Extended treatment duration and recurrence in successfully treated drug-susceptible pulmonary tuberculosis with cavity on chest radiograph Authors: Chang-Seok Yoon, Tae-Ok Kim, Hong-Joon Shin, Ju Sang Kim, Hyung Woo Kim, Eung Gu Lee, Sung Soo Jung, Jee Youn Oh, Jin Woo Kim, Heayon Lee, Seunghoon Kim, Sun-Hyung Kim, Yeonhee Park, Jiwon Lyu, Sun Jung Kwon, Yun-Jeong Jeong, Do Jin Kim, Hyeon-Kyoung Koo, Ganghee Chae, Sun Young Kyung, Jinsoo Min, Yong-Soo Kwon, Jung Yeon Heo, Jung Yeon Heo, Jung Yeon Heo, Jung Yeon Heo Published: 2026 ABSTRACT BackgroundCavity on chest radiograph is recognized as an independent risk factor for recurrence in pulmonary tuberculosis (TB), but evidence supporting extended treatment in these patients remains limited, particularly in real-world data from Korea.MethodsWe performed a retrospective analysis using data from a multicenter prospective cohort of patients with pulmonary tuberculosis enrolled at 18 Korean institutions, comparing recurrence rates between standard treatment (≤200 days) and extended treatment (>200 days) among drug-susceptible pulmonary TB patients with cavity on chest radiograph.ResultsAmong 159 patients, 58 (36.5%) received standard treatment and 101 (63.5%) received extended treatment. Baseline sputum smear positivity was higher in the extended group (29.3% vs. 56.4%, p = 0.002), but 2-month culture positivity (6.3% vs. 4.4%, p > 0.999) and time to culture conversion (8.00 ± 3.91 vs. 7.45 ± 3.73 weeks, p = 0.454) were similar. Recurrence during post-treatment follow-up occurred in 4 patients (2.5%) and did not differ between groups (3.4% vs. 2.0%, p = 0.966). Treatment interruption due to adverse drug reactions was more frequent with extended treatment (1.7% vs. 7.9%, p = 0.204), while overall adverse event rates were comparable. In a fixed-time analysis using Firth’s penalized logistic regression, extended treatment showed a numerically lower risk of recurrence, although statistical significance was not reached (adjusted odds ratio [aOR], 0.07; 95% confidence interval [CI], 0.01–1.06; p = 0.056). Among covariates, low body mass index was associated with recurrence, but this estimate was very imprecise and should be interpreted with caution.ConclusionCavity on chest radiograph alone may not be sufficient to justify routine extension of treatment beyond 200 days. Because only four recurrence events were observed, the adjusted association favoring extended treatment was imprecise and should be interpreted as exploratory rather than definitive. Larger prospective studies that systematically capture microbiologic response and detailed radiologic disease burden are needed to identify which patients with cavitary disease may benefit from individualized treatment extension. Introduction Pulmonary tuberculosis (TB) is an infectious disease caused by Mycobacterium tuberculosis and transmitted via inhalation of airborne droplet nuclei generated by patients with infectious pulmonary TB. It is a preventable and treatable disease, with treatment success rates approach 95% in drug-susceptible (DS) TB when managed with the standard 6-month regimen [1,2]. However, advanced TB can lead to significant lung destruction and the development of cavitary lesions. The prevalence of cavitary TB has been reported to range from 29% to 87% [3]. Cavitary lesions in the lungs are associated with severe disease, including higher bacterial loads in sputum, and have been linked to an increased risk of post-treatment recurrence [4–10]. Recurrence is clinically important because it is associated with morbidity and mortality and may contribute to the development of drug-resistant TB [1]. To reduce recurrence risk in higher-risk phenotypes, the 2016 guidelines from the American Thoracic Society (ATS), Centers for Disease Control and Prevention (CDC), and Infectious Diseases Society of America (IDSA) recommend extending the continuation phase by 3 months, based largely on historical trial evidence and expert opinion, for patients with cavity on chest radiograph, particularly when cultures remain positive at 2 months [8]. However, contemporary evidence supporting routine treatment extension in cavitary TB remains limited. Treatment extension increases cumulative drug exposure and may increase the risk of adverse drug reactions, treatment interruptions, and additional costs. Drug resistance in M. tuberculosis arises through selection of spontaneous mutations under drug pressure, and inadequate or interrupted therapy can facilitate acquired resistance [11]. Therefore, optimizing treatment duration requires balancing potential benefits against the burden of longer therapy. We aimed to evaluate whether extending treatment beyond 200 days is associated with a reduced risk of bacteriologically confirmed recurrence during the post-treatment follow-up period among patients with DS pulmonary TB who had cavity on baseline chest radiograph. Methods Population A prospective cohort of 1,204 pulmonary TB patients who were recruited from 18 centers in South Korea between July 2019 and June 2023 was analyzed. Details of the cohort protocol have been described previously [12]. The data for this study were accessed on 30 June 2024. To evaluate the efficacy and safety of extended treatment in patients at high risk of recurrence, those with cavity on chest radiograph were included. Patients with any evidence or history of drug-resistant TB were excluded, including multidrug-resistant or rifampin-resistant TB, isoniazid mono-resistant TB, and other forms of mono-resistant TB. Accordingly, the analytic population was restricted to patients with pan-susceptible TB, defined as susceptibility to all standard HRZE components. Among these patients, only those who received the standard isoniazid, rifampin, pyrazinamide, and ethambutol (HRZE) regimen and achieved treatment success were included. Therefore, patients who did not receive the HRZE regimen, experienced treatment failure, died, transferred to other clinics, were lost to follow-up, or discontinued treatment for reasons such as consent withdrawal or changes in diagnosis were excluded. The final analytic cohort consisted of patients with cavity on chest radiograph who completed post-treatment follow-up (Fig 1). Flow chart of participant enrollment. Abbreviations: TB: tuberculosis; MDR/RR: multi-drug resistance/rifampin-resistance; HRZE: H: isoniazid; R: rifampin; Z: pyrazinamide; E: ethambutol. Data collection For the selected patients, baseline demographic data such as age, sex, and comorbidities were recorded at the time of diagnosis. Comorbidities were defined based on documented diagnoses, relevant medication use, or available clinical information in the medical records. Diabetes mellitus and hypertension were defined by documented diagnoses or use of glucose-lowering or antihypertensive medications, respectively. Chronic pulmonary disease included asthma, chronic bronchitis, emphysema, or other chronic pulmonary diseases with ongoing respiratory symptoms. Chronic kidney disease and chronic liver disease were defined as documented chronic renal or hepatic disease. Cardiovascular disease included chronic cardiac conditions such as ischemic heart disease, heart failure, or arrhythmia; neurologic disease included chronic neurologic disorders such as stroke, dementia, or Parkinson’s disease; and malignancy included active or previous cancer. Additionally, initial clinical data, including vital signs, biochemical tests, and microbiological results, were documented. Biochemical tests, including complete blood count, liver function tests, renal function tests, and inflammation markers, were conducted through blood sampling. To evaluate cavity or multilobe infiltration, chest radiographs were assessed, and for microbiological evaluation, acid-fast bacilli (AFB) smear and culture were evaluated through sputum tests. Throughout the follow-up period, participants received regular biochemical evaluations and were closely monitored for adverse drug reactions at predefined intervals (14 days, 28 days, 2 months, and monthly thereafter). Adverse drug reactions were identified from the cohort medical records and categorized using operational definitions adapted from CTCAE version 5.0 where applicable. Although information on clinical severity was available in some medical records, formal CTCAE severity grades were not uniformly recorded in the cohort database. Therefore, adverse drug reactions were analyzed as binary outcomes based on their presence or absence rather than by severity grade. All patients were followed for a minimum of one year after treatment completion through regular outpatient visits, linkage with the national tuberculosis surveillance system, and telephone follow-up when necessary to ensure complete ascertainment of recurrence [12–14]. Group and study outcome The enrolled patients were categorized according to total treatment duration. Because the standard 6-month regimen corresponds to approximately 180 days, extended treatment was defined as treatment lasting >200 days, allowing an additional 2 weeks beyond the expected standard duration. This 2-week allowance was based on a previous study of prolonged TB treatment [15], which considered that treatment interruptions of ≥2 weeks during the intensive phase may require reintroduction of intensive-phase treatment and thereby prolong total treatment duration. Thus, the 200-day cutoff was selected to distinguish meaningful treatment extension from minor variations around the standard 6-month course. The primary outcome was recurrence, defined as bacteriologically confirmed tuberculosis occurring within the follow-up period after treatment completion in patients previously classified as cured or having successfully completed treatment. Statistical analysis Continuous variables are presented as means ± standard deviations, and categorical variables are presented as counts (%). When comparing the two groups, t-tests were used to compare the means of continuous variables, while Pearson’s chi-square test or Fisher’s exact test were used to analyze the associations between categorical variables. To identify factors associated with recurrence, logistic regression analyses were conducted. Because extended treatment was defined as a total treatment duration of >200 days, a fixed-time analysis at 200 days was used to ensure consistent exposure classification. Cox proportional hazards modeling was not used as the main analytic approach because total treatment duration was determined during treatment and was fully known only after treatment completion; therefore, treating it as a baseline exposure could introduce immortal-time bias. Time-dependent Cox modeling or modeling treatment duration as a continuous variable was also not applied as the main analytic approach because only four recurrence events were observed, making more complex models highly unstable. Given the small number of recurrence events and the presence of separation, both univariable and multivariable analyses were conducted using Firth’s penalized logistic regression. Univariable analyses were first performed using demographic characteristics, comorbidities, radiologic findings, and microbiologic results. Given the very small number of recurrence events, we sought to minimize overfitting by restricting the number of variables included in the multivariable model. Extended treatment was included as the main exposure of interest. Age and sex were included a priori as clinically important covariates, and variables with p < 0.05 in the univariable analysis were additionally considered for inclusion. All statistical analyses were performed using R software (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria). Ethics statement The study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Boards (IRBs) and Ethics Committees of all participating institutions. The study was coordinated and overseen by the representative institution, the Catholic University of Korea (IRB No. KC19ONDI0458), which represented and managed the study across the 17 other participating medical institutions. All participating institutions obtained IRB approval prior to study initiation, with the following approval numbers: SCHBC202012023001 (Soonchunhyang University Bucheon Hospital), UUH202103016 (Ulsan University Hospital), GCIRB2021−016 (Gachon University Gil Hospital), 2020GR0402 (Korea University Guro Hospital), VC19ONDI0168 (St. Vincent's Hospital), KANGDONG2020-06-012-001 (Kangdong Sacred Heart Hospital), KYUH2020-12-013-002 (Konyang University Hospital), SCHCA202012076 (Soonchunhyang University Cheonan Hospital), DUIH202101023001 (Dongguk University Ilsan Hospital), 2020-07-007 (Chungnam National University Hospital), 2020-08-012-003 (Chungbuk National University Hospital), DC19ONDI0052 (Daejeon St. Mary's Hospital), HC19ONDI0068 (Bucheon St. Mary's Hospital), UC19ONDI0106 (Uijeongbu St. Mary's Hospital), PC19ONDI0149 (Eunpyeong St. Mary's Hospital), OC19ONDI0073 (Incheon St. Mary's Hospital), and ISPAIK2020-11-003-001 (Ilsan Paik Hospital). Written informed consent was obtained from all adult participants before their enrollment in the study. Results Of the 1,204 patients in the cohort, 135 with multidrug-resistant or rifampin-resistant TB, 49 with isoniazid mono-resistant TB, and 13 with other mono-resistant TB were excluded. Additional exclusions included patients who did not receive the HRZE regimen (n = 36), experienced treatment failure (n = 4), died (n = 53), transferred to other clinics (n = 41), were lost to follow-up (n = 8), or discontinued treatment because of consent withdrawal or changes in diagnosis (n = 11). Consequently, 854 pan-susceptible DS TB patients who achieved treatment success with the standard HRZE regimen were identified; 164 had cavity on chest radiograph, and 159 completed follow-up after treatment completion and were included in the final analytic cohort (Fig 1). A total of 159 patients with cavity on chest radiograph were enrolled. Among them, 58 (36.5%) received standard treatment and 101 (63.5%) received extended treatment. The extended treatment group had higher AFB smear positivity than the standard group (56.4% vs. 29.3%, respectively, p = 0.002). However, apart from a lower prevalence of chronic kidney disease, there were no significant differences in demographic characteristics or other comorbidities between the two groups (Table 1). 10.1371/journal.pone.0357772.t001Table 1Baseline characteristics of pulmonary tuberculosis patients with cavity on chest radiograph.CharacteristicsTotal, n = 159Standard, n = 58Extended, n = 101P valueAge (years)54.75 ± 14.2956.40 ± 14.0753.80 ± 14.400.269Males, n (%)127 (79.9%)47 (81.0%)80 (79.2%)0.943BMI < 18.5 kg/m2, n/N (%)26/136 (19.1%)5/44 (11.4%)21/92 (22.8%)0.174Ever smoker, n (%)113(71.1%)42 (72.4%)71 (70.3%)0.919Comorbid conditionsHypertension, n (%)33 (20.8%)14 (24.1%)19 (18.8%)0.552Diabetes mellitus, n (%)49 (30.8%)16 (27.6%)33 (32.7%)0.624Cardiovascular disease, n (%)2 (1.3%)1 (1.7%)1 (1.0%)>0.999Chronic pulmonary disease, n (%)7 (4.4%)4 (6.9%)3 (3.0%)0.447Chronic liver disease, n (%)5 (3.1%)1 (1.7%)4 (4.0%)0.760Chronic kidney disease, n (%)3 (1.9%)3 (5.2%)0 (0.0%)0.089Cerebrovascular disease, n (%)2 (1.3%)1 (1.7%)1 (1.0%)>0.999Malignancy, n (%)3 (1.9%)1 (1.7%)2 (2.0%)>0.999Previous history of TB treatment, n (%)28 (17.6%)8 (13.8%)20 (19.8%)0.459Extrapulmonary TB, n (%)11 (6.9%)4 (6.9%)7 (6.9%)>0.999Multi-lobe infiltration, n (%)71 (44.7%)25 (43.1%)46 (45.5%)0.894AFB smear positivity, n (%)74 (46.5%)17 (29.3%)57 (56.4%)0.002Abbreviations: AFB: acid-fast bacilli; BMI: body mass index; TB: tuberculosis. Baseline characteristics of pulmonary tuberculosis patients with cavity on chest radiograph. Abbreviations: AFB: acid-fast bacilli; BMI: body mass index; TB: tuberculosis. Baseline laboratory abnormalities were additionally compared between the two groups and are presented in S1 Table. Baseline leukopenia was more frequent in the standard-treatment group than in the extended-treatment group, whereas the frequencies of anemia, thrombocytopenia, abnormal liver function, and abnormal renal function were not significantly different between groups. Recurrence did not differ between the two groups (3.4% in the standard group vs. 2.0% in the extended group, p = 0.966), nor did the rate of 2-month culture positivity (6.3% vs. 4.4%, respectively, p > 0.999). In addition, no significant difference was observed in the time to culture conversion between the groups (8.00 ± 3.91 weeks vs. 7.45 ± 3.73 weeks, respectively, p = 0.454) (Table 2). 10.1371/journal.pone.0357772.t002Table 2Comparison of treatment outcomes between standard and extended treatment in pulmonary tuberculosis with cavity on chest radiograph.VariablesTotal, N = 159Standard, n = 58Extended, n = 101p valueTreatment duration (days)251.75 ± 68.47181.98 ± 13.31291.81 ± 53.50<0.001Duration of follow up after treatment completion (days)404.10 ± 191.34414.48 ± 223.51398.14 ± 171.090.6312-month culture positivity, n/N (%)4/77 (5.2%)2/32 (6.3%)2/45 (4.4%)>0.999Time to culture conversion (weeks)7.66 ± 3.798.00 ± 3.917.45 ± 3.730.454Treatment interruption of > 7 days for any reason, n (%)13(8.2%)4 (6.9%)9 (8.9%)0.884Recurrence, n (%)4 (2.5%)2 (3.4%)2 (2.0%)0.966 Comparison of treatment outcomes between standard and extended treatment in pulmonary tuberculosis with cavity on chest radiograph. Among the four patients with bacteriologically confirmed recurrence, two were in the standard-treatment group and two were in the extended-treatment group. In the standard-treatment group, recurrence was detected 197 days (6.5 months) and 414 days (13.6 months) after treatment completion. In the extended-treatment group, recurrence was detected 295 days (9.7 months) and 378 days (12.4 months) after treatment completion. Because molecular typing was not available, relapse and reinfection could not be definitively distinguished. Based on timing alone, the earlier recurrences were clinically more compatible with relapse, whereas the later recurrences could represent either late relapse or reinfection (S2 Table). Medication interruption due to adverse drug events tended to be more frequent in the extended group. However, there was no significant difference in the overall incidence of adverse events (Table 3). 10.1371/journal.pone.0357772.t003Table 3Adverse drug reactions according to treatment duration in pulmonary tuberculosis patients with cavity on chest radiograph.VariablesTotal, N = 159Standard, n = 58Extended, n = 101p valueGastrointestinal adverse reactiona, n (%)63 (39.6%)21(36.2%)42 (41.6%)0.618Hepatotoxicityb, n (%)47(29.6%)19(32.8%)28 (27.7%)0.625Cutaneous adverse drug reactionc, n (%)52(32.7%)19(32.8%)33(32.7%)>0.999Leukopeniad, n (%)14 (8.8%)3(5.2%)11(10.9%)0.350Generalized weaknesse, n (%)34(21.4%)15(25.9%)19(18.8%)0.399Ototoxicityf, n (%)11(6.9%)4 (6.9%)7 (6.9%)>0.999Nephrotoxicityg, n (%)5 (3.1%)3 (5.2%)2 (2.0%)0.523Infectionh, n (%)10 (6.3%)4 (6.9%)6 (5.9%)>0.999Treatment interruption due to adverse drug reactions, n (%)9 (5.7%)1 (1.7%)8 (7.9%)0.204a Defined as the occurrence of nausea, vomiting, diarrhea, abdominal discomfort, or anorexiab Defined as elevation of AST, ALT, or total bilirubin above the institutional upper limit of normalc Defined as newly developed rash, pruritus, urticaria, or drug eruptiond White blood cell count < 4,000 cells/µLe Defined as newly documented generalized weakness during anti-tuberculosis treatment that was assessed by the treating physician and recorded in the medical records.f Defined as newly documented hearing impairment, tinnitus, vertigo, or vestibular symptoms considered related to anti-tuberculosis treatmentg Defined as serum creatinine elevation above the institutional upper limit of normal or ≥1.5 times baselineh Defined as a newly diagnosed infection during anti-tuberculosis treatment, supported by clinical, laboratory, radiologic, or microbiologic evidence and/or antimicrobial treatment. Adverse drug reactions according to treatment duration in pulmonary tuberculosis patients with cavity on chest radiograph. a Defined as the occurrence of nausea, vomiting, diarrhea, abdominal discomfort, or anorexia b Defined as elevation of AST, ALT, or total bilirubin above the institutional upper limit of normal c Defined as newly developed rash, pruritus, urticaria, or drug eruption d White blood cell count < 4,000 cells/µL e Defined as newly documented generalized weakness during anti-tuberculosis treatment that was assessed by the treating physician and recorded in the medical records. f Defined as newly documented hearing impairment, tinnitus, vertigo, or vestibular symptoms considered related to anti-tuberculosis treatment g Defined as serum creatinine elevation above the institutional upper limit of normal or ≥1.5 times baseline h Defined as a newly diagnosed infection during anti-tuberculosis treatment, supported by clinical, laboratory, radiologic, or microbiologic evidence and/or antimicrobial treatment. In univariable analyses using logistic regression, no baseline demographic, clinical, radiologic, or microbiologic variables were significantly associated with recurrence. Diabetes mellitus, chronic pulmonary disease, and thrombocytopenia showed nonsignificant trends toward higher recurrence, but confidence intervals were wide due to the small number of events (S3 Table). However, in Firth’s penalized logistic regression analysis for recurrence using a 200-day fixed-time approach, the exploratory adjusted analysis yielded an adjusted odds ratio of 0.07 for extended treatment, but this association did not reach statistical significance (95% CI, 0.01–1.06; p = 0.056). In contrast, low body mass index (BMI) (<18.5 kg/m²) was associated with recurrence in the multivariable Firth's penalized logistic regression model (OR, 81.92; 95% CI, 4.95–21,694.24; p < 0.001), although this estimate was highly imprecise due to sparse data (Table 4). 10.1371/journal.pone.0357772.t004Table 4Firth's penalized logistic regression analysis of factors associated with recurrence in pulmonary tuberculosis patients with cavity on chest radiograph using a 200-day fixed-time approach.VariablesUnivariateMultivariate bOR95% CIp valueOR95% CIp valueAge ≥ 65 years1.260.12-7.960.8160.33<0.01 −5.460.483Males2.360.24-316.870.5225.120.24 −1797.410.346Ever smoker0.960.15-10.110.968BMI < 18.5 kg/m232.913.05-4482.790.00281.924.95 −21694.24<0.001Hypertension0.41<0.01-3.960.502Diabetes mellitus5.490.88-57.830.069Cardiovascular disease6.820.05-102.730.327Chronic pulmonary disease9.860.87-71.760.062Chronic liver disease2.360.02-28.520.528Chronic kidney disease4.840.03-63.020.400Cerebrovascular disease6.820.05-102.730.327Malignancy4.840.03-63.020.399Previous history of TB treatment2.000.19-12.760.507Extrapulmonary TB1.400.01-14.430.833Multi-lobe infiltration1.240.19-8.250.809AFB smear positivity1.150.17-7.630.8762 months positivity a——>0.999Extended treatment0.570.09-3.770.5360.070.01 −1.060.056Abbreviations: OR: odds ratio; CI: confidence interval; BMI: body mass index; TB: tuberculosis; AFB: acid-fast bacillia Estimates were unstable (OR extremely large with infinite/undefined CI) due to sparse data and quasi-complete separation, and thus not reported.b The multivariable model included extended treatment, age, sex, and BMI. Because BMI data were missing in 23 patients, this model was conducted as a complete-case analysis without imputation and included 136 patients with 3 recurrence events. Firth's penalized logistic regression analysis of factors associated with recurrence in pulmonary tuberculosis patients with cavity on chest radiograph using a 200-day fixed-time approach. Abbreviations: OR: odds ratio; CI: confidence interval; BMI: body mass index; TB: tuberculosis; AFB: acid-fast bacilli a Estimates were unstable (OR extremely large with infinite/undefined CI) due to sparse data and quasi-complete separation, and thus not reported. b The multivariable model included extended treatment, age, sex, and BMI. Because BMI data were missing in 23 patients, this model was conducted as a complete-case analysis without imputation and included 136 patients with 3 recurrence events. Sensitivity analyses were performed to assess the robustness of the association between extended treatment and recurrence. When alternative thresholds were applied to define extended treatment duration using a 200-day fixed-time approach (≥180, ≥ 200, and ≥220 days), the results were consistent, with extended treatment showing a numerically lower point estimate for recurrence across all definitions, although none reached statistical significance (S4 Table). In addition, analyses using alternative model specifications with varying degrees of covariate adjustment showed a similar direction of association (S5 Table). Because diabetes mellitus and chronic pulmonary disease showed possible associations with recurrence in the univariable analysis, additional sensitivity analyses were performed by including these variables in the multivariable model. The direction of the association between extended treatment and recurrence remained consistent across these models, although estimates remained imprecise because of the small number of recurrence events (S6 Table). Discussion In this multicenter Korean cohort of DS pulmonary TB patients with cavity on chest radiograph who achieved treatment success with the HRZE regimen, a 200-day fixed-time approach combined with Firth’s penalized logistic regression was applied to address sparse events, and extended treatment beyond 200 days was not statistically significantly associated with a reduced risk of recurrence. Although the point estimate favored extended treatment, the analysis was based on only a small number of recurrence events and should therefore be interpreted cautiously. This finding is important for several reasons. First, to the best of our knowledge, there is a lack of studies that have directly compared the risk of recurrence according to treatment duration among DS pulmonary TB patients with cavity on chest radiograph, including within existing guideline frameworks. Although current guidelines recommend treatment extension in selected high-risk patients, the evidence directly evaluating the incremental benefit of extended treatment duration on recurrence prevention remains limited. In this context, our study provides meaningful real-world evidence by explicitly examining the association between treatment duration and recurrence in a contemporary multicenter cohort. Second, by directly comparing patients who received standard treatment with those who underwent extended treatment, our study provides an opportunity to explore objective evidence regarding the potential benefit of treatment extension. In routine clinical practice, decisions to extend therapy are largely influenced by perceived baseline severity and clinician judgment regarding response [7,8,15,16]. As observed in our study, baseline AFB smear positivity was substantially higher in the extended-treatment group than in the standard-treatment group (56.4% vs. 29.3%; p = 0.002), suggesting that treatment extension was selectively applied based on perceived bacillary burden rather than uniformly determined by guideline-based criteria. Treatment extension based primarily on clinical judgment, in the absence of objective supporting evidence, may inadvertently result in unnecessary extension of treatment, with potential implications for drug-related toxicity, compromised treatment adherence, and increased healthcare resource utilization. Especially, anti-TB therapy is associated with clinically relevant adverse drug reactions, including hepatotoxicity, and longer exposure increases time-at-risk and monitoring burden [17–20]. In our cohort, treatment interruption due to adverse reactions was numerically higher in the extended-treatment group (1.7% vs. 7.9%; p = 0.204). Although imprecise, this pattern is compatible with the expectation that cumulative exposure increases the probability of clinically significant intolerance and associated interruptions [21]. In this context, our findings underscore the need to identify and select high-risk individuals when considering treatment extension. These results support a more selective, evidence-based approach—prioritizing the identification of high-risk patients rather than relying solely on clinician-driven decision-making, thereby potentially avoiding unnecessary extension in lower-risk patients. Lastly, in addition to extended treatment, we observed that low BMI was significantly associated with tuberculosis recurrence. This finding is biologically and clinically plausible, as low BMI reflects undernutrition and impaired host immunity, both of which are well-established risk factors for poor treatment response and disease recurrence in tuberculosis. Previous studies have consistently shown that underweight patients have delayed sputum conversion, reduced bactericidal activity, and higher rates of post-treatment recurrence, likely due to compromised cell-mediated immunity and altered pharmacokinetics of anti-tuberculosis drugs [22,23]. Our results are therefore concordant with existing evidence and suggest that nutritional status may be clinically relevant to long-term treatment outcomes. However, because this estimate was based on only four recurrence events and had a very wide confidence interval, the magnitude of the association should be interpreted with caution. Our study has several limitations. First, the retrospective design, as opposed to a randomized prospective study, introduces inherent limitations such as selection bias and unmeasured confounding, despite our efforts to adjust for known covariates. In addition, our analytic cohort was restricted to patients with DS pulmonary tuberculosis who achieved treatment success with the standard HRZE regimen. Therefore, our findings should be interpreted as comparing recurrence after treatment completion among patients who successfully completed treatment, rather than as estimating the overall efficacy or net clinical benefit of extended treatment. This selection may have introduced survivorship bias and may have underestimated potential harms or competing risks associated with prolonged treatment. Further prospective studies including all eligible patients from treatment initiation are needed to better evaluate the net clinical benefit of extended treatment. Second, the low number of recurrence events represents a major limitation of this study. Bacteriologically confirmed recurrence was uncommon, with only four events observed (4/159, 2.5%). Programmatic factors may have contributed to this low recurrence frequency; in Korea, public–private mix–supported tuberculosis care has been associated with improved adherence support and systematic follow-up, which may reduce both on-treatment complications and post-treatment recurrence [24]. Consequently, this study was underpowered to draw definitive conclusions regarding the effect of extended treatment on recurrence, and the effect estimates are expected to be unstable, as reflected by the wide confidence interval. To mitigate this issue, we applied Firth’s penalized logistic regression, which is specifically designed to reduce small-sample bias and provide more reliable estimates in the presence of rare events. Nevertheless, even with this approach, the extremely small number of events limits the robustness and generalizability of the findings, and the results should be interpreted as exploratory rather than definitive. For the same reason, time-dependent Cox modeling or continuous treatment-duration modeling could not be reliably applied in the present study, as these approaches would have produced highly unstable estimates. This limitation also necessitates careful contextual interpretation in comparison with prior evidence and existing guideline frameworks. Current recommendations emphasize treatment extension particularly for patients with both baseline cavity and persistent culture positivity at two months, a combined phenotype consistently associated with a higher risk of recurrence in previous studies [1,3–10]. In contrast, our study evaluated a broader population defined by cavity on chest radiograph, rather than the narrower subgroup characterized by both cavity and delayed culture conversion. Moreover, two-month culture positivity was rare in our cohort among patients with available results (4/77, 5.2%), which limited our ability to adequately assess outcomes in the guideline-defined highest-risk subgroup. Because the observed effect estimate was based on only four recurrence events and was therefore unstable, we did not base the future sample size estimation on the observed odds ratio. For planning purposes, assuming a clinically meaningful reduction in recurrence from 5% to 1%, a two-sided α of 0.05, 80% power, and 1:1 allocation, a future study would require approximately 285 patients per group, or 570 patients in total. These findings underscore the need for a substantially larger study or a cohort enriched for patients at higher recurrence risk, such as those with both baseline cavitation and persistent culture positivity at two months. Third, there were missing data for variables associated with the outcome. Baseline BMI data were missing for 23 of 159 patients (14.5%), primarily because height and/or weight information was not available in the cohort records. Missing BMI data were more frequent in the standard-treatment group than in the extended-treatment group (24.1% vs. 8.9%; p = 0.017). Given the strong association between low BMI and recurrence, this differential missingness may have introduced selection bias and affected the precision and stability of the model estimates. Therefore, the association between low BMI and recurrence should be interpreted with caution. Similarly, although sputum smear and culture examinations were intended to be performed according to the cohort protocol, two-month culture data were missing in 26 of 58 patients (44.8%) in the standard-treatment group and 56 of 101 patients (55.4%) in the extended-treatment group, with no statistically significant between-group difference (p = 0.249). Sensitivity analyses treating missing two-month culture results as either negative or positive showed no statistically significant between-group difference in two-month culture positivity, and univariable logistic analyses showed no significant association between two-month culture positivity and recurrence under these assumptions (S7 Table). However, these analyses cannot fully overcome the limitation caused by the high proportion of missing two-month culture data; therefore, the prognostic importance of two-month culture positivity and the treatment effect in the guideline-defined subgroup of patients with both baseline cavitation and persistent culture positivity at 2 months could not be reliably assessed. Fourth, treatment extension in our cohort was not governed by predefined protocol-driven criteria, but was determined by the clinical judgment of the treating physicians at each participating center. As a result, patients selected for extended treatment may have had greater baseline disease burden or slower treatment response. Consistent with this possibility, the extended-treatment group had a significantly higher proportion of baseline AFB smear positivity than the standard-treatment group. In addition, the specific reasons for treatment extension, such as perceived disease severity, delayed clinical response, adherence concerns, or institutional practice patterns, were not systematically captured in the case report forms or available medical records and therefore could not be reliably reconstructed retrospectively. Thus, the similar recurrence rates observed between groups should not be interpreted as evidence of equivalence or absence of benefit, because they may reflect either a true protective effect of extended treatment offsetting higher baseline risk, or residual confounding by indication. Although methods such as propensity score matching or inverse probability of treatment weighting may be useful in larger observational cohorts, they were not applied in the present study because the small sample size and only four recurrence events would likely result in unstable propensity models or extreme weights. Fifth, although we adjusted for measured covariates, residual confounding is likely to have remained. In particular, detailed radiologic severity related to cavitary disease burden could not be fully captured in our analysis. Current guideline recommendations are based primarily on the presence or absence of cavitation on chest radiography, rather than detailed characterization of cavitary burden [8]. Consistent with this framework, our cohort captured the presence of cavitation and information on multilobar infiltration as an indicator of overall disease extent. However, previous studies have reported that cavitation-specific features, including the number and size of cavities and multilobar involvement, may serve as indicators of tuberculosis disease burden and may be associated with treatment response and recurrence risk.[6–8,10,25,26]. In our cohort, detailed cavitation-specific variables, such as cavity size, number of cavities, and the extent of cavitary involvement, were not systematically recorded and therefore could not be incorporated into the analysis. The absence of these detailed radiologic data limits our ability to fully adjust for baseline disease severity and may have contributed to residual confounding. Other clinically relevant factors that may have contributed to residual confounding include symptom burden, treatment adherence quality, drug exposure and pharmacokinetics, clinicians’ subjective assessment of treatment response, and additional features plausibly associated with recurrence risk, such as extensive disease, substantial treatment interruptions, immunosuppression, and poorly controlled diabetes [5,9,27]. Sixth, our study design is subject to limitations related to outcome and exposure definitions. The definition of the outcome as bacteriologically confirmed recurrence during the documented post-treatment follow-up period improves specificity but may lead to under-ascertainment if microbiologic testing is not performed at recurrence, if recurrence is diagnosed clinically without culture confirmation, or if recurrence occurred outside the observed follow-up period. Post-treatment recurrence may reflect either relapse or reinfection, and the relative contribution of reinfection may increase with longer intervals after treatment completion [28]. Because molecular typing was also not available, we were unable to distinguish relapse from reinfection. As treatment extension would be expected to preferentially reduce relapse rather than reinfection, the coexistence of these mechanisms may have attenuated the observed association between extended treatment and recurrence risk [29]. Lastly, the use of 200 days as the threshold for defining extended treatment was a pragmatic choice. Although 6 months of daily therapy corresponds to approximately 180 days, real-world practice involves scheduling variability, brief treatment interruptions, and administrative delays. The 200-day cutoff was intended to distinguish minor deviations from more intentional treatment extension [15]; nevertheless, some degree of exposure misclassification is possible. Future studies could incorporate measures of cumulative drug exposure or adherence (e.g., proportion of doses taken), model treatment duration as a continuous variable, or apply flexible methods to explore potential non-linear associations. Conclusion In this cohort of DS pulmonary TB patients with cavity on chest radiograph who successfully completed treatment, the small number of recurrence events limited our ability to definitively determine whether extending therapy beyond 200 days reduces bacteriologically confirmed recurrence. Although the adjusted estimate favored extended treatment, it was imprecise and should be interpreted as exploratory. Larger prospective studies are needed to clarify which patients with cavitary disease may benefit from individualized treatment extension. Supporting information Comparison of baseline laboratory abnormalities between the standard- and extended-treatment groups. (DOCX) Clinical description of bacteriologically confirmed recurrence cases. (DOCX) Univariable logistic regression analysis of baseline variables with recurrence counts. (DOCX) Sensitivity analysis using alternative definitions of extended treatment in a 200-day fixed-time Firth's penalized logistic regression model. (DOCX) Sensitivity analyses using alternative model specifications. (DOCX) Sensitivity analyses of the association between extended treatment and recurrence using alternative covariate adjustments. (DOCX) Sensitivity analyses for missing two-month culture results. (DOCX) 6 May 2026 Dear Dr. Kwon, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. As the corresponding author, your ORCID iD is verified in the submission system and will appear in the published article. PLOS supports the use of ORCID, and we encourage all coauthors to register for an ORCID iD and use it as well. Please encourage your coauthors to verify their ORCID iD within the submission system before final acceptance, as unverified ORCID iDs will not appear in the published article. Only the individual author can complete the verification step; PLOS staff cannot verify ORCID iDs on behalf of authors. We look forward to receiving your revised manuscript. Kind regards, Jung Yeon Heo Academic Editor PLOS One Journal Requirements: When submitting your revision, we need you to address these additional requirements. 1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. Thank you for stating the following financial disclosure: “This work was supported by the Research Program funded by a grant (BCRI24033) from Chonnam National University Hospital Biomedical Research Institute.” Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript." If this statement is not correct you must amend it as needed. Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf. 3. Please note that funding information should not appear in any section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form. Please remove any funding-related text from the manuscript. 4. You have indicated that data is available from [ whitevet81@korea.kr]. Please can we ask you to provide us with a general contact email address for the data requests, so readers can request access in perpetuity. If a general email is not available please provide a link to a website where readers can obtain access to data. 5. In the online submission form you indicate that your data is not available for proprietary reasons and have provided a contact point for accessing this data. Please note that your current contact point is a co-author on this manuscript. According to our Data Policy, the contact point must not be an author on the manuscript and must be an institutional contact, ideally not an individual. Please revise your data statement to a non-author institutional point of contact, such as a data access or ethics committee, and send this to us via return email. Please also include contact information for the third party organization, and please include the full citation of where the data can be found. 6. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information. 7. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. Additional Editor Comments : Thank you for your submission to PLOS ONE. I apologize for the delay in issuing this decision due to my clinical and academic commitments. The manuscript was reviewed by two reviewers, both of whom provided detailed and constructive comments. After considering their evaluations, the decision is major revision. Please revise the manuscript accordingly and submit a detailed point-by-point response to all comments. A careful and thorough revision will be important for further consideration of the manuscript. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? Reviewer #1: Partly Reviewer #2: Yes ********** 2. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #1: Yes Reviewer #2: Yes ********** 3. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy Reviewer #1: Yes Reviewer #2: No ********** 4. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #1: Yes Reviewer #2: Yes ********** Reviewer #1: Major Comments 1. Underpowered study and overstated conclusions The primary analysis yielded an aOR of 0.07 (95% CI 0.01–1.06, p = 0.06), based on only four recurrence events, and did not reach statistical significance. Describing this as "a protective trend" risks overstating the evidence. The conclusion should more explicitly acknowledge that this study was underpowered to draw definitive conclusions. A post-hoc power calculation or sample size estimation for a future adequately powered study should be provided. 2. Absence of predefined indications for treatment extension No protocol-defined criteria for extending treatment are described, and the Discussion acknowledges that extension was largely driven by clinician judgment. If the clinical reasons for treatment extension can be retrieved retrospectively, they should be reported. If not, this should be explicitly acknowledged as a limitation. 3. Radiologic severity of cavitation not captured Cavity size, number, and extent of pulmonary involvement are recognized prognostic factors yet were not incorporated into the analysis. The authors should explore whether these data were recorded and, if so, include them in the analysis. If unavailable, this should be discussed more prominently as a source of residual confounding. 4. Missing BMI data BMI was the strongest independent predictor of recurrence, yet data were available for only 136 of 159 patients (85.5%). The reasons for and distribution of missing BMI data across groups should be reported, and the potential impact on model estimates should be addressed. 5. IRB approval for all participating institutions Only a single IRB approval number is provided for an 18-center study. The authors should confirm that central IRB approval covered all sites or provide individual approval numbers for each institution. 6. Sputum collection intervals The timing and frequency of sputum collection are not described. Given that two-month culture positivity and culture conversion are key outcomes, and the number of patients with available data differed between groups (n = 32 vs. 45), the authors should confirm that sputum collection intervals were comparable between groups. Minor Comments 1. The abstract should include a brief description of the source cohort. 2. Please explicitly state that only patients susceptible to all HRZE components were included, using the term "pan-susceptible" or equivalent, to avoid confusion with the current sequential exclusion criteria. 3. Specific patient numbers in the exclusion process should be moved from the Methods to the Results section or flow diagram legend. 4. "Gender" should be replaced with "sex" throughout the manuscript, consistent with ICMJE recommendations. 5. The rationale for the 200-day cutoff should be more explicitly justified in the Methods, with a clearer explanation of how the referenced study (ref 15) informed this threshold. 6. The average duration of follow-up after treatment completion and its comparability between groups should be reported. 7. P-values in Supplementary Table 2 should be reported to at least three decimal places. 8. A brief clinical description of the four recurrence cases (timing, group assignment, suspected relapse vs. reinfection) would provide useful context for interpreting the primary outcome. Reviewer #2: Dear Authors, I appreciate the effort that went into this multicenter study evaluating the potential benefit of extended treatment in drug-susceptible pulmonary tuberculosis patients with cavitary disease. The study addresses a clinically relevant and timely question, given the limited evidence base supporting current guideline recommendations for treatment extension. The utilization of a well-established prospective cohort from 18 Korean institutions is a notable strength, and the manuscript is clearly written with an appropriate structure and length. However, there are several methodological and interpretive aspects that require further clarification and discussion before the manuscript can be considered for publication. I have outlined my specific comments below. 1. Study Population and Potential Survivorship Bias The study population is restricted to patients who achieved treatment success with the HRZE regimen, having excluded those who experienced treatment failure (n=4), death (n=53), loss to follow-up (n=8), and transfer to other clinics (n=41). This introduces a significant survivorship bias, as patients who received extended treatment but could not complete therapy due to adverse events or who died during treatment are systematically excluded from the analysis. Consequently, the study does not evaluate the “efficacy of extended treatment” as the title implies, but rather compares recurrence rates among patients who successfully completed their respective treatment courses. This distinction is important and should be reflected in the title, abstract, and conclusions. Additionally, the Discussion section should include a dedicated paragraph addressing how this selection may have influenced the results. 2. Extremely Low Number of Recurrence Events and Model Stability With only 4 recurrence events among 159 patients, the multivariable Firth’s penalized logistic regression model including 4 covariates (extended treatment, age, sex, and BMI) yields an events-per-variable ratio below 1. While the choice of Firth’s method is appropriate to address separation, the fundamental limitation imposed by the extremely low event count cannot be fully overcome by any statistical technique. This is clearly reflected in the confidence interval for low BMI (OR 81.92; 95% CI, 4.95–21,694.24), which spans over four orders of magnitude. Characterizing this variable as an independent predictor may overstate the precision and reliability of the estimate. While the direction of association is biologically plausible and consistent with prior literature, the magnitude should be interpreted with explicit caution. It would be beneficial to add appropriate caveats both in the Results and Discussion sections. 3. Confounding by Indication The extended treatment group had significantly higher baseline AFB smear positivity (56.4% vs. 29.3%, p<0.01), suggesting that clinicians selectively extended treatment for patients with perceived greater disease severity. This introduces confounding by indication—a well-recognized limitation in observational treatment-duration studies. If the extended treatment group had a higher baseline disease burden yet showed similar recurrence rates to the standard group, this observation could reflect either a genuine protective effect of extended treatment or residual confounding. The current analysis does not adequately address this issue. The authors should consider discussing the potential utility of methods such as propensity score matching or inverse probability of treatment weighting, and at minimum provide a thorough discussion of how confounding by indication may have affected the interpretation of results. 4. Variable Selection in the Multivariable Model The rationale for variable selection in the multivariable model is not fully transparent. Age and sex were included a priori, and BMI was included based on clinical relevance and consistency with prior evidence. However, in the univariable analysis, diabetes mellitus (OR 5.49, p=0.07) and chronic pulmonary disease (OR 9.86, p=0.06) demonstrated associations with recurrence that were at least as strong as, or stronger than, several included variables, yet they were not selected for the final model. Given the severe constraint imposed by the small number of events, it is understandable that not all variables could be included. Nevertheless, the authors should provide explicit criteria for variable selection and consider presenting alternative model specifications with different covariate combinations in sensitivity analyses to demonstrate the robustness of the findings. 5. Definition of Comorbidities The operational definitions for comorbid conditions are incompletely described. Only chronic pulmonary disease is explicitly defined. For other comorbidities—particularly diabetes mellitus, hypertension, chronic kidney disease, and chronic liver disease—it is unclear whether the diagnoses were based on patient self-report, diagnostic codes, or predefined laboratory criteria (e.g., HbA1c thresholds, eGFR cutoffs). This is especially relevant for diabetes mellitus, which showed a near-significant association with recurrence. The authors should provide clear operational definitions for all comorbid conditions in the Methods section. 6. Definition of Adverse Drug Reactions Table 3 reports multiple categories of adverse drug reactions including hepatotoxicity, nephrotoxicity, ototoxicity, and cutaneous reactions, but no diagnostic criteria or grading systems are provided in the Methods. Only leukopenia is defined (WBC < 4,000 cells/µL) in a table footnote. Since adverse events represent a key secondary outcome of this study, standardized definitions or grading criteria (e.g., CTCAE grades, specific AST/ALT thresholds for hepatotoxicity) should be provided to ensure reproducibility and allow comparison with existing literature. 7. Missing Baseline Laboratory Data The Methods section states that biochemical tests including complete blood count, liver function tests, renal function tests, and inflammation markers were collected at baseline. However, none of these results are presented in the baseline comparison (Table 1). These variables may serve as important confounders that differ between groups and could influence both the clinical decision to extend treatment and the risk of recurrence. It would be beneficial to present these data in the manuscript or supplementary materials, or to explain why they were omitted. 8. Missing Data for 2-Month Culture Results Two-month culture results were available for only 77 of 159 patients (48.4%). This is an important limitation, as the combination of baseline cavity and 2-month culture positivity is the guideline-defined indication for treatment extension. The high proportion of missing data limits the ability to evaluate this clinically relevant subgroup. The authors should discuss whether the missingness was random or systematic, and consider sensitivity analyses that account for missing data. 9. Analytical Approach and the 200-Day Threshold While the rationale for using 200 days rather than 180 days as the threshold is explained, the choice of a fixed-time logistic regression approach rather than time-to-event analysis (e.g., Cox proportional hazards model) or modeling treatment duration as a continuous variable warrants further justification. The Discussion mentions these alternatives as future directions, but a brief explanation of why they were not applied in the current analysis would strengthen the methodological transparency. 10. Tone of Conclusions The conclusion states that “cavity on chest radiograph alone may not be sufficient to justify routine treatment extension beyond 200 days.” Given that only 4 recurrence events were observed and the adjusted analysis showed a trend favoring extended treatment (aOR 0.07, p=0.06), this statement appears to extend beyond what the data can reliably support. A more measured conclusion—such as “this study was underpowered to definitively determine the benefit of routine treatment extension in patients with cavitary disease”—would better reflect the limitations of the evidence. Once again, I appreciate the authors’ efforts in conducting this multicenter study on a clinically important topic. If the above comments are adequately addressed, I believe the manuscript will provide a meaningful contribution to the evidence base regarding treatment duration in cavitary pulmonary tuberculosis. Best regards. ********** what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy Reviewer #1: No Reviewer #2: No ********** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation. NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications. 3 Jun 2026 Response to the editor and reviewers Ref: Submission ID PONE-D-26-05397 Title: Efficacy of Extended Treatment in Drug-Susceptible Pulmonary Tuberculosis with Cavity on Chest Radiograph We sincerely thank the Editor and the Reviewer for their thoughtful and constructive feedback on our manuscript. We have carefully considered and addressed every comment in the revised version of the paper. Below, we present a detailed point-by-point response, with each reviewer’s comment followed by our corresponding reply. All changes have been incorporated into the manuscript, and the tracked-changes document highlights all new or substantially revised sections. Reviewer #1: Major Comments 1. Underpowered study and overstated conclusions The primary analysis yielded an aOR of 0.07 (95% CI 0.01–1.06, p = 0.06), based on only four recurrence events, and did not reach statistical significance. Describing this as "a protective trend" risks overstating the evidence. The conclusion should more explicitly acknowledge that this study was underpowered to draw definitive conclusions. A post-hoc power calculation or sample size estimation for a future adequately powered study should be provided. We thank you for this important comment. We agree that the primary analysis, which was based on only four recurrence events, does not provide sufficient statistical evidence to support a definitive protective effect. Accordingly, we revised the wording throughout the manuscript to avoid overstating the findings. First, in the Abstract and Results section, we removed the term “trend” from the description of the Firth penalized logistic regression analysis and replaced it with more objective wording (lines 97 and 289–292). Similarly, in the sensitivity analysis, we removed the phrase “protective trend” and revised the text using more objective language (lines 306-307). In addition, at the beginning of the Discussion, we deleted the sentence containing the phrase “protective trend” to further avoid overstating the results (lines 318–324). In addition, we have removed the statements in the Results section regarding the benefits of extended treatment to avoid the overstatement you pointed out. Instead, we have emphasized the potential risks of unnecessary treatment extension and its associated adverse effects, and revised the text to highlight the need for careful selection of patients who would require treatment extension in future practice. (Lines 351-355 in result section) In the Limitations section, we also strengthened the statement that the small sample size limits the generalizability of the findings We replaced wording suggesting cautious interpretation with a clearer statement that the results should be interpreted as exploratory rather than definitive, thereby emphasizing the limited generalizability of the findings (Second paragraph in limitation section). Furthermore, we added a description of the design considerations for a future study, including a specific sample size estimation, at the end of the second paragraph of the Limitations section. Finally, consistent with Reviewer 2’s comment on the tone of the conclusion, we revised the concluding statement of the Discussion to avoid overstating our findings. We removed the phrase “suggest a potential benefit in selected patients” and instead emphasized that the study was underpowered, the adjusted estimate was imprecise, and the findings should be interpreted as exploratory. (Conclusion paragraph of the Discussion section) In addition, as part of our revision and statistical verification process, we carefully re-examined the Firth penalized logistic regression results presented in Table 4. During this review, we noted that the originally calculated univariable estimates for age and sex were not accurately reflected in the previous version of the table. We sincerely apologize for this oversight and have corrected these values in the revised table. Importantly, the multivariable primary model results, including the adjusted estimate for extended treatment, remained unchanged, and these corrections do not affect the overall interpretation of the study findings. 2. Absence of predefined indications for treatment extension No protocol-defined criteria for extending treatment are described, and the Discussion acknowledges that extension was largely driven by clinician judgment. If the clinical reasons for treatment extension can be retrieved retrospectively, they should be reported. If not, this should be explicitly acknowledged as a limitation. Thank you for your insightful comment. We agree that treatment extension in our cohort was not governed by predefined protocol-driven criteria but was determined by the clinical judgment of the treating physicians at each participating center. However, the reasons for treatment extension were not documented in the cohort records and therefore could not be reliably retrieved retrospectively. Accordingly, we have explicitly acknowledged this issue in the fourth limitation paragraph of the revised Discussion. We clarified that the absence of predefined indications for treatment extension and the inability to retrieve the specific rationale for each extension decision may have introduced residual confounding by indication. 3. Radiologic severity of cavitation not captured Cavity size, number, and extent of pulmonary involvement are recognized prognostic factors yet were not incorporated into the analysis. The authors should explore whether these data were recorded and, if so, include them in the analysis. If unavailable, this should be discussed more prominently as a source of residual confounding. We thank you for this important comment. As shown in our results, information on multilobar infiltration on chest radiography was available as an indicator of overall disease extent. However, this cohort did not include detailed radiologic data specifically characterizing cavitary disease burden, such as cavity size, number of cavities, or the extent of cavitary involvement. Therefore, these cavitation-specific severity measures could not be reliably retrieved retrospectively or incorporated into the analysis. Accordingly, we expanded the fifth limitation paragraph of the revised Discussion to address this issue more explicitly. We noted that recent studies have suggested that cavitation-specific features, including cavity size, number of cavities, and extent of cavitary involvement, may serve as indicators of tuberculosis disease burden and recurrence risk, and we additionally cited the relevant supporting references. We also clarified that the lack of detailed radiologic information beyond the binary presence of cavitation may have limited our ability to fully adjust for baseline disease severity. 4. Missing BMI data BMI was the strongest independent predictor of recurrence, yet data were available for only 136 of 159 patients (85.5%). The reasons for and distribution of missing BMI data across groups should be reported, and the potential impact on model estimates should be addressed. We thank you for this very insightful comment. (i) Reasons for missing BMI data: The missing BMI values were attributable to the absence of height and/or weight information in the cohort records, rather than intentional exclusion. Unfortunately, no additional documentation was available to determine the specific reasons for these missing measurements. (ii) Distribution of missing data across groups and potential impact on model estimates: BMI was missing in 14 of 58 patients (24.1%) in the standard-treatment group and in 9 of 101 patients (8.9%) in the extended-treatment group, with a statistically significant difference between groups (p = 0.017). We agree that this imbalance is important, as complete-case analysis may have introduced selection bias. Given that BMI was identified as an important predictor of recurrence, differential missingness in BMI may also have affected the precision and stability of the model estimates. Accordingly, we have added a new third paragraph to the Limitations section of the revised Discussion. In this paragraph, we describe the imbalance in missing BMI data between treatment groups and explicitly acknowledge that, because BMI was an important predictor of recurrence, this differential missingness may have introduced selection bias and should be considered when interpreting the findings. 5. IRB approval for all participating institutions Only a single IRB approval number is provided for an 18-center study. The authors should confirm that central IRB approval covered all sites or provide individual approval numbers for each institution. We thank the reviewer for this important comment. We confirm that the study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Boards (IRBs) and Ethics Committees of all participating institutions. The study was coordinated and overseen by the representative institution, the Catholic University of Korea (IRB No. C19ONDI0458), which served as the coordinating center for the 17 other participating medical institutions. In accordance with the reviewer's suggestion, we have now provided the individual IRB approval numbers for each participating institution in the revised 'Ethics Statement' subsection of the Methods section, as follows: The study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Boards (IRBs) and Ethics Committees of all participating institutions. The study was coordinated and overseen by the representative institution, the Catholic University of Korea (IRB No. KC19ONDI0458), which represented and managed the study across the 17 other participating medical institutions. All participating institutions obtained IRB approval prior to study initiation, with the following approval numbers: SCHBC202012023001-HE001 (Soonchunhyang University Bucheon Hospital), UUH202103016-HE002 (Ulsan University Hospital), GCIRB2021-016 (Gachon University Gil Hospital), 2020GR0402 (Korea University Guro Hospital), VC19ONDI016 (St. Vincent's Hospital), KANGDONG2020-06-012-001 (Kangdong Sacred Heart Hospital), KYUH2020-12-013-002 (Konyang University Hospital), SCHCA202012076-HE005 (Soonchunhyang University Cheonan Hospital), DUIH202101023001-HE001 (Dongguk University Ilsan Hospital), 2020-07-007 (Chungnam National University Hospital), 2020-08-012-003 (Chungbuk National University Hospital), DC19ONDI0052 (Daejeon St. Mary's Hospital), HC19ONDI0068 (Bucheon St. Mary's Hospital), UC19ONDI0106 (Uijeongbu St. Mary's Hospital), PC19ONDI0149 (Eunpyeong St. Mary's Hospital), OC19ONDI0073 (Incheon St. Mary's Hospital), and ISPAIK2020-11-003-001 (Ilsan Paik Hospital). Written informed consent was obtained from all adult participants before their enrolment in the study. 6. Sputum collection intervals The timing and frequency of sputum collection are not described. Given that two-month culture positivity and culture conversion are key outcomes, and the number of patients with available data differed between groups (n = 32 vs. 45), the authors should confirm that sputum collection intervals were comparable between groups. We sincerely thank you for this insightful and important comment. We fully agree that the timing and frequency of sputum collection are important considerations when interpreting two-month culture positivity and culture conversion outcomes. According to the cohort protocol, sputum smear and culture examinations were intended to be performed as part of routine follow-up during anti-tuberculosis treatment, including assessment at approximately 2 months after treatment initiation. Therefore, the intended sputum collection schedule was comparable between the standard- and extended-treatment groups. Nevertheless, two-month culture results were not available for all patients in this real-world multicenter cohort, despite the intended protocol-based follow-up schedule. This incomplete availability may have reflected several factors encountered in routine clinical practice, including inability to expectorate sputum after clinical improvement, variation in visit timing around the two-month assessment window, clinician discretion regarding follow-up testing, differences in institutional practice patterns, or incomplete capture of test results in the cohort records. In response to this comment, and in addition to the issue of missing BMI data addressed in Comment 4, we further expanded the third limitation paragraph of the revised Discussion to address missing data for clinically important variables more broadly. Specifically, we acknowledged that the incomplete availability of two-month culture data may have limited our ability to fully evaluate the prognostic importance of two-month culture positivity and its association with recurrence. Minor Comments 1. The abstract should include a brief description of the source cohort. We thank you for this helpful comment. We have revised the Abstract to include a clearer description of the source cohort. Specifically, we clarified that this study was a retrospective analysis using data from a multicenter prospective cohort of patients with pulmonary tuberculosis enrolled at 18 Korean institutions, and that the present analysis focused on patients with drug-susceptible pulmonary tuberculosis and cavity on chest radiograph. 2. Please explicitly state that only patients susceptible to all HRZE components were included, using the term “pan-susceptible” or equivalent, to avoid confusion with the current sequential exclusion criteria. Thank you for this helpful suggestion. To clarify the eligibility criteria and avoid confusion with the sequential exclusion criteria, we revised the Methods section to explicitly state that only patients with pan-susceptible tuberculosis were included. Specifically, we added that the study population was restricted to patients susceptible to all HRZE components (‘population’ subsection of Methods section) 3. Specific patient numbers in the exclusion process should be moved from the Methods to the Results section or flow diagram legend. Thank you very much for this helpful and thoughtful suggestion. We agree that presenting the specific patient numbers for each exclusion step in the Results section or flow diagram is clearer and more appropriate than including them in the Methods section. Accordingly, we have revised the Methods section to describe the eligibility and exclusion criteria without specific patient numbers, and have moved the detailed exclusion counts to the Results section, with reference to Figure 1. (‘Population’ Subsection of Method section) (Result section: lines 226-234) 4. "Gender" should be replaced with "sex" throughout the manuscript, consistent with ICMJE recommendations. Thank you very much for this important suggestion. We have revised the manuscript to replace “gender” with “sex” throughout, in accordance with ICMJE recommendations. (line 153) 5. The rationale for the 200-day cutoff should be more explicitly justified in the Methods, with a clearer explanation of how the referenced study (ref 15) informed this threshold. Thank you very much for this valuable suggestion. We agree that the rationale for the 200-day cutoff should be more explicitly described to improve the clarity and reproducibility of our Methods. Accordingly, we have revised the Methods section to clarify that the cutoff was based on the standard 6-month regimen, corresponding to approximately 180 days, with an additional 2-week allowance. We also clarified that this allowance was informed by a previous study of prolonged TB treatment [15], which noted that treatment interruptions of ≥2 weeks during the intensive phase may require reintroduction of intensive-phase treatment and thereby prolong the total treatment duration. Thus, the 200-day cutoff was selected to distinguish meaningful treatment extension from minor variations around the standard 6-month course. (‘Group and study outcome’ subsection of Methods section: lines 182-189) 6. The average duration of follow-up after treatment completion and its comparability between groups should be reported. Thank you very much for this helpful suggestion. We agree that reporting the follow-up Submitted filename: Response to review.docx 12 Jul 2026 Dear Dr. Kwon, Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Please submit your revised manuscript by Aug 26 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols. As the corresponding author, your ORCID iD is verified in the submission system and will appear in the published article. PLOS supports the use of ORCID, and we encourage all coauthors to register for an ORCID iD and use it as well. Please encourage your coauthors to verify their ORCID iD within the submission system before final acceptance, as unverified ORCID iDs will not appear in the published article. Only the individual author can complete the verification step; PLOS staff cannot verify ORCID iDs on behalf of authors. We look forward to receiving your revised manuscript. Kind regards, Jung Yeon Heo Academic Editor PLOS One Journal Requirements: 1. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. 2. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice. Additional Editor Comments : The multivariable Firth penalized logistic regression should be presented more explicitly as an exploratory analysis. Although Firth’s method is appropriate for sparse data and separation, it cannot overcome the fundamental limitation imposed by only four recurrence events. Moreover, because BMI was missing in 23 patients, the adjusted model appears to have been based on a complete-case sample and potentially fewer than four recurrence events. Please clearly report how missing BMI data were handled, as well as the exact number of patients and recurrence events included in the multivariable model, preferably in a footnote to Table 4. The adjusted estimates should not be described as demonstrating independent effects or model stability. Instead, they should be framed as highly imprecise, hypothesis-generating associations. The terms “primary analysis,” “independent predictor,” and “independent determinant” should therefore be replaced with more cautious wording such as “exploratory adjusted analysis” and “association with recurrence.” The Abstract should also avoid emphasizing the extreme adjusted OR for low BMI without clearly stating its substantial imprecision. [Note: HTML markup is below. Please do not edit.] Reviewers' comments: Reviewer's Responses to Questions Comments to the Author Reviewer #1: All comments have been addressed Reviewer #2: All comments have been addressed ********** 2. Is the manuscript technically sound, and do the data support the conclusions??> Reviewer #1: Partly Reviewer #2: Yes ********** 3. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #1: Yes Reviewer #2: Yes ********** 4. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy Reviewer #1: Yes Reviewer #2: Yes ********** 5. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #1: Yes Reviewer #2: Yes ********** Reviewer #1: I thank the authors for their thorough and, in several places, commendably candid response to the previous round of review. The revision meaningfully improves the manuscript: the title now reflects the analytic population, the tone of the Discussion conclusion is appropriately hedged, operational definitions for comorbidities and adverse drug reactions have been added, baseline laboratory data and missing-data patterns are now reported, and useful sensitivity analyses have been supplied. The fundamental constraint of this study ("four recurrence events") cannot be resolved by any analytic method, and the authors now reasonably frame their findings as exploratory. My remaining concerns are therefore not requests for new analyses, but requests that the reporting be made fully consistent with that exploratory framing and fully transparent about what the primary model actually estimates. Several of these should be addressed before the manuscript is suitable for publication. Major points 1. The Abstract conclusion remains overstated and is now inconsistent with the revised Discussion. In the response letter and in the Discussion (final paragraph), the authors state that they removed language suggesting a benefit in selected patients and reframed the findings as exploratory and imprecise. However, the Abstract conclusion still reads that "our findings indicate a potential benefit of extended treatment in selected patients, warranting prospective evaluation of risk-stratified extension strategies." As the Abstract is what most readers will see, please revise it to match the hedged wording of the Discussion conclusion. 2. The extreme low-BMI estimate is presented in the Abstract without any indication of imprecision. The Abstract reports an adjusted OR of 81.92 (95% CI, 4.95–21,694.24). This estimate arises from near-complete separation (0/110 vs. 3/26) and its magnitude is not interpretable. Consistent with the caveat the authors added to the Discussion, I recommend removing the numeric estimate from the Abstract and instead stating that low BMI was associated with recurrence but that the estimate was very imprecise. 3. The analytic sample size and number of events contributing to the multivariable model must be reported. The primary multivariable model includes BMI, which is missing in 23 patients (14.5%), and no imputation is described, implying a complete-case analysis of approximately 136 patients. Because three of the four recurrence events occurred in the low-BMI group and the fourth appears to fall among the BMI-missing patients, the multivariable model may be based on as few as three events. Please state explicitly (a) how missing BMI was handled, and (b) the actual N and number of recurrence events contributing to the multivariable model, ideally in a footnote to Table 4. This information is essential for readers to judge the primary estimate. Minor points 4. Given the numeric corrections required in this revision, I would ask the authors to independently re-verify all figures in the final version. Specific inconsistencies I noted: - The Discussion reports baseline AFB smear positivity in the standard group as 30.2%, whereas Table 1, the Abstract, and the Results state 29.3%. Please reconcile. - The coordinating-institution IRB number appears as "C19ONDI0458" in the Ethics Statement but as "KC19ONDI0458" elsewhere. Please use a single consistent number. - Two-month culture positivity is reported as 5.1% in the Discussion but 5.2% in Table 2; please standardize the rounding. 5. Supplementary materials. Please confirm that the final Supporting Information file is complete and that its table numbering matches the in-text citations (S1–S7), including the recurrence-case summary and the sensitivity analyses added in this revision. The numbering of the supplementary tables and the in-text references should be checked against one another. Overall, I consider these to be addressable revisions. Provided with appropriate minor revisions, the manuscript would represent a sound, appropriately hedged contribution of real-world data on this clinically relevant question. Reviewer #2: Dear Authors, Thank you for your thorough and thoughtful revision. The manuscript has been substantially improved, and I appreciate the considerable effort taken to address each of my previous comments. The revised title now accurately reflects the study population, the survivorship bias and confounding by indication are explicitly discussed, comorbidity and adverse drug reaction definitions have been added, baseline laboratory data and missing data analyses are now provided, and the Discussion conclusion has been appropriately tempered. The addition of a sample size estimation for future studies and the clinical description of the four recurrence cases are also valuable. I have only two remaining concerns, both of which should be straightforward to address. 1. Inconsistency Between the Abstract Conclusion and the Discussion Conclusion The Discussion conclusion has been revised to state that the study was underpowered and that findings should be interpreted as exploratory, which I believe is appropriate. However, the Abstract conclusion still retains the original wording: “Cavity on chest radiograph alone may not be sufficient to justify routine treatment extension beyond 200 days after treatment success. Nevertheless, our findings indicate a potential benefit of extended treatment in selected patients.” This language is more assertive than what the Discussion now conveys and creates an internal inconsistency within the manuscript. Because the Abstract is often the only section read by many clinicians, it is important that its tone matches the more measured Discussion conclusion. I would suggest revising the Abstract conclusion to align with the revised Discussion. 2. Residual Use of “Independent Determinant” in the Discussion In the Discussion paragraph on low BMI (line 378), the phrase “independent determinant of long-term treatment outcomes” remains. While the cautionary sentence that follows is appreciated, the term “independent determinant” still implies a level of certainty that is not supported by an estimate based on four events with a confidence interval spanning four orders of magnitude. Replacing this with more cautious language, consistent with the revisions already made elsewhere in the manuscript, would ensure internal consistency. These are minor points that should not require substantial rewriting. Once addressed, I believe the manuscript will be suitable for publication. Best regards. ********** what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy Reviewer #1: No Reviewer #2: No ********** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation. NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications. 16 Jul 2026 We sincerely thank the Editor and Reviewers for their thoughtful and constructive feedback on our manuscript. We are grateful for the opportunity to revise the manuscript further. We have carefully considered all comments and revised the manuscript accordingly to improve its clarity, consistency, and transparency. Below, we provide a point-by-point response to the Editor’s and Reviewers’ comments. All changes have been incorporated into the revised manuscript and are shown in the tracked-changes version. Additional Editor Comments: The multivariable Firth penalized logistic regression should be presented more explicitly as an exploratory analysis. Although Firth’s method is appropriate for sparse data and separation, it cannot overcome the fundamental limitation imposed by only four recurrence events. Thank you for this insightful comment. We agree that the multivariable Firth penalized logistic regression should be presented more explicitly as exploratory, given the very small number of recurrence events. We revised the wording throughout the manuscript, including the Abstract conclusion, Results, and Discussion, to clarify that the adjusted analysis was exploratory and that the findings should not be interpreted as definitive evidence. Moreover, because BMI was missing in 23 patients, the adjusted model appears to have been based on a complete-case sample and potentially fewer than four recurrence events. Please clearly report how missing BMI data were handled, as well as the exact number of patients and recurrence events included in the multivariable model, preferably in a footnote to Table 4. We also appreciate this important point regarding the handling of missing BMI data. Because BMI data were missing in 23 patients, the multivariable model including BMI was conducted as a complete-case analysis without imputation. Accordingly, this model included 136 patients and 3 recurrence events; one recurrence case with missing BMI was excluded. To make this information clear and transparent for readers, we added a footnote to Table 4 specifying the analytic sample size and the number of recurrence events included in the multivariable model. (Table 4, Lines 319-321) The adjusted estimates should not be described as demonstrating independent effects or model stability. Instead, they should be framed as highly imprecise, hypothesis-generating associations. The terms “primary analysis,” “independent predictor,” and “independent determinant” should therefore be replaced with more cautious wording such as “exploratory adjusted analysis” and “association with recurrence.” Thank you for these insightful comments. First, in the Statistical Analysis subsection of the Methods, we revised the wording in the sentences explaining why Cox proportional hazards modeling and time-dependent or continuous treatment-duration modeling were not applied. Specifically, we replaced the term “primary analysis” with “the main analytic approach” to avoid implying a definitive primary causal framework. (‘Statistical analysis’ subsection of Method, lines 206,210) In the Results section, we also replaced “primary analysis” with “exploratory adjusted analysis” when describing the multivariable Firth penalized logistic regression results, to more clearly reflect the exploratory nature of this analysis. (Result section, line 307) In addition, we revised the Discussion paragraph on low BMI by replacing phrasing that could imply low BMI was an “independent” determinant with more cautious language, stating that nutritional status may be clinically relevant when considering long-term treatment outcomes. We then linked this statement to the caution that interpretation is needed because the finding was based on very few recurrence events and had a wide confidence interval. (Discussion section, lines 382-384) In addition, because wording related to “stability” in the sensitivity analysis section could be misinterpreted as implying model stability, we revised the sentence to state more cautiously that analyses using alternative model specifications with varying degrees of covariate adjustment showed a similar direction of association. (Result section, lines 328-329) The Abstract should also avoid emphasizing the extreme adjusted OR for low BMI without clearly stating its substantial imprecision. We also revised the Abstract to avoid emphasizing the extreme adjusted OR for low BMI. Specifically, we removed the numerical adjusted OR and confidence interval for low BMI from the Abstract and instead stated that low BMI was associated with recurrence, although the estimate was highly imprecise. ( Result of Abstract, lines 100-101) Reviewer #1: I thank the authors for their thorough and, in several places, commendably candid response to the previous round of review. The revision meaningfully improves the manuscript: the title now reflects the analytic population, the tone of the Discussion conclusion is appropriately hedged, operational definitions for comorbidities and adverse drug reactions have been added, baseline laboratory data and missing-data patterns are now reported, and useful sensitivity analyses have been supplied. The fundamental constraint of this study ("four recurrence events") cannot be resolved by any analytic method, and the authors now reasonably frame their findings as exploratory. My remaining concerns are therefore not requests for new analyses, but requests that the reporting be made fully consistent with that exploratory framing and fully transparent about what the primary model actually estimates. Several of these should be addressed before the manuscript is suitable for publication. Major points 1. The Abstract conclusion remains overstated and is now inconsistent with the revised Discussion. In the response letter and in the Discussion (final paragraph), the authors state that they removed language suggesting a benefit in selected patients and reframed the findings as exploratory and imprecise. However, the Abstract conclusion still reads that "our findings indicate a potential benefit of extended treatment in selected patients, warranting prospective evaluation of risk-stratified extension strategies." As the Abstract is what most readers will see, please revise it to match the hedged wording of the Discussion conclusion. We thank the reviewer for this important observation and agree that the Abstract conclusion was inconsistent with the revised Discussion. We have revised the Abstract conclusion to align with the hedged, exploratory framing used in the Discussion, while retaining the study's core clinical message (i.e., that cavity alone does not justify routine extension) so that the Abstract continues to convey the paper's principal contribution rather than reading as a mere restatement of its limitations. We have also revised the wording so that the reference to microbiologic response and disease burden explicitly refers to variables that future studies should systematically capture, consistent with our Limitations section noting that detailed cavitation-specific variables (cavity size, number, extent) were not recorded in the present cohort and that two-month culture data had substantial missingness. The revised Abstract conclusion now reads: “Cavity on chest radiograph alone may not be sufficient to justify routine treatment extension beyond 200 days after treatment success. Because only four recurrence events were observed, the adjusted association favoring extended treatment was imprecise and should be interpreted as exploratory rather than definitive. Larger prospective studies that systematically capture microbiologic response and detailed radiologic disease burden are needed to identify which patients with cavitary disease may benefit from individualized treatment extension” 2. The extreme low-BMI estimate is presented in the Abstract without any indication of imprecision. The Abstract reports an adjusted OR of 81.92 (95% CI, 4.95–21,694.24). This estimate arises from near-complete separation (0/110 vs. 3/26) and its magnitude is not interpretable. Consistent with the caveat the authors added to the Discussion, I recommend removing the numeric estimate from the Abstract and instead stating that low BMI was associated with recurrence but that the estimate was very imprecise. Thank you for this insightful comment. We agree that presenting the numerical low-BMI estimate in the Abstract may overemphasize an unstable estimate based on very few recurrence events. Accordingly, we removed the adjusted OR and confidence interval from the Abstract and revised the sentence to state that low BMI was associated with recurrence, while emphasizing that the estimate was highly imprecise. The revised sentence in the Abstract Results now reads: "Among covariates, low body mass index was associated with recurrence, but this estimate was very imprecise and should be interpreted with caution." In addition, consistent with this revision, we also modified the wording in the Discussion regarding low BMI. Specifically, we replaced phrasing that could imply low BMI was an “independent” determinant with more cautious language, stating that nutritional status may be clinically relevant when considering long-term treatment outcomes. We then linked this statement to the caution that interpretation is needed because the finding was based on very few recurrence events and had a wide confidence interval. The revised sentence in the Discussion section now reads: “Our results are therefore concordant with existing evidence and suggest that nutritional status may be clinically relevant to long-term treatment outcomes.” Furthermore, although the numeric estimate was retained in the Results section of the main text for completeness and transparency, we revised that sentence to explicitly flag its imprecision, so that readers encountering the estimate outside the Abstract are not misled by its magnitude. The revised sentence in the Results section now reads: "In contrast, low body mass index (BMI) (<18.5 kg/m²) was associated with recurrence in the multivariable Firth penalized logistic regression model (OR, 81.92; 95% CI, 4.95–21,694.24; p < 0.001), although this estimate was highly imprecise due to sparse data (see Table 4 footnote)." 3. The analytic sample size and number of events contributing to the multivariable model must be reported. The primary multivariable model includes BMI, which is missing in 23 patients (14.5%), and no imputation is described, implying a complete-case analysis of approximately 136 patients. Because three of the four recurrence events occurred in the low-BMI group and the fourth appears to fall among the BMI-missing patients, the multivariable model may be based on as few as three events. Please state explicitly (a) how missing BMI was handled, and (b) the actual N and number of recurrence events contributing to the multivariable model, ideally in a footnote to Table 4. This information is essential for readers to judge the primary estimate. Thank you for this important and insightful comment. We agree that explicitly reporting the analytic sample size and the number of recurrence events contributing to the multivariable model would improve the transparency and interpretability of the results. Because BMI data were missing in 23 patients, the primary multivariable Firth penalized logistic regression model, which included BMI, was conducted as a complete-case analysis without imputation. Accordingly, this model included 136 patients and 3 recurrence events; one recurrence case with missing BMI was excluded. To make this clearer for readers, we added a footnote to Table 4 specifying the actual analytic sample size and event count used in the multivariable model. (Result section, Table 4 footnote, lines 319-321). In addition, in Table 4, we revised confidence interval lower bounds that had been rounded to 0.00 and reported them as <0.01 to avoid misinterpretation. (Table 4) Minor points 4. Given the numeric corrections required in this revision, I would ask the authors to independently re-verify all figures in the final version. Specific inconsistencies I noted: - The Discussion reports baseline AFB smear positivity in the standard group as 30.2%, whereas Table 1, the Abstract, and the Results state 29.3%. Please reconcile. Thank you for pointing out this inconsistency. We have corrected the baseline AFB smear positivity in the Discussion from 30.2% to 29.3% for the standard-treatment group, so that it is consistent with Table 1, the Abstract, and the Results section. (Discussion section, lines 358) - The coordinating-institution IRB number appears as "C19ONDI0458" in the Ethics Statement but as "KC19ONDI0458" elsewhere. Please use a single consistent number. Thank you for carefully noting this inconsistency. After rechecking the coordinating-institution IRB approval number, we standardized it as “KC19ONDI0458” throughout the manuscript to ensure consistency. (Methods section, lines 227) - Two-month culture positivity is reported as 5.1% in the Discussion but 5.2% in Table 2; please standardize the rounding. Thank you for carefully noting this inconsistency. We have standardized the reported proportion of two-month culture positivity as 5.2% in the Discussion, consistent with the value presented in Table 2. (Discussion section, lines 419) 5. Supplementary materials. Please confirm that the final Supporting Information file is complete and that its table numbering matches the in-text citations (S1–S7), including the recurrence-case summary and the sensitivity analyses added in this revision. The numbering of the supplementary tables and the in-text references should be checked against one another. Thank you for this helpful comment. We carefully reviewed the final Supporting Information file and cross-checked all supplementary table numbers against the corresponding in-text citations. We confirmed that the Supporting Information file includes Supporting Tables S1–S7, including the recurrence-case summary and the sensitivity analyses added during revision, and that the numbering of the supplementary tables is consistent with the in-text references. Reviewer #2: Dear Authors, Thank you for your thorough and thoughtful revision. The manuscript has been substantially improved, and I appreciate the considerable effort taken to address each of my previous comments. The revised title now accurately reflects the study population, the survivorship bias and confounding by indication are explicitly discussed, comorbidity and adverse drug reaction definitions have been added, baseline laboratory data and missing data analyses are now provided, and the Discussion conclusion has been appropriately tempered. The addition of a sample size estimation for future studies and the clinical description of the four recurrence cases are also valuable. I have only two remaining concerns, both of which should be straightforward to address. 1. Inconsistency Between the Abstract Conclusion and the Discussion Conclusion The Discussion conclusion has been revised to state that the study was underpowered and that findings should be interpreted as exploratory, which I believe is appropriate. However, the Abstract conclusion still retains the original wording: “Cavity on chest radiograph alone may not be sufficient to justify routine treatment extension beyond 200 days after treatment success. Nevertheless, our findings indicate a potential benefit of extended treatment in selected patients.” This language is more assertive than what the Discussion now conveys and creates an internal inconsistency within the manuscript. Because the Abstract is often the only section read by many clinicians, it is important that its tone matches the more measured Discussion conclusion. I would suggest revising the Abstract conclusion to align with the revised Discussion. Thank you for this important and helpful comment. We fully agree that the Abstract conclusion should be consistent with the more cautious wording of the Discussion conclusion, particularly given the limited number of recurrence events. Accordingly, we have revised the Abstract conclusion to more clearly emphasize the small number of recurrence events, the imprecisi Submitted filename: Response letter.docx 23 Aug 2026

Extended Treatment Duration and Recurrence in Successfully Treated Drug-Susceptible Pulmonary Tuberculosis with Cavity on Chest Radiograph PONE-D-26-05397R2 Dear Dr. Kwon, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support. If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. Kind regards, Jung Yeon Heo Academic Editor PLOS One Additional Editor Comments: IRB approval number remains inconsistent in the methods section. Please verify the correct IRB approval number and ensure that it is reported consistently throughout the manuscript. Reviewers' comments: Reviewer's Responses to Questions Comments to the Author Reviewer #1: All comments have been addressed Reviewer #2: All comments have been addressed ********** 2. Is the manuscript technically sound, and do the data support the conclusions??> Reviewer #1: Yes Reviewer #2: Yes ********** 3. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #1: Yes Reviewer #2: Yes ********** 4. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy Reviewer #1: Yes Reviewer #2: Yes ********** 5. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #1: Yes Reviewer #2: Yes ********** Reviewer #1: I thank the authors for their complete response. All three major points are fully addressed, and the minor numeric inconsistencies have also been reconciled. The manuscript now presents its findings honestly as exploratory, and I have no further substantive concerns. One production-stage point, not requiring further review: the coordinating-institution IRB number reads "KC19ONDI0458" in the Methods but "C19ONDI0458" in the Ethics Statement field; please make the two consistent. Reviewer #2: Dear authrors, Thank you for careful and sound response to comments from the reviewer. All requested revisions have been made accurately. The manuscript now presents a transparent and appropriately cautious analysis of an important clinical question. Despite the inherent limitation of four recurrence events, the study provides meaningful real-world evidence from a well-characterized multicenter Korean cohort and clearly frames its findings as hypothesis-generating. I have no further comments and am satisfied that the manuscript is suitable for publication. Best regards. ********** what does this mean?). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy Reviewer #1: No Reviewer #2: No ********** PONE-D-26-05397R2 PLOS One Dear Dr. Kwon, I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS One. Congratulations! Your manuscript is now being handed over to our production team. At this stage, our production department will prepare your paper for publication. This includes ensuring the following: * All references, tables, and figures are properly cited * All relevant supporting information is included in the manuscript submission, * There are no issues that prevent the paper from being properly typeset You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps. Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org. You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing. If we can help with anything else, please email us at customercare@plos.org. Thank you for submitting your work to PLOS One and supporting open access. Kind regards, PLOS One Editorial Office Staff on behalf of Dr. Jung Yeon Heo Academic Editor PLOS One