Correction: Correlation analysis between plasma concentration of nilotinib and clinical efficacy and safety in patients with chronic myeloid leukemia: a single–center retrospective cohort study
Year: 2025
Type: erratum
Abstract: Chronic myeloid leukemia (CML) is a malignant neoplasm characterized by the clonal proliferation of hematopoietic stem cells driven by the breakpoint cluster region-Abelson (BCR::ABL1) fusion gene (Alves et al., 2021; Deininger et al., 2020; Garcia-Ferrer et al., 2019), with more than 90% of patients exhibiting Philadelphia chromosome (Ph) abnormalities (Tian et al., 2018). Although tyrosine kinase inhibitors (TKIs) have substantially improved clinical outcomes, clinical evidence indicates that some patients develop resistance, experience intolerance, or show no response to first-generation BCR::ABL1 inhibitors such as imatinib (Sampaio et al., 2021; García-Gutiérrez et al., 2019; Dinić et al., 2024; He et al., 2021).To optimize therapeutic strategies and improve remission rates in CML, several second-generation BCR::ABL1tyrosine kinase inhibitors (TKIs), including nilotinib, dasatinib, ponatinib, and bosutinib, have been successfully developed (Sampaio et al., 2021; Tian et al., 2018). More recently, asciminib, a groundbreaking treatment for CML, has gained FDA approval as the first STAMP inhibitor, which offers a novel mechanism of action by targeting the myristoyl pocket of BCR::ABL1, proves effective in overcoming resistance to traditional TKIs, and demonstrates superior efficacy and a favorable safety profile in clinical trials (Akhtar et al., 2025;Choi, 2023). Notably, nilotinib exhibits approximately 20-fold greater inhibitory potency compared to imatinib, with around 50% of imatinib-resistant patients achieving complete cytogenetic response (CCyR) while maintaining acceptable tolerability (Dinić et al., 2024; He et al., 2021; Wang et al., 2018).Nilotinib, a second-generation BCR::ABL1 inhibitor, is primarily indicated for patients with chronicphase or accelerated-phase CML who exhibit resistance or intolerance. It shows a bioavailability of 30%, with 98% plasma protein binding and predominant hepatic metabolism via CYP3A4 (Ding et al., 2013; Tian et al., 2018). According to the 2024 National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology for Chronic Myeloid Leukemia (Shah et al., 2024a), nilotinib is recommended as first-line therapy for patients with intermediate-to high-risk CML (Fukuda, 2022; Wang et al., 2018). However, considerable interindividual variability in plasma concentration has been reported worldwide, and a consensus therapeutic window for nilotinib has not yet been established.Therapeutic drug monitoring (TDM), a clinical approach for quantifying drug concentrations in blood and evaluating therapeutic ranges, is recognized as a key strategy for optimizing nilotinib therapy (Miura, 2015; Mueller-Schoell et al., 2021; Tuzimski et al., 2020). Therefore, investigating the relationship between nilotinib plasma concentrations and clinical efficacy and safety profiles in CML patients provides essential evidence for assessing the feasibility of TDM implementation.This study investigates the clinical significance of nilotinib plasma concentrations in CML patients by evaluating efficacy and safety outcomes. Through detailed analysis of the complex relationships between plasma concentrations, therapeutic responses, and adverse drug reactions (ADRs), and by systematically examining the influence of factors such as gender, age, treatment duration, and hepatic or renal function indices, this study aims to provide essential evidence to support personalized nilotinib treatment strategies in CML management.Materials and MethodsThis Figure 1. Flowchart of the study design.An ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) method was employed to quantify nilotinib plasma concentrations. The UPLC-MS/MS system was calibrated using a series of standard solutions, and a calibration curve was generated with a linear regression analysis (y = mx+b, R² > 0.99), where y represents the peak area ratio of the analyte to the internal standard, and x represents the concentration of the analyte. Quality control samples were analyzed to verify the accuracy and precision of the assay:Lower limit of quantification (LLOQ): 1 ng•mL -1 ； Linear range: 1-5000 ng•mL -1 ；Intra-/inter-day precision: <15% CV. Concentration normalization was performed using an internal standard to account for potential matrix effects and instrument variability.Sampling Protocol: Blood samples were strictly collected within 30 minutes prior to the next scheduled dose(trough concentration, Cmin) between 8:00-8:30 AM, ensuring measurement of true steady-state concentrations after ≥8 days of continuous dosing. This sampling time was selected to target the trough concentration based on the known pharmacokinetic profile of nilotinib, which reaches peak serum concentration (Cmax) approximately 4 hours post-dose, has an elimination halflife (t₁/₂) of approximately 16 hours, and achieves steady-state plasma concentrations within 8 days of continuous administration (Ding et al., 2013;Wang et al., 2018).Analysis of Repeated Measures: Pharmacokinetic data were analyzed using a mixed-effects model to account for both inter-individual and intra-individual variability. The model included time as a fixed effect and individual random intercepts and slopes to capture the variability in concentration profiles among participants. Model fit was evaluated using the Akaike Information Criterion (AIC)and Bayesian Information Criterion (BIC) (Jones, 2011), with lower values indicating a better fit.Nonlinear mixed-effects modeling was conducted to estimate the structural model and population pharmacokinetic parameters and pharmacokinetic parameter estimates (Choi et al., 2020; Kang et al., 2009).According to the NCCN 2024 guidelines (Shah et al., 2024b), major molecular response (MMR) was used to evaluate therapeutic efficacy, defined as BCR::ABL1 transcript levels ≤0.1% on the International Scale (BCR::ABL1 IS) at any time point. Adverse events were graded based on the Common Terminology Criteria for Adverse Events (CTCAE) (Freites-Martinez et al., 2021) version 5.0, with systematic monitoring and documentation throughout the treatment period.Patient demographic data, including age, sex, prior treatment, and comorbidities, were collected.Information on nilotinib dosing regimens, plasma concentration measurements, BCR::ABL1 IS levels, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), serum creatinine (Cr), neutrophils (NEUT), red blood cell (RBC), hemoglobin (Hb), platelets (PLT), triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), glucose (GLU), amylase, lipase, as well as therapeutic efficacy outcomes and occurrences of adverse events during treatment, was also recorded.Data on BCR::ABL1 mutation status at baseline or during treatment were not routinely collected as part of standard clinical care for all patients in this cohort and are therefore unavailable for analysis.Similarly, pharmacogenetic profiling (e.g., for CYP3A4, CYP2C8, or UGT1A1 polymorphisms) and comprehensive screening for drug-drug interactions were not performed. While these factors represent potential confounders, their absence is a inherent constraint of the retrospective design.Future prospective studies are warranted to incorporate these valuable parameters.Statistical analyses were conducted using IBM SPSS Statistics (version 21.0, IBM Corp., Armonk, NY, USA). Continuous variables were described as mean ± standard deviation (Mean ± SD), and categorical variables were presented as percentages. Comparative analyses were performed using the X 2 test, the Mann-Whitney U test, and the Kruskal-Wallis H test. Spearman's rank correlation was used to assess associations. Receiver operating characteristic (ROC) curve analysis was applied to evaluate the predictive value of plasma concentrations for therapeutic efficacy and safety. Univariate and multivariate linear regression models were used to identify factors influencing nilotinib plasma concentrations. Data visualization was performed using GraphPad Prism (version 10.0, GraphPad Software, San Diego, CA, USA). A two-tailed p<0.05 was considered statistically significant.3 ResultsPatients were stratified into the effective group(n=67) and the ineffective group(n=54) based on achievement of MMR. Baseline characteristics of the two groups are presented in Supplementary Table S1. No significant differences were observed between the groups in baseline variables, including sex, age, prior treatment, comorbidities, and dosage(P>0.05). However, significant differences were noted in treatment duration(P=0.04).The median plasma concentration of nilotinib among the 121 patients was 870.04 ng•mL -1 (IQR:526.35-1219.97), with a mean value of 902.84±509.42 ng•mL -1 . The nilotinib plasma concentrations were significantly higher in the effective group (1036.40±463.67 ng•mL -1 ) compared to the ineffective group (737.14±518.97 ng•mL -1 ) (P<0.001; Figure 2A). A significant moderate positive correlation was observed between nilotinib plasma concentrations and MMR (Spearman's ρ=0.33, P<0.001).Using MMR attainment as the dependent variable and plasma concentration as the independent variable, ROC curve analysis yielded an area under the curve (AUC) of 0.693 (95% CI:0.596-0.791) for predicting MMR. At the maximal Youden index (J=0.373), the optimal plasma concentration threshold was identified as 636.99 ng•mL -1 , with a sensitivity of 83.6% and a specificity of 53.7% (Figure 2B)To mitigate potential bias from TDM-guided dose adjustments (where low-exposure patients receive higher doses and vice versa), we applied dose normalization using the formula:Cnorm= Cmin/(Dose/600mg) After normalization, median Cnorm was 907.02 ng•mL -1 (IQR:546.14-1274.47), mean 907.02 ± 600.41 ng•mL -1 . The efficacy correlation persisted:MMR-achievers maintained significantly higher Cnorm (1045.10±468.08 ng•mL -1 ) versus non-responders (858.34 ± 723.66 ng•mL -1 ; P<0.05, Figure 3A), with a sustained moderate correlation to response depth (ρ=0.27, P<0.05).ROC analysis using Cnorm showed an AUC of 0.655 (95% CI:0.554-0.755) and notably identified the identical optimal threshold of 636.99 ng•mL -1 (sensitivity 85.1%, specificity 46.3%; Figure 3B). This consistency confirms that while TDM practices attenuated statistical associations, the underlying exposure-efficacy relationship remains robust. (B) Receiver operating characteristic (ROC) curve analysis of Cnorm for predicting major molecular response. The persistence of a significant AUC after dose normalization strengthens the finding that the efficacy threshold is a physiologically relevant target, independent of administered dose.Abbreviations: AUC, area under the curve; CI, confidence interval; Cnorm, dose-normalized concentration.3.3 Relationship between plasma concentration and overall incidence of ADRs 4B) showed no statistically significant associations. Fourteen patients (11.57%) required dose modifications (reduction n=7, discontinuation n=7) due to intolerance, though no fatal events occurred. Collectively, these analyses demonstrate that nilotinib exposure levels -whether raw or dose-normalized -were not predictive of ADR risk or severity in this cohort.Among all ADRs, non-hematologic toxicities were predominant, with the most common being hyperbilirubinemia (28.1%), elevated ALT (26.4%), and hypercholesterolemia (24.0%), primarily grade 1-2 (27.3%, 25.6%, and 24.0%, respectively). Only one case (0.8%) of grade 3-4 toxicity was reported for either hyperbilirubinemia or elevated ALT. Hematologic toxicities mainly included anemia (25.7%) and thrombocytopenia (13.2%), with grade 3-4 incidences of 1.7% and 2.5%, respectively (Supplementary Table S2) Data are presented as mean ± SD. Ordinary one-way ANOVA revealed no statistically significant differences in nilotinib exposure across all comparison groups (all P > 0.05), indicating that plasma concentration was not a predictor of ADR risk or severity. This finding was consistent for both raw (A) and dose-normalized (B) concentrations, and aligns with the clinical profile of nilotinib where ADRs are common but predominantly low-grade and manageable.Abbreviations: ADR, adverse drug reaction; Cnorm, dose-normalized concentration. Based on the efficacy-derived threshold (636.99 ng•mL -1 ) 1 and pharmacokinetic characteristics (Ding et al., 2013;Tian et al., 2018) suggesting toxicity risk escalates above double this threshold (>1273.98 ng•mL -1 ), patients were stratified into three cohorts: low concentration (<636.99 ng•mL -1 , n=40), medium concentration (636.99-1273.98 ng•mL -1 , n=53) and high concentration (>1273.98 ng•mL -1 , n=28). The overall incidences of ADRs were 72.5%, 79.2% and 78.6%, respectively. Significant differences in plasma concentration distribution were observed among the three groups (P<0.0001;Figure 5). In all cohorts, grade 1-2 ADRs were predominant (57.5%, 69.8%, 67.9%), while grade 3-4 events were less frequent (15.0%, 9.4%, 10.7%).The distributions of ADRs, including rash, headache, myalgia, elevated ALT, elevated AST, increased lipase, increased amylase, elevated triglyceride, elevated cholesterol, thrombocytopenia, anaemia, and neutropenia, showed no significant differences among the groups (all P>0.05).However, a significant association was observed between nilotinib concentration groups and the incidence of hyperbilirubinemia (P=0.030). The high-concentration group exhibited a markedly higher incidence (50.0%) and severity of hyperbilirubinemia compared to the low (20.0%) and medium-concentration groups (22.6%), with grade 3-4 events occurring in 3.6% of highconcentration group, while no severe cases were reported in the other groups (Supplementary Table S3).Utilising the occurrence of hyperbilirubinemia as the dependent variable and plasma concentration as the independent variable, ROC curve analysis yielded an AUC of 0.656 (95% CI: 0.540-0.771) for predicting elevated TBIL. At the maximal Youden index (J = 0.274), the optimal plasma concentration threshold was identified as 1290.34 ng•mL -1 , resulting in a sensitivity of 41.2% and a specificity of 86.2% (Figure 6). ng/mL, n=40), medium (636.99-1273.98 ng/mL, n=53), and high (>1273.98 ng/mL, n=28) concentration groups. The overall incidence of ADRs was high and comparable across all groups (72.5%-79.2%), with Grade 1-2 events being predominant in each cohort. The distribution of most specific ADR types (e.g., rash, elevated ALT, anemia) did not differ significantly among groups (all P > 0.05). A notable exception was hyperbilirubinemia, which exhibited a significant concentration-dependent increase in incidence (P=0.030), with the high-concentration group experiencing a markedly higher rate (50.0%).Abbreviations: ADR, adverse drug reaction. The ROC curve was generated to evaluate the predictive performance of nilotinib plasma concentration for the occurrence of hyperbilirubinemia. The area under the curve (AUC) was 0.656 (95% CI: 0.540-0.771).The optimal predictive threshold was identified as 1290.34 ng/mL.Abbreviations: AUC, area under the curve; CI, confidence interval.We compared the treatment duration between the effective group and the ineffective group. The distribution of patients across different treatment phases (0-6 months, 6-12 months, >12 months) was significantly different between the two cohorts (P<0.001). A notably larger proportion (67.2%) of the effective group were found in the longer treatment duration groups (>12 months), whereas the ineffective group were more prevalent in the shorter treatment duration groups (0-6 months, 44.0%).This indicates that responders were consistently exposed to nilotinib for a longer period.Univariate linear regression analysis was performed to evaluate the associations between nilotinib plasma concentrations and clinical variables. The results showed that only treatment duration was significantly associated with plasma concentration (B=113.150, P=0.036), while PLT approached but did not reach statistical significance (B=1.294, P=0.073). No significant associations were observed for age, sex, dosage, ALT, AST, TBIL, Cr, NEUT, or Hb (P>0.05; Supplementary Table S4).As univariate linear regression identifies only independent associations without adjusting for confounding factors, variables with P < 0.05 (treatment duration, P=0.036), those approaching the significance threshold (PLT, P=0.073), and metabolically relevant parameters (TBIL and Hb) were included in the multivariate linear regression model to assess correlations with plasma concentrations.The results showed that although the overall model did not reach statistical significance (F=1.926, P=0.111), treatment duration remained a significant independent predictor of nilotinib concentrations (B=112.777, P=0.029). No significant associations were observed for TBIL, Hb, or PLT (P>0.05).These findings suggest from treatment duration, the included variables were to a predictive a second-generation BCR::ABL1 inhibitor, significantly clinical outcomes in imatinib-resistant or CML patients and is recommended as first-line therapy for intermediate-to high-risk CML by the NCCN (Shah et al., However, interindividual variability in plasma protein drug and metabolism (Alves et al., 2021; García-Gutiérrez et al., to pharmacokinetic to the In this analysis of nilotinib plasma concentrations in 121 CML patients at of revealed interindividual variability ng•mL -1 ), although the distribution was consistent with reported values et al., et al., Wang et al., 2018). findings support the for into evidence indicates a complex and relationship between nilotinib plasma concentrations and clinical A by Garcia-Ferrer et et al., found no statistically significant in mean concentrations between patients who MMR ± ng•mL -1 ) and non-responders ± ng•mL -1 ) = a study by et et al., reported no significant correlation between nilotinib concentrations and MMR a in studies of cohorts: dose adjustments driven by therapeutic in patients with low exposure or in those with et al., the observed exposure and the true this potential we applied dose normalization which estimates exposure under dosing. the efficacy association after the identical optimal threshold (636.99 ng•mL -1 MMR was and Cnorm remained significantly higher in responders consistency that while TDM practices attenuated statistical associations by the 636.99 ng•mL -1 threshold a physiologically relevant efficacy target independent of dosing study significantly higher concentrations in the effective groups compared to the ineffective groups ± ± ng•mL -1 ; P < with MMR at concentrations ng•mL -1 . This threshold is with findings from et (Fukuda, who an target of ng•mL -1 . Notably, cohort study on concentrations, while data, suggesting a predictive value The consistency between findings and those of the of the pharmacokinetic window during the of the predictive accuracy of therapeutic efficacy, comprehensive also profiles (e.g., BCR::ABL1 mutation and characteristics (e.g., CYP3A4 et al., et al., to the et al., ), consistent with the findings of the by Garcia-Ferrer et et al., with imatinib, nilotinib a higher risk of elevated et al., A comprehensive et al., reported a significantly increased risk of in patients compared with Notably, study a concentration-dependent risk of the high-concentration group exhibited a significantly higher incidence of elevated suggesting a potential threshold of ng•mL -1 analyses have in hyperbilirubinemia et al., consistent with of et al., Although a indicated a higher incidence of in patients compared with those imatinib et al., no concentration-dependent relationship for was observed in the study In the most common ADRs were anemia (25.7%) and thrombocytopenia (13.2%), with grade 3-4 incidences of 1.7% and 2.5%, respectively. with other clinical et al., 2024; He et al., study reported a lower incidence of grade 3-4 which to the and potential comparable overall ADR rates across different concentration for concentration-dependent toxicities in patients with high drug exposure and is primarily in the via and (Tian et al., 2018). Univariate regression analysis revealed a significant positive correlation between treatment duration and plasma concentrations (B=113.150, P=0.036), suggesting that therapy increase drug exposure CYP3A4 or pharmacokinetic finding is by clinical that the effective group a significantly longer treatment duration compared to the ineffective group the of patients who a molecular response are maintained on nilotinib, to longer patients with resistance or treatment are to resulting in a shorter observed treatment the longer treatment duration in the effective group both a and a of variables such as and showed no significant consistent with studies higher exposure in than in a the clinical threshold for (Tian et al., 2018). based pharmacokinetic in studies et al., et al., which the of and comparable steady-state between and with no observed consistent with Although is known to increase nilotinib bioavailability et al., variables were not in this the for predictive models in with statistical significance < treatment (PLT, P = and Hb) in univariate analyses were included in multivariate linear regression to evaluate associations with nilotinib plasma concentrations. Although the overall model statistical significance (F=1.926, P=0.111), treatment duration remained an independent predictor of plasma concentrations (B=112.777, whereas TBIL, Hb, and PLT showed no significant associations (all Notably, the analysis by Garcia-Ferrer et et al., showed a in the predictive of plasma concentrations with in variable findings the for larger and of to improve model study is by retrospective and which the of for adverse including CYP3A4 BCR::ABL1 mutation and other pharmacogenetic factors (e.g., and drug-drug were not of the Although the of systematic data is a common constraint in pharmacokinetic we potential confounding effect on cohort distribution and response outcomes. studies are to these and predictive by Garcia-Ferrer et et al., and Wang et et al., 2018) that evidence not support the of therapeutic drug monitoring (TDM), a by findings of threshold the prospective data from et (Fukuda, support the value of concentration monitoring and suggest that studies on concentration profiles during the treatment than on analysis to documentation of baseline risk (e.g., in this retrospective we not for these in multivariate This common in TDM studies et al., influence the precision of estimates but not the identified threshold consistency with data (Fukuda, 2022; Garcia-Ferrer et al., study that TDM of nilotinib provides for CML with on concentration to efficacy, while monitoring of bilirubin levels in patients we a exposure which the of these Clinical treatment the of TDM to therapeutic and toxicity CML into precision by
Source: Frontiers in Pharmacology
Authors: Ying Liu, Taoyan Lin, Boxin Zhao, Yilei Li, Ping Zheng
Institutions: Nanfang Hospital, Southern Medical University
Language: English
Subfield: Hematology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W4416231064
Cites:
Cited by:
Related to: