The tacrolimus concentration‐to‐dose ratio is associated with kidney function in heart transplant recipients

Abstract

Aim

Heart transplantation (HT) is frequently complicated by chronic kidney disease, of which tacrolimus‐related nephrotoxicity is an important cause. In kidney and liver transplant recipients, fast tacrolimus metabolism (defined as a low concentration‐to‐dose [C ₀/D] ratio), negatively affects kidney function. Here, the association between the C ₀/D ratio and kidney function in HT recipients was investigated.

Methods

This was a retrospective study including 209 HT recipients who received an immediate‐release tacrolimus formulation. The C ₀/D ratio and kidney function (estimated glomerular filtration rate [eGFR]) were assessed at 3, 6, 12, 36 and 60 months post‐HT. Patients were categorized as fast, intermediate and slow metabolisers, depending on their individual median C ₀/D ratio as calculated over the follow‐up period. A linear mixed‐effects model analysis was performed, in which the time‐varying eGFR was the dependent variable.

Results

The distribution of the individual median C ₀/D ratios ranged from 0.41 to 8.9 ng/mL/mg. At baseline, patients' kidney function was comparable. In the multivariable linear mixed‐effects model, fast metabolisers (C ₀/D ratio ≤1.53) had a significantly lower eGFR compared to slow metabolisers (C ₀/D ratio >2.27) (−6.8 mL/min/1.73 m², 95% CI −11.2, −2.4, p = 0.002). This association was confirmed when utilizing the individual median C ₀/D ratio as a continuous variable: for each 1 unit increase in the C ₀/D ratio there was a 2.8 mL/min/1.73 m² (95% CI 1.0, 4.5) increase in eGFR (P = 0.002).

Conclusion

Fast tacrolimus metabolism is significantly associated with worse kidney function in HT recipients in the first 5 years post‐HT when compared to recipients with intermediate and slow tacrolimus metabolism.

What is already known about this subject

• The tacrolimus concentration‐to‐dose (C ₀/D) ratio has been proposed as a surrogate marker for an individual's metabolism.

• In kidney and liver transplant recipients, a low C ₀/D ratio is associated with worse kidney function.

• In heart transplant recipients, chronic kidney disease is a severe complication, of which tacrolimus‐induced nephrotoxicity is an important cause.

What this study adds

• A low tacrolimus C ₀/D ratio is associated with worse kidney function in the first 5 years after heart transplantation.

• Simulation of concentration‐time curves demonstrates a higher exposure to tacrolimus in patients with a low C ₀/D ratio.

• Identification of patients with a low C ₀/D ratio presents an opportunity for closer monitoring of tacrolimus exposure in these patients by, for example, the measurement of an area under the concentration vs time curve.

1. INTRODUCTION

Heart transplantation (HT) is frequently complicated by chronic kidney disease (CKD), ¹ , ² , ³ and as many as 20% of HT recipients develops end‐stage kidney disease (ESKD). ⁴ , ⁵ This is associated with a six‐fold higher risk of death compared to patients who do not develop ESKD. ⁴ Although the causes of ESKD after HT are several, an important cause is the nephrotoxicity of the immunosuppressant tacrolimus. ⁶ , ⁷

Tacrolimus is a calcineurin inhibitor (CNI) that is metabolized by the cytochrome P450 (CYP) enzymes 3A4 and 3A5 in the intestine and liver. ⁸ The drug is known for its highly variable inter‐patient pharmacokinetics, ⁹ which are explained by several factors, including CYP3A single‐nucleotide polymorphisms (SNPs), age, drug‐drug interactions and sex. ¹⁰ SNPs in CYP3A explain up to 40‐50% of the inter‐patient variability in tacrolimus' pharmacokinetics, ¹¹ , ¹² with CYP3A5 expressers requiring approximately a 1.5‐fold higher tacrolimus dose compared to non‐expressers. ⁸ HT recipients are often treated with drugs that interact with tacrolimus causing further variability in tacrolimus' pharmacokinetics (reviewed by Sikma ¹³ ). Many of tacrolimus' side effects are related to its concentration in blood, ⁹ and since tacrolimus also has a narrow therapeutic index, therapeutic drug monitoring (TDM) is routinely performed. ¹⁴ The parameter most widely used for TDM is the whole blood, pre‐dose concentration (C ₀). ¹⁴

A simple way to estimate a recipients' oral tacrolimus clearance was proposed by Thölking. ¹⁵ Dividing the C ₀ by the corresponding daily dose (D) results in the C ₀/D ratio. In kidney and liver transplant recipients, a low C ₀/D ratio (indicating a fast metabolism) was associated with decreased kidney function, ¹⁵ , ¹⁶ , ¹⁷ a higher prevalence of tacrolimus‐induced nephrotoxicity, ¹⁷ , ¹⁸ a higher allograft rejection rate, ¹⁶ an increased risk of kidney allograft loss ¹⁶ , ¹⁹ and higher mortality ¹⁵ when compared to patients having higher C ₀/D ratios. These adverse events could be explained by the higher dose needed to get the tacrolimus exposure within the predefined target range, exposing these fast metabolisers to high peak concentrations (C _max). ¹⁸ , ²⁰ Additionally, the formation of certain tacrolimus metabolites (among which 13‐O‐desmethyl‐tacrolimus is the most important) ²¹ may be affected by a higher daily dose and this may have an effect on clinical outcomes, although this effect appears to be small. ²² , ²³

An advantage of calculating the tacrolimus C ₀/D ratio over genotyping for CYP3A is that it may be a better indicator of metabolic phenotype as it also takes into account the effect of drug‐drug interactions, and other demographic and biological variables. Calculation of the C ₀/D ratio may thus identify patients at risk for adverse events, but its association with long‐term kidney function in HT recipients has never been investigated. This study aimed to investigate the association between the tacrolimus C ₀/D ratio and long‐term kidney function in HT recipients.

2. METHODS

2.1. Study design

This was a retrospective study of HT recipients transplanted in the Erasmus MC, University Medical Center, Rotterdam, the Netherlands. This study was conducted in accordance with the Declaration of Helsinki. The study used data from the HT recipient database of our department (MEC number 2017‐421). A waiver of consent was obtained from the Medical Ethical Review Board of the Erasmus MC to retrieve information regarding immunosuppression from the patient files (MEC number 2023‐0077).

2.2. Patient selection

Patients ≥18 years, who received a HT from the year 2000 onwards (when tacrolimus was introduced as the primary immunosuppressant) and received an immediate‐release tacrolimus formulation as initial immunosuppressive medication post‐HT for at least 6 months were eligible for inclusion. Exclusion criteria were as follows: the HT was a re‐transplantation, deceased within 3 months post‐HT or renal replacement therapy (dialysis or kidney transplantation [KT]) within 3 months post‐HT (as we are interested in long‐term kidney function). No multi‐organ transplants including the heart are performed in our centre.

2.3. Immunosuppressive regime

The immunosuppressive therapy post‐HT in our institute consists of rabbit anti‐thymocyte globulin (r‐ATG) as induction therapy, and tacrolimus combined with mycophenolate mofetil (MMF) and prednisolone as maintenance therapy. The target tacrolimus C ₀ in the first year post‐HT was 10‐16 ng/mL, and this was lowered to 4‐8 ng/mL in case of no rejection episodes. ²⁴ Patients were switched from tacrolimus or MMF to everolimus in case of cardiac allograft vasculopathy or leukopenia, respectively. The target everolimus C ₀ was 3‐6 ng/mL.

2.4. Data on tacrolimus

All available tacrolimus C ₀ with the corresponding daily doses were collected at the pre‐defined time periods of 3 months ± 2 weeks (M3), 6 months ± 3 weeks (M6), 1 year ± 1 month (Y1), 3 years (Y3) ± 1 month and 5 years ± 1 month (Y5) post‐HT, only if tacrolimus was administered orally. All C ₀ were measured before the next tacrolimus dose but could differ in timing of the sample regarding the 12‐h interval. The tacrolimus C ₀/D ratio was then calculated according to Equation 1. Since the follow‐up period was at least 6 months, at least two C ₀/D ratios were included per patient.

$$C0/D\text{ratio}\left(\mathrm{ng}/\mathrm{mL}/\mathrm{mg}\right)=\text{tacrolimus}C0\left(\mathrm{ng}/\mathrm{mL}\right)/\text{daily corresponding tacrolimus dose}\left(\mathrm{mg}\right)$$ (1)

All tacrolimus concentrations were measured in our central hospital laboratory using three different methods. Samples from 2000‐2012 were measured with antibody conjugated magnetic immunoassay, from 2013‐2015 with enzyme multiplied immunoassay and from 2016 on with liquid chromatography‐tandem mass spectrometry. ²⁵

No information was available regarding potentially interacting drugs or compliance to tacrolimus.

2.5. Kidney function

The estimated glomerular filtration rate (eGFR) was assessed by means of the Chronic Kidney Disease Epidemiology Collaboration formula ²⁶ at the previously mentioned time periods. Since no cystatin C measurements were available, the newer equations to estimate GFR could not be used. ²⁷

2.6. Clinical course after transplantation

Data were collected on comorbidities, rejection episodes and heart function.

Information on the concomitant use of antihypertensive drugs was collected, which was scored when a patient received one of the following: angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers, mineralocorticoid receptor antagonists, loop diuretics or thiazide diuretics.

Routine endomyocardial biopsy (EMB) was performed during the first year post‐HT for rejection surveillance. Hereafter, EMBs were only taken when rejection was suspected. Histology was graded by three consecutive systems: Billinghams' original criteria, the 1990 standard of grading from the International Society for Heart and Lung Transplantation (ISHLT) and the revised 2005 standard of the ISHLT. ²⁸ , ²⁹ , ³⁰ , ³¹ Acute cellular rejections were treated with pulsed high‐dose corticosteroids or, in the event of a steroid‐resistant rejection, with r‐ATG, whenever “moderate rejection” or more, grade ≥3A (1990 grading) or grade ≥2R (revised 2005 grading), was diagnosed. Antibody‐mediated rejection was treated only in case of signs of graft failure in combination with histological and/or immunopathologic findings. ²⁴

Left ventricular function was assessed as good, moderate or poor based on wall motion results on annual echocardiography and left ventricular ejection fraction. ²⁴

2.7. Statistical analysis

Continuous variables are reported as mean ± standard deviation or median (25th, 75th percentile). Categorical variables are shown as number and percentages. Fisher's exact test was used to test for differences in categorical variables between groups. The Kruskal‐Wallis test was performed to examine differences in continuous variables.

The primary dependent variable for all analyses was the eGFR (mL/min/1.73 m²). Individual median C ₀/D ratios were calculated using all five time points to correct for the known intra‐patient variability of tacrolimus. ³² Tertiles of the distribution of these medians were used to define three groups of tacrolimus metabolisers (fast, intermediate and slow). Patients were censored on the date of a re‐HT or death, or in case of initiation of renal replacement therapy (either dialysis or KT).

The primary aim was to examine the association between the tacrolimus metaboliser group and eGFR during the 5‐year follow‐up. Since the longitudinal data included repeated measurements of the dependent variable eGFR within individual patients, a linear mixed‐effects model analysis was performed. Random effects were included to model the intra‐patient correlation and individual variability over time for eGFR. Random intercepts (allowing each patient to have their own baseline eGFR) and random slopes (allowing the rate of change in eGFR over time to vary between patients) were tested. The appropriate random‐effects structure was selected through likelihood ratio tests. For the fixed‐effect adjustment (accounting for variables that might influence eGFR over time in a consistent way across all patients) three levels were considered: (1) univariable (only the main variable of interest was included, ie, the tacrolimus metaboliser group); (2) adjusted for the confounding variables patient's age at time of HT ² , ³ , ³³ , ³⁴ , ³⁵ , ³⁶ and patient's sex ² , ³ , ³³ , ³⁴ , ³⁵ ; (3) adjusted for relevant confounding variables that might affect eGFR post‐HT based on previous studies (eGFR pre‐HT, ² , ³ , ³³ , ³⁴ , ³⁵ , ³⁷ , ³⁸ , ³⁹ continuous veno‐venous hemofiltration [CVVH], ² diabetes mellitus, ³ , ³⁵ , ³⁷ , ³⁸ use of antihypertensive drugs, ³⁵ mechanical heart support pre‐HT ⁴⁰ and cold ischemic time ³ ) and adjusted for year of transplant and assay used for tacrolimus measurement. To avoid overfitting of the model, we adhered to the rule of thumb of one variable per 10 observations. ⁴¹

Secondary aims included: (1) To endorse the association between the tacrolimus metaboliser group and eGFR, other linear mixed‐effects models were fitted for eGFR, with (i) the individual median C ₀/D ratio as a continuous variable and (ii) the C ₀/D ratio at specific time points (M6 and 1Y post‐HT). (2) To simulate the concentration‐time curves for the metaboliser groups as an exploratory analysis to demonstrate the difference between the groups in overall exposure to tacrolimus. The simulation was performed using NONMEM software (version 7.5.1; FOCE+I, ICON Development Solutions) based on the final estimates of a previous developed population pharmacokinetic (popPK) model for tacrolimus in de novo HT recipients by Sikma. ⁴² Concentration‐time profiles were simulated at M6 post‐HT for each metaboliser group, for which the administered dose was the median dose for that group in our cohort at M6 post‐HT. (3) Finally, a survival analysis was carried out using the Kaplan‐Meier method. Differences in patient survival between the defined metaboliser groups were assessed using the log‐rank test.

R (version 4.3.2) was used for data management, statistical analysis and graphical diagnostics. Fitting of the linear mixed‐effects model was done using the nlme package (version 3.1‐153). ⁴³ Two‐sided P < 0.05 was considered statistically significant.

2.8. 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 PHARMACOLOGY, and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22. ⁴⁴ , ⁴⁵

3. RESULTS

3.1. Study population

Between January 2000 and April 2022, 412 patients were transplanted, of whom 209 were included (Supporting Information Figure S1). The median age at time of HT was 50 years [42, 58], and the majority of patients were male (n = 135, 64.6%). The most frequent cause of heart failure was non‐ischemic cardiomyopathy (n = 141, 67.5%), of which the most common specifications were dilated cardiomyopathy (n = 92, 66.2%) and hypertrophic cardiomyopathy (n = 23, 16.5%). In almost a quarter of our cohort, a left ventricular assist device was implanted before HT (n = 50, 23.9%), while an intra‐aortic balloon pump or extracorporeal membrane oxygenation was needed in 14 patients (6.7%) (Table 1).

TABLE 1.

Patient characteristics.

	Full cohort (n = 209)	Fast metabolisers (n = 71)	Intermediate metabolisers (n = 69)	Slow metabolisers (n = 69)	P
Recipients
Age (years)	50 [42, 58]	52 [43, 57]	50 [45, 59]	49 [40, 57]	0.37 a
Sex					<0.001 b
Male (n)	135 (64.6%)	40 (56.3%)	37 (53.6%)	58 (84.1%)
Female (n)	74 (35.4%)	31 (43.7%)	32 (46.4%)	11 (15.9%)
Height (cm)	174 [168, 182]	172 [168, 184]	174 [167, 180]	176 [168, 182]	0.52 a
Weight (kg)	74 [65, 85]	73 [65, 81]	73 [62, 85]	76 [66, 84]	0.54 a
BMI (kg/m2)	24.4 [21.6, 26.9]	24.4 [21.7, 26.2]	24.7 [21.2, 26.8]	24.3 [22.2, 28.0]	0.79 a
Heart failure aetiology					0.31 b
Non‐ischemic cardiomyopathy (n)	139 (66.5%)	52 (73.2%)	48 (69.6%)	39 (56.5%)
Dilated CMP (n)	92 (66.2%)	40 (76.9%)	29 (60.4%)	23 (59.0%)
Hypertrophic CMP (n)	23 (16.5%)	7 (13.5%)	10 (20.8%)	6 (15.4%)
Myocarditis (n)	7 (5.0%)	1 (1.9%)	4 (8.3%)	2 (5.1%)
Toxic CMP (n)	3 (2.2%)	NA	NA	3 (7.7%)
Restrictive CMP (n)	2 (1.4%)	2 (3.8%)	NA	NA
ARVC (n)	5 (3.6%)	1 (1.9%)	2 (4.2%)	2 (5.1%)
Non‐compaction CMP (n)	4 (2.9%)	NA	3 (6.3%)	1 (2.6%)
CMP of unknown cause (n)	3 (2.2%)	1 (1.9%)	NA	2 (5.1%)
Ischemic heart diseases (n)	57 (27.3%)	16 (22.5%)	16 (23.2%)	25 (36.2%)
Other c (n)	13 (6.2%)	3 (4.2%)	5 (7.2%)	5 (7.2%)
Use of mechanical heart support					0.45 b
Permanent support (LVAD) (n)	50 (23.9%)	18 (25.4%)	16 (23.2%)	16 (23.2%)
Temporary support (IABP or ECMO) (n)	14 (6.7%)	3 (4.2%)	7 (10.1%)	4 (5.8%)
Cold ischemic time (min)	190 [159, 215]	190 [159, 219]	189 [164, 212]	194 [159, 213]	0.98 a
Serum creatinine pre‐HT (μmol/L)	118 [99, 143]	123 [100, 145]	116 [98, 138]	118 [97, 137]	0.77 a
eGFR pre‐HT (mL/min/1.73 m2)	57 [43, 72]	51 [41, 72]	53 [43, 75]	62 [48, 71]	0.19 a
Diabetes pre‐HT (n)	27 (12.9%)	6 (8.5%)	11 (15.9%)	10 (14.5%)	0.37 b
Donors
Age (years)	46 [33, 55]	44 [30, 53]	47 [32, 54]	49 [39, 56]	0.15 a
Sex					1.00 b
Male (n)	84 (40.2%)	28 (39.4%)	28 (40.6%)	28 (40.6%)
Female (n)	125 (59.8%)	43 (60.6%)	41 (59.4%)	41 (59.4%)
Height (cm)	172 [167, 180]	172 [168, 180]	172 [167, 182]	172 [168, 180]	0.98 a
Weight (kg)	75 [65, 80]	75 [65, 82]	75 [65, 80]	75 [65, 80]	0.67 a
BMI (kg/m2)	24.5 [22.2, 26.9]	24.5 [22.5, 26.6]	23.9 [21.5, 26.4]	24.5 [22.5, 27.0]	0.39 a
Clinical course post heart transplantation
Left ventricular function
First year post‐HT	n = 203	n = 70	n = 68	n = 65	0.33 b
Good (n)	201 (99%)	68 (97.1%)	68 (100%)	65 (100%)
Moderate (n)	2 (1%)	2 (2.9%)	NA	NA
Poor (n)	NA	NA	NA	NA
Third year post‐HT	n = 183	n = 64	n = 60	n = 59	0.85 b
Good (n)	180 (98.4%)	63 (98.4%)	59 (98.3%)	58 (98.3%)
Moderate (n)	3 (1.6%)	1 (1.6%)	1 (1.7%)	1 (1.7%)
Poor (n)	NA	NA	NA	NA
Fifth year post‐HT	n = 151	n = 53	n = 48	n = 50	0.61 b
Good (n)	146 (96.7%)	52 (98.1%)	45 (93.8%)	49 (98.0%)
Moderate (n)	4 (2.6%)	1 (1.9%)	2 (4.2%)	1 (2.0%)
Poor (n)	1 (0.7%)	NA	1 (2.0%)	NA
Comorbidities
PTDM (n)	98 (46.9%)	32 (45.1%)	30 (43.5%)	36 (52.2%)	0.55 b
Hypertension (n)	138 (66.0%)	50 (70.4%)	45 (65.2%)	43 (62.3%)	0.61 b
CMV infection within first year post‐HT (n)	32 (15.3%)	8 (11.3%)	14 (20.3%)	10 (14.5%)	0.36 b
Number of acute cellular/humoral rejections					0.26 b
Within the first year post‐HT
0 (n)	99 (47.4%)	29 (40.8%)	32 (46.4%)	38 (55.1%)
1 (n)	61 (29.2%)	24 (33.8%)	19 (27.5%)	18 (26.1%)
2 (n)	37 (17.7%)	10 (14.1%)	15 (21.7%)	12 (17.4%)
3 (n)	10 (4.8%)	6 (8.5%)	3 (4.4%)	1 (1.5%)
4 (n)	2 (1.0%)	2 (2.8%)	NA	NA
During full follow‐up					0.64 b
0 (n)	92 (44.0%)	28 (39.4%)	29 (42%)	35 (50.7%)
1 (n)	60 (28.7%)	23 (32.4%)	19 (27.5%)	18 (26.1%)
2 (n)	39 (18.7%)	11 (15.5%)	16 (23.2%)	12 (17.4%)
3 (n)	11 (5.3%)	6 (8.5%)	3 (4.4%)	2 (2.9%)
4 (n)	5 (2.4%)	3 (4.2%)	1 (1.5%)	1 (1.5%)
5 (n)	1 (0.5%)	NA	1 (1.5%)	NA
8 (n)	1 (0.5%)	NA	NA	1 (1.5%)
Kidney replacement therapy
CVVH immediately post‐HT (n)	37 (17.7%)	11 (15.5%)	13 (18.8%)	13 (18.8%)	0.83 b
Haemodialysis (n)	3 (1.4%)	1 (1.4%)	1 (1.5%)	1 (1.5%)	1.00 b
Months post‐HT	76 [72‐121] d	72	76	121
Kidney transplantation (n)	1 (0.5%)	1 (1.4%)	0	0	NA
Months post‐HT	103	103	NA	NA

Note: Demographic and clinical characteristics in 209 heart transplant recipients and donors at time of heart transplantation. Data are shown as number of patients (percentages) for categorical variables, or median [25th, 75th percentile] for continuous variables. Significant P values are shown in bold.

Abbreviations: ARVC, arrhythmogenic right ventricular cardiomyopathy; BMI, body mass index; CMP, cardiomyopathy; CMV, cytomegalovirus; CVVH, continuous veno‐venous hemofiltration; ECMO, extracorporeal membrane oxygenation; eGFR, estimated glomerular filtration rate; HT, heart transplantation; IABP, intra‐aortic balloon pump; LVAD, left ventricular assist device; NA, not applicable; PTDM, post‐transplant diabetes mellitus.

^a P value is from Kruskal‐Wallis test.

^b P value is from Fisher's exact test.

^c Other causes of heart failure: heart valve disease, congenital heart defects, dissection and long QT syndrome type 3.

^d Presented as median and range since this variable consists of only three patients.

3.2. Tacrolimus metaboliser groups

A total of 955 tacrolimus C ₀ values were included, with a minimum of two measurements per patient. For most of the patients, five samples were available (n = 151, 72%). The distribution of the individual median C ₀/D ratios is displayed in Figure 1 (minimum 0.41, 33rd percentile 1.53, 67th percentile 2.27, maximum 8.9). All patients were categorized in three groups, fast (n = 71 patients), intermediate (n = 69 patients) and slow (n = 69) metabolisers, for which the median C ₀/D ratios were ≤1.53, >1.53 and ≤2.27, and >2.27 ng/mL/mg, respectively. We did not identify an effect of the different assays used for tacrolimus measurement in this study on the tacrolimus C ₀/D ratio (Supporting Information Table S1).

FIGURE 1.

Distribution of the individual median tacrolimus C ₀/D ratio as calculated over full follow‐up in all patients. The distribution was as follows: minimum 0.41 ng/mL/mg, 33rd percentile 1.53 ng/mL/mg, 67th percentile 2.27 ng/mL/mg and maximum 8.9 ng/mL/mg. The dashed lines indicate the 33rd and 67th percentiles. C ₀/D ratio, concentration‐to‐dose ratio.

3.3. Patient characteristics according to metaboliser groups

The groups were similar at baseline, except for a significant difference between the groups regarding the sex of the recipient. No significant difference was observed in the clinical outcomes after HT between the groups (Table 1). Patients were comparable regarding their left ventricular function at Y1, Y3 and Y5 post‐HT. All other patient and donor characteristics are described in Table 1.

3.4. Immunosuppressive medication

The tacrolimus C ₀ (ng/mL) was significantly different between the groups at M3, Y1 and Y5 (Table 2). The median tacrolimus C ₀ in all three groups fell within the target tacrolimus C ₀ range for that period. The daily tacrolimus dose was significantly different between the groups for each period. For example, at M3 post‐HT, the median daily tacrolimus dose was 12.0 mg [9.0‐15.5], 6.0 mg [5.0‐8.0], and 4.3 mg [4.0‐6.0], for fast, intermediate and slow metabolisers, respectively. There were no differences in the daily dose of the concomitant immunosuppressant drugs prednisolone, MMF and everolimus, except for the MMF dose at M3, Y1, Y3 and Y5.

TABLE 2.

Immunosuppressive medication after heart transplantation.

	Fast metabolisers (n = 71)	Intermediate metabolisers (n = 69)	Slow metabolisers (n = 69)	P
Tacrolimus C 0 (ng/mL)
3 months	10.5 [8.9, 12.0] (n = 71)	11.3 [10.0, 13.8] (n = 69)	12.1 [10.3, 13.4] (n = 69)	0.004
6 months	10.4 [9.0, 12.0] (n = 71)	10.9 [9.5, 13.0] (n = 69)	12.4 [8.6, 14.7] (n = 69)	0.063
1 year	8.3 [6.9, 9.4] (n = 70)	8.6 [6.9, 9.8] (n = 68)	9.7 [7.9, 11.8] (n = 65)	0.009
3 years	5.9 [4.9, 7.7] (n = 64)	6.7 [5.8, 8.3] (n = 60)	6.9 [5.8, 8.6] (n = 59)	0.066
5 years	5.9 [5.0, 7.1] (n = 53)	6.4 [4.6, 7.8] (n = 48)	7.1 [5.8, 8.0] (n = 50)	0.036
Tacrolimus daily dose (mg)
3 months	12.0 [9.0, 15.5] (n = 71)	6.0 [5.0, 8.0] (n = 69)	4.3 [4.0, 6.0] (n = 69)	<0.001
6 months	10.0 [8.0, 14.0] (n = 71)	6.0 [5.0, 7.0] (n = 69)	4.0 [3.0, 5.0] (n = 69)	<0.001
1 year	8.5 [6.0, 10.0] (n = 70)	5.0 [4.0, 5.0] (n = 68)	3.0 [3.0, 4.0] (n = 65)	<0.001
3 years	5.0 [4.0, 7.0] (n = 64)	3.0 [3.0, 4.0] (n = 60)	2.0 [2.0, 3.0] (n = 59)	<0.001
5 years	4.0 [4.0, 6.0] (n = 53)	3.0 [2.8, 4.0] (n = 48)	2.0 [2.0, 3.0] (n = 50)	<0.001
MMF daily dose (mg)
3 months	1500 [1000, 2000] (n = 54)	1000 [1000, 1500] (n = 51)	1500 [1000, 1500] (n = 55)	0.004
6 months	1000 [1000, 1500] (n = 50)	1000 [500, 1500] (n = 47)	1500 [1000, 1500] (n = 51)	0.06
1 year	1000 [1000, 1500] (n = 39)	1000 [500, 1500] (n = 40)	1500 [1000, 1500] (n = 46)	0.02
3 years	1000 [1000, 1000] (n = 25)	1000 [500, 1000] (n = 29)	1000 [1000, 1500] (n = 35)	0.02
5 years	1000 [1000, 1000] (n = 19)	1000 [500, 1000] (n = 18)	1000 [1000, 1500] (n = 27)	0.007
Prednisolone daily dose (mg)
3 months	10.0 [10.0, 10.0] (n = 70)	10.0 [10.0, 10.0] (n = 69)	10.0 [10.0, 10.0] (n = 69)	0.92
6 months	10.0 [10.0, 10.0] (n = 71)	10.0 [10.0, 10.0] (n = 69)	10.0 [10.0, 10.0] (n = 69)	0.83
1 year	10.0 [5.0, 10.0] (n = 64)	10.0 [7.5, 10.0] (n = 62)	7.5 [5.0, 10.0] (n = 61)	0.60
3 years	7.5 [7.5, 10.0] (n = 50)	10.0 [5.0, 10.0] (n = 47)	7.5 [5.0, 10.0] (n = 41)	0.06
5 years	7.5 [5.0, 10.0] (n = 41)	10.0 [7.5, 10.0] (n = 37)	5.0 [5.0, 10.0] (n = 30)	0.09
Everolimus daily dose (mg)
3 months	1.5 [1.5, 2.3] (n = 3)	1.5 [1.5, 1.5] (n = 2)	1.8 [1.6, 1.9] (n = 2)	0.62
6 months	1.5 [1.5, 3.0] (n = 5)	2.0 [1.3, 2.8] (n = 3)	1.5 [1.5, 1.5] (n = 1)	0.82
1 year	1.5 [1.5, 1.5] (n = 10)	1.5 [1.5, 1.6] (n = 4)	1.5 [1.5, 1.5] (n = 1)	0.86
3 years	1.5 [1.5, 2.0] (n = 14)	1.5 [1.0, 2.0] (n = 8)	1.5 [1.3, 1.8] (n = 7)	0.79
5 years	1.5 [1.5, 2.0] (n = 13)	1.5 [1.5, 2.0] (n = 8)	1.3 [0.6, 1.5] (n = 10)	0.13

Note: Immunosuppressive medication during follow‐up. Data are shown as median [25th, 75th percentile], along with the number of patients at the time point. P values are from the Kruskal‐Wallis test, significant P values are shown in bold.

Abbreviations: C ₀, pre‐dose concentration; MMF, mycophenolate mofetil; M3, month 3 post‐transplantation; M6, month 6 post‐transplantation; Y1, year 1 post‐transplantation; Y3, year 3 post‐transplantation; Y5, year 5 post‐transplantation.

3.5. Kidney function

Patients were comparable at baseline regarding their eGFR (Table 1). After HT, the eGFR between the tacrolimus metaboliser groups was significantly different for all time points (Supporting Information Table S2). The change in eGFR for each group over time is displayed in Figure 2. For all patients, the eGFR at 5Y post‐HT (61 mL/min/1.73 m², interquartile range [IQR] 49, 74) was not significantly different from pre‐HT (57 mL/min/1.73 m², IQR 43, 72, P = 0.5). Regarding proteinuria, no differences were observed between the groups during follow‐up (Supporting Information Table S2). Some patients received CVVH in the immediate postoperative phase, but no significant difference in the incidence of CVVH was observed between the groups. After the follow‐up period of this study, three patients started chronic intermittent haemodialysis and one patient received a KT approximately 8.5 years after HT.

FIGURE 2.

Boxplot of the kidney function over time. Boxplot illustrating the estimated glomerular filtration rate during full follow‐up, separated by the tacrolimus metabolism (C₀/D ratio) group. Asterix above the plot indicate the significance level between the groups at that time‐point. C ₀/D ratio, concentration‐to‐dose ratio; ns, non‐significant. *P value <0.05, **P value <0.01, ***P value <0.001. HT, heart transplantation.

3.6. Influences on kidney function

A linear mixed‐effects model was fitted, in which the slow metabolisers were the reference group. The random‐effects structure including random intercepts and slopes for the time after HT fit the data best (Supporting Information Table S3).

Univariable analysis yielded a significantly lower mean eGFR for both fast metabolisers (−9.9 mL/min/1.73 m², 95% CI −15.5, −4.3, P = 0.0006) and intermediate metabolisers (−8.2 mL/min/1.73 m², 95% CI −13.8, −2.5, P = 0.005) compared to slow metabolisers. The size of effect for time after HT was small but had a significant effect on the mean eGFR (0.1 mL/min/1.73 m², 95% CI 0.1, 0.2, P < 0.0001) (Table 3, model 1). Multivariable analysis demonstrated a significantly lower mean eGFR only for fast metabolisers when adjusted for recipient's age and sex (−6.6 mL/min/1.73 m², 95% CI −11.4, −1.9, P = 0.006) compared to slow metabolisers. Time after HT, recipient's age and sex all had a significant effect on the mean eGFR (Table 3, model 2). The final model, including the previous variables and adjusted for the relevant variables eGFR pre‐HT, CVVH, diabetes mellitus, use of antihypertensive drugs, mechanical heart support pre‐HT, cold ischemic time, tacrolimus assay and year of transplantation in the fixed‐effects part, yielded a significantly lower eGFR for fast metabolisers, as well (−6.8 mL/min/1.73 m², 95% CI −11.2, −2.4, P = 0.002) compared to slow metabolisers. The size of effect and P values of all confounders are presented in Table 3, model 3. The β coefficient estimates (including their 95% CIs) for the metaboliser groups of all three models are displayed in Figure 3.

TABLE 3.

Linear mixed‐effects models for kidney function (eGFR) over 5 years and tacrolimus metaboliser groups.

Parameter	β	95% CI	P	Model fit
Univariable analysis
1Adjusted for time after transplantation
Intercept	59.9	[55.8, 64.0]	<0.0001	AIC: 7626 BIC: 7664 logLik: −3805
Slow metabolisers	Ref
Intermediate metabolisers	−8.2	[−13.8, −2.5]	0.005
Fast metabolisers	−9.9	[−15.5, −4.3]	0.0006
Time after HT (per month)	0.1	[0.1, 0.2]	<0.0001
Multivariable analysis
2Adjusted for recipient's age and sex
Intercept	98.4	[89.9, 106.9]	<0.0001	AIC: 7546 BIC: 7594 logLik: −3763
Slow metabolisers	Ref
Intermediate metabolisers	−3.3	[−8.1, 1.5]	0.17
Fast metabolisers	−6.6	[−11.4, −1.9]	0.006
Time after HT (per month)	0.2	[0.1, 0.2]	<0.0001
Age at HT (per year)	−0.8	[−1.0, −0.6]	<0.0001
Sex (female)	−8.4	[−12.5, −4.3]	0.0001
3Adjusted for the variables of model 2 and the relevant variables (eGFR pre‐HT, CVVH, DM, use of antihypertensive drugs, mechanical heart support pre‐HT, cold ischemic time, tacrolimus, and year of HT)
Intercept	82.5	[68.0, 96.8]	<0.0001	AIC: 7492 BIC: 7580 logLik: −3728
Slow metabolisers	Ref
Intermediate metabolisers	−3.2	[−7.7, 1.2]	0.15
Fast metabolisers	−6.8	[−11.2, −2.4]	0.002
Time after HT (per month)	0.2	[0.1, 0.2]	<0.0001
Age at HT (per year)	−0.5	[−0.7, −0.4]	<0.0001
Sex (female)	−7.7	[−11.6, −3.8]	0.0001
eGFR pre‐HT (per mL/min/1.73 m2)	0.2	[0.1, 0.3]	<0.0001
CVVH (yes)	−11.0	[−15.9, −6.1]	<0.0001
Diabetes mellitus (yes)	−0.9	[−4.5, 2.8]	0.64
Antihypertensive drugs (yes)	−3.3	[−5.1, −1.5]	0.0003
Mechanical heart support (yes)	−2.2	[−6.3, 2.0]	0.30
Cold ischemic time (per minute)	−0.02	[−0.1, 0.02]	0.33
Tacrolimus assay (immunoassay)	−0.9	[−4.0, 2.1]	0.55
Year of HT (2015 and onwards)	−2.4	[−7.1, 2.2]	0.30

Note: Significant P values for the metaboliser groups are shown in bold.

Abbreviations: β, beta coefficient; AIC, Akaike information criterion; BIC, Bayesian information criterion; CI, confidence interval; CVVH, continuous veno‐venous hemofiltration; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; HT, heart transplantation; logLik, log likelihood.

FIGURE 3.

Caterpillar plot of the parameter estimates for the linear mixed effects models. Caterpillar plot illustrating the differences in kidney function compared to the reference group (slow metabolisers) using linear mixed‐effects model analysis with three levels of adjustment. Coefficient estimates, 95% CIs and P values are presented. CI, confidence interval; CVVH, continuous veno‐venous hemofiltration; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; HT, heart transplantation; ns, non‐significant.

To endorse the identified association, other linear mixed‐effects models were fitted for eGFR, with (1) the individual median C ₀/D ratio as a continuous variable and (2) with the C ₀/D ratio at specific time points. A significantly higher eGFR per higher individual median C ₀/D ratio was identified when adjusted stepwise for the same variables as above (2.8 mL/min/1.73 m² higher per 1 unit increase in C ₀/D ratio, 95% CI 1.0, 4.5, P = 0.002) (Table 4, model 1c). The C ₀/D ratio at a single time point (6 M and 1Y post‐HT) demonstrated a significantly higher eGFR per higher C ₀/D ratio as well when adjusted for the same variables as the final model: 1.5 mL/min/1.73 m² higher per 1 unit increase in C ₀/D ratio (95% CI 0.1, 2.9, P = 0.03) and 1.6 mL/min/1.73 m² higher per 1 unit increase in C ₀/D ratio (95% CI 0.2, 3.0, P = 0.03) for 6 M and 1Y post‐HT, respectively (Table 4, model 2a‐b).

TABLE 4.

Linear mixed‐effects models for kidney function (eGFR) over 5 years and the tacrolimus C ₀/D ratio as a continuous variable.

Parameter	β	95% CI	P	Model fit
1Individual median tacrolimus C 0/D ratio
Univariable analysis
aAdjusted for time after transplantation
Intercept	45.3	[40.2, 50.5]	<0.0001	AIC: 7629 BIC 7663 logLik: −3808
Median C 0/D ratio (per ng/mL/mg)	4.3	[2.0, 6.5]	<0.0001
Time after HT (per month)	0.1	[0.1, 0.2]	<0.0001
Multivariable analysis
bAdjusted for recipient's age and sex
Intercept	88.8	[79.5, 98.0]	<0.0001	AIC: 7546 BIC: 7589 logLik: −3764
Median C 0/D ratio (per ng/mL/mg)	3.2	[1.3, 5.1]	0.001
Time after HT (per month)	0.2	[0.1, 0.2]	<0.0001
Age at HT (per year)	−0.8	[−1.0, −0.6]	<0.0001
Sex (female)	−8.5	[−12.4, −4.5]	<0.0001
cAdjusted for the variables of model 2 and the relevant variables (eGFR pre‐HT, CVVH, DM, use of antihypertensive drugs, mechanical heart support pre‐HT, cold ischemic time, tacrolimus, and year of HT)
Intercept	74.8	[59.3, 87.0]	<0.0001	AIC: 7495 BIC: 7578 logLik: −3731
Median C 0/D ratio (per ng/mL/mg)	2.8	[1.0, 4.5]	0.002
Time after HT (per month)	0.2	[0.1, 0.2]	<0.0001
Age at HT (per year)	−0.5	[−0.7, −0.4]	<0.0001
Sex (female)	−8.1	[−11.9, −4.2]	<0.0001
eGFR pre‐HT (per mL/min/1.73 m2)	0.2	[0.1, 0.3]	<0.0001
CVVH (yes)	−10.8	[−15.7, −6.0]	<0.0001
Diabetes mellitus (yes)	−0.8	[−4.4, 2.8]	0.66
Antihypertensive drugs (yes)	−3.3	[−5.1, −1.5]	0.0003
Mechanical heart support (yes)	−1.9	[−5.9, 2.2]	0.37
Cold ischemic time (per minute)	−0.01	[−0.1, 0.01]	0.35
Tacrolimus assay (immunoassay)	−1.0	[−4.1, 2.1]	0.52
Year of HT (2015 and onwards)	−2.2	[−6.8, 2.4]	0.34
2Tacrolimus C 0/D ratio at a specific time point
C 0/D ratio (6 months post‐HT)
Intercept	76.9	[62.5, 91.4]	<0.0001	AIC: 7501 BIC: 7583 logLik: −3733
C 0/D ratio at 6 M (per ng/mL/mg)	1.5	[0.1, 2.9]	0.03
Time after HT (per month)	0.2	[0.1, 0.2]	<0.0001
Age at HT (per year)	−0.5	[−0.7, −0.4]	<0.0001
Sex (female)	−8.9	[−12.7, −5.1]	<0.0001
eGFR pre‐HT (per mL/min/1.73 m2)	0.2	[0.1, 0.3]	<0.0001
CVVH (yes)	−11.5	[−16.4, −6.5]	<0.0001
Diabetes mellitus (yes)	−0.9	[−4.6, 2.8]	0.63
Antihypertensive drugs (yes)	−3.3	[−5.1, −1.5]	0.0004
Mechanical heart support (yes)	−2.2	[−6.3, 2.0]	0.30
Cold ischemic time (per minute)	−0.02	[−0.1, 0.02]	0.32
Tacrolimus assay (immunoassay)	−1.0	[−4.1, 2.0]	0.51
Year of HT (2015 and onwards)	−1.9	[−6.6, 2.7]	0.41
b C 0/D ratio (1 year post‐HT)
Intercept	77.6	[63.2, 91.9]	<0.0001	AIC: 7392 BIC: 7474 logLik: −3679
C 0/D ratio at 1Y (per ng/mL/mg)	1.6	[0.2, 3.0]	0.03
Time after HT (per month)	0.2	[0.1, 0.2]	<0.0001
Age at HT (per year)	−0.6	[−0.7, −0.4]	<0.0001
Sex (female)	−8.5	[−12.3, −4.7]	<0.0001
eGFR pre‐HT (per mL/min/1.73 m2)	0.2	[0.1, 0.3]	<0.0001
CVVH (yes)	−8.5	[−13.6, −3.5]	0.001
Diabetes mellitus (yes)	−0.4	[−4.0, 3.3]	0.83
Antihypertensive drugs (yes)	−3.4	[−5.3, −1.6]	0.0002
Mechanical heart support (yes)	−2.1	[−6.2, 2.1]	0.33
Cold ischemic time (per minute)	−0.01	[−0.05, 0.03]	0.55
Tacrolimus assay (immunoassay)	−1.0	[−4.1, 2.0]	0.51
Year of HT (2015 and onwards)	−1.7	[−6.4, 2.9]	0.46

Note: Significant P values for the tacrolimus C ₀/D ratio are shown in bold. The models with the tacrolimus C ₀/D ratio at a specific time point as predictor include all the following variables in the fixed‐effects part: time after HT, age at HT, sex, eGFR pre‐HT, CVVH, DM, use of antihypertensive drugs, mechanical heart support pre‐HT, cold ischemic time, tacrolimus and year of HT.

Abbreviations: β, beta coefficient; AIC, Akaike information criterion; BIC, Bayesian information criterion; C ₀/D ratio, concentration‐to‐dose ratio; CI, confidence interval; CVVH, continuous veno‐venous hemofiltration; eGFR, estimated glomerular filtration rate; HT, heart transplantation; logLik, log likelihood.

3.7. Simulation of tacrolimus exposure

The simulated concentration‐time curves for our tacrolimus metaboliser groups at 6 M post‐HT display the highest C _max and higher overall tacrolimus exposure for fast metabolisers (Supporting Information Figure S2).

3.8. Survival analysis

Finally, the mortality in our cohort was assessed. A total of 13 patients died during follow‐up: three, four and six patients among the fast, intermediate and slow metabolisers, respectively. The 5‐year survival rates were 0.96, 0.94 and 0.91 for fast, intermediate and slow metabolisers, respectively. No difference in survival rate between the metaboliser groups was observed (log‐rank test, P = 0.6).

4. DISCUSSION

This study demonstrates that a fast tacrolimus metabolism in HT recipients, estimated using a C ₀/D ratio ≤1.53 ng/mL/mg, is significantly associated with worse kidney function in the first 5 years after heart transplantation when compared to patients with intermediate and slow tacrolimus metabolisms. This is the first study to assess the association between the tacrolimus C ₀/D ratio and kidney function in HT recipients. The identified difference of 6.8 mL/min/1.73 m² in mean eGFR is important since impaired kidney function was repeatedly associated with poor survival. ⁴⁶ , ⁴⁷ An eGFR <60 mL/min/1.73 m² at 1Y post‐HT was an independent predictor of all‐cause mortality with an adjusted hazard ratio of 1.7. ⁴⁷ Importantly, the mean eGFR for fast metabolisers in our cohort was <60 mL/min/1.73 m² at each time point.

The association between the tacrolimus C ₀/D ratio and kidney function aligns with previous research, ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ although others did not find an association. ⁴⁸ , ⁴⁹ We adjusted for multiple confounders affecting kidney function post‐HT. Unlike prior studies, we grouped HT recipients based on their median C ₀/D ratio over 5 years, whereas others used the mean C ₀/D ratio in the first 6 months ¹⁵ or year ⁴⁹ post‐transplantation, or the C ₀/D ratio at 1, ⁴⁸ 3 ¹⁶ , ⁴⁸ or 6 months ¹⁷ post‐transplantation. We did not assess the C ₀/D ratio in the immediate post‐HT phase since tacrolimus' pharmacokinetics may differ greatly due to potential alterations in absorption caused by gastroparesis ⁵⁰ and significant hemodynamic changes within the first 6 months post‐HT. ⁵¹ Assessment of kidney function in the early post‐transplantation phase thus may not provide reliable information on CKD. Since we did demonstrate a correlation between both an individual's median C ₀/D ratio and an individual's C ₀/D ratio at 6 M and 1Y post‐HT and kidney function over 5 years, we believe this solidifies the identified association. Another difference is our cut‐off value for the C ₀/D ratio for fast metabolisers, which is higher than the cut‐off of ≤1.05 ng/mL/mg ¹⁵ , ¹⁶ , ¹⁸ , ¹⁹ , ⁴⁸ or ≤1.14 ng/mL/mg ⁴⁹ used in kidney, and ≤1.09 ng/mL/mg in liver transplant recipients. ¹⁷

Explanations for the worse kidney function in the fast metabolisers in our cohort are a higher C _max or higher area under the curve (AUC) because of the necessary higher daily tacrolimus dose. Alternatively, the necessary higher daily tacrolimus dose in fast metabolisers could be related to drug absorption instead of a fast metabolism. Possibly, these patients have a lower bioavailability of tacrolimus (high first‐pass effect), which prevents the drug from entering the bloodstream, ⁹ resulting in higher required tacrolimus doses to achieve the target C ₀. However, in case of a lower bioavailability, the C _max or AUC would not be increased and it would not be related to nephrotoxicity.

Interestingly, kidney function improved over time for all metaboliser groups. In contrast, Taiwo demonstrated a decline in kidney function over 5 years post‐HT in n = 230 HT recipients at a rate of 2.9 (95% CI 2.8‐3.0) mL/min/1.73m²/year in the normal/near‐normal kidney function pre‐HT group (eGFR ≥45 mL/min/1.73 m²) and 2.2 (95% CI 1.8‐2.6) mL/min/1.73m²/year in the impaired pre‐HT group (eGFR <45 mL/min/1.73 m2) (ns). ⁵² In another study, kidney function declined to 88% of pre‐HT measured GFR at the first year to 77% at 5 years and 38% at 20 years post‐HT. ⁵³ A study including n = 160 HT recipients demonstrated a higher eGFR for CYP3A5 expressers over 9 years post‐HT than non‐expressers. ⁵⁴ This study thus contradicts our findings that fast metabolisers receiving higher tacrolimus doses have poorer eGFR. However, the majority of patients in this particular study used cyclosporin A (78%) rather than tacrolimus. ⁵⁴ The increase in eGFR in our study could be influenced by the drop‐out of deceased patients. Additionally, as tacrolimus‐induced nephrotoxicity became better understood, ⁴ , ⁵ our standard care changed, resulting in a more nephroprotective treatment by treating hypertension and/or proteinuria in HT recipients more diligently. ⁷ , ⁵⁵ Finally, screening for post‐transplant diabetes mellitus has been implemented because diabetes mellitus is a major contributor to CKD. ⁷ , ⁵⁶ , ⁵⁷

Recipient's sex was significantly different between the metaboliser groups, with the majority of females being fast or intermediate metabolisers and the majority of males being slow metabolisers. This may be explained by the potential influence of oestrogen activity on CYP enzyme activity, ⁵⁸ with reported increased CYP3A5 activity and subsequently higher clearance of tacrolimus. ⁵⁹ Furthermore, oestrogen could induce P‐glycoprotein activity (encoded by ABCB1 ), ⁶⁰ with a potential lower bioavailability as result. Overall, fewer females receive a HT. ²⁴

No difference in 5‐year survival rate between the metaboliser groups was observed. Studies including KT recipients reported a lower 5‐year survival rate for non‐rapid metabolisers ⁴⁹ or for fast metabolisers. ¹⁶ These contradicting results are, however, from studies with KT recipients, not HT recipients, and our population was smaller compared to these studies (n = 209 in our cohort vs n = 1254 and n = 401, respectively) with fewer deaths.

Fast metabolisers suspected of having tacrolimus‐induced toxicity could benefit from several therapeutic interventions. First, fast metabolisers could be converted to an extended‐release tacrolimus formulation with a lower C _max (LCP‐tacrolimus). This was beneficial in KT recipients, where fast metabolisers demonstrated a significant recovery of mean eGFR already 1 month after conversion (48.5 ± 17.6 vs 41.5 ± 17.0 mL/min/1.73 m², P = 0.03). ⁶¹ Second, targeting lower tacrolimus exposure by the addition of everolimus. ⁶² , ⁶³ , ⁶⁴ This does not change their C ₀/D ratio but they might be exposed to a lower C _max. However, everolimus is also nephrotoxic and this risk should be taken into account. Third, AUC‐guided dosing might be a strategy for fast metabolisers. Measurement of AUC is a better estimate of total tacrolimus exposure and may detect overexposure not apparent from C ₀ alone. ¹⁴ This would potentially guarantee similar tacrolimus exposure for fast, intermediate and slow metabolisers.

Our study has limitations. First, it is important to consider that the tacrolimus C ₀/D ratio is an estimate of a recipients' tacrolimus metabolism. Although we corrected for certain variables influencing tacrolimus metabolism, ¹⁰ information on CYP3A5 SNPs was not available. However, the C ₀/D ratio is a direct pharmacokinetic (PK) phenotype marker, whilst information on SNPs is “only” a genotype marker. Furthermore, the C ₀/D ratio may not only serve as a surrogate marker for CYP3A5 metabolism (in the absence of formal genetic testing, which many centres lack capability for), but might be a better, more relevant and clinically more useful marker. Second, approximately half of all identified HT recipients were excluded, resulting in a small sample size. Third, only C ₀ was available due to the retrospective aspect of our study. We did, however, provide simulated concentration‐time curves, demonstrating a higher exposure to tacrolimus for fast metabolisers. Fourth, we could not control for the increased awareness over the years of the nephrotoxic effects of tacrolimus, but including the year of transplantation as a confounder did not yield a significant result. Fifth, we lacked data on drug‐drug interactions and exposure to other nephrotoxic drugs. Sixth, different assays for the measurement of tacrolimus were used, but no significant difference in the tacrolimus C ₀/D ratio between the three assays was identified. Lastly, no external validation was performed.

The results from this study are important for clinical care because conventional interventions for limiting the deterioration of the kidney function, such as hypertension management, are limited in efficacy and no curative treatment is available. ⁷ The identification of the tacrolimus C ₀/D ratio as a potential marker, with a biologically plausible mechanism of action, presents an opportunity for improved tacrolimus dosing in patients with a low ratio. This calls for further research in multi‐centre larger cohorts, preferably using either an intensive sampling (traditional, rich PK studies) or a popPK approach (sparse sampling and compartmental PKs). If comparable results are obtained, prospective studies should be devised to explore the possible nephroprotective effect of LCP‐tacrolimus and/or the opportunity for AUC‐guided tacrolimus dosing for fast metabolisers, based on either a limited‐sampling strategy (relying on three or four blood samples instead of a full PK profile) or a popPK model (taking into account the metaboliser status). Such findings would formulate potential adjustments to the clinical care of HT recipients receiving tacrolimus.

5. CONCLUSION

The tacrolimus C ₀/D ratio is associated with kidney function in HT recipients in the first 5 years post‐transplantation. Identification of fast metabolisers at risk for worse kidney function, defined by a C ₀/D ratio ≤1.53 ng/mL/mg, is important since ESKD remains a severe and frequent complication post‐transplantation.

AUTHOR CONTRIBUTIONS

Participated in research design: D.A.H., B.C.M.dW. and O.M. Participated in data collection: B.C.A.S., M.R.S. and O.M. Participated in data analysis: M.R.S., T.B.P. and B.C.A.S. Writing of the first draft of the manuscript: M.R.S. Critically reviewed manuscript and approved final version: T.B.P., B.C.A.S., J.J.B., K.C., A.A.C., I.K., B.C.M.dW., D.A.H. and O.M.

CONFLICT OF INTEREST STATEMENT

J.J.B. reports speaker engagement or advisory board fees from Astra Zeneca, Abbott, Bayer, Danchii Sankyo, Novartis and Vifor. All are outside the scope of this work. D.A.H. has received lecture fees and consulting fees from Astellas Pharma, Astra Zeneca, Chiesi Pharma, Medincell, Novartis Pharma, Sangamo Therapeutics and Vifor Pharma. He has received grant support from Astellas Pharma, Bristol‐Myers Squibb and Chiesi Pharma (paid to his institution). D.A.H. does not have employment or stock ownership at any of these companies, and neither does he have patents or patent applications. O.M. reports speaker engagement or advisory board fees from Astra Zeneca, Abbott, Boehringer Ingelheim, Daiichi Sankyo, Novartis, Novo Nordisk, Siemens and Vifor. All are outside the scope of this work. The other authors declare no conflicts of interest.

Supporting information

SUPPORTING INFORMATION TABLE S1 Effect of different assays on the tacrolimus C ₀/D ratio.

SUPPORTING INFORMATION TABLE S2 Kidney function after heart transplantation.

SUPPORTING INFORMATION TABLE S3 Comparison of the univariable linear mixed‐effects models for kidney function.

SUPPORTING INFORMATION FIGURE S1 Patient inclusion flow chart.

SUPPORTING INFORMATION FIGURE S2 Simulated tacrolimus concentration‐time curves for the metaboliser groups at 6 months post‐HT.

Schagen MR, Petersen TB, Seijkens BCA, et al. The tacrolimus concentration‐to‐dose ratio is associated with kidney function in heart transplant recipients. Br J Clin Pharmacol. 2025;91(8):2363‐2377. doi: 10.1002/bcp.70041

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.