• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Oremus M, Zeidler J, Ensom MHH, et al. Utility of Monitoring Mycophenolic Acid in Solid Organ Transplant Patients. Rockville (MD): Agency for Healthcare Research and Quality (US); 2008 Feb. (Evidence Reports/Technology Assessments, No. 164.)

Cover of Utility of Monitoring Mycophenolic Acid in Solid Organ Transplant Patients

Utility of Monitoring Mycophenolic Acid in Solid Organ Transplant Patients.

Show details

4Discussion

Discussion of the Evidence for the Key Questions

Question 1. What is the Evidence That Monitoring Mycophenolic Acid in Patients who Receive a Solid Organ Transplant Results in a Lower Incidence of Transplant Rejections and Adverse events Compared to Patients who are not Monitored?

Only three studies addressed this question. The first, by Meiser et al.,7,8 was really two case series reported together. The study was not designed to compare monitoring versus no monitoring, so the authors did not report important comparative data. For example, there was no presentation of mean plasma predose concentrations for concentration controlled patients who did not have rejection, nor was there a statistical comparison of intra- or inter-group differences. Therefore, one cannot conclude from this study that outcomes or adverse events were affected by monitoring versus no monitoring. The second study, by Flechner et al.,9 found no evidence to suggest that monitoring is associated with a lower incidence of rejection. In contrast, there was evidence to suggest that monitored patients could have a lower incidence of certain gastrointestinal adverse events. However, the evidence regarding rejection and gastrointestinal problems could have been confounded by starting dose. The initial dose of MMF (Mycophenolate Mofetil) was different (2 g in the fixed dose group and 1 g in the monitored group), so any potential effect of a higher starting dose in the monitored group could have been obscured as a result of the study design. As well, more gastrointestinal adverse events might have occurred in the fixed dose group regardless of monitoring because there is evidence of a positive association between larger MMF doses and adverse events.12

The third study,10 the first published randomized controlled trial (RCT) to compare monitoring versus no monitoring of mycophenolic acid (MPA) in any patient group, found a lower incidence of treatment failures (driven primary by a lower incidence of acute rejections) in the monitored (concentration-controlled) group. Although the RCT suggests a potential benefit for monitoring, it is limited to adult kidney transplant patients, so the efficacy of monitoring in other patient populations is still unknown. Likewise, the clinical applicability of the trial's limited area under the curve (AUC) sampling strategy, or the applicability of the 40 mg·h/L MPA target dose, to these other populations is also unknown.

Two further RCTs comparing concentration-controlled versus fixed dose patients have, at the time of this report, been completed yet not published. Some of the data from these trials are publicly available in abstract form. The first RCT is the Opticept study from the United States (Roche protocol number ML 17225).124126 This is a 2 year, open label RCT in kidney transplant patients designed to evaluate fixed dose MMF (1 g BID) versus concentration-controlled MMF (predose-based dose adjustments of 1.3 μg/mL or more in cyclosporine treated patients and 1.9 μg/mL in tacrolimus treated patients). The primary outcome is renal function measured as mean percent change in calculated GFR (Glomerular filtration rate). So far, the investigators have reported that baseline characteristics and renal function were similar between groups in a total sample of 522 persons. Personal correspondence with one of the study investigators (Roy Bloom) and Roche indicate that no results have been published in peer reviewed journals. As well, no timelines were available with respect to when further details of the study might be published.

The Fixed Dose Concentration Controlled (FDCC) RCT127129 is a multicenter RCT conducted in Europe, Canada, South America, Asia, and Australia. Kidney transplant patients (n=901) were randomized to fixed dose MMF (2 g daily for adults, 1.2 g daily per square meter for children) or concentration controlled MMF based on a target MPA AUC0–12 range of 30 to 60 h.mg/L. The primary outcome was a composite of patients who suffered any of the following: biopsy proven acute rejection, graft loss, death, or discontinuation of MMF therapy. According to personal correspondence with lead author Teun van Gelder, the main results have been submitted to the Lancet. The results from a substudy of the FDCC trial have been published.77 In the substudy, which reports on 290 patients, 147 received the fixed dose and 143 received the concentration controlled dose. The purpose of the substudy was to examine the incidence of diarrhea. The patients were further divided by type of concomitant therapy: cyclosporine and MMF (n=56 fixed dose; n=54 concentration-controlled) or tacrolimus and MMF (n=91 fixed dose; n=89 concentration controlled). Within the cyclosporine/MMF group and the tacrolimus/MMF group, there was no difference in the number of cases of diarrhea between fixed or concentration controlled patients (p>0.05). When the groups were compared to one another, the incidence of diarrhea was higher in the tacrolimus/MMF group (n=69 versus n=17 in the cyclosporine/MMF group [p<0.001]). MPA AUC0–12 values did not differ between patients who suffered diarrhea and patients who did not (p>0.05).

While the results of these other two multicenter studies are being anxiously awaited, it should be noted that the study populations involve kidney transplant recipients, so the results may not be directly applicable to other solid organ transplant subpopulations. Certainly, RCTs in these other subpopulations are warranted before the key question can be more fully answered.

Question 2. Does the Incidence Differ by any of the Following?

2a: MPA Dose and Dose Frequency

Overall, the evidence to support an association between MMF dosage and rejection is outweighed by the evidence against. However, an equal number of studies supported and refuted the association between MMF dosage and adverse events. Unfortunately, most of the evidence was in the form of case series. Furthermore, even the relatively few higher quality studies (e.g., cohort studies) were not designed to address whether MMF dosage is associated with rejection or adverse events. These factors, coupled with the diversity of other variables in the studies (e.g., concomitant medications, different lengths of followup, specific adverse events evaluated) make it difficult to provide a clear answer to the question. What is direly needed are RCTs that compare patients who are monitored to patients who are not monitored. Ideally, these trials would permit comparisons at different fixed doses and at different targets for concentration control.

2b: Type of MPA (mycophenolate mofetil [CellCept®], enteric-coated mycophenolate sodium [Myfortic®])

The recently introduced, enteric-coated, delayed release formulation of MPA (i.e., enteric-coated mycophenolate sodium (ECMPS)) was designed to reduce upper gastrointestinal adverse events. ECMPS delivers the same MPA exposure (AUC) as MMF and is therapeutically equivalent, but leads to higher C0 concentrations.130 None of the included studies directly compared ECMPS with MMF. Studies that were not helpful in answering question 2b included those without control group, e.g. with all patients switched from MMF to mycophenolate sodium. Due to small numbers, adverse events or rejection events were not observed or could not be correlated with PK parameters in many studies, so question 2b could not be answered.

Clinicians should be aware of the potential for higher predose plasma or serum concentrations (C0) with ECMPS compared to MMF. Full AUCs are not expected to be different between the two formulations, but are too difficult to use in standard practice situations. Predose concentrations or abbreviated sampling strategies are more realistic, but due to the delayed absorption of ECMPS, they will have to be validated separately from MMF. Future randomized concentration-controlled trials comparing no monitoring to monitoring with different target PK parameters could establish therapeutic concentrations for mycophenolate sodium and evaluate the utility of monitoring at the same time.

Question 3a: Does the Incidence Differ by Total Versus Free MPA, Albumin, Metabolites, Genetic Differences or by Analytical Method of MPA Monitoring?

Does the Incidence Differ by Total Versus Free MPA or Albumin?

Only free, protein unbound drug molecules are available for receptor binding. Therefore, measurements of free MPA (fMPA) may theoretically be expected to correlate better with outcomes than total MPA. However, none of the included studies confirmed this hypothesis, although free (not total) MPA was found to be associated with infections and haematological adverse events.13,14,17 Thus, there is potential for the utility of fMPA monitoring, but this has yet to be demonstrated in an RCT. Many of the studies in this report showed that impaired renal function and hypoalbuminemia coincide with elevated mycophenolic acid glucuronide (MPAG) and fMPA, but not total MPA. The mechanisms involved are complex. In renal failure, MPAG excretion is decreased, the accumulated metabolite displaces MPA from albumin, and the added fMPA is available not only for therapeutic or toxic effects, but also for hepatic clearance. Measures of total MPA do not reflect these processes and might even be decreased. Given the added complexity and limited availability of fMPA testing, an alternative would be to measure total MPA while taking renal function and serum albumin into account. Recently, however, Roche has introduced an Inosine 5′-monophosphate dehydrogenase (IMPDH) based assay for free and total MPA. A CEDIA assay is now available from Microgenics.

Does the Incidence Differ by Genetic Differences or Metabolite Concentrations?

The pharmacogenetic study by Naesens et al.79 showed that carriers of the two MRP2 (multidrug resistance protein) SNPs (single nucleotide polymorphisms) were protected from reduced MPA exposure in mild liver dysfunction. The other genetic study, by Satoh et al.,30 found associations between MPA and genes, genes and diarrhea, and MPA and rejection. The clinical relevance of both studies is unclear, as they do not suggest how monitoring of MPA could be augmented to prevent rejection or adverse events. The biochemical mechanisms are not well enough understood and genetic screening for the mentioned polymorphisms does not seem warranted. More basic and clinical research appears necessary.

The studies regarding metabolites yielded few positive results. The fat malabsorption results,16 based on five patients, apply to a very specialised population. The only other significant associations were those between AcMPAG (acyl glucuronide metabolite of mycophenolic acid), MPAG, and anemia, but not to other adverse events or efficacy endpoints.15 Monitoring of metabolites cannot be generally recommended based on these results. The pharmacokinetics of MPA is very complex, involving enterohepatic recirculation, competition of parent drug and MPAG for albumin binding, many drug-drug interactions and other complicating factors. Although the active metabolite (AcMPAG) may hold some promise in predicting toxicities, the mechanisms leading to adverse events, especially GI effects, are not yet understood and should be studied in the laboratory. Larger, randomized trials are necessary to establish the utility of monitoring MPA and its metabolites.

Does the Incidence Differ by Assay Method?

In two studies,26,27 HPLC (high-performance liquid chromatography) and EMIT (enzyme-multiplied immunoassay technique) performed similarly well in the assessment of acute rejection risk in pediatric kidney transplant patients. As expected, EMIT cut off values were higher than those derived from HPLC measurements. This is because immunoassays often show a positive bias compared to more specific chromatographic techniques. As well, the EMIT for MPA cross reacts with AcMPAG, an active metabolite of MPA.119 Theoretically, EMIT could be advantageous over HPLC because it might reflect total immunosuppressive activity better, although this is not certain because cross reactivities are concentration dependent, and the two studies did not find EMIT to be superior. Potentially higher cut off values for EMIT mean that target ranges for total MPA AUC0–12 or C0 will have to be derived separately for HPLC and EMIT.

The general implications of the findings are difficult to assess. Only two studies26,27 directly compared HPLC and EMIT; the study populations in both studies were pediatric patients. It remains to be seen whether diagnostic sensitivities and specificities would differ between methods in other populations. In one study,27 the age and sex distributions of pediatric patients were not provided, so it was difficult to know exactly to whom the diagnostic sensitivities and specificities were applicable. As well, there is currently no information about the comparative merits of HPLC or EMIT in conjunction with other assay methods, such as HPLC-MS, because no study was undertaken to make such comparisons.

Adverse events were considered in one study26 and MPA PK (pharmacokinetic) parameters were not found to predict them, regardless of assay method. However, there is some evidence in this report that PK parameters can distinguish between persons with and without adverse events. Perhaps the findings apply only to the specific profile of pediatric patients enrolled in the study. Another possibility is the potential for bias. Weber et al.26 did not explain the basis upon which their patients were chosen, thus raising the issue of selection bias. Verification bias may also have been present because some patients did not undergo biopsy, nor was there any reporting of stratification according to the factors that triggered biopsy.

Another issue with the two studies26,27 that are pertinent to Question 3aii was the lack of clarity concerning how the operating points on the ROC (receiver operator characteristic) curves were chosen. Other choices of decision levels and their corresponding sensitivity/specificity pairs may have been more appropriate, depending on the prior probability of rejection, the importance of correct classification, and the relative undesirability of false positive or false negative errors.

Ultimately, since the goal of monitoring is the prevention (not diagnosis) of rejection and adverse events, the utility of monitoring will have to be assessed in trials designed to study this goal. A factorial trial would be appropriate to study monitoring versus not monitoring in conjunction with the efficacy of measuring MPA using different assay methods, including the new assays for total and free MPA mentioned above. Alternatively, reference therapeutic PK parameters for different assay methods could first be derived from observational studies and then tested in an RCT. A similar strategy may apply for all key questions.

3b: Does the Incidence Differ by Method of MPA Monitoring (Full AUC or Limited Sampling Strategies [i.e., Predose Concentrations, 2 hour Post Dose Concentrations, Other])?

Overall, the evidence to support an association between full AUC (AUC0–12) and rejection outweighs the evidence against. The opposite is true for the association between full AUC and adverse events. There are more studies showing that predose (C0, Cmin, or C12) compared to full AUC measurements are associated with both rejection and adverse events, but there are an even greater number of studies demonstrating that trough has no association. Equal numbers of studies demonstrate positive versus no associations between monitoring using other limited sampling strategies and rejection, but when adverse events are considered there are more studies showing a lack of association rather than an association.

Since full AUC measurements are cumbersome and impractical to use clinically, and more studies demonstrate the lack of utility of trough in discriminating between patients with and without rejection or adverse events, we are left to consider other limited sampling strategies. To date, C2 has not been well studied and there appears to be no consensus regarding the utility of other limited sampling strategies in discriminating between rejectors and non rejectors. However, there are three times as many studies that demonstrate the lack of utility of other limited sampling strategies in predicting adverse events.

The evidence for answering this question is limited by the objectives of the included studies. Most of the studies were observational or case series designs developed with the intention of studying the biological or pharmacological effects of MMF dosing or MMF in combination with a calcineurin inhibitor. Some earlier exploratory studies were undertaken to obtain information on the associations between PK parameters and dosing, time, or other PK parameters. None of the studies were designed to compare the incidence of rejection or adverse events in groups of patients whose MMF doses were controlled using different sampling strategies. Although many studies had multiple sampling strategies measured on the same patients, these measurements were not used for dose adjustment. Rather, the authors of these studies sought to examine whether mean measurement values were associated with an outcome such as rejection or adverse events. These data are hypothesis generating because they can provide insight into the types of sampling strategies to use in monitoring, but they do not actually indicate whether monitoring and dose adjustment would have an affect on outcomes.

Question 3b can best be answered with head-to-head (RCT) comparisons of monitoring and dose adjustment using different sampling strategies. To date, there is only one published study comparing concentration-controlled and fixed dose MMF.10 In the concentration-controlled group, the investigators used a 3-sample limited sampling strategy (developed by Bayesian techniques) to predict MPA AUC. Although the concentration-controlled group had significantly lower treatment failures and acute rejections, there was no significant difference in incidence of most adverse events, save for the incidence of herpes infections, which was greater in the concentration-controlled group. As eloquently articulated in an editorial accompanying the published trial, “One is left to wonder that despite an elegant and elaborate algorithm for dose changes, could these same [adverse effect] results have been obtained by simply administering higher doses of MMF without MPA monitoring?”131

Question 4. Does the Evidence for Monitoring MPA Differ by any of the Following - Age, Gender, Ethnicity, Concomitant use of Calcineurin Inhibitors, Concomitant use of Other Medications, Comorbidity?

Across all parts of Question 4, most of the evidence from the literature search did not directly address the key question. Studies of direct relevance would have evaluated whether monitoring MPA in recipients of solid organ transplants would have led to a lower incidence of rejections or adverse events compared to not monitoring, with subanalyses (specified a priori) stratified by factors such as age, gender, ethnicity, concomitant use of medications, and comorbidities. To date, no such study exists.

The majority of included studies focused on adults and kidney transplant recipients. Few studies involved children, the elderly, or other solid organ transplants. Study findings were difficult to compare because measures of MPA in plasma or serum sometimes exhibit large intra- and inter-patient variability over time post transplant. Moreover, the factors of concern (e.g., age) in this question were not consistently addressed in all of the included studies. Inconsistency was also a hallmark of outcome definition or selection, thereby further detracting from comparability. For example, rejection was inconsistently defined, sometimes clinically via Banff criteria and sometimes using surrogate endpoints such as GFR or serum creatinine. A consistent basket of adverse events was also not the norm. Many studies looked at particular adverse events (e.g., gastrointestinal, liver dysfunction) or did not clearly define the types of adverse events that were under examination. Some published studies, primarily rapid communications such as the work of Behrend et al.,25 provided limited raw data to support descriptive results and conclusions.

Based on the evidence that could be gleaned from the included studies, certain patient demographics appeared to influence MPA PK parameters. Within pediatric populations, the evidence suggested that younger children may require a higher MMF dose to achieve a specified MPA concentration. Similarly, the evidence suggests that the elderly have lower MPA exposure compared to younger adults receiving the same dose of MMF. However, the bulk of the evidence indicated no association between patient age and MPA PK parameters in general (i.e., over all age ranges without stratification into pediatric and adult populations). Regarding gender, the evidence suggested AUC0–12 and predose concentrations might be higher in women, but the impact of these findings for monitoring rejection or adverse events was not studied. Race and ethnicity did not appear to influence PK parameters.

Calcineurin inhibitors are co-administered frequently with MMF and many studies examined the relationship between these drugs and MPA PK parameters. The evidence found that exposure to MPA is higher in patients receiving tacrolimus compared to cyclosporine, with lower doses of MMF required in combination with tacrolimus to achieve adequate MPA exposure. This difference is explained by the inhibition of the enterohepatic circulation of MPA by cyclosporine. Concomitant use of medications not only influences the MPA exposure but also may affect the utility of therapeutic drug monitoring (TDM). If a solid organ transplant recipient is receiving four different immunosuppressants with a low rejection risk, the overall immunosuppressant effects depend to a much lesser degree on the correct dosing of MPA, whereas in a regimen with only two immunosuppressants and a higher risk of rejection, the overall adequacy of immunosuppression depends heavily on the correct dosing and exposure of MPA.

The effect of renal function on MPA PK parameters was addressed in a number of studies, but the findings were inconsistent and inconclusive.

Question 5. What is the Short- and Long-Term Cost-Effectiveness of Avoiding Acute Rejection due to MPA Monitoring?

The published literature contains no data on the cost effectiveness of monitoring versus no monitoring in solid organ transplants. Therefore, it is not possible to answer this key question.

At the time this report is being written, the authors of the lone published RCT on monitoring versus no monitoring10 report that an economic evaluation of their trial results is ongoing. These results, once published, will be an important addition to the literature. For a monitoring strategy to be cost-effective, the additional costs of implementing the monitoring protocol would have to be exceeded by the savings associated with treating fewer rejections or adverse events. From the perspective of a public or private health insurer that is considering whether to reimburse the cost of monitoring, it is not sufficient to simply look at cost data. Effectiveness data (e.g., quality adjusted life years [QALYs]) should also be considered and evaluated using standard methods of cost effectiveness analysis.100 The result of such an analysis would be to obtain an incremental cost per unit of effect (e.g., cost per QALY). This ratio can be used to compare monitoring with other competing healthcare programs, thereby allowing insurers to determine which program is most effective per unit of cost. Such information can be used to help make decisions about which program(s) to reimburse.

Limitations of This Evidence Report

Only English language, published studies were included in the report. The available budget and timelines limited the McMaster University Evidence-based Practice Center's (MU-EPC's) ability to obtain, translate, and abstract non English or unpublished studies. In addition, study authors were not contacted to obtain supplemental data that were not presented in the published articles. It has been the MU-EPC's experience that the majority of authors do not respond in a timely fashion, if at all, to requests for information. These omissions may have introduced publication bias into this evidence report.

Virually all of the studies involve MMF, not ECMPS. The generalizability of MMF data to ECMPS should be handled with extreme caution because differences in absorption kinetics make, it difficult to substitute algorithms developed for limited sampling strategies in MMF to ECMPS. In addition, the utility of predose concentration measurements may be even more limited for persons receiving ECMPS than for persons receiving MMF because the enteric-coated formulation is particularly prone to delays in gastric emptying time. As a result, very high morning predose concentrations can be encountered.

The evidence report contains all of the relevant literature to address the key questions up to and including October 2007. This means that new and potentially important studies published after this date will not be included unless a future update of the report is commissioned.

Conclusions

The state of knowledge about therapeutic drug monitoring of MPA in solid organ transplants is still in its infancy. This is especially so for organs other than the kidney because the overwhelming majority of published studies involve kidney transplant patients. There is direct evidence from only one study10 to suggest that monitoring would reduce the incidence of rejection in adult kidney transplant patients. Two soon to be published trials (Opticept, FDCC) will supplement this limited evidence, but many issues will remain outstanding. These issues include the optimal method of MPA monitoring. The most complete and most studied method is the full AUC (AUC0–12), but this procedure requires at least eight blood samples over a 12 hour dose interval and is therefore impractical to use in most clinical settings. Evidence for the utility of limited sampling strategies (e.g., predose [C0, Cmin, C12]) is equivocal at best and largely based on case series or observational studies whose primary purpose was something other than to compare strategies. Other limited sampling strategies (e.g., C2, multiple sample strategies, etc.) have not been studied well enough to assess their utility for monitoring.

Another issue is the lack of an obvious MPA target concentration to govern dose adjustment. The selection of such a concentration depends on the sampling strategy and may be frustrated by the wide intra-patient variability in MPA plasma concentration time profiles, especially if the influence of time after transplantation is not accounted for. Even if a standardized target concentration can be agreed upon, there are too few studies to guide the choice of assay or suggest the best frequency for measuring MPA in the plasma or serum. At this point, there is no evidence to even suggest whether assay type matters.

The utility of monitoring MPA is further muddled by the fact that resolutions to all of the aforementioned issues may differ by type of drug (MMF, ECMPS), dose, population characteristics (adult, pediatric), comorbidity, concomitant medications, and type of organ transplanted. There is certainly evidence to suggest that these items matter (e.g., physicians targeting MPA predose concentrations must note the existence of higher morning C0 concentrations with ECMPS130), but the literature provides no clear guidance on how to operationalize them clinically. Furthermore, there is little data available on the long term pharmacokinetics of MPA. The extent to which changes in pharmacokinetic parameters over time post transplant can affect the utility of TDM needs to be the subject of investigation.

Another knowledge gap is in the area of economic evaluation. No published study has contained an examination of whether monitoring is cost effective versus no monitoring. The results of such an analysis could influence the reimbursement decisions of private or public health insurers. These decisions are important because they affect patient access to treatment.

Quality is also an issue. Reporting of some essential features of RCT design (e.g., method of randomization, blinding) and observational study design (e.g., blinding) was lacking in most studies. Additionally, only 28 of 75 observational studies reported attempts to control confounding. Since none of the observational studies contained direct evidence to address the key questions, the studies can be regarded as hypothesis generating rather than hypothesis confirming. The quality issue further reinforces the notion of hypothesis generation versus confirmation. Studies with quality challenges may not have valid results because of bias and confounding. Consequently, the results of these studies should be verified in future research, preferably using well-designed RCTs.

Overall, the published evidence on MPA monitoring is inconclusive, with some studies suggesting potential benefits and other studies suggesting no benefit. This makes the issuance of clinical recommendations difficult. There is no evidence, except for one published RCT, to suggest that monitoring is more or less beneficial than not monitoring. Until there is more evidence on the utility of routine MPA monitoring in solid organ transplant recipients, patients, clinicians, and other stakeholders (e.g., public and private insurers) will have to decide on a case by case basis whether the possible but uncertain benefits are worth the extra time and expense of monitoring.

PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...