• Allograft rejection;
  • genomic variability;
  • non-HLA-related rejection;
  • pharmacogenomics


  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetics of Transplantation Outcomes
  5. Future Perspectives
  6. Disclosure
  7. Electronic Resources Described

Over the last decade, advances in genetic technologies have accelerated our understanding of the genetic diversity across individuals and populations. Case–control and population-based studies have led to several thousand genetic associations across a range of phenotypes and traits being unveiled. Despite widespread and successful use of organ transplantation as a curative therapy for organ failure, genetic research has yet to make a major impact on transplantation practice aside from HLA matching. New studies indicate that non-HLA loci, termed minor histocompatibility antigens (mHAs), may play an important role in graft rejection. With several million common and rare polymorphisms observed between any two unrelated individuals, a number of these polymorphisms represent mHAs, and may underpin transplantation rejection. Genetic variation is also recognized as contributing to clinical outcomes including response to immunosuppressants, introducing the possibility of genotype-guided prescribing in the very near future. This review summarizes existing knowledge of the impact of genetics on transplantation outcomes and therapeutic responses, and highlights the translational potential that new genomic knowledge may bring to this field.


copy number variant


genome-wide association studies


homozygous-deleted copy number variant


hematopoietic stem cell transplantation




minor histocompatibility antigens


single nucleotide polymorphism


single nucleotide variant


  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetics of Transplantation Outcomes
  5. Future Perspectives
  6. Disclosure
  7. Electronic Resources Described

Over the last two decades, more than 300 000 solid organ transplantations have been performed in the United States alone [1]. However, despite improvements in surgical techniques and the development of more effective immunosuppressant therapies, allograft rejection still affects ∼60% of transplanted individuals and remains one of the major risk factors of graft loss [2, 3]. Up to 40% of graft recipients experience some form of rejection within the first postoperative year [4], with lung and heart recipients showing the highest rates of rejection, with 55% and 25% of patients, respectively [1, 5, 6], and kidney and liver the lowest, with ∼10% and ∼17% of patients experiencing rejection, respectively. Rejection can occur where genetic disparities exist between donors and recipients, which may lead to presentation of polymorphic peptides that the recipient's immune system recognizes as nonself. Although key HLA loci have traditionally been considered to be the main contributor to the genetic variability of allograft rejection, some degree of rejection still occurs in HLA-matched sibling transplantations, which may be the result of noncompatible loci beyond HLA between donor and recipient [7]. Indeed, new findings indicate that non-HLA polymorphisms can impact upon transplantation outcomes since they have the potential of generating histoincompatibilities [7-9] influencing allograft rejection [10], and impacting immunosuppressant responses [11]. Approximately 3.5 million common and rare polymorphisms exist between two unrelated individuals of European ancestry and up to 10 million variants in individuals of African ancestry [12]. However, investigations of non-HLA genetic determinants of clinical outcomes following organ transplantation have yet to be performed in any systematic fashion to date. Recent technological advances in genomics such as genome-wide association studies (GWAS) allow the characterization of hundreds of thousands to several million single nucleotide polymorphisms (SNPs) and copy number variants (CNVs) across the human genome rapidly and inexpensively [11, 13]. Figure 1 illustrates the concepts of structural variations, SNPs and insertion/deletion polymorphisms (InDels). Furthermore, whole exome and whole genome sequencing, which interrogates the coding regions and the entire human genome, respectively, are quickly becoming commonly used tools within the clinical diagnostic arena. These second-generation sequencing technologies have the ability to extensively characterize genome-wide sources of histoincompatibility between donors and recipients, potentially unraveling specific genetic risk factors influencing rejection and immunosuppressant responses or severe adverse effects [14, 15]. In this review we aim to overview the current knowledge from existing genetics studies recently conducted for transplantation outcomes and therapeutic responses to immunosuppression therapies and we also discuss the translational components from this genetic knowledge that may be rapidly implemented in this field.


Figure 1. Illustration of common forms of human genomic variations: structural variation (copy number variants [CNVs]), single-nucleotide variants (SNVs) and insertion–deletion polymorphisms (InDels). (A) Structural variation. CNVs are the most common form of structural variation and include deletions and multiplications (typically duplications) >1000 bp. CNVs can affect whole genes or parts of them and may be present in heterozygosis (e.g. deletion of one copy of gene A in IND2 or duplication of gene B in IND1) or homozygosis (e.g. deletion of one copy of gene D in IND3). (B) SNVs and InDels. The presence of an A in IND1 instead of a C on the reference genome represents a graphic illustration of an SNV and the two red Ns in IND3 represent two nucleotides deletion.

Download figure to PowerPoint

Genetics of Transplantation Outcomes

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetics of Transplantation Outcomes
  5. Future Perspectives
  6. Disclosure
  7. Electronic Resources Described

Incompatibility across key HLA alleles has traditionally been considered the main factor influencing rejection in stem cell and solid organ transplants, and has therefore been the focus of extensive investigation. Because of high genetic variability in the HLA locus, a detailed characterization of HLA mismatches between donors and recipients is not routinely achieved. The effect of specific HLA mismatches in kidney transplantation have been characterized for sometime, whereas the importance of HLA matching on outcomes in organs such as the liver is still under debate [6, 16-18]. Even in HLA identically matched kidney transplantation, some degree of rejection is still evident. Non-HLA or minor histocompatibility antigens (mHAs) resulting from a range of functional polymorphisms in the genome have been suggested to be capable of inducing strong cellular immune responses (reviewed in [7]). In contrast to the established role of HLA, Terasaki [2] estimated that over twice as many graft failures in HLA identical siblings at 10 years posttransplantation (38%) were due to immunological reactions to non-HLA factors compared to graft failures attributable to HLA (18%). Although our current knowledge of non-HLA antigens is still limited to a small number of loci [1, 19, 20], if one extrapolates the large number of genetic variants observed between two unrelated individuals within a donor–recipient pair, then the number of non-HLA discrepancies between any given donor and recipient would be expected to be very large.

The most rigorous way of investigating the association of mHAs with organ survival is within HLA-identical sibling donor–recipients, which is mostly restricted to hematopoietic stem cell transplantation (HSCT) and living kidney transplants. Therefore, the role of mHAs in other solid organ rejection may be more difficult to resolve due to incomplete or no matching.

The first source of mHAs was identified in the Y chromosome [21] with at least six Y chromosome genes encoding various antigens presented by multiple MHC alleles recognized by donor T cells as foreign peptides after gender mismatched transplantation [19, 22, 23]. Autosomal chromosomes also contain mHAs, with more than 40 described to date (reviewed in [20]). Such antigens arise as a consequence of common genetic variation, especially nonsynonymous SNPs in coding regions of the genome [24], leading to differences in the amino acid sequence proteins between donor and recipient. The number of identified autosomal mHAs identified to date is small considering the abundance of functional variants in the human genome, but it will undoubtedly grow with the recent genomic advances. Recently, Tennessen et al [25] applied whole exome sequencing to 15 585 human protein-coding genes in more than 2000 individuals of European and African ancestry and described more than 500 000 single nucleotide variants (SNVs) among the entire sample with an average of 13 595 variants per individual. The majority of these variants had not been previously described and were rare (minor allele frequency below 0.5%) and population-specific, with ∼2% predicted to impact the function of more than 300 genes per genome assessed [25], highlighting the need for a deeper genome-wide examination of the donor and recipient polymorphisms for potential mHAs.

A distinct form of loss-of-function (LoF) genetic variants, homozygous-deleted CNVs (hdCNVs), is gaining traction as playing a role in transplant rejection. hdCNVs have been identified through large-scale GWAS and sequencing data sets [9, 26]. Possession of hdCNVs may manifest in a “normal” phenotype, with the term “disposable” or “dispensable,” genes often used to describe such variants. Cases where unaltered phenotypes are observed with carriage of hdCNVs suggest that there is either compensation for the hdCNV(s), or their function is no longer necessary. For example, hdCNVs of UGT2B17, a gene expressed in graft-versus-host disease-affected tissues, have been characterized as mHAs in HSCT [8, 27]. McCarroll et al [8] recently analyzed six common CNV deletions spanning genes in three HSCT cohorts (totaling 1345 HLA-identical sibling donor–recipient pairs). The authors found that risk of acute graft-versus-host disease was greater in recipients where UGT2B17 hdCNV was mismatched, that is zero gene copies present in the donor but one or two copies present in recipient (OR = 2.5; 95% CI 1.4–4.6) [8]. However, they only used markers for CNVs that were detectable on the initial GWAS arrays, which were limited in content, as many CNV regions known now were not adequately probed or captured. Common hdCNVs may thus be important as they may play an active role in graft longevity.

Recent whole exome and genome sequencing indicates that each individual carries numerous genetic variants predicted to cause LoF of protein-coding genes. MacArthur et al recently studied 2951 putative LoF variants obtained from whole genome sequencing from 185 human individuals from the 1000 Genome Project (derived from typical population samples), identifying and validating rare and likely deleterious LoF alleles, as well as common LoF variants in nonessential genes in this data set [26]. They estimated that a typical human genome contains ∼100 genuine LoF variants with ∼20 genes inactivated in both copies, indicating unexpected redundancy in the human genome and suggesting that there are numerous mutations that are “private” to each personal genome. Since the immune system of an individual carrying LoF variants in both copies of a given gene may have had no previous exposure to protein(s) encoded by that gene(s), cellular or humoral immune recognition of that protein as an alloantigen in the grafted organ could very plausibly contribute to risk of rejection. LoF impacting the expression or translation of both copies of a given gene, especially in the transplanted organ of interest, is thus a plausible source of donor–recipient genomic incompatibilities, and may underpin rejection.

Need for second-generation sequencing

The emergence of large numbers of potential non-HLA incompatibilities with varying potential levels of immunogenicity highlights the clear need for deeper capture and assessment of SNVs, SNPs and CNVs in donors and recipients. Deep sequencing technologies will undoubtedly be key to unraveling common and rare LoFs and hdCNVs in population-scale diversity as well as capturing “private” individual-level polymorphisms to characterize the histocompatibility determinants involved in the biological mechanisms of alloimmunity. It should be noted that while there is low statistical power to detect such LoFs and hdCNVs at an association level, there is a lot of promise to follow up putative individual donor–recipient genomic incompatibility through the use of autoantibody testing in sera and/or tissues from the recipient postoperatively. Such approaches offer the promise of clinical utility through closer monitoring of a priori mHAs or potentially tolerizing recipients to a given gene product.

Genetic risk factors for transplantation outcomes beyond HLA

Over the last decade there have been unprecedented advances in the assessment of human genomic diversity across the major human populations through the development of high-throughput genotyping and deep sequencing technologies, as well as the development of population scale genomic maps such as the International HapMap Project [28]. These tools have been applied to deeply characterize the genetic architecture of common and rare diseases and evoked conditions such as drug severe adverse events [29]. Unfortunately, to date, the field of organ transplantation has not benefited from many of these advances. While over a thousand genetic association studies on organ rejection have been published so far, they are primarily candidate-gene based and suffer from many of the usual pitfalls of genetic association studies such as lack of adequate sample sizes, retrospective study designs, noninclusion of appropriate covariates such as ethnicity, lack of replication and proper statistical correction when multiple hypotheses are tested. Table 1 shows a review of appropriately designed published studies on organ transplantation outcomes. These studies were initially collated from PubMed and filtered for appropriate study size and design, defining appropriate as having at least a sample size of 200 individuals, including covariates, accounting for ethnicity and adjusting for multiple testing, when necessary. As observed in Table 1, genetic variation in up to 47 genes encoding cytokines, chemokines, cell-adhesion molecules, components of the renin–angiotensin–aldosterone pathway, coagulation and aggregation factors has been widely investigated and associated with outcomes such as delayed graft function, acute or chronic rejection or graft failure.

Table 1. Genetic association studies of organ transplantation outcomes
GeneVariantPMIDOutcomeSample sizeGraftResults
  1. Summary of the most relevant studies performed to date reporting genes/variants associated with transplantation outcomes. Low case letters a to h in PMID column indicate that, amongst all the variants, these particular studies investigate exclusively those variants specified by the letters. For the rest of the studies, all variants are investigated. In alphabetical order, abbreviations for outcomes are: AR = acute rejection; BOS = bronchiolitis obliterans syndrome; BPAR = biopsy proven acute rejection; CAN = chronic allograft nephropathy; DGF = delayed graft function; GF = graft function; GL = graft loss; GS = graft survival; IIF = infiltration of inflammatory cells; PNF = primary non-function; PS = patient survival; RHC = rejection with hemodynamic compromise; serCr = serum creatinine; SR = subclinic rejection. In sample size and results columns, D = donor and R = recipient. For results, ESRD = end-stage renal disease; NA = No Association

  11923700GF, serCr224KidneyNA
  21659963RHC532HeartRHC and rs4343: HR = 0.58 [0.36-0.95]; p = 0.031
AGTrs69912548125GF210KidneyIncreased Cr/t shorter t-to-sustained doubling of Cr, shorter t-to GL (p < 0.007)
  11923700GF, serCr224KidneyNA
AGTR1rs518611923700GF, serCr224Kidneyrs5186 and preserved long-term GF (p = 0.037)
ALCAMrs104424020220571GS954 + 1002KidneyNA
C3rs223019922176838AR, DGF, PNF1265D-RKidneyNA
  19246358AR, GF, PS1147KidneyNA
C4C4L/S, CNV21164027AR, GF, GL, GS1969KidneyNA
CCL2rs102461117989610AR, CAN, SR436KidneyNA
  12462338AR, GS209LiverNA
  12239249GF232Kidneyrs1024611GG vs A and GS: 67 ± 14 vs 95 ± 4 months; p = 0.0052
CCL5109T/C, rs2107538, rs2280788 (a)17989610AR, CAN, SR436KidneyNA
  12462338aAR, GS209LiverNA
CCR2rs179986415458467AR, CAN244DKidneyNA
  17989610AR, CAN, SR436KidneyNA
  12201365AR, GS, PS207LiverNA
CCR5rs1799987 (b), rs333 (c)19561149AR243D-RKidney# A alleles (D + R) and AR, RR
  15458467AR, CAN244DKidneyAR (p = 0.039)
  12462338bAR, GS209LiverNA
  12201365cAR, GS, PS207LiverNA
  17989610cAR, SR436KidneyNA
  17217435AR, BOS, GS226LungEarlier AR, BOS and rs2569190TT: HR = 1.65 [1.03–2.64], p = 0.04
COX2rs2041719788502GF603Kidneyrs20417C and GL HR = 2.43 [1.19-4.97], cohort 1; HR = 1.72 [0.99-3.77], cohort 2 (p < 0.051)
CUBNrs1801239, rs791897222574174Proteinuria1142R + 1186DKidneyESRD: OR = 1.39, elevated proteinuria 1y post (p < 0.015)
CXCL12rs180115712201365AR, GS, PS207Liverrs1801157 29.0% in dead patients compared to alive 17.1% and PS time (134 vs 98 months wt vs carriers) (p < 0.034)
  21304904BPAR, GS335D-RKidneyD-rs1801157A and BPAR: OR = 0.39, [0.20–0.76] and poor GS: HR = 3.01; [1.19–7.60]; p < 0.020
CXCR1rs267122221452410AR216KidneyAR and D rs2671222A (OR = 3.56 [1.37-9.27]; p = 0.009)
CXCR2rs1126579, rs467425821452410AR216KidneyNA
CXCR4rs222801421304904BPAR, GS335D-RKidneyNA
CYP11B2rs179999811923701GF, serCr224Kidneyrs1799998TT vs CC worse GF (p = 0.002)
DARCrs281477815327416AR, DGF222KidneyNA
FASrs180068221659963RHC532HeartRHC and rs1800682AA: HR = 1.84 [1.25–2.69]; p = 0.002
FCN2rs17514136, rs17549193, rs3124952, rs3124953, rs785169622173060BPAR, DGF, PNF, PS1272KidneyNA
FVrs602511858477AR, GF394KidneyGA no GF (RR = 2.87 [1.01-8.26]), #dialysis until GF, risk for at least one AR episode (RR = 3.83 [1.38-10.59]). Cr, protein excretion rate (p < 0.05).
HMOX1(GT)n, rs2071746, rs2071747, rs2071748, rs2071749, rs2285112, rs5755720, rs6518952, rs814066918640487DGF965RKidneyNA
IFNGrs243056115458467AR, CAN244DKidneyCAN (p < 0.008)
  17989610AR, CAN, SR436KidneyNA
  15599305AR, DGF291R + 206DKidneyNA
  20622753GF, IIF218KidneyNA
IL10rs1800871, rs1800872, rs180089615458467AR, CAN244DKidneyNA
  15599305AR, DGF291R + 206DKidneyR ACCACC, ATAATA, GCCATA AR: OR = 1.9 [1.1–3.1], p = 0.016
  20622753GF, IIF218Kidneyrs1800871 TT high IIF grade (OR = 3.27 [1.1–9.8]; p = 0.035)
  15367225GS2298 + 1901KidneyNA
  21659963RHC532Heartrs1800896 RHC: HR = 0.49 [0.27–0.90]; p = 0.020
 rs1554286, rs1878672, rs2222202, rs3021094, rs3024493, rs3024494, rs302449818640487DGF965RKidneyNA
IL12Brs321222720622753GF, IIF218KidneyNA
IL18rs18723819077897AR226 + 148CKidneyrs187238GG and AR: OR = 3.653; p = 0.015
IL1Brs1143634, rs1694420622753GF, IIF218Kidney
IL2(CA)m(CT)n11981437AR290HeartAllele 135 and risk of AR (p < 0.03)
 rs206976215458467AR, CAN244DKidneyNA
IL4Rrs180127515367225GS2298 + 1901KidneyNA
IL6rs180079515458467AR, CAN244DKidneyNA
  15599305AR, DGF291R + 206DKidneyNA
  20622753GF, IIF218KidneyNA
ITGB3rs591817928472AR445Kidneyrs5918TT and AR: OR = 3.4; p = 0.04
MASP2rs7255087022173060BPAR, DGF, PNF, PS1272KidneyNA
MBL2rs11003125, rs1800450, rs1800451, rs5030737, rs7095891, rs709620622173059BPAR, DGF, PNF, PS1271KidneyNA
  19104434BOS, GS277 D-RLungD-rs7096206 and GS and BOS free PS (p = 0.007)
MTHFRrs1801131, rs180113319349296GS262D-218RKidneyNA
NOS3rs179998319349296GS262D-218RKidneyGS: OR = 9.192; [1.09-76.9]; p = 0.0406
PECAM1rs12953, rs1131012, rs66820220571GS954 + 1002KidneyNA
SELLrs1131498, rs2205849, rs222956920220571GS964 + 1002KidneyNA
TGFB1rs1800471, rs1800470 (d), rs1800472 (e)15458467AR, CAN244DKidneyrs1800470 and AR (p = 0.027)
  17989610AR, CAN, SR436KidneyNA
  15599305AR, DGF291R + 206DKidneyR rs1800471GG +D rs1800871T and AR: OR = 1.8 [1.1–3.0], p = 0.027
  20622753GF, IIF218KidneyNA
  15367225GS2298 + 1901KidneyNA
TLR3rs3775290, rs3775291, rs377529619741468AR216D-RKidneyNA
TLR4rs10759932, rs192791419741468AR216D-RKidneyR rs10759932C vs TT and AR (OR = 0.25 [0.11-0.57]). rs10759932C higher rejection-free PS rates (p < 0.023)
TNFrs1800629 (f), rs1800628 (g), rs3093662 (h) rs36152512682890AR210D-RLiverNA
  17989610AR, CAN, SR436KidneyNA
  15599305AR, DGF291R + 206DKidneyR rs1800629AA and AR: OR = 5.0 [3.0–8.3]; D rs1800629GA and DGF OR = 1.6 [1.1–2.6], p < 0.040
  15367225AR, GS2298 + 1901KidneyGS rate among retransplants: 63.0% AA vs.79.5%G; p = 0.0116
  18640487f, g, hDGF965RKidneyNA
  20622753GF, IIF218KidneyNA
TP53rs1042522, rs12951053, rs1614984, rs1625895, rs17884306, rs4968187, rs989494618640487DGF965RKidneyNA
VEGFrs1005230, rs2010963, rs3556934, rs69994721659963RHC532HeartNA

However, results do not consistently replicate across studies. This lack of replication can be attributed to both the above-mentioned limitations of candidate-gene study designs and the complexity and diversity of clinical phenotypes. In addition to differences in protocols for immunosuppressive regimes, ascertainment of outcomes may differ by study—e.g. the criteria used to define acute rejection and chronic allograft dysfunction might vary among transplant centers: these all introduce further heterogeneity between studies. Specifically, in the case of rejection, studies may not always base the diagnosis on biopsy results and biopsy results themselves may be inaccurate due to interoperator differences. It is essential, therefore, to use objective and standard definitions for transplantation outcomes in order to obtain consistent results to homogenize studies.

Thus future studies should aim to define phenotypes with precision and use a rigorous genetic approach, which preferably incorporates a hypothesis-free design such as GWAS to gain the most insight into genetic risk factors for organ transplantation.

Pharmacogenetics of immunosuppressant responses

Genetic background is thought to account for as much as 95% of the variability in drug disposition and therapeutic effects [30]. Regimes for immunosuppressants are typically characterized by wide pharmacokinetic interindividual variability and narrow therapeutic indexes, which often makes the ideal balance between sufficient immunosuppression and drug toxicity difficult to achieve. Thus, the discovery of genetic markers responsible for the interindividual variation in response to immunosuppressive therapy is an intensive area of ongoing research in transplantation. Ekberg et al evaluated the immunosuppressant drug exposures in 1645 renal transplant patients randomly assigned to four treatment groups: (1) standard-dose cyclosporine, (2) low-dose cyclosporine, (3) low-dose tacrolimus and (4) low-dose sirolimus [31], and observed that up to 90% of patients experienced at least one adverse event during treatment. Both biopsy-proven acute rejection and severe adverse events were similar in groups, ranging from 2.3% to 37% and 43% to 53%, respectively, depending on the drug, with the highest events observed in the low sirolimus dose group [31]. In efforts to avoid such issues, therapeutic drug monitoring is being routinely performed, but it is currently assessed posttransplant and thus is not used for determining the optimal immunosuppressant starting dose, which is still established by an iterative postoperative approach.

Alternative strategies incorporating pharmacogenetics hold great promise as complementary tools in drug monitoring to better guide individual therapies and doses. While the number of studies focusing on the pharmacogenetics of immunosuppressants has increased dramatically over the last few years, to date, many of these are underpowered that suffer from the typical candidate gene association study pitfalls. One clear exception is the robust association observed for rs776746 in CYP3A5, the primary enzyme involved in the metabolism of tacrolimus, the most frequently prescribed immunosuppressant drug worldwide. While nongenetic factors such as recipient gender, age, diabetes status and calcium channel blockers, such as some newer class of antifungals and grapefruit juice, influence tacrolimus levels, the rs776746 SNP is very important in tacrolimus clearance, with dosing requirements as well as time to therapeutic concentrations explaining up to 45% of the dose and 30% of clearance variability [13]. The rs776746-A form is classically referred to as CYP3A5*1, while rs776746-G is referred to as CYP3A5*3. The latter is a noncoding variant that results in a cryptic splicing site, which causes 131 nucleotides of the intronic sequence to be inserted in the mRNA [32]. The consequence is the introduction of a premature stop codon that truncatesCYP3A5 and leads to complete lack of CYP3A5 translation in *3 homozygotes [32]. The prevalence of this CYP3A5*3 allele is as high as 90% in individuals of European ancestry while ∼90% of African Americans carry at least one of the common fully functional CYP3A5*1 alleles (*1/*1 or *1/*3). These frequency differences in CYP3A5*3 make it one the most important genetic markers of interindividual and interethnic differences observed in CYP3A-dependent drug responses and clearance [33]. There is also evidence of an additional rare variant, CYP3A5*6, which is associated with tacrolimus trough levels.

Tables 2-4 review genetic association studies for tacrolimus, cyclosporine and mycophenolic acid regarding pharmacokinetics, transplant outcomes and adverse events. Due to the previously mentioned limitations of the candidate gene approach, we only focus on studies with over 100 individuals and appropriate quality control for phenotype and genotype data. Several studies have confirmed the major impact of CYP3A5 on tacrolimus dose requirements and renal clearance; however, similar to genetic association studies on graft outcomes, it has been challenging to robustly replicate additional findings. Apart from CYP3A5, to date, only the associations of ABCB1 rs1045642, rs1128503, rs2032582 and cyclosporine pharmacokinetics and CYP3A4 rs2740574 and rs3559936 with tacrolimus pharmacokinetics appear to be consistent.

Table 2. Pharmacogenetic studies on immunosuppressant therapy pharmacokinetics
GeneVariantPrimary ISTPMIDSample sizeGraftResults
  1. Relevant studies investigating genes/variants associated with calcineurin inhibitor and mycophenolic acid pharmacokinetics. Similarly to Table 1, letters a to g in PMID column indicate that particular studies investigate exclusively those variants specified by the letters. For Primary IST, CsA = cyclosporine A; CNI = CsA + Tac; MPA = mycophenolic acid; Tac = tacrolimus. For sample size, D = donor and R = recipient. In results, AUC0–12 = area under the curve from 0 to 12 hours; B = bioavailability; C = concentration; Cl = clearance; Co = trough concentration; D = dose; MPAG/MPA = MPA glucuronide to MPA ratio; NA = No Asociation.

ABCB1rs1045642 (a), rs1128503, rs2032582 (b)CsA18192894104KidneyLower oral B: rs2032582GG, rs1128503TT, rs2229109G-rs1128503C-rs2032582G-rs1045642C (p < 0.05)
   11375290a124Kidney and heartNA
   20061922294HeartHigher Co/D-Co/weigh in rs1128503CC-rs2032582GG-rs1045642CC than TT-TT-TT (p < 0.05)
  Tac16753004a,b103Kidneyrs1045642 explains 3.7% of D variation (p = 0.009)
   17885626150DLiverrs1128503T and rs2032582T hepatic C greater than CC and GG (p < 0.035)
   18334918136Kidneyrs1045642TT lower D than C (p < 0.04)
ABCC2rs3740066, 1446C > G, rs717620 (c)CsA18192894104KidneyNA
  MPA + CNI19890249c185KidneyNA
CYP3A4rs2740574 (d) and rs28371759, rs35599367 (e), rs4987161, rs55785340, rs4986910CsA18192894104KidneyNA
   15592326151Kidney and heartrs2740574G higher Cl than AA (p < 0.05)
  CNI19005401d832Kidneyrs2740574AA-rs776746 GG (CYP3A5) less time to Tac C and higher C/D/Kg (p < 0.001)
   21903774e185Kidneyrs35599367T D lower than rs35599367CC (p = 0.018). Haplotype rs776746AA (CYP3A5)-rs35599367T or CC higher C than rs776746G (CYP3A5)-rs35599367CC (p < 0.03)
CYP3A5rs776746, rs10264272 (f), rs41303343 (g)CsA18192894104KidneyNA
   21806386f,g301Kidneyrs776746AA Tac Co/D than G (p≤0.0001)
  CNI21903774185KidneyHaplotype rs776746AA (CYP3A5)-rs35599367T or CC higher C than rs776746G (CYP3A5)-rs35599367CC (p < 0.03)
   19005401832Kidneyrs2740574(CYP3A4)AA-rs776746GG and less time to achieve Tac C and higher C/D/Kg (p < 0.001)
  Tac16753004103Kidneyrs776746A explains 35.3% of D variation (p < 0.001)
   17885626f150DLiverrs776746G higher D than AA (p < 0.05)
   21677300209D-RKidneyR rs776746G lower Co, Co/D (p < 0.05)
   18334918136Kidneyrs776746G lower Co/D (p < 0.003)
CYP3AP1*1/*3Tac15147425178Kidney*1 lower C, delay in achieving target C and *3/*3 C above target (p < 0.003)
   12490779180 CYP3AP1*1 vs *3 reduction in C (p < 0.05)
SLCO1B3rs4149117, rs60140950MPA + CNI19890249185KidneyTac patients rs4149117T decrease in MPAG/MPA; Cmax (p < 0.0003)
UGT1A8rs1042597, rs17863762MPA + CNI19494809338Kidneyrs1042597GG higher CsA AUC0–12; rs6714486A and/or rs17868320T lower CsA AUC0–12 (p < 0.05)
UGT1A9rs17868320, rs6714486, rs72551330, rs2741049 19890249185 NA
UGT2B7rs7438135, rs7439366, rs7668282MPA18641546332KidneyNA
  MPA + CNI19494809338KidneyNA
Table 3. Pharmacogenetic studies of transplantation outcomes
GeneVariantPrimary ISTPrimary outcomePMIDSample sizeGraftRESULTS
  1. Relevant pharmacogenetic studies in transplantation outcomes. Letters a to g in PMID column indicate that particular studies investigate exclusively those variants specified by the letters. For Primary IST, CsA = cyclosporine A; CNI = CsA + Tac; MPA = mycophenolic acid and Tac = tacrolimus. For sample size, D = donor and R = recipient. For primary outcome, AR = acute rejection; CAD = chronic allograft damage; DGF = delayed graft function; GL = graft loss; GS = graft survival. For sample size, D = donor and R = recipient and in results, NA = No Asociation.

ABCB1rs1045642 (a), rs1128503 (b), rs2032582 (c), rs717620 (d)CsAAR11375290a124KidneyNA
    17198259a,c170Heartrs1045642CC and haplotype rs2032582GG-rs1045642CC higher risk for >3A rejection (p = 0.02)
   AR, GL20505666227D-259RKidneyD haplotype rs1128503T-rs2032582T-rs1045642T higher risk for AR (p = 0.018) and GL (p = 0.0019)
   DGF18510642147Kidneyrs1045642T and rs2032582T higher risk for DGF (p < 0.034)
  CNIAR19005401832KidneyHaplotype rs1128503C-rs2032582G-rs1045642T higher risk for AR (p = 0.038)
  MPA + TacCAD19762492a-c252KidneyD and R rs104564TT higher IF/TA (p = 0.001)
  TacAR, DGF21677300a209D-RKidneyNA
ABCC2rs3740066, 1446C > G, rs717620MPA + CNIAR19494809338KidneyNA
   AR, GL20505666227D-259RKidneyNA
  MPA + TacCAD19762492252KidneyNA
  MPA + TacCAD19762492252KidneyNA
   AR, DGF21677300209D-RKidneyNA
CYP3AP1*1/*3TacAR15147425178Kidney*1 earlier AR (p = 0.005)
IMPDH1rs11770116, rs2228075, rs2288548, rs2288549, rs2288550, rs2288553, rs4731448 and rs2278293 (e), rs2278294 (f)MPAAR17851563191Kidneyrs2278293A and rs2278294A lower risk (p < 0.05)
    20679962e,f456Kidneyrs2278294A lower risk (p = 0.0075)
IMPDH2rs11706052 (g), rs4974081MPAAR17851563g191KidneyNA
PXRrs1523130, rs2276707, rs3814055, rs7643645CNIDGF22453193178D-RKidneyDonor 2276707TT higher risk for DGF (p < 0.05)
UGT1A8rs1042597, rs17863762      
UGT1A9rs6714486, rs17868320MPA + CNIAR19494809338KidneyTac patients with 6714486A and/or rs17868320T higher risk of AR (p < 0.05)
UGT2B7rs7438135, rs7439366, rs7668282      
Table 4. Pharmacogenetics of immunosuppressant therapy–induced adverse events
GeneVariantPrimary ISTAdverse eventPMIDSample sizeGraftRESULTS
  1. Pharmacogenetic studies on IST induced adverse events. Letters a to g in PMID column indicate that particular studies investigate exclusively those variants specified by the letters. For Primary IST, CsA = cyclosporine A; CNI = CsA + Tac; MPA = mycophenolic acid; Tac = tacrolimus. For sample size, D = donor and R = recipient. For sample size, D = donor and R = recipient and in results, NA = No Association.

ABCB1rs1045642 (a), rs1128503, rs2032582CsANephrotoxicity20061922294HeartNA
  Tac 21677300a209D-RKidneyNA
  MPADiarrhea,leukopenia, anemia, infections21142914a218KidneyNA
ABCC2rs2273697, rs3740066, 1446C > G, rs717620MPADiarrhea,leukopenia, anemia, infections21142914218KidneyNA
ACErs4343CNINephrotoxicity16477235233Liverrs4343DD higher risk of nephrotoxicity (p < 0.0001)
    20526235304Kidneyrs776746G higher risk of nephrotoxicity (p = 0.01)
  CsABlood pressure15454731399D-RKidneyNA
IMPDH1rs11770116, rs2228075, rs2288548, rs2288549, rs2288550, rs2288553, rs4731448 and rs2278293 (b), rs2278294 (c)MPADiarrhea,leukopenia, anemia, infections17851563 (leukopenia)191KidneyNA
    20679962b,c (leukopenia,infection)456Kidneyrs2278294 higher risk of leukopenia (p = 0.0139)
IMPDH2rs11706052, rs4974081MPALeukopenia, infections17851563 (leukopenia)191KidneyNA
SLCO1B1521T > C, rs2306283, rs4149015, rs4149056MPADiarrhea,leukopenia, anemia, infections21142914218Kidneyrs4149056C reduced risk of adverse events (p = 0.002)
UGT1A998T > C, rs17868320, rs6714486      
UGT2B7rs7438135, rs7439366MPADiarrhea,leukopenia, anemia, infections18641546 (diarrhea, leukopenia)332KidneyNA

All of the rigorously designed studies to date conclude that there is a need to dose tacrolimus patients by the CYP3A5 rs776746 genotype. Thervet et al [15], during the first 6 days posttransplantation, studied the effect of guiding tacrolimus dosing based on the number of CYP3A5 functional alleles, on plasma drug concentration. Patients were randomly assigned to either a standard initial dose of tacrolimus (0.2 mg/kg/day) or a genotype-adjusted dose, who received 0.3 mg/kg/day if carrying allele *1 and 0.15 mg/kg/day if *3/*3. This randomized clinical trial found significantly higher number of patients achieving optimal dosing in the genotype-based dosing group, more rapidly and with fewer dose modifications [15]. However, the trial was not designed to investigate hard clinical outcomes such as graft failure—the acid test of a pharmacogenetic test.

Passey et al [34] established a dosing algorithm for tacrolimus including clinical, genetic and ethnic information. From a cohort of more than 600 kidney recipients, compared to those with G/G genotype, those with the rs776746 A/G genotype experienced a 69% increase in tacrolimus clearance and A/A genotype a 100% increase in clearance. The final dosing algorithm included CYP3A5 rs776746 genotype, days posttransplantation, age, steroid and calcium channel blocker use as independent predictors of the outcome [34].

The incorporation of CYP3A5 genetic information into tacrolimus dosing algorithm will likely be a first major step toward precision genotype-guided dosing in the transplantation setting. Optimal use of pharmacogenetics will require a deeper knowledge of additional genetics factors governing drug disposition, efficacy and toxicity. The application of the genomic advances to this field has the real potential of optimizing dosing strategies for immunosuppressive drugs, avoiding serious adverse and improving patient management after organ transplantation.

Ancestry as a genetic risk factor for transplantation outcomes

It is also worth noting the large impact that donor and recipient ancestry has in transplantation. Many studies have reported greater risks of rejection and mortality in African Americans when their organs are used as donor grafts or when they are a recipient of a graft [35-37]. Callender et al also confirmed lower kidney survival rates among African Americans compared to all other ethnic groups and showed that this ethnic group also harbored the shortest half-life after kidney transplantation: only 5.3 years versus 12.2 years for Asians, 10.2 years for individuals of European descent and 9.0 years for Hispanics [36].

In 2010, Li et al [38] published a pivotal study showing that Chinese and African human genomes differed by approximately 5 Mb of unique sequence and over 240 potential genes differing between these sequences. Such “pan-genome” data sets illustrate the differences between populations and emphasize the need for further characterization of such genomic differences in the transplantation arena. In addition to the highly polymorphic differences in HLA loci already evident in the different ethnic groups, variants like LoFs or hdCNVs with the potential of acting as incompatibility antigens, or genetic risk factors for postoperative transplant outcomes as well as pharmacogenetic markers, exhibit substantial variability interethnically.

Future Perspectives

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetics of Transplantation Outcomes
  5. Future Perspectives
  6. Disclosure
  7. Electronic Resources Described

The current pace of genomic discovery has revolutionized our understanding of the genetic architecture of common diseases ( [29]. In contrast, the pipeline of genomic research in transplantation has been relatively weak. This is surprising because in essence transplantation involves the direct exposure of a foreign donor genome into a recipient, creating unparalleled opportunities to investigate the implications this has on graft function. For example, recipient by donor genomic interactions may play a critical role in graft survival. However, analyses of these types will require large well-characterized data sets that have been harmonized for important clinical traits and genotyped using GWAS or preferably second-generation sequencing. While small candidate studies have started the collation of DNA and appropriately phenotyped recipients and donors, very few studies have sufficient numbers to be statistically powered to look at polygenic transplant-related intermediate phenotypes and outcomes. There is a clear need for wider collaborations among all appropriately harmonizable data sets using robust genomic platforms and statistical methods to bolster statistical power among existing studies. With such initiatives, we expect the genetic architecture of transplant rejection ranging from identification of LoF variants, which introduce new epitopes to donors, to genes involved in drug metabolism to change the way transplanted grafts are matched and/or recipients are monitored posttransplant. Furthermore, established genetic techniques such as hypothesis-free analyses and Mendelian randomization can be conducted in this setting to investigate the causal effect of modifiable risk factors on graft function and survival with the potential to discover new therapeutic targets.

Large-scale NIH networks such as The electronic Medical Records and Genomics (eMERGE) network integrating large-scale hospital GWAS biobank resource with EMRs ( will allow us a framework to consent, counsel and manage the return of results to physicians and graft recipients in a clinical environment, which will lead to greater patient monitoring and care. As waiting lists for kidney transplantation continue to grow (n = 97 856, September 2013) and with solid organ transplants costing from ∼$200 000 to $850 000 (USD) per patient, efforts such as eMERGE are currently attempting to integrate actionable pharmacogenomics data into patients' EMRs, and tacrolimus-related dosing is being conducted in at least one eMERGE site thus far, which will likely result in a significant benefit to society in the very near future.

Although beyond the scope of this review, a number of protein, RNA and microRNA biomarkers for rejection and graft damage injury are emerging that may aid in the diagnosis of rejection episodes. Specific biomarkers are also under investigation for immune quiescence and minimization of immune-suppressants (reviewed in [39]). We are now entering a period in genomics when tools are becoming available, which will dramatically change our understanding of the molecular mechanisms underpinning graft outcomes, and patient-specific treatments will begin to benefit transplant patient care and to define the future of transplantation.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetics of Transplantation Outcomes
  5. Future Perspectives
  6. Disclosure
  7. Electronic Resources Described

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.


Electronic Resources Described

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetics of Transplantation Outcomes
  5. Future Perspectives
  6. Disclosure
  7. Electronic Resources Described

A Catalog of Published Genome-Wide Association Studies: The electronic Medical Records and Genomics (eMERGE) network: