Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

A model of prediction system for adverse cardiovascular reactions by calcineurin inhibitors among patients with renal transplants using gene-based single-nucleotide polymorphisms


The application of pharmacogenomic information to diagnostic assays is expected to improve the prediction of drug efficacy and toxicity, leading to appropriate therapeutic regimens for individual patients. Cardiovascular events are common and severe adverse drug reactions (ADRs) among transplant patients treated with calcineurin inhibitors (CNIs). We conducted case-control association studies using 50,947 gene-based single-nucleotide polymorphisms (SNPs) to identify genetic variations that might be associated with cardiovascular risk factors in 72 renal transplant recipients with CNI therapy. The overall incidence of cardiovascular events was 13.9% (10/72) among patients receiving cyclosporine or tacrolimus; arrhythmias in six patients (8.3%), ischemic heart diseases in two patients (2.8%), and heart failure in two patients (2.8%). On the basis of results of the genome-wide association studies, we attempted to establish a scoring system to predict individual risks for cardiovascular toxicity of cyclosporine and tacrolimus. Estimation of the predictive performance was carried out by the use of internal leave-one-out cross-validation test. When we combined arrhythmia, ischemic heart disease and heart failure cases as subjects with a cardiotoxicity phenotype, nine of ten ADR patients and 50 of 62 non-ADR patients were correctly classified into the respective categories using the top eight SNPs. In addition, the proportion of individuals in the control population (n=246) with scores over the cut-off (11.0%) was close to the cardiovascular ADR frequency (8.3%) among renal transplant patients in the previous clinical study. Our results open the possibility that prediction of CNI-induced cardiovascular complications can lead to better prognosis and quality of life among kidney-transplant patients, and to improved immunosuppressive regimens.


Outcomes of organ transplantation have improved over the last two decades due to the availability of calcineurin inhibitors (CNIs; e.g., cyclosporine and tacrolimus) as immunosuppressive agents for the prevention of acute and chronic rejections. Although these agents bind primarily to different molecular targets [cyclophilin or FK506-binding protein-12 (FKBP12), respectively], they exert similar pharmacological effects on signal-transduction pathways in T lymphocytes. It is generally accepted that by blocking dephosphorylation and subsequent translocation of the nuclear factor of activated T-cells (NFAT), CNIs inhibit the production of interleukin-2 (IL-2) (Shaw et al. 1988; McCaffrey et al. 1993). Extensive clinical experience has demonstrated that although both drugs are indispensable for immunosuppression after organ transplantation, due to a narrow window for therapeutic dosage it is difficult to find the dose appropriate to each patient. Under-dosing can lead to rejection of the transplanted organ, while over-dosing increases the risks of severe drug-specific adverse events including renal, pancreatic, and cardiovascular toxicity.

Cardiovascular complications, such as heart failure, myocardial infarction, and sudden cardiac death, are the leading causes of death following kidney transplantation, and CNIs are thought to be responsible for inducing these adverse cardiovascular reactions. Hypertension, myocardial ischemia, and arrhythmias have been reported in Japanese patients given tacrolimus therapy after renal transplantation (Seino et al. 2003), and multi-center studies in the United States and Europe have also documented hypertension, arrhythmias, and angina in recipients of kidney transplants who were treated with either cyclosporine or tacrolimus (Pirsch et al. 1997; Mayer et al. 1997; Margreiter, 2002).

The dosage of these drugs is usually decided by monitoring their concentrations in blood; however, the pharmacokinetic profiles of CNIs are not always associated with clinical efficacy or toxicity (Kahan and Grevel 1988; Kahan et al. 1990; Caruso et al. 2001). The evidence suggests that adverse drug reactions (ADRs) caused by CNIs might reflect not only differences in pharmacokinetics from one individual to another, but also variations in genes that alter drug response. Thus, a genome-wide approach seemed to be indicated for elucidating the mechanism of CNI-induced ADR.

We describe here the results of genome-wide association studies using single-nucleotide polymorphisms (SNPs) from the JSNP database (Haga et al. 2002) to identify genetic variations that might confer susceptibility to cardiovascular events under immunosuppressive regimens. On the basis of genotypic information from eight SNP loci at which we observed significant association with ADR, we developed a prediction-scoring system that was able to clearly separate two groups of patients; i.e., with or without ADRs.

Materials and methods


DNAs from a total of 72 patients who were treated with either cyclosporine or tacrolimus after renal transplantation were obtained at the University of Tokyo for a case-control association study. Controls consisted of 246 members of the general population recruited through the University of Tokyo. All subjects were Japanese and provided written informed consent to participate in the study in accordance with the process approved by the Ethical Committees at the SNP Research Center, The Institutes of Physical and Chemical Research (RIKEN), Yokohama, Japan, and The Institute of Medical Science at the University of Tokyo, Tokyo, Japan.

Dosage and use of these drugs were pursuant to the standards indicated for kidney transplantation by the Ministry of Health, Labor and Welfare, Japan, and the doses were adjusted for each patient to maintain blood concentrations within the target range. In the present study, the median loading doses given by oral administration were 2.8 and 0.082 mg/kg/day for cyclosporine and tacrolimus, respectively.

SNPs and genotyping

We selected 52,608 SNPs from the JSNP database according to tag-SNP information (Tsunoda et al. 2004) for genome-wide association studies. The SNPs were genotyped using the multiplex PCR-based Invader assay (Third Wave Technologies, Madison, Wis.). DNA extraction, design of PCR primers, multiplex PCR experiments, and Invader assay were performed as previously described (Iida et al. 2001; Ohnishi et al. 2001).

Statistical analyses and prediction-score system

We carried out statistical analysis for association by comparing two patient groups, with or without cardiovascular ADR, by means of Fisher’s exact test using dominant-inheritance and recessive-inheritance models. SNPs were rank-ordered according to the lowest P value in both models. When a “risk allele” contributed to cardiovascular events in a dominant-inheritance manner, we scored an individual as 2, 1 or 0 according to homozygosity for the risk allele, heterozygosity for the risk allele, or homozygosity for the low-risk allele, respectively. If a risk allele contributed to cardiotoxicity in a recessive-inheritance manner, we assigned scores of 2 to individuals with homozygosity for the risk allele and zero to individuals with other genotypes. Then we calculated a prediction score for each patient by adding his or her scores from each of the SNP loci. The discriminating SNPs for the prediction-scoring system were chosen according to a “leave-one-out” procedure.


Patient characteristics

Table 1 summarizes clinical information for the 72 CNI-treated patients in the study. Cardiovascular events were reported in ten patients (13.9%). Among ten patients with the cardiovascular ADR, we noted arrhythmias in six cases (8.3%) and premature contraction in three, as well as atrial fibrillation, sinus tachycardia, and QT prolongation in one patient each. The frequency of each of ischemic heart diseases and heart failure was observed in two each of these patients. The 12-h trough blood concentrations at the onset of cardiac symptoms were measured in six patients treated with cyclosporine and in two with tacrolimus. Mean trough concentrations were within the therapeutic range at the time of onset of adverse reactions in most of the patients with cardiovascular complications, and were not statistically different from those in patients without ADR (Fig. 1).

Table 1 Patient characteristics and cardiovascular events
Fig. 1

Twelve-hour trough concentrations of cyclosporine (a) and tacrolimus (b) in whole blood from renal-transplant recipients at the point of the absence or presence of ADR. Dotted lines represent the maximum whole-blood therapeutic range generally accepted for treatment of renal-transplant patients

Genome-wide association studies

Initially, we performed a case-control association study with six arrhythmia patients and 62 non-ADR patients by genotyping recipients of kidney transplants using a high-throughput multiplex PCR-Invader assay method and 52,608 gene-based SNPs. These SNPs were selected from the JSNP database based on the haplotype block structure estimated previously (Tsunoda et al. 2004). The success rate for this genotyping was 96.8% (50,947 SNPs), and the distribution of P values is shown in Table 2. We observed associations between two patient groups at ten SNP loci for which P values were <0.001 by a dominant-inheritance model, and at seven SNP loci with a recessive-inheritance model. Then, we combined arrhythmia, ischemic heart disease and heart failure cases as subjects with a cardiotoxicity phenotype and performed a case-control association study. We observed associations with P values of <0.001 at 14 SNPs according to the dominant-inheritance model between the two groups; eight SNPs revealed P values of <0.001 by the recessive-inheritance model (Table 2).

Table 2 Distribution of P values in genotyping of patients with or without cardiovascular events

Development of a prediction-scoring system

On the basis of results of the genome-wide association studies, we attempted to establish a scoring system to predict individual risks for cardiovascular toxicity of cyclosporine and tacrolimus. To determine the optimal number of the SNPs for prediction of ADR risk, an internal leave-one-out cross validation was employed. One of the patients was withheld, and a rank-ordered SNP list and prediction score were generated from the remaining subjects. The withheld patient was tested whether the patient was classified as ADR or non-ADR groups according to a cut-off score at the middle of the medians for ADR and non-ADR groups. The process was repeated for each of all subjects. The rates of correct classification into ADR and non-ADR groups, sensitivity and specificity, were calculated. When we combined arrhythmia, ischemic heart disease and heart failure cases, the number of the SNPs used for the scoring influenced the sensitivity and specificity in the prediction (Fig. 2). Using the eight SNPs that showed the most significant associations, nine of ten ADR patients and 50 of 62 non-ADR patients were correctly classified into the respective categories. Thus, the maximum values for sensitivity and specificity were 90.0 and 80.6%, respectively (Fig. 2). For comparison between arrhythmia and non-ADR groups, we obtained the highest values for sensitivity and specificity, 83.3 and 50.0%, respectively, when the top eight SNPs on each rank-ordered list were used for the scoring in the cross-validation (data not shown). In fact, the prediction-scoring system using the data from all subjects with the top eight SNPs for a combination of three cardiovascular ADRs clearly separated the two patient groups (Fig. 3). To evaluate the system further, we investigated DNA from 246 members of the general Japanese population. As shown in Fig. 3, the scores of these controls ranged from zero to ten with a median of four. When we set a cut-off score at 6.5 using the average for median values of ADR and non-ADR groups, the proportion of subjects in the control population with scores higher than the cut-off value was 11.0%.

Fig. 2

Optimization of the number of SNPs for prediction-scoring system using the data from combination analysis of arrhythmia, ischemic heart disease and heart failure cases. Solid and dotted lines represent the sensitivity and specificity calculated by the prediction scores in leave-one-out test for cross-validation using the top 100 SNPs on the rank-ordered lists

Fig. 3

Prediction scores using the top eight SNPs in the rank-ordered list with the data from combination analysis of arrhythmia, ischemic heart disease and heart failure cases. Filled circles indicate scores in 72 learning cases (ten patients with ADR, 62 without). Open circles represent scores for 246 individuals from the general Japanese population


ADRs often deteriorate the quality of life and are sometimes fatal. In an attempt to predict genetic risks for these events, many investigators have looked for associations between ADRs and polymorphisms in drug-metabolizing enzymes; however, few useful examples have emerged to allow consistent prediction of ADR prior to drug administration. ADRs due to overdosing frequently correlate with blood concentrations of parent drugs or active metabolites, an indicator of systemic drug exposure. However, in the present study, the mean trough concentrations of CNIs in renal transplant patients with cardiovascular ADRs were comparable to those in patients with normal cardiac function (Fig. 1), suggesting that some biological factor(s), not an excess of drug, might be responsible for the emergence of cardiovascular toxicity after administration of cyclosporine or tacrolimus. Therefore, we included two subjects with extremely higher whole-blood concentrations compared with the maximum therapeutic range in the case-control association studies.

We profiled the SNP-genotype patterns of renal transplant patients according to the number of risk alleles each carried, with a view to predicting risk for ADR by means of a scoring system based on data for the SNPs significantly associated with CNI-induced ADR in our patient samples. Because of the small number of subjects, estimation of the predictive performance was not done in an independent group, but by the use of internal leave-one-out cross-validation test. In the validation study, the maximum values for sensitivity and specificity were 90.0 and 80.6%, respectively, in the combination analysis of arrhythmia, ischemic heart disease and heart failure cases, while we obtained the highest values for sensitivity and specificity, 83.3 and 50.0%, respectively, in the comparison between arrhythmia and non-ADR groups. Thus, the prediction-scoring system using data from the combination analysis of three cardiovascular events was more promising compared with the comparison analysis between the arrhythmia and non-ADR groups, probably due to the small number of the arrhythmia subjects. Moreover, the proportion of individuals in the control population (n=246) with scores over the cut-off value (11.0%) was close to the cardiovascular ADR frequency (8.3%) among transplant patients in the previous clinical study (Seino et al. 2003), indicating the feasibility of using our scoring system as a means of predictive genetic testing to distinguish between ADR and non-ADR patients before administration of CNI drugs. Considering the CNI-induced cardiovascular ADR frequency, the positive predictive value was calculated to be only 29.6%, while the negative predictive value was 98.9%. These results indicate that about 70% of patients should show false-positive results, but suggest that our system can be used to exclude high-risk patients with CNI-induced cardiotoxicity among renal transplant recipients.

Coronary vasospasm is generally suspected as the cause of the CNI-associated arrhythmias and myocardial ischemia. In fact, reversible vasospasm was observed in several CNI-treated patients in the present study (data not shown). Proposed mechanisms by which these drugs cause coronary vasospasm include inhibition of expression of the gene encoding endothelial nitric oxide synthetase (eNOS) and activation of the endothelin (ET) gene (Weis et al. 2000). In support of this hypothesis, several groups have demonstrated that administration of CNIs significantly increases concentrations of ET, a potent vasoconstrictive peptide, in plasma from organ-transplant recipients (Haug et al. 1995; Slowinski et al. 2002). In addition, it is well-known that the endothelial dysfunction is often observed in patients with heart failure. Several in vitro and clinical studies demonstrated that tumor necrosis factor-α (TNF-α) induced the endothelial dysfunction as a result of downregulation of eNOS expression and an increase in apoptosis (Agnoletti et al. 1999; Fichtscherer et al. 2001). Thus, unknown factor(s) associated with endothelial function might be responsible for the risk of these three cardiovascular ADRs, arrhythmias, ischemic heart disease and heart failure.

Table 3 summarizes the top eight SNPs used for our prediction-scoring system. No association of any of those genes with cardiac functions has been reported; however, variations in the functions and/or expression levels of these genes do appear to contribute to the cardiovascular toxicity of CNI drugs. For example, one of them, SLC12A8, belongs to the cation-coupled chloride ion co-transporter family (SLC12) (Hebert et al. 2004), of which two other members (SLC12A6 and SLC12A7) possess transport activity in potassium ion efflux (Mount et al. 1999) and might play important roles in the genesis of arrhythmias following myocardial ischemia (Yan et al. 1996). Although the substrates transported by SLC12A8 are unknown, our data suggest that polymorphism in this gene may be an important determinant of risk for cardiovascular toxicity with CNI therapy.

Table 3 Top eight SNPs associated with arrhythmias, ischemic heart diseases and heart failure

In conclusion, through a genome-wide association study we identified genes that might be related to the adverse cardiovascular events often observed in renal transplantation recipients who are given CNI immunosuppressants. Although further validation and improvement of the scoring system described here will be required, using a very large number of patients, our results open the possibility that prediction of CNI-induced cardiovascular complications can lead to better prognosis and quality of life among kidney-transplant patients, and to improved immunosuppressive regimens.


  1. Agnoletti L, Curello S, Bachetti T, Malacarne F, Gaia G, Comini L, Volterrani M, Bonetti P, Parrinello G, Cadei M, Grigolato PG, Ferrari R (1999) Serum from patients with severe heart failure downregulates eNOS and is proapoptotic: role of tumor necrosis factor-alpha. Circulation 100:1983–1991

    CAS  Article  Google Scholar 

  2. Caruso R, Perico N, Cattaneo D, Piccinini G, Bonazzola S, Remuzzi G, Gaspari F (2001) Whole-blood calcineurin activity is not predicted by cyclosporine blood concentration in renal transplant recipients. Clin Chem 47:1679–1687

    CAS  Article  Google Scholar 

  3. Fichtlscherer S, Rossig L, Breuer S, Vasa M, Dimmeler S, Zeiher AM (2001) Tumor necrosis factor antagonism with etanercept improves systemic endothelial vasoreactivity in patients with advanced heart failure. Circulation 104:3023–3025

    CAS  Article  Google Scholar 

  4. Haga H, Yamada R, Ohnishi Y, Nakamura Y, Tanaka T (2002) Gene-based SNP discovery as part of the Japanese Millennium Genome Project: identification of 190,562 genetic variations in the human genome. Single-nucleotide polymorphism. J Hum Genet 47:605–610

    CAS  Article  Google Scholar 

  5. Haug C, Duell T, Lenich A, Kolb HJ, Grunert A (1995) Elevated plasma endothelin concentrations in cyclosporine-treated patients after bone marrow transplantation. Bone Marrow Transplant 16:191–194

    CAS  PubMed  Google Scholar 

  6. Hebert SC, Mount DB, Gamba G (2004) Molecular physiology of cation-coupled Cl- cotransport: the SLC12 family. Pflugers Arch 447:580–593

    CAS  Article  Google Scholar 

  7. Iida A, Sekine A, Saito S, Kitamura Y, Kitamoto T, Osawa S, Mishima C, Nakamura Y (2001) Catalog of 320 single nucleotide polymorphisms (SNPs) in 20 quinone oxidoreductase and sulfotransferase genes. J Hum Genet 46:225–240

    CAS  Article  Google Scholar 

  8. Kahan BD, Grevel J (1988) Optimization of cyclosporine therapy in renal transplantation by a pharmacokinetic strategy. Transplantation 46:631–644

    CAS  Article  Google Scholar 

  9. Kahan BD, Shaw LM, Holt D, Grevel J, Johnston A (1990) Consensus document: Hawk’s Cay meeting on therapeutic drug monitoring of cyclosporine. Clin Chem 36:1510–1516

    CAS  Article  Google Scholar 

  10. Margreiter R; European Tacrolimus vs Ciclosporin Microemulsion Renal Transplantation Study Group (2002) Efficacy and safety of tacrolimus compared with cyclosporine microemulsion in renal transplantation: a randomised multicentre study. Lancet 359:741–746

    CAS  Article  Google Scholar 

  11. Mayer AD, Dmitrewski J, Squifflet JP, Besse T, Grabensee B, Klein B, Eigler FW, Heemann U, Pichlmayr R, Behrend M, Vanrenterghem Y, Donck J, van Hooff J, Christiaans M, Morales JM, Andres A, Johnson RW, Short C, Buchholz B, Rehmert N, Land W, Schleibner S, Forsythe JL, Talbot D, Pohanka E, et al. (1997) Multicenter randomized trial comparing tacrolimus (FK506) and cyclosporine in the prevention of renal allograft rejection: a report of the European Tacrolimus Multicenter Renal Study Group. Transplantation 64:436–443

    CAS  Article  Google Scholar 

  12. McCaffrey PG, Luo C, Kerppola TK, Jain J, Badalian TM, Ho AM, Burgeon E, Lane WS, Lambert JN, Curran T et al (1993) Isolation of the cyclosporine-sensitive T cell transcription factor NFATp. Science 262:750–754

    CAS  Article  Google Scholar 

  13. Mount DB, Mercado A, Song L, Xu J, George AL Jr, Delpire E, Gamba G (1999) Cloning and characterization of KCC3 and KCC4, new members of the cation-chloride cotransporter gene family. J Biol Chem 274:16355–16362

    CAS  Article  Google Scholar 

  14. Ohnishi Y, Tanaka T, Ozaki K, Yamada R, Suzuki H, Nakamura Y (2001) A high-throughput SNP typing system for genome-wide association studies. J Hum Genet 46:471–477

    CAS  Article  Google Scholar 

  15. Pirsch JD, Miller J, Deierhoi MH, Vincenti F, Filo RS (1997) A comparison of tacrolimus (FK506) and cyclosporine for immunosuppression after cadaveric renal transplantation. FK506 Kidney Transplant Study Group. Transplantation 63:977–983

    CAS  Article  Google Scholar 

  16. Seino Y, Hori M, Sonoda T (2003) Multicenter prospective investigation on cardiovascular adverse effects of tacrolimus in kidney transplantations. Cardiovasc Drugs Ther 17:141–149

    CAS  Article  Google Scholar 

  17. Shaw JP, Utz PJ, Durand DB, Toole JJ, Emmel EA, Crabtree GR (1988) Identification of a putative regulator of early T cell activation genes. Science 241:202–205

    CAS  Article  Google Scholar 

  18. Slowinski T, Subkowski T, Diehr P, Bachert D, Fritsche L, Neumayer HH, Hocher B (2002) Interaction of the endothelin system and calcineurin inhibitors after kidney transplantation. Clin Sci (Lond) 103(Suppl 48):396S–398S

    CAS  Article  Google Scholar 

  19. Tsunoda T, Lathrop GM, Sekine A, Yamada R, Takahashi A, Ohnishi Y, Tanaka T, Nakamura Y (2004) Variation of gene-based SNPs and linkage disequilibrium patterns in the human genome. Hum Mol Genet 13:1623–1632

    CAS  Article  Google Scholar 

  20. Weis M, Wildhirt SM, Schulze C, Pehlivanli S, Rieder G, Wolf WP, Wilbert-Lampen U, Meiser BM, Enders G, von Scheidt W (2000) Coronary vasomotor dysfunction in the cardiac allograft: impact of different immunosuppressive regimens. J Cardiovasc Pharmacol 36:776–784

    CAS  Article  Google Scholar 

  21. Yan GX, Chen J, Yamada KA, Kleber AG, Corr PB (1996) Contribution of shrinkage of extracellular space to extracellular K+ accumulation in myocardial ischaemia of the rabbit. J Physiol 490:215–28

    CAS  Article  Google Scholar 

Download references


We thank the following: R. Ohishi for her excellent technical assistance; T. Yamazaki and K. Nakazono for their expertise in computer programming and statistical analyses; and all members of the SNP Research Center, The Institute of Physical and Chemical Research, for their contribution to the completion of our study.

Author information



Corresponding author

Correspondence to Yozo Ohnishi.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Mushiroda, T., Saito, S., Tanaka, Y. et al. A model of prediction system for adverse cardiovascular reactions by calcineurin inhibitors among patients with renal transplants using gene-based single-nucleotide polymorphisms. J Hum Genet 50, 442–447 (2005).

Download citation


  • SNP
  • Genome-wide
  • Cyclosporine
  • Tacrolimus
  • Cardiotoxicity
  • Association study
  • Prediction score

Further reading


Quick links