Vol. 53 No. 1 January 2017 Journal of Pain and Symptom Management 67

Original Article

Associations Between and Fatigue and Energy Levels in Women After Breast Cancer Surgery Jasmine Eshragh, RN, MS, Anand Dhruva, MD, Steven M. Paul, PhD, Bruce A. Cooper, PhD, Judy Mastick, RN, MN, Deborah Hamolsky, RN, MS, Jon D. Levine, MD, PhD, Christine Miaskowski, RN, PhD, and Kord M. Kober, PhD School of Nursing (J.E., S.M.P., B.A.C., J.M., D.H., C.M., K.M.K.) and School of Medicine (A.D., J.D.L.), University of California, San Francisco, California, USA

Abstract Context. Fatigue is a common problem in oncology patients. Less is known about decrements in energy levels and the mechanisms that underlie both fatigue and energy. Objectives. In patients with breast cancer, variations in neurotransmitter genes between lower and higher fatigue latent classes and between the higher and lower energy latent classes were evaluated. Methods. Patients completed assessments before and monthly for six months after surgery. Growth mixture modeling was used to identify distinct latent classes for fatigue severity and energy levels. Thirty candidate genes involved in various aspects of neurotransmission were evaluated. Results. Eleven single-nucleotide polymorphisms or haplotypes (i.e., ADRB2 rs1042718, BDNF rs6265, COMT rs9332377, CYP3A4 rs4646437, GALR1 rs949060, GCH1 rs3783642, NOS1 rs9658498, NOS1 rs2293052, NPY1R Haplotype A04, SLC6A2 rs17841327, and 5HTTLPR þ rs25531 in SLC6A4) were associated with latent class membership for fatigue. Seven single- nucleotide polymorphisms or haplotypes (i.e., NOS1 rs471871, SLC6A1 rs2675163, SLC6A1 Haplotype D01, SLC6A2 rs36027, SLC6A3 rs37022, SLC6A4 rs2020942, and TAC1 rs2072100) were associated with latent class membership for energy. Three of 13 genes (i.e., NOS1, SLC6A2, and SLC6A4) were associated with latent class membership for both fatigue and energy. Conclusions. Molecular findings support the hypothesis that fatigue and energy are distinct, yet related symptoms. Results suggest that a large number of play a role in the development and maintenance of fatigue and energy levels in breast cancer patients. J Pain Symptom Manage 2017;53:67e84 Ó 2016 American Academy of Hospice and Palliative Medicine. Published by Elsevier Inc. All rights reserved.

Key Words Fatigue, energy, neurotransmitter genes, growth mixture modeling, breast cancer, candidate genes

Introduction levels of fatigue in the first two months after surgery followed by mild-to-moderate levels of fatigue that per- Fatigue is the most common symptom associated sisted for 12 months after surgery. with cancer and its treatments.1 Although several The measurement of a patient’s level of energy has studies have examined fatigue in breast cancer pa- received little or no attention in the cancer literature. tients receiving chemotherapy (CTX)2 and radiation,3 Although energy level is commonly thought of as the studies on the occurrence of and predictors for fa- opposite of fatigue, evidence suggests that fatigue tigue after surgery are scarce. In a recent study,4 pa- and energy are distinct but related concepts.5,6 In tients with breast cancer reported relatively high oncology, fatigue is defined as a distressing and

Address correspondence to: Kord M. Kober, PhD, Department of Accepted for publication: August 3, 2016. Physiological Nursing, School of Nursing, University of Cal- ifornia, 2 Koret WaydN631Y, San Francisco, CA 94143-0610, USA. E-mail: [email protected]

Ó 2016 American Academy of Hospice and Palliative Medicine. 0885-3924/$ - see front matter Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jpainsymman.2016.08.004 68 Eshragh et al. Vol. 53 No. 1 January 2017

persistent sense of physical, emotional, and/or cogni- evening fatigue during CTX,23 between-group differ- tive tiredness or exhaustion related to cancer or its ences were identified in a number of neurotransmitter treatment that is not proportional to recent activity pathways. However, no studies were identified that and interferes with usual functioning.7 In contrast, en- evaluated for associations between neurotransmitter ergy is defined as an individual’s potential to perform genes and fatigue and energy levels in patients with physical and mental activity.6 In the only study that breast cancer. evaluated energy levels in patients with breast cancer The present study is based on our recent work that before surgery,8 while 32% of the women reported used growth mixture modeling (GMM) to identify clinically meaningful levels of fatigue, nearly 50% of distinct latent classes for fatigue severity and decre- these women reported clinically meaningful decre- ments in energy levels in women (n ¼ 398) before ments in energy levels. Findings from this study,8 and for six months after breast cancer surgery.16 Fa- and a study of patients who underwent radiation and tigue and energy levels were evaluated using the Lee their family caregivers9 and a study of patients with Fatigue Scale (LFS).24 In the GMM analysis for fatigue, HIV disease,10 support the hypothesis that energy is two distinct latent classes were identified (i.e., lower a distinct concept from fatigue. fatigue [38.5%] and higher fatigue [61.5%]). At Factors that contribute to fatigue severity are multi- enrollment, mean fatigue scores were 1.60 and 3.90 dimensional and include numerous biopsychosocial for the lower and higher fatigue classes, respectively. characteristics.11 Some of the predictors of fatigue af- In both fatigue classes, fatigue scores remained rela- ter breast cancer surgery include higher levels of anx- tively constant from the preoperative assessment to iety; the personality characteristic of extraversion;12 six months after breast cancer surgery. In the GMM increased fatigue prior to surgery;13 higher levels of analysis for energy, two distinct latent classes were emotional distress, mental fatigue, and pain;14 and identified (i.e., higher energy [32.0%] and lower en- depressive symptoms and receipt of CTX.4 ergy [68.0%]). At enrollment, mean energy scores Recent evidence suggests that genetic mechanisms were 5.82 and 4.35 for the higher and lower energy contribute to the severity of fatigue experienced by classes, respectively. In both energy groups, energy breast cancer patients. For example, in one study,15 a levels remained relatively constant from the preopera- number of pro-inflammatory cytokine genes were tive assessment to six months after breast cancer associated with fatigue. In another study,16 polymor- surgery. phisms in IL1B and IL10 were associated with fatigue Given the paucity of research on the role of various in women who underwent breast cancer surgery. Poly- neurotransmitters in the mechanism that underlie fa- morphisms in cytokine genes may contribute to fa- tigue severity or decrements in energy levels in pa- tigue severity through the modulation of pro- and tients with breast cancer, the purpose of this study anti-inflammatory pathways.17,18 was to evaluate for associations between variations in Although the majority of studies on genetic associa- a number of genes involved in neurotransmission, tions with fatigue has focused on cytokine dysregula- drug , and transport of molecules across tion, a number of additional pathways may influence cell membranes between the lower and higher fatigue fatigue and energy levels (for review, see Saligan latent classes and between the higher and lower en- et al.19). Neurotransmitter dysregulation may play an ergy latent classes. important role in the severity of fatigue and/or changes in energy levels. The most commonly cited neurotransmitter associated with fatigue is . For example, increased serum levels of serotonin Methods were linked to fatigue after prolonged exercise.20 Patients and Settings However, it is unlikely that a single neurotransmitter The study methods are described in detail else- is responsible for the development of/or changes in where.8,25,26 In brief, patients were recruited from fatigue and/or energy levels. Rather, it is more likely 2004 to 2008, from Breast Care Centers located in that several neurotransmitters contribute to interindi- a Comprehensive Cancer Center, two public hospi- vidual variability in fatigue and energy.21 Some neuro- tals, and four community practices. Patients were transmitter genes that were associated with fatigue and eligible to participate if they were adult women energy in a variety of populations include alterations ($18 years) who were scheduled to undergo unilat- in the dopaminergic system, specifically polymor- eral breast cancer surgery; were able to read, write, phisms in catechol-O-methyl- (COMT ), and understand English; agreed to participate; and -2 receptor (DRD2), and dopamine-1 trans- gave written informed consent. Patients were porter (DAT1).22 For example, in a recent study that excluded if they were having breast cancer surgery evaluated for differences in expression in breast on both breasts and/or had distant metastasis at cancer patients who reported lower vs. higher levels of the time of diagnosis. Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 69

Instruments included 5-hydroxytryptamine receptor (HTR) 1A The demographic questionnaire obtained informa- (HTR1A), HTR 1B (HTR1B), HTR 2A (HTR2A), tion on age, marital status, education, ethnicity, HTR 3A (HTR3A), SLC family 6 member 4dseroto- employment status, and living situation. Functional nin transporter (SLC6A4), and hydroxy- status was evaluated using the Karnofsky Performance lase 2 (TPH2). Genes involved in molecular Status (KPS) scale.27 The number and impact of co- transport and drug metabolism were adenosine morbid conditions was evaluated using the Self- triphosphateebinding cassette, subfamily B member administered Comorbidity Questionnaire (SCQ).28 1(ABCB1) and cytochrome P450, family 3, subfamily The LFS consists of 18 items designed to assess phys- A, polypeptide 4 (CYP3A4). A number of genes ical fatigue and energy.24 Each item was rated on a 0 to involved in various aspects of neurotransmission that 10 numeric rating scale. Total fatigue and energy scores were evaluated included brain-derived neurotrophic were calculated as the mean of the 13 fatigue items and factor (BDNF ), galanin (GAL), galanin receptor 1 the five energy items, with higher scores indicating (GALR1), galanin receptor 2 (GALR2), GTP cyclohy- greater fatigue severity and higher levels of energy. Pa- drolase 1 (GCH1), 1 (NOS1), ni- tients were asked to rate each item based on how they tric oxide synthase 2 (NOS2), neuropeptide Y (NPY ), felt ‘‘right now.’’ Cutoff scores of $ 4.4 and # 4.8 indi- neuropeptide Y receptor 1 (NPY1R), prodynorphin cate clinically meaningful levels of fatigue severity and (PDYN ), tachykinin precursor 1 (TAC1), and tachyki- low levels of energy.3 The LFS has well-established valid- nin receptor 1 (TACR1). All genes were identified ac- ity and reliability.24,29 Cronbach’s alphas for fatigue and cording to the approved symbol stored in the Human energy scales were 0.96 and 0.93, respectively. Genome Organization Gene Nomenclature Commit- tee database (http://www.genenames.org). Study Procedures Blood Collection and Genotyping. Of the 398 patients The study was approved by the Committee on Hu- who completed the enrollment assessment, 310 pro- man Research at the University of California, San vided a blood sample from which DNA could be iso- Francisco, and by the institutional review boards at lated from peripheral blood mononuclear cells each of the study sites. During the patients’ preopera- (PBMCs). Genomic DNA was extracted using the tive visit, a clinician explained the study and deter- PUREGene DNA Isolation System (Invitrogen, Carls- mined patients’ willingness to participate. The bad, CA). DNA was quantitated with a Nanodrop Spec- research nurse met with interested women, deter- trophotometer (ND-1000) and normalized to a mined eligibility, and obtained written informed con- concentration of 50 ng/mL(dilutedin10mMTris/ sent before surgery. After obtaining consent, patients 1 mM EDTA). Genotyping was performed using a completed the enrollment questionnaires an average custom array on the Golden Gate genotyping platform of four days before surgery. Patients completed the (Illumina, San Diego, CA) and processed according to LFS at enrollment and monthly for six months (i.e., the standard protocol using GenomeStudio (Illumina). seven assessments). Medical records were reviewed for disease and treatment information. Single-Nucleotide Polymorphisms Selection.Tagging single-nucleotide polymorphisms (SNPs) and Genomic Analyses literature-driven SNPs were selected for analysis. Gene Selection. Thirty candidate genes involved in Tagging SNPs were required to be common (defined various aspects of neurotransmission, drug meta- as having a minor allele frequency of $ 0.05) in public bolism, or transport of molecules across cell mem- databases. To ensure robust genetic association ana- branes were evaluated (Supplementary Table 1). lyses, quality control filtering of SNPs was performed. Genes involved in catecholaminergic neurotransmis- SNPs with call rates of less than 95% or Hardy- sion included alpha-1D-adrenergic receptor Weinberg P-values of < 0.001 were excluded. A total (ADRA1D), alpha-2A-adrenergic receptor (ADRA2A), of 249 SNPs among the 30 candidate genes passed all beta-2-adrenergic receptor (ADRB2), beta-3- the quality control filters and were included in the ge- adrenergic receptor (ADRB3), beta adrenergic recep- netic association analyses (Supplementary Table 1). tor kinase 2 (ADRBK2), solute-like carrier (SLC) fam- Localization of SNPs on the was ily 6 member 2dnoradrenaline transporter performed using the GRCh38 human reference as- (SLC6A2), SLC family 6 member 3ddopamine trans- sembly. Regional annotations were identified using porter (SLC6A3), hydroxylase (TH ), and the University of California Santa Cruz Human COMT. The gene involved in gamma-aminobutyric Genome Browser GRCh37/hg19 (http://genome. acid (GABA)-ergic neurotransmission was the SLC ucsc.edu/cgi-bin/hgTracks?db¼hg38). Potential regu- family 6 member 1dGABA transporter (SLC6A1). latory involvement of SNPs was investigated using Genes involved in serotonergic neurotransmission ENCODE.30 70 Eshragh et al. Vol. 53 No. 1 January 2017

Genotyping the Serotonin-Linked Polymorphic Region of For association tests, three genetic models were as- SLC6A4 sessed for each SNP: additive, dominant, and reces- The serotonin transporter-linked polymorphic re- sive. The genetic model that best fit the data, by gion (5-HTTLPR) occurs in the promoter region of maximizing the significance of the P-value, was the SLC6A4 gene. 5-HTTLPR occurs primarily as selected for each SNP. Logistic regression analysis, either a shorter (S, i.e., 14 repeats of 23 nucleotides) that controlled for significant covariates, and genomic or longer (L, i.e., 16 repeats) sequence.31 5-HTT estimates of (i.e., ancestry informative markers) and rs25531 is an SNP, which is present in either a com- self-reported race/ethnicity, was used to evaluate the mon (A) or rare (G) variant and is located immedi- associations between genotype and higher fatigue ately upstream of the 5-HTTLPR.32,33 The 5- and lower energy class memberships. Only those ge- HTTLPR/rs25531 haplotype (i.e., 5-HTTLPR triallelic netic associations identified as significant from the polymorphism) is known to influence SLC6A4 expres- bivariate analyses were evaluated in the multivariate sion levels. In vitro studies of the L allele suggest that analyses. A backward stepwise approach was used to the LA allele exhibits higher 5-HTT transcription create a parsimonious model. Except for race/ < and that the LG allele is more similar in function to ethnicity, only predictors with a P-value of 0.05 32,34 the S allele. In this study, the LA allele was used were retained in the final model. e as the reference allele. The 5-HTTLPR polymorphism As was done in our previous studies,9,46 50 based on was measured using polymerase chain reaction fol- the recommendations in the literature,51,52 and the lowed by resolution of polymerase chain reaction implementation of rigorous quality controls for products by gel electrophoresis.35 The rs25531 geno- genomic data, the nonindependence of SNPs/haplo- type was obtained by DNA cycle sequencing. types in linkage disequilibrium, and the exploratory nature of the analyses, adjustments were not made Statistical Analyses for the Phenotypic Data for multiple testing. In addition, significant SNPs iden- Data were analyzed using SPSS, version 22 (IBM tified in the bivariate analyses were evaluated further Corp, Armonk, NY),36 and STATA, version 13 (Stata- using logistic regression analyses that controlled for Corp LP, College Station, TX).37 As described previ- differences in phenotypic characteristics, potential ously,16 unconditional GMM with robust maximum confounding because of population stratification, likelihood estimation was carried out to identify latent and variations in other SNPs/haplotypes within the classes with distinct fatigue and energy trajectories us- same gene. Only those SNPs that remained significant e ing Mplus, version 5.21.38 40 Descriptive statistics and were included in the final presentation of the results. frequency distributions were generated for sample Therefore, the significant independent associations characteristics. Independent sample t-tests, Mann- reported are unlikely to be due solely to chance. Un- Whitney U tests, and chi-squared analyses were used adjusted associations, for all the SNPs evaluated, are to evaluate for differences in demographic and clin- found in Supplementary Table 1, to allow for subse- ical characteristics between the two latent classes for quent comparisons and meta-analyses. fatigue and energy. A P-value of < 0.05 was considered statistically significant. Results Statistical Analyses for the Genetic Data The genomic analyses are described in detail else- Differences in Demographic and Clinical where.16 In brief, allele and genotype frequencies Characteristics Between the Fatigue Classes were determined by gene counting. Hardy-Weinberg Differences between the two fatigue classes are equilibrium was assessed by the chi-square or Fisher listed in Table 1. Patients in the higher fatigue class exact tests. Measures of linkage disequilibrium (i.e., were significantly younger, had more years of educa- D0 and r2) were computed using Haploview 4.2. Haplo- tion, a lower KPS score, a higher SCQ score, a higher types were constructed using PHASE, version 2.1.41 number of lymph nodes removed, and a higher fa- Only inferred haplotypes that occurred with a fre- tigue severity score at enrollment. A larger percentage quency estimate of $ 15% were included in the asso- of patients in the higher fatigue class had received ciation analyses, assuming a dosage model. neoadjuvant CTX and had received CTX during the Ancestry informative markers were used to mini- first six months after breast cancer surgery. mize confounding because of population e stratification.42 44 Homogeneity in ancestry among Candidate Gene Analyses for Fatigue patients was verified by principal component anal- Genotype distributions differed between the lower ysis,45 using HelixTree (GoldenHelix, Bozeman, and higher fatigue classes for two SNPs and one haplo- MT). The first three PCs were used in all the logistic type in ADRB2; three SNPs in BDNF, one SNP in regression models. COMT, one SNP in CYP3A4, one SNP in GALR1, one Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 71

Table 1 Differences in Demographic and Clinical Characteristics Between the Lower Fatigue (n ¼ 153) and Higher Fatigue (n ¼ 244) Classes at Enrollment Lower Fatigue Higher Fatigue Class, Class, n ¼ 153 n ¼ 244 Characteristics (38.4%), Mean (SD) (61.3%), Mean (SD) Statistic and P-Value

Age (years) 57.8 (11.9) 53.1 (11.0) t ¼ 4.09, P # 0.0001 Education (years) 15.3 (2.5) 15.9 (2.8) t ¼2.02, P ¼ 0.04 Karnofsky Performance Status score 96.6 (7.0) 91.1 (11.4) t ¼ 5.86, P # 0.0001 Self-administered Comorbidity Questionnaire score 3.8 (2.6) 4.6 (3.0) t ¼2.64, P ¼ 0.009 Fatigue severity score at enrollment 1.6 (1.6) 4.1 (2.2) t ¼12.55, P # 0.0001 Number of breast biopsies in past year 1.5 (0.8) 1.5 (0.8) U, P ¼ 0.47 Number of positive lymph nodes 0.8 (1.9) 1.0 (2.4) t ¼0.88, P ¼ 0.38 Number of lymph nodes removed 4.8 (5.1) 6.4 (7.5) t ¼2.43, P ¼ 0.016

n (%) n (%)

Ethnicity c2 ¼ 2.82, P ¼ 0.42 White 100 (65.8) 155 (63.8) Black 19 (12.5) 21 (8.6) Asian/Pacific Islander 17 (11.2) 32 (13.2) Hispanic/mixed ethnic background/other 16 (10.5) 35 (14.4) Married/partnered (% yes) 64 (42.1) 100 (41.5) FE, P ¼ 0.92 Work for pay (% yes) 71 (46.4) 118 (49.0) FE, P ¼ 0.68 Lives alone (% yes) 40 (26.5) 54 (22.4) FE, P ¼ 0.40 Gone through menopause (% yes) 96 (63.6) 151 (64.3) FE, P ¼ 0.91 Stage of disease U, P ¼ 0.13 0 29 (19.0) 44 (18.0) I 66 (43.1) 85 (34.8) IIA and IIB 48 (31.4) 92 (37.7) IIIA, IIIB, IIIC, and IV 10 (6.5) 23 (9.4) Surgical treatment FE, P ¼ 1.00 Breast conservation 123 (80.4) 195 (79.9) Mastectomy 30 (19.6) 49 (20.1) Sentinel node biopsy (% yes) 130 (85.0) 197 (80.7) FE, P ¼ 0.34 Axillary lymph node dissection (% yes) 50 (32.7) 98 (40.3) FE, P ¼ 0.14 Breast reconstruction at the time of surgery (% yes) 33 (21.7) 53 (21.7) FE, P ¼ 1.00 Neoadjuvant chemotherapy (% yes) 21 (13.7) 58 (23.9) FE, P ¼ 0.014 Radiation therapy during the first six months (% yes) 87 (56.9) 137 (56.1) FE, P ¼ 0.92 Chemotherapy during the first six months (% yes) 36 (23.5) 97 (39.8) FE, P ¼ 0.001 FE ¼ Fisher exact test; U ¼ Mann Whitney U test.

SNP in GCH1, five SNPs and two haplotypes in NOS1, COMT rs9332377, carrying one or two doses of the one SNP and one haplotype in NPY1R, one SNP and rare C allele was associated with a 52% lower odds of one haplotype in SLC6A1, two SNPs and one haplo- belonging to the higher fatigue class (Fig. 1c). For type in SLC6A2, one SNP in SLC6A3, and two SNPs CYP3A4 rs4646437, carrying one or two doses of the and one haplotype in TAC1 and in the rare T allele was associated with a 52% lower odds of 5HTTLPR þ rs25521 haplotype in the SLC6A4 gene. belonging to the higher fatigue class (Fig. 1d). For GALR1 rs949060, carrying two doses of the rare C Regression Analyses for Fatigue allele was associated with a 2.46-fold higher odds of In these regression analyses that included genomic belonging to the higher fatigue class (Fig. 1e). For estimates of and self-reported race/ethnicity, the GCH1 rs3783642, carrying one or two doses of the only phenotypic characteristics that remained signifi- rare C allele was associated with a 53% lower odds of cant in the multivariate model were age, KPS score, belonging to the higher fatigue class (Fig. 1f). SCQ score, and receipt of CTX within six months after For NOS1, two SNPs were associated with member- breast cancer surgery. Eleven gene loci remained ship in the higher fatigue class. In the regression anal- significantly associated with fatigue class membership ysis, including both SNPs, for NOS1 rs9658498, in the regression analyses (Table 2). carrying two doses of the rare C allele was associated For ADRB2 rs1042718, carrying two doses of the rare with a 55% lower odds of belonging to the higher fa- A allele was associated with a 87% lower odds of tigue class (Fig. 2a). In the same regression analysis, belonging to the higher fatigue class (Fig. 1a). For for NOS1 rs2293052, carrying two doses of the rare T BDNF rs6265, carrying one or two doses of the rare allele was associated with a 4.58-fold higher odds of A allele was associated with a 50% lower odds of belonging to the higher fatigue class (Fig. 2b). For belonging to the higher fatigue class (Fig. 1b). For NPY1R Haplotype A04 (HapA04), that is composed 72 Eshragh et al. Vol. 53 No. 1 January 2017

Table 2 Table 2 Multiple Logistic Regression Analyses for Continued Neurotransmitter Genes and Lower Fatigue vs. Higher Odds Fatigue Classes Predictor Ratio SE 95% CI Z P-Value Odds Predictor Ratio SE 95% CI Z P-Value SCQ score 1.12 0.060 1.012, 1.246 2.19 0.029 Any chemotherapy 2.23 0.623 1.292, 3.858 2.88 0.004 c2 ¼ < 2 ¼ ADRB2 rs1042718 0.13 0.100 0.030, 0.582 2.67 0.008 Overall model fit: 52.30, P 0.0001, R 0.1280 Age 0.80 0.052 0.707, 0.912 3.39 0.001 ADRB2 ¼ adrenergic, beta-2 receptor, surface; any chemotherapy ¼ receipt of KPS score 0.56 0.097 0.396, 0.783 3.36 0.001 chemotherapy within six months after surgery; BDNF ¼ brain-derived neuro- SCQ score 1.11 0.062 0.998, 1.243 1.92 0.054 trophic factor; CI ¼ confidence interval; COMT ¼ catechol-O-methyltransfer- Any chemotherapy 2.31 0.669 1.307, 4.072 2.88 0.004 ase; CYP3A4 ¼ cytochrome P450, family 3, subfamily A, polypeptide 4; Overall model fit: c2 ¼ 59.87, P < 0.0001, R2 ¼ 0.1479 GALR1 ¼ galanin receptor 1; GCH1 ¼ GTP cyclohydrolase 1; ¼ ¼ BDNF rs6265 0.50 0.149 0.278, 0.897 2.32 0.020 Hap haplotype; 5-HTTLPR serotonin-linked polymorphic region; KPS ¼ Karnofsky Performance Status; NOS1 ¼ 1; Age 0.80 0.052 0.707, 0.910 3.43 0.001 ¼ ¼ NPY1R neuropeptide Y receptor Y1; SCQ Self-administered Comorbidity KPS score 0.57 0.101 0.406, 0.810 3.16 0.002 Questionnaire; SLC6A2 ¼ solute carrier family 6 (neurotransmitter trans- SCQ score 1.13 0.063 1.010, 1.256 2.14 0.032 porter, noradrenaline) member 2. Any chemotherapy 2.50 0.727 1.414, 4.420 3.15 0.002 Multiple logistic regression analyses of candidate gene associations with lower Overall model fit: c2 ¼ 56.84, P < 0.0001, R2 ¼ 0.1404 fatigue vs. higher fatigue classes (n ¼ 301). For each model, the first three COMT rs9332377 0.48 0.158 0.256, 0.919 2.22 0.026 principal components identified from the analysis of ancestry informative Age 0.82 0.052 0.723, 0.928 3.13 0.002 markers, as well as self-reported race/ethnicity, were retained in all models KPS score 0.55 0.095 0.389, 0.767 3.49 <0.0001 to adjust for potential confounding due to race/ethnicity (data not shown). SCQ score 1.13 0.063 1.011, 1.260 2.15 0.031 For the regression analyses, predictors evaluated in each model included ge- notype (ADRB2 rs1042718: CC þ CA vs. AA; BDNF rs6265: GG vs. GA þ AA; Any chemotherapy 2.41 0.697 1.370, 4.251 3.05 0.002 þ þ c2 ¼ < 2 ¼ COMT rs9332377: TT vs. TC CC; CYP3A4 rs4646437: CC vs. CT TT; Overall model fit: 56.34, P 0.0001, R 0.1392 GALR1 rs949060: GG þ GC vs. CC; GCH1 rs3783642: TT vs. TC þ CC; CYP3A4 0.48 0.157 0.253, 0.914 2.24 0.025 NOS1 rs9658498: TT þ TC vs. CC; NOS1 rs2293052: CC þ CT vs. TT; rs4646437 NPY1R HapA04: haplotype composed of the rs9764 common T allele and Age 0.81 0.052 0.710, 0.914 3.36 0.001 the rs7687423 common G allele; SLC6A2 rs17841327: CC þ CA vs. AA; and KPS score 0.55 0.098 0.392, 0.783 3.34 0.001 5HTTLPR þ rs25531 [the 5-HTTLPR triallelic polymorphism]: zero doses of SCQ score 1.12 0.063 1.005, 1.251 2.04 0.041 LA allele vs. one or two doses of LA allele), age (five-year increments), func- Any chemotherapy 2.40 0.691 1.365, 4.221 3.04 0.002 tional status (KPS score in 10 unit increments), number of comorbid condi- tions, and receipt of chemotherapy within six months after surgery. Overall model fit: c2 ¼ 56.43, P < 0.0001, R2 ¼ 0.1394 GALR1 rs949060 2.46 0.950 1.150, 5.244 2.32 0.020 Age 0.81 0.053 0.713, 0.920 3.25 0.001 KPS score 0.58 0.100 0.413, 0.814 3.15 0.002 SCQ score 1.12 0.063 1.007, 1.253 2.09 0.037 Any chemotherapy 2.55 0.738 1.444, 4.496 3.23 0.001 of alleles at two SNPs (i.e., rs9764 [common T allele] Overall model fit: c2 ¼ 56.98, P < 0.0001, R2 ¼ 0.1411 and rs7687423 [common G allele]), each additional GCH1 rs3783642 0.47 0.144 0.260, 0.859 2.46 0.014 dose of NPY1R HapA04 was associated with a 1.77- Age 0.81 0.052 0.713, 0.917 3.31 0.001 KPS score 0.58 0.102 0.411, 0.818 3.10 0.002 fold higher odds of belonging to the higher fatigue SCQ score 1.12 0.064 1.006, 1.256 2.07 0.039 class (Fig. 3). For SLC6A2 rs17841327, carrying two Any chemotherapy 2.40 0.690 1.364, 4.216 3.04 0.002 doses of the rare A allele was associated with a 10.31- Overall model fit: c2 ¼ 57.66, P < 0.0001, R2 ¼ 0.1424 NOS1 rs9658498 0.45 0.164 0.223, 0.920 2.19 0.029 fold higher odds of belonging to the higher fatigue NOS1 rs2293052 4.58 2.429 1.621, 12.953 2.87 0.004 class (Fig. 2c). For the 5-HTTLPR þ rs25531 polymor- Age 0.80 0.053 0.705, 0.913 3.33 0.001 phism in SLC6A4, carrying one or two doses of the L KPS score 0.54 0.095 0.383, 0.762 3.51 <0.0001 A SCQ score 1.11 0.063 0.991, 1.240 1.80 0.072 allele was associated with a 47% lower odds of Any chemotherapy 2.45 0.721 1.373, 4.361 3.04 0.002 belonging to the higher fatigue class (Fig. 2d). Overall model fit: c2 ¼ 69.13, P < 0.0001, R2 ¼ 0.1708 NPY1R Haplotype 1.77 0.346 1.207, 2.595 2.92 0.003 A04 Differences in Demographic and Clinical Age 0.81 0.052 0.711, 0.917 3.31 0.001 Characteristics Between the Energy Classes KPS score 0.55 0.099 0.388, 0.784 3.32 0.001 SCQ score 1.11 0.063 0.994, 1.241 1.85 0.064 Differences between the two energy classes are listed Any chemotherapy 2.58 0.756 1.454, 4.584 3.24 0.001 in Table 3. Patients in the lower energy class had a Overall model fit: c2 ¼ 60.22, P < 0.0001, R2 ¼ 0.1487 lower KPS score, a higher SCQ score, and a lower SLC6A2 10.31 8.139 2.195, 48.439 2.96 0.003 rs17841327 mean energy score at enrollment. In addition, a high- Age 0.81 0.053 0.717, 0.924 3.18 0.001 er percentage of patients with more advanced disease KPS score 0.56 0.101 0.395, 0.797 3.23 0.001 were in the lower energy class. SCQ score 1.13 0.064 1.007, 1.257 2.08 0.037 Any chemotherapy 2.68 0.784 1.514, 4.756 3.38 0.001 Overall model fit: c2 ¼ 65.01, P < 0.0001, R2 ¼ 0.1606 Candidate Gene Analyses for Energy þ 5HTTLPR 0.53 0.148 0.305, 0.914 2.28 0.023 Genotype distributions differed between the higher rs25531 in SLC6A4 energy and lower energy classes for: one SNP in Age 0.81 0.505 0.720, 0.919 3.33 0.001 COMT, two SNPs in HTR2A, one SNP in NOS1, one KPS score 0.59 0.101 0.422, 0.825 3.09 0.002 SNP in NOS2A, four SNPs and three haplotypes in (Continued) SLC6A1, four SNPs in SLC6A2, one SNP in SLC6A3, Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 73

Fig. 1. aef) Differences between the fatigue latent classes in the percentages of patients who were homozygous for the com- mon allele or heterozygous or homozygous for the rare allele in for each of the polymorphism identified. Values are plotted as unadjusted proportions with corresponding P-value. three SNPs and one haplotype in SLC6A4, one SNP in rs37022, carrying two doses of the rare A allele was TAC1, and one SNP in TACR1. associated with a 9.75-fold higher odds of belonging to the lower energy class (Fig. 4d). For SLC6A4 Regression Analyses for Energy rs2020942, carrying two doses of the rare A allele was In these regression analyses that included genomic associated with a 64% lower odds of belonging to estimates of and self-reported race/ethnicity, the the lower energy class (Fig. 4e). For TAC1 rs2072100, phenotypic characteristics that remained significant carrying two doses of the rare G allele was associated in the multivariate model were KPS score and receipt with a 2.11-fold higher odds of belonging to the lower of CTX within six months after breast cancer surgery. energy class (Fig. 4f). Seven gene loci remained significantly associated with energy class membership in the multivariate logistic regression analyses (Table 4). Discussion For NOS1 rs471871, carrying two doses of the rare T allele was associated with a 72% lower odds of A discussion of the differences in phenotypic char- belonging to the lower energy class (Fig. 4a). For acteristics between the fatigue latent classes, and be- tween the energy latent classes, is found in our SLC6A1, one SNP and one haplotype were associated 16 with membership in the lower energy class. For recent article. This discussion is focused on the SLC6A1 rs2675163, carrying one or two doses of the genotypic findings. rare C allele was associated with a 1.85-fold higher odds of belonging to the lower energy class Fatigue Polymorphisms (Fig. 4b). In the same regression analysis, for The ADRB2 receptor is part of the G-proteine SLC6A1 HapD01, that is composed of alleles at three coupled receptor family that influences sympathetic SNPs (i.e., rs10514669 [common C allele], rs2697138 nervous system responses and plays a role in the regu- [common C allele], and rs1062246 [common A lation of lipid metabolism. Polymorphisms in ADRB2 allele]), each additional dose of SLC6A1 HapD01 are associated with bronchodilation, insulin secretion, was associated with a 40% lower odds of belonging gluconeogenesis, and glycogenolysis in to the lower energy class (Fig. 5). For SLC6A2 and also increased cardiac output, arterial dilation, rs36027, each additional dose of the rare G allele and lipolysis.53 Sarpeshkar and Bentley53 hypothesized was associated with a 41% lower odds of belonging that alterations in this gene may be responsible to the lower energy class (Fig. 4c). For SLC6A3 for enhanced aerobic capacity and delayed 74 Eshragh et al. Vol. 53 No. 1 January 2017

Fig. 2. aed) Differences between the fatigue latent classes in the percentages of patients who were homozygous for the com- mon allele or heterozygous or homozygous for the rare allele in for each of the polymorphism identified. Values are plotted as unadjusted proportions with corresponding P-value. exercise-induced fatigue. In addition, ADRB2 receptor rare C allele for COMT rs9332377 had a 52% lower stimulation inhibits production of type 1 pro- odds of belonging to the higher fatigue class. This in- inflammatory cytokines,54 and underexpression of tronic SNP is located near the 30 untranslated region ADRB2 receptors is associated with chronic fatigue of the COMT gene. Although this location suggests syndrome.55 that this polymorphism has a regulatory function In our study, patients who were homozygous for the and may affect COMT expression,64 we found no sup- rare A allele for ADRB2 rs1042718 had a 87% lower port for nearby regulatory regions when the ENCODE odds of belonging to the higher fatigue class. Polymor- data were reviewed. Only three studies have reported phisms in ADRB2 rs1042718 produce a silent muta- significant associations between COMT rs9332377 tion. Although no studies were identified that and clinical phenotypes (i.e., hearing loss,65 suicidal evaluated associations between this SNP and fatigue, ideation,66 nicotine dependence64). No studies have significant associations were found between evaluated for associations between COMT rs9332377 rs1042718 and enhanced longevity56 and negative and fatigue. However, in the study of suicidal emotions.57 In the negative emotions study,57 individ- ideation,66 individuals who were homozygous for the uals who were heterozygous or homozygous for the rare C allele of COMT rs9332377 reported lower irrita- rare allele in rs1042718 were less likely to report feel- bility scores on the Questionnaire for Measuring Fac- ings of uselessness, loneliness, and anxiety. Given that tors of Aggression. This finding supports our previous studies of oncology patients found associa- association between rs9332377 and decreased fatigue tions between higher levels of psychological distress when one considers COMT’s role in the manifestation e and fatigue,58 60 our findings are consistent with of emotions, a possible marker for chronic fatigue those reported by Zheng et al.57 syndrome.67 COMT is a key responsible for the meta- BDNF is a neural growth factor found throughout bolism and inactivation of dopamine, norepineph- the central nervous system (CNS). BDNF is associated rine, and epinephrine.61 Alterations in the COMT with overall brain health because it plays a role in the gene were associated with fatigue and pain in breast promotion of neurogenesis, neuroprotection, mental cancer patients through interactions with two stress performance, and cognitive function.68 Altered pathways (i.e., hypothalamic-pituitary-adrenal axis, BDNF levels are associated with fibromyalgia syn- e the sympathetic nervous system).61 63 In our study, pa- drome,69 chronic fatigue syndrome,70 and tients who were heterozygous or homozygous for the depression.71 Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 75

BDNF levels in the brain, where it may play a greater role in the perception of fatigue, remains unknown. The CYP3A4 gene is a part of the cytochrome P450 superfamily. Cytochrome P450 are respon- sible for catalyzing multiple reactions involved in lipid synthesis and drug metabolism. These enzymes are responsible for the metabolism of approximately one third of anticancer drugs.74 The rs4646437 SNP is located in intron 7 of CYP3A4. Although no studies evaluated for associations between CYP3A4 rs4646437 and fatigue, in one study,75 an association was found between CYP3A4 rs4646437 and in vitro CYP3A expression and activity. In this study, women who carried the rare T allele of rs4646437 had higher expression and activity of the CYP3A4 enzyme. Consid- ering CYP3A4’s role in metabolizing anticancer drugs, one can hypothesize that women who are able to more effectively metabolize CTX would be less likely to experience higher levels of fatigue. This hypothesis is supported by our findings that being heterozygous or homozygous for rare T allele for rs4646437 was asso- ciated with a 52% lower odds of belonging to the high- er fatigue class. Galanin, a neuropeptide found throughout the CNS, has an inhibitory effect on multiple neurotrans- mitters.76 Polymorphisms in the galanin gene are asso- e Fig. 3. NPYR1 linkage disequilibrium (LD) based heatmap ciated with eating disorders,77 cancer,78 Alzheimer’s and haplotype analysis. The top white bar represents the disease,79,80 depression, and anxiety.76 Within the physical distance along the human . Reference sequence identifiers (rsIDs) for each single-nucleotide poly- CNS, the functional effects of galanin are mediated morphism (SNP) are plotted on the white bar and equidis- by three G-proteinecoupled receptor subtypes, tantly to render the pairwise LD estimates. The correlation including GALR1. The GAL1 receptor has an inhibi- 2 0 statistics (r and D ) are provided in the heatmap. The tory effect on adenylate cyclase through coupling haplotype is indicated in a bolded triangle and its compo- 0 with the G proteins Gi/Go. This inhibition affects nent SNPs are rendered in bold font. Pairwise D value (range: 0e1, inclusive) was rendered in color, with the dark- adenosine triphosphate metabolism and plays an 80 er red diamond representing D 0 value approaching 1.0. important role in cellular energy pathways. Of When the r2 (range 0e100, inclusive) is not equal to 0 or note, Staines81 hypothesized that dysfunctions in G 100, it is provided in a given diamond. The two-SNP haplo- proteinecoupled receptors (e.g., GALR1) contribute type associated with fatigue is composed of rs9764 and to the development of fatigue. In our study, patients rs7687423. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version who are homozygous for the rare C allele for GALR1 of this article.) rs949060 had a 2.46-fold higher odds of belonging to the higher fatigue class. GALR1 rs949060 is located BDNF rs6265 is a missense mutation that results in a on chromosome 18, approximately 3 kilobases up- nonsynonymous conservative change in the amino stream of the GALR1 gene in the promoter region. acid sequence from valine (Val) to methionine However, no nearby regulatory element was identified (Met). In two studies,71,72 decreases in serum BDNF in the ENCODE data. levels were associated with the Met allele. In our study, GCH1 is the rate-limiting enzyme involved in the being heterozygous or homozygous for the rare allele synthesis of (BH4). BH4 plays a was associated with a reduction in the odds of role in nitric oxide (NO) production and hydroxyl- belonging to the higher fatigue class. One might hy- ation of aromatic amino acids. Polymorphisms in pothesize that lower levels of BDNF would be associ- GCH1 are associated with pain,82 altered cognitive per- ated with membership in the higher fatigue class formance,83 and dopa-responsive dystonia.84 In our given that lower levels of BDNF were associated with study, being heterozygous or homozygous for the depression71 and chronic fatigue syndrome.70 Howev- rare C allele for GCH1 rs3783642 was associated with er, findings regarding changes in serum levels of a 53% lower odds of belonging to the higher fatigue BDNF associated with the Met allele are inconsis- class. Although no studies reported on GCH1 tent.73 In addition, the effect of the Met allele on rs3783642, one study85 found a protective association 76 Eshragh et al. Vol. 53 No. 1 January 2017

Table 3 Differences in Demographic and Clinical Characteristics Between the Higher Energy (n ¼ 127) and Lower Energy (n ¼ 270) Classes at Enrollment Higher Energy Lower Energy Class, Class, n ¼ 127 n ¼ 270 Characteristics (31.9%), Mean (SD) (67.8%), Mean (SD) Statistic and P-Value

Age (years) 56.5 (10.8) 54.2 (11.8) t ¼ 1.88, P ¼ 0.061 Education (years) 15.7 (2.2) 15.7 (2.8) t ¼ 0.01, P ¼ 0.994 Karnofsky Performance Status score 95.4 (9.4) 92.2 (10.6) t ¼ 3.06, P ¼ 0.002 Self-administered Comorbidity Questionnaire score 3.6 (2.3) 4.6 (3.0) t ¼3.47, P ¼ 0.001 Mean energy score at enrollment 6.1 (2.7) 4.4 (2.2) t ¼6.26, P # 0.0001 Number of breast biopsies in past year 1.5 (0.8) 1.5 (0.8) U, P ¼ 0.604 Number of positive lymph nodes 0.8 (2.0) 1.0 (2.3) t ¼ 0.76, P ¼ 0.450 Number of lymph nodes removed 5.0 (6.3) 6.1 (6.9) t ¼1.51, P ¼ 0.132

n (%) n (%)

Ethnicity c2 ¼ 1.75, P ¼ 0.627 White 86 (68.3) 169 (62.8) Black 10 (7.9) 30 (11.2) Asian/Pacific Islander 16 (12.7) 33 (12.3) Hispanic/mixed ethnic background/other 14 (11.1) 37 (13.8) Married/partnered (% yes) 50 (39.7) 114 (42.7) FE, P ¼ 0.586 Work for pay (% yes) 66 (52.4) 123 (45.9) FE, P ¼ 0.236 Lives alone (% yes) 29 (23.0) 65 (24.4) FE, P ¼ 0.801 Gone through menopause (% yes) 84 (68.3) 163 (62.0) FE, P ¼ 0.256 Stage of disease U, P ¼ 0.040 0 29 (22.8) 44 (16.3) I 51 (40.2) 100 (37.0) IIA and IIB 39 (30.7) 101 (37.4) IIIA, IIIB, IIIC, and IV 8 (6.3) 25 (9.3) Surgical treatment FE, P ¼ 0.686 Breast conservation 100 (78.7) 218 (80.7) Mastectomy 27 (21.3) 52 (19.3) Sentinel node biopsy (% yes) 103 (81.1) 224 (83.0) FE, P ¼ 0.673 Axillary lymph node dissection (% yes) 40 (31.7) 108 (40.0) FE, P ¼ 0.120 Breast reconstruction at the time of surgery (% yes) 28 (22.2) 58 (21.5) FE, P ¼ 0.896 Neoadjuvant chemotherapy (% yes) 22 (17.5) 57 (21.1) FE, P ¼ 0.421 Radiation therapy during the first six months (% yes) 75 (59.1) 149 (55.2) FE, P ¼ 0.515 Chemotherapy during the first six months (% yes) 34 (26.8) 99 (36.7) FE, P ¼ 0.054 FE ¼ Fisher Exact test; U ¼ Mann Whitney U test. between other polymorphisms in GCH1 and fibromy- and one intronic SNP (rs7687423). Although no algia syndrome. This locus resides in putative studies were identified that reported on NPYR1 CCCTC-binding factor (CTCF) and RAD22 transcrip- HapA04, polymorphisms in rs9764 and rs7687423 tion factorebinding sites that suggests a possible role were associated with nicotine90 and methamphet- in the regulation of GCH1. amine91 dependence, respectively. No studies were NPY1R is part of a family of G proteinecoupled re- identified that reported on associations with either ceptors that binds NPY. NPY acts in both the CNS and SNP and fatigue. peripheral nervous system. Peripherally, NPY is a neurotransmitter that is released from sympathetic nerve endings. Centrally, NPY acts on receptors pre- Energy Polymorphisms sent in those areas of the brain that are involved The SLC6A1 gene encodes for one of the four with emotion.86 NPY is involved in sleep regulation, GABA transporters found in the brain. The role of anxiety, memory, pain, and energy homeostasis.87,88 this transporter is to remove GABA from the synaptic Alterations in NPY are implicated in chronic fatigue cleft that decreases extracellular levels of GABA. The syndrome86 and depression.89 Alterations in NPY inhibitory neurotransmitter GABA is important for signaling through variations in NPY1R may have an ef- normal brain function. Based on studies of knockout fect on any of the aforementioned processes, mice,92 deficiencies in SLC6A1 were associated with including fatigue. depression, reduced aggression, and reduced anxiety. In our study, each additional dose of NPY1R Furthermore, an association was found between poly- HapA04 was associated with a 1.77-fold higher odds morphisms in SLC6A1 and anxiety disorders.93 In a of belonging to the higher fatigue class. HapA04 genome-wide association study,94 an SNP in SLC6A1 comprised a 30 untranslated region SNP (rs9764) was associated with symptoms of inattention and Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 77

Table 4 Multiple Logistic Regression Analyses for Neurotransmitter Genes and Higher Energy vs. Lower Energy Classes Predictor Odds Ratio SE 95% CI Z P-Value

NOS1 rs471871 0.28 0.138 0.103, 0.736 2.57 0.010 KPS score 0.65 0.101 0.483, 0.884 2.75 0.006 Any chemotherapy 1.73 0.479 1.002, 2.972 1.97 0.049 Overall model fit: c2 ¼ 24.43, P ¼ 0.0037, R2 ¼ 0.0638 SLC6A1 rs2675163 1.85 0.507 1.082, 3.166 2.25 0.025 SLC6A1 Haplotype D01 0.60 0.116 0.413, 0.880 2.62 0.009 KPS score 0.68 0.105 0.503, 0.921 2.49 0.013 Any chemotherapy 1.56 0.440 0.898, 2.714 1.58 0.114 Overall model fit: c2 ¼ 30.86, P ¼ 0.0006, R2 ¼ 0.0810 SLC6A2 rs36027 0.59 0.107 0.415, 0.844 2.90 0.004 KPS score 0.66 0.102 0.484, 0.889 2.72 0.007 Any chemotherapy 1.75 0.485 1.014, 3.010 2.01 0.044 Overall model fit: c2 ¼ 26.25, P ¼ 0.0019, R2 ¼ 0.0686 SLC6A3 rs37022 9.75 10.612 1.155, 82.302 2.09 0.036 KPS score 0.66 0.103 0.484, 0.895 2.67 0.008 Any chemotherapy 1.75 0.487 1.017, 3.022 2.02 0.043 Overall model fit: c2 ¼ 24.77, P ¼ 0.0032, R2 ¼ 0.0647 SLC6A4 rs2020942 0.36 0.144 0.161, 0.787 2.55 0.011 KPS score 0.66 0.103 0.488, 0.898 2.65 0.008 Any chemotherapy 1.73 0.482 1.006, 2.991 1.98 0.047 Overall model fit: c2 ¼ 24.16, P ¼ 0.0041, R2 ¼ 0.0631 TAC1 rs2072100 2.11 0.718 1.083, 4.113 2.19 0.028 KPS score 0.67 0.102 0.498, 0.905 2.61 0.009 Any chemotherapy 1.73 0.480 1.007, 2.983 1.98 0.047 Overall model fit: c2 ¼ 22.78, P ¼ 0.0067, R2 ¼ 0.0595 Any chemotherapy ¼ receipt of chemotherapy within six months after surgery; CI ¼ confidence interval; Hap ¼ haplotype; KPS ¼ Karnofsky Performance Status; NOS1 ¼ nitric oxide synthase 1; SCQ ¼ Self-administered Comorbidity Questionnaire; SLC6A1 ¼ solute carrier family 6 (neurotransmitter transporter, GABA) member 1; SLC6A2 ¼ solute carrier family 6 (neurotransmitter transporter, noradrenaline) member 2; SLC6A3 ¼ solute carrier family 6 (neurotransmitter trans- porter, dopamine) member 3; SLC6A4 ¼ solute carrier family 6 (neurotransmitter transporter, serotonin) member 4; TAC1 ¼ tachykinin, precursor 1. Multiple logistic regression analyses of candidate gene associations with higher energy vs. lower energy classes (n ¼ 301). For each model, the first three principal components identified from the analysis of ancestry informative markers, as well as self-reported race/ethnicity, were retained in all models to adjust for potential confounding due to race/ethnicity. For the regression analyses, predictors evaluated in each model included genotype (NOS1 rs471871 genotype: AA þ AT vs. TT; SLC6A1 rs2675163 genotype: TT vs. TC þ CC; SLC6A1 HapD01 haplotype: composed of the rs10514669 common C allele, the rs2697138 common C allele, and the rs1062246 common A allele; SLC6A2 rs36027 genotype: AA vs. AG vs. GG; SLC6A3 rs37022 genotype: TT þ TA vs. AA; SLC6A4 rs2020942 genotype: GG þ GA vs. AA; TAC1 rs2072100 genotype: AA þ AG vs. GG), functional status (KPS score in 10-unit increments), and receipt of chemotherapy within six months after surgery. hyperactivity in attention-deficit/hyperactivity disor- Although the majority of the literature on polymor- der (ADHD). phisms in the SLC6A3 gene has focused on In our study, one SNP and one haplotype in the ADHD,97,98 associations were found between dopami- SLC6A1 gene were associated with membership in nergic polymorphisms and fatigue22 and decreases in the lower energy class. Being heterozygous or homozy- mental energy and sustained attention.99 In our study, gous for the rare C allele of SLC6A1 rs2675163 was being homozygous for the rare A allele of SLC6A3 associated with a 1.85-fold higher odds of belonging rs37022 was associated with a 9.75-fold higher odds to the lower energy class, whereas each additional of belonging to the lower energy class. No studies dose of SLC6A1 HapD01, that is composed of alleles were identified that reported on polymorphisms in at three SNPs (i.e., rs10514669, rs2697138, and this SNP. rs1062246), was associated with a 40% lower odds of The TAC1 gene encodes for a group of tachykinin belonging to the lower energy class. No studies were peptide hormones (e.g., Substance P) that function identified that reported on polymorphisms in any of as neurotransmitters. Substance P is implicated in fi- the SLC6A1 SNPs and energy. bromyalgia syndrome100 and with fatigue and other The SLC6A3 gene encodes for a dopamine trans- negative mood states.101 Therefore, polymorphisms porter. The dopamine transporter protein is respon- in tachykinin pathway genes may have an effect on fa- sible for re-uptake of dopamine from the synaptic tigue and energy levels. In our study, being homozy- cleft that results in decreased extracellular levels of gous for the rare G allele of TAC1 rs2072100 was dopamine.95 Decreased levels of dopamine are hy- associated with a 2.11-fold higher odds of belonging pothesized to play a role in the development of central to the lower energy class. Although the rs2072100 fatigue because of dopamine’s known effects on initi- polymorphism was linked with increased risk for colo- ation of movement.96 Therefore, alterations in dopa- rectal cancer102 and susceptibility to multiple scle- minergic circuits, including its transport receptors, rosis,103 no studies were identified that reported on may affect an individual’s energy level and fatigue. associations with energy. 78 Eshragh et al. Vol. 53 No. 1 January 2017

Fig. 4. aef) Differences between the energy latent classes in the percentages of patients who were homozygous for the com- mon allele or heterozygous or homozygous for the rare allele in for each of the polymorphism identified. Values are plotted as unadjusted proportions with corresponding P-value.

Fatigue and Energy Polymorphisms In addition, fatigue is a common symptom associated 113 Three genes (i.e., NOS1, SLC6A2, and SL6A4) were with PD and may share similar susceptibility gene associated with latent class membership for both fa- polymorphisms. A different SNP in the NOS1 gene tigue and energy. NOS1 is part of a group of NOS was associated with energy levels. Being homozygous responsible for the synthesis of NO. NO mediates vaso- for the rare T allele of rs471871 was associated with a dilation, neural regulation of skeletal muscle, and 72% lower odds of belonging to the lower energy class. neurotransmission.104 Elevated NO levels are impli- No studies were identified that reported on NOS1 cated in chronic fatigue syndrome,105 fatigue in rs471871. muscular dystrophies,106,107 and fatigue in post- The SLCA2 gene encodes for the norepinephrine radiation syndrome.108 Although no studies were iden- transporter (NET) protein. The NET found at norad- tified on associations between NOS polymorphisms renergic synapses is responsible for the removal of NE and fatigue, other studies found associations between from the synaptic cleft and plays a major role in NE polymorphisms in the NOS1 gene and depression109 homeostasis.114 Impairments in the NET protein and anxiety.110 may contribute to the development of fatigue.115 Mu- In our study, two SNPs in the NOS1 gene were asso- tations in the SLCA2 gene are associated with ortho- ciated with membership in the higher fatigue class. static intolerance, a syndrome that includes fatigue 114,116 Being homozygous for the rare C allele of rs9658498 as a significant symptom. In addition, polymor- was associated with a 55% lower odds of belonging phisms in the SLCA2 gene were associated with major to the higher fatigue class, whereas carrying two doses depression, a condition that includes fatigue as a ma- 117 of the rare T allele of rs2293052 was associated with a jor symptom. In our study, being homozygous for 4.58-fold higher odds of belonging to the higher fa- the rare A allele of SLC6A2 rs17841327 was associated tigue class. No studies were identified that reported with a 10.31-fold higher odds of belonging to the high- on NOS1 rs9658498. However, an association was er fatigue group. In addition, a different SNP in the found between rs2293052 and Parkinson’s disease SLC6A2 gene was associated with energy levels. Each (PD).111 These results support our findings of an asso- additional dose of the rare G allele of SLC6A2 ciation between this SNP and increased fatigue rs36027 was associated with a 41% higher odds of because similar to the aforementioned fatigue syn- belonging to the lower energy class. No studies were dromes, PD is associated with increased NO levels.112 identified that reported on either SLC6A2 SNP. Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 79

Fig. 5. SLC6A1 linkage disequilibrium (LD)ebased heatmap and haplotype analysis. The top white bar represents the phys- ical distance along the human chromosome. Reference sequence identifiers (rsIDs) for each single-nucleotide polymorphism (SNP) are plotted on the white bar and equidistantly to render the pairwise LD estimates. The correlation statistics (r2 and D 0) are provided in the heatmap. The haplotype is indicated in a bolded triangle and its component SNPs are rendered in bold font. Pairwise D 0 value (range 0e1, inclusive) was rendered in color, with the darker red diamond representing D 0 value ap- proaching 1.0. When the r2 (range 0e100, inclusive) is not equal to 0 or 100, it is provided in a given diamond. The three-SNP haplotype associated with energy is composed of rs10514669, rs2697138, and rs1062246. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

The SLC6A4 gene encodes for a membrane protein was associated with a 47% decrease in the odds of that is responsible for re-uptake of serotonin from the belonging to the higher fatigue class. Although synaptic cleft. The serotonergic neurotransmitter sys- rs2020942 was linked with obsessive-compulsive symp- tem is hypothesized to play a role in cancer-related fa- toms122 and risk for nonsyndromic cleft lip with or tigue.118,119 Serotonin is involved in various human without cleft palate,123 no studies have reported on as- behaviors including sleep, mood, appetite, memory, sociations between SLC6A4 rs2020942 and energy level. and learning. Increased levels of serotonin in the Although the functional consequences of SLC6A4 brain are hypothesized to contribute to fatigue rs2020942 are not known, a considerable amount of through its interaction with the hypothalamic- research has evaluated the functional consequences pituitary-adrenal axis leading to a sensation of of the A > G polymorphism (rs25531) within the L reduced potential to perform physical activity.118 Ya- allele of 5-HTTLPR. In this triallelic polymorphism 120 mamoto et al. demonstrated a reduced density of (i.e., S, LA, and LG), nearly equivalent expression of serotonin transporters in the rostral subdivision of the serotonin transporter is observed with the S and the anterior cingulate of patients with chronic fatigue LG alleles and increased transcriptional activity is 32,34 syndrome. In addition, an association was found be- observed with the LA allele. In addition, other tween polymorphisms in the promoter of the studies found that being homozygous for the LA allele SLC6A4 gene and chronic fatigue syndrome.121 was associated with increased serotonin transporter In our study, being homozygous for the rare A allele binding.124,125 Because the serotonin transporter of SLC6A4 rs2020942 was associated with a 64% lower modulates the concentration of serotonin at the syn- odds of belonging to the lower energy class. In addi- aptic cleft, an increase in the transcription of the tion, carrying one or two doses of the LA allele for transporter would decrease the amount of serotonin the 5-HTTLPR þ rs25531 polymorphism in SLC6A4 at the synaptic cleft. In a number of studies, 80 Eshragh et al. Vol. 53 No. 1 January 2017

individuals who possessed one or two doses of the low- Christine Miaskowski is an American Cancer Society activity alleles (S or LG) were more likely to develop Clinical Research Professor and is supported by a behavioral disorders (e.g., depression) when they K05 award from the NCI (CA168960). This project is experienced stressful life events.126,127 Our findings supported by National Institutes of Health (NIH)/ are consistent with these observations in that patients NCRR UCSF-CTSI grant number UL1 RR024131. Its who carried one or two doses of the LA allele were less contents are solely the responsibility of the authors likely to be in the higher fatigue class. One can hy- and do not necessarily represent the official views of pothesize that patients with the LA allele have the NIH. The authors declare no conflicts of interest. increased transcription of the serotonin transporter that would decrease the concentration of serotonin at the synaptic cleft. The decreased concentration of References serotonin would result in decreased levels of fatigue. 1. Prue G, Rankin J, Allen J, Gracey J, Cramp F. Cancer- related fatigue: a critical appraisal. Eur J Cancer 2006;42: 846e863. Conclusions 2. Siefert ML. Fatigue, pain, and functional status during outpatient chemotherapy. Oncol Nurs Forum 2010;37: Several limitations must be acknowledged. e Although our sample size was sufficient, additional E114 E123. studies with independent samples are needed to 3. Dhruva A, Dodd M, Paul SM, et al. Trajectories of fa- confirm the latent classes and the genetic associations. tigue in patients with breast cancer before, during, and after radiation therapy. Cancer Nurs 2010;33:201e212. To increase the generalizability of these results, women were recruited from seven different centers 4. Huang HP, Chen ML, Liang J, Miaskowski C. Changes and approximately 30% of the patients were ethnic in and predictors of severity of fatigue in women with breast cancer: a longitudinal study. Int J Nurs Stud 2014;51: minorities. However, the single diagnosis of breast can- 582e592. cer limits the generalizability of the findings to other 5. Lerdal A. A theoretical extension of the concept of en- cancer diagnoses. Finally, recent studies suggest that ergy through an empirical study. Scand J Caring Sci 2002;16: diurnal variations in fatigue and energy levels consti- 197e206. tute distinct symptom phenotypes.9,128,129 Future 6. Lerdal A. A concept analysis of energy. Its meaning in studies need to evaluate for differences in the levels the lives of three individuals with chronic illness. Scand J of morning and evening fatigue and decrements in en- Caring Sci 1998;12:3e10. ergy levels in women after breast cancer surgery. 7. Berger AM, Mooney K, Alvarez-Perez A, et al. Cancer- The molecular findings from this study support the related fatigue, version 2.2015. J Natl Compr Canc Netw hypothesis that fatigue and energy are distinct, yet 2015;13:1012e1039. related symptoms. Only three of the 13 genes identified 8. Van Onselen C, Aouizerat BE, Dunn LB, et al. Differ- in this study were associated with membership in both ences in sleep disturbance, fatigue and energy levels be- the fatigue and energy latent classes. Additional sup- tween women with and without breast pain prior to breast port for this hypothesis comes from recent studies cancer surgery. Breast 2013;22:273e276. that explored the concepts of fatigue and energy in pa- 9. Aouizerat BE, Dhruva A, Paul SM, Cooper BA, tients with cancer and their family caregivers9,46 and in Kober KM, Miaskowski C. Phenotypic and molecular evi- women with HIV.130 Lerdal et al.130 concluded that fa- dence suggests that decrements in morning and evening en- ergy are distinct but related symptoms. J Pain Symptom tigue and energy are distinct constructs and should Manage 2015;50:599e614. not be used interchangeably, neither clinically nor in research. Additional studies are needed to determine 10. Aouizerat BE, Gay CL, Lerdal A, Portillo CJ, Lee KA. Lack of energy: an important and distinct component of which common and distinct phenotypic and molecular HIV-related fatigue and daytime function. J Pain Symptom characteristics are associated with increased fatigue Manage 2013;45:191e201. severity and decrements in energy. Findings from these 11. Berger AM, Gerber LH, Mayer DK. Cancer-related fa- types of studies will provide insights into the mecha- tigue: implications for breast cancer survivors. Cancer nisms that underlie one or both of these symptoms 2012;118:2261e2269. and facilitate the development and testing of interven- 12. De Vries J, Van der Steeg AF, Roukema JA. Determi- tions to decrease fatigue and/or increase energy levels nants of fatigue 6 and 12 months after surgery in women of patients undergoing cancer treatment. with early-stage breast cancer: a comparison with women with benign breast problems. J Psychosom Res 2009;66: 495e502. Disclosures and Acknowledgments 13. Lockefeer JP, De Vries J. What is the relationship be- tween trait anxiety and depressive symptoms, fatigue, and This study was funded by grants from the National low sleep quality following breast cancer surgery? Psychoon- Cancer Institute (NCI CA107091 and CA118658). Dr. cology 2013;22:1127e1133. Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 81

14. Bortolon C, Krikorian A, Carayol M, Brouillet D, 31. Nakamura M, Ueno S, Sano A, Tanabe H. The human Romieu G, Ninot G. Cancer-related fatigue in breast cancer serotonin transporter gene linked polymorphism (5- patients after surgery: a multicomponent model using par- HTTLPR) shows ten novel allelic variants. Mol Psychiatry tial least squares-path modeling. Psychooncology 2014;23: 2000;5:32e38. 444e451. 32. Hu XZ, Lipsky RH, Zhu G, et al. Serotonin transporter 15. Bower JE, Ganz PA, Irwin MR, Castellon S, Arevalo J, promoter gain-of-function genotypes are linked to obsessive- Cole SW. Cytokine genetic variations and fatigue among pa- compulsive disorder. Am J Hum Genet 2006;78:815e826. tients with breast cancer. J Clin Oncol 2013;31:1656e1661. 33. Wendland JR, Martin BJ, Kruse MR, Lesch KP, 16. Kober KM, Smoot B, Paul SM, Cooper BA, Levine JD, Murphy DL. Simultaneous genotyping of four functional Miaskowski C. Polymorphisms in cytokine genes are associ- loci of human SLC6A4, with a reappraisal of 5-HTTLPR ated with higher levels of fatigue and lower levels of energy and rs25531. Mol Psychiatry 2006;11:224e226. in women following breast cancer surgery. J Pain Symptom Manage 2016;52:695e708. 34. Praschak-Rieder N, Kennedy J, Wilson AA, et al. Novel 5-HTTLPR allele associates with higher serotonin trans- 17. Illi J, Miaskowski C, Cooper B, et al. Association be- porter binding in putamen: a [(11)C] DASB positron emis- tween pro- and anti-inflammatory cytokine genes and a sion tomography study. Biol Psychiatry 2007;62:327e331. symptom cluster of pain, fatigue, sleep disturbance, and depression. Cytokine 2012;58:437e447. 35. Kaiser R, Tremblay PB, Schmider J, et al. Serotonin transporter polymorphisms: no association with response 18. Collado-Hidalgo A, Bower JE, Ganz PA, Irwin MR, to antipsychotic treatment, but associations with the schizo- Cole SW. Cytokine gene polymorphisms and fatigue in paranoid and residual subtypes of . Mol Psy- breast cancer survivors: early findings. Brain Behav Immun chiatry 2001;6:179e185. 2008;22:1197e1200. 36. SPSS. IBM SPSS for windows (version 23). Chicago: 19. Saligan LN, Olson K, Filler K, et al. The biology of SPSS, Inc., 2015. cancer-related fatigue: a review of the literature. Support Care Cancer 2015;23:2461e2478. 37. StataCorp. Stata statistical software: Release 14. College Station, TX: Stata Corporation, 2015. 20. Meeusen R, Roelands B. Central fatigue and neuro- transmitters, can thermoregulation be manipulated? Scand 38. Jung T, Wickrama KAS. An introduction to latent class J Med Sci Sports 2010;20 Suppl 3:19e28. growth analysis and growth mixture modeling. Social Per- sonal Psychol Compass 2008;2:302e317. 21. Meeusen R, Watson P, Hasegawa H, Roelands B, Piacentini MF. Brain neurotransmitters in fatigue and over- 39. Nylund KL, Asparouhov T, Muthen BO. Deciding on training. Appl Physiol Nutr Metab 2007;32:857e864. the number of classes in latent class analysis and growth mixture modeling: a Monte Carlo simulation study. Struct 22. Malyuchenko NV, Schegolkova JV, Kulikova MA, Equ Model 2007;14:535e569. Timofeeva MA, Shlepzova VA, Sysoeva OV, et al. Effects of genetic variations in the dopaminergic system on fatigue 40. Tofighi D, Enders CK. Identifying the correct number in humans: gender aspects. Bull Exp Biol Med 2010;149: of classes in growth mixture models. Charlotte, NC: Informa- 226e232. tion Age Publishing, 2008. 23. Kober KM, Dunn L, Mastick J, et al. 41. Stephens M, Smith NJ, Donnelly P. A new statistical profiling of evening fatigue in women undergoing chemo- method for haplotype reconstruction from population therapy for breast cancer. Biol Res Nurs 2016;18:370e385. data. Am J Hum Genet 2001;68:978e989. 24. Lee KA, Hicks G, Nino-Murcia G. Validity and reliability 42. Halder I, Shriver M, Thomas M, Fernandez JR, of a scale to assess fatigue. Psychiatry Res 1991;36:291e298. Frudakis T. A panel of ancestry informative markers for esti- 25. McCann B, Miaskowski C, Koetters T, et al. Associations mating individual biogeographical ancestry and admixture from four continents: utility and applications. Hum Mutat between pro- and anti- inflammatory cytokine genes and e breast pain in women prior to breast cancer surgery. J Pain 2008;29:648 658. 2012;13:425e437. 43. Hoggart CJ, Parra EJ, Shriver MD, et al. Control of con- 26. Miaskowski C, Cooper B, Paul SM, et al. Identification founding of genetic associations in stratified populations. e of patient subgroups and risk factors for persistent breast Am J Hum Genet 2003;72:1492 1504. pain following breast cancer surgery. J Pain 2012;13: 44. Tian C, Gregersen PK, Seldin MF. Accounting for e 1172 1187. ancestry: population substructure and genome-wide associa- 27. Karnofsky D, Abelmann WH, Craver LV, Burchenal JH. tion studies. Hum Mol Genet 2008;17:R143eR150. The use of nitrogen mustards in the palliative treatment of e 45. Price AL, Patterson NJ, Plenge RM, et al. Principal com- carcinoma. Cancer 1948;1:634 656. ponents analysis corrects for stratification in genome-wide 28. Sangha O, Stucki G, Liang MH, Fossel AH, Katz JN. association studies. Nat Genet 2006;38:904e909. The Self-Administered Comorbidity Questionnaire: a new 46. Dhruva A, Aouizerat BE, Cooper B, et al. Cytokine gene method to assess comorbidity for clinical and health services e associations with self-report ratings of morning and evening research. Arthritis Rheum 2003;49:156 163. fatigue in oncology patients and their family caregivers. Biol 29. Gay CL, Lee KA, Lee SY. Sleep patterns and fatigue in Res Nurs 2015;17:175e184. e new mothers and fathers. Biol Res Nurs 2004;5:311 318. 47. Miaskowski C, Cataldo JK, Baggott CR, et al. Cytokine 30. Encode Project ConsortiumDunham I, Kundaje A, gene variations associated with trait and state anxiety in et al. An integrated encyclopedia of DNA elements in the oncology patients and their family caregivers. Support human genome. Nature 2012;489:57e74. Care Cancer 2015;23:953e965. 82 Eshragh et al. Vol. 53 No. 1 January 2017

48. Dunn LB, Aouizerat BE, Langford DJ, et al. Cytokine 64. Greenbaum L, Kanyas KS, Rigbi A, Alkelai A, Kohn Y, gene variation is associated with depressive symptom trajec- Lerer B. Why do young women smoke? VII COMT as a risk tories in oncology patients and family caregivers. Eur J On- modifying gene for Nicotine dependence - role of gene- col Nurs 2013;17:346e353. gene interaction, personality, and environmental factors. Hum Psychopharmacol 2010;25:536e542. 49. Miaskowski C, Dodd M, Paul SM, et al. Lymphatic and angiogenic candidate genes predict the development of sec- 65. Ross CJ, Katzov-Eckert H, Dube MP, et al. Genetic vari- ondary lymphedema following breast cancer surgery. PLoS ants in TPMT and COMT are associated with hearing loss in One 2013;8:e60164. children receiving cisplatin chemotherapy. Nat Genet 2009; 41:1345e1349. 50. Miaskowski C, Cooper BA, Dhruva A, et al. Evidence of associations between cytokine genes and subjective reports 66. Calati R, Porcelli S, Giegling I, et al. Catechol-o-methyl- of sleep disturbance in oncology patients and their family transferase gene modulation on suicidal behavior and per- caregivers. PLoS One 2012;7:e40560. sonality traits: review, meta-analysis and association study. J Psychiatr Res 2011;45:309e321. 51. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990;1:43e46. 67. Goertzel BN, Pennachin C, de Souza Coelho L, Gurbaxani B, Maloney EM, Jones JF. Combinations of single 52. Bakan D. The test of significance in psychological e nucleotide polymorphisms in neuroendocrine effector and research. Psychol Bull 1966;66:423 437. receptor genes predict chronic fatigue syndrome. Pharmaco- 53. Sarpeshkar V, Bentley DJ. Adrenergic-beta(2) receptor genomics 2006;7:475e483. polymorphism and athletic performance. J Hum Genet e 68. Seifert T, Secher NH. Sympathetic influence on cere- 2010;55:479 485. bral blood flow and metabolism during exercise in humans. 54. Elenkov IJ, Wilder RL, Chrousos GP, Vizi ES. The sym- Prog Neurobiol 2011;95:406e426. d pathetic nerve an integrative interface between two super- 69. Nugraha B, Karst M, Engeli S, Gutenbrunner C. Brain- systems: the brain and the immune system. Pharmacol Rev derived neurotrophic factor and exercise in fibromyalgia e 2000;52:595 638. syndrome patients: a mini review. Rheumatol Int 2012;32: e 55. Light AR, White AT, Hughen RW, Light KC. Moderate 2593 2599. exercise increases expression for sensory, adrenergic, and 70. Chen R, Liang FX, Moriya J, et al. Chronic fatigue syn- immune genes in chronic fatigue syndrome patients but drome and the central nervous system. J Int Med Res 2008; e not in normal subjects. J Pain 2009;10:1099 1112. 36:867e874. 56. Zhao L, Yang F, Xu K, et al. Common genetic variants 71. Lotrich FE, Albusaysi S, Ferrell RE. Brain-derived neu- of the beta2-adrenergic receptor affect its translational effi- rotrophic factor serum levels and genotype: association ciency and are associated with human longevity. Aging Cell with depression during interferon-alpha treatment. Neuro- e 2012;11:1094 1101. psychopharmacology 2013;38:985e995. 57. Zeng Y, Cheng L, Zhao L, et al. Interactions between so- 72. Ozan E, Okur H, Eker C, Eker OD, Gonul AS, cial/behavioral factors and ADRB2 genotypes may be associ- Akarsu N. The effect of depression, BDNF gene val66met ated with health at advanced ages in China. BMC Geriatr polymorphism and gender on serum BDNF levels. Brain 2013;13:91. Res Bull 2010;81:61e65. 58. So WK, Marsh G, Ling WM, et al. The symptom cluster 73. Terracciano A, Martin B, Ansari D, et al. Plasma BDNF of fatigue, pain, anxiety, and depression and the effect on concentration, Val66Met genetic variant and depression- the quality of life of women receiving treatment for breast related personality traits. Genes Brain Behav 2010;9: cancer: a multicenter study. Oncol Nurs Forum 2009;36: 512e518. E205eE214. 74. Oneda B, Crettol S, Jaquenoud Sirot E, Bochud M, 59. Fox SW, Lyon D, Farace E. Symptom clusters in patients Ansermot N, Eap CB. The P450 genotype with high-grade glioma. J Nurs Scholarsh 2007;39:61e67. is associated with CYP3A activity in vivo as measured by the midazolam phenotyping test. Pharmacogenet Genomics 60. Hwang SS, Chang VT, Rue M, Kasimis B. Multidimen- e sional independent predictors of cancer-related fatigue. 2009;19:877 883. J Pain Symptom Manage 2003;26:604e614. 75. Schirmer M, Rosenberger A, Klein K, et al. Sex-depen- 61. Light KC, White AT, Tadler S, Iacob E, Light AR. Ge- dent genetic markers of CYP3A4 expression and activity in human liver microsomes. Pharmacogenomics 2007;8: netics and gene expression involving stress and distress path- e ways in fibromyalgia with and without comorbid chronic 443 453. fatigue syndrome. Pain Res Treat 2012;2012:427869. 76. Holmes A, Heilig M, Rupniak NM, Steckler T, Griebel G. Neuropeptide systems as novel therapeutic tar- 62. Fernandez-de-Las-Penas C, Cantarero-Villanueva I, Fer- gets for depression and anxiety disorders. Trends Pharmacol nandez-Lao C, et al. Influence of catechol-o- Sci 2003;24:580e588. methyltransferase genotype (Val158Met) on endocrine, sym- pathetic nervous and mucosal immune systems in breast can- 77. Leibowitz SF. Regulation and effects of hypothalamic cer survivors. Breast 2012;21:199e203. galanin: relation to dietary fat, alcohol ingestion, circulating lipids and energy homeostasis. Neuropeptides 2005;39: 63. Fernandez-de-las-Penas C, Fernandez-Lao C, Cantarero- 327e332. Villanueva I, et al. Catechol-O-methyltransferase genotype (Val158met) modulates cancer-related fatigue and pain 78. Misawa K, Ueda Y, Kanazawa T, et al. Epigenetic inacti- sensitivity in breast cancer survivors. Breast Cancer Res Treat vation of galanin receptor 1 in head and neck cancer. Clin 2012;133:405e412. Cancer Res 2008;14:7604e7613. Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 83

79. Wrenn CC, Crawley JN. Pharmacological evidence sup- monitoring-related cortical evoked potentials that are associ- porting a role for galanin in cognition and affect. Prog Neu- ated with ADHD. Biol Psychol 2010;85:19e32. ropsychopharmacol Biol Psychiatry 2001;25:283e299. 98. Yang B, Chan RC, Jing J, Li T, Sham P, Chen RY. 80. Ogren SO, Kuteeva E, Hokfelt T, Kehr J. Galanin recep- A meta-analysis of association studies between the 10- repeat tor antagonists : a potential novel pharmacological treat- allele of a VNTR polymorphism in the 3’-UTR of dopamine ment for mood disorders. CNS Drugs 2006;20:633e654. transporter gene and attention deficit hyperactivity disorder. Am J Med Genet B Neuropsychiatr Genet 2007;144B: 81. Staines D. Are vasoactive neuropeptide autoimmune e fatigue-related disorders mediated via G protein-coupled re- 541 550. ceptors? Med Hypotheses 2005;65:29e31. 99. Lim J, Ebstein R, Tse CY, et al. Dopaminergic polymor- 82. Lotsch J, Klepstad P, Doehring A, Dale O. A GTP cyclo- phisms associated with time-on-task declines and fatigue in 1 genetic variant delays cancer pain. Pain 2010; the Psychomotor Vigilance Test. PLoS One 2012;7:e33767. e 148:103 106. 100. Iqbal R, Mughal MS, Arshad N, Arshad M. Pathophysi- 83. Yasuda Y, Hashimoto R, Ohi K, et al. A functional poly- ology and antioxidant status of patients with fibromyalgia. morphism of the GTP cyclohydrolase 1 gene predicts atten- Rheumatol Int 2011;31:149e152. e tion performance. Neurosci Lett 2014;566:46 49. 101. Lieberman HR, Kellogg MD, Kramer FM, Bathalon GP, 84. Theuns J, Crosiers D, Debaene L, et al. Guanosine Lesher LL. Lipid and other plasma markers are associated triphosphate cyclohydrolase 1 promoter deletion causes with anxiety, depression, and fatigue. Health Psychol 2012; dopa-responsive dystonia. Mov Disord 2012;27:1451e1456. 31:210e216. 85. Kim SK, Kim SH, Nah SS, et al. Association of guano- 102. Yu Y, Pan Y, Jin M, et al. Association of genetic variants sine triphosphate cyclohydrolase 1 gene polymorphisms in tachykinins pathway genes with colorectal cancer risk. Int with fibromyalgia syndrome in a Korean population. J Colorectal Dis 2012;27:1429e1436. J Rheumatol 2013;40:316e322. 103. Cunningham S, O’Doherty C, Patterson C, et al. The 86. Fletcher MA, Rosenthal M, Antoni M, et al. Plasma neu- neuropeptide genes TAC1, TAC3, TAC4, VIP and PACA- ropeptide Y: a biomarker for symptom severity in chronic fa- P(ADCYAP1), and susceptibility to multiple sclerosis. tigue syndrome. Behav Brain Funct 2010;6:76. J Neuroimmunol 2007;183:208e213. 87. Dyzma M, Boudjeltia KZ, Faraut B, Kerkhofs M. Neuro- 104. Colasanti M, Suzuki H. The dual personality of NO. peptide Y and sleep. Sleep Med Rev 2010;14:161e165. Trends Pharmacol Sci 2000;21:249e252. 88. Chambers AP, Woods SC. The role of neuropeptide Y 105. Pall ML. Elevated, sustained peroxynitrite levels as the in energy homeostasis. Handb Exp Pharmacol 2012;209: cause of chronic fatigue syndrome. Med Hypotheses 2000; e 23 45. 54:115e125. 89. Morales-Medina JC, Dumont Y, Quirion R. A possible 106. Angelini C, Tasca E. Fatigue in muscular dystrophies. role of neuropeptide Y in depression and stress. Brain Res Neuromuscul Disord 2012;22 Suppl 3:S214eS220. 2010;1314:194e205. 107. Heydemann A, McNally E. NO more muscle fatigue. 90. Wei J, Chu C, Wang Y, et al. Association study of 45 J Clin Invest 2009;119:448e450. candidate genes in nicotine dependence in Han Chinese. Addict Behav 2012;37:622e626. 108. Pall ML. Post-radiation syndrome as a NO/ONOO- cycle, chronic fatigue syndrome-like disease. Med Hypothe- 91. Okahisa Y, Ujike H, Kotaka T, et al. Association between ses 2008;71:537e541. neuropeptide Y gene and its receptor Y1 gene and metham- phetamine dependence. Psychiatry Clin Neurosci 2009;63: 109. Galecki P, Maes M, Florkowski A, et al. Association be- 417e422. tween inducible and neuronal nitric oxide synthase poly- morphisms and recurrent depressive disorder. J Affect 92. Zhou Y, Danbolt NC. GABA and glutamate transporters e in brain. Front Endocrinol (Lausanne) 2013;4:165. Disord 2011;129:175 182. 93. Thoeringer CK, Ripke S, Unschuld PG, et al. The 110. Kurrikoff T, Lesch KP, Kiive E, et al. Association of a GABA transporter 1 (SLC6A1): a novel candidate gene for functional variant of the nitric oxide synthase 1 gene with personality, anxiety, and depressiveness. Dev Psychopathol anxiety disorders. J Neural Transm (Vienna) 2009;116: e 649e657. 2012;24:1225 1235. 94. Ebejer JL, Duffy DL, van der Werf J, et al. Genome-wide 111. Hancock DB, Martin ER, Vance JM, Scott WK. Nitric association study of inattention and hyperactivity-impulsivity oxide synthase genes and their interactions with environ- measured as quantitative traits. Twin Res Hum Genet 2013; mental factors in Parkinson’s disease. Neurogenetics 2008; e 16:560e574. 9:249 262. 95. Sikora M, Gese A, Czypicki R, et al. Correlations be- 112. Beal MF. Excitotoxicity and nitric oxide in Parkinson’s tween polymorphisms in genes coding elements of dopami- disease pathogenesis. Ann Neurol 1998;44:S110eS114. nergic pathways and body mass index in overweight and e 113. Wen HB, Zhang ZX, Wang H, et al. Epidemiology and obese women. Endokrynol Pol 2013;64:101 107. clinical phenomenology for Parkinson’s disease with pain 96. Foley TE, Fleshner M. Neuroplasticity of dopamine cir- and fatigue. Parkinsonism Relat Disord 2012;18 Suppl 1: cuits after exercise: implications for central fatigue. Neuro- S222eS225. e molecular Med 2008;10:67 80. 114. Robertson D, Flattem N, Tellioglu T, et al. Familial 97. Althaus M, Groen Y, Wijers AA, et al. Variants of the orthostatic tachycardia due to norepinephrine transporter SLC6A3 (DAT1) polymorphism affect performance deficiency. Ann N Y Acad Sci 2001;940:527e543. 84 Eshragh et al. Vol. 53 No. 1 January 2017

115. Morriss RK, Robson MJ, Deakin JF. Neuropsychological 123. Mostowska A, Hozyasz KK, Wojcicka K, Lianeri M, performance and noradrenaline function in chronic fatigue Jagodzinski PP. Polymorphisms of stress-related genes and syndrome under conditions of high arousal. Psychopharma- the risk of nonsyndromic cleft lip with or without cleft pal- cology 2002;163:166e173. ate. Birth Defects Res A Clin Mol Teratol 2011;91:948e955. 116. Shannon JR, Flattem NL, Jordan J, et al. Orthostatic 124. Baldinger P, Kraus C, Rami-Mark C, et al. Interaction intolerance and tachycardia associated with between 5-HTTLPR and 5-HT1B genotype status enhances norepinephrine-transporter deficiency. N Engl J Med 2000; cerebral 5-HT1A receptor binding. Neuroimage 2015;111: 342:541e549. 505e512. 117. Ryu SH, Lee SH, Lee HJ, et al. Association between 125. Lothe A, Boni C, Costes N, et al. 5-HT1A gene pro- norepinephrine transporter gene polymorphism and major moter polymorphism and [18F]MPPF binding potential in depression. Neuropsychobiology 2004;49:174e177. healthy subjects: a PET study. Behav Brain Funct 2010;6:37. 118. Ryan JL, Carroll JK, Ryan EP, et al. Mechanisms of 126. Caspi A, Hariri AR, Holmes A, Uher R, Moffitt TE. Ge- cancer-related fatigue. Oncologist 2007;12 Suppl 1:22e34. netic sensitivity to the environment: the case of the seroto- nin transporter gene and its implications for studying 119. Alexander S, Stone P, White S, Andrews P, Nussey S, complex diseases and traits. Am J Psychiatry 2010;167: Bano G. Evaluation of central serotonin sensitivity in breast 509e527. cancer survivors with cancer-related fatigue syndrome. J Pain e 127. Caspi A, Sugden K, Moffitt TE, et al. Influence of life Symptom Manage 2010;40:892 898. stress on depression: moderation by a polymorphism in 120. Yamamoto S, Ouchi Y, Onoe H, et al. Reduction of se- the 5-HTT gene. Science 2003;301:386e389. rotonin transporters of patients with chronic fatigue syn- 128. Wright F, D’Eramo Melkus G, Hammer M, et al. Trajec- e drome. Neuroreport 2004;15:2571 2574. tories of evening fatigue in oncology outpatients receiving e 121. Narita M, Nishigami N, Narita N, et al. Association be- chemotherapy. J Pain Symptom Manage 2015;50:163 175. tween serotonin transporter gene polymorphism and 129. Wright F, D’Eramo Melkus G, Hammer M, et al. Predic- chronic fatigue syndrome. Biochem Biophys Res Commun tors and trajectories of morning fatigue are distinct from 2003;311:264e266. evening fatigue. J Pain Symptom Manage 2015;50:176e189. 122. Lei X, Chen C, He Q, et al. Sex determines which sec- 130. Lerdal A, Kottorp A, Gay CL, Lee KA. Lee fatigue and tion of the SLC6A4 gene is linked to obsessive-compulsive energy scales: exploring aspects of validity in a sample of symptoms in normal Chinese college students. J Psychiatr women with HIV using an application of a Rasch model. Psy- Res 2012;46:1153e1560. chiatry Res 2013;205:241e246. Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 84.e1

Appendix

Supplementary Table 1 Summary of Single-Nucleotide Polymorphisms Analyzed for Neurotransmitter Genes and the Growth Mixture Model Analyses for Fatigue (i.e., Lower vs. Higher) and Energy (i.e., Higher vs. Lower) Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

Catecholaminergic neurotransmitters Receptors Alpha-1D adrenergic receptor ADRA1D rs3787441 4205059 20 0.268 T > C 2.162 0.339 A 0.886 0.642 A ADRA1D rs6084664 4207929 20 0.159 T > C 0.421 0.810 A 1.962 0.375 A ADRA1D rs2326478 4208246 20 0.326 C > T 2.629 0.269 A 1.064 0.587 A ADRA1D rs835880 4208894 20 0.225 A > G 2.733 0.255 A 0.373 0.830 A ADRA1D rs8183794 4210447 20 0.182 C > T 1.239 0.538 A 0.565 0.754 A ADRA1D rs6116268 4211439 20 0.480 C > T 0.231 0.891 A 0.382 0.826 A ADRA1D rs946188 4215315 20 0.236 A > G 1.912 0.385 A 1.632 0.442 A ADRA1D rs1556832 4215556 20 0.461 C > T 0.036 0.982 A 1.230 0.541 A ADRA1D rs8118409 4216662 20 0.229 G > A 1.469 0.480 A 1.247 0.536 A ADRA1D rs4815670 4216863 20 0.467 G > A 1.559 0.459 A 0.012 0.994 A ADRA1D rs6076639 4219257 20 0.206 C > T 0.024 0.988 A 1.069 0.586 A ADRA1D rs4815675 4223453 20 0.423 T > C 0.441 0.802 A 2.968 0.227 A ADRA1D HapA01 1.244 0.537 0.303 0.859 ADRA1D HapA03 2.644 0.267 0.360 0.835 ADRA1D HapB02 1.672 0.433 1.494 0.474 ADRA1D HapB03 0.272 0.873 0.321 0.852 ADRA1D HapC01 1.608 0.448 0.020 0.990 ADRA1D HapC02 0.668 0.716 0.408 0.816 ADRA1D HapC03 1.469 0.480 1.247 0.536 ADRA1D HapD01 0.217 0.897 2.698 0.260 ADRA1D HapD02 0.477 0.788 0.252 0.881 Alpha-2A adrenergic receptor ADRA2A rs521674 112835589 10 0.364 A > T n/a n/a n/a n/a n/a n/a ADRA2A rs3750625 112839600 10 0.079 C > A 2.186 0.335 A 2.644 0.267 A Beta-2 adrenergic receptor ADRB2 rs2400707 148205051 5 0.401 G > A 3.901 0.142 A 1.770 0.413 A ADRB2 rs11168070 148205926 5 0.357 C > G FE 0.046 D 2.522 0.283 A ADRB2 rs1042718 148206916 5 0.203 C > A FE 0.023 R 2.519 0.284 A ADRB2 rs1042719 148207446 5 0.315 G > C 3.870 0.144 A 1.158 0.560 A ADRB2 HapA01 0.754 0.686 0.538 0.764 ADRB2 HapA02 8.497 0.014 1.642 0.440 ADRB2 HapA05 4.652 0.098 2.265 0.322 Beta-3 adrenergic receptor ADRB3 rs4994 37823797 8 0.092 T > C 2.520 0.284 A 0.959 0.619 A Beta-adrenergic receptor kinase 2 ADRBK2 rs1008673 25994012 22 0.148 A > G 0.651 0.722 A 1.077 0.584 A ADRBK2 rs3817819 26075187 22 0.421 C > T 0.264 0.876 A 2.243 0.326 A ADRBK2 rs5761159 26102307 22 0.438 G > T 0.210 0.901 A 0.325 0.850 A ADRBK2 rs9608416 26111017 22 0.468 A > G 1.498 0.473 A 1.133 0.567 A ADRBK2 HapA01 1.253 0.534 1.533 0.465 ADRBK2 HapA04 0.378 0.828 0.364 0.834 Transporters Solute carrier family 6 member 2dnoradrenaline transporter SLC6A2 rs2242446 55690424 16 0.242 T > C 5.004 0.082 A 2.579 0.275 A SLC6A2 rs17841327 55694252 16 0.321 C > A FE 0.001 R FE 0.038 D SLC6A2 rs3785143 55695105 16 0.087 C > T 3.800 0.150 A FE 0.034 D SLC6A2 rs192303 55700223 16 0.291 G > C 1.667 0.434 A 2.487 0.288 A SLC6A2 rs6499771 55700670 16 0.155 A > G 1.543 0.462 A 1.102 0.576 A SLC6A2 rs36027 55702779 16 0.439 A > G 4.109 0.128 A 6.269 0.044 A SLC6A2 rs36024 55706390 16 0.403 C > T FE 0.033 D FE 0.009 D SLC6A2 rs36021 55711949 16 0.416 T > A 1.343 0.511 A 2.959 0.228 A SLC6A2 rs40147 55716839 16 0.323 C > T 0.664 0.717 A 0.969 0.616 A SLC6A2 rs1814270 55717076 16 0.404 T > C 3.383 0.184 A 0.264 0.876 A SLC6A2 rs36017 55718817 16 0.438 C > G 2.140 0.343 A 0.073 0.964 A SLC6A2 rs3785155 55722389 16 0.138 G > A 5.059 0.080 A 2.325 0.313 A SLC6A2 rs47958 55726461 16 0.433 C > A 1.663 0.435 A 0.937 0.626 A (Continued) 84.e2 Eshragh et al. Vol. 53 No. 1 January 2017

Supplementary Table 1 Continued Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

SLC6A2 rs5568 55730123 16 0.315 A > C 1.182 0.554 A 0.153 0.926 A SLC6A2 rs1566652 55731574 16 0.321 G > T 3.038 0.219 A 0.768 0.681 A SLC6A2 rs5569 55731834 16 0.303 C > T 0.526 0.769 A 0.289 0.866 A SLC6A2 rs998424 55731945 16 0.303 C > T 1.076 0.584 A 0.760 0.684 A SLC6A2 HapA01 11.238 0.004 4.147 0.126 SLC6A2 HapC01 2.585 0.275 0.913 0.634 SLC6A2 HapC10 3.172 0.205 0.141 0.932 SLC6A2 HapD01 0.585 0.746 0.235 0.889 SLC6A2 HapD04 0.955 0.620 0.912 0.634 Solute carrier family 6 member 3ddopamine transporter SLC6A3 rs3863145 1392710 5 0.219 C > T 3.510 0.173 A 0.973 0.615 A SLC6A3 rs40184 1395076 5 0.419 G > A 0.951 0.621 A 1.541 0.463 A SLC6A3 rs11564773 1396812 5 0.052 A > G FE 0.706 A FE 0.557 A SLC6A3 rs6876225 1406035 5 0.035 C > A n/a n/a n/a n/a n/a n/a SLC6A3 rs6347 1411411 5 0.265 A > G 4.120 0.127 A 0.673 0.714 A SLC6A3 rs37022 1415628 5 0.216 T > A 2.452 0.294 A FE 0.015 R SLC6A3 rs2975292 1419931 5 0.447 C > G 3.724 0.155 A 0.318 0.853 A SLC6A3 rs11564758 1420587 5 0.323 G > C 1.653 0.438 A 0.783 0.676 A SLC6A3 rs464049 1423904 5 0.465 T > C 0.512 0.774 A 0.122 0.941 A SLC6A3 rs10053602 1428134 5 0.213 T > C 1.840 0.399 A 0.207 0.902 A SLC6A3 rs463379 1431163 5 0.253 C > G 2.283 0.319 A 2.907 0.234 A SLC6A3 rs403636 1438353 5 0.207 G > T 0.333 0.846 A 0.064 0.969 A SLC6A3 rs6350 1443198 5 0.060 C > T FE 0.212 A FE 0.851 A SLC6A3 rs2937639 1443727 5 0.471 G > A FE 0.032 R 1.541 0.463 A SLC6A3 HapA01 0.635 0.728 0.220 0.896 SLC6A3 HapA07 2.116 0.347 3.043 0.218 SLC6A3 HapA09 1.570 0.456 0.437 0.804 SLC6A3 HapA10 1.786 0.409 0.684 0.710 Synthesis TH rs2070762 2186334 11 0.500 T > C 0.021 0.990 A 0.738 0.691 A TH rs6357 2188237 11 0.243 G > A 0.755 0.686 A 1.481 0.477 A TH rs6356 2190950 11 0.403 G > A 1.687 0.430 A 1.292 0.524 A TH HapA01 1.354 0.508 1.319 0.517 TH HapA02 0.733 0.693 0.005 0.998 TH HapA04 FE 0.555 FE 0.538 Metabolism Catechol-O-methyltransferase COMT rs5748489 19927145 22 0.388 C > A 1.476 0.478 A 0.394 0.821 A COMT rs2020917 19928883 22 0.263 C > T 1.755 0.416 A 0.401 0.818 A COMT rs737866 19930108 22 0.265 A > G 1.569 0.456 A 0.293 0.864 A COMT rs1544325 19931667 22 0.397 G > A 1.210 0.546 A 0.246 0.884 A COMT rs5993882 19937532 22 0.234 T > G 1.818 0.403 A FE 0.034 R COMT rs5993883 19937637 22 0.495 T > G 0.294 0.863 A 0.739 0.691 A COMT rs740603 19945176 22 0.495 G > A 0.581 0.748 A 0.730 0.694 A COMT rs4646312 19948336 22 0.371 T > C 2.192 0.334 A 0.899 0.638 A COMT rs165656 19948862 22 0.489 C > G 0.689 0.709 A 2.150 0.341 A COMT rs6269 19949951 22 0.391 A > G 2.017 0.365 A 0.770 0.680 A COMT rs4633 19950234 22 0.472 C > T 0.669 0.716 A 2.295 0.317 A COMT rs6267 19950262 22 0.002 G > T n/a n/a n/a n/a n/a n/a COMT rs740601 19950762 22 0.399 A > C 2.671 0.263 A 1.847 0.397 A COMT rs5031015 19951102 22 0.001 G > A n/a n/a n/a n/a n/a n/a COMT rs4818 19951206 22 0.387 C > G 1.935 0.380 A 0.704 0.703 A COMT rs4680 19951270 22 0.475 G > A 0.919 0.632 A 2.802 0.246 A COMT rs165774 19952560 22 0.288 G > A 0.705 0.703 A 1.270 0.530 A COMT rs174699 19954457 22 0.098 T > C 3.521 0.172 A 0.875 0.646 A COMT rs9332377 19955691 22 0.129 T > C FE 0.029 D 1.310 0.519 A COMT rs165599 19956780 22 0.338 A > G 0.173 0.917 A 0.123 0.940 A COMT HapA01 0.624 0.732 0.106 0.949 COMT HapA06 1.618 0.445 5.354 0.069 COMT HapA10 1.939 0.379 0.260 0.878 COMT HapB02 0.789 0.674 1.102 0.576 COMT HapB20 1.404 0.496 0.244 0.885 COMT HapC01 0.173 0.917 0.123 0.940 COMT HapC02 4.286 0.117 1.028 0.598 COMT PAIN LPS 2.839 0.242 1.586 0.452 COMT PAIN APS 0.707 0.702 1.737 0.420 (Continued) Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 84.e3

Supplementary Table 1 Continued Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

GABAergic neurotransmission Transporter Solute carrier family 6 member 1dGABA transporter SLC6A1 rs2697149 11036479 3 0.221 T > G 4.406 0.110 A 1.138 0.566 A SLC6A1 rs2601126 11036623 3 0.407 C > T 7.247 0.027 A 9.249 0.010 A SLC6A1 rs1710885 11038806 3 0.192 T > C 0.932 0.627 A 1.105 0.575 A SLC6A1 rs1710886 11039654 3 0.333 G > C 1.273 0.529 A 2.153 0.341 A SLC6A1 rs1710887 11039959 3 0.395 G > T 0.483 0.786 A 0.833 0.659 A SLC6A1 rs9990174 11040438 3 0.326 G > T 3.317 0.190 A 0.073 0.964 A SLC6A1 rs1568072 11041605 3 0.220 C > T 0.107 0.948 A 0.725 0.696 A SLC6A1 rs1728811 11041869 3 0.426 C > T 1.060 0.589 A FE 0.019 D SLC6A1 rs11718132 11045019 3 0.134 G > T 2.251 0.324 A 2.314 0.314 A SLC6A1 rs2697144 11051098 3 0.251 A > G 1.457 0.483 A 2.461 0.292 A SLC6A1 rs2928079 11055113 3 0.425 A > T 0.727 0.695 A 5.034 0.081 A SLC6A1 rs1170695 11055337 3 0.309 T > C 0.011 0.994 A 3.093 0.213 A SLC6A1 rs2933308 11055623 3 0.366 G > A 0.001 0.999 A 5.113 0.078 A SLC6A1 rs10510403 11066669 3 0.141 A > G 1.210 0.546 A 3.984 0.136 A SLC6A1 rs2675163 11075013 3 0.231 T > C 4.566 0.102 A FE 0.007 D SLC6A1 rs10514669 11075911 3 0.194 C > T 3.846 0.146 A 0.410 0.815 A SLC6A1 rs2697138 11076906 3 0.145 C > A 3.498 0.174 A 0.902 0.637 A SLC6A1 rs1062246 11080168 3 0.417 A > G 0.752 0.686 A 6.731 0.035 A SLC6A1 HapA01 7.675 0.022 9.703 0.008 SLC6A1 HapA02 5.233 0.073 3.699 0.157 SLC6A1 HapA04 4.102 0.129 0.977 0.614 SLC6A1 HapB01 0.350 0.840 3.012 0.222 SLC6A1 HapB03 1.060 0.589 7.111 0.029 SLC6A1 HapC01 0.001 0.999 5.113 0.078 SLC6A1 HapC02 0.039 0.981 1.084 0.582 SLC6A1 HapC03 0.011 0.994 3.093 0.213 SLC6A1 HapD01 0.142 0.932 9.291 0.010 SLC6A1 HapD02 0.083 0.960 4.475 0.107 Serotonergic neurotransmission Receptors 5-Hydroxytryptamine receptor 1A HTR1A rs6449693 63256017 5 0.437 A > G 1.721 0.423 A 0.092 0.955 A 5-Hydroxytryptamine receptor 1B HTR1B rs6296 78172259 6 0.313 G > C 1.927 0.382 A 3.466 0.177 A 5-Hydroxytryptamine receptor 2A HTR2A rs6314 47409033 13 0.078 C > T 4.127 0.127 A 0.589 0.745 A HTR2A rs7322347 47410102 13 0.420 T > A 0.976 0.614 A 1.289 0.525 A HTR2A rs1923882 47411660 13 0.223 C > T 1.334 0.513 A 3.683 0.159 A HTR2A rs7997012 47411984 13 0.380 G > A 1.799 0.407 A 0.053 0.974 A HTR2A rs3742278 47419576 13 0.189 A > G 0.134 0.935 A 0.032 0.984 A HTR2A rs1923884 47421835 13 0.167 C > T 0.410 0.815 A 0.083 0.959 A HTR2A rs1923886 47423290 13 0.427 T > C 3.229 0.199 A 0.284 0.868 A HTR2A rs7330636 47423591 13 0.364 C > T 2.559 0.278 A 2.276 0.320 A HTR2A rs9567739 47424943 13 0.374 G > C 0.777 0.678 A 0.180 0.914 A HTR2A rs2296972 47428470 13 0.330 G > T 2.576 0.276 A 1.113 0.573 A HTR2A rs9534495 47429227 13 0.114 A > G FE 0.889 A FE 0.384 A HTR2A rs9534496 47431107 13 0.182 G > C 4.086 0.130 A 6.131 0.047 A HTR2A rs4942578 47432609 13 0.264 G > T 0.672 0.715 A 0.393 0.822 A HTR2A rs2770292 47435105 13 0.162 C > G 1.095 0.578 A 0.787 0.675 A HTR2A rs1928042 47437215 13 0.218 A > C 1.288 0.525 A 2.168 0.338 A HTR2A rs2770293 47438973 13 0.376 C > T 1.724 0.422 A 2.884 0.236 A HTR2A rs1328674 47441706 13 0.044 G > A n/a n/a n/a n/a n/a n/a HTR2A rs2770298 47446846 13 0.260 T > C 1.500 0.472 A 0.512 0.774 A HTR2A rs1928040 47447235 13 0.480 T > C 3.163 0.206 A FE 0.044 R HTR2A rs972979 47449163 13 0.373 G > A 0.861 0.650 A 0.310 0.856 A HTR2A rs731779 47452037 13 0.171 T > G 2.539 0.281 A 0.810 0.667 A HTR2A rs2770304 47455364 13 0.333 A > G 0.124 0.940 A 0.749 0.688 A HTR2A rs927544 47456050 13 0.255 T > C 0.687 0.709 A 1.617 0.445 A HTR2A rs594242 47458051 13 0.169 C > G 2.157 0.340 A 0.357 0.837 A HTR2A rs4941573 47464856 13 0.447 A > G 2.597 0.273 A 0.611 0.737 A HTR2A rs1328684 47466229 13 0.314 T > C 2.964 0.227 A 1.412 0.494 A HTR2A rs6304 47466548 13 0.010 A > G n/a n/a n/a n/a n/a n/a HTR2A rs2296973 47466780 13 0.281 G > T 0.486 0.784 A 0.144 0.930 A (Continued) 84.e4 Eshragh et al. Vol. 53 No. 1 January 2017

Supplementary Table 1 Continued Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

HTR2A rs2070037 47467069 13 0.216 T > C 0.222 0.895 A 0.603 0.740 A HTR2A rs9534511 47468579 13 0.445 C > T 3.026 0.220 A 0.195 0.907 A HTR2A rs6313 47469939 13 0.450 C > T 3.482 0.175 A 0.891 0.640 A HTR2A HapA03 1.394 0.498 3.597 0.166 HTR2A HapA07 1.790 0.409 0.517 0.772 HTR2A HapB01 0.296 0.862 0.034 0.983 HTR2A HapB02 2.558 0.278 2.390 0.303 HTR2A HapB03 3.229 0.199 0.284 0.868 HTR2A HapC01 1.400 0.497 2.117 0.347 HTR2A HapC05 2.004 0.367 0.646 0.724 HTR2A HapD01 0.672 0.715 0.393 0.822 HTR2A HapD02 0.617 0.734 0.810 0.667 HTR2A HapE01 1.288 0.525 2.168 0.338 HTR2A HapF01 3.173 0.205 4.584 0.101 HTR2A HapF02 4.177 0.124 1.625 0.444 HTR2A HapF03 1.500 0.472 0.512 0.774 HTR2A HapG01 0.456 0.796 0.687 0.709 HTR2A HapH01 2.201 0.333 0.306 0.858 HTR2A HapH06 1.304 0.521 2.825 0.244 HTR2A HapI01 3.916 0.141 0.835 0.659 5-Hydroxytryptamine receptor 3A HTR3A rs1985242 113848272 11 0.370 T > A 1.548 0.461 A 0.170 0.919 A HTR3A rs11214796 113854678 11 0.261 T > C 0.845 0.655 A 1.080 0.583 A HTR3A rs10160548 113856680 11 0.378 T > G 2.139 0.343 A 0.480 0.787 A HTR3A HapA01 1.214 0.545 0.206 0.902 HTR3A HapA04 1.218 0.544 0.763 0.683 Transporter Solute carrier family 6 member 4dserotonin transporter SLC6A4 rs3813034 28524803 17 0.476 A > C 0.931 0.628 A 1.254 0.534 A SLC6A4 rs1042173 28525010 17 0.478 T > G 0.808 0.667 A 1.473 0.479 A SLC6A4 rs4325622 28526474 17 0.473 T > C 0.846 0.655 A 0.974 0.614 A SLC6A4 rs3794808 28531792 17 0.469 G > A 0.394 0.821 A 2.851 0.240 A SLC6A4 rs140701 28538531 17 0.464 G > A 0.270 0.874 A 3.873 0.144 A SLC6A4 rs140700 28543388 17 0.089 G > A 0.995 0.608 A 1.979 0.372 A SLC6A4 rs2020942 28546913 17 0.346 G > A 0.611 0.737 A FE 0.016 R SLC6A4 rs8076005 28547209 17 0.214 A > G 2.102 0.350 A FE 0.018 D SLC6A4 rs6354 28549897 17 0.180 A > C 0.617 0.735 A FE 0.041 D SLC6A4 rs2066713 28551664 17 0.345 C > T 0.999 0.607 A 4.337 0.114 A SLC6A4 HapA01 0.386 0.825 2.173 0.337 SLC6A4 HapA11 0.432 0.806 2.324 0.313 SLC6A4 HapB01 0.041 0.980 2.573 0.276 SLC6A4 HapB04 0.535 0.765 6.867 0.032 SLC6A4 5HTTLPR 3.365 0.186 A 2.566 0.277 A SLC6A4 5HTTLPR þ rs25531 FE 0.013 D 4.461 0.107 A Synthesis Trytophan hydroxylase 2 TPH2 rs11179000 72338627 12 0.268 A > T 0.383 0.826 A 3.130 0.209 A TPH2 rs7955501 72350025 12 0.357 A > T 0.108 0.948 A 2.892 0.236 A TPH2 rs1487275 72410291 12 0.259 T > G 0.097 0.952 A 0.154 0.926 A Molecular transport and drug metabolism ATP-binding cassette, subfamily B (MDR/TAP) member 1 ABCB1 rs2235048 87138510 7 0.471 T > C 0.100 0.951 A 0.297 0.862 A ABCB1 rs6961419 87172135 7 0.400 T > C 0.994 0.608 A 0.379 0.828 A ABCB1 rs1128503 87179600 7 0.433 C > T 1.306 0.520 A 0.129 0.938 A ABCB1 rs1922241 87185893 7 0.299 G > A 2.837 0.242 A 3.502 0.174 A ABCB1 rs10264990 87202614 7 0.293 T > C 0.868 0.648 A 1.805 0.405 A ABCB1 rs1989830 87205662 7 0.309 C > T 2.162 0.339 A 1.293 0.524 A ABCB1 rs1858923 87221215 7 0.445 T > C 0.027 0.987 A 0.960 0.619 A ABCB1 rs9282564 87229439 7 0.089 A > G 2.773 0.250 A 0.744 0.689 A ABCB1 rs13233308 87244959 7 0.438 C > T 0.438 0.803 A 1.249 0.535 A ABCB1 rs10267099 87278759 7 0.213 A > G 4.187 0.123 A 0.050 0.975 A ABCB1 HapA01 1.328 0.515 0.104 0.949 ABCB1 HapA05 2.796 0.247 3.493 0.174 ABCB1 HapB01 0.574 0.751 1.448 0.485 ABCB1 HapB02 0.312 0.855 1.712 0.425 Cytochrome P450, family 3, subfamily A, polypeptide 4 (Continued) Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 84.e5

Supplementary Table 1 Continued Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

CYP3A4 rs4646437 99365082 7 0.163 C > T FE 0.031 D 1.013 0.602 A Genes involved in various aspects of neurotransmission Brain-derived neurotrophic factor BDNF rs7124442 27677040 11 0.290 T > C 4.385 0.112 A 2.754 0.252 A BDNF rs6265 27679915 11 0.222 G > A FE 0.042 D 2.636 0.268 A BDNF rs11030101 27680743 11 0.409 A > T 2.614 0.271 A 0.792 0.673 A BDNF rs11030102 27681595 11 0.205 C > G 6.132 0.047 A 3.223 0.200 A BDNF rs11030104 27684516 11 0.233 A > G 4.247 0.120 A 3.035 0.219 A BDNF rs2049045 27694240 11 0.156 G > C 2.743 0.254 A 1.505 0.471 A BDNF rs11030107 27694834 11 0.205 A > G 6.253 0.044 A 3.303 0.192 A BDNF rs7103411 27700124 11 0.243 T > C 5.365 0.068 A 3.811 0.149 A BDNF rs16917237 27702382 11 0.231 G > T 4.987 0.083 A 2.871 0.238 A BDNF rs6484320 27703187 11 0.243 A > T 5.365 0.068 A 3.811 0.149 A BDNF rs7127507 27714883 11 0.295 T > C 3.149 0.207 A 2.151 0.341 A BDNF rs2049046 27723774 11 0.464 A > T 2.727 0.256 A 1.592 0.451 A BDNF HapA01 2.384 0.304 0.673 0.714 Galanin GAL rs694066 68452984 11 0.104 G > A 0.433 0.805 A 0.571 0.751 A GAL rs3136540 68456409 11 0.249 C > T 1.936 0.380 A 0.753 0.686 A GAL rs1042577 68458469 11 0.334 G > A 2.473 0.290 A 3.691 0.158 A GAL HapA01 2.443 0.295 3.508 0.173 GAL HapA04 1.838 0.399 0.762 0.683 Galanin receptor 1 GALR1 rs949060 74958937 18 0.381 G > C FE 0.017 R 4.518 0.104 A Galanin receptor 2 GALR2 rs2443168 74066446 17 0.443 T > A 1.118 0.572 A 0.066 0.968 A GALR2 rs2598414 74067098 17 0.391 C > T 0.043 0.979 A 1.314 0.518 A GALR2 HapA01 0.043 0.979 1.314 0.518 GALR2 HapA03 1.333 0.513 0.085 0.958 GTP cyclohydrolase 1 GCH1 rs7142517 55306803 14 0.297 C > A 1.053 0.591 A 0.077 0.962 A GCH1 rs841 55310491 14 0.236 C > T 1.116 0.572 A 2.634 0.268 A GCH1 rs752688 55311568 14 0.236 C > T 1.116 0.572 A 2.634 0.268 A GCH1 rs7155309 55322850 14 0.234 T > C 1.085 0.581 A 2.556 0.279 A GCH1 rs12587434 55325582 14 0.236 T > G 0.675 0.713 A 2.942 0.230 A GCH1 rs9671371 55328634 14 0.337 C > T 3.455 0.178 A 0.920 0.631 A GCH1 rs2183081 55336750 14 0.409 T > C 0.566 0.754 A 3.866 0.145 A GCH1 rs17128050 55343878 14 0.148 T > C 2.307 0.316 A 3.276 0.194 A GCH1 rs3783637 55348117 14 0.155 C > T 3.040 0.219 A 3.939 0.140 A GCH1 rs3783638 55348372 14 0.187 G > A 4.716 0.095 A 2.287 0.319 A GCH1 rs998259 55355030 14 0.168 C > T 4.400 0.111 A 2.700 0.259 A GCH1 rs3783642 55360202 14 0.461 T > C FE 0.003 D 3.142 0.208 A GCH1 HapA01 1.855 0.395 2.752 0.253 GCH1 HapA05 1.122 0.571 2.616 0.270 GCH1 HapA06 1.100 0.577 0.091 0.955 GCH1 HapB01 1.922 0.383 2.672 0.263 GCH1 HapB03 5.712 0.058 2.942 0.230 Nitric oxide synthase 1 NOS1 rs2682826 117652837 12 0.311 C > T 0.946 0.623 A 0.143 0.931 A NOS1 rs816361 117655130 12 0.318 C > G 1.353 0.508 A 0.359 0.836 A NOS1 rs816363 117660466 12 0.458 C > G FE 0.042 D 1.261 0.532 A NOS1 rs9658498 117668524 12 0.409 T > C FE 0.041 R 1.186 0.553 A NOS1 rs1353939 117675352 12 0.261 G > A 1.739 0.419 A 1.016 0.602 A NOS1 rs1047735 117685269 12 0.346 C > T 1.896 0.387 A 1.227 0.542 A NOS1 rs12829185 117694019 12 0.243 C > T FE 0.025 R 0.295 0.863 A NOS1 rs2293054 117701713 12 0.299 G > A 3.410 0.182 A 3.123 0.210 A NOS1 rs6490121 117708194 12 0.364 A > G 2.852 0.240 A 0.548 0.760 A NOS1 rs2293052 117715619 12 0.358 C > T FE 0.001 R 1.925 0.382 A NOS1 rs7977109 117730339 12 0.418 A > G 0.957 0.620 A 1.830 0.401 A NOS1 rs3782206 117745088 12 0.116 C > T 4.155 0.125 A 1.577 0.455 A NOS1 rs7295972 117747367 12 0.445 G > A 1.150 0.563 A 0.033 0.984 A NOS1 rs11068447 117747686 12 0.124 C > T 2.235 0.327 A 1.408 0.495 A NOS1 rs547954 117754505 12 0.206 C > T 1.188 0.552 A 1.085 0.581 A NOS1 rs3782212 117755401 12 0.270 C > T 0.320 0.852 A 0.069 0.966 A NOS1 rs12578547 117763346 12 0.266 T > C 2.973 0.226 A 0.642 0.726 A NOS1 rs471871 117765517 12 0.246 A > T 3.255 0.196 A FE 0.039 R (Continued) 84.e6 Eshragh et al. Vol. 53 No. 1 January 2017

Supplementary Table 1 Continued Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

NOS1 rs545654 117777048 12 0.496 T ¼ C 0.913 0.633 A 1.220 0.543 A NOS1 rs1552227 117779034 12 0.257 C > T 3.750 0.153 A 0.096 0.953 A NOS1 rs10507279 117780273 12 0.122 G > A 0.206 0.902 A 1.890 0.389 A NOS1 rs693534 117784717 12 0.382 G > A 2.797 0.247 A 0.826 0.662 A NOS1 rs1123425 117786104 12 0.439 A > G 12.001 0.002 A 3.429 0.180 A NOS1 rs3782221 117795880 12 0.270 G > A 4.488 0.106 A 1.189 0.552 A NOS1 HapA02 3.993 0.136 0.878 0.645 NOS1 HapA04 1.299 0.522 0.429 0.807 NOS1 HapB02 1.739 0.419 1.016 0.602 NOS1 HapB03 5.382 0.068 1.186 0.553 NOS1 HapC01 1.896 0.387 1.227 0.542 NOS1 HapC03 6.036 0.049 0.295 0.863 NOS1 HapD01 9.804 0.007 2.102 0.350 NOS1 HapD02 2.376 0.305 0.208 0.901 NOS1 HapD03 1.204 0.548 1.340 0.512 NOS1 HapE01 5.117 0.077 1.019 0.601 NOS1 HapE03 1.150 0.563 0.033 0.984 NOS1 HapF01 1.914 0.384 2.593 0.273 NOS1 HapF02 4.204 0.122 1.325 0.516 NOS1 HapF04 1.045 0.593 1.299 0.522 NOS1 HapF06 3.648 0.161 2.065 0.356 Nitric oxide synthase 2 NOS2A rs9906835 26089373 17 0.413 A > G 0.061 0.970 A 1.357 0.507 A NOS2A rs2297512 26092554 17 0.385 A > G 1.117 0.572 A 1.156 0.561 A NOS2A rs2297516 26095729 17 0.416 A > C 1.368 0.505 A 0.896 0.639 A NOS2A rs2297518 26096596 17 0.145 G > A 1.666 0.435 A 3.351 0.187 A NOS2A rs2248814 26100320 17 0.393 G > A 0.840 0.657 A 2.349 0.309 A NOS2A rs1137933 26105931 17 0.170 C > T 0.889 0.641 A 0.817 0.665 A NOS2A rs4795067 26106674 17 0.278 A > G 0.266 0.876 A 0.278 0.870 A NOS2A rs3729508 26109029 17 0.422 G > A 1.730 0.421 A 2.031 0.362 A NOS2A rs944725 26109570 17 0.382 C > T 0.689 0.709 A FE 0.034 D NOS2A rs3730013 26125917 17 0.342 C > T 2.534 0.282 A 0.021 0.990 A NOS2A rs10459953 26127517 17 0.366 G > C 4.128 0.127 A 1.975 0.372 A NOS2A rs2779248 26127831 17 0.347 T > C 0.452 0.798 A 0.340 0.844 A NOS2A HapA01 1.518 0.468 1.018 0.601 NOS2A HapA04 1.088 0.580 1.051 0.591 NOS2A HapB01 2.660 0.265 3.648 0.161 NOS2A HapB02 1.177 0.555 1.658 0.436 NOS2A HapC01 0.649 0.723 0.520 0.771 NOS2A HapC02 3.667 0.160 1.396 0.498 NOS2A HapC03 2.487 0.288 0.123 0.940 Neuropeptide Y NPY rs16148 24322337 7 0.424 T > C 0.669 0.716 A 1.443 0.486 A NPY rs16147 24323409 7 0.496 A > G 0.401 0.818 A 0.807 0.668 A NPY rs16478 24324607 7 0.290 C > T 1.867 0.393 A 2.658 0.265 A NPY rs16139 24324878 7 0.029 A > G n/a n/a n/a n/a n/a n/a NPY rs1468271 24326980 7 0.027 A > G n/a n/a n/a n/a n/a n/a NPY rs5574 24329132 7 0.429 C > T 0.466 0.792 A 0.667 0.717 A NPY HapA01 0.621 0.733 0.501 0.779 NPY HapA04 2.576 0.276 1.807 0.405 NPY HapA05 1.867 0.393 2.658 0.265 Neuropeptide Y receptor Y1 NPY1R rs9764 164245404 4 0.282 T > C 2.934 0.231 A 4.479 0.107 A NPY1R rs7687423 164250796 4 0.410 G > A FE 0.008 D 2.179 0.336 A NPY1R HapA01 3.258 0.196 4.432 0.109 NPY1R HapA04 7.788 0.020 2.296 0.317 Prodynorphin PDYN rs6045868 1967277 20 0.334 G > A 2.867 0.239 A 0.441 0.802 A PDYN rs2235751 1969933 20 0.361 G > A 0.229 0.892 A 1.737 0.420 A Tachykinin precursor 1 TAC1 rs7793277 97359584 7 0.267 C > G 0.788 0.674 A 2.257 0.323 A TAC1 rs2072100 97361783 7 0.476 A > GFE<0.0001 R FE 0.019 R TAC1 rs1229434 97365841 7 0.429 A > G FE 0.004 R 3.535 0.171 A TAC1 rs4526299 97367628 7 0.195 C > T 1.723 0.422 A 3.950 0.139 A TAC1 HapA01 9.317 0.009 3.141 0.208 TAC1 HapA05 1.700 0.427 4.030 0.133 (Continued) Vol. 53 No. 1 January 2017 Neurotransmitter Genes and Fatigue and Energy 84.e7

Supplementary Table 1 Continued Fatigue Energy

Gene SNP Position Chr MAF Alleles Chi Square P-Value Model Chi Square P-Value Model

TAC1 HapA06 0.857 0.651 2.232 0.328 Tachykinin receptor 1 TACR1 rs1106855 75277986 2 0.243 G > A 3.306 0.191 A 0.598 0.741 A TACR1 rs4439987 75287105 2 0.385 A > G 0.259 0.879 A 0.461 0.794 A TACR1 rs11688000 75293156 2 0.390 A > G 0.135 0.935 A 0.174 0.917 A TACR1 rs6546952 75301762 2 0.399 T > C 0.374 0.829 A 0.362 0.834 A TACR1 rs17564182 75302305 2 0.224 C > G 1.544 0.462 A 3.227 0.199 A TACR1 rs3771810 75307652 2 0.167 T > C 0.303 0.860 A 1.636 0.441 A TACR1 rs34242711 75321179 2 0.199 G > A 3.557 0.169 A 1.303 0.521 A TACR1 rs2111378 75354603 2 0.315 C > T 1.493 0.474 A 2.217 0.330 A TACR1 rs3771825 75355479 2 0.197 C > T 0.748 0.688 A 1.295 0.523 A TACR1 rs3771827 75361863 2 0.453 T > C n/a n/a n/a n/a n/a n/a TACR1 rs741418 75363185 2 0.440 A > G 5.424 0.066 A 2.041 0.360 A TACR1 rs9808455 75369568 2 0.479 T > C 1.683 0.431 A 2.945 0.229 A TACR1 rs3771836 75380951 2 0.484 T > G 0.595 0.743 A 3.431 0.180 A TACR1 rs759588 75384548 2 0.378 C > T 1.174 0.556 A 3.500 0.174 A TACR1 rs3821318 75387310 2 0.458 C > T 2.240 0.326 A 1.267 0.531 A TACR1 rs6733933 75387833 2 0.189 A > G 0.007 0.996 A 0.040 0.980 A TACR1 rs13428269 75395778 2 0.169 C > T 0.333 0.847 A FE 0.018 R TACR1 rs3771853 75401613 2 0.407 C > T 0.808 0.668 A 0.092 0.955 A TACR1 rs12477554 75402064 2 0.462 G > A 2.724 0.256 A 0.624 0.732 A TACR1 rs4853116 75411277 2 0.334 A > G 1.126 0.570 A 0.301 0.860 A TACR1 rs3821320 75414091 2 0.410 A > G 1.363 0.506 A 0.755 0.685 A TACR1 rs4853119 75416295 2 0.229 T > C 1.598 0.450 A 1.375 0.503 A TACR1 rs3771863 75419713 2 0.195 C > T 2.710 0.258 A 2.490 0.288 A TACR1 HapA01 3.414 0.181 0.394 0.821 TACR1 HapA04 0.342 0.843 0.668 0.716 TACR1 HapB01 1.493 0.474 2.217 0.330 TACR1 HapB02 1.846 0.397 0.081 0.961 TACR1 HapB03 0.748 0.688 1.295 0.523 TACR1 HapC01 5.149 0.076 1.896 0.388 TACR1 HapC04 1.691 0.429 2.971 0.226 TACR1 HapD03 0.416 0.812 4.246 0.120 TACR1 HapD05 1.174 0.556 3.500 0.174 TACR1 HapE01 2.217 0.330 0.617 0.735 TACR1 HapE04 1.250 0.535 0.124 0.940 A ¼ additive model; ABCB ¼ ATP-binding cassette, subfamily B (MDR/TAP) member 1; ADRA1D ¼ adrenergic, alpha-1D receptor; ADRA2A ¼ adrenergic, alpha- 2A receptor; ADRB2 ¼ adrenergic, beta-2 receptor, surface; ADRB3 ¼ adrenergic, beta 3 receptor; ADRBK2 ¼ adrenergic, beta, receptor kinase 2; BDNF ¼ brain- derived neurotrophic factor; Chr ¼ chromosome; COMT ¼ catechol-O-methyltransferase; CYP3A4 ¼ cytochrome P450, family 3, subfamily A, polypeptide 4; D ¼ dominant model; FE ¼ Fisher’s Exact; GAL ¼ galanin; GALR1 ¼ galanin receptor 1; GALR2 ¼ galanin receptor 2; GCH1 ¼ GTP cyclohydrolase 1; Hap ¼ haplotype; HTR1A ¼ 5-hydroxytryptamine receptor 1A, G proteinecoupled; HTR1B ¼ 5-hydroxytryptamine receptor 1B, G protein coupled; HTR2A ¼ 5-hydroxytryptamine receptor 2A, G protein coupled; HTR3A ¼ 5-hydroxytryptamine receptor 3A, ionotropic; MAF ¼ minor allele frequency; n/ a ¼ not assayed because SNP violated Hardy-Weinberg expectations (P < 0.001) or because MAF was <0.05; NOS1 ¼ nitric oxide synthase 1; NOS2A ¼ nitric oxide synthase 2, inducible; NPY ¼ neuropeptide Y; NPY1R ¼ neuropeptide Y receptor Y1; PDYN ¼ prodynorphin; R ¼ recessive model; SLC6A1 ¼ solute carrier family 6 (neurotransmitter transporter, GABA) member 1; SLC6A2 ¼ solute carrier family 6 (neurotransmitter transporter, noradrenaline) member 2; SLC6A3 ¼ solute carrier family 6 (neurotransmitter transporter, dopamine) member 3; SLC6A4 ¼ solute carrier family 6 (neurotransmitter transporter, seroto- nin) member 4; SNP ¼ single-nucleotide polymorphism; TAC ¼ tachykinin, precursor 1; TACR1 ¼ tachykinin receptor 1; TH ¼ tyrosine hydroxylase; TPH2 ¼ 2.