Pediatric patients that underwent kidney transplantation with a deceased or living-related donor between 2001 and 2019 at the referral transplant center Hospital Luis Calvo Mackenna (Santiago, Chile) were invited to participate in this descriptive, prospective, longitudinal study. In our center, 85–90% of the transplantations were performed with deceased donors each year. We studied 104 recipients who received TAC and MPA as immunosuppressive therapy combined with steroids. Patients were induced with basiliximab on day 0 during surgery before reperfusion and on day 4 post-transplantation. TAC (starting dose 0.15 mg/kg twice per day), MPA (starting dose 800 mg/m2 twice per day), and steroids were administered according to the institutional protocol. Steroid use was individualized according to immunological risk and graft function in each patient. Starting steroid therapy was methylprednisolone 2 mg/kg/day for 2 days and switched to prednisone at day 3. The patients undergoing steroid withdrawal were treated successively with 2, 1, 0.5, and 0.25 mg/kg/day, each dose for one day. The patients that continued steroid therapy switched to prednisone 2 mg/kg/day at day 3 and reduced gradually until reaching 0.12 mg/kg/day at month 6, and this was kept constant from that month onwards. Therapeutic drug monitoring of TAC was applied regularly to all recipients and dose adjustment was performed when necessary to achieve target trough levels (10–15 ng/ml in the first 3 months and 5–7 ng/ml after 3 months post-transplantation). At the end of a 6-months follow-up of TAC therapy, associations of the CYP3A5 SNP with drug concentrations (TAC-C₀), daily dose normalized by weight (TAC-D mg/kg) and drug concentrations normalized by dose (TAC- C₀/D) were analyzed. MPA levels were regularly monitored and associations of the MRP2 and UGT1A9 SNPs with drug concentrations (MPA-C₀), daily dose normalized by body surface (MPA-D mg/m2) and drug concentrations normalized by dose (MPA-C₀/D) were analyzed at one time point, once stable immunosuppression had been achieved. The analysis of the area under the curve in 12 h normalized by dose (AUC_0–12h/D) was conducted on 98 and 44 patients with stable immunosuppression based on TAC or MPA therapy, respectively. Relative frequencies of ABO and Rh blood groups demonstrated that the study population was representative of the Chilean population (data not shown).

The study was conducted according to the declaration of Helsinki and was approved by the Ethics Committees of the Universidad de Chile and Hospital Luis Calvo Mackenna. Patients with evidence of graft dysfunction during the first year post-transplantation, with change of immunosuppressive medication to azathioprine or cyclosporine A within the first 6 months after transplantation, or parents that denied participation were excluded.

Nucleic acids were extracted from whole peripheral blood samples using QIAmp DNA Blood Mini Kit (QIAGEN) or MagNA Pure Compact Nucleic Acid Isolation Kit I (Roche) following the manufacturer’s instructions. CYP3A5, MRP2, and UGT1A9 were selected as genes with variants of interest for this study, because they have been proposed as the major factors to explain the variability related to plasma drug levels (TAC and MPA) and have shown similar results in different populations. The presence of the CYP3A5 c. A6986G polymorphism (rs776746) was determined using the PCR-RFLP technique with specific primers (forward 5′-CAT GAC TTA GTA GAC AGA TGA-3 ′and reverse 5′-GGT CCA AAC AGG GAA GAA ATA-3′) and the restriction enzyme SspI, according to previously described protocols (Rong et al., 2010). The assessment of the polymorphisms UGT1A9 2152C > T (rs17868320), UGT1A9 -275T > A (rs6714486) and MRP2 -24C > T (rs717620) was performed with the TaqMan™ Drug Metabolism Genotyping Assay (ThermoFisher). All SNPs were processed on LightCycler^® 480II and analyzed with the LightCycler^® 480II software (Roche). Each assay included controls confirmed previously by Sanger sequencing for the three genotypes (homozygous reference allele, heterozygous, homozygous variant allele) if available and negative controls with an equal volume of nuclease-free pure water.

To measure plasma drug concentration, four blood samples of 2 ml were taken in EDTA tubes, corresponding to C₀ (trough level before dose), C₁, C₂, and C₄ time points. Whole blood levels of TAC were measured using the Abbott Architect i1000 immunoassay (Abbott Laboratories). Plasma levels of MPA were measured using a liquid-liquid extraction and high-performance liquid chromatography equipment with a diode array detector (Agilent 1260, Agilent Technologies). The value of TAC AUC_0–12h was calculated in 98 patients using the abbreviated equation: AUC_0–12h = 10 + 1.4∗C₀ + 0.8∗C₁ + 1.6∗C₂ + 5.5∗C₄) (Wong et al., 2000). The value of MPA AUC_0–12h was calculated in 44 patients using the abbreviated equation: AUC_0–12h = 8.217 + 3.163∗C₀ + 0.994∗C₁ + 1.334∗C₂ + 4.183∗C₄ (Filler, 2004). The AUC_0–12h values were normalized by dose (TAC or MPA) for the following reasons: 1) AUC_0–12h was performed in each patient according to the clinical need to determine empirically drug exposure and was required at different time points, before and after 3 months of transplantation, that differ in AUC_0–12h and C₀ therapeutic targets, requiring higher doses in the first 3 months period; 2) the study cohort is a pediatric population with different etiologies/co-morbidities and their ages ranged from 1.5 to 15.3 years at the time of transplantation, causing differences in drug metabolism that confer, in consequence, variability in dose requirements to reach therapeutic targets.

Pearson chi-square (χ²) goodness-of-fit test for the Hardy Weinberg equilibrium was applied to assess deviation of allele and genotype frequencies. Normality and homoscedasticity assumptions were checked for all variables or transformed with log to achieve normal distribution. AUC_0–12h/TAC-D was analyzed by a one-way ANOVA to compare CYP3A5 genotypes. The analysis of AUC_0–12h without TAC dose normalization to compare CYP3A5 genotypes are shown as supplemental material (Supplementary Figures S1, S2). TAC blood concentrations (TAC-C₀), TAC daily dose requirement (TAC-D), TAC blood concentrations normalized by daily dose requirement (TAC- C₀/D), and creatinine levels throughout the 6-months follow-up were analyzed by a repeated-measures ANOVA with genotype and time as main effects using the ez package (Lawrence, 2013). Post hoc tests were applied in the case of significant differences between groups by Tukey HSD test. MPA blood concentrations (MPA-C₀), MPA dose requirement (MPA-D), and MPA blood concentrations normalized by dose requirement (MPA- C₀/D) were analyzed by Wilcoxon-Mann Whitney test, grouping the heterozygous and homozygous carriers of the variant allele. The AUC_0–12h/MPA-D values between carriers and non-carriers of the variant allele were compared with an unpaired one-tailed Wilcoxon-Mann Whitney test. The analysis of AUC_0–12h without MPA dose normalization to compare UGT1A9 -275 variant carriers are shown as supplemental material (Supplementary Figure S3). Outlier values (mean ± 3*SD) suggestive of errors in sampling procedure, technical measurements or data manipulation were excluded from the analysis. Box plots show median and interquartile range (IQR) and whiskers represent the 1.5*IQR. All statistical analyses were performed with R (Wilson and Norden, 2015) and plots were made using ggpubr (Kassambara, 2020a) and rstatix (Kassambara, 2020b) packages for R. P-values < 0.05 were considered to be statistically significant.

Altogether 104 pediatric kidney recipients from a single referral transplant center were included in this study. Clinical characteristics and CYP3A5 genotype data are shown in Table 1. The median age at the time of transplantation was 8.1 years, and the leading cause of the end stage renal disease to require transplantation was a structural anomaly.

The SNP rs776746 in CYP3A5 was analyzed in the entire cohort and we found that 8.7% presented the CYP3A5*1/*1 genotype, 28.8% the heterozygous CYP3A5*1/*3 genotype and 62.5% the CYP3A5*3/*3 genotype (Table 1). The frequency of the CYP3A5*3 variant allele resulted in 76.9%. Genotype and allele frequencies were not significantly different than expected if the population was in Hardy-Weinberg equilibrium (χ² = 3.66, p = 0.06). Compared to the Latino/Admixed American population reported in the gnomAd database v2.1.1 in December 2020 (https://gnomad.broadinstitute.org/), no statistical differences were found in our population in terms of genotype (χ² = 3.09, p = 0.21) or allele (χ² = 0.49, p = 0.48) frequencies of this SNP.

The analysis of AUC_0–12h/TAC-D in 97 patients revealed significant differences (p < 0.01) among the CYP3A5 genotypes (Figure 1). The CYP3A5*3/*3 patients showed the highest values (1058 [IQR 688.6–1460] ng*hr/ml/mg/kg), which was 1.56-fold higher than CYP3A5*1/*3 patients (676.9 [IQR 546.0–883.6] ng*hr/ml/mg/kg; p < 0.01) and 1.91-fold higher than CYP3A5*1/*1 patients (554.7 [IQR 239.7–1117]ng/h*ml/mg/kg; p < 0.05), respectively. AUC_0–12h/TAC-D values did not differ between CYP3A5*1/*3 and CYP3A5*1/*1 patients (p = 0.79).

Regarding the steroid therapy, the analysis showed some differences between CYP3A5 genotypes (Figure 2). When analyzing patients with early steroid withdrawal, the carriers of the CYP3A5*3/*3 genotype showed AUC_0–12h/TAC-D values (1099 [IQR 930.7–1537] ng*hr/ml/mg/kg), 1.58-fold significantly higher than the carriers of the CYP3A5*1 ancestral allele (CYP3A5*1/*3 n = 7 and CYP*1/*1; n = 2) (697 [IQR 404.5–1108] ng*hr/ml/mg/kg) (p < 0.05). No statistical differences were observed between CYP3A5 genotypes that continued steroid therapy (p = 0.34).

In the analysis performed in 52 patients with 6 months of follow-up, the TAC blood concentrations (TAC-C₀), TAC daily dose requirement (TAC-D) and TAC blood concentrations normalized by dose requirement (TAC- C₀/D) differed significantly between the CYP3A5 genotypes (TAC-C₀ p < 0.001, TAC-D p < 0.01, TAC- C₀/D p < 0.001) (Figures 3A–C). Specifically, patients with the CYP3A5*3/*3 genotype showed higher TAC-C₀ values than patients with the CYP3A5*1/*3 and CYP3A5*1/*1 genotypes (p < 0.01 and p < 0.001, respectively). Additionally, patients with the CYP3A5*1/*3 genotype also exhibited higher TAC-C₀ values than patients with the CYP3A5*1/*1 genotype (p < 0.001). Monitoring within the first three months, showed that 5–15% of the patients carrying the CYP3A5*1 allele had achieved target levels, lower than the 22–43% of the patients with the CYP3A5*3/*3 genotype. At the sixth month, 44.6% of the patients had achieved TAC-C₀ in the range between 5 and 7 ng/mL, regardless of their CYP3A5 genotype (Figure 3A). According to the analysis of TAC-D, patients with CYP3A5*1/*1 genotype showed higher TAC-D values than patients with the CYP3A5*1/*3 genotype and patients with the CYP3A5*3/*3 genotype (p < 0.01 and p < 0.001, respectively). Additionally, patients with the CYP3A5*1/*3 genotype also displayed higher TAC-D values than patients with the CYP3A5*3/*3 genotype (p < 0.001). Interestingly, significant differences between genotypes were not detected until the third month of follow-up, which is corroborated by an interaction between genotype and time (p < 0.01) (Figure 3B). Finally, patients with the CYP3A5*3/*3 genotype showed higher TAC-C₀/D values than patients with the CYP3A5*1/*3 and CYP3A5*1/*1 genotypes (p < 0.001 in both cases). Similar to the observations in the TAC- C₀ analysis, patients with the CYP3A5*1/*3 genotype also exhibited higher TAC-C₀/D values than patients with the CYP3A5*1/*1 genotype (p < 0.001) (Figure 3C). No significant temporal variation (TAC-C₀ p = 0.32; TAC-D p = 0.43; and TAC- C₀/D p = 0.44) or genotype-time interaction were observed (TAC-C₀ p = 0.18; and TAC- C₀/D p = 0.56), except for the genotype-time interaction for TAC-D (see above).

The change in dose requirements comparing the initial prescribed TAC dose with the required TAC dose at the end of the 6-months follow-up was different between CYP3A5 genotypes, experiencing a 209% increase in carriers of the CYP3A5*1 ancestral allele (CYP3A5*1/*3 and CYP*1/*1 patients; 0.314 ± 0.151 mg/kg, p < 0.01). By contrast, the TAC dose requirement of CYP3A5*3/*3 patients remained unaltered at the end of the follow-up in comparison to the initial prescribed TAC dose for these patients (0.164 ± 0.073 mg/kg, p = 0.99). Creatinine clearance was stable during the 6 months, regardless of the genotype (p = 0.76) (data not shown).

When analyzing the SNPs in MRP2 and UGT1A9 that might influence MPA pharmacokinetics in a subgroup of 53 patients, genotype and allele frequencies were not significantly different than expected if the population was in Hardy-Weinberg equilibrium (MRP2 -24C > T χ² = 1.54, p = 0.21; UGT1A9 -275T > A χ² = 0.26, p = 0.61; UGT1A9 -2152C > T χ² = 0.046, p = 0.83), although a low frequency of some variant alleles was observed (Table 2). In particular, the frequencies of the MRP2 -24A, UGT1A9 -275A and UGT1A9 -2152T alleles were 22.1, 6.6, and 2.9%, respectively. No significant differences between genotypes were observed in MPA-C₀, MPA-D or MPA-C₀/D (MRP2 -24C > T: MPA-C₀ p = 0.93, MPA-D p = 0.66, MPA-C₀/D p = 0.86; UGT1A9 -275T > A: MPA-C₀ p = 0.27, MPA-D p = 0.50, MPA- C₀/D p = 0.54; UGT1A9 2152C > T: MPA-C₀ p = 0.30, MPA-D p = 0.99, MPA- C₀/D p = 0.36).

Since the UGT1A9 and MRP2 alleles have been reported as variants associated with higher enzyme and transporter activities and higher MPA dose requirements, AUC_0–12h/MPA-D was compared between carriers and non-carriers of the variant alleles. Interestingly, patients carrying the UGT1A9-275A variant allele had lower AUC_0–12h/MPA-D (0.053 [IQR 0.040–0.100] ug*hr/ml/mg/m2) than patients carrying only the UGT1A9-275T ancestral allele genotype (0.117 [IQR 0.058–0.150]ug*hr/ml/mg/m2) with a difference of marginal significance (p = 0.05) (Figure 4). This difference was not observed in the other UGT1A9 (p = 0.33) or MRP2 (p = 0.29) alleles explored in this study.