Effects of combined test dose and therapeutic drug monitoring strategy in exposure-directed busulfan

Although exposure-directed busulfan (BU) dosing can improve allogeneic hematopoietic stem cell transplantation outcomes, there is still large variability in BU exposure with test dose alone due to changes in BU clearance caused by drug interactions. We conducted a single-arm phase II trial using the combined test dose and therapeutic drug monitoring strategy (PK-guided group) and compared the outcomes with an external historical cohort receiving a fixed-dose (fixed-dose group). The first eight and second eight doses were adjusted based on the area under the blood concentration-time curve (AUC) of the test and first doses, respectively, targeting a total AUC of 82.1 mg·h/L. All patients received either BU and cyclophosphamide conditioning (BU/CY) or fludarabine (FLU)-containing conditioning. The BU clearance at the first dose decreased more in patients receiving FLU than in those receiving BU/CY; however, BU clearance also declined over time in patients who received BU/CY. The simulated total AUC (sAUC) with test dose only was significantly higher in patients who received FLU than in those who received BU/CY, but sAUC with the combined strategy was comparable. The 100-day progression-free survival was 85.5% (95% confidence interval [CI]: 71.9–92.8%), and was not inferior to that in the fixed-dose group. For the FLU-containing regimens, the PK-guided group showed decreased relapse (0.0% vs. 26.9%, p = 0.03), and favorable overall survival (75.1% vs. 57.0%, p = 0.07) at 1 year. The combined strategy effectively controlled the BU exposure close to the target levels, potentially improving efficacy, especially in patients receiving the FLU-containing regimen. Clinical evaluation of efficacy of dose-modified intravenous busulfan in allogeneic hematopoietic stem cell transplantation for hematological malignancy (#UMIN000014077, June 15th, 2014).


Introduction
Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is a curative treatment for hematologic malignancies or non-malignancies. Historically, busulfan and high-dose cyclophosphamide (BU/CY) or high-dose cyclophosphamide and 12 Gy total body irradiation (CY/TBI) has been used for standard pretransplant conditioning with myeloablative intensity [1]. Fludarabine and 4-day busulfan (FLU/ BU) regimen was developed to reduce treatment-related toxicity while maintaining the antitumor effects in elderly patients or patients with comorbidities [2]. According to a prospective randomized study from the Center for International Blood and Marrow Transplant Research comparing BU-based (mainly FLU/BU or BU/CY) and TBI-based regimens in patients with acute myeloid leukemia (AML) or myelodysplastic syndromes (MDS), the 2-year overall survival (OS) in patients who received BU-based regimens was significantly superior to that in patients who received the TBI-based regimen; thus, a BU-based regimen is now widely used for pretransplant conditioning [3].
In Japan, intravenous BU (IV BU) was approved for pretransplant conditioning in 2003. Although IV BU was assumed to be pharmacokinetically less variable compared to oral BU, there is a three-to-four-fold difference in the area under the blood concentration-time curve (AUC) of the serum BU concentration among patients receiving the same IV dose [4,5]. BU is eliminated by glutathione S-transferase (GST)-catalyzed conjugation with glutathione in the liver, and GSTA1 is the major isoform catalyzing BU conjugation [6]. GSTA1 gene polymorphism is known to be related to BU pharmacokinetics (PK) [7][8][9]. Other factors, such as obesity, are also related to lower clearance (CL) of BU [7,10]. The difference in BU CL resulting from these factors is believed to be responsible for the interpatient variability in BU-PK [11].
It is known that the AUC of BU is related to post-transplant efficacy and BU toxicity. In the context of conditioning with BU every 6 h for 4 days, HSCTs with a single-dose AUC of 950-1520 µM·min exhibited favorable outcomes [12], the risk of engraftment failure or relapse increased in HSCTs with a single-dose AUC of < 1350 µM·min [13], and the risk of treatment-related toxicity including sinusoidal obstruction syndrome (SOS) increased in HSCTs with a single-dose AUC of > 1500 µM·min [14]. These findings suggest that controlling the AUC of BU within a target range would improve HSCT outcomes. In fact, Andersson et al. reported that PK-guided BU, one dose per day for 4 days, in combination with fludarabine, reduced relapse or treatmentrelated mortality, and improved OS and progression-free survival (PFS) in patients with MDS or AML not in remission when compared to the fixed-dose group [15]. Accordingly, personalized BU dosing was recommended as per the guideline proposed by the American Society for Transplantation and Cellular Therapy in 2016 [16]. However, it still remains unclear whether an efficacious, exposure-directed BU dosing can be implemented in Asian patients. Moreover, the strategies for therapeutic drug monitoring (TDM) or test dose adjustment are still under investigation. GSTA1*B allele carriers who have poor GSTA1 expression are less common in Asians than in Caucasians [7,8,17]. In addition, obesity, which can decrease the BU CL, is less prevalent in Asia than in America. Therefore, we assumed that exposure-directed BU dosing would also be more beneficial for Asian patients. Furthermore, there is still uncertainty in the AUC obtained after dose adjustment based on a test dose because BU CL can change due to interactions with other drugs started after the test dose [16,18]. Besides drug-drug interactions, BU CL is also suggested to decline over time [11,19]. Therefore, TDM might be needed to strictly control BU exposure.
In this phase II trial, we used a combined test dose and TDM strategy in the pretransplant conditioning regimen (PK-guided group), prospectively assessed the effects of the combined strategy on BU-PK, and evaluated the transplantation outcomes against an external historical cohort from the Kyoto Stem Cell Transplantation Group (KSCTG) as a fixed-dose group, where a standardized transplantation protocol was used.

Eligibility criteria
Patients aged 16 to 70 years with hematologic malignancies who underwent transplantation in Kurashiki Central Hospital and Kyoto University Hospital between 2014 and 2018 were enrolled in the prospective, single-arm trial. The eligibility criteria were further defined as acceptable organ function and favorable performance status (Supplementary File). A donor matching ≥ 5/6 alleles of HLA-A, -B, or -DRB1 was required for bone marrow transplantation (BMT) or peripheral blood stem cell transplantation (PBSCT), and a donor matching ≥ 4/6 HLA-A, -B, or -DR antigens was required for cord blood transplantation (CBT). Our protocol complied with the Declaration of Helsinki and was approved by the institutional review board of each participating institution and the central review board of Kyoto University Hospital. Written informed consent was obtained from each patient. This study was registered at the Japanese University Hospital Medical Information Network (UMIN) Clinical Trial Registry (clinical evaluation of efficacy of dose-modified intravenous busulfan in allogeneic hematopoietic stem cell transplantation for hematological malignancy; UMIN000014077).

BU dose, sampling, and PK analysis
Either a BU/CY or FLU-containing regimen was administered for pretransplant conditioning. For the BU/CY regimen, BU (Busulfex; Otsuka Pharmaceutical, Tokyo, Japan) was administered on days − 7 to − 4, followed by CY (60 mg/ kg Endoxan; Shionogi, Osaka, Japan) on days − 3 and − 2. For the FLU-containing regimen, FLU (25 mg/m 2 Fludara; Sanofi, Tokyo, Japan) was administered on days − 6 to − 2, and BU was administered on days − 5 to − 2. Low-dose TBI (2-4 Gy) or melphalan (80-140 mg/m 2 Alkeran; Aspen Japan, Tokyo, Japan) was administered following the final dose of busulfan within the FLU-containing regimen as necessary according to the patient's condition or disease status. To calculate BU dose, adjusted body weight (ideal body weight + 0.25 × [actual body weight − ideal body weight]) was calculated if body mass index (BMI) was > 25 kg/m 2 . As for the fixed-dose group, the single BU dose was 0.8 mg/ kg (actual body weight or adjusted body weight if BMI was > 25 kg/m 2 ).
A single busulfan dose was administered for 2 h, and blood samples were collected immediately before and 0, 0.5, 1, 2, and 4 h after BU infusion, and ethylenediaminetetraacetic acid disodium was used for anticoagulation. The blood samples were immediately centrifuged, and the obtained plasma was frozen and then transferred to the Department of Clinical Pharmaceutics at Doshisha Women's College of Liberal Arts for PK analysis. The analysis results were reported on the next day, and the therapeutic doses were adjusted.
The test dose of BU (0.8 mg/kg) was administered 2 to 7 days prior to the start of the therapeutic conditioning regimen. We set a target total AUC of 82.1 mg·h/L (20,000 µM·min) and administered the first through eighth doses based on the test dose AUC. We further adjusted the ninth to sixteenth doses based on the first and test dose AUC values. We calculated and simulated total AUC without dose adjustment, with test dose alone (dose adjustment × 1) or after applying a combined strategy (dose adjustment × 2) where AUC and CL values were calculated from the test, first, and thirteenth dose according to the following equation: AUC = Dose CL . A schematic overview of the protocol and the calculations for dose adjustment and simulated total AUC (sAUC) are shown in Fig. 1 and Eqs. 1-5, respectively. The plasma BU levels were calculated by highperformance liquid chromatography, and the pharmacokinetic parameters including AUC or clearance were analyzed with one-compartment model using Phoenix WinNonlin 6.4 (Certara LP, Princeton, NJ, USA). The detailed methods for plasma concentration measurement and PK analysis have been described previously [20].

Transplantation procedures
If available, HLA-matched related donors were the top priority in donor selection, and HLA-matched unrelated donors were an alternative option. If an HLA-matched donor was unavailable, a one-allele mismatched donor or cord blood graft was selected as the HSCT donor.
The conditioning regimens were at the discretion of the attending physicians and were determined based on disease status or patient characteristics. Generally, younger patients (< 55 years old) with acceptable comorbidities received the BU/CY regimen, and elderly patients or patients with comorbidities were assigned to the FLU-containing regimen. Voriconazole, itraconazole, or metronidazole administration (2)  was avoided during preconditioning. Sodium valproate was administered to all patients prophylactically.
Graft-versus-host disease (GVHD) prophylaxis consisted of calcineurin inhibitor (CNI) alone, CNI plus methotrexate (MTX), and/or mycophenolate mofetil (MMF) (Supplementary File). All transplant protocols were in accordance with the guidelines published by the Japanese Society for Transplantation and Cellular Therapy.

Data collection
Because temporary or cumulative BU doses can be higher than the target when using the combined strategy, the primary endpoint was to compare the 100-day PFS after HSCT with that of the external cohort from KSCTG as the fixeddose group to evaluate the non-inferiority of early BU toxicity or early relapse due to excessive or inadequate BU dosing. The secondary endpoints included OS, PFS, relapse, non-relapse mortality (NRM), and the incidence of adverse events at 1 year after HSCT. The patients were followed up and the engraftment, adverse events, and outcomes were evaluated at 30 days, 100 days, and 2 years after transplantation. The KSCTG cohort data were extracted from the Transplant Registry Unified Management Program, sponsored by the Japanese Society for Transplantation and Cellular Therapy and the Japanese Data Center for Hematopoietic Cell Transplantation [21].

Sample size
According to the previous study [3], we assumed that the 100-day PFS would be 85% for the PK-guided group and 80% for the fixed-dose group. As we anticipated that a fifth of all patients in the KSCTG would be enrolled in the PKguided group, we estimated that 46 patients (184 patients for the fixed-dose group) would be required in the PK-guided group to show non-inferiority in the 100-day PFS compared to the fixed-dose group, with a power of 80%, a one-sided alpha level of 0.05, and a non-inferiority margin of 10%. The target number was set to 50 patients after considering the dropouts.

Statistical analysis
The percentage of the patients whose CL decreased was compared between patients receiving BU/CY conditioning and FLU-containing conditioning using chi-squared tests. The variance of the sAUC was compared using the F test. The averages of the sAUC without dose-adjustment, with test dose only (dose adjustment × 1), and with combined test dose and TDM strategy (dose adjustment × 2) were compared between patients receiving BU/CY conditioning and FLU-containing conditioning using t-tests. Patient characteristics were compared between the PK-guided and fixed-dose groups using chi-squared tests or t-tests. Noninferiority of 100-day PFS was confirmed if the 95% confidence interval (CI) of the probability difference between the two groups did not exceed the non-inferiority margin of 10%. NRM or relapse was estimated using the cumulative incidence function and compared with the Fine-Gray proportional hazards model, considering relapse or death as a competing risk, respectively. OS and PFS were estimated using the Kaplan-Meier method and compared with the Cox proportional hazards model. Variables with p < 0.1 in the univariate analysis were selected for multivariate analysis. The cumulative incidence of SOS 1 year after allo-HSCT was calculated considering death as a competing event. The incidence of adverse events was compared between the PKguided and fixed-dose groups using the Fine-Gray proportional hazards model. All statistical analyses were performed using Stata (version 16.1, StataCorp, College Station, TX, USA).

Patient characteristics
Fifty-two patients were registered, but three patients failed to start the study due to deteriorated cardiac function or uncontrollable infection. Consequently, 49 patients were enrolled in this study. The median age of the patients was 54 years ( Table 1). All patients were Asian, and the median body weight and BMI were 59.1 kg and 22.7 kg/m 2 , respectively. Twenty-four patients received the BU/CY regimen, and 25 patients received the FLU-containing regimen.

Results of PK analysis and dose adjustment
Compared to the BU CL for the test dose, CL at the first dose decreased more in patients who received the FLU-containing regimen than in those who received BU/CY regimen (76.0% vs. 41.6%, p = 0.02). CL of patients who received BU/CY regimen also declined over time (CL 13th was lower than CL 1st by an average of − 0.913 L/h in patients who received BU/ CY regimen) (Fig. 2a, b; Table 2).
Based on the test dose, the first through the eighth dose were increased in 31 patients (63.3%) and decreased in 17 patients (34.7%). The median (range) dose during this phase was 0.85 (0.53-1.16) mg/kg. Based on the TDM of the first dose, the ninth through the sixteenth dose were increased in 24 patients (49.0%) and decreased in 24 patients (49.0%) compared to the test dose. Additionally, compared to the first dose, the ninth through the sixteenth dose were increased in 16 patients (32.7%), and decreased in 31 patients (63.3%). The median (range) dose during this phase was 0.80 (0.40-1.62) mg/kg. The median (range) cumulative dose (including the test dose and sixteen therapeutic doses) was 14.08 (8.30-21.64) mg/kg. In the fixed-dose group, sixteen doses of BU 0.8 mg/kg were administered, and the cumulative dose was 12.8 mg/kg (Fig. 2c). The average (standard deviation) sAUC values without dose adjustment, with dose adjustment × 1, and with dose adjustment × 2 were 79.02 (14.71), 86.73 (10.32), and 84.06 (2.84) mg·h/L, respectively. The variance in sAUC with dose adjustment × 2 was significantly reduced compared to that without dose adjustment and with dose adjustment × 1 (p < 0.001 and p < 0.001, respectively). The sAUC was outside the target AUC ± 10% range (< 73.89 mg·h/L or > 90.31 mg·h/L) in 29 patients without dose adjustment, 18 patients with dose adjustment × 1, and 3 patients with dose adjustment × 2. Notably, the sAUC was < 73.89 mg·h/L (target AUC − 10%) in 20 patients without dose adjustment, 4 patients with dose adjustment × 1, and no patient with dose adjustment × 2 (Fig. 2d). The sAUC without dose adjustment in patients who received the FLU-containing regimen was significantly higher than that of the patients who received the BU/ CY regimen (85.36 vs. 72.41 µM·min, p = 0.001), and the sAUC with dose adjustment × 1 was still higher in those who received FLU-containing regimen (89.63 vs. 83.70 µM·min, p = 0.04). The sAUC with dose adjustment × 2 of the patients who received the FLU-containing regimen was not significantly different from that of the patients who received the BU/CY regimen (83.80 vs. 84.33 mg·h/L, p = 0.52) (Fig. 2e). During BU administration, two patients developed visual and auditory hallucinations; one patient was deemed grade 2 and the other was deemed grade 1 according to the Common Terminology Criteria for Adverse Events Ver 5.0. Both patients received an FLU-containing regimen. For the patient with grade 2 hallucinations, the AUC of the first dose was as high as 6.25 mg·h/L (average of all patients was 4.99 mg·h/L). Both patients recovered without sequelae when the therapeutic conditioning dose was completed.

Patient characteristics of the fixed-dose group
Next, we compared the results with the retrospective data from the KSCTG cohort. A total of 251 patients were extracted based on inclusion and exclusion criteria that were similar to those of the PK-guided group (Fig. 3). For the BU/CY regimen, the BMI was higher, the ratio of patients with AML was smaller, and the ratio of PBSCT was smaller in the PK-guided group. For the FLU-containing regimen, body weight and BMI were higher, more patients had a hematopoietic cell transplantationcomorbidity index (HCT-CI) of ≥ 3, and more patients used CNI + MTX + MMF for GVHD prophylaxis in the PK-guided group ( Table 3).

Impact of combined test dose and TDM strategy on post-transplantation adverse events
In terms of post-transplantation adverse events, no significant increase in SOS was observed (4.2% vs. 3.6% at 1 year after transplantation, p = 0.86). For the BU/CY regimen, the incidence of overall viral infection was significantly higher in the PK-guided group. As for the FLU-containing regimen, there was a tendency towards increased bacterial infection and viral cystitis. Persistence of recipient hematopoiesis (mixed chimera or recovery of recipient hematopoiesis) was less frequent in the PK-guided group  for both conditioning regimens, although not statistically significant (Table 5).

Discussion
Our study demonstrated two major findings: (1) Using combined test dose and TDM strategy, we were able to reduce deviations from the target of total AUC ± 10% more than test dose alone. (2) The combined strategy led to lower relapse and improved OS and PFS, especially in patients with FLUcontaining conditioning compared to that of the fixed-dose group. Some drugs, such as fludarabine, metronidazole, and itraconazole, are known to influence the metabolism or PK of BU [16]. Therefore, an unpredictable AUC increase could occur due to interactions with combination drugs even after the first dose adjustment based on the test dose. Indeed, the BU CL decreased significantly at the first dose in patients who received FLU-containing regimen compared to BU/CY regimen, and the sAUC tended to be higher in the patients receiving FLU-containing regimen without dose adjustment or with test dose only. Furthermore, BU CL declined over time during pretransplant conditioning. Therefore, there was still considerable variability in sAUC with test dose alone. In our study, the combined test dose and TDM strategy successfully corrected the deviation. Thus, the total BU exposure remained well-controlled close to the target value in almost all patients. As a result, the cumulative BU dose differed by up to ~ 2.6-fold between patients, in accordance with previous literature [4,5]. Because BU administration can cause seizures, anticonvulsants were prophylactically administered to all patients in our study, and there were no patients who experienced seizures. The patients who developed visual and auditory hallucinations did not take itraconazole or metronidazole during pretransplant conditioning. Posterior reversible encephalopathy syndrome was excluded because magnetic resonance imaging of the brain revealed no abnormalities. Visual and auditory hallucinations during or after BU administration were reported previously, and all these hallucinatory events were grade 1 [22]. The relationship between high plasma BU concentration and the cause of hallucinatory events is not known. N,N-Dimethylacetamide, an additive in the BU injection, can also cause hallucinations [23]. The efficacy of the combined test dose and TDM strategies has been reported for the pretransplant conditioning of ALL patients [24][25][26]; however, all of these reports were of single-arm studies. To the best of our knowledge, this is the first study to compare the efficacy of the combined Table 3 Comparison of characteristics of PK-guided group and fixed-dose group Abbreviations: CR, complete remission; HLA, human leukocyte antigen; ATG , anti-thymocyte globulin; PK, pharmacokinetics; BMI, body mass index; AML, acute myeloid leukemia; MDS, myelodysplastic syndromes; ALL/LBL, acute lymphoblastic leukemia/lymphoblastic lymphoma; allo-HSCT, allogeneic hematopoietic stem cell transplantation; HCT-CI, hematopoietic cell transplant-comorbidity index; BM, bone marrow; PB, peripheral blood; CB, cord blood; BU, busulfan; CY, cyclophosphamide; FLU, fludarabine; TBI, total body irradiation; MEL, melphalan; strategy with that of a control arm. In the present study, the average total exposure was slightly higher than the target (max 92.17 mg·h/L). The 100-day PFS in the PK-guided group was not inferior to that in the fixed-dose group; thus, this overreach was not critical for early BU toxicity. We were able to eliminate cases with the total AUC below 73.89 mg·h/L with the combined strategy. As a result, we were able to reduce relapse and improve prognosis, particularly in patients receiving the FLU conditioning regimen. Therefore, the conventional fixed dose per body size may be insufficient for some patients, possibly because poor metabolizers with the GSTA1*B allele or obese patients are relatively infrequent in Japan [17,27]. In contrast, we could not find any significant difference in the outcome of patients who received the BU/CY regimen. The reason for this is unknown; however, one possible reason is that a high dose of CY may offset the insufficient BU dose. The difference in patient backgrounds may be another reason. For example, among the patients who received the BU/CY regimen, there were more patients who received PBSCT in the fixed-dose group than in the PK-guided group, which might reduce relapse in the fixed-dose group. However, further investigation is required to confirm the long-term benefits of disease relapse. Although we set the target total AUC to 82.1 mg·h/L, the optimal target AUC has not yet been determined. Oran et al. reported that the PFS in patients with a target AUC of 20,000 µM·min was significantly improved compared with that in patients with an AUC of 16,000 µM·min because of reduced progression [28]. Bartelink et al. reported that improved clinical outcomes are likely to be achieved by targeting the BU AUC to 78-101 mg·h/L, with reduced graft failure or relapse and acceptable toxicity [29]. Andersson et al. reported improved OS and PFS in patients with a targeted daily AUC of 6000 µM·min (total of 24,000 µM·min) compared with the dose-fixed group [15]. Esteves et al. indicated a higher risk of SOS with a daily AUC above 5000 µM·min (total > 20,000 µM·min) [30]. Accordingly, targeting a total AUC of approximately 78-99 mg·h/L is now the most commonly accepted strategy. However, in our study, infections in the PK-guided group may have increased, while the persistence of recipient hematopoiesis appeared to be less frequent in the PK-guided group than in the fixed-dose group. Indeed,    personalized BU dosing was reported to impact donor chimerism [31]. Therefore, optimization of the target total AUC in individual patients (for example, setting a lower target in patients with a lower risk of relapse) requires further investigation. The present study revealed the efficacy of the combined test dose and TDM strategy in pretransplant conditioning. However, there are few limitations to this study that must be considered. First, due to the nature of this phase II study, which included a relatively small number of patients, caution should be exercised in interpreting results other than the primary endpoint. Second, because the transplant outcomes of the patients of the prospective trial were compared with historical cohort data, there may have been a selection bias (i.e., patients with relatively favorable physical conditions or disease status may have been recruited in the study). Differences in patient characteristics, such as BMI, HCT-CI, and GVHD prophylaxis, can also skew the evaluation of outcomes. Therefore, a prospective randomized study is required to overcome those problems.
In summary, our study revealed that the conventional BU fixed-dose strategy is insufficient for some patients, and there was still considerable variability in total exposure with test dose only due to change in CL during pretransplant conditioning. The combined test dose and TDM strategy successfully reduced deviation from the targeted AUC range. Our study showed decreased relapse and favorable survival with the combined strategy, especially in patients who received FLU-containing conditioning regimens, suggesting that this strategy enables us to enhance the antitumor effect without increasing fatal complications.