Article Text

Original research
Initial dosage optimisation of cyclosporine in Chinese paediatric patients undergoing allogeneic haematopoietic stem cell transplantation based on population pharmacokinetics: a retrospective study
  1. Huanwen Feng1,2,
  2. Xianggui Wang1,2,
  3. Wei Zheng3,
  4. Sha Liu4,
  5. Hua Jiang4,
  6. Yuxian Lin5,
  7. Haojie Qiu3,
  8. Teng Fong Chan1,2,
  9. Min Huang1,2,
  10. Yan Li6,
  11. Xiaolan Mo3,
  12. Jiali Li1,2
  1. 1Institute of Clinical Pharmacology, Sun Yat-Sen University School of Pharmaceutical Sciences, Guangzhou, Guangdong, China
  2. 2Guangdong Provincial Key Laboratory of New Drug Design and Evaluation, Sun Yat-Sen University School of Pharmaceutical Sciences, Guangzhou, Guangdong, China
  3. 3Department of Pharmacy, Guangzhou Women and Children’s Medical Center, Guangzhou Medical University, Guangzhou, Guangdong, China
  4. 4Department of Hematology/Oncology, Guangzhou Women and Children's Medical Center, Guangzhou Medical University, Guangzhou, Guangdong, China
  5. 5Department of Pharmacy, The Fifth Affiliated Hospital of Guangzhou Medical University, Guangzhou Medical University, Guangzhou, Guangdong, China
  6. 6Guangzhou Cord Blood Bank, Guangzhou Women and Children’s Medical Center, Guangzhou Medical University, Guangzhou, Guangdong, China
  1. Correspondence to Dr Jiali Li; lijiali5{at}; Dr Xiaolan Mo; allenmor{at}; Dr Yan Li; firely{at}


Objective Improved understanding of cyclosporine A (CsA) pharmacokinetics in children undergoing allogeneic haematopoietic stem cell transplantation (allo-HSCT) is crucial for effective prevention of acute graft-versus-host disease and medication safety. The aim of this study was to establish a population pharmacokinetic (Pop-PK) model that could be used for individualised therapy to paediatric patients undergoing allo-HSCT in China.

Design, setting and participants A retrospective analysis of 251 paediatric HSCT patients who received CsA intravenously in the early post transplantation period at Women and Children’s Medical Center in Guangzhou was conducted.

Analysis measures The model building dataset from 176 children was used to develop and analyse the CsA Pop-Pk model by using the nonlinear mixed effect model method. The basic information was collected by the electronic medical record system. Genotype was analysed by matrix-assisted time-of-flight mass spectrometry. The stability and predictability of the final model were verified internally, and a validation dataset of 75 children was used for external validation. Monte Carlo simulation is used to adjust and optimise the initial dose of CsA in paediatric allo-HSCT patients.

Results The typical values for clearance (CL) and volume of distribution (Embedded Image) were 14.47 L/hour and 2033.53 L, respectively. The body weight and haematocrit were identified as significant variables for V, while only body weight had an impact on CL. The simulation based on the final model suggests that paediatrics with HSCT required an appropriate intravenous dose of 5 mg/kg/day to reach the therapeutic trough concentration.

Conclusions The CsA Pop-PK model established in this study can quantitatively describe the factors influencing pharmacokinetic parameters and precisely predict the intrinsic exposure to CsA in children. In addition, our dosage simulation results can provide evidence for the personalised medications

Trial registration number ChiCTR2000040561

  • pharmacology

Data availability statement

Data are available on reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Relevant studies on population pharmacokinetic (Pop-PK) in paediatric haematopoietic stem cell transplantation (HSCT) patients are still limited. As a result, determining the optimal dose to achieve individualised medication is challenging, particularly for paediatric patients receiving allo-HSCT in China.


  • Physiological factors and genetic polymorphisms were used as potential covariates in this study to comprehensively explain variations in cyclosporine A (CsA) PK properties. Our findings suggest that body weight and haematocrit are influential factors in the variability of CsA disposition.

  • We proposed an initial dose of 5 mg/kg/day for paediatric patients with variable HCT weighing 16.5 kg, which permits steady-state CsA concentrations to swiftly reach the therapeutic window, minimising potential adverse reaction.


  • The Pop-PK model established in this study can further be adopted to optimise the initial dose of CsA in Chinese paediatrics with allo-HSCT, reducing the frequency of dosage adjustments and serving as a reference for dose guidance in clinical care during the early post-transplantation.


Acute graft-versus-host disease (aGVHD) is the leading cause of non-relapse mortality, and it is also one of the most serious complications of allo-haematopoietic stem cell transplantation (HSCT) surgery. Relevant research shows that patients have a higher incidence of aGVHD after HSCT, ranging from 40% to 60%.1–3 Cyclosporine A (CsA), a conventional calcineurin inhibitor, is known as the first-line drug for preventing aGVHD. However, due to its narrow therapeutic window, which means the narrow range between immunosuppressive and toxic concentrations, as well as the large pharmacokinetic (PK) variability (total drug and unconjugated drug area under the curve can vary by twofold intraindividually and by more than thrice between individuals),4 namely a variety of factors, such as demographics, concomitant medications, liver function and genetics, contribute to unique behaviours in the absorption and disposition of CsA among different individuals, so that patients applying same dose can exhibit a wide range of blood concentrations.5–9 These characteristics may expose patients to an increased risk of toxicity if overdosed or allograft rejection if underdosed. Therefore, CsA requires frequent dose adjustments via therapeutic drug monitoring(TDM) to maintain through concentration in therapeutic range set between 150 and 200 ng/mL,10 11 particularly early after transplantation. Adjusting the dose based on TDM results may cause treatment to be delayed because it is a lagging method that cannot be used for initial dose formulation. Only by identifying the factors that influence CsA PK and developing an individualised initial dose for each patient can CsA’s safety and efficacy be improved.

Population PK (Pop-Pk) is an effective tool to optimise the individual administration of CsA. Despite the fact that Pop-PK research on CsA for organ transplantation has been widely conducted in recent years,12–15 there have been few studies in paediatric patients with haematological diseases.16–18 According to studies, the PK parameters of CsA differ significantly between different organ transplantation types and populations (such as adults and children), resulting in high uncertainty in the safety and effectiveness of CsA PK extrapolation.19 20

Currently, research aimed at CsA Pop-PK in Chinese children receiving allo-HSCT is severely lacking. Li et al were the first to report the Pop-PK of CsA in Chinese children with malignant haematological disorders who underwent allo-HSCT.21 However, an optimised initial dosage of intravenous infusion was not proposed in the existing study. In addition, the genetic factors in the model variables were still contradictory. Furthermore, the size of the participants is small, so it is hard to access the performance of the model by using external validation. As a result, larger sample size studies are required to better demonstrate the potential effects of CsA on PK.

To minimise side effects and formulate an effective initial CsA dose, a paediatric Pop-PK model with good precision, which is used to analyse the CsA PK process, not only provides evidence-based support for the formulation of clinical dosing regimens but also greatly compensates for the lack of empirical medication. Recently, in paediatric Chinese patients who underwent bone marrow transplants, Chen et al proposed that patients weighing 5–30 kg have a greater probability of reaching an effective whole blood concentration at a dose of 6 mg/kg/day CsA, based on the simulated findings of the Pop-PK model.22 This finding is crucial for achieving and maintaining therapeutic concentration. However, the optimal initial dose of CsA in Chinese paediatric patients undergoing HSCT remains unclear, and no reports have been published on optimising the initial dose using the CsA Pop-PK model.

The aim of this study was to establish a Pop-PK model of CsA in Chinese paediatric HSCT patients and quantify the covariates influencing the PK process of CsA by incorporating physiological factors and genetic polymorphisms. In addition, we recommended an optimal initial dose of CsA in children based on predicting individual exposure indicators under specific administration regimens using the empirical Bayesian estimation method.


Participants and data collection

This study retrospectively enrolled paediatric patients (≤18 years old) who underwent allo-HSCT and received CsA for aGVHD prophylaxis at Guangzhou Women and Children’s Medical Center between January 2016 and December 2020. Patients with incomplete information were excluded. The relevant clinical data were collected from electronic medical records, including age, sex and etc (see online supplemental table S1). CsA trough concentrations were measured using an enzyme-multiplied immunoassay technique assay with the Viva-E analyser in the hospital.

Supplemental material

This study was registered with the Chinese Clinical Trial Registry (Reference number: ChiCTR2000040561). Regular contact with members and representatives of Guangzhou Women and Children’s Medical Center made the outcome measures of this study clear for patients in making informed decisions about treatment. Patients were not involved in setting the research question, the outcome measures or the design or implementation of the study.

Dosage regimen

CsA was typically administered twice daily, with an initial dose of 3 mg/kg/day administered intravenously on day 1 of transplantation. Patients with thalassaemia frequently received CsA at an initial dose of 1.5 mg/kg/day once every 12 hours as an intravenous drip infusion 10 days before transplantation, then adjusted to 3 mg/kg/day once every 12 hours as an intravenous drip infusion on day 1 of transplantation. The CsA dosage was adjusted based on TDM data acquired the morning after the last dose to maintain a trough blood concentration of 150–200 ng/mL.

DNA extraction and genotyping

Before the transplant, a 2 mL of the patient’s peripheral venous blood was taken, and genomic DNA was extracted using a DNA extraction kit (TIANGEN, Beijing, China). The ABCB1 (rs1045642, rs3842, rs1128503, rs34800935), CYP3A4 rs2242480, CYP3A5 rs776746, POR rs17685 and NR1I3 rs2307424 genotypes were determined using the Agena Bioscience MassARRAY technology. Hardy-Weinberg equilibrium testing is performed using a χ2 goodness-of-fit (GOF) test.

Pop-PK modelling

The data were divided into training and validation sets at a ratio of 7:3 based on the patient’s transplant time. The Pop-PK model of CsA in Chinese allo-HSCT paediatric patients was developed using the training dataset (n=176), applying the nonlinear mixed effect model method in Phoenix (V.7.0; Certara). Interindividual and intraindividual variability of PK parameters was evaluated by using the first-order conditional estimation-extended least squares algorithm. Clearance (CL) and volume of distribution (Embedded Image) were the fundamental PK parameters of the structure-function model.

Structure model

The additive and proportional models that reflect the interindividual variation of parameters were compared, as were various error models such as additive, proportional and mixed (additive and proportional). The objective function value (OFV) was criterion for model selection. The results of the comparison are shown in online supplemental table S2. Interindividual variability of PK parameters was calculated using an exponential model and expressed as follows: (Eq. 1)

Embedded Image(1)

Where Embedded Image is the estimated value of the PK parameter for i-th individual, Embedded Image is typical population value of the PK parameter and Embedded Image is the random effect of i-th individual. Embedded Image follows the normal distribution with a mean of 0 and variance of ω2. The residual variability was described using a proportional model and expressed as follows: (Eq. 2)

Embedded Image(2)

Where Embedded Image is the observation and C is the individual predicted concentration. ε obeys the Gaussian distribution with a mean of 0 and a variance of Embedded Image .

Covariate analysis

After imputing the missing values (for categorical variables, filled with mode and abnormally distributed continuous variables, filled with median), a stepwise method was used to screen covariates in the basic model. The change in OFV was used as the selection criterion. During the forward inclusion process, adding covariates that significantly (p<0.01) decreased the OFV by more than 6.63. Retaining covariates that significantly(p<0.001) increased OFV by more than 10.83 during the backward-exclusion procedure.

Model validation

The GOF plots were used for preliminary evaluation. The visual predictive check (VPC) was also used to verify the prediction performance of the final model by using the K-means method to divide the time periods and then generating 1000 sets of simulated data. And the bootstrap method was performed to assess the robustness of the final model. There is no difference in dataset and settings between internal validation and final model. The validation set (n=75) was substituted into the final model to calculate the predicted concentration of CsA and analyse the correlation between the predicted concentration and the observations. The data were compared by calculating individual prediction error (IPE%), mean prediction error (MPE%), mean absolute percentage error (MAPE%), root mean square error (RMSE%) as follows: (Eq. 3), (Eq. 4), (Eq. 5), (Eq. 6)

Embedded Image(3)

Embedded Image(4)

Embedded Image(5)

Embedded Image(6)

where Embedded Image is the individual predicted concentration, Embedded Image is the observation, n is the sample size. The percentage of IPE within 20% and 30% (IF20, IF30) can be used to further measure the precision and accuracy of the Pop-PK model. Smaller MPE%, MAPE% and RMSE% values suggest less bias and more accuracy.


Based on the validated final model, 1000 Monte Carlo simulations were performed to generate the steady-state trough concentration on day 7 after transplantation in typical patients (weight 16.5 kg) with different HCT levels (10%, 30%, 50%) while receiving continuous intravenous infusions of different CsA dosages (2 mg/kg/day, 3 mg/kg/day, 4 mg/kg/day and 5 mg/kg/day).The recommended dose was presented in combination with the target therapeutic concentration range (150 μg/L–250 μg/L).


Characteristics of patients

The current Pop-PK modelling analysis involved 865 whole blood CsA concentrations in samples from 251 paediatric HSCT recipients. The demographics of the patients are shown in table 1. Detailed characteristics for patients in training dataset and validation dataset were shown in online supplemental table S3. The frequency distributions of allele and genotype are shown in the table 2.

Table 1

Demographic data of 251 paediatric patients

Table 2

Pharmacogenetics and HWE analyses

Pop-PK modelling

From the data obtained in 176 paediatric patients, the PK characteristics of CsA were best described by the one-compartment model with first-order absorption and elimination. The results of the stepwise method for covariate screening are shown in online supplemental table S4. They showed that body weight (BW) and HCT were both significantly influenced the Embedded Image, while only BW was significantly influenced the CL. Other covariates were not found to have statistical significance on PK parameters. The following equations (Eq. 7) and (Eq. 8)) describe the CL and Embedded Image for final covariate models:

Embedded Image(7)

Embedded Image(8)

The parameter estimations of the final model are shown in table 3. From the above equations, the final relationship describing the CL and Embedded Image can be deduced: the CL and Embedded Image of CsA increased as the patient’s BW increased. On the contrary, as the patient’s HCT increased, the Embedded Image of CsA declined.

Table 3

Population pharmacokinetic parameters of CsA and bootstrap results

The GOF diagrams of the basic model and the final model are shown in figures 1 and 2, respectively. In contrast, the final model fits better.

Figure 1

Goodness-of-fit graphs for the basic model. (A) Observed concentration versus population prediction; (B) observed concentration versus individual prediction; (C) conditional weighted residuals versus population prediction; (D) conditional weighted residuals versus time. CWRES, conditional weighted residuals.

Figure 2

Goodness-of-fit graphs for the final model. (A) Observed concentration versus population prediction; (B) observed concentration versus individual prediction; (C) conditional weighted residuals versus population prediction; (D) conditional weighted residuals versus time. CWRES, conditional weighted residuals.

Model validation

The results of the VPC of the final model are displayed in figure 3. The success rate of bootstrapping (n=1000) was 100%. The estimated typical values of CL and Embedded Image were comparable to the median resulting from the bootstrap method; additionally, they are both within the 95% CI (14.47 (13.00 to 15.73) for CL and 20.33 (1657.30 to 2416.52) for Embedded Image). The detailed results were reported in table 3. The plot of observed CsA concentration versus individual predicted CsA concentration in the validation dataset (n=75) is displayed in figure 4. The predicted concentration of CsA in paediatric patients has a statistical correlation with the observed concentration (r=0.863, p<0.01). In addition, the IF20 and IF30 of the final model were 64.29% and 84.21 %, the MPE%, MAPE% and RMSE% were 3.12%, 14.46% and 25.24%, respectively. It confirmed that the model developed in this study can better predict the plasma concentration of CsA intravenous infusion in children undergoing allo-HSCT.

Figure 3

Prediction-corrected visual predictive check (n=1000). The blue dots are the observed concentrations. The red dash lines reflect the median and 95th percentile of observations, and the semitransparent red shading area represents the 90% CI of the predicted median. The red solid line reflects the 5th percentile of observed concentrations. The semitransparent blue shading area represents 90% CI of the predicted 5th and 95th percentile based on prediction. DV, observed data; IVAR, time after the first dose.

Figure 4

Plot of observed CsA concentration versus individual predicted CsA concentration in the validation group. CsA, cyclosporine A.


The steady-state trough concentrations on day 7 after transplantation in typical patients with different haematocrit levels and different dosages are presented in figure 5. Compared with an initial dose of 2–4 mg/kg/day, the probability of reaching the effective whole blood concentration in paediatric patients weighing 16.5 kg who were given an initial dose of 5 mg/kg/day was higher.

Figure 5

Simulated CsA trough concentration of the typical patient with different haematocrit levels under different dosage regimens (the shaded area is the therapeutic range: 150–250 μg/L; the horizontal line in the middle of the box represents the median values, and the whiskers represent the maximum and minimum values). CsA, cyclosporine A.


In this study, we successfully established a Pop-PK model of CsA in Chinese children with HSCT and provided an optimised initial dose for reaching the target concentration of 150–250 ng/mL. The typical values of CL and Embedded Image in the final model were 14.47 L/hour and 2033.53 L, respectively, which were comparable to the results of Ni et al and Wang et al. The comparison of CsA pop-PK parameters in paediatric patients with previous studies is shown in online supplemental table S5.

It has been proven that weight is the main covariate for PK parameters.20 23 24 Given that there is a statistically significant correlation between BW and age, it is always difficult to distinguish PK differences from age-related factors and size-related factors due to collinearity, we, therefore, decided to retain only BW in the final model. Currently, dosage selection based on weight is commonly used in the clinical treatment of children, which was confirmed in our findings.

Additionally, the HCT was considered to have a significant impact on PK parameters in prior research since CsA is an 11-amino acid lipophilic cyclic polypeptide, widely distributed in red blood cells and highly binding to proteins in plasma after administration.19 25 26 Fanta et al constructed a study in paediatric patients who underwent renal transplants, and statistically significant associations were observed between V/F and HCT.27 In this study, we discovered a similar result: an inverse correlation was found between HCT and V. Considering the high lipophilicity of CsA, changes in HCT may affect its distribution between blood and adipose tissue. Lower HCT levels increase CsA distribution to fat, resulting in a larger apparent volume of distribution,28 implying that as HCT declines, the amount of unbound CsA distributed into peripheral tissues rises, as does the risk for toxicity. To adjust the dose regimen, physicians must regularly monitor patients’ HCT levels.16

CsA is a substrate for the CYP3A4 and CYP3A5 enzymes and is carried out of cells by P-glycoprotein (encoded by ABCB1/MDR1). The heterogeneity in CYP3A4, CYP3A5 and ABCB1 expression may contribute to interindividual variability in CsA PK. Several studies have examined the relationship between CsA and genetic variants, particularly CYP3A5*3, because mutation frequency is high in the Chinese population. Tao et al indicated that carriers of CYP3A5 *3/*3 have higher Cmax of cyclosporine than carriers of CYP3A51*/*3.29 Moreover, there is evidence that the C0/D of CsA in CYP3A5 expressers is significantly lower than that in CYP3A5 non-expressers in paediatric renal transplant recipients.30 In addition, P450 oxidoreductase (POR), which is required for catalytic activity, was discovered to influence tacrolimus CYP3A5 activity in CYP3A5 expressers in Elens et al’s research. Also, they found that POR*28/*28 increases CYP3A4 activity for CsA in CYP3A5 nonexpressers with a CYP3A4*22 loss-of-function allele, resulting in lower dose-adjusted predose concentrations of CsA.31 Moreover, pregnane X receptor (PXR, encoded by NR1I2)is a critical nuclear receptor that regulating the expression of metabolic enzymes. Several studies reported that the NR1I2 genotype may influence PXR expression because multiple SNPs in NR1I2 have been linked to CYP3A4, CYP3A5 and ABCB1 activity.32 33 Our laboratory has indicated that the −24622A>T in the 5’-untranslated region and the −24446C>A in exon 1 of the NR1I2 gene increased PXR activity in Han Chinese in previous studies.34 Although genetic polymorphisms as one of the important factors contributing to variations in CsA PK,35 36 the findings are still controversial. Anglicheau et al found no correlation between polymorphisms in the CYP3A5 and MDR1 genes and the PK parameters of CsA in 106 kidney transplant recipients.37 Li et al reported that CYP3A5 and ABCB1 polymorphisms were not found to have significant effects on the PK process of CsA in a total of 86 Chinese children who received allo-HSCT.21 Relevant outcomes have been reported in the study constructed by Xue et al.38 Similarly, none of the genetic polymorphisms in ABCB1, CYP3A4, CYP3A5, POR and NR1I3 were significant covariates in the PK of CsA in our study.

Evidence shows that genotypes such as CYP3A5*3 could influence the correlation of CYP3A4/MDR1/NR1I2 genetic polymorphisms with tacrolimus concentration.39 Additionally, Qiu et al demonstrated that the effect of MDR1 polymorphisms on CsA PK in Chinese renal transplant recipients may be masked by CYP3A polymorphisms.40 As a result, we presumed that the contribution of SNPs to interindividual variance in CsA PK variation may be obscured by the involvement of metabolising enzyme polymorphisms in the model. Other reasons leading to inconsistent results may be attributed to differences in sample size, age, liver function, ethnicity and CsA concentration detection methods.

Furthermore, the final model was applied to imitate the optimal initial dose of CsA in order to effectively prevent rejection by using Monte Carlo simulation. Even though CsA is generally given by continuous intravenous infusion before the transplant at an initial dose of 3 mg/kg per day, there is still contradiction due to the lower level of evidence and lack of established data support.11 According to our simulation, the 3 mg/kg per day commonly used in clinical might be too low for children who received HSCT. The best initial dosage regimen for typical patients weighing 16.5 kg with different HCT should be 5 mg/kg per day so that the steady-state trough concentration of CsA in the early stage of transplantation can largely reach the treatment window. Recently, researchers have reported that CsA concentrations in the early post-transplantation period were significantly linked to severe aGVHD in paediatric transplant recipients. For instance, Izumi et al demonstrated that inadequate exposures to CsA can be a considerable risk factor for developing aGVHD.41 Bianchi et al indicated a significant linkage between the higher incidence of aGVHD and patients with CsA levels <200 µg/L in the first 10 days of allo-HCST. Moreover, their results have shown that only a small proportion of patients reached a CsA level >201 µg/L on day 0 after receiving the starting dose of CsA was 3 mg/kg /day intravenously on day −3 with concomitant drugs. These results highlight the urgent need for optimal adjustments of CsA dosing in children to maintain therapeutic CsA levels above 195 µg/L in the first 10 days of allo-HCST to prevent the onset of aGVHD.42 Because dose escalation clinical trials in children are typically difficult to perform, our simulation, which is similar to the observation made in the above research, has a high reference value for clinical work aimed at reaching target concentrations of CsA as quickly as possible in Chinese paediatric patients who underwent allo-HSCT.

Finally, in comparison to Ti et al’s article published in 2019, their pop-PK model only adopts internal validation, which is insufficient for extrapolation to future paediatric patients. Our research compensates for the lack of sample size in their study by enrolling 251 paediatric HSCT patients, allowing us to obtain a reasonable prediction model by validating it externally. Even though they demonstrated that CYP3A4*1G and eGFR are both significant covariates impacting CL, the association between renal function and genetic polymorphism on CsA PK in children, particularly with HSCT, remains disputed.17 43 Thus, the similarities and variances in our research findings are meaningful for formulating individualised medication and serve as a reference for future research. The retrospective design of our study is one of its limitations; in order to confirm the recommended dosage in the future, a prospective study will be required. Despite the limitations listed above, this study may assist with individualised and precision medicine in CsA therapy.


In summary, the accuracy and constancy of the one-compartment Pop-PK model with first-order absorption and elimination to evaluate CsA exposure in Chinese children receiving allo-HSCT were verified internally and externally. The influences of BW and HCT were included in our model, which was further used to perform a simulation of different dosing regimens and HCT levels to obtain an optimal initial dose. We suggested an intravenous infusion dose of 5 mg/kg/day every 12 hours may be relatively appropriate for Chinese paediatric recipients of allo-HSCT for better prevention of aGVHD and improvement of prognosis.

Data availability statement

Data are available on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and was approved by Ethics Committee of Guangzhou Women and Children’s Medical Center (Guangzhou Women and Children’s Medical Center, Issue No.63100(2020), EC). Participants gave informed consent to participate in the study before taking part.


We thank the physicians and nurses from the department of haematology/oncology for their cooperation. At the same time, we dedicated to the 20th anniversary of School of Pharmaceutical Sciences, Sun Yat-sen University.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • HF, XW and WZ contributed equally.

  • Contributors MH, YL, XM and JL designed and planned the study. HF, XW, WZ, HQ and TFC performed the experiments. HF, XW, WZ, SL, HJ and YL collected and collated the data. HF, XW and WZ conducted the analyses. HF, XW, WZ, MH, YL, XM and JL wrote and edited the manuscript. JL as the guarantor of the overall content.

  • Funding This research was supported by grants from Guangdong Provincial Key Laboratory of Construction Foundation (No.2017B030314030, No.2020B1212060034), Guangzhou Science and Technology Bureau program (No. 202102010237, No. SL2022B03J01382), Natural Science Foundation of Guangdong Province (No. 2021A1515011308), Guangdong Pharmaceutical Association Program (No. 2021A35), Traditional Chinese Medicine Bureau of Guangdong Province (No. 20201302), Hospital Pharmacy Youth Talent Project of Hospital Pharmacy Special Committee of Chinese Pharmaceutical Society (No.CPA-Z05-ZC-2021-003) and Wu Jieping Medical Foundation (No.320.6750.2022-20-26).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.