The evaluation of different strategies to improve the diagnosis of tuberculosis in people living with HIV in resource-limited settings

Dhana, Ashar Vijay

The evaluation of different strategies to improve the diagnosis of tuberculosis in people living with HIV in resource-limited settings

Doctoral Thesis

2023

Abstract

Background The 2019 WHO screening and diagnostic algorithm for tuberculosis in people living with HIV (PLHIV) has 2 components: the WHO Xpert MTB/RIF (Xpert) algorithm and WHO Alere Determine TB-LAM (AlereLAM) algorithm. According to the WHO Xpert algorithm, WHO recommends that PLHIV be routinely screened for tuberculosis with the WHO foursymptom screen (W4SS; comprising any one of current cough, fever, night sweats, or weight loss) and, if the screen is positive, receive Xpert or Xpert MTB/RIF Ultra (Xpert Ultra) confirmatory testing. According to the WHO AlereLAM algorithm, WHO also recommends that PLHIV be routinely screened for tuberculosis using screening criteria and, if the screen is positive, receive urine lateral-flow lipoarabinomannan (LF-LAM) confirmatory testing with AlereLAM. We aimed: i. To determine the diagnostic accuracy of the W4SS and alternative screening tools and strategies in ambulatory PLHIV, including key subgroups, and to compare the diagnostic accuracy of the WHO Xpert algorithm with Xpert confirmatory testing for all ambulatory PLHIV ii. To determine the performance of the W4SS and alternative screening tools and strategies in HIV-positive inpatients and to compare the diagnostic accuracy of the WHO Xpert algorithm with Xpert confirmatory testing for all HIV-positive inpatients iii. To determine the performance of WHO screening criteria and alternative screening tools and strategies to guide LF-LAM testing in HIV-positive inpatients and to compare the performance of the WHO AlereLAM algorithm with AlereLAM and Fujifilm SILVAMP TB-LAM (FujiLAM; a novel LF-LAM test) confirmatory testing in all HIV-positive inpatients. iv. To develop and validate novel clinical prediction models (CPMs) for tuberculosis screening in outpatient PLHIV and to determine the clinical utility of these CPMs and WHO-recommended screening tools Methods We conducted a systematic review and individual participant data (IPD) meta-analysis. We updated a search of PubMed (MEDLINE), Embase, Cochrane Library, and conference 2 abstracts for publications from Jan 1, 2011, to March 12, 2018, done in a previous systematic review to include the period up to August 2, 2019 (objectives i and iv) and March 1, 2020 (objectives ii and iii). We also screened reference lists of identified pieces and contacted experts in the field. We included prospective cross-sectional studies, observational studies, and randomized trials that enrolled adult and adolescent (age ≥10 years) PLHIV irrespective of symptoms and signs of tuberculosis. We also included studies that enrolled outpatient PLHIV with a positive W4SS (objective iv only). We extracted study-level data using a standardized data extraction form, and we requested IPD from study authors. The reference standards were culture (objectives i, ii, and iv) and culture or Xpert (objective iii). For screening tools and strategies, we also used separate reference standards of Xpert (objective i and ii), AlereLAM (objective iii), and FujiLAM (objective iii). We selected these confirmatory tests as reference standards since these tests are the most likely confirmatory tests used in practice. We obtained pooled proportion estimates with a random-effects model, assessed diagnostic accuracy (i.e., sensitivity and specificity) by fitting random-effects bivariate models, and assessed diagnostic yield (i.e., proportion of total tuberculosis cases with a positive confirmatory test) descriptively. For CPMs, we first used logistic regression, allowing for non-linear relations, to develop an extended CPM (using backwards selection of C-reactive protein [CRP] and other predictors) and a CRP-only CPM (which only included CRP along with spline transformations); we then used internal-external cross-validation to evaluate discrimination, calibration, and clinical utility (i.e., decision curve analysis) of both CPMs and other screening strategies. Decision curve analysis plots net benefit across a range of risk thresholds. This systematic review has been registered with PROSPERO, CRD42020155895. Results i. We obtained data for 22 of 25 studies (n= 15,666 participants; 4,347 on antiretroviral therapy [ART]). W4SS sensitivity was 82% (95% CI 72, 89) and specificity was 42% (29, 57). CRP (≥10 mg/L) had similar sensitivity (77% [61, 88]), but higher specificity (74% [61, 83]; n=3571). Cough (lasting ≥2 weeks), haemoglobin (< 8 g/dL), body mass index (<18.5kg/m²), and lymphadenopathy had high specificities (80–90%) but low sensitivities (29–43%). The WHO Xpert algorithm had a sensitivity of only 58% (50,66) and a specificity of 99% (98, 100); Xpert for all had a sensitivity of 68% (57–76) and similar specificity. In the only study that compared both tests, the sensitivity of sputum Xpert Ultra was higher than sputum Xpert (73% [62, 81] vs 57% [47, 67]) and specificities were similar. Among outpatients on ART, W4SS sensitivity was 53% (35, 71) and specificity was 71% (51, 85). In this population, a parallel strategy (two or more screening tests offered at the same time) of W4SS with any chest X-ray abnormality had higher sensitivity (89% [70, 97]) and lower specificity (33% [17, 54]; n=2,670) than W4SS alone; at a 5% tuberculosis prevalence, this strategy would require 379 more Xpert tests per 1,000 PLHIV than W4SS but detect 18 more cases. Among outpatients not on ART, W4SS sensitivity was 85% (76, 91) and specificity was 37% (25, 51). CRP (≥10 mg/L) had a similar sensitivity (83% [79, 86]), but higher specificity (67% [60, 73]; n=3,187) and a sequential strategy (second screening test offered only if first screening test is positive) of W4SS then CRP (≥5 mg/L) also had similar sensitivity (84% [75, 90]) but higher specificity (64% [57, 71]; n=3187); at 10% tuberculosis prevalence, these CRP-based strategies would require 272 and 244 fewer Xpert tests per 1,000 PLHIV than W4SS but miss two and one more cases, respectively. ii. We obtained data for all six eligible studies (n=3,660 participants). The pooled proportion of inpatients eligible for Xpert was 90% (89, 91; n=3,658). Among screening tools to guide Xpert testing, W4SS and CRP (≥5 mg/L) had highest sensitivities (≥96%) but low specificities (≤12%); cough (≥2 weeks), haemoglobin (< 8 g/dL), body mass index (sensitivity of sputum Xpert Ultra was higher than sputum Xpert (73% [62, 81] vs 57% [47, 67]) and specificities were similar. Among outpatients on ART, W4SS sensitivity was 53% (35, 71) and specificity was 71% (51, 85). In this population, a parallel strategy (two or more screening tests offered at the same time) of W4SS with any chest X-ray abnormality had higher sensitivity (89% [70, 97]) and lower specificity (33% [17, 54]; n=2,670) than W4SS alone; at a 5% tuberculosis prevalence, this strategy would require 379 more Xpert tests per 1,000 PLHIV than W4SS but detect 18 more cases. Among outpatients not on ART, W4SS sensitivity was 85% (76, 91) and specificity was 37% (25, 51). CRP (≥10 mg/L) had a similar sensitivity (83% [79, 86]), but higher specificity (67% [60, 73]; n=3,187) and a sequential strategy (second screening test offered only if first screening test is positive) of W4SS then CRP (≥5 mg/L) also had similar sensitivity (84% [75, 90]) but higher specificity (64% [57, 71]; n=3187); at 10% tuberculosis prevalence, these CRP-based strategies would require 272 and 244 fewer Xpert tests per 1,000 PLHIV than W4SS but miss two and one more cases, respectively. ii. We obtained data for all six eligible studies (n=3,660 participants). The pooled proportion of inpatients eligible for Xpert was 90% (89, 91; n=3,658). Among screening tools to guide Xpert testing, W4SS and CRP (≥5 mg/L) had highest sensitivities (≥96%) but low specificities (≤12%); cough (≥2 weeks), haemoglobin (< 8 g/dL), body mass index (<18.5 kg/m²) and lymphadenopathy had higher specificities (61–90%) but low sensitivities (12–57%). The WHO Xpert algorithm had sensitivity of 76% (67, 84) and specificity of 93% (88, 96; n=637). Xpert for all had similar accuracy to the WHO Xpert algorithm: sensitivity was 78% (69, 85) and specificity was 93% (87, 96; n=639). We obtained data from all 5 identified studies (n=3,504). The pooled proportion of inpatients eligible for AlereLAM testing using WHO criteria was 93% (91, 95). Among screening tools to guide LF-LAM testing, WHO criteria, CRP (≥5 mg/L), and CD4 count (< 8 g/dL), body mass index (<18.5 kg/m2) lymphadenopathy, and WHO-defined danger signs had higher specificities but suboptimal sensitivities. AlereLAM for all had the same sensitivity (62% [47, 75]) and specificity (88% [64, 97]) as WHO AlereLAM algorithm. Sensitivities of FujiLAM and AlereLAM were 69% and 48%, while specificities were 88% and 96%, respectively. In 2 studies that 4 collected sputum and non-sputum samples for Xpert and/or culture, diagnostic yield of sputum Xpert was 40–41%, AlereLAM was 39–76%, and urine Xpert was 35– 62%. In one study, FujiLAM diagnosed 80% of tuberculosis cases (vs 39% for AlereLAM), and sputum Xpert combined with AlereLAM, urine Xpert, or FujiLAM diagnosed 61%, 81%, and 92% of all cases, respectively. Conclusion These findings informed the updated 2021 WHO guidelines on tuberculosis screening in PLHIV. Among outpatient PLHIV, the WHO-recommended W4SS has suboptimal diagnostic accuracy and clinical utility. CRP reduces the need for further Xpert confirmatory testing compared with W4SS without compromising sensitivity and has been included in the updated WHO tuberculosis screening guidelines. CRP also shows utility when used in a CPM. However, CRP data were scarce for outpatients on ART, necessitating future research on the accuracy of CRP in this subgroup. Chest X-ray can be useful in outpatients on ART when combined with W4SS. The WHO Xpert algorithm has suboptimal sensitivity; Xpert for all offers slight sensitivity gains and may be considered if resources permit. Among HIV-positive inpatients, WHO screening criteria and other potential screening tools to guide Xpert and AlereLAM testing have suboptimal performance. Based on these findings, WHO now strongly recommends Xpert testing in all medical HIV-positive inpatients in settings where tuberculosis prevalence is higher than 10%. The findings in this thesis also support that AlereLAM testing be implemented in all HIV-positive medical inpatients. 5 Routine FujiLAM testing in all HIV-positive medical inpatients may substantially improve tuberculosis diagnosis, but prospective evaluation of this novel assay is required.

Keywords

medicine

Reference:

Collections

PhD / Doctoral

Full item page