Clinical Infectious Diseases MAJOR ARTICLE

Transcription

1 Clinical Infectious Diseases MAJOR ARTICLE Implications of Failure to Routinely Diagnose Resistance to Second-Line Drugs in Patients With Rifampicin- Resistant Tuberculosis on Xpert MTB/RIF: A Multisite Observational Study Karen R. Jacobson, 1 Marinus Barnard, 2 Mary B. Kleinman, 3 Elizabeth M. Streicher, 4 Elizabeth J. Ragan, 1 Laura F. White, 5 Ofer Shapira, 6 Tania Dolby, 2 John Simpson, 2 Lesley Scott, 7 Wendy Stevens, 7 Paul D. van Helden, 4 Annelies Van Rie, 8,9 and Robin M. Warren 4 1 Section of Infectious Diseases, Boston University School of Medicine, Massachusetts; 2 National Health Laboratory Service, Cape Town, South Africa; 3 Infectious Disease Prevention and Health Services Bureau, Prevention and Health Promotion Administration, Maryland Department of Health and Mental Hygiene, Baltimore; 4 Department of Science and Technology/National Research Foundation Centre of Excellence in Biomedical Tuberculosis Research/South Africa Medical Research Council for Molecular Biology and Human Genetics, Stellenbosch University, Tyberberg; 5 Department of Biostatistics, Boston University School of Public Health; 6 Department of Cancer Biology, Dana-Farber Cancer Institute, Cambridge, Massachusetts; 7 Department of Molecular Medicine and Haematology, School of Pathology, University of the Witwatersrand and National Health Laboratory Service, National Priority Program, Johannesburg, South Africa; 8 Department of Epidemiology, University of North Carolina, Chapel Hill; 9 Department of Epidemiology and Social Medicine and Epidemiology for Global Health Institute, University of Antwerp, Belgium Background. Xpert MTB/RIF (Xpert) detects rifampicin-resistant tuberculosis (RR-tuberculosis), enabling physicians to rapidly initiate a World Health Organization recommended 5-drug regimen while awaiting second-line drug-susceptibility test (DST) results. We quantified the second-line DST results time and proportion of patients potentially placed on suboptimal therapy. Methods. We included RR-tuberculosis patients detected using Xpert at the South African National Health Laboratory Services (NHLS) of the Western Cape between November 2011 and June 2013 and at Eastern Cape, Free State, and Gauteng NHLS between November 2012 and December We calculated time from specimen collection to phenotypic second-line DST results. We identified isoniazid and ethionamide resistance mutations on line probe assay and performed pyrazinamide sequencing. Results. Among 1332 RR-tuberculosis patients, only 44.7% (596) had second-line DST for both fluoroquinolones and second-line injectable: 55.8% (466 of 835) in the Western Cape and 26.2% (130 of 497) in the other provinces. Patients with smear negative disease and age 10 years were less likely to have a result (risk ratio [RR] = 0.72; 95% CI, and RR = 0.49; 95% CI, ). Median time to second-line DST was 53 days (range, 8 259). Of the 252 patients with complete second-line DST, 101 (40.1%) potentially initiated a suboptimal regimen: 46.8% in the Western Cape and 25.3% in the other provinces. Conclusions. Many South Africans diagnosed with RR-tuberculosis by Xpert initiate a suboptimal regimen, with information to adjust therapy available in half of all patients after a median 7 weeks. Algorithm completion and time delays remain challenging. Keywords. multidrug-resistant tuberculosis; tuberculosis; South Africa; drug susceptibility; drug resistance. Improved control of the drug-resistant tuberculosis epidemic requires early diagnosis and initiation of effective therapy to stop the spread and improve treatment outcomes. In 2010, the World Health Organization (WHO) endorsed the molecular test Xpert MTB/RIF (Xpert; Cepheid, California) as the initial diagnostic for tuberculosis and multidrug-resistant (MDR) tuberculosis [1]. The Xpert assay rapidly detects Mycobacterium tuberculosis (MTB) and rifampicin (RIF) resistance, a marker of MDR tuberculosis, enabling physicians to simultaneously identify Received 6 September 2016; editorial decision 20 January 2017; accepted 9 February 2017; published online February 12, Correspondence: K. R. Jacobson, Section of Infectious Diseases, Boston Medical Center, Department of Medicine, Boston University School of Medicine, 801 Massachusetts Ave, Crosstown 2nd floor, Boston, MA (kjacobso@bu.edu). Clinical Infectious Diseases 2017;64(11): The Author Published by Oxford University Press for the Infectious Diseases Society of America. All rights reserved. For permissions, journals.permissions@oup.com. DOI: /cid/cix128 tuberculosis patients and triage those who require MDR tuberculosis therapy. Use of the Xpert assay can result in same-day diagnosis and treatment initiation when placed at the point of care [2]. In South Africa, where Xpert tests are performed in 207 smear microscopy laboratories across the country, time from sample collection to result reporting is 4 6 days [3, 4]. MDR tuberculosis patients should initiate a treatment regimen containing 4 5 drugs to which their infecting strain is likely susceptible to have good treatment response [5, 6]. Because the Xpert assay only reports susceptibility to RIF, subsequent drug-susceptibility testing (DST) is required to determine the isolate s resistance profile to first- and second-line drugs. South Africa has the fifth highest burden of MDR tuberculosis globally [7]. In 2011, the South African Department of Health (DOH) implemented Xpert as the initial tuberculosis diagnostic test. If the initial Xpert assay is positive for RIFresistant tuberculosis, the South African algorithm requires 1502 CID 2017:64 (1 June) Jacobson et al

2 that a second sample be evaluated by the MTBDRplus line probe assay (LPA; Hain Lifescience GmbH, Nehren, Germany) to confirm MDR tuberculosis (testing for isoniazid [INH] and confirming RIF) and by culture-based DST to determine resistance to second-line injectable drugs (amikacin [AMI] or kanamycin [KAN]) and fluoroquinolones (ofloxacin [OFX]) [8]. The national guidelines instruct that patients start the standardized MDR tuberculosis regimen within 5 days of the Xpert result [8, 9]. In line with the WHO guidelines, the standardized regimen consists of 5 drugs: KAN, moxifloxacin (MOXI), ethionamide (ETH), pyrazinamide (PZA), and terizidone. Our aim in this study was to determine the proportion of patients evaluated using second-line DST (culture for AMI or KAN plus OFX susceptibility), the time until availability of second-line DST results, the variables associated with lack and/or delay of second-line DST results, and the proportion of RIFresistant tuberculosis patients potentially placed on empiric suboptimal therapy, defined as fewer than 4 likely effective drugs. METHODS Patient Population Western Cape Province A retrospective observational study included all patients diagnosed with RIF-resistant tuberculosis by Xpert performed at the National Health Laboratory Services (NHLS), Cape Town, South Africa, from November 2011 to June With the exception of a few hospitals and clinics, all samples from Western Cape patients with presumed tuberculosis are sent to NHLS. According to the Western Cape DOH tuberculosis guidelines, 2 sputum specimens are sent to the NHLS: 1 for Xpert and 1 for LPA and culture-based DST if Xpert indicates RIF resistance. The request for 2 samples at screening is unique to this province and thus an interesting comparator for the other provinces. Using the NHLS database, we identified all patients with RIF-resistant Mycobacterium tuberculosis on Xpert and linked patients first and second sputum specimens by matching name, age, sex, and location. We abstracted acid-fast bacilli (AFB) sputum grade, RIF and INH LPA resistance results, and culture-based second-line DST results and result date. The Stellenbosch University s Health Research Ethics Committee and the Boston University Medical Center Institutional Review Board (IRB) approved the study. Eastern Cape, Free State, and Gauteng Province A prospective observational study included a sample of adults (aged 18 years) diagnosed with RIF-resistant MTB by Xpert between November 2012 and December 2013 in the NHLS of 3 South African provinces (Eastern Cape, Free State, and Gauteng). These provinces follow national DOH guidelines, which require collection of 1 sputum specimen at baseline for Xpert testing and a second subsequent sample for additional testing (LPA and culture-based DST) only if RIF-resistant tuberculosis is diagnosed on the first specimen. We collected data on age, sex, AFB sputum grade, RIF and INH LPA resistance results, culture-based second-line DST results, and result date through review of the NHLS database and clinic files. The Stellenbosch University Health Research Ethics Committee and the University of North Carolina IRB approved the study. Specimen Preparation and Drug-Susceptibility Testing At NHLS, specimens are decontaminated in a standard method. An initial aliquot is used for auramine smear; a second aliquot for MTBDRplus LPA (version 2) [10] is used to assess the rpob gene (RIF resistance) and the katg and inha promoter genes (INH and ETH resistance) [11]; and a third aliquot is used for BACTEC MGIT 960 system culture (BD Diagnostics Systems, Maryland). DST is performed on all positive cultures using the 1% proportion method on 7H11 agar slants that contain OFX (2 µg/ml), AMI (4 µg/ml), or KAN (6 µg/ml) [12]. Secondline injectable resistance is defined as AMI or KAN resistance, given their high levels of cross resistance [13]. OFX resistance is used as the indicator for MOXI, the fluoroquinolone used in the standard MDR tuberculosis regimen [14]. Drug-resistant tuberculosis isolates have been collected from NHLS and deposited in our laboratory at Stellenbosch University since For this study, we found stored isolates from identified patients that were viable and performed PZA testing through amplification of the pnca gene, as previously described [15]. PZA testing is not performed at NHLS and is not routinely available. When results from the MTBDRplus assay were unavailable, targeted gene sequencing was also performed in our laboratory to confirm RIF resistance and determine resistance to INH and ETH. Terizidone DST is not performed due to its instability in culture [16]. Statistical Analyses Descriptive statistics were used to calculate the proportions of patients with second-line DST results available and distribution of drug-resistance profiles. We calculated the proportion of patients potentially on a suboptimal regimen as those with 3 or fewer effective drugs based on second-line DST, MTBDRplus, and PZA sequencing, divided by all of the patients with those results. The χ 2 test was used to examine the association between patient characteristics and having second-line DST results. To assess for changes over time in acquisition of second-line DST, we used visual inspection to look for linear trends and a generalized additive model with logit link to look for nonlinear trends in Western Cape Province and separately in the other provinces. Logistic regression was used to study potential correlates with second-line DST. Turnaround time for second-line DST results was calculated by subtracting the date of specimen collection and/or Xpert test request from the date of NHLS DST result Second Line Susceptibility RIF-Resistant Tuberculosis CID 2017:64 (1 June) 1503

3 reporting. The χ 2 test was used to compare the probability of second-line DST results by availability of pnca results. A negative binomial regression model was used to assess turnaround time by smear status, provincial group, sex, and age (dichotomized as 10 years and >10 years). Analyses were conducted using Matlab, version R2012B, and SAS, version 9.3. All statistical tests were 2-sided and significance was determined at P =.05. RESULTS Characteristics of the Study Population Among 1332 patients identified with RIF-resistant tuberculosis in the 4 provinces, only 44.7% (596) correctly completed the South African algorithm, resulting in second-line DST results for both fluoroquinolones and injectables. In the Western Cape, we identified 835 patients with RIF-resistant MTB Xpert results (Figure 1A). A second sample was processed in 700 patients (83.8%), but cultures grew for second-line DST in only 642 (76.9%). Results on second-line injectables and fluoroquinolones were available in only 466 (55.8%) due to second-line DST not done (n = 71), contamination (n = 56), loss of viability (n = 45), or incomplete DST (n = 4). We included 497 patients with RIF-resistant MTB/RIF Xpert results from the Eastern Cape, Free State, and Gauteng provinces (Figure 1B). A second sample was processed in 447 (89.9%) of these patients, but second-line DST for injectables and fluoroquinolones was only available in 130 (26.2%) due to contamination (n = 96), failure to grow on culture (n = 88), second-line DST not performed (n = 53), culture not done (n = 23), error (n = 8), nontuberculous mycobacteria (n = 6), or incomplete results (n = 43). Western Cape patients were more likely to have second-line DST results than patients from other provinces (risk ratio [RR] = 2.13; 95% confidence interval [CI], ). A 835 samples with RIF-resistant MTB on Xpert, November 2011 June 2013 Retrospec ve collec on 700 pa ents had 2nd sample received 642 cultures grew for second-line DST 466 (55 8%) had second-line DST results 135 pa ents no addi onal tes ng found 58 no MTB culture grew (includes ini ally contaminated samples) 71 second-line DST not done 56 addi onal samples contaminated 45 lost viability 4 incomplete second-line DST results Of the 596 patients diagnosed with RIF-resistant tuberculosis on Xpert with available second-line DST results, mean age was 34.1 years (standard deviation, 11.5), 245 (41.1%) were female, and 332 (58.3%) were smear positive (Table 1). Of the 585 (98.2%) patients with LPA or targeted sequencing results, 468 (80.0%) had INH-resistance mutations and 574 (98.6%) had RIF-resistance mutations detected. Patients from the Western Cape were younger (P =.0001), due to inclusion of children in that province, and had more INH resistance detected (P =.003) than patients from the other provinces (Table 1). Using a generalized additive model, we found in the Western Cape the odds of getting a second-line test decreased by 3% every 30 days (P =.03) and was 0.56 (95% CI, ). The probability of a second-line test in the other provinces did not significantly change over time (P =.70) and was estimated to be 0.26 (95% CI, ). In Western Cape, children (aged 10 years; n = 32) were 51% less likely to have second-line DST than patients aged >10 years (RR = 0.49; 95% CI, ). Patients with smear-negative disease were less likely to have second-line DST results than patients with smear-positive disease (RR = 0.72; 95% CI, ). Gender was not associated with likelihood of having a second-line DST result (RR = 1.1; 95% CI, ) and did not significantly change after controlling for province. Second-Line DST Turnaround Time and Results Median turnaround time for second-line DST was 53 days (range, 8 259). Turnaround time was 1.22 (95% CI, ) times longer for patients with smear-negative compared to smear-positive disease and 1.09 times longer for females compared to males (95% CI, ). Time to results was 1.15 (95% CI, ) times longer for patients from Western Cape Province than for those from other provinces. B 497 pa ents with RIF-resistant MTB on Xpert, November 2012 December 2013 Prospec ve collec on 447 pa ents had 2nd sample received 295 cultures grew for second-line DST 130 (26 2%) had second-line DST results 50 pa ents no addi onal tes ng found 88 culture nega ve 27 culture contaminated 23 culture not done 6 NTM 8 error 53 second-line DST not done 69 addi onal samples contaminated 43 incomplete second-line DST Figure 1. Flow diagrams of patients from Western Cape Province (A) and Eastern Cape, Free State, and Gauteng provinces (B) included in the study. Abbreviations: DST, drug-susceptibility test; MTB, Mycobacterium tuberculosis; NTM, nontuberculous mycobacteria; RIF, rifampin CID 2017:64 (1 June) Jacobson et al

4 Table 1. n = 596 Characteristics of Patients Diagnosed With Rifampin-Resistant Tuberculosis on Xpert Who Had Second-Line Drug-Susceptibility Test Results: Characteristic All Provinces (N = 596) Western Cape (N = 466) Other Provinces (N = 130) P Value Mean age, y (standard deviation) 34.1 (11.5) 33.1 (11.5) 37.5 (11.0).0001 Women (%) 245 (41.1) 183 (39.3) 62 (47.7).08 Smear positive (%) 332 (58.3) 253 (57.1) 79 (62.2).30 Rifampin resistant on LPA (%) 574 (98.6) 458 (98.7) 116 (98.3).67 Isoniazid resistant on LPA (%) 468 (80.0%) 383 (82.5%) 85 (70.3%).003 Data are proportion of patients (%) unless otherwise indicated. Smear positive includes scanty results. Missing: smear positive, 26; rifampin resistant on LPA, 14; isoniazid resistant on LPA, 11. Continuous variables were compared between groups using the Student t test for normal distribution. Categorical variables were compared with χ 2 test if cell values were 5, otherwise Fisher exact test was used. Abbreviation: LPA, MTBDRplus line probe assay. Among the 549 patients with DST results for AMI/KAN, OFX, and ETH, 326 (59.4%) had resistance to 1 or more second-line drugs, including 71 (12.9%) with preextensively drug-resistant (pre-xdr) tuberculosis and 38 (6.9%) with XDR tuberculosis (Table 2). These proportions were similar for the Western Cape and other provinces, except that patients from the Western Cape were more likely to have any additional resistance (RR = 1.92; 95% CI, ) due to more ETH resistance (RR = 4.06; 95% CI, ; Table 2). Among these 549 patients, 252 (45.9%) had pnca results, of which 137 (54.4%) were PZA resistant based on any mutation in the pnca gene and promoter region. Characteristics of patients with and without pnca data were similar except that Western Cape patients without pnca results were more likely to have ETH resistance (RR = 1.41; 95% CI, ) than Western Cape patients with pnca results (Table 2). In the 252 patients with DST data on AMI/KAN, OFX, ETH, and PZA, the standard MDR tuberculosis regimen (KAN, MOXI, ETH, PZA, and terizidone) would only provide 5 effective drugs to 31.4% and 4 effective drugs to 28.6% of patients (if terizidone is assumed effective in all; Table 3). The percentage of patients with likely effective drugs if started on the standard regimen is depicted in Figure 2. Following the South African DOH guidelines, an ineffective regimen would be initiated empirically in 40.1% of these patients, with 7.9% receiving only 1, 8.3% receiving 2, and 23.8% receiving 3 effective drugs. The standard MDR tuberculosis regimen was more likely to contain only 3 effective drugs (RR = 2.96; 95% CI, ) or to be a suboptimal baseline regimen (RR = 1.84; 95% CI, ) for patients from Western Cape than patients from the other provinces (Table 3). Table 2. Second-Line Drug-Resistance Patterns in All Patients With Second-Line Drug-Susceptibility Test Results (Defined as Having Results for Fluoroquinolone, Kanamycin or Amikacin, and Ethionamide Susceptibility) and the Subset of Patients With pnca Results (Representing PZA Susceptibility) Drug Resistance Western Cape Province (n = 466) Other Provinces (n = 130) Combined Provinces (n = 596) Patients With Second- Line Results (%), n = 451 Subset of Patients With pnca Results (%), n = 173 Patients With Second- Line Results (%), n = 98 Subset of Patients With pnca Results (%), n = 79 Patients With Second- Line Results (%), n = 549 Subset of Patients With pnca Results (%), n = 252 KAN a only 3 (0.7) 2 (1.2) 3 (3.1) 3 (3.8) 6 (1.1) 5 (2.0) ETH only b 206 (45.7) c 63 (36.4) c 11 (11.2) 8 (10.1) 217 (39.5) 71 (28.2) OFX d only 13 (2.9) 7 (4.1) 5 (5.1) 5 (6.3) 18 (3.3) 12 (4.8) KAN + ETH 25 (5.5) 8 (4.6) 8 (8.2) 6 (7.6) 33 (6.0) 14 (5.6) KAN + OFX d 4 (0.9) 2 (1.2) 1 (1.0) 1 (1.3) 5 (0.9) 3 (1.2) ETH + OFX d 14 (3.1) 8 (4.6) (2.6) 8 (3.2) KAN + ETH + OFX d 28 (6.2) 15 (8.7) 5 (5.1) 5 (6.3) 33 (6.0) 20 (7.9) Any additional 293 (65.0) 105 (60.7) 33 (33.7) 28 (35.4) 326 (59.4) 133 (52.8) resistance e Pre-XDR 55 (12.2) 25 (14.5) 16 (16.3) 14 (17.7) 71 (12.9) 39 (15.5) XDR 32 (7.1) 17 (9.8) 6 (6.1) 6 (7.6) 38 (6.9) 23 (9.1) pnca results represent pyrazinamide susceptibility. Categorical variables were compared with χ2 test if cell values were 5, otherwise Fisher exact test was used. Abbreviations: ETH, ethionamide; FQ, fluoroquinolone; KAN, kanamycin; OFX, ofloxacin; XDR, extensively-drug resistant. a KAN resistance tested by amikacin susceptibility. b ETH only resistance differed between patients from Western Cape and those from other provinces (P <.0001). c Among Western Cape patients, ETH-only resistance differed between patients with and without pnca results (P =.0018). d Fluoroquinolone resistance tested by OFX susceptibility. e More patients from Western Cape had any additional resistance than patients from the other provinces (P <.0001). Second Line Susceptibility RIF-Resistant Tuberculosis CID 2017:64 (1 June) 1505

5 Table 3. Number of Likely Effective Drugs in Patient s Likely Standardized 5-Drug Baseline Multidrug-Resistant Tuberculosis Regimen No. Effective Drugs All Provinces n = 252 (%) Western Cape n = 173 (%) Other Provinces n = 79 (%) P Value 5 79 (31.4) 48 (27.7) 31 (39.2) (28.6) 44 (25.4) 28 (35.5) (23.8) 52 (30.1) 8 (10.1) (8.3) 14 (8.1) 7 (8.9) (7.9) 15 (8.7) 5 (6.3).52 Suboptimal ( 3) 101 (40.1) 81 (46.8) 20 (25.3).0012 Categorical variables were compared with χ 2 test if cell values were 5, otherwise Fisher s exact test was used. DISCUSSION Timely first- and second-line DST results are critical for ensuring effective MDR tuberculosis regimens. The successful South African Xpert implementation revealed both operational and technical issues in the phenotypic DSTs used for the second-line tuberculosis drugs. Operational challenges include inconsistent receipt and testing of the second sample. Technical challenges include high rates of sample contamination, loss of viability, and inherent delays. In this South African study, less than half (44.7%) of RIF-resistant tuberculosis patients had second-line DST results for both fluoroquinolones and second-line injectables, and of those with results, only 3 of 5 (59.9%) would have received effective empiric treatment (defined as containing 4 or 5 drugs to which the MTB bacillus is sensitive) when initiating the standard South African MDR-tuberculosis treatment regimen (KAN+MOXI+ETH+PZA+terizidone). Furthermore, a median turnaround time for injectable and fluoroquinolone DST results of 53 days following RIF-resistance diagnosis on Xpert placed many patients at risk of exposure to ineffective therapy for weeks before susceptibility patterns could inform treatment adjustments [17 19]. The severity of the problem differed by region. Compared to patients in the other provinces, Western Cape patients were twice as likely to have second-line DST results (44.7% vs 26.2%) but almost half as likely to initiate an effective MDR-tuberculosis treatment regimen (46.8% vs 25.3%). We found a slight decreasing trend in likelihood of having a second-line DST in the Western Cape, reflecting the challenge of maintaining this algorithm. The higher prevalence of available DST results in the Western Cape likely reflects the unique provincial guidelines that recommend collection of 2 baseline samples. The lower effectiveness of the standard MDR-tuberculosis regimen in the Western Cape was due to the higher prevalence of ETH resistance. The finding that many patients, especially those with smear-negative disease and children, failed to have second-line DST results reflects the technical challenge of getting good-quality specimens for culture-based tests [20, 21]. Our data confirm prior studies that showed that operational challenges in getting a second sample contribute to difficulty in constructing an effective treatment regimen [22] and support guidelines that recommend the collection of 2 samples at baseline in all people with presumptive tuberculosis so that the first specimen can be dedicated to Xpert testing and a second to LPA and/or phenotypic DST [23]. This recommendation of 2 baseline specimens should be extended to all countries where the tuberculosis diagnostic algorithm mirrors that in South Africa. Few studies have reported second-line DST results to the same level of granularity or explored the time element that we present. Our findings are in line with those published by the Figure 2. Percentage of patients with number of likely effective drugs in standardized 5-drug multidrug-resistant tuberculosis regimen: Western Cape Province (N = 173) and Eastern Cape, Free State, Gauteng provinces (N = 79) CID 2017:64 (1 June) Jacobson et al

6 Preserving Effective TB Treatment Study (PETTS), conducted in 9 countries between 2005 and The PETTS group found that 43.7% of MDR-tuberculosis patients had resistance to at least 1 second-line drug, ranging from 33.3% in Thailand to 62% in Latvia [24]. In South Africa, 44.7% of MDR-tuberculosis patients had resistance to at least 1 second-line drug, but this cohort excluded patients previously treated for MDR tuberculosis and did not perform PZA DST; half of MDR tuberculosis specimens are resistant to PZA globally, so their South Africa resistance estimate is likely a lower boundary [25]. The NHLS recently released the South African Tuberculosis Drug Resistance Survey, [26]. They report frequent second-line drug resistance among MDR cases across South Africa, including 44.7% (95% CI, 25.9% 63.6%) ETH resistance, 59.1% (95% CI, 49% 69.1%) PZA resistance, and 13% (95% CI, 5% 21%) for both second-line injectables and OFX, which is similar to what we report in our cohort. Additionally, they report 4.9% (95% CI, 1% 8.8%) XDR tuberculosis, which is similar to our finding of 6.9% XDR tuberculosis. The WHO recently endorsed the use of new drugs (bedaquiline and delamanid) and a novel, shortened 7-drug MDR tuberculosis regimen [27]. A caveat of this recommendation is that patients with known or highly suspected baseline resistance to any of the 7 drugs (except INH), including KAN, MOXI, ETH, and PZA, should not be started on the shortened regimen. Nearly half of our cohort would have resistance to at least 1 drug. The high frequency of resistance raises the question of this shorter regimen s effectiveness in South African patients. Ideally, standardized regimens should be built on local epidemiologic evidence regarding baseline resistance patterns in communities and contain new drugs with less risk of baseline resistance. Currently, information on ETH resistance can be obtained by inha promoter mutation presence on the MTBDRplus LPA assay [11]. Resistance to fluoroquinolones and second-line injectables can also be obtained more rapidly using the MTBDRsl LPA (Hain Lifescience GmbH, Nehren, Germany), which was recently endorsed by the WHO [28]. Changes in the South African algorithm to implement the MTBDRsl would overcome many of the technical challenges our study revealed, including shortening turnaround time and reducing reliance on culture viability. As new drugs such as bedaquiline, delamanid, and linezolid become available [29], careful monitoring for resistance will also need to occur. While our study has many strengths, including the relatively large sample size and inclusion of a representative sample of 4 South African provinces, we do note limitations. First, the proportion of patients with second-line DST results may have been underestimated in the Western Cape as linkage of first and second specimens was done through a matching algorithm. Second, classification of ETH resistance based on inha promoter mutation information may underestimate the prevalence of ETH resistance, as other molecular mechanisms are known to cause ETH resistance [11]. Third, generalizability may be limited beyond South Africa, especially to regions with less mature MDR-tuberculosis epidemics and those that do not use rapid diagnostics to screen for RIF resistance. Fourth, dictated by data availability, the study periods between the provinces were nearly, but not perfectly, overlapping. However, no major policy changes other than Xpert implementation were made during these years. Last, because we did not have access to treatment registries, we use the phrase potentially started ineffective treatment ; all patients who did start treatment would receive the stated empiric regimen. The successful South African implementation of Xpert to screen all tuberculosis suspects means that more MDR-tuberculosis patients are identified and started on therapy earlier [30]. Although a major achievement, the combination of high baseline resistance rates to drugs included in the South African standardized MDR-tuberculosis regimen, the failure to generate second-line DST information in more than half of all RIF-resistant tuberculosis patients, and the long turnaround time of second-line DST could fuel amplification of resistance and treatment failure and worsen the resistance epidemic [19 21]. The possible negative consequences of screening for resistance to a single drug (RIF) and use of a standard MDR-tuberculosis regimen highlight the need for implementation of available rapid diagnostics for resistance to a broader panel of drugs, as well as the need for individualized MDR-tuberculosis treatment based on comprehensive knowledge of drug-resistance patterns. Notes Author contributions. K. J. and R. M. conceived this study and developed the analysis plan. M. K., E. R., L. W., and O. S. did the data cleaning, final analyses, and prepared all tables and figures. L. W. provided biostatistical methods support. K. J., A. V., and R. M. interpreted the findings. M. B., E. S., T. D., J. S., L. S., W. S., and A. V. provided data and reviewed results. K. J. prepared the first draft. All authors provided critical feedback and edits to finalize the manuscript. Acknowledgments. We thank the participants of this study and the National Health Laboratory Service for providing the laboratory data used in our analysis. Financial support. This work was supported by the National Institutes of Health, Fogarty International Center (K01 TW to K. J.), Burroughs Wellcome Fund/American Society of Tropical Medicine and Hygiene Postdoctoral Fellowship in Infectious Diseases (to K. J.), and the National Institute of Allergy and Infectious Diseases (R01 AI to A. V.]. Potential conflicts of interest. All authors: No potential 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. Tuberculosis Diagnostics: Xpert MTB/RIF Test Available at: eng.pdf. Accessed on 14 July Van Rie A, Page-Shipp L, Hanrahan CF, et al. Point-of-care Xpert MTB/RIF for smear-negative tuberculosis suspects at a primary care clinic in South Africa. Int J Tuberc Lung Dis 2013; 17(3): Lawn SD, Brooks SV, Kranzer K, et al. Screening for HIV-associated tuberculosis and rifampicin resistance before antiretroviral therapy using the Xpert MTB/RIF assay: a prospective study. PLoS Med. 2011; 8:e Second Line Susceptibility RIF-Resistant Tuberculosis CID 2017:64 (1 June) 1507

7 4. Cohen GM, Drain PK, Noubary F, Cloete C, Bassett IV. Diagnostic delays and clinical decision making with centralized Xpert MTB/RIF testing in Durban, South Africa. J Acquir Immune Defic Syndr 2014; 67:e Ahuja SD, Ashkin D, Avendano M, et al. Multidrug resistant pulmonary tuberculosis treatment regimens and patient outcomes: an individual patient data meta-analysis of 9,153 patients. PLoS Med 2012; 9:e Franke MF, Appleton SC, Mitnick CD, et al. Aggressive regimens for multidrug-resistant tuberculosis reduce recurrence. Clin Infect Dis 2013; 56(6): World Health Organization. Global Tuberculosis Report Geneva: WHO; Department of Health Republic of South Africa. National Tuberculosis Management Guidelines Available at: TB%20management%20guidelines% pdf. Accessed 23 March Boehme CC, Nicol MP, Nabeta P, et al. Feasibility, diagnostic accuracy, and effectiveness of decentralised use of the Xpert MTB/RIF test for diagnosis of tuberculosis and multidrug resistance: a multicentre implementation study. Lancet 2011; 377: Barnard M, Albert H, Coetzee G, O Brien R, Bosman ME. Rapid molecular screening for multidrug-resistant tuberculosis in a high-volume public health laboratory in South Africa. Am J Respir Crit Care Med 2008; 177(7): Vadwai V, Ajbani K, Jose M, et al. Can inha mutation predict ethionamide resistance? Int J Tuberc Lung Dis 2013; 17: World Health Organization. Guidelines for surveillance of drug resistance in tuberculosis, 5th edition Available at: eam/10665/174897/1/ _eng.pdf. Accessed 30 July Jugheli L, Bzekalava N, de Rijk P, Fissette K, Portaels F, Rigouts L. High level of cross-resistance between kanamycin, amikacin, and capreomycin among Mycobacterium tuberculosis isolates from Georgia and a close relation with mutations in the rrs gene. Antimicrob Agents Chemother 2009; 53: Farhat MR, Jacobson KR, Franke MF, et al. Gyrase mutations are associated with variable levels of fluoroquinolone resistance in Mycobacterium tuberculosis. J Clin Microbiol 2016; 54(3): Streicher EM, Maharaj K, York T, et al. Rapid sequencing of the Mycobacterium tuberculosis pnca gene for detection of pyrazinamide susceptibility. J Clin Microbiol 2014; 52(11): Pfyffer GE, Bonato DA, Ebrahimzadeh A, et al. Multicenter laboratory validation of susceptibility testing of Mycobacterium tuberculosis against classical second-line and newer antimicrobial drugs by using the radiometric BACTEC 460 technique and the proportion method with solid media. J Clin Microbiol 1999; 37(10): Chihota VN, Müller B, Mlambo CK, et al. Population structure of multi- and extensively drug-resistant Mycobacterium tuberculosis strains in South Africa. J Clin Microbiol 2012; 50: Klopper M, Warren RM, Hayes C, et al. Emergence and spread of extensively and totally drug-resistant tuberculosis, South Africa. Emerg Infect Dis 2013; 19(3): Streicher EM, Müller B, Chihota V, et al. Emergence and treatment of multidrug resistant (MDR) and extensively drug-resistant (XDR) tuberculosis in South Africa. Infect Genet Evol 2012; 12: Dodd PJ, Gardiner E, Coghlan R, Seddon JA. Burden of childhood tuberculosis in 22 high-burden countries: a mathematical modelling study. Lancet Glob Heal 2014; 2:e Zar HJ, Workman L, Isaacs W, Dheda K, Zemanay W, Nicol MP. Rapid diagnosis of pulmonary tuberculosis in African children in a primary care setting by use of Xpert MTB/RIF on respiratory specimens: a prospective study. Lancet Glob Heal 2013; 1:e Dlamini-Mvelase NR, Werner L, Phili R, Cele LP, Mlisana KP. Effects of introducing Xpert MTB/RIF test on multi-drug resistant tuberculosis diagnosis in KwaZulu-Natal South Africa. BMC Infect Dis 2014; 14: Barnard M, Warren R, Gey Van Pittius N, et al. Genotype MTBDRsl line probe assay shortens time to diagnosis of extensively drug-resistant tuberculosis in a high-throughput diagnostic laboratory. Am J Respir Crit Care Med 2012; 186(12): Dalton T, Cegielski P, Akksilp S, et al; Global PETTS Investigators. Prevalence of and risk factors for resistance to second-line drugs in people with multidrug-resistant tuberculosis in eight countries: a prospective cohort study. Lancet 2012; 380(9851): Whitfield MG, Soeters HM, Warren RM, et al. A global perspective on pyrazinamide resistance: systematic review and meta-analysis. PLoS One 2015; 10:e National Institute for Communicable Diseases Division of the National Health Laboratory Service. South African Tuberculosis Drug Resistance Survey Available at: DR%20Survey%20Report.pdf. Accessed 23 March World Health Organization. WHO treatment guidelines for drug-resistant tuberculosis 2016 update. Geneva: WHO; World Health Organization. The use of molecular line probe assays for the detection of resistance to second-line anti-tuberculosis drugs: Policy guidance Available at: Accessed 21 July Gualano G, Capone S, Matteelli A, Palmieri F. New antituberculosis drugs: from clinical trial to programmatic use. Infect Dis Rep 2016; 8: Cox HS, Daniels JF, Muller O, et al. Impact of decentralized care and the Xpert MTB/RIF test on rifampicin-resistant tuberculosis treatment initiation in Khayelitsha, South Africa. Open Forum Infect Dis 2015; 2:ofv CID 2017:64 (1 June) Jacobson et al