A new donors’ CYP3A5 and recipients’ CYP3A4 cluster predicting tacrolimus disposition, and new-onset hypertension in Chinese liver transplant patients

Aim The purpose of the current study was to investigate individualized therapy of tacrolimus (Tac), as well as complications after liver transplantation (LT) with the known genetic determinants and clinical factors. Methods In this retrospective study, two cohorts (n=170) from the China Liver Transplant Registry (CLTR) database from July 2007 to March 2015 were included. Results Both donors’ CYP3A5*3 and recipients’ CYP3A4*1G had a correlation with Tac pharmacokinetics at four weeks (all P<0.05), except recipients’ CYP3A4*1G nearly had an association at week 2 (P=0.055). The model of donors’ CYP3A5*3, recipients’ CYP3A4*1G, and total bilirubin (TBL), for the prediction of Tac disposition, was better than donors’ CYP3A5*3 only at week 1, 2, and 3 (P=0.010, P=0.007, and P=0.010, respectively), but not apparent at week 4 (P=0.297). Besides, when the P value was greater than or equal to 0.6685 after considering the false-positive rate R=10%, the patients were considered to have a faster metabolism, according to the mentioned model. Interestingly, we found that if more than or equal to two alleles A were present in the combination of donors’ CYP3A5*3 and recipients’ CYP3A4*1G genotype, there was a lower Tac C/D ration at week 1, 2, and 3 (P<0.001, P=0.001, and P<0.001), except at week 4 (P=0.082), and the probability of new-onset hypertension was lesser (P<0.001). Conclusions These data provided a potential basis for a comprehensive approach to predicting the Tac dose requirement in individual patients and provided a strategy for the effective prevention, early diagnosis of new-onset hypertension in Chinese LT recipients.


INTRODUCTION
Tacrolimus (Tac) is an immunosuppressant drug that belongs to the class of calcineurin inhibitors and has an important role in the prevention of allograft rejection in liver transplantation (LT) [1,2]. It is characterized by a narrow therapeutic index and large interpatient variabilities in its pharmacokinetic and pharmacodynamic profiles, and it displays a wide range of potentially severe drug-related toxicities [3][4][5][6]. Regardless of these unfavorable characteristics, Tac is recognized as one of the most important immunosuppressants in solid-organ transplantation [7,8]. Research Paper www.impactjournals.com/oncotarget CYP3A5 is the main catalyst of Tac metabolism, as we known. The 6986A > G variant in intron 3 of CYP3A5 (CYP3A5*3) (rs776746), is known as one of the most important single nucleotide polymorphisms (SNPs) in CYP3A5 [9]. The CYP3A5*3 AA or AG genotype (i.e., patients expressing CYP3A5 protein) have a substantially higher Tac clearance, resulting in markedly higher Tac dose requirements when compared with that of CYP3A5*3 GG genotype (i.e. patients not expressing CYP3A5 protein) [10][11][12]. On the basis of these findings, it has been hypothesized that individualizing the initial Tac dosage based on CYP3A5 genotype (i.e., 0.30 mg/ kg/day for CYP3A5*3 AA or AG genotype carriers and 0.15 mg/kg/day for CYP3A5*3 GG genotype patients, instead of the standard 0.20 mg/kg/day for all patients) might help avoid underexposure and overexposure to Tac early after LT [13]. As underexposure to Tac is related with an increased risk of acute rejection [14] and overexposure is related with an increased risk of drug-related toxicities, such as new-onset hypertension and new-onset diabetes mellitus after LT [15][16][17], individualized dosing might improve the quality of life and clinical outcomes after transplantation. There was a prospective randomized study showed that more number of patients within the desired Tac target range early after transplantation and a faster achievement of Tac trough (C0) levels according to the CYP3A5 genotype of the patient [18], but it is essential to realize some limitations that are focusing solely on the CYP3A5 genotype [19,20].
Many factors including both clinical [e.g. age, sex, hemoglobin (Hb), albumin, total bilirubin (TBL)] and genetics [e.g. CYP3A5 and CYP3A4 SNPs] have been identified to affect the pharmacokinetics of Tac [21][22][23]. Regarding the genetics, it is greatly recognized that the CYP3A5 *3 genotype has a noticeable effect on Tac pharmacokinetics, whereas research on the effect of CYP3A4 SNPs is limited [24,25].
The CYP3A4*1G allele (rs2242480), a novel G-to-A substitution at position 82266 in intron 10 has been identified in the Japanese population [19,26]. CYP3A4*1G can increase the activity of CYP3A4 enzyme [27,28], and several studies indicate that this SNP is related to the pharmacokinetics of Tac [2,29], as well as responsible for the interindividual differences in cyclosporine disposition [30,31]. Hence, we advocated that the interindividual differences in Tac pharmacokinetics in vivo might also be partially owing to the interindividual differences in the CYP3A4*1G activity. Based on this frame, we investigated the relationship between the CYP3A5*3 and CYP3A4*1G genotypes in liver transplant donors and recipients, and on the pharmacokinetics of Tac, and on the complication of liver transplantation (e.g. new-onset diabetes and new-onset hypertension), considering the known clinical determinants of Tac disposition.

Clinical characteristics
The clinical characteristics for all population (n=170), training set (n =100) and validating set (n=70) were shown in Table 1. All patients were Chinese in this study and tested 4 weeks in both the training set and validating sets after LT. The average age of all patients was 47.9±9.5 years, and the average weight was 66.3±12.3 kg. The age of training set was younger than that of the validating set; however, this trend did not to be statistically significant. The majority causes of our transplant patients were hepatocellular carcinoma caused by hepatitis B virus.

Effect of SNPs on Tac C/D ratios
The Tac C/D ratios at different time periods after LT were compared among patients with different donors' and recipients' CYP3A5*3 polymorphisms, as well as CYP3A4*1G polymorphisms in the training set (Table  3). There was a correlation of recipients' CYP3A4*1G genotype with Tac C/D ratios at week 1, 3, and 4(P = 0.046, 0.015, and 0.024, respectively), and nearly at week 2(P=0.055). However, the association between donors' CYP3A4 *1G gene polymorphisms and Tac C/D ratios was not found. Contrary to CYP3A4 *1G genotype, donors' CYP3A5*3 had a correlation with Tac pharmacokinetics at week 1, 2, 3, and 4(P< 0.001, P =0.032, P = 0.048, and P = 0.003, respectively), but for recipients' CYP3A5*3, the association with Tac disposition was just found at week 1 and week 3 (P = 0.014 and P = 0.038 respectively).

Effect of combination SNPs on Tac C/D ratios
Donors' CYP3A5*3 allele A and recipients' CYP3A4*1G allele A were shown to be related to faster Tac metabolism as stated in Table 3. Hence, the allele A was further explored in a combination analysis in the training set. The associations between the number of alleles A with a fast metabolism and Tac C/D ratios were shown in Table 4. When the number of alleles A was greater than or equal to two, the patients were found to have lower Tac C/D ratios at week 1, 2, and 3 (P < 0.001, P =0.001, and P < 0.001, respectively), and closed to significant at week 4 (P = 0.082). Table 5 showed the multivariate linear regression models predicting Tac daily dose requirements, Tac C0 level, dose-corrected Tac C0 level at week 1, 2, 3, and 4 in the training set. The incorporated cofactors included Hb, the number of allele A (combination of donors' CYP3A5 *3 and recipients' CYP3A4*1G genotype), TBL as well as glutamic oxalacetic transaminase. All these factors had been reported to have potential effects on Tac pharmacokinetics and entered into the validating set in the multivariate linear regression analysis (Table 6), which showed that TBL was the second explanatory variable to be retained after the number of allele A, in the list of Tac pharmacokinetic parameters.

The number of alleles A (combination of donors' CYP3A5 *3 and recipients' CYP3A4*1G genotype) and TBL predicting Tac disposition: multivariate linear regression analysis
Donors' CYP3A5 genotype, recipients' CYP3A4 genotype, and TBL predicting the daily Tac dose corrected by weight: receiver operating characteristic (ROC) curve As shown in Figure 1   Data was presented as mean±standard deviation. Comparison between groups was performed by t-test. P<0.05 was considered significant. The data was presented as mean±standard deviation. The comparison between groups was performed by Chi-square. P<0.05 was considered significant. <2: GGGG, GGGA; ≥2: GGAA, GAAA, AAAA.  Table 7, diagnosis point of 0.6685 obtained after the false positive rate R=10% was taken. If the prediction probability value was greater than or equal to 0.6685 it was considered positive (i.e. diagnosed with a fast metabolism), whereas a value less than 0.6685 was diagnosed with a slow metabolism. www.impactjournals.com/oncotarget

New-onset hypertension according to the combination of CYP3A4 and CYP3A5 polymorphisms
As shown in Table 8 that in our all study population, the association was observed between new-onset hypertension and the amounts of allele A (P=0.001), which the combination of donors' CYP3A5 *3 genotype and recipients' CYP3A4*1G genotype. Besides, there was a difference between alleles A with greater than two and lesser than or equal to two. With increasing number of alleles associated with fast metabolism, the patients were found to have an increasingly low probability of the occurrence of hypertension. However, it was not apparent in the aspect of new-onset diabetes (P=0.637) and newonset hyperlipidemia (P=0.941).

DISCUSSION AND CONCLUSIONS
It was well known that CYP3A5*3 played a crucial role in the metabolism of Tac disposition. Individuals with AA or AG genotype were CYP3A5 expressors and metabolized Tac; however, those with the GG genotype barely metabolized Tac [10][11][12]. Our patients were mostly carriers of CYP3A5*3 GG and AG genotype; only 11 patients (donors and recipients) were carriers of AA genotype in the training set, and the predominant allele of CYP3A5*3 was up to 70-74%. In African-Americans, the CYP3A5*3 allele frequency was up to 55% compared with that observed in Caucasian subjects (85-95%) [9].
The data obtained in this study were between findings in American and Caucasian populations. Similarly, CYP3A4*1G allele frequency varied among different ethnic groups: 24.9% in Japanese and 22.1% in Chinese [19,26]. In this study, the CYP3A4*1G 22-29% allele distribution in our patients was similar to that reported previously. In this context, compared to the liver, the role of the small intestine in the metabolism of drugs in receptor organisms could not be ignored. Since most clinical drugs delivers was oral, the role of the small intestine in drug absorption link was over-emphasized in the past, and the ability of absorption was underestimated.
The expression of small intestinal CYP3A4 was much higher than that of CYP3A5, accounting for 73% of the total CYP3A [33]. Thus, we were intrigued to examine the impact of combination with CYP3A5 and CYP3A4 polymorphisms on Tac pharmacokinetics in LT, considering the known clinical determinants. First of all, our data indicated that donor CYP3A5and recipient CYP3A4-mediated Tac metabolisms were both critical to Tac disposition in vivo [2]. This was consistent with the fact that CYP3A4 existed mostly in the gastrointestinal tract of the recipient and CYP3A5 presented mainly in the liver of the donor, both being sites of drug metabolism. Almost all studies had reported a lower Tac exposure and/or a higher dose requirement in individuals who were CYP3A5 expressors (harboring CYP3A5*3 AA or AG genotype) than that in nonexpressors (CYP3A5*3 GG genotype) [34]. Besides, Qiu XY et al. observed in their study performed in 103 renal transplant recipients that the CYP3A4*1G AA genotype was found to have a lower dose-adjusted concentration [27]. The study was consistent with our results that the potentially higher metabolic capacity of CYP3A4*1G in patients with A allele in LT and we supposed that the association of recipients' CYP3A5*3 with Tac disposition was because of the gene linkage disequilibrium with recipients' CYP3A4*1G partially.
Secondly, TBL was the third variable that independently predicted that Tac pharmacokinetics in addition to donors' CYP3A5 and recipients' CYP3A4 genotype. Plasma TBL was mainly related to hemeoxygenase (HO) which would influence the metabolism of heme consisted of hemoglobin [35], and in line with the strong binding of Tac to the red cell [36]. Besides, TBL could represent the function of the donor's liver, where Tac was mostly metabolized. Notably, in our study, the selection of the study population might be responsible for the fact that no other biochemical or clinical variable predicted Tac pharmacokinetics. The choice of the Chinese population of adult liver transplant recipients who were tested 4 weeks after transplantation showed that some variables known to be associated with Tac disposition (i.e., ethnicity) could not affect Tac pharmacokinetics in our study. In addition, major drug-drug and drugfood interactions were avoided because the use of drugs and food that were known to either inhibit or induce CYP3A isoenzymes or to interfere with the absorption, distribution, metabolism, or excretion of Tac was not allowed, other than corticosteroids. Thirdly, in the multivariate linear regression analysis, when the recipients' CYP3A4*1G and donors' CYP3A5*3 polymorphisms were combined, we found that the number of allele A, which combination of donors' CYP3A5*3 and recipients' CYP3A4*1G genotypes, correlated notably with Tac C/D variation at four weeks. Besides, extensive metabolizers, with the number of alleles A greater than or equal to two, showed lower dose-adjusted blood concentration than that of poor metabolizers with the number of alleles A less than two. Furthermore, the model of donors' CYP3A5*3, recipients' CYP3A4*1G, and TBL, for the prediction of Tac disposition was better than the model of donors' CYP3A5*3 only at week 1, 2, and 3, the reasons for did not apparent at week 4 maybe were that (1) the concentration of Tac became steady at week 4 and the role of gene turned into small, especially CYP3A4*1G; (2) liver function presented by TBL transformed into normal. However, these views need to be confirmed in the further study with larger samples. We established a digital model to guide the clinically use of Tac; when the calculated P value was greater than or equal to 0.6685, the patients belonged to the category of a faster metabolism, which was consistent with our experimental results.
Fourthly, when combining the donors' CYP3A5*3 and recipients' CYP3A4*1G polymorphisms, there was a correlation between new-onset hypertension and the number of allele A, which in the combination of donors' CYP3A5*3 genotype and recipients' CYP3A4*1G genotype. If the number of alleles A more than two, the likelihood of new-onset hypertension was less. One of the reasons may be that allele A could affect the metabolism of Tac, which caused the development of hypertension by activating the renal sodium chloride co-transporter (NCC) [3], and the previous study also reported that Tac led to renal vasoconstriction and nephrotoxicity further confirmed our results [37,38].
Overall, our results suggested that the donors' CYP3A5*3, recipients' CYP3A4*1G genotype, and TBL had a major influence on Tac exposure. Notably, to our knowledge, this was the first time to define individualized Tac doses in liver transplant patients according to our digital model consisted of genotypic CYP3A5*3, CYP3A4*1G, and clinical TBL, in a Chinese population. However, two limitations need to be acknowledged. Firstly, this study was based on small Chinese cohorts, most of them had hepatitis B virus-related liver diseases.
Secondly, this was an observational study, and the basis of every important finding still required further explanation. In the future, a well-controlled clinical study was warranted to investigate this issue.
Furthermore, to date, this was the first study to explore the number of alleles A was associated with Tac disposition in LT in Chinese patients according to CYP3A5 *3 and CYP3A4*1G combinations. This may allow the prevention of liver graft rejection and improve the safety profile of Tac. Besides, to our limited knowledge, this was the first time to define that the number of alleles A was associated with new-onset hypertension in liver transplant patients. This finding could be clinically relevant for the effective prevention, early diagnosis, and treatment of new -onset hypertension in Chinese LT recipients.  The data was presented with count (percentage). The comparison between groups was performed by Chi-square. P<0.05 was considered significant. ≤2: GGGG, GGGA, GGAA; >2: GAAA, AAAA.

Ethics statement
Informed consent was obtained from all donors and recipients. Each organ donation or transplant was approved by the Institutional Review Board, First Hospital Affiliated Shanghai Jiao Tong University, strictly under the guidelines of the Ethics Committee of the hospital, the current regulation of the Chinese Government, and the Declaration of the Helsinki. No donor's livers were harvested from executed prisoners.

Data collection
On the basis of previous studies, we used the Tac serum concentration to dose ratios (C/D ratio) for 28 days after transplantation as an index of Tac pharmacokinetics [32]. Trough blood concentration of Tac (ng/mL) was detected by PRO-Trac TM II Tac ELISA kit (DiaSorin Inc., USA) with microparticle enzyme immunoassay (ELx800NB analyzer, BioTek, USA). The daily dose (mg) of Tac was recorded, and the weight-adjusted dosage (mg/ kg/d) was calculated. The Tac C/D ratio was calculated by dividing the Tac trough concentration (C0) by the corresponding weight-adjusted dosage. The results of the laboratory tests were also recorded. The average clinical data in the different periods were calculated to represent the corresponding clinical status.

Statistical analysis
The Kolmogorov-Smirnov test was used to check for normality. The Hardy-Weinberg equilibrium test was performed using an appropriate χ 2 test. Pairwise r 2 and D′ values for linkage disequilibrium were calculated using SHEsis software (http://analysis.bio-x.cn/myAnalysis. php). SPSS version 19.0 (SPSS Inc., Chicago, IL, USA) was used to complete other statistical analyses. Quantitative variables were expressed as mean ± SD and compared by Student's t-test or Wilcoxon-Mann-Whitney test. Categorical variables were presented as values (percentages) and compared using Fisher's exact test and Pearson's χ 2 test. Exploratory univariate correlation analysis (Spearman's correlation coefficient) was performed to explore whether a specific covariate potentially affected the Tac pharmacokinetics. Tac does, Tac C0, and dose-corrected Tac C0 were used as dependent variables. All covariates that correlated with the Tac pharmacokinetic parameters at a P value < 0.2 in univariate correlation analysis were retained and entered into the multivariate linear regression model. The models were calculated in binary logistic regression, transformed in a new variable, and then compared using receiver operating characteristic (ROC) curve. According to the linear interpolation method, the false positive rate R = 10% was taken after points for diagnosis. In addition, categorical covariates were coded with a dummy variable set arbitrarily at 0 or 1 depending on the absence or presence of a specific feature. In a multivariate regression analysis, significant covariates of Tac pharmacokinetics were selected using the backward elimination procedure. A two-sided P < 0.05 was considered to be statistically significant. www.impactjournals.com/oncotarget Abbreviations Tac: tacrolimus; LT: liver transplantation; CLTR: China Liver Transplant Registry; TBL: total bilirubin; Hb: hemoglobin; C/D: dose-corrected Tac concentration; D/W, weight-corrected Tac dose; SNPs: single nucleotide polymorphisms; OLT: orthotopic liver transplantation (OLT); ROC: receiver operating characteristic; AUC: area under the curve; BMI: body mass index; GPT: glutamate pyruvate transaminase; PCR: polymerase chain reaction.

Author contributions
Yuan Liu, Tao Zhang and Xiaoqing Zhang carried out the studies, participated in collecting data, performed the statistical analysis and drafted the manuscript. Zhihai Peng and Junwei Fan participated in its design. Ling Ye, Haitao Gu, Lin Zhong, Hongcheng Sun and Chenlong Song helped draft the manuscript. All authors read and approved the final manuscript.

ACKNOWLEDGMENTS
I want to take this chance to thanks to my tutor, Dr. Junwei Fan, a professor in Shanghai Jiao Tong University school of medicine. In the process of composing this paper, he gives me many academic and constructive advise.
At the same time, I do need to thanks my father, Chunsheng Liu, my mother, Meihong Yang, who give me a lot of mental and economic support making me grow up well.