Introduction Individualised treatment through therapeutic drug monitoring (TDM) may improve tuberculosis (TB) treatment outcomes but is not routinely implemented. Prospective clinical studies of drug exposure and minimum inhibitory concentrations (MICs) in multidrug-resistant TB (MDR-TB) are scarce. This translational study aims to characterise the area under the concentration–time curve of individual MDR-TB drugs, divided by the MIC for Mycobacterium tuberculosis isolates, to explore associations with markers of treatment progress and to develop useful strategies for clinical implementation of TDM in MDR-TB.
Methods and analysis Adult patients with pulmonary MDR-TB treated in Xiamen, China, are included. Plasma samples for measure of drug exposure are obtained at 0, 1, 2, 4, 6, 8 and 10 hours after drug intake at week 2 and at 0, 4 and 6 hours during weeks 4 and 8. Sputum samples for evaluating time to culture positivity and MIC determination are collected at days 0, 2 and 7 and at weeks 2, 4, 8 and 12 after treatment initiation. Disease severity are assessed with a clinical scoring tool (TBscore II) and quality of life evaluated using EQ-5D-5L. Drug concentrations of pyrazinamide, ethambutol, levofloxacin, moxifloxacin, cycloserine, prothionamide and para-aminosalicylate are measured by liquid chromatography tandem-mass spectrometry and the levels of amikacin measured by immunoassay. Dried blood spot on filter paper, to facilitate blood sampling for analysis of drug concentrations, is also evaluated. The MICs of the drugs listed above are determined using custom-made broth microdilution plates and MYCOTB plates with Middlebrook 7H9 media. MIC determination of pyrazinamide is performed in BACTEC MGIT 960.
Ethics and dissemination This study has been approved by the ethical review boards of Karolinska Institutet, Sweden and Fudan University, China. Informed written consent is given by participants. The study results will be submitted to a peer-reviewed journal.
Trial registration number NCT02816931; Pre-results.
- clinical pharmacology
- public health
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: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
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.
Strengths and limitations of this study
To our knowledge, this is a novel study approach which fully investigates the distribution of drug exposure in relation to minimum inhibitory concentration (MIC) for Mycobacterium tuberculosis (Mtb) isolates from patients with multidrug-resistant tuberculosis (TB) along with biomarkers (eg, time to positivity), culture conversion and the clinical scoring tool TBscore II to assess treatment outcome.
We used a translational approach with experts from research centres across the world to design a study protocol including both MIC determinations, drug exposure estimation using novel technology, as well as microbiological and clinical surrogate markers for improvement, to enable strategies for therapeutic drug monitoring use in TB treatment.
The patients’ drug exposure will be compared with individual Mtb MICs, exploring pharmacokinetics–pharmacodynamics indices in multidrug-resistant TB treatment.
Dried blood spot as a method to simplify blood sampling by finger prick instead of venous sampling will be investigated.
A limitation of the study is the low target number of patients for inclusion, due to a laborious and costly study protocol, which might partly be compensated for by using pharmacometric modelling.
Despite programmatic management of tuberculosis (TB), the incidence of multidrug-resistant TB (MDR-TB), defined as Mycobacterium tuberculosis (Mtb) resistant to rifampicin and isoniazid, is steadily increasing.1 Inconsistent treatment, due to poor treatment adherence, lack of drugs, as well as subtherapeutic dosing are contributing factors. For many TB drugs, the administered dose is not predictive of the drug exposure and clinical effect in the patient.2 A hollow-fibre study indicated that pharmacokinetic variability may be an underestimated cause of drug resistance development3 and low drug concentrations in the treatment of drug-susceptible TB have been associated with poor outcome in some prospective studies.4 5
Therapeutic drug monitoring (TDM) is a strategy to personalise treatment by measuring systemic drug levels in blood/plasma as a guide for individual dose adjustments.6 Specifically for infectious diseases, the drug efficacy not only depends on the drug exposure but also on the susceptibility level of the bacteria, the minimum inhibitory concentration (MIC).7 The MIC is defined as the lowest concentration of a drug that inhibits visible growth of bacilli and should be exceeded to cure the infection.8 TDM has been recommended during MDR-TB treatment by several organisations, for example, the Infectious Disease Society of America.9 The pharmacokinetic studies that have been performed have shown that subtherapeutic drug levels in TB treatment are common, although with conflicting results regarding association between drug exposure and treatment outcome.10–12 However, studies on MDR-TB are limited and only a few studies have included drug concentrations as well as the individual MICs of the bacteria.13–15
An optimal estimation of drug exposure (ie, area under the concentration versus time curve (AUC)) traditionally requires multiple venous blood samples, often followed by prompt centrifugation and sample storage at −80°C. A simplified strategy for collection and transportation of blood samples needed for TDM would aid its implementation in clinical practice. Dried blood spot (DBS) allows minimal blood sampling by capillary finger pricking on filter paper, which can be transported without a cold chain, simplifying transportation and storage.16 DBS is a well-established and validated method, but has only been evaluated for a few second-line TB drugs, for example, moxifloxacin and linezolid.16–18 A clinical implementation of DBS could enable TDM for TB treatment in remote areas and reduce costs.16
There are scarce data regarding drug exposure and treatment outcome in MDR-TB treatment. Assessing end-of-treatment outcome in MDR-TB studies is cumbersome due to long treatment durations. Other strategies include using interim endpoints such as time to positivity (TTP) in liquid culture media, a surrogate of bactericidal activity,19 and sputum culture conversion after 2 or 3 months of treatment,20 the latter commonly used in drug efficacy studies. A clinical composite scoring system, TBscore II, can be used as a surrogate marker for TB disease severity and to predict failure.21 Patients’ quality of life can be objectified using the validated EQ-5D-5L tool assessing five different dimensions (mobility, self-care, typical activity, pain/discomfort and anxiety/depression), an often overlooked tool in clinical treatment studies.
China has the second highest burden of MDR-TB in the world and has existing resources to perform TDM, thus making it an ideal setting for pharmacokinetic/pharmacodynamic (PK/PD) studies. The overall incidence of TB in China was 895 000 TB cases in 2017, of which 8.2% were MDR-TB.22
We describe a new comprehensive approach to TDM studies, assessing drug exposure, individual MICs as well as clinical outcome markers. The primary aim of the study is to investigate the distribution of AUC/MIC and Cmax/MIC for MDR-TB drugs during MDR-TB treatment in China. Secondary aims are to analyse AUC/MIC in relationship to markers of clinical improvement, such as sputum culture conversion, TTP, TBscore II, body mass index (BMI) and qualitative measures of well-being (European Quality of Life scale, EQ-5D-5L). Signs of acquired resistance are assessed by investigating changes in MICs and genetic mutations, during the first 3 months of treatment. A clinical implementation of DBS as well as a method of simultaneous MIC determination are assessed to simplify the use of TDM in clinical practice.
Methods and analysis
We are conducting a prospective cohort study of TB drug exposure and MICs in patients with MDR-TB in Xiamen, China. This is a joint project between the School of Public Health Fudan University Shanghai, Department of Medicine Karolinska Institutet, Department of Pharmaceutical Biosciences University of Uppsala and the Public Health Agency of Sweden, in collaboration with the Centre for Disease Control (CDC) in Xiamen. The study protocol conforms with the Strengthening the Reporting of Observational Studies in Epidemiology Statement for cohort studies.23
Patient and public involvement
The original study protocol by the coauthors was changed by reducing the number of blood samples after feedback from patients included in a pilot study. The result of the study can be obtained in Mandarin on request at the Xiamen CDC. Patients were not involved in the recruitment and the conduct of the study.
The study is carried out in Xiamen, Fujian region in Southeast China, where the incidence of TB in 2016 was 42.4 cases/100 000 inhabitants and of the 1661 confirmed cases that year, there were 28 MDR-TB patients (1.7%).24
The study hospital is the designated TB hospital in Xin Ling, Xiamen, a large teaching hospital with a specialised TB ward with 105 beds as well as a negative pressure ward (12 beds) with specialised TB physicians and nurses. Recruitment of patients is performed by the Xiamen CDC, which also keeps a screening log. Patients are routinely admitted for 2 months of in-patient treatment. The study is registered at ClinicalTrials.gov (NCT02816931) and opened on 17 April 2016.
A total number of 30 fully evaluable patients, according to the criteria below, will be included.
Consenting adults (≥18 years) with a verified diagnosis of pulmonary MDR-TB, by routine drug susceptibility testing (DST) admitted to the Xin Ling TB Hospital, Xiamen.
Eligible for and consent to MDR-TB treatment in Xiamen.
Patients admitted to the intensive care unit
Confirmed extensively drug-resistant TB by DST
Ongoing medication for MDR-TB (ie, five active drugs or more for more than 1 day.)
The overall study outline is shown in figure 1. After informed consent, a completed inclusion questionnaire with demographic and clinical information, baseline blood and sputum samples are collected from the patient by a designated study nurse. Treatment regimens adhere to WHO guidelines and are adjusted following DST results. The first day of MDR-TB treatment is defined as ‘day 0’. Clinical data are collected at inclusion, day 2, weeks 1, 2, 4, 8 and week 12 after treatment initiation. Adverse events, routine blood tests and vital signs are closely monitored to ensure the safety of the study patients. The final treatment outcome is recorded at the end of MDR-TB treatment.
Drug concentrations of second-line TB drugs are measured at steady state at 2, 4 and 8 weeks after treatment initiation. In order to estimate the AUC, multiple blood samples for drug concentration analysis (ie, rich sampling) are collected at week 2 (0, 1, 2, 4, 6, 8 and 10 hours after drug intake). A sparse-sampling strategy is applied at weeks 4 and 8 (0, 4 and 6 hours). Whole blood samples are simultaneously collected and pipetted directly onto DBS cards. Finger prick blood samples are collected on DBS cards at week 2 (0, 4 and 6 hours after drug intake) (figure 2). The drug concentrations in plasma and DBS will be analysed using liquid-chromatography tandem mass spectrometry (LC-MS/MS) and immunoassay.25 In order to assess delayed absorption and possible interactions, information of concomitant drugs is noted in the medical records. Additionally, detailed food intake is noted by the patient in a diary on the days of blood sample collection. Pharmacometric modelling and simulation will be performed in the analysis phase.
Sputum is collected at days 0 and 2 and weeks 1, 2, 4, 8 and 12 in order to evaluate changes in TTP. Whole genome sequencing (WGS) and MIC determination for TB drugs (pyrazinamide (PZA), ethambutol, levofloxacin, moxifloxacin, ofloxacin, cycloserine, ethionamide, para-aminosalicylic acid (PAS), amikacin, kanamycin, rifampicin and isoniazid) are performed at baseline and for any positive culture after 1 month or more of treatment, to assess development of acquired resistance. Time to sputum culture conversion is defined as the day from starting treatment until the day of the first of two consecutive negative sputum cultures, collected at least 30 days apart.
Disease severity is estimated and monitored using inflammatory markers such as C reactive protein and erythrocyte sedimentation rate, presence of cavity on chest X-ray as well as the total score obtained in TBscore II, based on the following variables; cough, dyspnoea, chest pain, anaemia (pale lower conjunctivas), BMI and mid-upper arm circumference.21 Quality of life during the first 3 months of treatment is estimated using the validated EQ-5D-5L-5L (Mandarin version).26 The patients are followed up until treatment completion or loss to follow-up, whichever occurs first, through information accessible from the TB-registry, Xiamen CDC.
Drug concentration measurement
A combined assay for drug concentration analysis using LC-MS/MS is under development at the Xiamen CDC to measure the plasma concentrations of PZA, ethambutol, levofloxacin, moxifloxacin, cycloserine, prothionamide and PAS.27 28 The second-line injectable drug amikacin will be analysed with a commercial immunoassay kit (amikacin assay kit, Beckman Coultier). The collected venous blood samples will be centrifuged at 3500 rpm for 10 min within 1 hour from sampling. Aliquots of plasma are then frozen at −70°C awaiting analysis.
A puncture from the DBS card will be immersed in extraction solution as previously described18 and analysed through LC-MS/MS. Plasma concentrations will be compared with blood concentrations collected by DBS.18
Microbiology: TTP, DST and MIC
All microbiological tests are carried out at a biosafety laboratory level 3 at the Xiamen CDC, apart from routine DST testing which is partly performed in local hospital laboratories and WGS analysis performed at the Public Health Agency of Sweden.
Time to culture positivity
Sputum samples are treated according to Chinese National standards based on a WHO recommended protocol.29 In short, NALC-NaOH is added to the sputum, then shaken using a vortex shaker until fully liquefied, followed by incubation for 15 min in room temperature. Phosphate buffer is added to reach a total volume of 45 mL, after which the solution is centrifuged for 15 min at 3000 g. The supernatant is removed and 1 mL phosphate buffer is added. Finally, 0.5 mL of the solution is transferred using a pipette to two labelled MGIT tubes, gently tilted for 1 min and incubated in the BACTEC MGIT 960 machine at 37°C. TTP is done in duplicate and is automatically recorded by the BACTEC MGIT 960.
Drug susceptibility testing
Routine DST is performed according to Chinese National Guidelines with the proportion method on Lowenstein-Jensen (LJ) medium, according to WHO’s recommendations.30
Simultaneous MIC determination of Mtb using TREK Sensititre broth microdilution plates
Since MIC testing in BACTEC MGIT is labour-intense and time-consuming, a high throughput broth microdilution plate has been developed to test up to 12 antibiotics simultaneously in Middlebrook 7H9 on a single MIC plate. We have designed a custom-made Sensititre plate for the drugs used at the study site, with concentration ranges including wild-type isolates, manufactured by Thermofisher (figure 3). The reference isolate H37Rv ATCC 27294 is always included in each test run and compared with previously published quality control target ranges for each drug.31 The Thermofisher Sensititre MYCOTB plate is used for internal validation of ethionamide (range 0.5 mg/L–32 mg/L), which was not stable in the pretrial validation of the customised plate.
After positive culture of Mtb in the BACTEC MGIT 960 and recording of TTP, the isolates are stored at −80°C awaiting MIC determination. After thawing and reculturing on LJ media, bacterial suspensions are prepared from Mtb isolates which are no more than 2 weeks old. Bacterial suspension together with Middlebrook 7H9 stock solution are then added to each well, according to the manufacturer’s instructions.32 The plates are sealed and left to incubate in 37°C. Manual reading is done after 10–21 days, depending on growth, assisted by an inverted mirror (figure 4).
Due to specific pH requirements, PZA susceptibility is determined using BACTEC MGIT 960 PZA Susceptibility Test, with a pH of 5.9 as previously described.33 In short, colonies of Mtb no older than 2 weeks are suspended in Middlebrook 7H9 broth with phosphate-buffered saline. A bacterial suspension, corresponding to a McFarland turbidity of 0.5, is prepared. Following the test protocol provided by the manufacturer (Becton Dickinson Biosciences, Sparks, Maryland, USA), the suspension is thereafter diluted 1:5 (inoculum A), from which a 1:10 diluted control is prepared (inoculum B). From inoculums A and B, 0.5 mL is then added to the MGIT 960 PZA tubes and the proportional growth control tube, respectively. The tubes are incubated in 37°C and read automatically by the BACTEC MGIT.
Whole genome sequencing
All baseline study isolates, as well as any viable isolate after at least 1 month of treatment or more, will be analysed using WGS to detect new resistance mutations. In brief, DNA is extracted from Mtb LJ cultures using a chloroform/CTAB (N-cetyl-N,N,N-trimethyl ammonium bromide)-based protocol,34 transported to Sweden and sequenced using Illumina technology (Illumina, San Diego, California, USA).
Mapping to a set of resistant genes from the Mtb H37Rv reference genome (GeneBank accession nr NC_000962.3) and extraction of variants are performed in CLC Genomics Workbench 8 (Qiagen, Hilden, Germany) using the following filters: minimum coverage: 10x; minimum count of reads calling variants: 2; minimum frequency of reads calling variants: 10%; minimum frequency of reads calling variants in each direction: 5%. In addition, pyro-error variants in homopolymer regions with a minimum length of 3 and a frequency below 0.8 are removed. The remaining variants are then compared with our in-house database of resistance mutations.
Data analysis plan
Regular study monitoring is performed quarterly by the Swedish and Chinese researchers as well as biweekly reports from the study site. Study data from the case report forms are entered in EpiData with a range check by two independent researchers and results compared for coherence.
The distribution of AUC/MIC and Cmax/AUC will be presented and visualised in graphs. The agreement between drug exposure in plasma and DBS will be assessed. An exploratory analysis of the PK/PD indices for key TB drugs, such as fluoroquinolones, in relation to sputum culture conversion, TTP and changes in TBscore II during treatment will be performed. Pharmacometric modelling will assess the relationship between dose, concentrations and effect and population models will be applied, using the non-linear mixed-effects modelling software NONMEM (Icon Development Solutions, Ellicot city). Time-to-event data with censoring will be analysed using the Cox regression model, whereas binary outcomes will be analysed with logistic regression and continuous outcome with linear regression, if data are normally distributed. The validated Chinese value set of the quality of life tool EQ-5D-5L will be used and quality of life perception described.
For analysis of trends in drug exposure over time, the dependent nature of the data will be taken into account using mixed-effect models. Missing values will not be imputated. A p value of <0.05 will be considered as statistically significant.
Information of potential confounders such as age, gender, BMI, concomitant treatment and comorbidities and disease severity assessed by TBscore II will be collected and evaluated during data analysis. As this is a feasibility and hypothesis-generating study, no power calculation was performed.
Ethics and dissemination
The study is performed in accordance with Good Clinical Practice and the Declaration of Helsinki. Ethical approval was obtained.
Prior to the study start, a designated study team of nurses, doctors and laboratory staff participated in training workshops of the study protocol and ethical considerations, led by the main study investigators from Fudan University and Karolinska Institutet. Patients are informed about the study orally and in writing, including information that neither study participation nor study termination will result in any changes in their treatment. An informed consent is signed or, in the case of illiteracy, a fingerprint given under observation by a witness. A travel grant to enable follow-up is offered to all the study participants. The sum was set so as not to create financial motivation to accept study participation. All patients are treated according to standard of care at the designated MDR-TB hospital and patients’ safety ensured by regular monitoring.
Extensive blood sampling is a sensitive issue in China and should be avoided in severely ill patients. Therefore, the number of blood samples collected have been reduced to a minimum for the estimation of the AUC. Moreover, extensive blood sampling should be minimised in severely ill patients. An intravenous line is inserted to minimise patient discomfort. The increased sputum sample collection is a potential hazardous risk for other patients, hospital and laboratory staff. Therefore, biosafety and awareness training, as well as an upgrade of biosafety equipment, have been implemented.
We aim to present our data in international conferences and to publish our results in a peer-reviewed journal, regardless of study results. Any significant protocol amendments will be reported to the respective ethical boards in Sweden and China.
In this prospective observational cohort study, we present a comprehensive, translational approach to TDM studies in MDR-TB, likely to be of benefit in future trials in the area. Multiple blood sampling and individual MIC determination will enable exploration of AUC/MIC for MDR-TB drugs, a poorly investigated research area. In a key study, the level of peak drug concentrations and AUC of PZA, rifampicin and isoniazid strongly influenced treatment outcome, although no comparison with the Mtb MICs was performed.4 Bacterial MIC has also been found to influence treatment outcome of patients with MDR-TB, with a sixfold increased odds of failure when comparing MIC of gatifloxacin of ≤0.25 mg/L to 1 mg/L, although both concentrations are still regarded as susceptible.35 There are very few tentative targets for most second-line TB drugs, although an AUC/MIC >100 for fluoroquinolones has been suggested.36 So far, the tentative targets have not been correlated with clinical outcome. To our knowledge, this is the first study to assess both AUCs and individual MICs for the most commonly used second-line drugs in MDR-TB regimen.
Not only have optimal PK/PD targets not been established, the critical concentrations used for DST are poorly validated.37 MIC determination provides more information of the level of the resistance, but it has the drawback of being time-consuming. Fortunately, commercial MIC plates are available, facilitating fast MIC determination and will be assessed in this study.38 When interpreting individual MICs, the clinician should bear in mind the innate variability may be up to ±1 twofold MIC dilution step, but often less in a meticulous laboratory, which impacts on PK/PD indices estimates. Furthermore, when results of MIC testing are reported, it is important to note that there is no reference method for MIC testing of Mtb.
A limitation of this study protocol is the limited target number of included patients, common with other studies in the field, mainly due to MDR-TB incidence in Xiamen and costly and cumbersome sampling procedures. This may not allow us to perform extensive analysis of PK/PD indices in relation to treatment outcome, especially since all patients are treated with multiple drugs. However, pharmacometric mathematical modelling and simulation enables reduced sample sizes in clinical trials39 and may partly compensate for the limited number of study patients in our study. Also, we use markers of early clinical improvement using microbiological surrogate measurements, such as TTP and sputum culture conversion, due to the nature of long treatment periods and follow-up. Sputum culture conversion is an imperfect surrogate marker of the final treatment outcome but it is, nevertheless, commonly used in clinical trials and is a sign of clinical improvement.40
This study will provide useful insights of the PK/PD relations in MDR-TB treatment and highlight the importance of individualised treatment, taking both drug concentrations and MICs and innovative surrogate markers of improvement into account. With a simultaneous method for drug concentration analysis and blood sample collection simplified through DBS, TDM would be more feasible in clinical practice, including low-resource and high-endemic settings. We hope that this study will inspire future randomised controlled studies for TDM for both drug susceptible and MDR-TB, including treatment groups such as children, pregnant women, diabetic and HIV-infected patients who are prone to altered PK characteristics.
We thank all the study patients, the staff at Xiamen CDC and the Xiamen TB hospital, as well as Brian Davies for language revision.
Contributors LDF, KN, YH, RZ, XZ, JW, JP, USHS, EE, J-WA, SH, BX, TS and JB designed the study. RK, WC, YH, CH, YL, YG, JW, XZ, TS, LDF, KN and MM developed the plan for the microbiological part of the study. USHS, EE, TS, XZ, YH, LDF, KN, JB, JK, WC and J-WA developed the pharmacokinetic part of the study. LDF wrote the first draft of the manuscript together with KN and YH. All authors contributed and approved the final version.
Funding This work was supported by the Swedish Heart Lung Foundation (grant number 20150508), the Swedish National Research Council (grant number 540-2013-8797) and the National Research Foundation of China (grant number 81361138019).
Competing interests None declared.
Patient consent Not required.
Ethics approval Ethical Review Board of Stockholm (approval number EPN: 2015/646 31/1) and the Institutional Review Board of the School of Public Health, Fudan University, China (approval number IRB 2015-09-0565).
Provenance and peer review Not commissioned; externally peer reviewed.