Allogeneic hematopoietic transplantation is frequently the only curative standard therapeutic method for many hematological diseases. For a successful outcome after transplantation, it is essential to use immunosuppression, usually for a limited period after the procedure, to allow new immunocompetent cells to develop immune tolerance and to prevent the onset of graft-versus-host disease (GVHD). A calcineurin inhibitor, in combination with low-dose methotrexate or mycophenolate mofetil (MMP), is used in the majority of European centers, and usually, T-cell depletion is performed (The EBMT Handbook, 2024). Cyclosporine and tacrolimus, as the two calcineurin inhibitors, are accepted and are practically interchangeable in this indication (Nash et al., 2005; Ratanatharathorn et al., 1998).

Tacrolimus is increasingly being used over cyclosporine, largely due to its more predictable pharmacokinetics (Ratanatharathorn et al., 1998; Nash et al., 2000). However, it has a narrow therapeutic window, and monitoring of whole blood concentrations is necessary after transplantation. Despite its widespread use, the recommended blood level for patients after allogeneic hematopoietic transplantation is not precisely defined as the target concentration has been extrapolated from solid organ transplantations, with the range of 5–20 ng/mL (Torlen et al., 2016; Mori et al., 2012; Sharma et al., 2021). During the first year after transplantation, immune tolerance usually develops, with immune reconstitution through the thymic pathway, as opposed to solid organ transplantation. Therefore, the immunosuppressive strategy is quite different from solid organ transplantation (van den Brink et al., 2015).

Therapeutic blood levels of calcineurin inhibitors are maintained until the start of drug withdrawal, usually at approximately day 100 after transplantation (Sharma et al., 2021; Ganetsky et al., 2016). Some studies suggest the possibility of earlier dose reduction, which may lead to better disease control, especially in high-risk disease (Abraham et al., 1997; Schmid et al., 2006; Offer et al., 2015). The association between low levels of calcineurin inhibitors and the occurrence of acute graft-versus-host disease (aGVHD) in the early phase (first month) after transplantation is widely described (Mori et al., 2012; Sharma et al., 2021; Ganetsky et al., 2016; Offer et al., 2015; Hagen et al., 2019; Malard et al., 2010; Bianchi et al., 2019). However, only a single study describes prolonged tacrolimus monitoring (up to 6 weeks) and the association between its low levels, that is, levels below 10 ng/mL, and aGVHD (Tian et al., 2019). Here, aGVHD is associated with low levels of calcineurin inhibitors, and optimal values from these studies seem to be above 10 ng/mL. On the other hand, high calcineurin inhibitor levels may lead to inappropriate immunosuppression that may cause disease relapse or opportunistic infections; however, few studies associate higher levels of calcineurin inhibitors with lower overall survival or risk of relapse (Wong et al., 2005; Craddock et al., 2010; Bacigalupo et al., 1991; Braidotti et al., 2024). The greatest mortality after transplantation for malignant diseases is caused by relapse of the disease, which is thus the most important predictor for overall survival (Schmalter et al., 2024). Again, these studies only describe a short period of drug levels monitoring after stem cell transplantation, with a maximum of 21 days, as described by Craddock et al. (2010). Therefore, it is not possible to reliably extrapolate optimal therapeutic calcineurin inhibitor levels in the later phase following transplantation needed to achieve the best clinical outcomes.

The primary aim of this study was to investigate the relationship between both short-term and long-term tacrolimus exposure and overall survival in the context of other clinical covariates possibly affecting patients’ overall survival after allogeneic stem cell transplantation. The secondary objective was to develop a tacrolimus population pharmacokinetic model in order to propose model-based individualization of tacrolimus dosing.

The period during which tacrolimus levels were analyzed in this study and related to survival and clinical data is exceptionally long. Most previous analyses of tacrolimus levels aimed to prove a connection between the drug levels and the occurrence of aGVHD after an allogeneic stem cell transplant. These studies showed similar results for the minimum tacrolimus whole blood trough concentration needed to avoid an increased risk of aGVHD. In the study by Hagen et al., the optimal trough levels of tacrolimus were 10–15 ng/mL for the first 4 weeks (aGVHD grade II–IV was present in 26.5% of patients), and according to the study by Sharma el al., tacrolimus levels should be ≥10.15 ng/mL during the first week (aGVHD grade II–IV was present in 40% of patients) for lower risk of aGVHD (Sharma et al., 2021; Hagen et al., 2019). Ganetsky el al. reported that first-week levels should be >12 ng/mL (aGVHD grade II–IV of 42.8%), and Offer et al. found that in children, levels should be above 10 ng/mL during the third week (aGVHD grade II–IV 33%) (Ganetsky et al., 2016; Offer et al., 2015). Yoshikawa et al. suggested that prolonged exposure to tacrolimus blood levels ≥10 ng/mL early post-transplantation may avert the severity of acute graft-versus-host disease (Yoshikawa et al., 2025). We were not able to find a connection between tacrolimus exposure and the risk of aGVHD. First, the incidence of aGVHD grade II–IV was low (19.8%) in our study, likely due to the uniform use of T-cell depletion in all patients, as opposed to the study by Ganetsky, where no antithymocyte globulin was used, and to the study by Sharma et al. and Offer et al., where only 62.7% and 56.7% received immunotherapy with antithymocyte globulin or alemtuzumab (Sharma et al., 2021; Ganetsky et al., 2016; Offer et al., 2015). Second, tacrolimus trough levels during the first month after transplant in our population mostly exceeded the previously reported minimum tacrolimus trough values. The median (Q1–Q3) tacrolimus trough values by days 10 and 28 were 10 ng/mL and 11.87 (9.2–13.7) ng/mL, respectively (van den Brink et al., 2015; Ganetsky et al., 2016; Abraham et al., 1997; Schmid et al., 2006; Offer et al., 2015).

Our study did not demonstrate a statistically significant effect of tacrolimus exposure/levels during the first month on the overall survival in the multivariate model. In contrast, Wong at al. found that tacrolimus exposure during the first 11 days correlated with survival and described the optimal tacrolimus trough range of 7–9 ng/mL (Wong et al., 2005). In the abovementioned study, Sharma et al. reported that tacrolimus trough levels ≥11.55 ng/mL during the second week correlated with relapse incidence (Sharma et al., 2021). However, in both of these studies, the authors followed only a very short period of time after transplantation, and therefore, nothing can be inferred from these studies about tacrolimus exposure during the entire treatment period. In our cohort, it appeared that the longer period we analyzed, the higher level of statistical significance we achieved. Therefore, in the case of short time period of analysis, the results may be inaccurate as they fail to capture tacrolimus exposure over time.

A limitation of our study is that due to the very stable and predictable pharmacokinetics of tacrolimus and well-performed routine dose-adjusted TDM, there were only a minimum number of patients in our cohort who did not achieve minimal therapeutic drug exposure (Venkataramanan et al., 1995). Therefore, we were not able to identify minimum tacrolimus exposures/levels predictive of clinical outcomes (Venkataramanan et al., 1995).

On the other hand, using ROC analysis, we identified an upper limit for each of the 3 months that strongly correlated with the overall survival in the case of months 2 and 3, as well as for the entire three-month period.

Maximum tacrolimus exposure limits decreased over time since transplantation, suggesting that tacrolimus levels should be adjusted/reduced to achieve the best outcome of allogeneic transplantation during the approximately three-month therapeutic period of tacrolimus administration. During the first, second, and third months, the tacrolimus exposure, expressed as AUC_tc, should not exceed 279 ng h/mL (corresponding to a trough concentration of 11.6 ng/mL), 258 ng h/mL (corresponding to a trough concentration of 10.8 ng/mL), and 160 ng h/mL (corresponding to a trough concentration of 6.7 ng/mL), respectively. This principle of decremental exposure over time is consistent with previously published studies, where early tacrolimus dose reduction was associated with very good clinical outcomes (Abraham et al., 1997; Schmid et al., 2006). In fact, adjusting higher blood levels allows for earlier tapering of immunosuppression.

This further highlights the need to monitor tacrolimus levels even in the later periods after transplantation, when disproportionately high exposure and resulting immunosuppression negatively affect the prognosis of patients after allogeneic stem cell transplantation.

Although our study was not designed to directly compare the importance of TDM at different time points after transplantation, when integrating the available data from previous studies with our results, we conclude that in the first month after transplantation, it is probably more important to achieve a minimum therapeutic level to prevent aGVHD, and later, it may be more important not to overexpose tacrolimus/over-suppress the immune system, which increases the risk of relapse and adverse survival outcomes.

There are no substantial discrepancies between univariate and multivariate analyses. Although the non-myeloablative conditioning was not a significant parameter for survival in the univariate analysis, it later turned out as a significant covariate in the multivariate analysis. This apparent discrepancy is probably due to an imbalance between patients receiving ablative and non-myeloablative regimens, as non-myeloablative conditioning is used for older and frail patients, whereas most of younger patients treated for malignancy receive ablative regimens. Thus, the worse survival of older patients in univariate analysis likely overshadowed the less significant effect of non-myeloablative conditioning. Therefore, non-myeloablative conditioning only emerged as a significant covariate after the correct adjustment for age in the multivariate analysis.

We developed a population pharmacokinetic model for tacrolimus, in which the typical value of volume of distribution was 127 L and was not affected by any of the covariates tested, whereas clearance decreased with age according to the following equation: CL = 7.94 × e − 0.0085 × age . Clearance was further reduced by 23% in patients who underwent a repeat transplantation. Our observation that tacrolimus clearance decreases with age seems physiologically plausible. Tacrolimus is completely metabolized before its elimination from the organism, mainly via CYP3A enzymes (Venkataramanan et al., 1995). Although this subfamily of enzymes possesses relatively few age-dependent changes in its functional capacity, some reports indicate a decrease in the hepatic CYP3A content and the expression pattern in the elderly (Konstandi and Johnson, 2023).

As tacrolimus is a substrate of CYP3A enzymes, exposure to tacrolimus may also be altered by polymorphisms in these enzymes. Although a rather minor contribution of the CYP3A genotype to intravenous tacrolimus exposure was observed in patients after allogeneic stem cell transplantation (Pasternak et al., 2022), another study in a renal transplant population reported that CYP3A5 expressors required approximately twice the dose of tacrolimus to maintain comparable exposure compared to non-expressors, and the frequency for CYP3A5 expressors was significantly higher in blacks (100%) and mixed-ancestry patients (76%) than in the white population (12%) (Muller et al., 2020). Due to the fact that CYP3A genotyping is not routinely performed prior to tacrolimus administration and that our study was performed in a retrospective design, we could not examine this marker as a covariate in our model. It should be noted that there were only white patients in our study population. In addition, possible inclusion of the genotype in the model would limit its practical utility, as most clinical sites would also not have the genotype available. Nevertheless, it should be acknowledged that the CYP3A genotype may be a source of undescribed variability in the pharmacokinetics of tacrolimus in the population, particularly where there is a higher proportion of black people.

Our expectation that co-administration of azole antifungals (as known inhibitors of the CYP3A family) would result in a drug–drug interaction characterized by increased exposure to tacrolimus (as a substrate of CYP3A) was not detected. Voriconazole and posaconazole have been reported to be potent CYP3A inhibitors, and their co-administration can lead to up to 5-fold increase in tacrolimus dose-corrected trough concentrations (Vanhove et al., 2017). However, the CYP3A genotype and several clinical variables, such as age and hematocrit, have been identified as factors that blunt this interaction. However, the main reason why this interaction did not prove significant in our analysis is probably the fact that all patients were treated with azole antifungals. In total, 28 patients received voriconazole, 20 patients received posaconazole, and 52 patients were treated with fluconazole, which is considered a moderate inhibitor of CYP3A (Gaies et al., 2011). There was a trend toward lower tacrolimus clearance in voriconazole and posaconazole subgroups than in the fluconazole subgroup, with a median (min–max) tacrolimus clearance in voriconazole, posaconazole, and fluconazole subgroups of 4.50 (1.58–10.70), 4.77 (2.74–8.93), and 5.12 (2.69–8.50) L/h, respectively. Therefore, small differences in tacrolimus clearance among patient subgroups who had a concomitant azole agent and were thus exposed to the drug–drug interaction, combined with the relatively high variability (which may be caused, i.e., by the CYP3A genotype or other clinical factors), made it impossible to detect a statistically significant difference in the effect of individual antifungal agents on tacrolimus clearance. Other co-medications that could potentially influence the pharmacokinetics of tacrolimus are MMP and glucocorticoids. However, as all patients were co-treated with MMP and no patient received glucocorticoids, it was not possible to observe the possible effect of these drugs (due to the absence of a control group). In addition to its direct effect on tacrolimus exposure, immunosuppressive co-medication may also affect the observed outcomes (Pidala et al., 2012). Therefore, it is important to note that our findings may be fully transferable only if the immunosuppressive regimen used is identical.

Tacrolimus is partially distributed in the blood into erythrocytes, which is why it is recommended to measure whole blood levels of this drug. This fact has important implications—when the red blood cell count decreases, tacrolimus levels in total blood may decrease quite significantly, whereas the concentration of the free fraction of the drug may be the same or even increase (which may be associated with a higher risk of toxicity) (Sikma et al., 2020; Sikma et al., 2015). In clinically unstable patients with fluctuating red blood cell counts, some authors, therefore, recommend adjusting the tacrolimus level (and subsequently dosage) according to hematocrit. Nevertheless, no validation studies have yet compared hematocrit-corrected whole blood tacrolimus levels against measured concentrations of unbound drug that could reliably confirm this hypothesis. Patients after allogeneic stem cell transplantation are mostly clinically stable with very predictable hemoglobin (Hb) levels. Around stem cell infusion, their Hb level ranges between 70 and 80 g/L and steadily and uniformly increases to normal levels during first 3 months after transplantation. They do not experience fluid shifts or bleeding complication as they undergo no surgery procedure. We therefore did not consider using hematocrit-adjusted tacrolimus levels in our analyses. On the other hand, it should be noted that the findings of our study are only applicable to comparable patient populations.

Based on the observed association between tacrolimus clearance and age, described in our population model, and considering the emerging AUC_tc cutoff value of 279 ng h/mL, we propose a nomogram for estimating the maximum initial daily dose of tacrolimus in patients undergoing first stem cell transplantation (Figure 2). This nomogram was constructed based on the equation among dose, clearance, and drug exposition (dose = CL × AUC). This dose should be further reduced by 23% for patients undergoing repeat transplantation. Following this proposed dosage, the cutoff value should not be exceeded. However, due to the relatively high residual variability in tacrolimus pharmacokinetics, it is still necessary to subsequently measure tacrolimus levels and adjust the dose if required, according to TDM principles.

Despite the very uniform protocol for immunosuppression, our study was conducted in a retrospective way, which is its main limitation. Furthermore, identification of cutoff values for tacrolimus exposure using overall survival as an endpoint and then using these cutoff values in the survival analysis of the same dataset may lead to the overestimation of significance of findings. Therefore, the findings of this study need to be validated in a prospectively conducted study. Finally, some studies indicated that CMIA tends to overestimate tacrolimus levels in comparison with the LC/MS method (Becker et al., 2020; Levine et al., 2011). This potential bias may have led to a slight overestimation of the identified tacrolimus thresholds, if applied at a site where tacrolimus is measured using the LC/MS method. On the other hand, to the best of our knowledge, this is the first study to cover such a long period of tacrolimus exposure and to report its time dependence with respect to allogeneic stem cell transplantation. Moreover, it proposes a very practical clinical recommendation for tacrolimus dose optimization.

We identified a cutoff value for tacrolimus exposure, above which patients’ overall survival is reduced after allogeneic stem cell transplantation. Exceeding the average trough tacrolimus level in the second and third months after transplantation, above 10.8 and 6.7 ng/mL, respectively, reduces the overall survival by approximately three-fold. We also observed that long-term exposure to tacrolimus (a three-month period) was more important for overall survival than exposure in the initial phase (the first month). Other factors influencing survival included the disease risk category, age ≥55 years, and non-myeloablative conditioning. Furthermore, we described a tacrolimus population pharmacokinetic model, in which tacrolimus clearance decreases with age and is reduced by 23% in patients undergoing a repeat transplantation. Based on these observations, we propose a nomogram for determining the maximum initial tacrolimus daily dose in patients treated with allogeneic stem cell transplantation.