Journal Logo

Original Clinical Science—General

High Intrapatient Tacrolimus Variability Is Associated With Worse Outcomes in Renal Transplantation Using a Low-Dose Tacrolimus Immunosuppressive Regime

Whalen, Henry R.; Glen, Julie A.; Harkins, Victoria; Stevens, Katherine K.; Jardine, Alan G.; Geddes, Colin C.; Clancy, Marc J.

Author Information
doi: 10.1097/TP.0000000000001129
  • Free

Calcineurin inhibitors (CNIs) have been the mainstay of clinical immunosuppression for more than 30 years. Initially, cyclosporine use predominated with tacrolimus-based regimes, gaining popularity in the last decade.1,2 The narrow therapeutic index of CNIs necessitates therapeutic drug monitoring to achieve efficacy and avoid toxicity. Achieving that balance is complicated by the considerable pharmacokinetic variability between patients (interpatient variability) and within individual patients (intrapatient variability [IPV]).

Intrapatient variability represents the range of spread of blood concentrations measured when therapeutic drug monitoring is applied to an immunosuppressive agent. Given that the purpose of therapeutic drug monitoring is to avoid the complications of too little or too much of the agent, the target range may be seen as the ideal “width of the path” and the variability as a measure of how frequently and how widely this was strayed from while following its course.

Therapeutic drug monitoring entails regular measurement of drug concentration to maintain levels within a prespecified range, based on interpatient variability. This premise entails the assumption that IPV is not a sufficiently large factor to disrupt matters at an individual patient level. In theoretical terms, the higher the IPV, the greater potential for outlying individual drug levels within an overall “on target” mean drug exposure.

Although a vast nonadherence literature makes it unquestionable that frequent or sustained outlying drug levels are associated with negative clinical consequences,3,4 it remains unclear whether more frequent subtherapeutic or excessive trough levels, potentially concealed within mean drug levels measured as on target, are associated with meaningful negative clinical consequences. Theoretically, subtherapeutic immunosuppressant levels may predispose to acute rejection, whereas supratherapeutic levels may cause drug toxicity. Although subclinical, allograft injury during such episodes may contribute to worse outcomes in the long term. Several studies have established that high IPV of CNI levels are associated with worse outcomes5-8 but any potential causal association remains unproven.

After reporting of the Symphony Study9 in 2007, many renal transplant units switched to “low-dose” tacrolimus regimes combined with corticosteroids and mycophenolic acid preparations (MPAs), because this combination improved 1-year rejection rates and renal function compared with analogous cyclosporine-based regimens.

Statistically, however, low-dose tacrolimus treatment may tend to increase tacrolimus IPV, because the absolute change in drug concentration required to lie outside the desired therapeutic window is reduced, and the fractional effect of an absolute change in drug concentration is magnified. The undoubted overall population benefits of low-dose tacrolimus may therefore be associated with increased tacrolimus IPV. For certain individual patients, this might lead to avoidable rejection episodes and/or allograft losses.

This study aims to assess whether the previously demonstrated association between tacrolimus variability and negative outcomes is evident in patients managed with a “Symphony style” low-dose tacrolimus immunosuppressive regimen and to evaluate the magnitude of any difference in outcome.

MATERIALS AND METHODS

Data were prospectively collected in an electronic database (West of Scotland Electronic Renal Patient Record) for all adult kidney transplants performed at our center between January 01, 2007, and December 31, 2011. Data for each patient were collected from the date of transplant until November 19, 2013. At this point, all data were compiled, and outcomes were examined.

Patients were excluded from the study if within the 1st post transplant year, they:

  • –suffered allograft loss for any reason,
  • –transferred to another unit or were lost to follow-up,
  • –were converted from tacrolimus to other immunosuppression,
  • –died.

Induction immunosuppression consisted of basiliximab (20 mg intravenously at day 0 and day 4), methylprednisolone (1 g intravenously), mycophenolate mofetil (MMF) (1 g orally), and tacrolimus (0.05 mg/kg orally). Subsequent maintenance triple therapy immunosuppression was initially 20 mg prednisolone daily, tapering to 5 mg daily by 3 months, MMF (1 g twice daily), and tacrolimus (0.05 mg/kg twice daily adjusted to achieve 12 hour trough blood tacrolimus levels of 5-8 μg/L). Patients with functioning allografts routinely remained on triple therapy immunosuppression in the long term.

Twelve-hour trough tacrolimus levels were measured at every clinic visit by a tandem mass spectrometric method. Lower limit of detection was less than 0.5 μg/L; however, clinical samples were reported using a lower limit of 1.6 μg/L, levels below which were considered as zero for the calculation of variability. An intra-assay coefficient of variation of 8.8% based on 54 samples of fixed concentrations between 3.5 and 16 μg/L was observed.

Calculation of Tacrolimus Variability 6 to 12 Months Posttransplant

Only trough tacrolimus levels considered truly representative were included in the study. Supratherapeutic levels taken when outpatients had taken tacrolimus in error the morning of clinic attendance were excluded from the study. All subtherapeutic levels were included in the study. Because not all patients were treated with a stable dose of tacrolimus during 6 and 12 months posttransplantation, the tacrolimus plasma were levels were corrected for drug dosage.

Using tacrolimus trough levels 6 to 12 months posttransplant, IPV was calculated according to the formula shown below:

Xmean is the mean tacrolimus level for all available samples, n is the number of samples, X1 is the first level available, X2 is the second, and so on.

Median population tacrolimus variability was calculated using all true tacrolimus trough levels between 6 and 12 months posttransplant. Patients with IPV below the observed median were assigned to low variability (LV) group and those with a value equal to, or greater than the observed median considered high variability (HV) group.

The LV and HV groups were compared for the following outcomes:

  • –allograft survival;
  • –patient survival;
  • –MDRD estimated glomerular filtration rate (eGFR) at 1, 2, 3, and 4 years posttransplant, with failed transplants being assigned a GFR of 0;
  • –biopsy-proven acute rejection episodes;
  • –new-onset diabetes posttransplant as defined according to International Consensus Guidelines.10,11

Subsequently, 2 separate subgroup analysis were conducted. The first excluded all patients suffering an acute rejection episode within the first posttransplant year. The second subgroup analysis excluded all patients undergoing retransplantation. In these analyses, median tacrolimus variability was calculated as previously described. Patients whose IPV value was less than the median were assigned to the LV group, and those equal to, or greater than, the median were assigned to the HV group.

Group Baseline Characteristics

Data were collected regarding group baseline characteristics (Table 1). Data regarding MMF and steroid dosing were collected from all patients included in the study and used to calculate mean MMF and steroid doses at 6 months posttransplant. Data regarding donor kidney characteristics were also compiled.

TABLE 1
TABLE 1:
Baseline characteristics of low versus high variability groups

Statistical Analysis

Baseline characteristics of high and low IPV groups were then compared in univariate analysis to determine the effect of IPV on allograft outcomes.

Allograft failure, patient survival, rejection-free survival, and late acute rejection rates in the study groups were compared using Kaplan-Meier analysis and log rank test. χ2 tests were performed to investigate the prevalence of acute rejection within the first year posttransplant. Comparison of mean eGFR was by unpaired t tests. Such data were analyzed using Graphpad PRISM 6.0d. P values less than 0.05 were considered statistically significant.

To assess the independent effect of variables on graft failure after 12 months, a logistic regression was performed using graft failure as a primary outcome. Clinically important factors considered for this model were tacrolimus variability, early biopsy proven acute rejection (E-BPAR), late biopsy proven acute rejection (L-BPAR), delayed graft function, MDRD GFR at 1 year, total MLA mismatch, HLA mismatch greater than 1, donor and recipient ages, total cold ischemic time, retransplanted patients, cytomegalovirus donor-positive/recipient-negative transplants and recipient gender. Criterion for entry into the model was a P value less than 0.2 on univariate testing. The model was constructed using SPSS version 22.

RESULTS

A total of 432 renal transplants were performed in the study period. Fifty-six patients were excluded: during the first posttransplant year, 27 patients (6%) suffered allograft loss, 17 patients (4%) died, and 7 patients (2%) transferred to other centers (ie, overseas) or were lost to follow-up. Five recipients (1%) were converted from tacrolimus to other immunosuppression during the first year (Figure 1). The final study population comprised 376 patients.

FIGURE 1
FIGURE 1:
Overview of reasons for patient exclusion from study.

The median IPV for the study population of 376 patients was 15%. There were 186 patients (49.5%) assigned to the LV group and 190 to the HV group (50.5%). The baseline characteristics of the LV and HV groups were similar and are shown in Table 1.

Twelve patients (3%) died and 20 grafts (5%) failed during the follow-up period (Figures 2 and 3). Allograft function as assessed by MDRD eGFR were compared in the LV vs HV groups at 1, 2, 3, and 4 years of follow-up (see Figure 4). The LV group was found to have significantly better (P < 0.0001) allograft function at all follow-up points.

FIGURE 2
FIGURE 2:
Allograft survival is significantly worse in the high variability group. HR, hazard ratio; CI, confidence interval.
FIGURE 3
FIGURE 3:
No significant difference in patient survival was noted between LV and HV groups.
FIGURE 4
FIGURE 4:
Transplant function as assessed by MDRD GFR and serum creatinine is significantly better in the LV group at all follow-up points (error bars represent SD).

Fifty-one patients (14%) suffered E-BPAR, defined as proven acute rejection occurring within the first posttransplant year. Thirty patients (8%) experienced L-BPAR after 1 year (see Figure 5). More episodes of both E-BPAR and L-BPAR were observed in the HV group, suggesting a relationship between high IPV and increased incidence of acute rejection at all time points posttransplant (see Figure 6).

FIGURE 5
FIGURE 5:
A, Compared with the LV group, significantly more patients with BPAR were observed in the HV group (8.6% vs 18.4%; P = 0.0064) in the first posttransplant year and (B) also during subsequent follow up (4.8% vs 11.1%; P = 0.0351).
FIGURE 6
FIGURE 6:
Rejection-free survival is significantly better in the low variability group.

No difference in incidence of new-onset diabetes after transplantation was found between the LV and HV groups (P = 0.2425).

Analysis of Outcomes After Censoring Allografts Suffering From E-BPAR

Fifty-one patients suffered E-BPAR during the first posttransplant year and were excluded from this subgroup analysis. Median IPV was 14%. One hundred fifty-four patients were assigned to the LV group and 171 patients to the HV group. Significantly worse allograft survival was observed in the HV group (hazard ratio, 4.347; 95% confidence interval, 1.252-15.10; P = 0.0207). Glomerular filtration rate values were observed to be significantly worse in the HV group at 1 year (P = 0.0104), 2 years (P = 0.0039), 3 years (P = 0.0028), and 4 years of follow-up. (P = 0.0152).

Analysis of Outcomes After Censoring Patients Undergoing Retransplantation

Sixty-two patients included in the primary analysis underwent retransplantation and were overrepresented in the HV group (P = 0.0178). To investigate the effect of this, a subgroup analysis was performed, excluding all patients undergoing retransplantation. Three hundred fourteen patients were included, with a median tacrolimus variability of 14%. One hundred forty-six patients were assigned to the LV group and 168 to the HV group. Significantly worse allograft survival was observed in the HV group (hazard ratio, 5.535; 95% confidence interval, 1.903-16.10; P = 0.0017). Glomerular filtration rate values were observed to be significantly worse in the HV group at 1 year (P < 0.0001), 2 years (P < 0.0001), 3 years (P < 0.0001), and 4 years of follow-up (P < 0.0001).

The following clinically important parameters were entered into logistical regression to predict transplant failure, were found to have a P value less than 0.2, and hence were included in the model: high tacrolimus variability, E-BPAR, L-BPAR, age of recipient at transplantation, GFR at 1 year posttransplantation, and delayed graft function.

Previous transplantation, cold ischemic time, cytomegalovirus donor-positive/recipient-negative, greater than 1 HLA mismatch, total HLV mismatch, and proportion of live donations were found to have a P value greater than 0.2 and were excluded.

High tacrolimus variability, age of recipient, L-BPAR, and GFR at 1 year were found to be independent predictors of graft loss after 12 months. The model is robust as demonstrated by an AUC of 0.923.

DISCUSSION

In this study, we demonstrate that the previously described association between high IPV and worse clinical outcomes (acute rejection, lower eGFR at up to 4 years of follow-up, and reduced allograft survival) holds true in patients treated with the widely used combination of low-dose tacrolimus, MPA, and corticosteroids. To the best of our knowledge, this is the largest patient cohort receiving a ‘Symphony’ style immunosuppression regime, with outcomes stratified according to their tacrolimus variability.

Early posttransplant fluctuations in tacrolimus levels may be caused by multiple exogenous events, such as infections, antibiotics, paralytic ileus, intravenous steroid treatments, and so on. We therefore focused on the more stable period of 6 to 12 months postrenal transplantation to study, replicating the methods used in the study by Borra et al.7 Use of this timeframe to calculate IPV reflects the time by which most patients have become established onto a regular tacrolimus dose, yet are still attending outpatient follow-up frequently enough to allow meaningful numbers of tacrolimus trough levels to be recorded and accurate IPV values to be calculated. However, a recent cohort study12 reports inferior allograft outcomes to be associated with the extremes of tacrolimus variability in a time-dependent survival analysis. These findings are broadly in keeping with our observations and suggest that tacrolimus variability is important at all posttransplant time points.

In contrast to the findings of Borra et al,7 we observe significantly more episodes of acute rejection in our HV group, both during the 1st post transplant year (E-BPAR) and during subsequent follow-up (L-BPAR). However, the retrospective nature of this study means that any causal relationship between high IPV and rejection remains unproven.

One key factor that may contribute to variability is drug nonadherence, although unpublished data from our unit has shown that pretransplant dietary phosphate compliance is not a predictor for high IPV after transplantation. This suggests that factors beyond patient adherence may contribute to IPV. Intuitively, patient nonadherence with tacrolimus dosing schedules would seem likely to be a key factor in IPV, although we are not aware of any studies that confirm or quantify this assumption. Patient education and support measures aimed at improving compliance with both medication and diet are relatively easily implemented and would seem a sensible strategy to adopt to reduce IPV.13,14

Multiple daily drug dosing has also been associated with increased risk for nonadherence15 and Wu et al16 have reported a reduction in IPV after conversion from twice daily to once daily tacrolimus formulation. In view of this report and the findings of other studies,17-19 evaluating the effect of switching high IPV patients from twice daily to once daily tacrolimus preparations would be of interest. Further factors associated with high IPV, such as genetic polymorphisms, diet, alcohol intake and other lifestyle factors, are also worthy of further study.

Although we demonstrate inferior allograft outcomes in the HV group, the causal relationship that links high IPV with increased E-BPAR episodes in our study remains a point for speculation. Frequent subtherapeutic tacrolimus levels predispose to acute rejection episodes, and rejection episodes are well known to effect allograft outcomes adversely.15,20 This may explain the associations we observe between reduced allograft survival, worse renal function, and high IPV. However, it is also possible that medical management strategies used to treat acute rejection during hospital admission cause changes in tacrolimus pharmacokinetics and lead to high IPV. Theoretically, therefore, acute rejection may cause high IPV, and it may be incorrect to associate high IPV with worse allograft outcomes, when they are in fact the result of acute rejection episodes.

However, when all patients who experience E-BPAR are excluded from our analysis, we still observe high IPV patients to suffer significantly reduced allograft survival and worse eGFR at 1, 2, 3, and 4 years of follow-up. This finding supports the hypothesis that high IPV is associated with worse allograft outcomes independent of E-BPAR episodes.

Furthermore, we find HV patients to be at increased risk of subsequent L-BPAR. This observation is consistent with the findings of a previous study which concluded that increased IPV was a risk factor for L-BPAR and graft loss independent of earlier rejection episodes.6

Table 1 highlights some differences between the LV and HV group characteristics at baseline. Although statistically significant, published evidence would suggest it is unlikely that small differences in mean daily MPA dose (183 mg) are solely responsible for the large and divergent differences in group outcomes we report.21,22 Although acknowledging the presence of more retransplanted patients in the HV group, a subgroup analysis excluding retransplanted patients reveals the HV group to have significantly worse allograft survival (P = 0.0017), and worse GFR at 1, 2, 3, and 4 years of follow-up (P < 0.0001). This suggests retransplantation is not a contributing factor to outcomes we report. Furthermore, the argument that high IPV has a negative effect on allograft outcomes is supported by our logistical regression analysis using graft failure 12 months posttransplantation as the primary outcome. This multivariate analysis suggests that high IPV 6 to 12 months posttransplantation is associated with earlier subsequent graft failure than those patients exhibiting lower IPV in the 6- to 12-month posttransplant period.

A potential weakness of our study is the finding that more tacrolimus levels were measured 6 to 12 months posttransplant in the HV group (P < 0.0001.)

As both inpatient and outpatient tacrolimus trough levels were used to calculate IPV, we speculate that more measured tacrolimus levels reflect increased surveillance of poorly performing allografts or individuals whose trough levels were frequently outside the therapeutic target range. Increased measured variation as a result of hospital admission may be associated with adverse allograft outcomes and introduce study bias. Unfortunately, we are unable to separate inpatient from outpatient tacrolimus trough levels in our analysis. However, the subgroup analysis of those patients rejection free up to 1 year posttransplant is aimed to at least partly address this potential bias because admission rates in this rejection-free group are likely to have been extremely low, and in this subgroup analysis, the strong association with worse outcome holds true.

Nevertheless, our observations support the hypothesis that high IPV either contributes causally to negative clinical outcomes or is associated with worse results due to covariation with factors undetermined, such as subclinical rejection. Although ambiguity remains regarding the causal nature of the observed relationship between high IPV and negative clinical outcomes, these data make a compelling argument that strategies that reduce IPV might be applied in the context of a clinical trial.

In summary, we report that high IPV is associated with allograft rejection, allograft loss, and inferior allograft function after renal transplantation in the context of low-dose tacrolimus, MPA, and corticosteroids. The nature of this association remains unclear. High IPV seems to identify a subset of patients at increased risk for rejection and allograft loss. This potentially allows increased surveillance and targeted interventions to improve long-term allograft survival and function. Prospective trials of interventions to reduce IPV are justified and should be undertaken.

TABLE 2
TABLE 2:
Independent predictor of graft failure after 12 mo on multivariate analysis (binary logistic regression)

REFERENCES

1. Matas AJ, Smith JM, Skeans MA, et al. Special issue: organ procurement and transplantation network and scientific registry of transplant recipients 2011 data report. OPTN/SRTR 2011 annual data report. Am J Transplant. 2013;13:11.
2. Recipients OPaTNaSRoT. Department of Health and Human Services, Health Resources and Services administration, Healthcare Systems Bureau, Division of Transplantation (ed). Am J Transplant. 2012;13(Suppl 1):11–46.
3. Dunn TB, Browne BJ, Gillingham KJ, et al. Selective re-transplant after graft loss to non-adherence: success with a second chance. Am J Transplant. 2009;9:1337–1346.
4. Pinsky BW, Takemoto SK, Lentine KL, et al. Transplant outcomes and economic costs associated with patient noncompliance to immunosuppression. Am J Transplant. 2009;9:2597–2606.
5. Kahan BD, Welsh M, Urbauer DL, et al. Low intra-individual variability of cyclosporine A exposure reduces chronic rejection incidence and health care cost. J Am Soc Nephrol. 2000;11:1122–1131.
6. Pollock-Barziv SM, Finkelstein Y, Manlhiot C, et al. Variability in tacrolimus blood levels increases the risk of late rejection and graft loss after solid organ transplantation in older children. Pediatr Transplant. 2010;13:968–975.
7. Borra LC, Roodnat JI, Kal JA, et al. High within-patient variability in the clearance of tacrolimus is a risk factor for poor long-term outcomes after kidney transplantation. Nephrol Dial Transplant. 2010;25:2757–2763.
8. Waiser J, Slowinski T, Brinker-Paschke A, et al. Impact of the variability of cyclosporine A trough levels on long-term renal allograft function. Nephrol Dial Transplant. 2002;17:1310–1317.
9. Ekberg H, Tedesco-Silva H, Demirbas A, et al. Reduced exposure to calcineurin inhibitors in renal transplantation. N Engl J Med. 2007;357:2562–2575.
10. Davidson J, Wilkinson AH, Dantal J, et al. New-onset diabetes after transplantation: 2003 International Consensus Guidelines. Transplantation. 2003;7:SS3–SS24.
11. Wilkinson AH, Davidson J, Dotta F, et al. Guidelines for the treatment and management of new-onset diabetes after transplantation. Clin Transplant. 2005;19:291–298.
12. Sapir-Pichhadze R, Wang Y, Famure O, et al. Time-dependent variability in tacrolimus trough blood levels is a risk factor for late kidney transplant failure. Kidney Int. 2014;85:1404–1411.
13. Annunziato RA, Emre S, Shneider BL, et al. Transitioning health care responsibility from caregivers to patient: a pilot study aiming to facilitate medication adherence during this process. Pediatr Transplant. 2008;12:309–315.
14. Vicari-Christensen M, Repper S, Basile S, et al. Tacrolimus: review of pharmacokinetics, pharmacodynamics, and pharmacogenetics to facilitate practitioners' understanding and offer strategies for educating patients and promoting adherence. Prog Transplant. 2009;19:277–284.
15. Morrissey PE, Reinert S, Yango A, et al. Factors contributing to acute rejection in renal transplantation: the role of noncompliance. Transplant Proc. 2005;37:2044–2047.
16. Wu MJ, Cheng CY, Chen CH, et al. Lower variability of tacrolimus trough concentration after conversion from prograf to advograf in stable transplant recipients. Transplantation. 2011;92:648–652.
17. Tinti F, Mecule A, Poli L, et al. Improvement in graft function after conversion to once daily tacrolimus of stable kidney transplant patients. Transplant Proc. 2010;42:4047.
18. Sessa A, Esposita A, Iavicoli G, et al. Cardiovascular risk factors in renal transplant patients after switch from standard tacrolimus to prolonged release tacrolimus. Transplant Proc. 2012;44:1901–1906.
19. Gaber AO, Alloway RR, Bodziak K, et al. Conversion from twice-daily to once-daily extended release tacrolimus (LCPT): a phase 2 trial of stable renal transplant recipients. Transplantation. 2013;96:191–197.
20. Joosten SA, van Kooten C, Sijpkens YW, et al. The pathobiology of chronic allograft nephropathy: immune-mediated damage and accelerated aging. Kidney Int. 2004;65:1556–1559.
21. Watson C, Johnson R, Birch R, et al. A simplified donor risk index for predicting outcome after deceased donor kidney transplantation. Transplantation. 2012;93:314–318.
22. Su VC, Greanya ED, Ensom MH. Impact of mycophenolate mofetil dose reductions on allograft outcomes in kidney transplant recipients on tacrolimus based regimes. Ann Pharmacother. 2011;45:248–257.
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.