Monitoring mycophenolate in liver transplant recipients: Toward a therapeutic range



Predose plasma mycophenolic acid (MPA) concentrations measured with a semi-automated enzyme-multiplied immunoassay were related to adverse events (e.g., rejection, leukopenia, infection), drug dose, and clinical status in 147 adult and 63 pediatric liver allograft recipients receiving adjunctive immunosuppression with mycophenolate mofetil (MMF). In 12 of 13 acute rejection episodes, predose MPA levels were below the 1 mg/L cut-off defined using receiver operating characteristic (ROC) curve analysis. The relative risk of developing infection or leukopenia increased more than 3-fold above predose MPA levels of 3 to 4 mg/L. Plasma MPA levels correlated weakly (r2 = 0.081) with MMF dose and the dose / level relationship was variably influenced by age, the indication for MMF, concentrations of serum albumin and creatinine, and comedication with tacrolimus or cyclosporine. The median mycophenolate dose required per unit mycophenolate level was 50% lower in children than in adults. Comparable drug requirements were also decreased by renal dysfunction (by 40 and 43% in adults and children, respectively), and in patients prescribed MMF alone rather than with tacrolimus or cyclosporine. However, in patients with serum albumin less than 35g/L, MMF dose requirements were higher than in those with normal albumin levels (by 2.1- and 2.6-fold in adults and children, respectively). In adults, 44.7% achieved clinically acceptable therapeutic MPA concentrations at a dose less than 1 g MMF twice daily and only 6.3% required 1.5 g twice daily as suggested by the manufacturer. The immunoassay was a rapid, reliable, and acceptably precise technique in which only 10.8% of measurements were unproductive. In conclusion, our data suggests that MPA predose level monitoring is both clinically- and cost-effective and that a therapeutic range of 1 to 3.5mg/L (by immunoassay) is applicable in liver allograft recipients given adjunctive MMF. (Liver Transpl 2004;10:492–502.)

Mycophenolic acid (MPA) is the active immunosuppressant derived from its 2,4-morpholino ester, CellCept, (mycophenolic acid mofetil: MMF1) and its sodium salt, Myfortic. It is a potent suppressor of purine (guanine) synthesis, principally via uncompetitive inhibition of inosine monophosphate dehydrogenase II (EC This mechanism of action is distinct from that of the primary immunosuppressants tacrolimus and cyclosporine, calcineurin inhibitors (CNI) that reduce lymphocyte clonal expansion induced by transcription factors whose migration to the nucleus requires calcineurin-dependent dephosphorylation. Associated with these different mechanisms are dissimilar profiles of side effects for MPA versus the CNI agents. Consequently, adjunctive immunosuppression with MMF is increasingly sought either for the treatment of acute or chronic rejection episodes or for reducing doses of the CNI agents (“CNI-sparing”) and minimizing their side effects, particularly nephrotoxicity.3

An early milestone clinical trial with MMF in renal transplant recipients showed an association of its efficacy with drug exposure (measured as area under the plasma concentration versus time curve) and of a corresponding increase of side effects with drug exposure.4 More recently, several laboratories have reported interindividual differences in the pharmacokinetic parameters for MPA of approximately 10-fold in adult renal,5 pediatric renal,6 adult liver,7 pediatric liver,8 and cardiac transplant recipients.9 This variability exists together with a narrow therapeutic window for MPA10 and the established value of monitoring in confirming compliance and differentiating immunosuppression-related side effects from recipient pathology. Consequently, assertions that categorical dosing was appropriate (e.g., in the CellCept package leaflet) and MPA monitoring was unnecessary (cited in Holt10) are in conflict with conventional recommendations for therapeutic drug monitoring of immunosuppressants.11 Given the considerable resource implications of monitoring immunosuppressive agents, we have evaluated the clinical benefit of immunosuppressive drug monitoring of MPA in liver graft recipients over the first 25 months of monitoring from February 2000. The aims were: 1) to establish a target range of plasma levels which minimized adverse therapeutic events, 2) to define factors influencing MMF dose / MPA plasma level relationships in adult and pediatric liver graft recipients, 3) to document the characteristics of the routine assay method employed, and 4) to demonstrate whether there was any cost effective justification for routine MPA monitoring.


MPA, mycophenolic acid; MMF, mycophenolate mofetil; ROC, receiver operator characteristic; CNI, Calcineurin Inhibitor; EMIT, Enzyme multiplied immunoassay technique.

Patients and Methods

From February 1, 2000 to February 28, 2002, predose blood samples were monitored for plasma mycophenolic acid according to a preliminary protocol for 3 assays per week with an enzyme multiplied immunoassay technique (EMIT).10 A reference range of predose MPA levels of 1 to 5 mg/L was proposed based on contemporary published data.12–14 Of the 4816 total samples, 2178 were from adult and 774 from pediatric patients, 786 were taken as additional samples in pharmacokinetic or investigative studies and 1078 (22%) were quality control, calibrator, or quality assessment scheme samples. This study considers only the control data and results on those samples from discrete cohorts of liver graft recipients where sequential samples were available.



The 147 adults (78 males) received liver allografts at a median age of 50.1 (16.9–71.8) years for biliary cirrhosis (37 patients), alcoholic liver disease (22 patients including 7 with a secondary diagnosis), hepatitis C viral cirrhosis (20 patients including 6 with a secondary diagnosis), autoimmune hepatitis (13 patients), cryptogenic cirrhosis (13 patients), acute liver failure (9 patients), non-A to E acute viral hepatitis (6 patients), hepatitis B viral cirrhosis (5 patients), Budd Chiari syndrome (4 patients), Wilson's disease (3 patients), α-1 antitrypsin deficiency (2 patients), cystic fibrosis (2 patients), hemochromatosis (2 patients), oxaluria (2 patients), Crigler Najaar syndrome (1 patient), and noncirrhotic portal hypertension (1 patient). Fourteen were retransplanted before or during the study (5 for chronic rejection, 4 for recurrent disease, 2 for primary graft nonfunction, and 3 for other causes). MMF was introduced at a starting dose of 500 mg twice a day at 2.9 (2.3–6.9) months before transplantation for autoimmune hepatitis in 4 patients and 23.6 (0–158.5) months afterward in the remainder: 10% first received MMF within 2 weeks of transplantation and 27% within 3 months. MMF doses were increased over time as required, and with increasing reference to MPA levels. Additional immunosuppression consisted of glucocorticoids (in 122), tacrolimus (in 99), and cyclosporine (in 40). Thirteen had received azathioprine before commencing MMF. Data from 5 patients admitted to an ongoing trial of MPA monitoring techniques were excluded from the analysis of MPA levels because their MMF dose adjustment was not based on monitoring results.


Sixty-three children (32 females) received liver grafts at a median age of 3.5 years (range: 0.3–19.5 years) for extrahepatic biliary atresia (25 patients), α-1 antitrypsin deficiency (PiZZ) (6 patients), Alagille syndrome (4 patients), metabolic disorders (4 patients), cryptogenic cirrhosis (3 patients), hepatoblastoma (3 patients), intra-hepatic cholestatic disorders (3 patients), non-A to E acute hepatitis (3 patients), Wilson's disease (3 patients), cystic fibrosis (2 patients), and 1 patient each of acute viral hepatitis B, autoimmune hepatitis, congenital hepatic fibrosis, conjugated hyperbilirubinemia, Crigler Najjar Type 1 disease, neonatal hepatitis, and liver trauma. Fifteen were retransplanted before or during the study (5 for hepatic artery thrombosis, 5 for chronic rejection, 2 for primary graft non-function, 1 for de novo autoimmune hepatitis, and 2 for other causes).

MMF was introduced 2.4 (0–129.9) months posttransplant, with 19% receiving the drug in the first 2 weeks and 54% within 3 months postoperatively. Dosing commenced at 5 mg/kg body weight twice a day and was increased over time to 10 and 20 mg/kg twice a day if tolerated and required. Additional immunosuppression consisted of glucocorticoids (in 60 patients), tacrolimus (in 40 patients), and cyclosporine (in 22 patients). Nineteen had received azathioprine before commencing MMF. MPA levels were used increasingly for making dosage adjustments as the study progressed.

Tacrolimus, cyclosporine, and mycophenolate monitoring was performed throughout the study as required clinically— usually at the time of outpatient clinic appointments or during inpatient hospital stays. An initial MPA sample was taken at least 3 days after starting MMF to allow attainment of dosage equilibrium. Blood samples for immunosuppressive drug monitoring were taken predose (in EDTA anticoagulant) and submitted with dosage information. Variability in outpatient clinic appointment times meant that such predose blood samples were drawn within the wide range of 8 to 21 h postdose. Since this reflected normal clinical practice, only results on samples taken outside these limits were excluded. Episodes of adverse responses seemingly associated with mycophenolate were retrieved from summary notes and details of rejection episodes were confirmed from histopathology reports. Leukopenia was defined as white cell counts less than 4 × 109/L; neurological complications included persistent headaches, migraine, depression and neuralgia; gastrointestinal events included diarrhea, nausea, vomiting, ulceration, and bleeding; infections included those clinically significant episodes involving viruses, bacteria, and fungi and requiring treatment. Low serum albumin and high serum creatinine levels were defined relative to the normal range (35–50 g/L and 45–120 μmol/L, respectively).


Where possible, plasma for mycophenolate measurements was isolated by centrifugation within 6 h of phlebotomy to minimize hydrolysis of mycophenolate glucuronide. Otherwise, blood was stored at 4 °C before centrifugation. Mycophenolate concentrations were measured by EMIT (Dade-Behring, Marburg, Germany) on a Cobas Mira analyzer (Roche, Basel, Switzerland). MPA quality control samples were obtained from Dade-Behring and Chromsystems GmbH (Munich, Germany). The minimum detection limit of the assay was 0.2 mg/L and a cut-off of 0.3 mg/L MPA was set in reporting results.

Data Processing and Statistics

Plasma MPA concentrations in samples taken outside the 8 to 21 h postdose interval mentioned above were excluded from the analysis. MPA levels were equalized to a standard MPA dose by calculating the plasma concentration per 1g MMF dose administered twice a day. MMF doses (in g) were equalized by calculating that dose required to achieve a 1 mg/L plasma MPA concentration. This dose per unit plasma MPA concentration is also termed apparent relative clearance since it equates to the volume of plasma cleared of drug per dose interval.

Analysis of the 2501 eligible MPA test results was based on selected cohorts. Side effects were analyzed in relation to MMF doses and MPA levels only in the samples from 147 adult patients because of insufficient numbers of adverse events in the 63 children. The effects of comedication with tacrolimus or cyclosporine on MMF dose / MPA level were compared only in 1421 samples from adult patients where MPA levels were available for making dose adjustments from the time MMF was commenced. This excluded data from 47 patients already started on MMF in whom CNI withdrawal was already ongoing and the extent of interactions with comedication was incomplete. Comparisons of adult and pediatric data on MMF dose / MPA level interrelationships were made in the entire collection of 1927 adult and 574 pediatric samples. Similarly, the influence on dose/level interactions of indication for MMF therapy (CNI sparing vs. other causes) used the complete populations.

Data are presented as median (range) unless otherwise stated. Statistical analysis was performed with SPSS version 10 software using the tests described and with significance accorded when P < .05. Grouped data were compared by the Mann Whitney U test. To establish the extent to which plasma MPA concentrations were associated with adverse events, ROC curve analysis was performed using incremental MPA levels to calculate sensitivity and specificity in patients with no recorded adverse events and the specific event of interest (e.g., rejection). Relative risk was calculated from the proportions of risk (i.e., incidence) in respective adverse event and control cohorts.



Of the 147 adult recipients, 100 were prescribed MMF for CNI sparing, 36 for supplemental immunosuppression in acute (22 patients) and chronic rejection (14 patients), and 11 patients for disease recurrence. Of the 63 pediatric transplant recipients, 20 patients were prescribed MMF for CNI sparing, 37 patients for supplemental immunosuppression in acute (20 patients), chronic (2 patients) and ongoing rejection/de novo autoimmune hepatitis (15 patients), and 6 patients for other causes.

In the 95 adults prescribed MMF for CNI sparing and monitored throughout treatment, mean cyclosporine doses and levels and tacrolimus doses fell significantly (by 25 mg and 21 μg/L and by 0.5 mg, respectively, P < .001 in all cases) and there was also a slight reduction in mean prednisolone dose (P = .023). The 10 adults treated for rejection were significantly younger (median 40.2 vs. 54.7 y) and commenced MMF significantly earlier (median 4.4 vs. 32 mo) after transplantation than those given MMF for CNI sparing (P < .05 for each by post-hoc Mann Whitney U test after ANOVA). Eight were coprescribed tacrolimus and 2 cyclosporine, and the median doses of tacrolimus (3 vs. 2 mg twice a day) and glucocorticoids (8.8 vs. 5 mg twice a day) were higher than in the CNI-sparing group as were median trough tacrolimus levels (10.2 vs. 6 μg/L) (P < .012 in every case).


A total of 106 adverse events possibly associated with MMF therapy were well documented in the 147 adult patients, together with corresponding details of immunosuppressive therapy (including adjunctive immunosuppression). Comparative information was recorded on 279 occasions either from these same patients at times when there was no evidence of adverse events or in patients who were free from recorded adverse events at all times. Comparable adverse events in the pediatric population were too few to be analyzed meaningfully. In adults, the adverse events consisted of 10 episodes of acute rejection (8 confirmed on biopsy), 12 episodes of upper and 30 of lower gastrointestinal dysfunction, 9 cases of leukopenia and 1 of neutropenia, 7 neurological episodes, 18 infectious episodes (3 viral, 13 bacterial, and 2 fungal) and 18 other problems (including 5 with malignancy, 3 with pruritis, 3 with lethargy, and 2 with thrombocytopenia). Those adverse events affecting the gastrointestinal tract occurred more frequently early after beginning MMF, but others occurred at all stages (Table 1).

Table 1. Mycophenolate Dose and Levels in Relation to Adverse Events in Adult Liver Transplant Recipients
Adverse EventnPost MMF (days), median (range)MMF Dose (mg), median (range)MPA Level (mg/L), median (range)
  • Abbreviations: GI, gastrointestinal; AE, adverse events.

  • *

    P < .05 (for comparison with the AE-free group).

  • P < .01(for comparison with the AE-free group).

  • P < .10(for comparison with the AE-free group).

  • §

    Other adverse events include 5 cases of malignancy, 4 of pruritis, 3 of liver disease recurrence, 3 of fatigue/lethargy, 2 of thrombocytopaenia, and 1 each of hair loss and multiple nonspecific problems.

None279175 (4–1304)750 (250–2000)1.4 (<0.3–7.4)
Acute rejection10368 (42–834)750 (250–1000)0.5 (<0.3–1.8)*
All gastrointestinal4247 (5–560)750 (250–1500)1.6 (<0.3–9.5)
 Upper GI1272 (6–401)500 (500–1000)1.8 (0.4–9.5)
 Lower GI3080 (5–560)*750 (250–1500)1.5 (<0.3–9.5)
Low total white cell counts10122 (40–575)1000 (750–1000)2.8 (0.9–7.8)
Leucopenia997 (40–348)1000 (750–1000)2.8 (0.9–7.8)
Neurological7218 (7–557)750 (500–1500)1.9 (0.6–3.3)
All infections18173 (12–768)500 (250–1000)1.8 (0.3–6.0)
Bacterial infections13152 (12–695)500 (500–1000)1.6 (0.3–4.4)
Other AE§19198 (28–1281)875 (500–1500)2.0 (0.8–5.2)*

The 10 episodes of acute rejection all occurred late after transplantation (i.e., > 2.5 mo, median 43 months) and when plasma MPA concentrations were significantly lower than in patients without any adverse events (median 0.5 vs. 1.4 mg/L, P < .001 by Mann Whitney U test) (Table 1). Nine of the 10 episodes were associated with plasma MPA concentrations less than 1 mg/L, with the exception occurring at 1.8mg/L in a patient whose serum albumin was 31g/L and creatinine 236 μmol/L. MMF doses in the patients with rejection were not different from those in the control cohort. Episodes of leukopenia were associated both with higher median plasma MPA levels (2.8 vs. 1.4 mg/L, P = .004) and higher MMF doses than in the control cohort (1000 vs. 750 mg twice a day, P = .004. Median MMF doses were lower (500 vs. 750 mg/L) but MPA levels were higher (1.8 vs. 1.4 mg/L) during episodes of bacterial, fungal and viral infections, although this trend failed to achieve significance (P = .081 for doses and .056 for levels). There were no differences in either median MPA levels or MMF doses associated with neurological episodes or gastrointestinal side effects compared with the control cohort (Table 1). However, median plasma levels were also higher (2.0 vs. 1.4 mg/L, P = .036) during the other miscellaneous adverse event episodes described in Table 1.

As a guide for the use of MPA levels, we have presented the cumulative proportion of patients with each of the adverse events of rejection, leukopenia, infection, and gastrointestinal disturbances against MPA levels (Fig. 1A). This indicates optimal efficacy and fewest complications in our patient population at a predose MPA level around 1mg/L. In addition, we have performed ROC curve analysis to define optimal cut-offs for MPA concentrations and specific adverse events, and calculations of relative risk at incremental plasma MPA levels (i.e., the respective proportions of patients in adverse event cohort of interest versus control patients). The relative risk of rejection (95% confidence intervals) increased 4.2-fold (2.34–7.49), 2.5-fold (1.92–3.22), and 1.6-fold (1.28–2.03) at plasma MPA levels less than 0.5, 1.0, and 1.5 mg/L (P = .003, .002, and .058, respectively) (Fig. 1B). Emphasizing this point are the additional observations that first, all 3 episodes of acute rejection noted in pediatric liver recipients treated with MMF occurred at plasma MPA levels less than 0.5 mg/L; secondly, that there was a significant association of rejection with MPA levels in ROC curve analysis (Fig. 2) where area under the ROC curve was 0.840 (0.725–0.954) (95% confidence intervals); and finally, that the cut-off so defined in the adults was 0.85 mg/L. Corresponding increases more than 3-fold in the relative risks for leukopenia, infection, and gastrointestinal disturbances were all noted at plasma MPA levels of 3 to 4 mg/L (Fig. 1B). The corresponding specific cut-offs defined from the ROC curve analysis were 2.85 mg/L in infectious episodes (area under the plasma concentration versus time curve: 0.634 (0.499–0.770), P = .056) and 2.25 mg/L in leukopenia (area under the plasma concentration versus time curve: 0.780 (0.642–0.919), P = .003) (Fig. 2). An additional association with MMF dose prevailed in episodes of leukopenia (area under the plasma concentration versus time curve: 0.750 (0.662 – 0.837), P = .007). The relative risk of gastrointestinal adverse events also rose as MPA levels increased (Fig. 1B) but there was no significant association with either MMF dose or MPA level on ROC curve analysis (P > .5) The remaining specified or miscellaneous adverse events listed in Table 1 were also not significantly associated with either MMF dose or MPA level.

Figure 1.

Relationship of plasma mycophenolic acid concentrations to (A) the incidence of major adverse events of therapy and (B) the relative risk of experiencing adverse events. The dashed line in B represents a relative risk of 1 (i.e., no change in risk).

Figure 2.

Receiver operating curve analysis of plasma mycophenolic acid concentrations in relation to episodes of acute rejection, leucopenia, and infection.

Defining a Therapeutic Range of MPA Levels

Based on the definition of a reference range describing 95% of the control population (in a non-normally distributed cohort), we determined the predose plasma MPA concentrations within 95 centiles of the median in those patients with no recognized adverse events. These limits defined a suggested therapeutic range for predose plasma MPA levels of 0.3 to 5.2 mg/L. While highly sensitive for defining patients without side effects, this range fails to consider the increased proportion of adverse events described above at plasma MPA concentrations exceeding 3 to 4 mg/L and therefore lacks specificity.

MMF Doses and MPA Test Results

There was no close correlation between MMF dose and MPA levels in either the complete adult or pediatric datasets. For example, in adults, the 2 variables were related by: MPA level = (0.003 (± 0.001 SE) × MMF dose) + 1.342 (± 0.09), with r2 = 0.01 by linear regression analysis. The Spearman correlation coefficient in the entire 2501 samples was 0.241 (P < .001).

At the end of the study, when the influence of monitoring on dose was greatest, 45.7% of adult patients received 1000 mg MMF twice a day, with 44.7% prescribed lower doses (10.6% at 750, 29.8% at 500, and 4.3% at 250 mg twice a day) and only 9.6% receiving higher doses. To achieve therapeutic levels, only 6.3% of adults required doses of 1.5g twice a day as recommended in the prescribing literature.

Interpatient variability in MPA levels (expressed as the mean coefficient of variation) was lower in adults (50.5%) than in children (62.5%) (P = .027). Median MPA levels equalized for dose did not change significantly in either adults or children over time, but a significant differential between adults and children was maintained: median 1.8 versus 3.2 mg/L per g MMF dose at the start of monitoring (P = .003) and 2.0 versus 3.0 mg/L per g MMF at the end (P = .009). This was consistent with a 77% higher apparent relative clearance of MPA at the start of therapy (i.e., MMF dose per unit MPA predose level) in the adult patients (556 vs. 313 L per dose interval in children, P = .003) that fell to 50% at the end of monitoring (500 vs. 333 L, P = .010).


In the cohort of 95 adult patients monitored from the start of MMF therapy, mean MMF doses were higher in tacrolimus than in cyclosporine comedicated patients (889 vs. 804 mg twice a day, P = .003) and both mean doses were higher than in those patients where CNIs were withdrawn (P < .001 for tacrolimus vs. MMF and P = .001 for cyclosporine, Table 2). In contrast, median MPA levels were lower with tacrolimus (1.1 mg/L) than with either cyclosporine (1.4 mg/L, P < .001) or no CNI comedication (1.7 mg/L, P < .001) (Table 2). Further analysis of dose-equalized MPA levels showed that values during tacrolimus comedication approached 57% of those achieved with comparable doses of MMF during cyclosporine comedication and 53% of those during treatment with MMF alone (Table 2). Corresponding doses required per unit MPA plasma concentration were higher in the tacrolimus comedicated patients (Table 2) but subsequent analysis showed this was also influenced by the indication for mycophenolate and by serum albumin concentrations.

Table 2. MMF Doses and Plasma Levels in Relation to Tacrolimus or Cyclosporine Comedication in Adult Liver Graft Recipients
VariableSummary Data by Comedication, median (range)
Tacrolimus (n = 1,041)Cyclosporine (n = 182)None (n = 103)
  • Abbreviation: NS, not significant.

  • NOTE. Data are presented for the 95 adult liver graft recipients in whom MPA level was monitored on a total of n occasions throughout MMF therapy. Patients were coprescribed tacrolimus (73) or cyclosporine (22) when MMF therapy began, but tacrolimus was subsequently withdrawn in 8 and cyclosporine in 3, while 1 further subject was converted from tacrolimus to cyclosporine. At the study end, MMF was prescribed alone in 11, with tacrolimus in 64 and with cyclosporine in 20. The initial result for each patient was excluded because no dosage adjustment had been possible using MPA levels. Values expressed as median range.

  • Statistical comparisons (Mann-Whitney U test):

  • *

    tacrolimus vs. cyclosporine.

  • cyclosporine vs. MMF.

  • MMF vs. tacrolimus.

MMF dose (mg)1,0001,000500
Plasma MPA (mg/L)
MPA/mg MMF dose1.22.12.3
 (mg/L per g MMF dose)(0.1–14)(0.3–8.4)(0.15–21.2)
MMF dose/MPA level820476444
 (mg per mg/L MPA)(71–10000)(119–3333)(47–6667)

Univariate Analysis

A univariate analysis was performed to identify determinants of MPA level in the total population of adult and pediatric liver graft recipients. With MMF dose as a determinant factor, patient age (pediatric or adult) and comedication were identified as significant covariates, confirming the observations above. Additional contributions were made by serum albumin concentrations early after transplantation and by a renal sparing indication for MMF and a possibly related effect of serum creatinine concentrations in adults but not children.

Patients with low serum albumin concentrations (<35 g/L) frequently failed to achieve the therapeutic levels of MPA, particularly when MMF was first administered. Figure 3A illustrates the discrete populations of adult patients with normal and low serum albumin levels (defined by 95% confidence intervals). In adults with serum albumin less than 35g/L, median MPA levels were 42% of those in patients with normal serum albumin levels given corresponding doses (P < .001). Correspondingly even lower values (19% of normal, P = .002) prevailed in children (Fig. 3B). Consequently, patients with serum albumin levels less than 35g/L initially required higher MMF doses (2.1-fold in adults, 2.6-fold in children) than their counterparts with normal albumin levels. Serum albumin concentrations increased with time after transplantation in both adults and children, and because patients were given MMF for renal sparing later after transplant than for other indications (43.3 vs. 2.7 months), median serum albumin levels were higher in the CNI sparing cohort (39 vs. 35.5 g/L, respectively, P = .012).

Figure 3.

Influence of serum albumin levels on plasma mycophenolate concentrations in adult (A) and pediatric (B) liver graft recipients. MMF doses and plasma mycophenolate levels are shown in the initial post-transplant sample from patients with serum albumin below or within the normal range (35–55 g/L). Curved lines show 95% confidence intervals.

In adult patients elevated serum creatinine levels (>120 μmol/L) were associated with higher MPA levels per unit MMF dose (median increase 38% early and 50% late after transplantation, P < .04). Unsurprisingly, median serum creatinine concentrations were initially higher in those adults receiving MMF for CNI sparing of nephrotoxicity than for other indications (143 vs. 115 μmol/L, in adults P = .003), but they also differed significantly with comedication (120 vs. 143 vs. 193 in the tacrolimus- or cyclosporine-comedicated and MMF alone sub-groups, respectively; P ≤ .001 for all comparisons). The corresponding variations in MMF dose and MPA levels with comedication and the indication for MMF therapy are illustrated in Fig. 4. The MMF doses required per unit MPA level (shown as apparent relative clearance) were 27% lower in the CNI sparing group overall, (median 476 vs. 652 L per dose interval, P < .001). The lowest apparent relative clearance value was in patients receiving MMF alone who showed the highest median serum creatinine concentrations (Fig. 4). Corresponding differences in children failed to achieve significance, perhaps because growth makes serum creatinine particularly inadequate as an index of renal function.

Figure 4.

Relative clearance of mycophenolic acid in relation to comedication and indication for therapy. MMF was prescribed for renal sparing (“renal indications”) or for acute or chronic rejection, disease recurrence or graft dysfunction (“other indications”).

Assay Performance

During the 25 months of the study (which commenced shortly after the introduction of the Mira analyzer into our laboratory), the EMIT assay required 19 recalibrations, 3 of which were due to reagent lot number changes. Mean (SD) calibration curve lifetime was 5.7 (3.2) weeks. Over the final 3 months of the study, values for interassay precision (coefficients of variation) were 7.7, 6.4%, and 6.2% at 1.1, 2.1, and 5.3 mg/L MPA. Over the prior 22 months analogous values were 12.4%, 11.1%, and 11.5%, respectively, using a second analyzer replaced at our request on the basis of continuing suboptimal performance despite intensive corrective maintenance by engineers. There was good agreement of results in patient and control samples analyzed by EMIT and high performance liquid chromatography throughout this period: EMIT = 1.102 (1.074–1.132) × high performance liquid chromatography + 0.094 (0.041–0.413) (95% confidence intervals) by Passing and Bablock regression with a median proportional bias of 15.6%). Over the final 3 months of the study and the subsequent 10 months to December 2002, mean calibration curve lifetime has been 11.2 wks and unproductive samples (calibrators, quality assessment, and quality control samples) represented 10.8% of the total samples analyzed.


We believe our results show that monitoring of plasma mycophenolic acid concentrations is indicated for 3 reasons: first, drug levels were related to efficacy and major adverse events in liver graft recipients; secondly, plasma MPA concentrations within patient groups were not closely correlated with the dose of mycophenolate mofetil but were dependent upon the indication for prescribing MMF, age and serum albumin and creatinine concentrations; and finally, levels were influenced by the type of immunosuppressant comedication.

To facilitate the monitoring process, a therapeutic range of predose MPA plasma levels of 1 to 3.5 mg/L (by EMIT) was defined in this study. It aimed to avoid acute rejection at the lower extreme, and episodes of infection, leucopenia, and gastrointestinal disturbances shown by us as more prevalent at higher concentrations. It is a more realistic target than the 95% confidence intervals of 0.3 to 5.2 mg/L we identified in patients in whom no adverse effects were documented. The lower therapeutic limit of 1mg/L was based on our observations that 12 of 13 (92%) rejection episodes occurred below this level. The higher limit derived from a more than 3-fold increase in relative risk of adverse events at MPA levels of 3 to 4 mg/L, with 60% of these occurring above 3 mg/L. Both were confirmed by ROC curve analysis. The finding that low predose MPA levels limit efficacy confirms data from the earliest trials with MMF in renal transplantation15 and subsequent results in adult heart and renal16 and pediatric renal transplantation.17 The 1mg/L lower limit also corresponds with that suggested previously in cardiac18 and pediatric renal graft recipients,12 although a second study in heart transplantation19 suggested a cut-off at more than 2 mg/L MPA. There were too few episodes of rejection in this study to analyze whether a MPA level of 1 mg/L is equally applicable during MPA monotherapy (plus corticosteroids) as it is during CNI comedication. Clearly, it represents an acceptable minimum because rejection risk would likely increase in the absence of adjunctive immunosuppression.

The potential for MPA toxicity at higher predose plasma levels is in agreement with the findings in renal transplant recipients of Mourad and colleagues20 who identified a 3 mg/L threshold using ROC curve analysis and with Smak Gregoor et al. who defined no specific threshold MPA level.14 Episodes of leukopenia and infection have been associated with high free MPA concentrations in pediatric kidney transplant recipients17 but not with high total MPA levels as here. The higher MPA levels we noted during these episodes occurred at correspondingly higher MMF doses in leukopenia but at lower doses during significant infections. In contrast, the disproportionally low MPA levels in rejection were associated with normal doses of MMF. This emphasizes the deficiencies of categorical dosing and the benefit of a therapeutic range of predose MPA levels to dosage adjustment. The range suggested (1– 3.5 mg MPA/L) provides the best combination of specificity and sensitivity for adverse event-free therapy with MMF and a buffer of almost 100% above the maximum levels associated with rejection in this study. Arguments that monitoring a single predose sample may provide lesser discriminatory power17, 21, 22 than area under the plasma concentration versus time curve monitoring, must be set against advantages in terms of speed, cost and convenience. Given agreement with earlier recommendations,13 it may be possible to extrapolate our findings for use in recipients of allografts more immunogenic than liver. A further challenging issue to be resolved is that of synergy in relation to therapeutic range setting during cotherapy with multiple immunosuppressive agents.

Our data revealed no close correlation between MMF doses and plasma MPA levels, paralleling previous observations with the immunosuppressants tacrolimus and cyclosporine.23, 24 While efficacy should certainly be achieved with the MMF dose of 1.5 g twice a day suggested for liver allograft recipients by the manufacturer in the CellCept package leaflet, our data suggest this dose is rarely required. Thus, in adult liver recipients stabilized on MMF with the benefit of MPA monitoring, only 6.3% were considered to need 1.5 g twice a day clinically, and 44.7% were prescribed less than 1 g twice a day per day. The saving in drug costs in this latter group alone will fund the consumables cost of continual MPA testing of these patients (at £13.50 per sample) once every 4 days based on current drug and EMIT MPA kits costs. Usually, twice weekly testing is adequate during inpatient stays and a mean of 13.5 tests was performed per adult patient during the mean period of 354 days receiving MMF in our study. Consequently, savings in drug costs in a proportion should more than fund the consumables costs of MPA monitoring of all patients. Additional cost benefit may derive from reduction in the incidence of rejection episodes and drug-related adverse effects together with their associated investigations and treatment. Clinical benefits to the patient will be to reduce both the significant long-term risks of overimmunosuppression, including infection and neoplasia,25 and the short-term risk of other adverse effects. A further reason not to use fixed doses, even when commencing MMF, relates to our observations of altered requirements when serum albumin concentrations were low or renal dysfunction prevailed.

The third indication for MPA monitoring highlighted by our results was the variable influence of immunosuppressant comedication, as noted in previous studies of mycophenolate pharmacokinetics during cyclosporine or tacrolimus cotherapy.8, 26, 27 In a study of 61 children, MPA clearance was greatest during cyclosporine comedication, least during tacrolimus comedication and intermediate when MMF was given without either CNI agent (and without transplantation in 50%).28 The underlying mechanism is suggested to be an inhibition of the excretion and enterohepatic recirculation of MPA glucuronide by cyclosporine compared with inhibition of mycophenolate glucuronidation by tacrolimus.8, 29 Not only does this complicate MMF therapy if conversion between cyclosporine and tacrolimus occurs but it also prompts caution during dose changes of either single agent. Our present data confirmed that differences in MMF dosage requirements and circulating MPA levels prevailed during cyclosporine versus tacrolimus cotherapy. However, in our complete population of adult patients prescribed MMF, the apparent relative clearance of MPA appeared highest during tacrolimus comedication, lowest without comedication and intermediate with cyclosporine, differing from previous observations mentioned above.28 Dissimilarities of renal and hepatic function in the different groups may be responsible, related to the interval posttransplant. For example, patients coprescribed tacrolimus began mycophenolate earlier after transplantation than those on cyclosporine (median 8.2 vs. 62.5 months posttransplant, P < .001). Consequently, serum albumin and creatinine concentrations were lower in the tacrolimus- than the cyclosporine-comedicated cohorts, with the former relating to recovery from chronic liver disease pretransplant, and the lower creatinines to a shorter exposure to nephrotoxic CNI agents posttransplant. Low serum albumin concentrations decrease MPA protein binding and increase clearance of free drug, so lowering MPA levels.22, 30 In our study, apparent relative clearance was more than twice as fast in the patients with low serum albumin than in those with albumin levels in the normal range, hence the faster clearance in the tacrolimus comedicated cohort. In contrast, renal dysfunction impairs MPA clearance via poor excretion of MPA glucuronide, the primary metabolic pathway for MPA removal.5, 12 Our adults with the worst renal function were those in whom CNI had been withdrawn (i.e., the cohort receiving mycophenolate alone, followed by the cyclosporine comedicated cohort with long CNI exposure). MPA accumulation in both is likely given the reduction of 27% in apparent relative clearance of in our patients with renal dysfunction and may contribute to the seemingly conflicting effects of comedication we observed. Our data will need confirmation using liquid chromatography in which the minor acyl glucuronide conjugate of MPA, which may accumulate in renal dysfunction, does not contribute to MPA concentrations as it does with the EMIT assay.31

On the basis of the evidence we have presented, the management of mycophenolate therapy should be improved using drug level monitoring. Such monitoring may be performed cost-effectively and with acceptable precision using the semi-automated EMIT, which has proved faster and more reliable than the corresponding high performance liquid chromatography assay in our hands. The suggested therapeutic range of predose plasma MPA concentrations of 1 to 3.5 mg/L was derived principally in adult liver transplant recipients, and predominantly in those prescribed MMF as adjunctive immunosuppression at a median 24 months after transplantation. It provides reference values as progress is made toward a more widely applicable therapeutic range and contributes positively to the debate on MPA monitoring appraised very recently by Cox and Ensom.32 Mycophenolate monitoring will not only identify those individuals who are at risk from rejection, or the adverse side effects of high plasma mycophenolate levels despite conventional dosage, but will also minimize likely overimmunosuppression from currently recommended definitive doses in the majority, as was concluded recently in a study in pediatric renal transplant recipients.33 It will also accommodate variable MMF dosage requirements in patients whose indications for mycophenolate differ, in patients with changing serum albumin or creatinine concentrations and in those whose comedication with tacrolimus or cyclosporine is changing (e.g., during renal sparing or conversion). Coupled with additional benefits in maintaining therapeutic levels after dose changes over time or when other comedicants are introduced (e.g., antibiotics, antivirals),13, 15 as well as to monitor compliance, the benefits appear substantial. The frequency of monitoring will clearly depend on the clinical condition of the patient in addition to the above factors, but a policy of decreasing testing with time after transplantation and with increasing clinical stability seems appropriate, analogous to that used with tacrolimus or cyclosporine. Additional studies are required in patients prescribed MMF earlier in their postoperative course and for other transplant indications to establish how widely these suggestions apply.


The authors acknowledge the contributions of Professor Giorgina Mieli-Vergani and Drs. Suzanne Norris and John O'Grady to clinical care and of Marian Aw, Ascencio Garcia Lopez, Gill Rogers, Louise Taylor, and Rachel Taylor to data collection.