Relationship between allograft cyclosporin concentrations and P‐glycoprotein expression in the 1st month following renal transplantation

Abstract

The immunosuppressant cyclosporin is a P‐glycoprotein (P‐gp) substrate whose impaired function has been associated with an increased risk of cyclosporin‐induced nephrotoxicity following renal transplantation. This study investigated the relationship between blood and allograft cyclosporin concentration, and the effect of P‐gp expression. Fifty biopsy samples were obtained from 39 renal transplant recipients who received cyclosporin as part of maintenance immunosuppression. Blood cyclosporin concentrations (2 hours postdose) were obtained from clinical records, matching allograft cyclosporin concentrations were measured in frozen biopsy tissue by liquid chromatography–tandem mass spectrometry, and allograft P‐gp expression was assessed by immunohistochemistry. Blood and allograft cyclosporin concentrations in the 1st month post‐transplantation ranged from 505–2005 μg/L and 0.01–16.7 ng/mg tissue, respectively. Dose was the only significant predictor of allograft cyclosporin concentrations (adjusted R ² = .24, F‐statistic = 11.52, P = .0019), with no effect of P‐gp expression or blood cyclosporin concentrations. P‐gp expression is not the major determinant of allograft cyclosporin concentrations.

1.

What is already known about this subject

• Calcineurin inhibitors (CNIs) are P‐glycoprotein substrates.

• Low renal allograft P‐glycoprotein expression is associated with increased risk of allograft tubulointerstitial damage in patients receiving CNIs, despite dosage adjustment based on blood CNI concentrations.

• Whether allograft cyclosporin concentration variability is due to P‐glycoprotein expression and/or blood cyclosporin concentrations has been inadequately studied.

What this study adds

• Whole blood concentrations of cyclosporin and allograft P‐glycoprotein expression do not predict renal allograft cyclosporin concentrations.

• Dose is the only significant pharmacokinetic predictor of renal allograft cyclosporin concentrations in the 1st month post‐transplant.

2. INTRODUCTION

The calcineurin inhibitors (CNIs) tacrolimus and cyclosporin (CsA) form part of maintenance immunosuppression following renal transplantation, commonly co‐administered with mycophenolate mofetil and prednisolone. However, CNIs are nephrotoxic and a major cause of chronic allograft dysfunction.1, 2, 3 Their low therapeutic indices, and substantial interindividual variability in pharmacokinetics, compounded by pharmacogenetic factors, necessitates therapeutic drug monitoring (TDM) to individualise dosing. Nonetheless, renal allograft CsA concentrations do not correlate with blood concentrations,4, 5 and in liver transplantation it has been suggested that allograft CNI concentrations may better predict rejection.6, 7 P‐glycoprotein (P‐gp) transports CsA8 and is expressed at the brush border of renal tubules,9 facilitating CsA efflux from tubular cells, thereby decreasing intracellular exposure. Inhibition of P‐gp increases intracellular accumulation and cytotoxicity of CsA in cultured human renal epithelial cells.10 Decreased or low renal P‐gp expression in rats is a potential mediator of in vivo CsA‐induced nephrotoxicity.11 In addition, clinical studies following renal transplantation indicate that: low allograft P‐gp expression in tubular epithelium is associated with chronic tubulointerstitial damage12 and histological evidence of CNI induced nephrotoxicity13; lack of allograft P‐gp upregulation following transplantation is associated with increased incidence of biopsy‐proven CsA nephrotoxicity9; and donor allografts with ABCB1 genetic polymorphisms associated with decreased P‐gp function contribute to higher risk of CsA nephrotoxicity,14 tubulointerstitial damage12 or allograft loss.15

These observations suggest that increased allograft and not blood CsA exposure may be an underlying factor determining CsA nephrotoxicity. However, no clinical studies have adequately investigated the relationship between blood and allograft CsA concentrations, or their relationship to allograft P‐gp expression. Therefore, the aims of this study were to determine the relationship between blood and allograft CsA concentrations, and the effect of allograft P‐gp expression on allograft CsA concentrations.

3. METHODS

3.1. Subjects

The study was approved by the Human Research Ethics Committee of The Queen Elizabeth Hospital (approval number 2008178). Thirty‐nine subjects who received kidney transplants between 2004 and 2010 gave informed consent to participate. Maintenance immunosuppression consisted of CsA, mycophenolic acid and prednisolone. Subjects were aged between 17 and 71 (median 47) years; 25 received a deceased donor allograft; 30 were male; cold ischaemia times ranged from 2.5 to 25 h (median 11 h) and HLA donor‐recipient mismatches varied from 0 to 6 (median 4.5). Whole blood CsA concentrations 2 h postdose (C2) were measured as part of routine TDM (CEDIA Plus, Thermofisher, Scoresby, VIC, Australia) on a Hitachi 912 analyser (Boehringer Mannheim Corporation, Indianapolis, IN, USA). CsA doses were obtained from case notes.

3.2. Measurement of allograft CsA concentrations

Allograft CsA concentrations were measured by a validated liquid chromatography–tandem mass spectrometry (LC–MS/MS) method5 in excess tissue from routine postimplantation clinical care renal biopsies stored at −80°C (TissueTEK O.C.T., Sakura, Japan). They were matched to C2 concentrations taken within 24 h of the biopsy. Where an individual had more than 1 biopsy taken, only their 1st biopsy with same‐day C2 concentration was considered for comparisons with dose or P‐gp expression. For 2 patients with multiple post‐transplant biopsies, the relationship between C2 and allograft CsA concentrations was examined visually.

3.3. Immunohistochemistry assessment of P‐gp expression

Formalin‐fixed paraffin embedded (FFPE) pre‐ and postimplantation tissues were available for every patient to allow determination of P‐gp expression at baseline (preimplantation) and at the time of clinical care biopsy. Immunohistochemistry P‐gp staining was performed on 4 replicates of 4‐μm formalin‐fixed FFPE sections. Kidney cortex tissue from a nontransplant nephrectomy patient was used as a positive control. Sections were dried overnight then deparaffinised in xylene, rehydrated using an ethanol/water/phosphate‐buffered saline (PBS) gradient, followed by antigen retrieval in sodium citrate buffer (0.01 M, pH 6) at ~100°C for 20 min. Non‐specific binding of the primary antibody was blocked by incubation in PBS containing bovine serum albumin (BSA; 1%) and normal sheep serum (10%) for 45 min at room temperature. The primary antibody, monoclonal mouse anti‐P‐glycoprotein (MDR) Clone F4 (Sigma, Castle Hill, NSW, Australia), was diluted 1:100 in BSA/PBS, prior to overnight incubation at 4°C; negative control tissue was the nontransplant nephrectomy tissue sections incubated in BSA/PBS without primary antibody. All sections were then blocked against endogenous peroxidase activity (3% hydrogen peroxide in PBS), followed by incubation for 1 h with secondary antibody (ECL anti‐mouse IgG, horseradish peroxidase‐linked whole antibody from sheep; GE Healthcare, Piscataway, NJ, USA) diluted 1:100 in PBS. PBS was used to wash between each step. Positive antibody–epitope interaction was visualised using the Vector NovaRed peroxidase substrate kit (Abacus‐ALS, East Brisbane, QLD, Australia), following the manufacturer's instructions. Sections were counterstained using Gill's haematoxylin (Abacus‐ALS) then dehydrated and mounted for assessment by light microscopy.

Two blinded investigators (B.D.N., J.K.C.) independently scored each stained section semi‐quantitatively into categories according to the staining intensity at the brush border membrane (BB) of proximal tubules (PT), as follows: 0 (negative), absence of any positive staining; 1 (borderline positive), <50% of morphologically clear BB exhibited positive staining and overall PT staining was of weak intensity; 2 (positive), >50% of morphologically clear BB exhibited definite positive staining and overall PT staining was of moderate intensity; or 3 (strongly positive), 100% of morphologically clear BB exhibited intense positive staining and overall PT staining was of strong intensity. The median score of the four replicates from each investigator was calculated. When 2 median scores differed by ≤0.5, an overall median score was calculated by pooling all 8 replicate scores. Where the 2 median scores differed by 1, the final score was reached by consensus, allowing a half‐score for samples that were clearly borderline between 2 categories.

3.4. Data analyses

Comparisons of C2, allograft CsA concentrations and the allograft/C2 concentration ratio between different times post‐transplantation and between different categories of P‐gp expression were performed using Kruskal–Wallis analysis (with Dunn's correction for multiple comparisons). Comparison of changes in P‐gp expression between pre‐ and postimplantation biopsy were assessed by Wilcoxon matched pairs signed rank test. Parametric Pearson correlation analysis was performed to investigate the relationships between: (i) C2 and allograft CsA concentrations; and (ii) CsA dose and C2, allograft CsA concentrations or the concentration ratio, as the data were normally distributed.

A linear multivariate regression model was used to investigate the relationship between allograft CsA concentrations and dose, C2 concentrations and postimplantation P‐gp score using R, 16 with false discovery rate adjusted P‐values <.05 considered significant.

3.5. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY17 and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18.18

4. RESULTS

Fifty biopsies were available from 39 renal transplant recipients. The majority of samples were taken within 2 weeks of transplantation (n = 35), but some were available 1–3, 3–12 and > 12 months post‐transplantation (Figure 1). CsA doses, C2 and allograft concentrations were significantly decreased after 1 month (Figure 1A‐C). In the 1st month there was an ~4‐fold range in C2 (505–2005 μg/L), a 5.5‐fold range in dose‐corrected C2 (0.72–3.97 10⁻³/L), and substantial variability in allograft concentrations (range: 0.01–16.7 ng/mg tissue), even when normalised for dose or C2. In contrast, the median allograft/C2 concentration ratios did not change significantly with time (Figure 1D).

Figure 1.

(A) CsA dose, (B) C2 blood concentrations, (C) matching allograft biopsy concentrations within 24 h of C2, and (D) the allograft/C2 concentration ratios, shown with respect to time post‐transplantation. Line indicates the median

For all biopsies taken within the 1st month post‐transplantation, there was no correlation between C2 and allograft concentrations (R ² = .096; P = .07, Figure 2A). However, in 2 patients where multiple biopsies were available, changes in C2 were reflected by similar changes in allograft concentrations (Figure 2B and C). There was a significant correlation between dose and allograft concentrations (R ² = .26, P = .002, Figure 3B), but no correlation with C2 (R ² = 0.11, P = .052, Figure 3A). Median (range) plasma creatinine concentrations and eGFR at the time of sample collection were 207 (101–846) μmol/L and 29 (5–66) mL/min/1.73m², respectively. There was no correlation of these parameters with either C2 or allograft concentrations, or the allograft/C2 ratio (data not shown).

Figure 2.

Lack of relationship between C2 and allograft concentrations for all biopsies collected in the 1st month post‐transplantation (A); and serial C2 and allograft concentrations in 2 individual transplant recipients (B, C)

Figure 3.

Relationships between dose and (A) C2 concentrations or (B) allograft concentrations (Pearson correlation showing 95% CIs); and effect of postimplantation allograft proximal tubule P‐gp expression score on (C) dose‐corrected C2 concentrations or (D) dose‐corrected allograft concentrations in the 1st month post‐transplantation. Line indicates medians

In the 1st month post‐transplantation, 29 patients were evaluated for P‐gp expression, with the remainder excluded either because FFPE tissue was unavailable or no kidney cortex was identified in tissues. In these patients, there was a small but statistically significant decrease in allograft P‐gp expression score between pre‐ and postimplantation biopsies (median [range] 1.5 [0–3] vs 1.0 [0–3], respectively, P = .02). However, there was no effect of postimplantation renal P‐gp expression score on either dose‐corrected allograft or C2 concentrations (Figure 3C and D). Similarly, there was no effect of preimplantation P‐gp expression (or change in P‐gp expression) on dose‐corrected C2 or allograft concentrations (P ≥ .06, data not shown). A linear multivariate regression model was also used to investigate the relationship between allograft concentrations and dose, C2 concentrations and postimplantation allograft P‐gp score. The only significant predictor of allograft concentrations was dose, accounting for 24% variability (multiple R ² = .27, adjusted R ² = .24, F‐statistic = 11.52, P = .0019).

5. DISCUSSION

It has been proposed that chronic nephrotoxicity may, at least in part, be related to allograft CNI exposure,1 with P‐gp indirectly implicated as an important determinant of allograft CNI concentrations. Garcia del Moral et al.19 1st attempted to quantitate both allograft P‐gp expression and CsA exposure using immunohistochemistry in renal transplant biopsies. Although they reported a significant correlation between the intensities of tissue P‐gp and CsA immunohistochemistry staining, they used an anti‐CsA antibody that had significant cross reactivity with CsA metabolites.19 In contrast, the present study, which specifically measured allograft CsA concentrations by LC–MS/MS, found that allograft tubular P‐gp expression was not the major determinant of allograft concentrations. Instead, dose was the only significant predictor, accounting for approximately 25% variability. In support of our observations, dose has previously been reported as a strong clinical predictor of chronic nephrotoxicity.2 The negligible effect of allograft P‐gp expression on allograft concentrations is similarly supported by lack of association between allograft P‐gp expression and a histological marker commonly attributed to CNI nephrotoxicity, de novo arteriolar hyalinosis.12, 20

The lack of correlation between C2 and allograft concentrations or dose may in part be due to CsA's extensive binding to erythrocytes and plasma proteins, and subsequent low (<5%) and variable unbound fractions in blood and plasma.21 However, only unbound plasma CsA (determined by dose and unbound clearance) is available for allograft uptake. Thus, whilst differences between individuals in unbound fractions, which we were not able to measure, may have contributed to the substantial variability in C2, they do not contribute to variability in unbound plasma CsA and, hence, allograft concentrations. This may partly explain the stronger correlation of allograft (vs C2) concentrations with dose. The use of C2 concentrations may also have been less than ideal, as concentrations in the early postdose period are rapidly changing, reflecting the absorption process, and may not be in equilibrium with those in tissue. In addition, C2 blood and allograft samples were not collected simultaneously and CsA concentrations were measured by different analytical methods (immunoassay and LC–MS/MS, respectively). The difference in collection times and the well described cross‐reactivity of CsA metabolites in immunoassays may also have contributed to the lack of correlation. However, despite this substantial interindividual variability and lack of correlation between blood C2 and allograft concentrations, in 2 individual patients with multiple biopsy samples, changes in C2 concentrations were closely mirrored by similar changes in allograft concentrations, consistent with relatively stable protein binding, allograft uptake and efflux, and elimination of CsA, at least within individuals.

Our study has several limitations. Firstly, it is possible that, as a result of ABCB1 genetic polymorphisms P‐gp expression is not a good indicator of overall function.22 Secondly, the observation that allograft concentrations were significantly higher than those in blood suggests the possibility of extensive allograft binding, and it may be that P‐gp expression is better correlated to unbound allograft concentrations, which, as discussed above, we were unable to measure. In addition, the use of semiquantitative immunohistochemistry may have limited our determination of P‐gp expression and the small sample size may have limited our statistical power. Further, P‐gp also transports cytokines23 and may modulate apoptosis,24 so it is possible that changes in its activity or expression may directly affect allograft inflammation and overall function.1 CsA is also an inhibitor of P‐gp and so may indirectly affect allograft function by altering P‐gp activity.24 Alternatively, some CsA metabolites may be nephrotoxic,25 and P‐gp may play a more important role in determining allograft concentrations of these metabolites. As our study was limited to the 1st month post‐transplantation, we could not address chronic nephrotoxicity. Whilst plasma creatinine concentrations (and eGFR) showed substantial interindividual variability, this was primarily due to delayed allograft function (in approximately 1/3 of the cohort), rather than nephrotoxicity.

In conclusion, this study demonstrated a relationship between dose and allograft CsA concentrations, whilst P‐gp expression was not a major determinant of allograft concentrations. However, further work is clearly necessary to confirm these relationships in a large clinical study, particularly in transplant recipients administered tacrolimus, which has now largely replaced CsA following renal transplantation.

COMPETING INTERESTS

There are no competing interests to declare.

CONTRIBUTORS

B.C.S. study design, funding, data analyses and interpretation, preparation of manuscript. B.D.N. conduct of experiments, data analyses, review of manuscript. J.K.C. study design, funding, data analyses and interpretation, conduct of experiments, review of manuscript. J.T. data analyses and interpretation, review of manuscript. G.R. study design, patient consent and clinical care, review of manuscript. A.A.S. study design, funding, data analyses and interpretation, review of manuscript. All authors have approved the final manuscript and have agreed to be accountable for all aspects of the work.

Sallustio BC, Noll BD, Coller JK, Tuke J, Russ G, Somogyi AA. Relationship between allograft cyclosporin concentrations and P‐glycoprotein expression in the 1st month following renal transplantation. Br J Clin Pharmacol. 2019;85:1015–1020. 10.1111/bcp.13880