Abstract
Background
Individual pharmacokinetic variability is a driver of poor tuberculosis (TB) treatment outcomes. We developed a method for measurement of rifampin concentrations by urine colorimetry and a mobile phone photographic application to predict clinically important serum rifampin pharmacokinetic measurements in children treated for TB.
Methods
Among spiked urine samples, colorimetric assay performance was tested with conventional spectrophotometric and the mobile phone/light box methods under various environmental and biologic conditions. Urine rifampin absorbance (Abs) was then determined from timed specimens from children treated for TB in Tanzania, and compared to serum pharmacokinetic measurements collected throughout the dosing interval.
Results
Both the mobile phone/light box and spectrophotometry demonstrated excellent correlation across a wide range of urine rifampin concentrations (7.8–1000 mg/L) in intra- and interday trials, 24-hour exposure to ambient light or darkness, and varying urinalysis profiles (all r ≥ 0.98). In 12 Tanzanian children, the urine mobile phone/light box measurement and serum peak concentration (Cmax) were significantly correlated (P = .004). Using a Cmax target of 8 mg/L, the area under the receiver operating characteristic curve was 80.1% (range, 47.2%–100%). A urine mobile phone/light box threshold of 50 Abs correctly classified all patients (n = 6) with serum measurements below target.
Conclusions
The urine colorimetry with mobile phone/light box assay accurately measured rifampin absorbance in varying environmental and biological conditions that may be observed clinically. Among children treated for TB, the assay was sensitive for detection of low rifampin serum concentrations. Future work will identify the optimal timing for urine collection, and operationalize use in TB-endemic settings.
Keywords: colorimetry, pediatric tuberculosis, pharmacokinetics; rifampin, therapeutic drug monitoring
Individual pharmacokinetic variability of rifampin compromises tuberculosis treatment outcomes. Measurement of rifampin absorbance by urine colorimetry and mobile phone application was reproducible and correlated with serum pharmacokinetics, which could deliver personalized dosing to children treated for tuberculosis in endemic settings.
In 2018, an estimated 10.0 million people were diagnosed with tuberculosis (TB), the world’s leading killer from a single infectious disease [1]. The majority of TB diagnoses were from 30 high-burden countries, which also suffered the highest TB mortality rates [2]. Children < 15 years of age accounted for 11% of global cases [1, 3]. Despite the childhood burden, only recently have efforts prioritized strategies for pediatric-specific TB treatment optimization. In 2014, the World Health Organization (WHO) revised weight-based dosing recommendations for children increase pharmacokinetic exposures of first-line anti-TB drugs [4]. Despite the revised WHO doses, we found among a study of children treated for TB in rural Tanzania that peak serum concentrations (Cmax) were below the target for all studied; for example, the median Cmax of rifampin was 2.17 µg/mL (interquartile range, 0.59–4.61 µg/mL) where target ranges are typically 8–24 µg/mL [5]. In addition to mg/kg dose, malnutrition was also significantly associated with lower rifampin and isoniazid Cmax in multivariate modeling.
Increasing evidence suggests that suboptimal pharmacokinetic exposure is an important driver of TB treatment failure, acquired drug resistance, and death [6–8]; thus, measurement of individual pharmacokinetic exposures and personalized dose correction is likely to improve pediatric TB outcomes [9, 10]. Such an approach, termed therapeutic drug monitoring (TDM), is typically performed by utilizing serum collected at scheduled times during the dosing interval to estimate Cmax or other measurements such as the total area under the concentration–time curve (AUC), which are predictive of microbial kill and prevention of acquired resistance for the concentration dependent anti-TB drugs [11]. Despite recommendations for the use of TDM in children with severe TB disease or those with concern for malabsorption, TDM has been largely underutilized in TB-endemic countries due to cost; availability of the analytic tools (mass spectrometry or high-performance liquid chromatography); cold chain requirement for serum transport; and, specifically relevant to malnourished children, the necessity of multiple blood draws [12, 13]. A promising alternative methodology to overcome these barriers to more widespread implementation of TDM has been to employ urine as the analytic specimen and spectrophotometric testing (colorimetry) as a readout that may eliminate the need for mass spectrometry or chromatography [9].
Of the first-line anti-TB drugs, rifampin demonstrates the most pharmacokinetic variability, has the greatest capacity for dose increase without dose-related toxicity, and exerts both bactericidal and sterilizing activity [14]. Resistance to rifampin, such as can be acquired from suboptimal treatment, necessitates an entirely different multidrug treatment regimen of more prolonged duration and with higher rates of treatment failure. Thus, rifampin represents an important candidate for novel TDM assays. Previously, we have demonstrated that urine colorimetric testing could be used for prediction of rifampin serum pharmacokinetic targets among adults treated for TB in Botswana [15]. In an effort to deliver this personalized treatment approach closer to the point of care, we hypothesized that the spectrophotometric analysis could be replicated by use of a mobile phone photographic application and standardized light environment utilizing a 3D printed light box. Additionally, we compared the spectrophotometric and mobile phone/light box approaches with urine samples from children undergoing a comprehensive pharmacokinetic study from rural Haydom, Tanzania, to determine the assay’s ability to predict clinically important serum thresholds concentrations in children from a region with a high prevalence of malnutrition.
METHODS
Rifampin Urine Colorimetric Assay
The Sunahara method for extraction of total rifamycins from body fluid was adapted for use with urine and the mobile phone/light box apparatus [16]. In brief, 500 μL of 100 mM phosphate buffer (pH 7, Sigma-Aldrich) was added to 1 mL of urine, followed by 100 μL of isoamyl alcohol (Fisher Scientific). Each sample was vortexed for 20 seconds and centrifuged at 4000 rpm at room temperature for 5 minutes or until there was clear separation between layers. The aqueous phase was transferred in triplicate to a 96-well plate (100 μL) for conventional spectrophotometry, and 750 µL to a cuvette for mobile phone/light box measurement. Absorbance (Abs) was then measured using a spectrophotometer (Bio-Rad, IMark) at 475 nm as well the mobile phone/light box, as described below. Calibration curves were generated using seven 2-fold serial dilutions of 1000 mg/mL stock rifampin (Sigma-Aldrich) with urine pooled from 3 healthy donors.
Mobile Phone/Light Box
Following rifamycin extraction, cuvettes containing processed urine were placed into a 3D printed light box containing a central LED bulb (Figure 1). Images of each sample were recorded with an iPod touch (Apple), and the ColorMeterRGB Colorimeter mobile application was used to measure absorbance. The “Area Mode” toggle was placed around the center of color density and the “magenta” parameter was selected, which was previously determined to be the most accurate setting using urine with known rifampin concentrations (data not shown). Measurements were then recorded for analysis.
Figure 1.
Mobile phone/light box apparatus.
Comparison Between Mobile Phone/Light Box and Conventional Spectrophotometry Utilizing Stock Urine and Leftover Clinical Urine Samples
The performance of the mobile phone/light box and conventional spectrophotometric methods was tested under a variety of conditions utilizing both stock urine and leftover clinical samples. To determine intraday and interday variability, urine stock with known concentrations of rifampin was created using healthy donor urine in the same manner as the calibration curves. The colorimetric assay was then performed and absorbance measured using both the mobile phone/light box and spectrophotometer, and the urine stock was stored at –20°C. The assay and measurements were then repeated at 3.5, 24, and 27.5 hours after urine collection. To test the effects of light exposure, urine stock with known concentrations of rifampin was prepared and 1-mL aliquots were stored at room temperature for 24 hours on a windowsill with direct sunlight cycle, ambient indoor light, or complete darkness prior to completion of the assay and absorbance measurements. Finally, to simulate a range of biological variability in pH, proteinuria, and other patient-dependent factors, 10 distinct urine samples from hospitalized patients with variable urinalysis results were obtained from the University of Virginia Medical Center clinical laboratory. Known concentrations of rifampin were added to the urine samples and serial dilutions were performed in the same manner as described above, and the colorimetric assay was performed with absorbance measured using the mobile phone/light box and spectrophotometer.
Urine Colorimetric Assay Comparison to Serum Pharmacokinetic Measurements Among Children Treated for TB in Tanzania
Urine samples and serum pharmacokinetic data were derived from pediatric TB patients participating in a prospective pharmacokinetic study among children in rural Haydom, Tanzania, in a protocol approved by the institutional review boards for human subjects research at the University of Virginia and the Tanzania National Institute for Medical Research with reference NIMR/HQ/R.8a/Vol.IX/2120 (ClinicalTrials.gov identifier NCT03559582). Inclusion criteria were children < 15 years of age with confirmed or probable TB as defined by the National Institutes of Health Consensus Case Definitions for TB research in children and having been started on first-line TB treatment. All parents/guardian signed written informed consent and children > 7 years of age provided assent. Each patient received weight-based dosing of isoniazid, pyrazinamide, ethambutol, and rifampin per Tanzanian national guidelines. Blood collection occurred at 14 days after medication start to allow for rifampin autoinduction of clearance and steady-state kinetics. Blood was collected at 1, 2, and 6 hours after observed medication administration, and serum was stored at –80°C until batch shipment to the University of Florida Infectious Diseases Pharmacokinetics Laboratory where rifampin concentrations were measured by a validated liquid chromatography–tandem mass spectrometry assay, including prior confirmation of drug milligram content in the administered fixed-dose combination pills [5]. Urine samples were also collected on the same day as blood collection and children were encouraged to void prior to medication administration. The first urine voided after medication administration was collected and timed at 2 hours after administration for children able to adhere. Urine volumes were recorded and then aliquoted and stored at –80°C and removed from light until batch shipment to the University of Virginia. Samples were then selected for the urine colorimetric testing from participants to include those with the greatest range of serum Cmax values. The urine colorimetric assays were then run in triplicate on thawed samples as described and absorbance was measured with both spectrophotometer and mobile phone/light box.
Statistical Analysis
The mobile phone/light box and conventional spectrophotometry methods were compared using linear correlation. As a log-linear relationship was observed with the rifampin calibration curves, mobile phone/light box measurements were log-transformed prior to analysis. Demographic characteristics of study participants were calculated using simple frequencies. Rifampin Cmax and estimated AUC over 24 hours (AUC0–24) were determined by noncompartmental analysis using Phoenix WinNonlin version 8.0 (Certara USA, Princeton, New Jersey), which allowed AUC0–24 estimates in those with 6-hour values within the elimination phase. Urine assay results and serum rifampin concentrations were compared with linear correlation. Serum rifampin Cmax and AUC0–24 were categorized as at or below target levels (8 mg/L and 35 mg × hour/L, respectively) based on previous work that found that concentrations below these breakpoints predicted poor long-term treatment outcome [13], and comparisons between groups were made with the Wilcoxon-Mann-Whitney test. Receiver operating characteristic (ROC) curves were generated, the area under the ROC curve was calculated, and 95% confidence intervals were determined using 2000 bootstrap replicates. Youden J statistic was used to determine the optimal assay cutoff, and sensitivity and specificity were calculated using this value. Data were analyzed in R software (version 3.6.1, http://r-project.org), with ROC analysis performed using the pROC package [17].
RESULTS
Comparison Between Mobile Phone/Light Box and Conventional Spectrophotometry for Control and Leftover Urine Samples
An 8-point calibration curve was created and accuracy of the mobile phone/light box and spectrophotometer was compared under the differing environmental and biological conditions utilizing the control urine. Calibration curves under standard conditions are shown in Figure 2. Using the spectrophotometer to measure absorbance, extraction of rifampin using the Sunahara method demonstrated a linear relationship between rifampin concentration from 7.8 to 1000 mg/L (r = 1), which is similar to previous reports [15]. The mobile phone/light box method produced a log-linear curve from 7.8 to 1000 mg/dL (r = 0.99).
Figure 2.
Calibration curve for the urine colorimetric assay using the mobile phone/light box (A) and conventional spectrophotometer (B) to measure absorbance (Abs) in stock urine samples.
Performance of the mobile phone/light box and spectrophotometric methods under the various environmental conditions are displayed in Table 1. Both the mobile phone/light box and spectrophotometer measurements demonstrated excellent correlation with urine rifampin concentrations in intra- and interday trials (all r ≥ 0.99). Exposure to sunlight somewhat reduced the consistency of measurements from the spectrophotometer (mean r = 0.94) but not the mobile phone/light box (mean r = 0.99), while both spectrophotometry and mobile phone/light box measurements were unaltered with 24-hour exposure to ambient light and darkness (all r ≥ 0.98). For the testing of urine from 10 hospitalized patients, pretesting urinalyses presented a variety of abnormal findings, including abnormal coloration, cloudy appearance, elevated specific gravity, proteinuria, glucosuria, and ketonuria (Table 1). Both the mobile phone/light box and spectrophotometer measurements continued to demonstrate acceptable correlation with urine rifampin concentration in the hospitalized patient samples (mean, r = 0.99 for mobile phone/light box and r = 0.98 for spectrophotometer).
Table 1.
Correlation Between Urine Rifampin Concentration and Measurements by Light Box/Mobile Phone and Spectrophotometer in Varying Conditions in Stock Urine and Leftover Patient Samples
| Mean r Value | ||
|---|---|---|
| Testing Condition | Light Box/Mobile Phone | Spectrophotometer |
| Intraday and interday variability (n = 2) | ||
| Intraday | 0.99 | 1.00 |
| Interday | 0.99 | 0.99 |
| Environmental exposure (n = 2) | ||
| Ambient light | 0.99 | 1.00 |
| Sunlight | 0.99 | 0.94 |
| Darkness | 0.99 | 0.98 |
| Patient samples with varying urinalysis resultsa (n = 10) | ||
| Patient samples | 0.99 | 0.98 |
All samples were run in triplicate.
aPatient sample urinalyses ranges: color, yellow to dark; appearance, clear to cloudy; specific gravity, 1.012 to 1.031; protein, absent to 1+; pH 5.0–7.0; glucose, absent to 3+; ketones, absent to large; bilirubin, all samples absent; urobilinogen, 0.2–1.0; blood, all samples absent; red blood cells, absent to 3–5 per high-power field; nitrite, all samples absent; white blood cells, absent to 3–5 per high-power field; casts, absent to rare; epithelial cells, all samples absent.
Urine Colorimetric Assay Comparison to Serum Pharmacokinetic Measurements Among Children Treated for TB in Tanzania
Twelve urine samples were selected from children treated for TB that had serum pharmacokinetics reflective of the widest range of serum rifampin concentrations. The median age was 3 years, 5 (42%) were female, and 11 (92%) had moderate or severe malnutrition, with median weight-for-age, height-for-age, and body mass index (BMI) z scores of –3.6, –2.6, and –1.8, respectively. All children were human immunodeficiency virus negative. Median rifampin dose was 12.7 mg/kg. Serum rifampin concentration curves are shown in Figure 3. Six of 12 patients (50%) had a rifampin Cmax greater than the target of 8 mg/L, and 5 of 11 (45%) had an AUC0–24 greater than the target of 35 mg × hour/L (AUC0–24 was unable to be calculated in 1 patient). All patients with a rifampin Cmax > 8 mg/L also reached target AUC0–24. Time of maximal rifampin concentration was 2 hours in the majority of patients (9 of 12 [75%]).
Figure 3.
Serum rifampin concentrations over time among children treated for tuberculosis in Tanzania.
Urine samples were collected a median of 3.3 hours after rifampin dose. The spectrophotometer was unable to quantify absorbance in 1 of 12 samples (8.3%), whereas the mobile phone/light box produced measurements for all samples tested. As shown in Figure 4, the urine assay results correlated with serum rifampin Cmax (r = 0.76 and r = 0.84 for mobile phone/light box and spectrophotometer, respectively) and AUC0–24 (r = 0.81 and r = 0.91) (all P values < .005). Participants who reached the target rifampin Cmax had a higher median mobile phone/light box readout than patients with a low Cmax (65.3 vs 32, P = .09; Figure 5). Similar trends were seen for AUC0–24 (P = .18) and when using a spectrophotometer to measure absorbance (P = .18 for Cmax and P = 0.35 for AUC0–24).
Figure 4.
Correlation between peak serum rifampin concentration (Cmax) and the urine colorimetric assay with absorbance (Abs) measured by mobile phone/light box (A) and conventional spectrophotometry (B), and correlation between rifampin area under the concentration–time curve from 0 to 24 hours (AUC0–24) and absorbance measured by mobile phone/light box (C) and conventional spectrophotometry (D) among children treated for tuberculosis in Tanzania.
Figure 5.
Urine assay results among children treated for tuberculosis in Tanzania with peak serum concentration (Cmax) of rifampin > 8 or ≥ 8 mg/L (A) and with rifampin area under the curve from 0 to 24 hours (AUC0–24) > 35 or ≥ 35 mg × hour/L (B), using the mobile phone/light box to measure absorbance (Abs).
ROC curves for the mobile phone/light box urine assay to identify patients with subtarget serum rifampin concentrations are shown in Figure 6. Using a Cmax target of 8 mg/L, the area under the ROC curve was 80.1% (range, 47.2%–100%); for an AUC0–24 target of 35 mg × hour/L, the area under the ROC curve was 77% (40%–100%). A mobile phone/light measurement of 50 Abs was determined to be the optimal cutoff by Youden J statistic. Using this threshold, the assay detected patients with a rifampin Cmax < 8 mg/L with 100% sensitivity and 67% specificity, and detected patients with rifampin AUC0–24 < 35 mg × hour/L with 100% sensitivity and 60% specificity. The classification table using this cutoff is shown in Table 2; the mobile phone/light box correctly distinguished 6 of 6 patients (100%) with rifampin Cmax below the target level, and correctly identified 4 of 6 (67%) with Cmax within the target range. ROC analysis was similar for AUC0–24, which showed sensitivity and specificity of 100% and 60%, with correct classification of 6 of 6 patients with a low AUC0–24 and 3 of 5 patients who had AUC0–24 within the target range.
Figure 6.
Receiver operating characteristic (ROC) curves for the mobile phone/light box assay to identify patients with peak serum rifampin concentration (Cmax) < 8 mg/L (A), and patients with area under the rifampin serum concentration curve from 0 to 24 hours (AUC0–24) < 35 mg × hour/L (B) among children treated for tuberculosis in Tanzania.
Table 2.
Classification Table for the Mobile Phone/Light Box Assay to Detect Patients With Low Rifampin Peak Serum Concentration Among Tanzanian Children
| Rifampin Cmax | ||
|---|---|---|
| Above or Below Assay Cutoff | < 8 mg/L | ≥ 8 mg/L |
| Below 50 Abs | 6 | 2 |
| Above 50 Abs | 0 | 4 |
Abbreviations: Abs, absorption; Cmax, peak serum concentration.
DISCUSSION
The main findings of this study were that a mobile phone/light box readout was comparable to spectrophotometry for quantifying urine rifampin concentrations across a range of environmental and biological conditions, and that even among urine specimens collected at a single time point in the dosing interval from children treated for TB in rural Tanzania, the urine colorimetry adequately predicted clinically relevant serum pharmacokinetic targets. Unexpectedly, the mobile phone/light box methodology out-performed the spectrophotometer under certain environmental conditions that may be relevant to clinical and laboratory conditions. While we do not suggest that urine colorimetry should replace serum quantification for rigorous pharmacokinetic studies, the use of the urine-based assay as a semi-quantitative test to predict serum concentrations below a threshold that would trigger dose increase is nevertheless promising for delivery of personalized dosing to children in TB-endemic settings.
This work differed from others’ and our prior studies of urine colorimetry for detection of anti-TB drug concentrations in several important ways [9, 15]. To our knowledge, this is the first study of urine colorimetry in children treated for TB, and was completed among a cohort from a high-TB-burden country in a population more generalizable to the global pediatric TB burden. As a substrate for pharmacokinetic measurement, urine is more readily available from children in whom blood draws may be limited, and our series of environmental and biological parameter testing reveals urine can be utilized for colorimetric detection of rifampin under a wide range of host-related and logistical conditions. This study further described the optimization of the assay for the detection of rifampin by making use of a mobile phone/light box design as a proxy for spectrophotometric analysis. While the chemical processing steps require a certain laboratory capacity, the precision of results produced utilizing the mobile phone application and a standardized light environment represent an important proof of concept for directing future experimental design in simplifying measurement, reducing instrumental footprint, and ultimately delivering personalized dosing directly in real-world clinical settings.
Although the mobile phone/light box urine colorimetric assay performed well in this cohort of patients using an assay threshold of 50 Abs, and correctly identified all patients with below-target rifampin serum Cmax and AUC, the cutoff requires validation in a separate cohort [18]. Additionally, further examination of factors that may influence the effectiveness of urine rifampin measurement as a surrogate for serum rifampin pharmacokinetics should be explored, such as TB disease severity, renal function, and other body composition or drug metabolism–related enzymatic maturation through childhood [19]. For example, of the 12 patients tested in this study, the 2 patients in whom the urine assay incorrectly predicted serum rifampin levels had more severe malnutrition and anemia, suggesting that disease severity may impact utility of the assay. Finally, the spectrophotometer was unable to measure absorbance in 1 sample, despite successful measurement by the mobile phone/light box. This patient had the greatest serum rifampin exposure of the samples tested here, suggesting that further refinement of the protocol may be necessary, such as inclusion of dilution procedures in such a case.
There were limitations to our approach that impact the immediate application of our findings. While the serum pharmacokinetic data included complete calculations of Cmax and AUC based on an accepted blood sampling strategy, the urine was collected only at a single time point. Consequently, there may be other strategies for timing or volume of urine collection during the dosing interval that better correlate with serum pharmacokinetics [20]. We are currently enrolling for such a study among adults with TB in the United States and children with TB from Tanzania, which will expand our understanding of optimal collection times and conditions in varying clinical settings. Last, given the evolving international consensus toward higher dosing strategies and the significant individual pharmacokinetic variability of rifampin, we tested rifampin first for the mobile phone/light box methodology. While other anti-TB drugs may also be fitting to apply the urine colorimetric approach, there may be differing drug class–related limitations to assay development, such as the need to further modify processing steps [10].
In summary, children treated for TB in endemic settings experience suboptimal pharmacokinetics despite WHO’s revised dosing recommendations and consequently are at risk of treatment failure. The urine colorimetric assay for rifampin detection with mobile phone/light box readout performed well under various biological and environmental conditions, and was sensitive in predicting attainment of clinically important serum pharmacokinetic targets among malnourished children from rural Tanzania. Future work will aim to develop a fully point-of-care test for personalized dosing of anti-TB drugs for children in the most vulnerable of settings.
Notes
Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (grant numbers U01AI115594 to S. K. H. and R01AI137080 to C. V., and training grant T32AI007046 to D. V. A.).
Potential conflicts of interest. All authors: No reported conflicts of interest.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1. World Health Organization. Global tuberculosis report 2019. Available at: http://www.who.int/tb/publications/global_report/en/. Accessed December 14, 2019.
- 2. Gandhi NR, Nunn P, Dheda K, et al. Multidrug-resistant and extensively drug-resistant tuberculosis: a threat to global control of tuberculosis. Lancet 2010; 375:1830–43. [DOI] [PubMed] [Google Scholar]
- 3. Marais BJ, Gie RP, Schaaf HS, et al. Childhood pulmonary tuberculosis: old wisdom and new challenges. Am J Respir Crit Care Med 2006; 173:1078–90. [DOI] [PubMed] [Google Scholar]
- 4. World Health Organization. Guidance for national tuberculosis programmes on the management of tuberculosis in children, 2nd ed. Geneva, Switzerland: WHO, 2014. [PubMed] [Google Scholar]
- 5. Justine M, Yeconia A, Nicodemu I, et al. Pharmacokinetics of first-line drugs among children with tuberculosis in rural Tanzania. J Pediatr Infect Dis Soc 2018; 10.1093/jpids/piy106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ramachandran G, Kumar AK, Kannan T, et al. Low serum concentrations of rifampicin and pyrazinamide associated with poor treatment outcomes in children with tuberculosis related to HIV status. Pediatr Infect Dis J 2016; 35:530–4. [DOI] [PubMed] [Google Scholar]
- 7. Pasipanodya JG, McIlleron H, Burger A, et al. Serum drug concentrations predictive of pulmonary tuberculosis outcomes. J Infect Dis 2013; 208:1464–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Sekaggya-Wiltshire C, von Braun A, Lamorde M, et al. Delayed sputum culture conversion in tuberculosis-human immunodeficiency virus–coinfected patients with low isoniazid and rifampicin concentrations. Clin Infect Dis 2018; 67:708–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Zentner I, Modongo C, Zetola NM, et al. Urine colorimetry for therapeutic drug monitoring of pyrazinamide during tuberculosis treatment. Int J Infect Dis 2018; 68:18–23. [DOI] [PubMed] [Google Scholar]
- 10. Ghimire S, Van’t Boveneind-Vrubleuskaya N, Akkerman OW, et al. Pharmacokinetic/pharmacodynamic-based optimization of levofloxacin administration in the treatment of MDR-TB. J Antimicrob Chemother 2016; 71:2691–703. [DOI] [PubMed] [Google Scholar]
- 11. Alffenaar J-WC, Gumbo T, Dooley KE, et al. Integrating pharmacokinetics and pharmacodynamics in operational research to End TB [manuscript published online ahead of print 28 September 2019]. Clin Infect Dis 2019. doi:10.1093/cid/ciz942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Cruz A, Starke J. Monitoring treatment of childhood tuberculosis and the role of therapeutic drug monitoring. Indian J Pediatr 2019; 86:732–9. [DOI] [PubMed] [Google Scholar]
- 13. Alffenaar JC, Heysell SK, Mpagama SG. Therapeutic drug monitoring: the need for practical guidance. Clin Infect Dis 2019; 68:1065–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Alsultan A, Peloquin CA. Therapeutic drug monitoring in the treatment of tuberculosis: an update. Drugs 2014; 74:839–54. [DOI] [PubMed] [Google Scholar]
- 15. Zentner I, Schlecht HP, Khensouvann L, et al. Urine colorimetry to detect Low rifampin exposure during tuberculosis therapy: a proof-of-concept study. BMC Infect Dis 2016; 16:242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Sunahara S, Nakagawa H. Metabolic study and controlled clinical trials of rifampin. Chest 1972; 61:526–32. [DOI] [PubMed] [Google Scholar]
- 17. Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 2011; 12:77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Sackett DL, Haynes RB. The architecture of diagnostic research. 2002; 324:3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Zvada SP, Denti P, Donald PR, et al. Population pharmacokinetics of rifampicin, pyrazinamide and isoniazid in children with tuberculosis: in silico evaluation of currently recommended doses. J Antimicrob Chemother 2014; 69:1339–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Stott KE, Pertinez H, Sturkenboom MGG, et al. Pharmacokinetics of rifampicin in adult TB patients and healthy volunteers: a systematic review and meta-analysis. J Antimicrob Chemother 2018; 73:2305–13. [DOI] [PMC free article] [PubMed] [Google Scholar]