of a path analysis to examine a biopsychosocial model of depressive symptoms provided the opportunity to describe a comprehensive clinical picture of older adults at risk for depressive symptoms. Thus, interventions could be implemented to identify older adults at risk for depressive symptoms.

About 25% of adults aged 65 and older suffer from clinically significant depressive symptoms that do not meet criteria for major depressive disorder (Forlani et al., 2013; Meeks, Vahia, Lavretsky, Kulkarni, & Jeste, 2011). The depressive symptoms associated with so-called “minor depression” cause substantial emotional suffering, increase the prevalence of illness and disability, decrease quality of life, escalate healthcare utilization, and cost expenditure by as much as 50%, increase suicidal ideation, and increase mortality (Lyness, Chapman, McGriff, Drayer, & Duberstein, 2009; Lyness et al., 2007; Meeks et al., 2011). In addition, depressive symptoms increase the risk of developing major depressive disorder (Lyness et al., 2009; Meeks et al., 2011).

Biopsychosocial Conceptualization of Depressive Symptoms in Older Adults

The Biopsychosocial model has been used as a way to understand many factors that influence depressive symptoms among older adults. This model suggests that biological, psychological, and social factors are all interlinked, and each plays a role in the development of depressive symptoms (Engel, 1980). Biologic factors include availability of monoamine neurotransmitters, such as serotonin, dopamine, and norepinephrine in the brain synapses and maintenance of healthy brain structures and functional pathways. Common psychological stressors experienced by older adults, such as chronic illnesses, pain, and fear of falling contribute to depressive symptoms. Social and behavioral factors such as lack of engagement in physical activity and/or meaningful activity (e.g., volunteering) also contribute to the development of depressive symptoms in late life.

Psychological and Behavioral Contributors to Depressive Symptoms in Older Adults

Common symptoms associated with aging including pain or fear of falling can have both direct and/or indirect associations with depressive symptoms. Pain and depressive symptoms are persistent, strongly directly associated, and, in combination, lead to a poor prognosis in older adults (Chou, 2007; Geerlings, Twisk, Beekman, Deeg, & van Tilburg, 2002). In addition, pain indirectly influences depressive symptoms as it decreases the likelihood that individuals will engage in behaviors known to alleviate depressive symptoms such as volunteering (Yanay & Yanay, 2008) or physical activity (Lim & Taylor, 2005). Fear of falling is also directly associated with depressive symptoms in older adults (Klinedinst & Resnick, 2014; van Haastregt, Zijlstra, van Rossum, van Eijk, & Kempen, 2008) and indirectly through reductions in physical activity (Lim & Taylor, 2005).

A sedentary lifestyle is associated with depressive symptoms (Owiti & Bhui, 2012; Vance, Wadley, Ball, Roenker, & Rizzo, 2005), and increased physical activity decreases depressive symptoms among older adults (Barbour & Blumenthal, 2005; Sjösten & Kivelä, 2006). Participation in meaningful activities, such as volunteering, decrease the likelihood of experiencing depressive symptoms among older adults (Kim & Pai, 2010).

Genetics and Depressive Symptoms

Genetics plays an integral role in the development of depressive symptoms. Twin studies demonstrated that 37–50% of the etiology of mood disorders can be explained by genetics; environmental factors are assumed to explain the rest (Sadock, Kaplan, & Sadock, 2007; Sullivan, Neale, & Kendler, 2000). While a single “depression gene” has yet to be identified, commonly studied candidate genes associated with depression are those known to influence neurotransmitters or neuroplasticity. Often single nucleotide polymorphisms (SNPs) are examined in candidate gene studies. SNPs are alternative nucleotides (A, T, C, or G) than usually occurs in a population, at a particular position in the DNA gene sequence.

Mounting evidence links depression with the Solute Carrier Family 6 (Neurotransmitter Transporter), Member 4 (SLC6A4) gene which influences serotonin (Karg, Burmeister, Shedden, & Sen, 2011). The 5-Hydroxytryptamine (Serotonin) Transporter (5-HTTLPR) gene region within SLC6A4 contains repetitive sections of nucleotides with variation in the numbers of repeats resulting in (s) short allele or (l) long alleles. Interactions between the experience of negative life events and the s allele of 5-HTTLPR have been associated with a higher risk of depressive symptoms (Beaver, Vaughn, Wright, & Delisi, 2012; Caspi et al., 2003; Zannas, McQuoid, Steffens, Chrousos, & Taylor, 2012), but meta-analytic evidence remains inconclusive (Risch et al., 2009). Wendland and colleagues (2008) also found a significant relationship on a novel SNP: rs25532 and obsessive–compulsive disorder among Caucasian adults.

Catechol-O-methyltransferase (COMT), an enzyme responsible for the degradation of dopamine in neural synapses is encoded for by the COMT gene (Antypa, Drago, & Serretti, 2013; Du et al., 2014). A common SNP, rs4680, known as the Val158Met polymorphism encodes a substitution of methionine (Met) for valine (Val) at codon 158. Individuals who are homozygous for the Val allele likely have lower synaptic levels of dopamine (Antypa et al., 2013), a known risk factor for depression. However, associations between the Val158Met polymorphism and depression have been inconclusive (Baekken, Skorpen, Stordal, Zwart, & Hagen, 2008; Illi et al., 2010; Wray et al., 2008).

Brain-derived neurotrophic factor (BDNF) is a growth factor that supports the survival and differentiation of existing neurons (Hyman et al., 1991). Depressed adults often have lower serum BDNF levels than nondepressed individuals. Antidepressant treatment increases serum BDNF levels in responders, which corresponds to remission of depressive symptoms (Castrén & Rantamäki, 2010). The Met allele of the Val66Met polymorphism, rs6265, in the BDNF gene has been shown to moderate the relationship between stress and depression (Hosang, Shiles, Tansey, McGuffin, & Uher, 2014), but evidence of a direct relationship with depression is inconclusive (Chen et al., 2008).

Another related gene involved in neurotrophin signaling is the neurotrophic tyrosine kinase receptor-3 (NTRK3) gene, the putative receptor for neurotrophin-3 (NT-3). NT-3 helps to support the survival and differentiation of existing neurons. Low expression of the NTRK3 gene is associated with decreased synaptic plasticity and depression in studies of young adults (Farhang et al., 2014; Feng et al., 2008; Verma et al., 2008). Commonly studied NTRK3 polymorphisms associated with depression and other psychiatric disorders include rs1369430, rs1435403, rs3784441, rs7180942, rs2059588, rs1110306, and rs3784406 (Feng et al., 2008; Mercader et al., 2008; Verma et al., 2008).

Genes such as SLC6A4, COMT, BDNF, and NTRK3 are also associated with perceived pain and/or willingness to engage in physical or meaningful activity (Caldwell Hooper, Bryan, & Hagger, 2014; Sah, Ossipo, & Porreca, 2003; Vogel et al., 2003; Vossen et al., 2010). Therefore, these genes may indirectly influence depression in older adults through pain, or engagement in physical activity or volunteering.

The majority of the genetic research on depression has been conducted with young adults and few studies, if any, consider the additional psychosocial factors that may contribute to depressive symptoms among older adults. The purpose of this study was to gain a more comprehensive understanding of depressive symptoms among older adults. Using the Biopsychosocial model (Figure 1), we hypothesized that SNPs in each SLC6A4, COMT, BDNF, and NTRK3 genes when combined with pain, fear of falling, volunteering, physical activity would directly and indirectly be associated with depressive symptoms among older adults.

Figure 1.

Hypothesized biopsychosocial model of depressive symptoms in older adult.

Methods

Study Design

This was a cross-sectional, descriptive study based on data collected during a single face-to-face interview and blood draw conducted with older adults living in a continuing care retirement community (CCRC). The CCRC is a single building unit containing independent living apartments, assisted living apartments, and a nursing home. Apartments in both independent living and assisted living are equipped with kitchens but residents also have access to congregate dining facilities where meals are prepared for them. All residents have access to housekeeping services. Residents were eligible to participate if they resided in the independent or assisted living apartments of the CCRC, could describe the expectations of the study, and scored at least two out of three on the three item recall of the Mini-Cog (Borson, Scanlan, Brush, Vitaliano, & Dokmak, 2000). The study was approved by a University-based Institutional Review Board. All participants gave written informed consent.

Sample

There were 244 residents living in the CCRC during the year in which the interviews occurred. A convenience sample of 149 (61%) residents consented to participate in the study. Thirty-one residents (13%) refused to participate and the remaining 64 (26%) were not reachable during the study period (e.g., out of town, unavailable). Among those consented, two were not eligible to participate due to cognitive issues leaving 147 participants who completed the interview. Of the residents who participated in the interview, 121 (82% of participants) consented to the genetics portion of the study. The remaining individuals either refused blood or we were not able to schedule the blood draw at a time that was convenient for the participant. The DNA from five individuals was not of sufficient quantity for analysis, further reducing the sample size to 116. The majority of the sample (98%) was Caucasian of European descent. In order to control for racial and ethnic differences in genotype, residents who did not identify as Caucasian (n = 2) where excluded in analyses, leaving a final sample size of 114.

Questionnaires

Depressive symptoms were measured using the three-item Useful Depression Screening Tool (UDST) due to its brevity and specific design in screening for depressive symptoms among older adults in congregate housing settings (Klinedinst & Resnick). The first two items in the UDST are those used in the Patient Health Questionnaire (PHQ-2) (Kroenke, Spitzer, & Williams, 2003) and inquire (a) how often participants felt down, depressed or hopeless; and (b) how often the participant experienced little interest or pleasure in doing things. Responses included: 0 = not at all; 1 = several days; 2 = more than half the days; 3 = nearly every day. The third question, “How often do you feel useful to your family and friends?” addresses whether or not the individual feels useful (Gruenewald, Karlamangla, Greendale, Singer, & Seeman, 2007) and responses included: 3=never, 2=rarely, 1=sometimes, or 0=frequently. A score of ≥ 4 is indicative of a need for further evaluation for depression. The UDST has established reliability and validity for use with older adults in congregate living settings (Klinedinst & Resnick).

Volunteerism was measured by asking participants to respond (Yes/No) to 16 items about different types of volunteer work outside the CCRC setting. In addition, participants indicated whether they volunteered within the CCRC and in which activities. The individual was considered to volunteer if he or she engaged in at least one volunteer activity either inside or outside of the CCRC (Resnick, Klinedinst, Dorsey, & Holtzman, 2013).

Physical activity was measured using the Yale Physical Activity Survey (YPAS) (Dipietro, Caspersen, Ostfeld, & Nadel, 1993). The YPAS includes five categories of physical activity: housework, yard work, caretaking, exercise and recreational activities performed during a typical week. The YPAS includes a wide range of lower intensity activities commonly performed by older adults. Prior use provided evidence of 2 week repeatability (r = 0.63, p < .001), and the YPAS has been validated against several physiological variables that are indicative of habitual activity (Dipietro et al., 1993; Pescatello, DiPietro, Fargo, Ostfeld, & Nadel, 1994) and other physical activity surveys (Pescatello et al., 1994).

Demographic variables including age and number of chronic illnesses were collected from the participants’ medical record. Participants were asked their gender. Fear of falling was evaluated by asking the participant to rate her fear of falling on a scale of 0–4 (Resnick, 1998) and current overall pain was evaluated using the 0–10 Numeric Rating Scale (Herr & Mobily, 1991, 1993; Herr, Spratt, Mobily, & Richardson, 2004). Higher scores indicate more fear of falling and more perceived pain, respectively.

DNA Isolation and SNP Analysis

Between 3 and 7.5ml of peripheral whole blood was drawn into a BD vacutainer® ACD blood collection tube (BD, Ref# 364606). The sample was transported on ice to the Translational Core Lab (University of Maryland School of Medicine, Baltimore MD) and DNA was either isolated immediately or frozen at −20 °C within 1hr of blood draw. If blood was frozen, DNA was isolated within 3 weeks. All DNA was isolated using the QIAamp® DNA blood maxi kit (Qiagen, Valencia, CA; Cat#51194) following the manufacturer protocol. DNA was stored at −80 °C until ready for study.

Ten SNPs in four candidate genes were identified a priori from a literature review (Table 1) and genotyped as part of a 64-plex genotyping assay created using online design software from Life Technologies (Carlsbad, CA). Eighteen of the 64 assays were custom assays, designed by the software; the others were off-the-shelf Taqman assays. Assays were run on the QuantStudio 12K Flex with the OpenArray block, following published protocols (Applied Biosystems, 2012).

Table 1.

Characteristics of Candidate Genes and Associated SNPs in Sample

Gene	SNP	Chr	Position	Function	Alleles	N (%)
NTRK3	rs1435403	15q25	88437069	Intron	GG	55 (47)
					GA	45 (12)
					AA	12 (10)
NTRK3	rs1110306	15q25	88680100	Intron	AA	24 (21)
					GA	52 (45)
					GG	28 (24)
NTRK3	rs1369430	15q25	88430769	Intron	AA	42 (36)
					AG	56 (48)
					GG	15 (13)
NTRK3	rs3784441	15q25	88451765	Intron	GG	58 (50)
					GA	46 (40)
					AA	12 (10)
NTRK3	rs7180942	15q25	88674576	Intron	TT	29 (25)
					GT	60 (52)
					GG	24 (21)
NTRK3	rs2059588	15q25	88678680	Intron	TT	29 (25)
					GT	55 (47)
					GG	25 (22)
NTRK3	rs3784406	15q25	88685330	Intron	CC	34 (29)
					CT	58 (50)
					TT	21 (18)
BDNF	rs6265	11p13	27679916	Exonic - missense	GG	70 (60)
					GA	38 (33)
					AA	1 (1)
SLC6A4	rs25532	17q11.2	28564170	Upstream of gene	CC	82 (71)
					CT	26 (22)
					TT	4 (3)
COMT	rs4680	22q11.21	19951271	Exonic - missense	GG	30 (26)
					GA	54 (47)
					AA	30 (26)

Notes: Chr = chromosome; SNP = Single nucleotide polymorphism.

Data Analysis

Descriptive analyses were conducted to describe the sample characteristics. Correlations were performed to determine significant associations between the predictors and the outcome variable of depressive symptoms to develop a parsimonious model. There were no significant relationships between gender or comorbidities and any of the endogenous factors or outcomes so these were not included in the model. As shown in Figure 1, a model of the factors influencing depressive symptoms was tested using path analysis and the AMOS statistical program. The sample covariance matrix was used as input and a maximum likelihood solution sought. The chi-square statistic, the normed fit index (NFI), and Steigers Root Mean Square Error of Approximation (RMSEA) were used to estimate model fit. The larger the probability associated with the chi-square, the better the fit of the model to the data (Bollen, 1989; Loehlin, 1998). The chi-square statistic is sample size dependent, therefore we considered chi-square divided by degrees of freedom (df) and used a ratio of 3 or less to be indicative of a good fit of the model to the data (Bollen, 1989; Loehlin, 1998). The NFI tests the hypothesized model against a baseline model and should be 1.0 if there is perfect model fit. The NFI is “normed” so that the values cannot be below 0 or above 1. The RMSEA is a population-based index and consequently is insensitive to sample size. An RMSEA of < 0.10 is considered good, and <0.05 is very good. Path significance was based on the Critical Ratio (CR), which is the parameter estimate divided by the standard error. A CR > 2 in absolute value was considered significant (Arbuckle, 2006).

Each SNP was tested in the model independently. Initially an additive SNP model was tested to establish if the increase in a particular allele contributed to depressive symptoms, pain, physical activity, and volunteering. In this model, the SNP was coded 0 for wild type, 1 for heterozygous for a minor allele, and 2 for homozygous for two minor alleles. In addition, each SNP was tested in a recessive model where the SNP was coded 1 as homozygous for the polymorphic allele or 0 not homozygous for the polymorphic allele. A total of 20 models were tested. Significance for path estimates was set at p ≤ .05.

Because several SNPs in the NTRK3 gene were significant in our model and were in high linkage disequilibrium, common haplotypes in this candidate gene were also tested using the same path analysis as described above. Haplotype phase was estimated using SAS genetics version 9.3 using the EM algorithm (Carey, NC).

Results

Sample characteristics for major study variables are displayed in Table 2. Participants were mostly female (75%), 87.0 (SD = 6) years old (range 71–103 years), and had between three and four comorbidities. They had low levels of pain with a mean score of 3.59 (SD = 3.33) and low levels of depressive symptoms (mean score 1.80, SD = 1.72). However, 14.1% (n = 16) of the sample scored ≥ 4 on the UDST indicating a high risk for depression.

Table 2.

Sample Characteristics for Major Study Variables

Characteristic	Mean (SD)	N (%)
Age	86.40 (6.31)
Gender
Male		29 (25)
Female		87 (75)
Comorbidities	3.28 (1.98)
Pain	3.53 (3.00)
Fear of Falling	1.51 (1.29)
Volunteering
Yes		58 (50)
No		58 (50)
Physical activity (min/week)	181.96 (144.73)
Depressive symptoms Scored ≥ 4	1.77 (1.72)	16 (14)

Each of the 10 SNPs conformed to the expectations of Hardy Weinberg equilibrium and had sufficiently high call rate (>90%) to be included in the analysis. Of the 20 SNP models tested (additive and recessive for each SNP), four models contained SNPs that were directly related to depressive symptoms. Path estimates for these models and associated significance of these paths are displayed in Table 3. Figure 2 shows the significant paths for each of the models. Age and physical activity directly influenced volunteer behavior such that those who were older and spent more time in physical activity were more likely to volunteer. Fear of falling, physical activity, pain and four SNPs from NTRK3 gene: rs7180942, rs2059588, rs1110306, and rs3784406 were directly associated with depression. Those who had more fear of falling, pain, engaged in less physical activity and had a higher risk allele for each of three the SNPs (rs2059588-T allele, rs1110306-G allele, and rs3784406- T allele) or were homozygous for either allele in rs7180942 (heterozygosis of rs7180942 was protective) were more likely to have depressive symptoms. Volunteering was directly associated with more physical activity. Both age and volunteering indirectly influenced depressive symptoms through physical activity. Taken together all variables accounted for 28–30% of the variance in depressive symptoms depending on the specific SNP included in the model. There was a fair fit of the model to the data across all models with a χ²/df ratio that was between 4 and 5; an NFI of .86, and RMSEA between .14 and .15.

Table 3.

Path Estimates and Significance for Models with Significant Genetic Association with Depressive Symptoms

Model Pathway			Model 1 rs7180942 (T)		Model 2 rs2059588 (T)		Model 3 rs1110306 (G)		Model 4 rs3784406 (T)		Model 5 GGAC
			Path estimate	p	Path estimate	p	Path estimate	p	Path estimate	p	Path estimate	p
Pain	<---	Age	−0.03	.70	−0.03	.69	−0.03	.71	−0.03	.75	−0.03	.69
Pain	<---	SNP/haplotype	0.06	.52	0.08	.40	−0.03	.72	−0.07	.46	.04	.64
Volunteer	<---	Age	−0.32	<.001	−0.32	<.001	−0.32	<.001	−0.32	<.001	−0.32	<.001
Volunteer	<---	Fear of falling	−0.00	.98	−0.00	.98	−0.00	.97	−0.00	.97	−0.00	.98
Volunteer	<---	Pain	−0.07	.40	−0.07	.40	−0.07	.40	−0.06	.45	−0.07	.39
Volunteer	<---	SNP/haplotype	−0.02	.78	−0.01	.94	0.02	.82	0.11	.23	0.00	1
Physical Activity	<---	Age	−0.25	.001	−0.26	.001	−0.25	.001	−0.27	<.001	−0.26	.001
Physical Activity	<---	Fear of Falling	0.03	.73	0.03	.72	0.03	.73	0.03	.74	0.03	.72
Physical Activity	<---	Volunteer	0.28	<.001	0.28	<.001	0.28	<.001	0.26	<.001	0.28	<.001
Physical Activity	<---	Pain	−0.03	.70	−0.03	.70	−0.03	.73	−0.02	.80	−0.03	.70
Physical Activity	<---	SNP/haplotype	0.04	.68	0.03	.74	0.00	.97	0.13	.13	0.05	.54
Depressive Symptoms	<---	Fear of falling	0.26	<.001	0.26	<.001	0.25	<.001	0.26	<.001	0.26	<.001
Depressive Symptoms	<---	Age	0.14	.07	0.15	.06	0.13	.08	0.11	.17	0.15	.05
Depressive Symptoms	<---	Pain	0.16	.02	0.17	.02	0.16	.02	0.17	.02	0.16	.02
Depressive Symptoms	<---	SNP/haplotype	0.30	<.001	0.22	<.01	0.29	<.001	0.23	<.01	−0.17	.04
Depressive Symptoms	<---	Volunteer	−0.09	.23	−0.09	.25	−0.09	.23	−0.11	.18	−0.09	.25
Depressive Symptoms	<---	Physical activity	−0.28	<.001	−0.28	<.001	−0.29	<.001	−0.32	<.001	−0.28	<.001

Figure 2.

Biopsychosocial model of depressive symptoms in older adults—significant paths only.

There were a total of three common haplotypes from these four SNPs (rs7180942, rs2059588, rs1110306, rs3784406) with frequencies greater than 1% in our population (indicated in block 2 of Figure 3). These were GGAC, TTGT, and TTGC with observed frequencies of 46, 43 and 7%, respectively. Each of these haplotypes was then entered in the model coded based on if a participant had 0, 1, or 2 copies of the haplotype. Only the GGAC haplotype was directly associated with depressive symptoms. Taken together, fewer copies of the GGAC haplotype, pain, fear of falling and less physical activity explained 28% of the variance in depressive symptoms.

Figure 3.

Linkage disequilibrium plot for NTRK3 SNPs.

Discussion

The results of this study provide continued support for the biopsychosocial framework of depressive symptoms in older adults. In addition to finding that the well-established clinical factors (pain, fear of falling, and engagement in physical activity) were associated with depressive symptoms, several intronic variants in the NTRK3 gene significantly influenced depressive symptoms among older adults in this sample.

The genetic findings from this study support prior research describing a relationship between the NTRK3 gene and depression. Verma and colleagues (2008) initially argued for the biologic plausibility for polymorphic variation in the NTRK3 gene as part of the etiology for major depressive disorder, suggesting that alterations in neurotrophins, particularly BDNF and NT-3, could contribute to depression through loss of neuroprotective effects. While they found nine SNPs in NTRK3 that were nominally associated with depression among adults of European ancestry, these SNPs were not significant after correcting for multiple testing. In our study, heterozygosis of rs7180942, is protective from depressive symptoms among older adults. Prior examination of HapMap data, heterozygosis for rs7180942 correlated with lower expression levels of NTRK3 (Mercader et al., 2008). This demonstrates a heterozygous advantage model for depressive symptoms when considered in light of other findings noting a relationship between lower expression levels of NTRK3 and depression (Feng et al., 2008; Verma et al., 2008). Consistent with prior findings that demonstrate lower presence of the GAC haplotype (consisting of the rs2059588, rs1110306, and rs3784406 variants, respectively) among childhood-onset mood disorders (Feng et al., 2008), lower presence GGAC haplotype analyzed in this study (including rs7180942, in addition to rs2059588, rs1110306, rs3784406) was likewise more indicative of depressive symptoms in older adults.

In contrast to the findings of Feng et al. (2008) which focused on childhood-onset mood disorders, the SNPs closer to the 3′ end of the NTRK3 gene (rs1369430, rs1435403, rs3784441) were not significantly associated with depressive symptoms in our analyses. The differences in findings may be due to biologic variances in age or depressive phenotype. In addition, we tested our hypothesis using path analysis which considers all the direct and indirect model pathways together. When testing within the full model, a small genetic association may not explain a large amount of the variance in depressive symptoms beyond the effects of the other variables.

As has been found in other genetic studies of depression, neither the SNP in the COMT gene nor the SNP in the SLC6A4 gene were significantly associated with depressive symptoms (Baekken et al., 2008; Illi et al., 2010; Wray et al., 2008) Most prior research in the SLC6A4 gene has included variable tandem repeater length in 5HTTLPR, which we did not include. The SNP we tested, rs25532, has only been found to be significantly associated with obsessive compulsive disorder in combination with the long 5HTTLPR allele and in a younger population (Wendland et al., 2008). Future studies should consider looking at the interaction of 5TTLPR long and short allele with genetic variation in rs25532 among depressed and nondepressed older adults.

Our hypothesis that genetics could indirectly influence depressive symptoms via pain, volunteer behavior, or physical activity, was not supported by any of our models. Many of the genes we chose were associated with pain in other studies (Sah et al., 2003; Vogel et al., 2003; Vossen et al., 2010; Young, Lariviere, & Belfer, 2011). The lack of association between the genes and pain may be due to the fact that we only asked about pain at the time of testing. Moreover, there was limited variance in pain among our sample (mean 3.53, SD = 3.00).

Our findings were not consistent with others who found genetic associations between neurotropic, serotonergic, or dopaminergic pathways and motivation to engage in physical activity (Caldwell Hooper et al., 2014; Garland et al., 2011). This sample of older adults was very sedentary. It is possible that among older adults, the influence exerted by BDNF, SLC6A4, or COMT genes is not enough of a motivation to exercise when faced with the daily hassles associated with advanced age.

Several limitations in this study should be considered. First, this was a small sample of older adults from a single CCRC who completed a one-time survey that was potentially influenced by challenges to recall. We included only a few SNPs from selected candidate genes that were previously examined in relation to depression and the functional significance of many of these SNPs is not yet known. Further, we recognize that our measure of depressive symptoms, although valid and reliable, was brief and we may have benefitted from a more comprehensive measure of depressive symptoms. In addition, only a small portion of our sample (14%) had substantial depressive symptoms, although our rates are similar to those published about similar settings of older adults. This study could also have been strengthened by using an objective measure of physical activity. Despite these limitations, this is one of few studies to examine the influence of candidate genes on depressive symptoms, as opposed to major depressive disorder, among older adults at risk for depression. Our use of a path analysis to examine a biopsychosocial model of depressive symptoms among older adults provided the opportunity to describe a comprehensive clinical picture of older adults at risk for depressive symptoms. Continued research on the genetic influence of depressive symptoms within the larger clinical picture of psychosocial risk factors can lead to the development of targeted interventions to identify older adults at risk for depressive symptoms and treatment could be started early to reduce the chance that the depressive symptoms would turn in to major depression. For example, if polymorphisms in the NTRK3 gene are replicated as important factors contributing to depressive symptoms in older adults, then behavioral interventions to improve neuroplasticity or other functional genetic consequences in older adults with the risk allele may be beneficial for reducing depressive symptoms among that group.

Funding

This work was supported by theNational Hartford Centers of Geriatric Nursing Excellence Claire M. Fagin Fellowship program funded by theHartford Foundation, The Mayday Fund (to N.J.K.), and theNIH/NINR (P30NR011396 to S.G.D.).