Background: Nucleos(t)ide reverse transcriptase inhibitors (NRTIs), such as tenofovir, require intracellular phosphorylation for pharmacologic activity. Drug transporters may contribute to the intracellular disposition of NRTIs.
Objective: We characterized intracellular tenofovir diphosphate (TFV-DP) concentrations in HIV-infected patients (n = 30), and investigated associations between TFV-DP concentrations and polymorphisms in the drug transporter genes SLC22A6, ABCC2, and ABCC4.
Methods: Subjects were genotyped for 6 single-nucleotide polymorphisms: 2 in SLC22A6 (encodes influx transporter, human organic anion transporter 1), 728G>A and 453G>A; 2 in ABCC2 (encodes efflux transporter, multidrug resistance protein [MRP] 2), −24C>T and 1249G>A; and 2 in ABCC4 (encodes efflux transporter, MRP4), 3463A>G and 4131T>G.
Results: The mean TFV-DP was 76.1 fmol/106 cells (range: 16.3 to 212 fmol/106 cells). Tenofovir apparent oral and renal clearances were significantly predictive of intracellular TFV-DP concentrations. For every 1-L/h decrease in tenofovir renal clearance, there was, on average, an 8% increase in TFV-DP (P = 0.002). We identified a novel relation between ABCC4 3463A>G genotype and TFV-DP. ABCC4 3463G variants had TFV-DP concentrations 35% higher (29 fmol/106 cells) than wild type (P = 0.04).
Conclusion: This study provides direction for future investigations to elucidate the contribution of clinical characteristics and drug transporter genotype to TFV-DP safety and efficacy.
From the *School of Pharmacy, University of Colorado at Denver, and Health Sciences Center, Denver, CO; and the †Department of Medicine, Division of Infectious Diseases, University of Colorado at Denver, and Health Sciences Center, Denver, CO.
Received for publication July 3, 2007; accepted October 15, 2007.
Research funded by National Institutes of Health (NIH) grant RO1 AI 33835 (Principal Investigator: C. V. Fletcher), NIH grant P30-AI054907 (Principal Investigator: C. V. Fletcher), and grant M01 RR00051.
Portions of this study were presented at the 13th Conference on Retroviruses and Opportunistic Infections, February 5-8, 2006, Denver, CO (poster 570) and at the Seventh International Workshop on Clinical Pharmacology of HIV Therapy, April 20-22, 2006, Lisbon, Portugal (oral abstract 34).
Correspondence to: Courtney V. Fletcher, PharmD, College of Pharmacy, University of Nebraska Medical Center, 98600 Nebraska Medical Center, Omaha, NE 68198-6000 (e-mail: email@example.com).
Tenofovir disoproxil fumarate (TDF), the prodrug of the acyclic nucleotide tenofovir, is widely used in the treatment of HIV infection and is being investigated for use in the treatment of hepatitis B virus1 and for the prevention of HIV transmission.2
The 2 ester groups of the TDF prodrug are removed by means of esterase hydrolysis to form tenofovir.3 Tenofovir is not metabolized by cytochrome P450 or other hepatic enzymes to a significant degree but is instead eliminated unchanged by a combination of glomerular filtration and active tubular secretion. Like the nucleoside reverse transcriptase inhibitors (NRTIs), tenofovir requires intracellular phosphorylation to exert its antiviral effect. Whereas most NRTIs undergo 3 sequential phosphorylation steps by human kinases, however, tenofovir already contains a phosphate group, and thus requires only 2 phosphorylation steps by means of adenosine monophosphate kinase and 5′ nucleoside diphosphate kinase to become the active moiety, tenofovir diphosphate (TFV-DP). In addition to mediating the antiviral effects, the intracellular concentrations of NRTIs are responsible for many of their toxicities, including lactic acidosis, pancreatitis, peripheral neuropathy, and lipoatrophy.4
The 4-year safety data with TDF have revealed a low incidence of adverse events;5 however, there are several published case reports of renal severe adverse events with TDF, including acute renal failure, Fanconi syndrome, and nephrogenic diabetes insipidus.6-9 The mechanism(s) for this nephrotoxicity are not completely understood. Structurally, tenofovir is similar to adefovir. The nephrotoxicity from adefovir is concentration dependent;10 thus, tenofovir-induced nephrotoxicity may also be concentration dependent. Alterations in influx/efflux membrane transporter expression or function at the level of the renal proximal tubule cells could affect the renal clearance of tenofovir, thereby increasing TFV-DP concentrations within the renal proximal tubule cells and other cells in the body.
Human organic anion transporter 1 (hOAT1), on the basolateral side of the renal proximal tubule cell, mediates the uptake of tenofovir and the other nucleoside phosphonates, adefovir and cidofovir.11 The transporters responsible for tenofovir efflux on the apical side are less well characterized. There are data suggesting that tenofovir is a substrate for multidrug resistance protein (MRP) 212,13 and MRP4.14,15 Single nucleotide polymorphisms (SNPs) in the SLC22A6 (encodes hOAT1),16ABCC2 (encodes MRP2),17-20 and ABCC4 (encodes MRP4)21 drug transporter genes may result in altered hOAT1, MRP2, or MRP4 expression or function, which may affect the intracellular pharmacokinetics of tenofovir.
There are few data describing the intracellular concentrations of TFV-DP. Furthermore, there are no data on the association between polymorphisms in the genes that encode transporters responsible for tenofovir movement across cells and intracellular TFV-DP pharmacokinetics. We hypothesized that SLC22A6, ABCC2, and ABCC4 SNPs may have functional consequences that could alter intracellular TFV-DP concentrations in vivo. We selected 6 SNPs based on putative functional consequences and reported variant allele frequencies of >5%: 2 in SLC22A6 (728G>A and 453G>A), 2 in ABCC2 (−24C>T and 1249G>A), and 2 in ABCC4 (3463A>G and 4131T>G). The objective of this work was to characterize intracellular TFV-DP concentrations in HIV-infected patients and to investigate the associations between these concentrations and drug transporter genetics.
Data were obtained from 30 HIV-infected individuals enrolled in a pharmacokinetic study to determine the effects of lopinavir/ritonavir on the renal clearance of tenofovir. The complete study methods and primary results have been described elsewhere.22 Briefly, HIV-infected persons between the ages of 18 and 65 years with normal renal function, HIV-1 RNA values <50 copies/mL, and on a TDF-containing antiretroviral regimen for at least 4 weeks were eligible. Half of the study subjects were on an antiretroviral regimen containing lopinavir/ritonavir, and the other half were on an antiretroviral regimen that excluded any protease inhibitors. The 2 groups of subjects were matched on gender, race, and age. This study was approved by the Colorado Multiple Institutional Review Board, and all participants provided written informed consent. All procedures were in accordance with the Helsinki Declaration of 1975, as revised in 2000. All subjects underwent a 24-hour intensive pharmacokinetic evaluation for quantitation of tenofovir in the blood and urine. In addition, 16 mL of blood was collected before dosing and 5 and 24 hours after the observed dose for isolation and counting of human peripheral blood mononuclear cells (PBMCs) and measurement of intracellular TFV-DP concentrations. Tenofovir in plasma and urine and intracellular TFV-DP in PBMCs were quantified with validated liquid chromatography/tandem mass spectrometry (LC/MS/MS) procedures as we have previously described.23,24
Tenofovir plasma pharmacokinetic characteristics were determined with noncompartmental procedures (WinNonLin version 5.0.1; Pharsight Corporation, Mountain View, CA). Renal clearance was determined by multiplying the fraction of tenofovir excreted in the urine over a dosing interval by apparent oral clearance (CL/F). Intracellular exposure was determined as the average of TFV-DP concentrations before dosing and 5 and 24 hours after the observed dose for each patient.
The complete genotyping methods have been previously described.22 Subjects were genotyped for 6 SNPs: 2 in SLC22A6 (encodes hOAT1): 728G>A (Arg50His; rs11568626) and 453G>A in the 5′ untranslated region (UTR) (rs4149170); 2 in ABCC2 (encodes MRP2): −24C>T in the promoter (rs717620) and 1249G>A (Val417Ile; rs2273697); and 2 in ABCC4 (encodes MRP4): 3463A>G (Lys1116Lys; rs1751034) and 4131T>G in the 3′ UTR (rs3742106).
Human genomic DNA (hDNA) was isolated from whole blood using a commercially available kit (QIAamp DNA Mini Kit; Qiagen, Valencia, CA) according to the manufacturer's protocol. SLC22A6, ABCC2, and ABCC4 genotypes were determined by polymerase chain reaction (PCR; MyCycler; Bio-Rad Laboratories, Hercules, CA), followed by pyrosequencing analysis (PSQ 96 MA; Biotage AB, Uppsala, Sweden).
Linear regression was used to examine the effects of covariates, including polymorphisms in the SLC22A6, ABCC2, and ABCC4 drug transporter genes, on intracellular TFV-DP exposure. Adjustments were made in the linear regression analyses for treatment group (lopinavir/ritonavir vs. no protease inhibitor), Modification of Diet in Renal Disease (MDRD)-estimated glomerular filtration rate (GFR), and race if this differed between genotypes. For each polymorphism, heterozygotes were grouped with homozygous variants for data analysis and are referred to as variant carriers. TFV-DP concentrations were log-transformed for analysis to stabilize the variance. P < 0.05 was considered significant. No adjustments were made for multiple comparisons.
Thirty-four subjects were screened; 30 were eligible and completed the study. The average age of participants was 41.5 years (range: 25 to 60 years). Six subjects were black, and 8 were female. The average CD4 count of the subjects was 566 cells/mm3 (range: 200 to 1485 cells/mm3). In addition to TDF, subjects were also on lopinavir/ritonavir (n = 15), efavirenz (n = 11), nevirapine (n = 2), lamivudine (n = 18), zidovudine (n = 6), emtricitabine (n = 9), stavudine (n = 3), abacavir (n = 4), and didanosine (n = 2).
Characterization of Intracellular Tenofovir Diphosphate Concentrations
The average intracellular TFV-DP concentration was 76.1 fmol/106 cells (range: 16.3 to 212 fmol/106 cells; coefficient of variation [CV] = 52%). The GFR, estimated using the equation,25 was significantly predictive of TFV-DP concentrations. For every 10-mL/min decrease in GFR, there was, on average, an 8% increase in intracellular TFV-DP concentrations (P = 0.04). CL/F and renal clearance of tenofovir were significantly predictive of intracellular TFV-DP concentrations. For every 1-L/h decrease in tenofovir CL/F, there was, on average, a 2% increase in intracellular TFV-DP concentrations (Fig. 1A; P = 0.002). For every 1-L/h decrease in tenofovir renal clearance, there was, on average, an 8% increase in intracellular TFV-DP concentrations (see Fig. 1B; P = 0.002). There were no differences in intracellular TFV-DP concentrations by weight, age, gender, race, CD4 cell count, or treatment with lopinavir/ritonavir.
Genotypes and Intracellular Tenofovir Diphosphate Concentrations
hDNA was isolated from 27 of the 30 subjects. DNA samples were inadvertently not obtained from the first 3 subjects enrolled during their intensive pharmacokinetic visits. All genotypes were in Hardy-Weinberg equilibrium. The variant allele frequencies for the SLC22A6 453G>A and 728G>A SNPs were 9.3% and 1.9%, respectively. Because of the low frequency of the SLC22A6 728A allele, this SNP was not studied in further analyses. The variant allele frequencies for the ABCC2 −24C>T and 1249G>A SNPs were 16.7% and 22.2%, respectively. The variant allele frequencies for the ABCC4 3463A>G and 4131T>G SNPs were 20.4% and 40.7%, respectively. There was a trend (P = 0.09, χ2 test for equal proportions) toward more ABCC4 3463G variant carriers in black versus nonblack patients. Thus, race was controlled for in analyses with this SNP. There were no differences in SLC22A6 453G>A or ABCC2 genotypes by treatment group (ie, those on lopinavir/ritonavir vs. those not on a protease inhibitor) or race.
ABCC4 3463G variant carriers had TFV-DP concentrations that were 29 fmol/106 cells (35%) higher than wild-type homozygotes after adjusting for race, treatment group, and GFR (P = 0.04). Figure 2 shows the raw intracellular TFV-DP concentrations by ABCC4 3463A>G genotype. Adjusting for tenofovir plasma area under the concentration time curve (AUC), ABCC4 3463G variants had, on average, intracellular TFV-DP concentrations 21 fmol/106 cells higher than wild type; this difference, however, was not statistically significant (P = 0.1). No associations with SLC22A6 453G>A or ABCC2 −24C>T and 1249G>A genotypes and intracellular TFV-DP concentrations were identified.
This study describes the first association between intracellular TFV-DP concentrations and polymorphisms in genes encoding drug transporters relevant for tenofovir movement across cells; it is also one of the largest published data sets describing the intracellular concentrations of TFV-DP in persons with HIV. We found that as tenofovir oral and renal clearances decrease, there was an increase in intracellular TFV-DP concentrations. We also found that ABCC4 3463G variant carriers had TFV-DP concentrations that were 35% (29 fmol/106 cells) higher than wild type after adjusting for race, treatment group, and GFR (P = 0.04). If the concentrations of TFV-DP in PBMCs are a surrogate for those in renal proximal tubule cells, on the basis of these findings, subjects who have impaired renal function and/or the ABCC4 3463G variant allele may have increased intracellular TFV-DP concentrations in the renal proximal tubule cells and may be at increased risk for tenofovir-associated adverse effects, including nephrotoxicity. The findings of higher intracellular TFV-DP concentrations in the PBMCs of ABCC4 3463G variant carriers may also have important implications for virologic efficacy, although this could not be assessed in our study patients because an HIV-1 RNA level <50 copies/mL was required for study enrollment.26
hOAT1,11 MRP2,12,13 and MRP414,15 are drug transporters associated with tenofovir movement across the renal proximal tubule cells. hOAT1 is encoded by SLC22A6, and hOAT1 expression has been shown to mediate the nephrotoxicity induced by the acyclic nucleoside phosphonates.11 An arginine-to-histidine substitution at position 50 (R50H; 728G>A) in SLC22A6 has been associated with an increased affinity of the nucleoside phosphonates, including tenofovir, for hOAT1 in vitro.16 There was only 1 subject in our study, a black woman, who was heterozygous for the SLC22A6 728G>A SNP; thus, we were unable to evaluate the association between the 728G>A SNP and TFV-DP concentrations. Of note, her plasma tenofovir AUC was 6387 ng·h/mL and her average intracellular TFV-DP concentration was 119 fmol/million cells. Both values were above the 75th percentile for the other study subjects, which seems to argue against this SNP conferring an increased affinity of tenofovir for hOAT1 in this female subject. There are few data on hOAT expression in PBMCs, however, and at least 1 in vitro study found no hOATs expressed in PBMCs or purified T cells.27 The 5 subjects heterozygous for SLC22A6 453G>A did not have tenofovir plasma or TFV-DP concentrations that differed from the study population averages.
In our study, ABCC4 3463G variant carriers had higher intracellular concentrations of TFV-DP than wild type. This is a novel relation in vivo. In vitro experiments have shown that MRP4 overexpressing cells have decreased intracellular TFV-DP concentrations.15 Overexpression of MRP4 has also been shown to confer resistance to adefovir in vitro.26 Polymorphisms have been identified in ABCC4, which encodes MRP4. Although there are few data on the impact of ABCC4 SNPs on drug pharmacokinetics, the ABCC4 3463G variant has an increased probability of altered messenger RNA (mRNA) splicing and potentially altered MRP4 protein expression based on exonic splicing enhancer analyses.21 This SNP may therefore result in decreased MRP4 protein expression, which may influence intracellular concentrations of MRP4 substrates. This relation between ABCC4 3463G variant carriers and higher TFV-DP is consistent with our previous findings of an association between ABCC4 3463A>G genotype and plasma tenofovir pharmacokinetics in these patients.22,28 Specifically, ABCC4 3463G variant carriers had slower tenofovir renal clearance values and higher plasma tenofovir AUCs than wild type. We did not observe an association between ABCC2 genotype and intracellular TFV-DP concentrations in this study. MRP2 is encoded by ABCC2. In a prior case-control study, the ABCC2 CATC haplotype (composed of the SNPs −24C>T, 1249G>A, 3563T>A, and 3972C>T) was found in a higher percentage of patients who developed tenofovir-induced renal proximal tubulopathy compared with patients who did not develop tubulopathy.29 MRP230 and MRP431 are reportedly expressed on PBMCs.
There are limitations to this study. First, the blood volume and processing requirements only allowed us to obtain 3 PBMC samples for intracellular TFV-DP quantification per subject. An increased number of samples per patient may have provided an improved estimate of an individual's intracellular TFV-DP pharmacokinetics. The long intracellular half-life of tenofovir suggests that the intracellular concentrations should be relatively stable over a dosing interval,32 however, such that 3 observations would be highly representative of overall exposure. Second, the degree to which intracellular TFV-DP concentrations in PBMCs reflect those in the renal proximal tubule cells is unknown. Third, the small number of variant carriers for some SNPs (particularly in SLC22A6) and the small sample size limited the statistical power of our genetic association study. We did not make adjustments for multiple comparisons, given the exploratory nature of the study, nor did we evaluate the influence of haplotypes on intracellular TFV-DP concentrations. Finally, this was a hypothesis-generating study, which retrospectively evaluated the association between 6 SNPs in the genes that encode for drug transporters relevant for tenofovir movement across cells and intracellular TFV-DP concentrations. The association between intracellular tenofovir pharmacokinetics and ABCC4 3463A>G genotype noted in this investigation requires confirmation in larger studies.33
In conclusion, we found relations between patient characteristics and intracellular concentrations of the pharmacologically active moiety of tenofovir. These data provide the scientific basis for prospective investigations to elucidate the contribution of decreased renal function and ABCC4 genotype to systemic and intracellular tenofovir pharmacokinetics and the safety and efficacy of this drug.
The authors acknowledge the study participants, the University of Colorado Hospital General Clinical Research Center, and Pamela Wolfe, MS.
1. van Bommel F, Zollner B, Sarrazin C, et al. Tenofovir for patients with lamivudine-resistant hepatitis B virus (HBV) infection and high HBV DNA level during adefovir therapy. Hepatology
2. Peterson L, Taylor D, Clarke EEK, et al. Findings from a double-blind, randomized, placebo-controlled trial of tenofovir disoproxil fumarate (TDF) for prevention of HIV-infection in women [ThLB0103]. Presented at: XVI International AIDS Conference; 2006; Toronto.
3. Kearney BP, Flaherty JF, Shah J. Tenofovir disoproxil fumarate: clinical pharmacology and pharmacokinetics. Clin Pharmacokinet
4. Anderson PL, Kakuda TN, Lichtenstein KA. The cellular pharmacology of nucleoside- and nucleotide-analogue reverse-transcriptase inhibitors and its relationship to clinical toxicities. Clin Infect Dis
5. Nelson MR, Katlama C, Montaner JS, et al. The safety of tenofovir disoproxil fumarate for the treatment of HIV infection in adults: the first 4 years. AIDS
6. Zimmermann AE, Pizzoferrato T, Bedford J, et al. Tenofovir-associated acute and chronic kidney disease: a case of multiple drug interactions. Clin Infect Dis
7. Peyriere H, Reynes J, Rouanet I, et al. Renal tubular dysfunction associated with tenofovir therapy: report of 7 cases. J Acquir Immune Defic Syndr
8. Karras A, Lafaurie M, Furco A, et al. Tenofovir-related nephrotoxicity in human immunodeficiency virus-infected patients: three cases of renal failure, Fanconi syndrome, and nephrogenic diabetes insipidus. Clin Infect Dis
9. James CW, Steinhaus MC, Szabo S, et al. Tenofovir-related nephrotoxicity: case report and review of the literature. Pharmacotherapy
10. Izzedine H, Launay-Vacher V, Deray G. Antiviral drug-induced nephrotoxicity. Am J Kidney Dis
11. Cihlar T, Ho ES, Lin DC, et al. Human renal organic anion transporter 1 (hOAT1) and its role in the nephrotoxicity of antiviral nucleotide analogs. Nucleosides Nucleotides Nucleic Acids
12. Louie S, Lam JT, Neely M, et al. Multidrug resistance protein 2 (MRP2) inhibition by ritonavir increases TFV-associated cytotoxicity [abstract 23]. Presented at: Sixth International Workshop on Clinical Pharmacology of HIV Therapy; Québec City, Canada, 2005.
13. Mallants R, Van Oosterwyck K, Van Vaeck L, et al. Multidrug resistance-associated protein 2 (MRP2) affects hepatobiliary elimination but not the intestinal disposition of tenofovir disoproxil fumarate and its metabolites. Xenobiotica
14. Imaoka T, Kusuhara H, Adachi M, et al. Functional involvement of multidrug resistance-associated protein 4 (MRP4/ABCC4) in the renal elimination of the antiviral drugs adefovir and tenofovir. Mol Pharmacol
15. Ray AS, Cihlar T, Robinson KL, et al. Mechanism of active renal tubular efflux of tenofovir. Antimicrob Agents Chemother
16. Bleasby K, Hall LA, Perry JL, et al. Functional consequences of single nucleotide polymorphisms in the human organic anion transporter hOAT1 (SLC22A6). J Pharmacol Exp Ther
17. Itoda M, Saito Y, Soyama A, et al. Polymorphisms in the ABCC2 (cMOAT/MRP2) gene found in 72 established cell lines derived from Japanese individuals: an association between single nucleotide polymorphisms in the 5′-untranslated region and exon 28. Drug Metab Dispos
18. Haenisch S, Zimmermann U, Dazert E, et al. Influence of polymorphisms of ABCB1 and ABCC2 on mRNA and protein expression in normal and cancerous kidney cortex. Pharmacogenomics J
19. Ito S, Ieiri I, Tanabe M, et al. Polymorphism of the ABC transporter genes, MDR1, MRP1 and MRP2/cMOAT, in healthy Japanese subjects. Pharmacogenetics
20. Toh S, Wada M, Uchiumi T, et al. Genomic structure of the canalicular multispecific organic anion-transporter gene (MRP2/cMOAT) and mutations in the ATP-binding-cassette region in Dubin-Johnson syndrome. Am J Hum Genet
21. Anderson PL, Lamba J, Aquilante CL, et al. Pharmacogenetic characteristics of indinavir, zidovudine, and lamivudine therapy in HIV-infected adults: a pilot study. J Acquir Immune Defic Syndr
22. Kiser JJ, Carten ML, Aquilante CL, et al. The effect of lopinavir/ritonavir on the renal clearance of tenofovir in HIV-infected patients. Clin Pharmacol Ther
. 2007 June 20 [EPub ahead of print].
23. King T, Bushman L, Kiser J, et al. Liquid chromatography-tandem mass spectrometric determination of tenofovir-diphosphate in human peripheral blood mononuclear cells. J Chromatogr B Analyt Technol Biomed Life Sci
24. Delahunty T, Bushman L, Fletcher CV. Sensitive assay for determining plasma tenofovir concentrations by LC/MS/MS. J Chromatogr B Analyt Technol Biomed Life Sci
25. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med
26. Schuetz JD, Connelly MC, Sun D, et al. MRP4: a previously unidentified factor in resistance to nucleoside-based antiviral drugs. Nat Med
27. Minuesa G, Purcet S, Erkizia I, et al. Expression of nucleoside analog uptake transporters and entry mechanisms of AZT in immune cells [abstract 559]. Presented at: 14th Conference on Retroviruses and Opportunistic Infections; 2007; Los Angeles.
28. Kiser JJ, Aquilante CL, Anderson PL, et al. Effect of multidrug resistance proteins 2 and 4 polymorphisms on tenofovir pharmacokinetics in HIV-infected patients [abstract 34]. Presented at: Seventh International Workshop on Clinical Pharmacology of HIV Therapy; 2006; Lisbon.
29. Izzedine H, Hulot JS, Villard E, et al. Association between ABCC2 gene haplotypes and tenofovir-induced proximal tubulopathy. J Infect Dis
30. Janneh O, Owen A, Chandler B, et al. Modulation of the intracellular accumulation of saquinavir in peripheral blood mononuclear cells by inhibitors of MRP1, MRP2, P-gp and BCRP. AIDS
31. De Pasquale MP, Hulgan T, Sutton L, et al. HIV-1 infection is associated with changes in drug transporter gene expression in vivo [abstract 476]. Presented at: 10th Conference on Retroviruses and Opportunistic Infections; 2003; Boston.
32. Hawkins T, Veikley W, St. Claire RL 3rd, et al. Intracellular pharmacokinetics of tenofovir diphosphate, carbovir triphosphate, and lamivudine triphosphate in patients receiving triple-nucleoside regimens. J Acquir Immune Defic Syndr
33. Ioannidis JP, Ntzani EE, Trikalinos TA, et al. Replication validity of genetic association studies. Nat Genet