The transplanted child: New immunosuppressive agents and the need for pharmacokinetic monitoring

Abstract

BACKGROUND:

Pharmacokinetic monitoring has been insufficiently studied in paediatric solid organ transplantation, especially because some agents are relatively new to paediatric use, are of new formulation modification or are being used in combinations not previously well studied. The choice of immunosuppressive drugs after paediatric renal transplantation is increasing. Cyclosporine A (CyA), tacrolimus and mycophenolate mofetil (MMF) use has become routine. While pharmacokinetic monitoring of CyA and tacrolimus is routine, few paediatric data on tacrolimus pharmacokinetics exist, and, for MMF, pharmacokinetic monitoring is performed in only a few Canadian centres. The aim of the present article is to provide guidelines for the use of these three drugs by using a large number of full pharmacokinetic profiles in children.

METHODS:

One hundred forty-nine full pharmacokinetic 10-point profiles on cyclosporine microemulsion, 103 on the classic cyclosporine, 118 on tacrolimus and 114 on MMF were retrospectively analyzed. All pharmacokinetic profiles were obtained from paediatric renal transplant patients in steady state.

RESULTS:

For pharmacokinetic monitoring of the classic cyclosporine formulation, evaluation of the trough levels suffices to estimate the area under the curve (AUC). For microemulsified cyclosporine, the trough levels do not provide a useful tool, and blood concentrations at 2 or 3 h (C2 or C3) after intake should be measured instead. Tacrolimus trough levels sufficiently estimate the AUC, but measuring the C4 yields the best prediction of the AUC. Nonetheless, C2 also provides a superior tool than the trough levels. Tacrolimus and CyA AUCs change substantially over time after renal transplantation. There is only a poor correlation between the trough level and the AUC for mycophenolic acid (MPA). No single time point provides a surrogate marker of the AUC. At least three time points are required to accurately estimate the AUC, and C1, C2 and C6 serve as the best markers. The article also describes the interaction between MPA and differing concomitant immunosuppression, as well as the variation of the MPA AUC with differing concentrations of cyclosporine.

CONCLUSIONS:

Pharmacokinetic monitoring of these three drugs is mandatory in paediatric renal transplantation because it is impossible to predict the drug interactions and blood levels from a given dose. Target AUCs for a given time point after transplantation remain to be established.

The choice of immunosuppressive medications after paediatric renal transplantation is increasing. Twenty-five years ago Azathioprine and steroids were considered to be standard. In the early 1980s, the availability of cyclosporine (1) fundamentally altered the treatment, and a cyclosporine-based immunosuppressive therapy with or without azathioprine and steroids became standard. In the early 1990s, tacrolimus (2) became available and is gradually replacing cyclosporine. Mycophenolate mofetil (MMF) is a new immunosuppressive drug that acts by impairment of de-novo purine synthesis (3) by inhibiting inosine monophosphate dehydrogenase (IMPDH) (4). It has been used in paediatric renal transplantation since the mid-90s and within a relatively short time it replaced azathioprine. In several Canadian centres, the use of MMF instead of azathioprine is considered to be standard (5).

Pharmacokinetic monitoring is needed whenever one deals with either a narrow therapeutic window (ie, when the range between toxicity and subtherapeutic levels is small) or when drug levels are unpredictable in patients because of inter- or intraindividual variation. Furthermore, pharmacokinetic monitoring is mandatory if drug interactions affect drug levels. While the area under the time concentration curve (AUC) most closely resembles drug exposure, usually predose trough levels (C0) are measured because often there is a good correlation between the AUC and the trough level. The need for pharmacokinetic monitoring of cyclosporine in children is well established (6). There was substantial interindividual variability. In the mid-90s, microemulsified cyclosporine became available. The unpredictable pharmacokinetics became much more predictable; however, the old tools of measuring trough levels are not sufficient to estimate the AUC, and at least the 2- and/or 3-h concentrations (C2 and C3) also have to be measured (7,8). Tacrolimus is a calcineurin inhibitor like cyclosporine, and it appears that rejection episodes are reduced compared with cyclosporine (9). Tacrolimus shares the unpredictability of the blood levels from a given dose with cyclosporine. The need for pharmacokinetic monitoring of tacrolimus in children is well established (10).

MMF has been shown to reduce the frequency of rejection in renal transplantation (11). MMF is a prodrug of mycophenolic acid (MPA), which can be measured by high performance liquid chromatography or enzyme-multiplied immunoassay technique (EMIT). Few data are available on the dosing of MMF in paediatric kidney transplantation, and data are restricted to MMF in combination with cyclosporine (12,13). Very few data exist about dosing in combination with tacrolimus or without concomitant calcineurin inhibitors in paediatric patients, although it has become clear that dosing may be influenced by concomitant medication and that there is substantial interindividual variation (14,15). Reduced dosing of MMF in combination with tacrolimus was also noted in adults (16). Therapeutic drug monitoring (TDM) of MMF is not generally accepted for the treatment of adult patients; however, there is increasing evidence that TDM may help to diminish both short and long term side effects of MMF (17,18). The data of a single centre are reviewed, with special attention given to abbreviated formulas and drug interactions.

PATIENTS AND METHODS

Study design

Retrospective chart review and analysis of existing pharmacokinetic profiles.

Patients

One hundred thirty-nine paediatric patients receiving either the classic cyclosporine formulation, cyclosporine microemulsion, tacrolimus or MMF were investigated. Twenty-four of these patients were treated at the Children’s Hospital of Eastern Ontario, Ottawa. The remaining 115 children had been transplanted at the Charité Children’s Hospital, Humboldt University, Berlin, Germany. The pharmacokinetic profiles were obtained after a mean of 4.1±3.3 years (median 4.1 years, range 0.03 to 11.9 years after renal transplantation). All pharmacokinetic profiles were obtained in the same institution as the transplantation, except for 12 patients who were followed in Ottawa who received their transplants in Toronto. Mean age at transplantation was 10.1±3.5 years, median 9.8, minimum 2.05 and maximum 17.2 years. Mean age at the time of the pharmacokinetic profiles was 13.8±4.0 years, with a median of 13.6 years (range 2.33 to 21.7 years). A total of 103 pharmacokinetic profiles on the classic cyclosporine (45 patients), 149 full pharmacokinetic 10-point profiles on cyclosporine microemulsion (63 patients), 118 pharmacokinetic profiles on tacrolimus (49 patients) and 114 pharmacokinetic profiles on MMF (61 patients) were analyzed. All pharmacokinetic profiles were obtained from paediatric renal transplant patients in steady state. Patients had a median of two full pharmacokinetic profiles (range one to five).

Several combinations of immunosuppressive drugs were given. The combinations in children enrolled in this study are listed in Table 1. Sixty-one patients received MMF. Of these patients, 19 did not receive a concomitant calcineurin inhibitor, 24 received additional cyclosporine and 18 received concomitant tacrolimus. Tacrolimus alone or in combination with azathioprine and/or steroids was given to 31 patients. Sixty-five patients received cyclosporine in combination with steroids and azathioprine. Mean age of these patients was 15.0±4.0 years at the time of first pharmacokinetic profile. Mean age at transplantation of the transplanted patients was 11.7±3.4 years. Except for two patients (one in the cyclosporine A group and one in the group without concomitant calcineurin inhibitor) with chronic persistent hepatitis C, but with normal liver enzymes, liver disease was absent and liver function tests were normal in all patients.

TABLE 1.

Immunosuppressive drug combinations used by patients

Cyclosporine			Tacrolimus		None
Classic	Microemulsion
± Aza	± Aza	± MMF	± Aza	± MMF	MMF
± steroids	± steroids	± steroids	± steroids	± steroids	steroids

± With or without. Aza Azathioprine; MMF Mycophenolate mofetil

All patients underwent TDM after the establishment of stable trough concentration. All patients had at least one full pharmacokinetic profile with the first profile after a median of 21 days. The investigations were performed between 1993 and 2001.

Pharmacokinetic monitoring

Cyclosporine whole blood concentrations were measured using an automated EMIT assay, and tacrolimus whole blood concentrations were measured by microparticle enzyme immunoassay using a modified, sensitive Abbott Tacrolimus I Assay (Abbott Laboraties, USA) (19), and after 1997 using the newly available Abbott Tacrolimus II Assay (Abbott Laboraties, USA). Mycophenolic acid (MPA) was measured by an automated EMIT assay (20).

Pharmacokinetic profiles were obtained after inserting an intravenous cannula and obtaining a baseline trough level at seven o’clock in the morning. The patients were then asked to take their usual morning doses of MMF afterwards, and immediately thereafter they had a standard meal with rolls, butter, jam and fruit tea. They had free access to nondairy product drinks during the day (except grapefruit and orange juice) and had a normal lunch at noon. Additionally to the C0, 2 mL ethylenediamine-tetra-acetic acid whole blood samples were taken for duplicate measurements of MPA concentration at 0.5, 1, 1.5, 2, 3, 4, 6, 8 and 12 h, respectively, for a 10-point, 12-h pharmacokinetic profile. No saline was injected into the cannula – it was sealed after each blood sampling with a sterile heparinized mandrin instead. AUC was calculated according to the trapezoid rule.

Statistics

All data are presented as mean ± SD for normally distributed data, median and range for data or percentages that are not normally distributed. Continuous data were tested for normal distribution using the Kolmogorow-Smirnov test. The student’s t-test was used for normally distributed continuous variables and the Mann-Whitney test was used for continuous variables that are not normally distributed. Standard correlation analysis and linear regression analysis also were performed. By using stepwise linear regression analysis, an abbreviated AUC was calculated from all 10 sampling points and any combination of up to three parameters. All statistical analyses were performed using GraphPad Software (GraphPad Software Inc, USA) for Science Version 2.01 (GraphPad Software Inc, USA), or SPSS (SPSS Inc, USA) for Windows, Version 8.0.0 (Microsoft Corporation, USA).

RESULTS

Classic cyclosporine formulation

A total of 102 pharmacokinetic profiles were obtained from 45 paediatric renal transplant patients. Only data on twice daily dosing were analyzed. The mean dose was 7.4±3.4 mg/kg/day or 227±88 mg/m² body surface area per day. The mean (±1 SD) AUC was 3810 (±1885) ng×h/mL. There was a reasonably good correlation between the pre-dose trough level and AUC, with a correlation coefficient of r²=0.6230, P<0.0001. Only for the 6-h concentration (C6) was there a better correlation coefficient of r²=0.6602, P<0.0001, which was not significantly better than the trough concentration. Hence, using trough levels to estimate AUC in patients on the classic formulation of cyclosporine appears to be reasonable.

Multiple stepwise regression analysis was used to investigate whether a combination of two or three time points would yield a better prediction of the AUC. No combination of two time points achieved a multiple regression correlation coefficient greater than 0.90. However, when using three sampling points at 0, 3 and 6 h, an excellent model for an abbreviated AUC could be found. The AUC can be estimated using the following formula:

$$\text{AUC}\hspace{0.17em}(\text{estimated})\hspace{0.17em}=\hspace{0.17em}(473+5.60)\times (\text{C}0+2.56)\times (\text{C}3+3.54)\times \text{C}6$$

When using this formula, the abbreviated AUC had a coefficient of determination of 0.9188, a multiple correlation coefficient of 0.9585 and a residual standard deviation of 545.

Results are shown in Table 2.

TABLE 2.

Parameters of multiple stepwise regression analysis for an abbreviated area under the curve on the classic cyclosporine formulation

Independent variables	Coefficient	Standard error	t	P
(Constant)	473.49006
C0	5.59522	0.75117	7.449	<0.0001
C3	2.55784	0.21086	12.131	<0.0001
C6	3.53544	0.26015	13.590	<0.0001

C0 Predose trough concentration; C3 Concentration at 3 h; C6 Concentration at 6 h

Cyclosporine microemulsion

A total of 149 pharmacokinetic profiles were obtained from 63 paediatric renal transplant patients. Only data on twice daily dosing were analyzed. The mean dose was 5.2±2.1 mg/kg/day or 161±55 mg/m² body surface area per day, significantly lower than the dose of the classic cyclosporine formulation. The mean AUC was 3823±1453 ng×h/mL, which was identical to the AUC obtained by the classic cyclosporine formulation. There was a reasonable significant correlation between AUC and dose per body surface area (r²= 0.6544, P<0.0001). There was only a poor correlation between predose trough levels and AUC with a correlation coefficient of r²=0.4053, P<0.0001. Figure 1 shows the relationship between the AUC and the trough level for both the classic formulation and the microemulsion.

Figure 1).

The relationship between cyclosporine area under the curve (AUC) and trough level for the classic formulation (top) and microemulsion (bottom). There was a significant correlation for both formulations; however, the correlation coefficient (r²) for the microemulsion was significantly worse for the new microemulsion. The data clearly show that cyclosporine microemulsion therapy in children cannot be monitored by the trough level alone

Significantly better correlation coefficients were found between AUC and C2 and C3. The correlation coefficient was best for C3 (r²=0.7920, P<0.0001) and similar for the C2 (r²=0.7803, P<0.0001). Figure 2 shows the relationship between the AUC and C2 and C3.

Figure 2).

The relationship between cyclosporine area under the curve (AUC) and 2-h concentration (C2) (top) and 3-h concentration (C3) (bottom). The advantage of C3 lies in the narrower range. The correlation coefficient was best for C3 (r²=0.79, P<0.0001), although the slightly lower correlation coefficient for the C2 (r²=0.7803, P<0.0001) was not significantly different

Tacrolimus

A total of 118 pharmacokinetic profiles were obtained from 49 paediatric renal transplant patients. As for this study, patients received both 1 or 5 mg capsules, and, thus, occasionally differing doses for morning and evening applications. More recently, both a liquid formulation and 0.5 mg capsules were introduced. The mean dose was 0.09±0.06 mg/kg/day or 2.7±1.7 mg/m² body surface area per day. The mean AUC was 128±49 ng×h/mL. There was no correlation between the dose per body surface area and the AUC (r²=0.098), absolutely necessitating pharmacokinetic monitoring in these patients. The mean morning trough level was 8.1±3.5 ng/mL. Only for tacrolimus was it noted that the mean morning trough level was significantly different from the evening trough level, which was 7.5±2.6 ng/mL. This was due to a different dosing for the morning and evening dose (3.5±2.2 mg in the morning and 4.0±2.2 mg in the evening, respectively). The reason for the differing doses lies in the lack of the availability of a liquid preparation and the fact that the smallest capsule size is 1.0 mg. Patients usually received a higher dose in the evening when unequal doses were given. There was a significant correlation between the C0 and AUC with a correlation coefficient of r²=0.68, P<0.0001. The best correlation was with 4-h concentration (C4). The C2 also yields a reasonably good correlation with AUC (r²=0.76, P<0.0001). Figure 3 shows the relationship between the AUC and the trough level, as well as for the C4.

Figure 3).

The relationship between tacrolimus area under the curve (AUC) and trough level (C_trough) (top) and 4-h concentration (C4) (bottom). There was a positive correlation between the AUC and the ²=0.68, P<0.0001). The best relationship between a given C_trough (r time point and the tacrolimus AUC existed at 4 h with a correlation coefficient of r²=0.83, P<0.0001

Changes with time after renal transplantation for cyclosporine and tacrolimus

The AUC changed with time after renal transplantation. Figure 4 summarizes the change with time for both cyclosporine AUC and tacrolimus AUC. These data do not permit the determination of target AUCs, but the data help to give an idea about the pharmacodynamic alterations with time.

Figure 4).

The relationship between time after renal transplantation and AUC. Also given is the non-linear regression line using a model of one-side binding (hyperbola). These graphs only serve as a retrospective analysis and do not necessarily provide guidelines. Target AUCs have not been established for these drugs in children

MPA

A total of 114 pharmacokinetic profiles were obtained from 61 patients with a median of two pharmacokinetic profiles per patient. Mean AUC was 63.19±24.42 μg×h/mL, and the AUC was normally distributed. Mean C0 was 3.8±2.4 μg/mL and mean postdose trough concentration (C12) was 3.4±2.2 μg/mL. There was only poor correlation between the AUC and the C0 (r²=0.29, P<0.0001). Correlation between AUC and C12 was r²=0.48, P<0.0001.

There was no difference with regard to AUC and concomitant immunosuppression. Mean AUC was 62.8±23.3 μg×h/mL for concomitant tacrolimus, 62.7±9.1 μg×h/mL for concomitant cyclosporine and 64.5±19.3 μg×h/mL in the absence of a concomitant cal-cineurin inhibitor. Therefore, all pharmacokinetic profiles were pooled. Only the dose-normalized AUCs differed with regard to concomitant immunosuppressive medication (0.122±0.038 μg×h×m²/mL×mg for concomitant tacrolimus, 0.067±0.037 for concomitant cyclosporine and 0.084±0.031 μg×h×m²/mL×mg in the absence of a concomitant calcineurin inhibitor).

Using linear regression analysis, the best correlations between AUC and concentrations at various time points were found at 2 h (C2, r²=0.59, P<0.0001), 3 h (C3, r²=0.52, P<0.0001), 1.5 h (C1.5, r²=0.50, P<0.0001) and 6 h (C6 r²=0.43, P<0.0001).

Previously, stepwise linear regression analysis was used and showed that the following formula can be used for estimating AUC (20):

$$\text{AUC}=(10.75+0.98)\times (\text{C}1+2.38)\times (\text{C}2+4.86)\times \text{C6}$$

By using this formula, there was excellent correlation between the full 10-point AUC and estimated three-point AUC (r²=0.87, Pearson r=0.93, 95% CI 0.89 to 0.95).

Drug interactions between cyclosporine and tacrolimus with MPA

Differing doses of MMF are required in combination therapy with tacrolimus compared with cyclosporine. The recommended dosage for MMF in combination with cyclosporine A for paediatric kidney transplant recipients is 600 mg/m² bid. Recently, pharmacokinetic profiles of MMF in combination with tacrolimus were published to keep the MPA C0 between 2 and 5 μg/mL and to avoid side effects, mean MMF doses were reduced to 300 mg/m² bid. To investigate whether this striking difference was due to alterations of MPA clearance by cyclosporine A or tacrolimus, pharmacokinetic profiles from 13 patients who received MMF without cyclosporine A or tacrolimus were analyzed, and these data were compared with those from 14 patients who received a combination of MMF and tacrolimus, and 15 patients who received MMF and cyclosporine A. Mean AUC in all pharmacokinetic profiles was 61.9±23.8 μg×h/mL. Although the AUCs did not differ between the groups, the dose per square meter was significantly lower in patients receiving concomitant tacrolimus than that with cyclosporine A, and the dose-normalized AUC was significantly higher in the tacrolimus-treated group. The MMF doses were 1158±301 mg/m²/day in the cyclosporine A group, 555±289 mg/m²/day in the tacrolimus group, and 866±401 mg/m²/day in the group without concomitant calcineurin inhibitor treatment.

Even when using MMF only in combination with cyclosporine A, differing dose-normalized AUCs result with differing cyclosporine blood concentrations. Full 10-point profiles for both MPA and cyclosporine A in 23 paediatric patients receiving MMF and cyclosporine microemulsion were retrospectively analyzed. Because the majority of patients were treated with low doses of cyclosporine after adding MMF, the AUC for cyclosporine showed a wide scatter, ranging from 296 to 6400 ng×h/mL. The mean cyclosporine dose was 100±76 mg/m²/day (range 28 to 331). There was no correlation between MPA AUC and MPA dose, and there was substantial interindividual variation. However, there was a significant negative correlation between dose-normalized MPA AUC and cyclosporine AUC (r²=0.23, P<0.0220). When dividing the MPA profiles into two groups (11 and 12 patients) with a cyclosporine A AUC of less than or greater than 1600 ng×h/mL, there was a significantly higher 8-h concentration in the patients with the lower cyclosporine A AUC, secondary to a higher second peak. The data demonstrate that the cyclosporine AUC is a determining factor for the MPA AUC and that MPA dose should be reduced when cyclosporine dose is reduced to achieve the same MPA AUC. The significantly higher peak in the group with a lower cyclosporine A profile supports the concept of a dose-dependent cyclosporine-induced inhibition of MPA glucuronidation.

DISCUSSION

TDM of cyclosporine and tacrolimus is well accepted. Relatively novel is the finding that the correlation between trough levels and AUCs for the new microemulsified cyclosporine formulation actually is significantly worse than that of the classic formulation (7). Recently, it was advocated in adults to replace trough level monitoring for cyclosporine microemulsion with C2 (22,23). Our data support the same concept. For tacrolimus pharmacokinetic monitoring, using the trough level probably suffices, but the C2 and, in particular, C4 concentrations yield significantly better correlations with the AUC. This has not previously been described.

TDM of MMF is not generally accepted for the treatment of adult patients; however, there is increasing evidence that TDM may help to diminish side effects of MMF (24) and long term over-immunosuppression (25). TDM is not yet fully established and does not reflect the immunosuppressive action on the key enzyme IMPDH (26). The measurement of IMPDH is not generally available. A 50% inhibition of IMPDH, proposed to be sufficient for immunosuppression, was found at an average AUC of 59 μg×h/mL and MPA trough concentrations between 2 and 5 μg/mL (26), and, therefore, most clinicians use these ranges as target ranges. However, they have not yet been defined, particularly not in a paediatric patient cohort. In this view, we have used measurements of MPA concentrations using the commercially available EMIT assay, although there are validated high performance liquid chromatography methods in paediatric populations and the EMIT concentrations are usually slightly higher (20). The debate about the usefulness of the trough concentration of MPA because of the second peak (usually at 6 h, sometimes later) due to enterohepatic recirculation via the main metabolite, MPA glucuronide (27), is ongoing. We also found a rather poor correlation between the AUC and the C0. Only the use of a limited sampling strategy using three samples provided a significantly better correlation between AUC calculated with all 10 sampling points. Using the limited sampling procedure proposed from C1, C2 and C6, the costs are 66% less than a full profile, and there is good agreement with the AUC calculated from the full pharmacokinetic profile (28). More importantly, there are humanistic benefits of the abbreviated AUC, including less loss of blood, lower required time to stay for blood sampling, and, of particular importance for children, less trauma. We recommend using trough concentrations for monitoring MMF treatment, and in case of suspected toxicity or under-immunosuppression, the sampling of C1, C2 and C6.

The need for lower dosing of MMF in combination with tacrolimus than that with cyclosporine has only recently been recognized (24,28). While a starting dose of 600 mg/m² twice daily is appropriate for cyclosporine, that dose has to be reduced to 250 to 300 mg/m² when using MMF in combination with tacrolimus (24). But, even when using differing concentrations with the same concomitant medication, the dose-normalized MPA AUC changes (21). An increase of the cyclosporine AUC requires an increase of the MPA dosing to achieve the same MPA AUC (21). A dose-dependent effect of cyclosporine on the metabolism and subsequently on the trough concentration of MPA has been shown both in vitro and in vivo (29). Zucker et al (29) described that cyclosporine inhibited the enzyme uri-dine diphosphate-glucuronosyltransferase (UDPGT) in a dose-dependent manner. They used cyclosporine concentrations of 200 and 1000 ng/mL, which are well in the expected cyclosporine blood level ranges in children and also were observed in some of the patients. The mode of inhibition was thought to be due to a direct competition of cyclosporine with the active site of the UDPGT, and it was concluded that cyclosporine serves as a legitimate target for glucuronidation (30). Other studies have shown increased MPA-glucoronide levels with increasing cyclosporine AUCs and decreasing MPA-G levels when cyclosporine is withdrawn (31). All these drug interactions are difficult to predict and absolutely necessitate the use of TDM for MMF therapy.

CONCLUSIONS

The present paper shows that the simple assumption of the trough level correlating with the AUC of the drug does not hold in this paediatric renal transplant patient cohort. A reasonable correlation with the AUC and the C0 exists only for the classic cyclosporine and tacrolimus. The trough level does not reveal a high or low maximum concentration of cyclosporine microemulsion, and the variable second peak of MPA on MMF explains the poor correlation for this drug. Revised strategies for abbreviated AUCs for the cyclosporine microemulsion and MMF have to be derived. Though not part of this study, similar considerations apply for other paediatric solid organ transplant recipients, which may often involve those of the youngest ages. The present manuscript did not specifically address the influence of very young age on the pharmacokinetics, but the faster hepatic clearance in infants is well described. Another important group that necessitates pharmacokinetic monitoring is that of adolescents. In these patients, both compliance issues and hormonal changes present additional challenges. Only with complete pharmacokinetic profiles can these changes be revealed. The authors propose to perform at least one complete pharmacokinetic profile after transplantation once stable levels and graft function have been accomplished, and additionally in case of toxicity or inefficacy. Despite the useful data derived from this retrospective analysis, the limitations of such work are well recognized and future prospective trials to establish paediatric target AUC ranges and pharmacodynamics are warranted.