Lung function impairment at pulmonary TB and HIV/TB diagnosis: a South African cross-sectional cohort analysis Authors: Michael Marll, Tlhago Ngwanto, Nomsa Mofokeng, Lydia Mngomezulu, Sally El Sammak, Anisha Rajaratnam, Fay Willis, Matthew Magee, Lungile Gabuza, Hardy Kornfeld, Yun Li, Pholo Maenetje, Gregory P. Bisson, Sara Auld Published: 2026 ABSTRACT AbstractObjectivesTo characterise the patterns and severity of impaired lung function at pulmonary tuberculosis (TB) diagnosis and assess the impact of HIV co-infection on TB-associated lung injury.SettingA cross-sectional analysis of adults with and without HIV who presented with newly diagnosed drug-susceptible pulmonary TB in Johannesburg, South Africa.ParticipantsA total of 258 adults who were newly diagnosed with drug-susceptible pulmonary TB underwent comprehensive pulmonary function tests (PFT).Primary and secondary outcome measuresFive key PFT manoeuvres were evaluated: forced expiratory volume in one second (FEV1), forced vital capacity (FVC), FEV1/FVC, total lung capacity (TLC) and haemoglobin-corrected diffusion capacity of the lungs for carbon monoxide (DLCO). Three primary outcomes were modelled: (1) normal versus abnormal lung function, defined as any key PFT manoeuvre z-score < lower limit of normal, (2) continuous z-scores for each PFT and (3) clinical patterns of lung function (normal, restriction, obstruction, mixed disorder and isolated low DLCO). Secondary outcomes included respiratory health-related quality-of-life measures assessed using the St. George’s Respiratory Questionnaire (SGRQ).ResultsHIV/TB co-infection was associated with lower odds of any abnormal lung function (adjusted OR (aOR) 0.43, 95% CI 0.24 to 0.74) and lower odds of restrictive (aOR 0.46, 95% CI 0.24 to 0.87) and mixed patterns (aOR 0.22, 95% CI 0.07 to 0.66) compared with those with TB alone. HIV co-infection was also associated with better FEV1, FVC, FEV1/FVC and TLC z-scores. While those with HIV had less severe radiographic disease, respiratory symptoms were similar between groups (SGRQ total score: HIV-negative 22 (IQR: 13–38) vs HIV-positive 26 (IQR: 8–44), p>0.9).ConclusionsAdults with HIV/TB co-infection demonstrated a distinct clinical phenotype of TB-associated lung injury at TB diagnosis characterised by less severe radiographic and lung function impairment compared with those without HIV. WHAT IS ALREADY KNOWN ON THIS TOPIC Tuberculosis (TB) is the leading infectious killer of people living with HIV worldwide, yet core measures of lung function in HIV/TB co-infection at TB diagnosis remain under characterised. Tuberculosis (TB) is the leading infectious killer of people living with HIV worldwide, yet core measures of lung function in HIV/TB co-infection at TB diagnosis remain under characterised. WHAT THIS STUDY ADDS In this large cross-sectional cohort study of South African adults with newly diagnosed rifampicin-susceptible TB, people with HIV co-infection had significantly better lung function at TB diagnosis. Therefore, HIV co-infection manifests with a distinct clinical phenotype of lung function at the time of TB diagnosis. In this large cross-sectional cohort study of South African adults with newly diagnosed rifampicin-susceptible TB, people with HIV co-infection had significantly better lung function at TB diagnosis. Therefore, HIV co-infection manifests with a distinct clinical phenotype of lung function at the time of TB diagnosis. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Longitudinal clinical-translational studies are needed to characterise the mechanistic drivers of TB-associated lung injury in HIV co-infection to meaningfully improve long-term lung health outcomes for those afflicted by this infectious syndemic. Longitudinal clinical-translational studies are needed to characterise the mechanistic drivers of TB-associated lung injury in HIV co-infection to meaningfully improve long-term lung health outcomes for those afflicted by this infectious syndemic. Introduction Tuberculosis (TB) and HIV are two of the leading infectious causes of death worldwide, with over 10 million new TB cases estimated annually and nearly one in 10 occurring among people living with HIV.1 Despite increasing access to anti-retroviral therapy (ART), the risk of developing TB in people living with HIV is approximately 16-times greater than those without HIV.2 Furthermore, HIV co-infection complicates TB care, as evidenced by the higher rates of TB-related hospitalisations, drug resistance, treatment failure and disease recurrence seen among those with HIV.36 As a result, TB is the leading infectious killer of people living with HIV worldwide.2 Despite the profound impact of HIV on TB susceptibility, treatment success and survival, relatively little is known about the nature and severity of lung injury in HIV/TB co-infection that may underlie these disparate outcomes. Concurrently, HIV is recognised as a major risk factor for non-infectious pulmonary disease. A recent systematic meta-analysis found individuals with HIV face a greater risk of obstructive lung disease, impairment in gas exchange and accelerated declines in pulmonary function even in the era of widespread ART use.7 8 While the recent literature sheds light on the role of HIV in mediating chronic non-infectious pulmonary disease, the impact of HIV on TB-associated lung injury remains less clear.9 Previous cohort studies characterising patterns of lung function impairment in people with HIV/TB co-infection have shown contradicting associations with either preserved or more severe functional impairments after TB treatment completion.1013 These discordant findings may be attributable to differences in HIV status and ART use, the timing of lung function assessments and the reliance on spirometry alone, an insensitive diagnostic tool for characterising the full spectrum of pulmonary physiology and disease.14 Therefore, assessment of comprehensive pulmonary function tests (PFTs) including spirometry, lung volumes and diffusion capacity at TB diagnosis in people with and without HIV is needed to understand the physiological impacts of TB on lung function and clarify the role of HIV on TB-associated lung injury. In this cross-sectional analysis of a prospective cohort study conducted in Johannesburg, South Africa, we evaluated adults with first-time, rifampicin-susceptible pulmonary TB with and without HIV co-infection using comprehensive PFT, chest X-ray (CXR) imaging and health-related quality-of-life metrics. By capturing early and multidimensional measures of lung health in TB and HIV/TB co-infection, we aim to clarify the role HIV may have in TB-associated lung injury. Methods Study design and population The Inflammation and Fibrosis in Pulmonary Tuberculosis study is an ongoing prospective observational cohort based in Johannesburg, South Africa. Adults with and without HIV were enrolled within 2 weeks of a first-time episode of rifampicin-susceptible pulmonary TB between March 2022 and May 2025. Study visits were conducted at the Tembisa Clinical Research Centre of The Aurum Institute where participants were identified through local TB registries. Inclusion criteria were age ≥18 years, diagnosis of rifampicin-susceptible pulmonary TB using the Xpert MTB/RIF assay (Cepheid, California, USA), initiation of anti-tubercular therapy within 2 weeks of enrolment and ability to provide informed consent. Exclusion criteria included inability to schedule a baseline visit, prior TB, advanced clinical disease precluding study procedures, inability to produce a sputum sample, pregnancy and COVID-19 infection (figure 1). To characterise the impact a single pulmonary TB episode may have on respiratory health, individuals with a history of TB episodes were excluded to minimise the contributions from prior TB-associated lung damage. All participants received standard first-line TB treatment regimens and ART per South African guidelines in local Department of Health clinics. Participant enrolment flow diagram. hCG, Human chorionic gonadotropin; TB, tuberculosis. Patient and public involvement Participants and the public were not involved in the design, conduct or interpretation of this secondary analysis. Study procedures At screening, participants completed a structured interview and examination in their preferred language to collect demographic data, medical history, body mass index (BMI), COVID-19 screening results and HIV history. Participants without known HIV or recent documentation of a negative HIV test as part of their initial evaluation underwent HIV testing at enrolment. Those with HIV co-infection underwent additional testing for CD4+T cell count and HIV RNA viral load quantification (Abbott, Illinois, USA). CXRs were performed within 2 weeks of the baseline visit and scored by study physicians as minimal, moderately advanced or far advanced using validated severity grading criteria (online supplemental figure S1).15 Enrolment initially prioritised participants with moderate-to-far advanced radiographic disease. After 76 enrolments, eligibility expanded to include all radiographic disease severities consistent with pulmonary TB. Respiratory symptoms were assessed using the St George’s Respiratory Questionnaire (SGRQ). A subset of participants (n=70) completed an abbreviated SGRQ form prior to identification of missing items. All subsequent participants completed the full SGRQ. Participants with missing baseline SGRQ data were excluded from respiratory symptom analyses. PFTs, including spirometry, lung volumes by multiple-breath nitrogen washout and diffusion capacity of the lungs for carbon monoxide (DLCO), were performed within 2 weeks of TB diagnosis using the EasyOne Pro LAB Pulmonary Function System (ndd Medical Technologies, Massachusetts, USA). DLCO values were adjusted for participants’ haemoglobin levels measured at the time of enrolment. The Global Lung Initiative (GLI) —global reference ranges were used to generate z-scores according to American Thoracic Society (ATS)/Europena Respiratory Society (ERS) guidelines.16 Patterns of physiological impairment (obstruction, restriction, mixed pattern and isolated low DLCO) were defined using ATS/ERS criteria.16 When total lung capacity (TLC) was unavailable, spirometry was used to identify a restrictive pattern (ie, forced vital capacity (FVC) < lower limit of normal (LLN) and forced expiratory volume in one second (FEV1) / FVC > LLN), allowing for inclusion of participants unable to complete lung volume testing (online supplemental table S1). Statistical analysis The primary objective was to assess the association between HIV status and measures of lung function at TB diagnosis. Clinical variables, PFTs and radiographic scores were compared by HIV status using Student’s t-tests or Wilcoxon rank-sum tests for continuous variables and χ2 or Fisher’s exact tests for categorical variables. Regression modelling was used to assess the relationship of HIV and abnormal pulmonary function at the time of TB diagnosis. Five key PFT manoeuvres (FEV1, FVC, FEV1/FVC, TLC and haemoglobin-corrected DLCO) were evaluated. Three outcomes of lung function were then modelled: (1) a binary indicator of normal versus abnormal lung function, defined as any key PFT manoeuvre z-score < LLN, (2) continuous z-scores for each key PFT manoeuvre and (3) clinical patterns of lung function (normal, restriction, obstruction, mixed disorder and isolated low DLCO). Binary outcomes were modelled using logistic regression, continuous outcomes using linear regression and categorical outcomes using multinomial regression. Potential confounders were identified by directed acyclic graph theory and data-driven analyses of exposure-outcome relationships (online supplemental figure S2). Potential confounders included age, sex, BMI and ever-smoking status. Radiographic severity was not included in multivariable models because it likely lies on the causal pathway between HIV and lung function abnormalities in pulmonary TB. Exclusion of radiographic severity allows models to assess the overall association between HIV status and lung function measures. To assess the robustness of the primary findings, the following sensitivity analyses were conducted. First, the prevalence of abnormal lung function and the distribution of clinical PFT patterns were re-assessed after stratifying by HIV and ART use (ie, HIV-negative, HIV-positive on ART, HIV-positive not on ART). Second, unadjusted and adjusted logistic regression models were generated to assess the potential impact of ART use prior to TB diagnosis and absolute CD4+T cell levels at TB diagnosis (per 100 cell/mm3 increase) with abnormal versus normal lung function among participants with HIV co-infection. Last, multivariable regression models of lung function were restricted to participants with far advanced radiographic disease to assess the assumption that structural lung injury mediates the effect of HIV on lung function. Given the multiple comparisons performed, we applied the Benjamini-Hochberg procedure to control the false discovery rate (FDR). Statistical significance was defined as an FDR adjusted p value <0.10. Analyses and data visualisation were performed using R V.4.4.3. Results Clinical demographics and radiography Of 349 participants screened for eligibility, 91 were excluded due to inability to schedule a baseline visit (n=27), minimal radiographic disease on CXR (prior to protocol modification as above; n=23), drug resistance on Xpert MTB/RIF testing (n=9), severe clinical disease precluding completion of study procedures (n=9) and other reasons (n=23) (figure 1). 258 participants were enrolled, with a median age of 36 years, of whom 172 (67%) were men and 122 (47%) had HIV co-infection (table 1). Although most participants reported ever-smoking (n=142, 55%), participants with HIV were less likely to have ever smoked (47% vs 63%, p=0.04). The completion of PFT, radiographic assessments and participant questionnaires did not differ by HIV status (online supplemental table S2). Table 1Baseline demographic and clinical characteristics by HIV statusOverall, n=258*HIV (−), n=136*HIV (+), n=122*P value†‡Clinical characteristics Age36 (30, 44)33 (28, 43)38 (34, 46)0.01 Sex0.2 Male172 (67%)97 (71%)75 (61%) Female86 (33%)39 (29%)47 (39%) Ever smoker142 (55%)85 (63%)57 (47%)0.04 Smoking pack years15 (8, 22)12 (8, 20)16 (10, 23)0.3 BMI (kg/m2)19.5 (17.5, 22.0)19.4 (17.5, 21.0)19.9 (17.5, 23.0)0.2 Race/ethnicity0.5 Black252 (98%)132 (97%)120 (98%) Coloured5 (1.9%)4 (2.9%)1 (0.8%) White1 (0.4%)0 (0%)1 (0.8%)CXR involvement<0.01 Mild27 (10%)3 (2%)24 (20%) Moderately advanced81 (31%)42 (31%)39 (32%) Far advanced150 (58%)91 (67%)59 (48%)HIV-specific characteristics ART use54 (44%) No prior ART use68 (56%) HIV viral load (copies/mL)14 722(40, 187 332) CD4+ count (cells/µL)230 (101, 371) Undetectable viral load (<200 copies/mL)45 (38%)Missing data: HIV viral load (n=3), CD4 cell count (n=3).*Median (Q1, Q3); n (%)—percentages are column percentages.†Wilcoxon rank sum test; Pearson’s χ2 test; Fisher’s exact test.‡Statistical significance was defined as a FDR-adjusted p value <0.10.ART, antiretroviral therapy; BMI, body mass index; CXR, chest X-ray; FDR, false discovery rate. Baseline demographic and clinical characteristics by HIV status Missing data: HIV viral load (n=3), CD4 cell count (n=3). Median (Q1, Q3); n (%)—percentages are column percentages. Wilcoxon rank sum test; Pearson’s χ2 test; Fisher’s exact test. Statistical significance was defined as a FDR-adjusted p value <0.10. ART, antiretroviral therapy; BMI, body mass index; CXR, chest X-ray; FDR, false discovery rate. All 258 enrolled participants underwent CXR imaging at the screening visit. Far advanced radiographic disease, defined by any cavity ≥4 cm or slight to moderate density lesions in more than half the chest or dense lesions in more than one-sixth of the chest, was the most common overall pattern on CXR and occurred more often among participants without HIV co-infection than among those with HIV (67% vs 48%, p<0.01; table 1). At the time of TB diagnosis, 122 participants (56%) with HIV co-infection had not previously received ART. Among HIV co-infected participants on ART, median HIV viral loads (40 copies/mL (IQR:40–60) vs 144 531 copies/mL (IQR: 27 386–3 58 053), p<0.001) were significantly lower while median CD4+T cell count (338 cells/mm3 (IQR: 234–591) vs 132 cells/mm3 (IQR: 57–279), p<0.001) was significantly higher compared with those not on ART. In those with HIV co-infection, 38 (72%) individuals on ART and two (3%) individuals not on ART were found to have ‘undetectable levels’ of viral RNA (<50 copies/mL).17 18 Pulmonary function While median PFT z-scores across the entire cohort were above the LLN (table 2), 139 (56.3%) participants had abnormal lung function, defined as any of the study outcome PFT z-scores < LLN (ie, z-score <−1.65). Participants with HIV had significantly higher median z-scores for FEV1 (−1.01 vs −1.84, p<0.01), FVC (−1.03 vs −1.55, p=0.01), FEV1/FVC ratio (−0.09 vs −0.41, p=0.02) and TLC (−0.82 vs −1.77, p<0.01) than those without HIV. DLCO z-scores did not differ by HIV status (−0.56 vs −0.81, p=0.30). Table 2Comprehensive pulmonary function testing stratified by HIV statusCharacteristicOverall, n=247*HIV (−), n=131*HIV (+), n=116*P value‡‡Abnormal versus normal lung function<0.01 Abnormal139 (56.3%)86 (65.6%)53 (45.7%) Normal108 (43.7%)45 (34.4%)63 (54.3%)PFT z-scores FEV1−1.39 (−2.51, −0.34)−1.84 (−2.81, −0.83)−1.01 (−2.17, −0.12)<0.01 FVC−1.35 (−2.23, −0.35)−1.55 (−2.45, −0.58)−1.03 (−1.88, −0.19)0.01 FEV1/FVC−0.24 (−1.36, 0.55)−0.41 (−1.73, 0.30)−0.09 (−1.01, 0.78)0.01 TLC−1.24 (−2.47, −0.07)−1.77 (−2.97, −0.52)−0.82 (−1.60, 0.25)<0.01 DLCO−0.65 (−1.84, 0.48)−0.81 (−2.05, 0.55)−0.56 (−1.80, 0.48)0.30*Median (Q1, Q3); n (%)—percentages are row percentages.†Wilcoxon rank sum test; Pearson’s χ2 test; Fisher’s exact test.‡Statistical significance was defined as an FDR-adjusted p value<0.10.DLCO, diffusion capacity of the lung for carbon monoxide corrected for participant haemoglobin level; FDR, false discovery rate; FEV1, forced expiratory volume in one second; FVC, forced vital capacity; PFT, pulmonary function test; TLC, total lung capacity. Comprehensive pulmonary function testing stratified by HIV status Median (Q1, Q3); n (%)—percentages are row percentages. Wilcoxon rank sum test; Pearson’s χ2 test; Fisher’s exact test. Statistical significance was defined as an FDR-adjusted p value<0.10. DLCO, diffusion capacity of the lung for carbon monoxide corrected for participant haemoglobin level; FDR, false discovery rate; FEV1, forced expiratory volume in one second; FVC, forced vital capacity; PFT, pulmonary function test; TLC, total lung capacity. Within patterns of impaired lung function (obstruction, restriction, mixed pattern and isolated low DLCO), restriction was the most common pattern seen among all participants (figure 2). There was no significant difference in the proportion and severity of impairment patterns by HIV status at TB diagnosis (online supplemental table S3). Categorisation and severity of PFT patterns of impairment at the time of TB diagnosis stratified by HIV status. Severity was determined using ATS/ERS 2021 guideline recommendations for z-score cutoffs of mild (−1.65 to −2.5), moderate (−2.51 to −4.0) and severe (<−4.0). Participants with HIV co-infection were more likely to have ‘Normal’ patterns of pulmonary function at the time of TB diagnosis (p=0.01; Online supplemental table S4). ATS, American Thoracic Society; DLCO, diffusion capacity of the lung for carbon monoxide corrected for participant haemoglobin level; ERS, European Respiratory Society; PFT, pulmonary function test; TB, tuberculosis. Several complementary approaches were used to examine the association between HIV co-infection and impaired lung function while adjusting for differences in age, sex, BMI, and ever-smoking status. First, in an adjusted logistic regression model, participants with HIV had 57% lower odds of abnormal lung function compared with those without HIV (adjusted OR (aOR) 0.43, 95% CI 0.24 to 0.74, p<0.01). Using adjusted linear regression models to examine individual PFT measures as z-scores, HIV co-infection was associated with better lung function across several key measures. Participants with HIV had a 0.53 higher FEV1 z-score, indicating better lung function, compared with those without HIV (95% CI 0.18 to 0.89, p<0.01). Similar findings were seen for FVC (β=0.39, 95% CI 0.05 to 0.73, p=0.07) and FEV1/FVC (β=0.47, 95% CI 0.08 to 0.86, p=0.09). Participants with HIV exhibited the greatest difference in TLC z-score compared with other key PFT measures (β=1.10, 95% CI 0.56 to 1.7, p<0.01), while there was no significant difference in DLCO (β=0.07, −0.41 to 0.55, p=0.8). Finally, in adjusted multinomial models of clinical PFT patterns (obstruction, restriction, mixed pattern and isolated low DLCO) with normal lung function as the referent category, HIV was associated with lower odds of restrictive lung disease (aOR 0.46, 95% CI 0.24 to 0.87, p=0.07) and mixed pattern (aOR 0.22, 95% CI 0.07 to 0.66, p<0.01). HIV status was not significantly associated with obstructive lung disease or isolated low DLCO (aOR 0.80, 95% CI 0.26 to 2.45, p=0.80) at TB diagnosis. Across all three approaches for assessing pulmonary function outcomes (abnormal vs normal, continuous PFT z-scores and clinical PFT patterns), there was no meaningful change in the effect size or statistical significance between unadjusted and adjusted models (table 3). Table 3Unadjusted and adjusted logistic, linear and multinomial regression models to estimate the association between HIV and pulmonary function outcomes at TB diagnosisLogistic regressionUnadjustedAdjustedHIV (+) versus HIV (−) OR(95% CI)P value*HIV (+) versus HIV (−) OR(95% CI)P value*Abnormal lung function0.44(0.26 to 0.73)<0.010.43(0.24 to 0.74)<0.01Normal lung functionRef--Ref--Linear regressionUnadjustedAdjustedβ HIV (+) versus HIV (−)(95% CI)P value*β HIV (+) versus HIV (−)(95% CI)P value* FEV1 z-score0.65(0.29 to 1.0)<0.010.53(0.18 to 0.89)<0.01 FVC z-score0.49(0.14 to 0.84)0.030.39(0.05 to 0.73)0.07 FEV1/FVC z-score0.51(0.13 to 0.89)0.030.47(0.08 to 0.86)0.09 TLC z-score1.1(0.51 to 1.70)<0.011.1(0.54 to 1.8)<0.01 DLCO z-score0.2(−0.30 to 0.70)0.60.07(−0.41 to 0.55)0.8Multinomial regressionUnadjustedAdjustedHIV (+) versus HIV (−) OR(95% CI)P value*HIV (+) versus HIV (−) OR(95% CI)P value* Restriction0.49(0.27 to 0.89)0.080.46(0.24 to 0.87)0.07 Obstruction0.42(0.18 to 1.00)0.140.40(0.16 to 1.00)0.2 Mixed pattern0.20(0.07 to 0.57)0.020.22(0.07 to 0.66)0.05 Isolated low DLCO0.82(0.28 to 2.41)0.80.80(0.26 to 2.45)0.8 NormalRef--Ref-- Multivariable models adjusted for age, sex, BMI and ever-smoking status*Statistical significance was defined as an FDR-adjusted p value <0.10.BMI, body mass index; DLCO, diffusion capacity of the lung for carbon monoxide corrected for participant haemoglobin level; FDR, false discovery rate; FEV1, forced expiratory volume in one second; FVC, forced vital capacity; TB, tuberculosis ; TLC, total lung capacity. Unadjusted and adjusted logistic, linear and multinomial regression models to estimate the association between HIV and pulmonary function outcomes at TB diagnosis Multivariable models adjusted for age, sex, BMI and ever-smoking status Statistical significance was defined as an FDR-adjusted p value <0.10. BMI, body mass index; DLCO, diffusion capacity of the lung for carbon monoxide corrected for participant haemoglobin level; FDR, false discovery rate; FEV1, forced expiratory volume in one second; FVC, forced vital capacity; TB, tuberculosis ; TLC, total lung capacity. Sensitivity analyses were performed to assess the robustness of the main findings. First, stratification of HIV status by ART use prior to enrolment yielded similar frequencies and distributions of abnormal lung function and clinical PFT patterns (online supplemental table S4). Among participants with HIV co-infection, prior ART exposure (aOR 0.98, 95% CI 0.43 to 2.24, p>0.9) and CD4+T cell level at the time of TB diagnosis (per 100 cell/mm3 increase; aOR: 0.99, IQR: 0.84–1.16, p>0.9) were not associated with abnormal lung function at TB diagnosis (online supplemental table S6 and S7). Second, when restricting the multivariable logistic, linear and multinomial models to participants with far-advanced radiographic disease on CXR, HIV status remained significantly associated with better continuous measures of lung function (online supplemental table S7). Specifically, in this subgroup of far-advanced radiographic disease, HIV was associated with the following estimate differences in PFT z-scores: FEV1, β=0.55 (95% CI0.09 to 1.0, p=0.05); FEV1/FVC, β=0.70 (95% CI 0.16 to 1.2, p=0.06); TLC, β=1.2 (95% CI 0.35 to 2.0, p=0.03). Respiratory symptoms 70 participants (27%) within the eligible cohort had incomplete SGRQ results due to an error in the original study case report forms. There was no significant difference in HIV status between those with missing SGRQ and complete SGRQ results (online supplemental table S8). Among 188 participants with complete SGRQ scores, the median overall total SGRQ score was 23 (IQR: 12–40) and median overall scores were similar for symptom (32 (IQR: 23–43)), activity (30 (IQR: 0–52)) and impact (30 (IQR: 0–52)) domains. There were no significant differences in domain scores (symptoms, activity, impact) or total scores by HIV status (online supplemental table S9). Discussion In this study of South African adults with rifampicin-susceptible pulmonary TB, individuals with HIV/TB co-infection demonstrated better lung function at TB diagnosis compared with those without HIV. Among participants with impaired lung function, HIV/TB co-infection was associated with significantly lower odds of restrictive and mixed pattern disease compared with TB alone. Despite these differences in pulmonary function outcomes, respiratory symptoms were highly prevalent and did not differ by HIV status. To date, studies characterising the impact of HIV on TB-related impairments in lung function remain limited. Consistent with prior studies where impaired spirometry was observed in 48–66% of individuals at HIV/TB diagnosis, we found abnormal lung function to be common irrespective of HIV status (56.3%), though more frequent in those without HIV co-infection (HIV-positive: 45.6% vs HIV-negative: 65.6%, p<0.01).13 19 Across the cohort, abnormal lung function was predominantly characterised by restrictive physiology, aligning with previous studies that defined a restrictive pattern based on a reduced FVC.13 19 However, direct comparative analyses assessing TB-associated lung function in groups with and without HIV co-infection are rare. To our knowledge, only one prior study evaluated this relationship at TB diagnosis, reporting a lower unadjusted odds of any abnormal spirometry among those with HIV/TB co-infection compared with those without HIV (OR 0.66, 95% CI 0.47 to 0.99, p=0.05).20 However, prior studies on TB-associated lung function relied on spirometry alone, an insensitive modality for capturing the full spectrum of respiratory disease, thereby limiting the characterisation of HIV’s impact on TB-related lung damage. This analysis sought to clarify the relationship of HIV and pulmonary function at TB diagnosis by using spirometry, lung volumes and diffusion capacity assessments soon after TB diagnosis. After adjustment for age, sex, BMI and smoking history, HIV co-infection was associated with lower odds of any abnormal lung function (aOR 0.43, 95% CI 0.24 to 0.74, p<0.01), better continuous measures of spirometry (FEV1, FVC, FEV1/FVC) and lung volumes (TLC) and lower odds of restrictive (aOR 0.46, 95% CI 0.24 to 0.87, p=0.07) and mixed pattern (aOR 0.22, 95% CI 0.07 to 0.66, p<0.01) impairments. The relative preservation of lung function observed in people with HIV at the time of TB diagnosis may reflect an altered host immune response to Mycobacterium tuberculosis. Prior studies demonstrated that individuals with advanced HIV (CD4+count <200/mm3) more frequently present with normal CXRs and are less likely to develop cavitary TB disease.21 Consistent with this, HIV/TB co-infection in our cohort was associated with significantly milder radiographic disease at diagnosis. The atypical radiographic appearance of HIV/TB co-infection is thought to arise from an impaired adaptive immune response within the lungs, marked by a depletion in effective Mtb-specific CD4+T cells, an expansion of impaired CD8+T cells, and an attenuated pro-inflammatory cytokine response to Mtb.22 23 This diminished immune response to Mtb may underlie the milder radiographic disease and preserved lung function measures observed in our study. However, differences in radiographically apparent lung damage did not fully account for the better lung function outcomes seen with HIV/TB co-infection. In sensitivity analyses restricted to participants with far-advanced radiographic disease, HIV co-infection remained associated with higher FEV1, FEV1/FVC and TLC measures. This finding likely reflects the limited discriminatory capacity of subjective CXR severity scoring to characterise discrete features of lung injury in TB and HIV/TB co-infection. Moreover, similar radiographic patterns of lung injury on CXR may overlap with heterogeneous physiologic impairments influenced by the presence of HIV co-infection. These observations highlight the need to characterise molecular endotypes underlying HIV/TB-associated lung injury beyond radiographic and functional impairments. In addition to performing lung volume testing, this study measured single-breath diffusion capacity, an assessment of pulmonary gas exchange that has not been previously well characterised in African populations living with HIV or at the time of pulmonary TB diagnosis.8 24 HIV is a known risk factor for chronic diffusion impairments in high-resource settings, particularly after a prior TB episode.2426 Interestingly, our study found the prevalence of an isolated low DLCO pattern to be relatively rare at TB diagnosis (6.1%) and continuous DLCO z-scores were less severely impacted across both HIV co-infected and HIV uninfected groups (HIV-negative: −0.56 (IQR: −1.80 to 0.48), HIV-positive: −0.81 (IQR: −2.05 to 0.55), p=0.30) compared with median z-scores of spirometric and lung volume measures. Similar findings were recently reported by Byanova et al in a cross-sectional analysis of Ugandan adults with and without HIV where an isolated low DLCO pattern was uncommon (6.5%) and the prevalence of DLCO z-scores below the LLN did not differ by HIV serostatus (HIV-positive: 25.2%, HIV-negative: 18.5%, p=0.09). However, direct comparisons between these studies are limited as we assessed lung function at the time of TB diagnosis whereas Byanova et al studied participants with a significant prior pneumonia history.24 Overall, we hypothesise that the absence of substantial differences in DLCO outcomes by HIV serostatus may reflect competing biological mechanisms across groups. For individuals without HIV co-infection, greater structural lung damage, as evidenced by more severe CXR involvement at TB diagnosis, may drive lower DLCO values, while HIV/TB co-infected participants with less severe radiographic involvement on average may have a propensity towards HIV-related deficits in DLCO. Given the strong association between persistent diffusion impairments and increased mortality in other chronic lung diseases, future analyses identifying predictors of recoverable versus persistent diffusion deficits after TB treatment will be critical to understanding long-term morbidity and mortality in post-TB lung disease (PTLD).2729 Despite better lung function measures seen with HIV co-infection at TB diagnosis, respiratory symptoms assessed using SGRQ did not differ by HIV status. This result is supported by prior literature demonstrating comparable rates of TB-associated cough, weight loss and night sweats among those with and without HIV.3032 The discordance in people with HIV co-infection between better physiological function yet a similarly high prevalence of respiratory symptoms may be partially explained by the modest correlation between SGRQ scores and spirometry (−0.12 to −0.29, p<0.05) previously demonstrated in TB survivors.33 Furthermore, people living with HIV report high rates of respiratory symptoms even in the absence of an identifiable pulmonary abnormality.34 Last, HIV-related immunosuppression appears to drive TB-related morbidity, where lower CD4+T cell counts strongly correlate with increasing respiratory symptoms at TB diagnosis.3537 Therefore, ongoing research assessing the effect of immune function on respiratory symptoms early in TB diagnosis may shed light on the long-term drivers of TB-related respiratory morbidity that are not otherwise captured on PFTs. Strengths of this study include a robust analytic approach using multiple definitions of lung function impairment, multivariable regression modelling to address confounding and correction for multiple comparisons. Importantly, this study uniquely presents comprehensive pulmonary function results (spirometry, lung volumes and diffusion capacity) early in TB diagnosis, allowing for a more complete characterisation of TB-associated lung injury. Several limitations warrant consideration. First, the cross-sectional design of this analysis precludes the assessment of lung function over time, particularly in response to TB and HIV-specific treatments. As discussed above, clinically significant declines in lung function may occur later in HIV/TB co-infection following restoration of the immune system through ART initiation, as seen in tuberculosis-immune reconstitution inflammatory sydrome.38 39 Future analyses of this cohort study will focus on differences in lung function trajectories by HIV status during and after treatment. Second, although GLI—global reference equations were used to generate standardised z-scores, these equations are based on populations that under-represent South Africans and may not capture the normative range of lung function in this population. Third, unmeasured determinants of lung function not accounted for in our multivariable modelling may remain, such as variable duration of disease onset prior to TB diagnosis, differences in access to medical care and socioeconomic status. Notably, time since HIV diagnosis, an important driver of respiratory morbidity in people living with HIV, was not assessed. While sensitivity analyses incorporating prior ART use and CD4+ levels did not reveal clear associations with abnormal lung function, additional analyses are needed to assess the impact of HIV disease duration, control and severity on lung function during and after TB treatment. Furthermore, residual confounding may be present through improved access to healthcare among individuals with HIV receiving ART prior to study enrolment compared with those with TB disease alone. Last, the study population included those with rifampicin-susceptible TB and was initially enriched for those with more advanced radiographic disease, which may not be generalisable to populations with drug-resistant TB or milder presentations of rifampicin-susceptible TB. In conclusion, HIV co-infection was associated with better lung function at the time of pulmonary TB diagnosis, despite a persistently high burden of respiratory symptoms. Given that impaired lung function at TB diagnosis is a strong predictor of persistent lung function impairment after TB cure, understanding HIV’s influence on early lung injury during TB treatment has significant implications for improving long-term outcomes and reducing rates of PTLD among TB survivors with HIV.1319 3941 These findings emphasise the need for longitudinal and mechanistic studies to characterise the unique pathophysiological drivers of HIV/TB-associated lung injury in this highly vulnerable population. Supplementary material