Variants associated with tacrolimus troughs in European American kidney transplant recipients: A genome wide association study Jacobson PA 1 ; Miller MB 2 , Schladt D 3 , Israni A 4 , Sanghavi K 1 , Dorr C 4 , Remmel RP 5 , Guan W 6 ; Matas AJ 7 ; Oetting WS 1 for the DeKAF and GEN03 investigators 1 Experimental and Clinical Pharmacology, College of Pharmacy, University of Minnesota, 2 Department of Psychology, University of Minnesota, 3 Minneapolis Medical Research Foundation, Hennepin County Medical Center, Minnesota, 4 Department of Nephrology, Hennepin County Medical Center and University of Minnesota, 5 Department of Medicinal Chemistry, College of Pharmacy, University of Minnesota, 6 Department of Biostatistics, University of Minnesota 7 Department of Surgery, University of Minnesota RESULTS CONCLUSION We identiTied CYP3A5*3, CYP3A4*22 and CYP3A4*3 as top most variants important towards tacrolimus trough using GWAS. Days posttransplant, recipient age, GRF at time of trough, weight at baseline, primary disease, calcium channel blocker use, aceinhibitor use and antiviral use were clinical factors signiTicant towards tacrolimus dosenormalized trough concentrations. These variants should be tested in future studies in European Americans. INTRODUCTION Tacrolimus is an immunosuppressive agent highly dependent on CYP3A4 and CYP3A5 for its metabolism.[1,2] Variability in its pharmacokinetics and narrow therapeutics index necessitates close monitoring of tacrolimus concentrations. Compared to African Americans, European Americans require lower doses to achieve target concentrations.[2,3] In our previous study, conducted in 695 kidney transplant recipients, clinical factors and genotypes together explained ~50% of variability in dose normalized trough concentrations.[4] Majority of European Americans (~9095%) carry the nonfunctional CYP3A5*3 allele and hence are poor CYP3A5 substrate metabolizers. In our study the CYP3A5*3 was the most inTluential variant. [1,4] After accounting for CYP3A5*3, a large part of the pharmacokinetic variability remained unexplained. We hypothesize that there are additional genetic variants that further inTluence tacrolimus variability. The purpose of this study was to perform a genome wide search to Tind additional genetic variants speciTic to tacrolimus metabolism in the the European American population. METHODS REFERENCES 1. Barbarino JM, Staatz CE, Venkataramanan R, Klein TE, Altman RB. Pharmacogenet Genomics. 2013;23:56385 2. Staatz CE, Tett SE. Clin Pharmacokinet. 2004;43:62353. 3. Venkataramanan R, Swaminathan A, Prasad T, Jain A, Zuckerman S, Warty V, et al. Clin Pharmacokinet. 1995;29:40430. 4. Jacobson PA, Oetting WS, Brearley AM, Leduc R, Guan W, Schladt D, et al. Transplantation. 2011;91:3008 ACKNOWLEDGMENTS The study was funded by National Institute of Allergy and Infectious Disease. We acknowledge the dedication and hard work of our coordinators: Nicoleta Bobocea, Tina Wong, Adrian Geambasu, Alyssa Sader, Myrna Ross, Kathy Peters, Mandi DeGrote, Jill Nagorski, Lisa Berndt, Tom DeLeeuw, Wendy Wallace, Tammy Lowe, Catherine Barker, and Tena Hilario. We also acknowledge the dedicated work of our research scientists: Marcia Brott, Becky Willaert and Amutha Muthuswamy. Table 1: Patient Characteristics . N No. of subjects 1446 No. of trough concentrations 25255 No. of male subjects (%) 908 (62.79) Daily dose (mg) a 5.50 (0.1036.00) Tacrolimus trough (ng/mL) a 8.40 (0.3075.90) No. of subjects in each baseline weight group (kg) (%) 069 7081 82 to 95 >95 350 (24.20) 330(22.82) 370 (25.59) 396 (27.39) No. of recipients in each age category (%) 1834 years 3564 years >64 years 165 (11.41) 1048 (72.48) 233 (16.11) No. of troughs with antiviral drug (%) 14127 (55.94) No. of troughs with steroid (%) 15915 (63.02) No. of troughs with calcium channel blocker (%) 9213 (36.48) No. of troughs with ace inhibitor use (%) 3238 (12.82) Diabetes at baseline (%) 564 (39.00) Simultaneous kidneypancreas transplant (%) 120 (8.30) Donor status Living (%) Deceased(%) 961 (66.46) 485 (33.54) Variable Group Estimate (95% CI) pvalue For each day posttransplant b 1.06(1.061.07) 7.60X10 70 Additional effect for each day after day 9 posttransplant c 0.94(0.930.95) 9.00X10 68 Age in yrs compared to >64 yrs 1834 0.75(0.690.81) 6.30X10 12 3564 0.87(0.830.93) 6.60X10 06 GFR compared to >84 ml/min 55 1.06(1.021.09) 4.30X10 04 5668 1.04(1.011.07) 2.00X10 03 6984 1.03(1.011.06) 1.70X10 03 Weight in kg compared to >95 kg 69 1.05(1.011.09) 2.50X10 02 7081 1.06(1.021.1) 1.20X10 03 8295 1.04(1.011.07) 1.30X10 02 Diabetes at time of transplant 1.11(1.061.16) 7.20X10 06 Living donor 1.06(1.011.11) 1.40X10 02 Male donor 1.03(0.991.08) 1.30X10 01 Steroid use at time of trough 0.98(0.951.01) 1.10X10 01 Ca Channel blocker use at time of trough 1.05(1.031.07) 3.10X10 09 ACEinhibitor use at time of trough 0.97(0.950.99) 6.00X10 03 Antiviral use at time of trough 1.04(1.031.05) 2.70X10 11 Antibody induction combination 0.87(0.760.99) 2.90X10 02 monoclonal 0.98(0.931.04) 5.20X10 01 none 1.31(1.161.49) 2.10X10 05 CYP3A5*3 (effect of 1 G allele) 1.85 (1.741.20) 2.30X10 92 CYP3A4*22 (effect of 1 T allele) 1.28(1.21.38) 1.90X10 12 CYP3A4*3 (effect of 1 C allele) 1.37(1.151.62) 3.00X10 04 Table 3: Final regression model for the tacrolimus dose normalized troughs in Virst 6 months posttransplant a Subjects were European American kidney transplant recipients (n=1446) enrolled in our multicenter DEKAF genomics study (NCT00270712) who received tacrolimus maintenance therapy. Tacrolimus trough concentrations were obtained from each subject in the Tirst 6months (twice each week for the Tirst 2 months and then twice in each month up to 6 months). Trough concentrations were targeted to 812 ng/mL for the Tirst 3 months and 610 ng/mL for 36 months posttransplant. Table 1 shows demographic and clinical characteristics of the subjects. Genotyping: DNA was from peripheral blood and genotyped using a custom exomeplus Affymetrix TxArray SNP chip containing 450,130 markers after QC. Data quality control was conducted using PLINK software. Samples were dropped if they had less than 98% call rate, were monomorphic, did not pass Hardy Weinberg equilibrium testing, Hapmap concordance rate > 2%, gender mismatch, or were identical by descent, had minor allele frequency <1%. Principal component analysis and visual inspection was used to conTirm European ancestry. rs number (variant) Allele Frequencies rs776746 (3A5*3) G=93.20% A=6.80% rs35599367 (3A4*22) C=94.36% T=5.64% rs4986910 (3A4*3) T=99.15% C=0.85% Table 2: Genotype frequencies of the top variants Figure 1 shows the GWAS Manhattan plots of association of variants towards the tacrolimus dose normalized trough concentrations. In the initial unadjusted analysis, 55 variants were signiTicant (p<108, Figure 1A) towards trough concentrations. CYP3A5*3 was the top most signiTicant variant (p=6.88X 1049). We then adjusted the analysis for CYP3A5*3 and seven variants remained signiTicant (Figure 1B). The top variant was CYP3A4*22 (p=2.77 X 1013). We then adjusted the analysis for CYP3A5*3 and CYP3A4*22 and no variants were GWAS signiTicant although CYP3A4*3 was top most although it did not meet genome wide signiTicance (p=0.001, Figure 1C). The allele frequencies are shown in Table 2. The Tinal parameter estimates of clinical and genetic factors obtained after multivariate analysis are in Table 3. Figure 2 shows the plots of dose normalized concentrations vs time by genotypes. Statistics: Linear mixed effects regression models were used to test for associations between natural log (ln) transformed dose normalized troughs and genotypes. Visual inspection showed that weight normalized trough concentrations initially started low, rose quickly until day 9 posttransplant. Therefore, we initially tested the association of variants from GWAS with the estimated day 9 trough levels from a simple time trend model. For the Tinal model (Table 3), the top variants were adjusted for clinical factors that were identiTied using backward selection with a retention pvalue of 0.10 a data are median (range) in the Tirst 6 months posttransplant a Data shown as untransformed estimates, b For each day posttransplant (day 1 to 180) there is a daily 1.06 (6%) increase in dosenormalized tacrolimus troughs. c There is an additional effect for each day after day 9 (day 10180) where dosenormalized tacrolimus troughs are reduced by 0.94(6%). Figure 1. Manhattan plots of association of dose normalized troughs and variants Figure 2: Dosenormalized tacrolimus trough concentrations vs time by genotype A Chromosome log10p B Chromosome log10p C Chromosome log10p Unadjusted analysis Adjusted for CYP3A5*3 Adjusted for CYP3A5*3 and CYP3A4*22 A. CYP3A5*3 B. POR*28 C. CYP3A4*3 D. CYP3A4*22 0 0.5 1 1.5 2 2.5 0 50 100 150 200 250 Dose-Normalized Tacrolimus Trough (ng/mL) / (mg) Time from Tx (Days) *3/*3 *3/*1 *1/*1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 100 150 200 250 Dose-normalized tacrolimus troughs (ng/ml)/(mg) Time from Tx (Days) T/T C/T 0 0.5 1 1.5 2 2.5 0 50 100 150 200 250 Dose-Normalized Tacrolimus Trough (ng/mL) / (mg) Time from Tx (Days) C/C C/T T/T 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 100 150 200 250 Dose-Normalized Tacrolimus Trough (ng/mL) / (mg) Time from Tx (Days) C/C C/T T/T