Mortality from tuberculosis (TB) remains high among people living with HIV, accounting for nearly half a million cases and 32% of AIDS-related deaths in 2017.1 Underdiagnosis of TB, especially in places without widespread health care access, is a principal barrier to combatting the disease. In resource-limited settings, microscopy often remains a primary diagnostic method despite its poor sensitivity. Improved molecular diagnoses such as Xpert MTB/RIF have good sensitivity and excellent specificity but are still not widely available. An additional challenge is that these techniques rely on sputum samples that can be difficult to produce, especially for very ill HIV-positive patients. Transferring sputum samples from peripheral health facilities to laboratories for testing can additionally cause delays or losses. As a result, clinicians in low-resource, peripheral facilities often rely on their clinical judgement to diagnose TB.
Consequently, the emergence of easier to use, point-of-care (POC) tests using urine to identify TB [detecting the mycobacterial lipoarabinomannan (LAM) antigen] have been a welcome addition to the TB diagnostics environment.2,3 The lateral-flow TB LAM Ag [(lateral-flow lipoarabinomannan assay (LF-LAM)] assay has shown encouraging sensitivity and specificity (45% and 92%),4 and 2 large trials have shown a substantial reduction in mortality among hospitalized patients who immediately initiated treatment after a positive result.5,6 International guidance currently recommends that LAM may be used to assist TB diagnosis in HIV-positive ambulatory outpatients with signs and symptoms of TB who are severely immunocompromised (CD4 count ≤100 cells/µL) as well as those who are seriously ill.7 Yet, for most outpatients (who are not seriously ill), these recommendations are based on the assumption that a patient's CD4 cell count is readily available, when in fact this is often far from certain. Many resource-limited contexts still struggle with limited and inconsistent access to CD4 testing, an issue which may be compounded as viral load is the preferred technology to monitor ART efficacy in patients with HIV.
Thus, we investigated the diagnostic yield of urine LF-LAM in HIV positive, outpatients with symptoms of TB regardless of whether patients had a CD4 cell count immediately available. We assessed these test results by level of immunosuppression and the risk of mortality at 6 months. We also explored whether clinical signs could be used as a proxy for CD4 count to determine LAM testing eligibility.
Design and Population
This prospective, observational study, conducted between September 2015 and April 2017, consecutively included all ambulatory, ≥15-year-old, TB symptomatic (self-reported cough, weight loss, fever, or night sweats), HIV-positive patients presenting to the Outpatient Department (OPD) of Chiradzulu District Hospital or to any of 3 health centers in the district (Namitambo, Milepa, and Mauwa) in southern Malawi. The 4 facilities were active clinical sites run by the Malawi Ministry of Health with free TB and HIV care supported by Médecins Sans Frontières (MSF). Those who had taken fluoroquinolones or anti-TB drugs the month before their first consultation were excluded. Nonavailability of an immediate CD4 count result or the lack of a urine or sputum sample was not grounds for exclusion from the study. This study is part of a multicountry study conducted in 6 health facilities of Malawi and Mozambique and designed to assess the usefulness and feasibility of the LF-LAM in programmatic conditions. The results on the diagnostic value of including LF-LAM in TB diagnostic algorithms in HIV-positive patients severely immunocompromised have been previously reported.8
Participants' initial evaluation included an examination conducted by a clinical officer and the request of a urine sample, 2 spontaneously expectorated sputum samples (on spot and early morning), and a chest X-ray. Fresh urine was tested for LF-LAM [determine TB-LAM Ag test; Abbott, Waltham, MA (formerly Alere)] on the same day with results interpreted using a 4-grade scale, with grade 1 or above constituting a positive result (according to the manufacturer's instructions). Per the Malawi National TB Program's request at the time of the study implementation, LAM test results did not inform patient management or treatment initiation during the first 10 (of 14) months of the study recruitment period. Thus, tests were conducted at Chiradzulu District Hospital during the first 10 months of the study recruitment period (from September 2015 to June 2016), and at the POC during the latter period (July 2016–October 2016). Smear microscopy on sputum used auramine staining and light-emitting diode fluorescence, and the presence of at least one acid fast bacilli per 100 high-power fields on one slide was considered smear positive. All facilities could perform microscopy on-site except Mauwa. Xpert MTB/RIF technology (Cepheid, Sunnyvale, CA) was only available at Chiradzulu District Hospital, and sputum samples from the peripheral centers were transferred there. Chest X-ray was performed only at Chiradzulu District Hospital and only on selected days each week. In the absence of clinical or radiological findings suggestive of TB, patients were prescribed a broad-spectrum antibiotic (eg, amoxicillin 3 g/d) for 1 week and were reassessed 5 days after the first consultation. The treating clinicians decided whether or not to start TB treatment at any time during the diagnostic process based on their clinical assessment, the biological test results, and the chest X-ray findings, according to the national guidelines for TB diagnosis and treatment. TB was defined as bacteriologically confirmed if Xpert was positive.
Descriptive analyses explored demographic and clinical characteristics recorded during a patient's initial consultation as well as testing and treatment results. Continuous variables were summarized as median and interquartile ranges (IQRs) and compared with Wilcoxon rank-sum testing. Categorical variables were expressed as counts and percentages and compared with χ2 tests. We calculated and reported “time-to-result” (days from initial consult to the clinician receiving the result) and “time-to-treatment” (initial consultation to TB treatment initiation). To explore whether some clinical signs could be used to determine LAM testing eligibility, we compared LAM positivity in patients presenting a specific clinical sign and in those not presenting it. We used χ2 tests to compare the proportions.
Cox-proportional hazards regression was used to assess the association between LAM results by grade (negative, positive grade 1, positive grade 2–4) and mortality in the 6 months after a patient's initial consultation. As no death occurred in LAM-positive patients with CD4 ≥200 cells/µL, these analyses were restricted to patients with CD4 <200 cells/µL. Univariate models included age (≥30 years, <30 years), sex (women, men), antiretroviral treatment (on ART, not on ART), body mass index (BMI ≥ 18, <18 kg/m2), CD4 count (<100, 100–199 cells/µL), health care setting at first consultation (hospital OPD, peripheral health center), whether a patient was treated for TB (yes, no), or whether a patient was seriously ill (defined as temperature >39°C, respiratory rate >30 respirations/minute, cardiac rate >120 beats/minute, or an inability to walk without help). We built the multivariate model including clinically relevant variables and variables with P-value <0.2 in univariate analyses. We used a backward elimination stepwise approach to select the variables included in the final model while maintaining variables that had an effect in the association between LAM and mortality. We checked for confounders and effect modifiers to ensure that the results were robust. Hazard ratios (HRs) were estimated along with their 95% confidence intervals (CI) using an alpha level of 5%.
Data were analyzed using Stata 13 (College Station, TX).
The study was approved by the Médecins Sans Frontières Ethical Review Board and the National Ethical Review Committee in Malawi. Eligible participants provided written informed consent, and, for those aged 15–17 years, a legal guardian's consent was provided in addition to the assent of the participant.
Study Population and TB Diagnosis
A total of 485 ambulatory patients were included, 55.3% (268) in the peripheral health centers and 44.7% (217) in the hospital OPD. Median CD4 count was 341 cells/µL (IQR: 129–546), 171 (35.3%) had a CD4 <200 cells/µL, and 32 (7.2%) were seriously ill. The study cohort's demographic and clinical characteristics are detailed in Table 1.
Overall, 99.0% (480) received a urine-LAM test result, 87.6% (425) a microscopy result, 64.3% (312) an Xpert result, and 14.2% (69) a chest X-ray result. The median time to receive an LAM, Xpert or X-ray result was 0 days (IQR 0–0), 2 days (IQR 1–5), and 6 days (IQR 1–14), respectively. The median time to TB treatment was 4 (IQR 2–7) days. More than half, 57.8% (155) of the patients followed in the peripheral health centers did not have an Xpert result compared with 9.3% (18) of those seen at the hospital OPD. Conversely, patients followed in the health centers were less frequently LAM positive, 11.7% (31) compared with 23.4% (50) of those attending the hospital OPD, probably as a result of higher CD4 counts: 21.3% (57) with CD4 <200 cells/µL in the health centers compared with 52.5% (114) among those in the hospital OPD.
Among patients with a result, 16.9% (81) had a positive LAM, 14.1% (44) a positive Xpert, and 10.6% (45) a positive microscopy. Of the 173 patients without an Xpert result, almost all (172) had a LAM result and 13.3% (23) were LAM positive (ie, potentially missed patients) (Fig. 1). Using the 2 methods together (LAM and Xpert) identified 102 (21.0%) positive participants, a 2.3-fold increase over what identified Xpert alone.
During the study period, 65 (13.4%) participants initiated anti-TB therapy. Yet, if LAM-positive results had been used to initiate patients' treatment for the entire duration of the research period, an additional 51 LAM-positive patients (who were not identified by other tests and were therefore not treated for TB) could have initiated anti-TB therapy, a 1.8-fold increase over what was actually achieved.
LAM Results by Immunosuppression Level
LAM positivity was 24.9% in patients with CD4 <200 cells/µL and 12.5% in patients with CD4 ≥200 cells/µL. More pronounced LAM positivity (grades 2–4) was more frequently found among the more immunocompromised: 50.0% (<200 cells/µL) vs. 12.8% (≥200 cells/µL; P < 0.001) (Table 2). And, patients with higher degrees of positivity (LAM grades 2–4) were also more frequently Xpert positive (58.8%) than those with grade 1 (31.7%), (P = 0.055). However, LAM positivity among Xpert-positive patients was not statistically different across immunosuppression levels: 56.7% (CD4 <200) vs. 42.9% (CD4 ≥200) (P = 0.393).
Clinical Signs Associated With LAM-Positive Results
Although patients with some clinical symptoms at presentation (reported weight loss and ≥38°C measured temperature) were more frequently LAM positive (see Table S1, Supplemental Digital Content, http://links.lww.com/QAI/B387), these clinical signs (separately or together) were not instructive for determining LAM-testing eligibility in the absence of CD4. This is because weight loss was reported by 81.2% of all patients, and a ≥38°C temperature occurred in only 11.1% of those with LAM-positive results. LAM positivity was higher in “seriously ill” patients compared with those not seriously ill, but the difference was not statistically significant.
LAM Positivity and Mortality
Of all patients, 478 (98.6%) had their vital status assessed 6 months after their initial consultation, among whom 44 (9.2%) died. The cause of death of the patients was not ascertained. LAM-positive patients had higher mortality than LAM-negative participants (17.5% vs. 7.4%; P = 0.004), as did those with more pronounced LAM-positive results (grades 2–4: 36.0%; grade 1: 9.1%; negative: 7.4%; P < 0.001). However, no death occurred in LAM-positive patients with CD4 ≥200 cells/µL. In patients with CD4 <200 cells/µL, mortality was 18.6% (31/136), and the adjusted model showed that LAM-positive patients had a higher risk of mortality than LAM negatives (adjusted HR: 3.2, 95% CI: 1.4 to 7.2, P = 0.006), particularly those with LAM grades 2–4 patients (adjusted HR: 4.9, 95% CI: 1.8 to 13.3, P = 0.002; Table 3 and Fig. 2). In addition, during the period when LAM was not used for patients' management, the mortality among untreated LAM-positive patients with CD4 <200 cells/µL (ie, missed TB patients) was 41.2% (7/17) compared with 25.0% (4/16) among those treated, although this difference was not statistically significant, P = 0.325. Mortality was the greatest among untreated LAM-positive grades 2–4 patients: 66.7% (4/6).
This was an investigation of TB symptomatic, outpatients, HIV positive, tested with LAM regardless of whether they had a CD4 cell count available. Although LAM and Xpert positivity were predictably higher among the most immunocompromised, a considerable proportion of those less immunocompromised was also LAM positive, indicating that excluding this group from routine LAM testing may be missing opportunities to identify TB. We show that integrating urine-LAM into standard care could have doubled the number of TB-positive patients identified in the cohort. In addition, LAM grade can identify patients at higher risk of death. In this cohort, clinical signs did not serve as a proxy for CD4 to determinate eligibility for LAM-testing.
Previous research has found LAM positivity in 9%–35% of outpatients regardless of CD4 levels.9–13 In primary care settings without same-day X-ray facilities, POC-LAM has been shown to significantly increase same-day treatment initiation.10 Our findings add to this chorus of research suggesting that current international recommendations may not go far enough, and that the criteria recommending LAM use in only the severely immunocompromised (CD4 < 100) may need to be expanded (WHO guidelines on the use of LAM are currently under review). The recommendation is largely based on previous studies showing low-test sensitivity in patients with higher CD4 counts.4 In our study, LAM positivity among Xpert-positive patients with CD4 > 200 cells/µL was relatively high, around 43%. However, we should also consider that despite the high specificity of LAM in patients less immunocompromised reported in other studies,4 the lower TB prevalence in this group may lead to low positive-predictive values, and some LAM-positive patients may not have TB. Using a greater LAM grade for a positive test, threshold could improve further the specificity but would also reduce the diagnostic sensitivity.14 Yet, a test's value should not only depend on whether it has the highest accuracy, but also on whether other diagnostic tools are available in a facility and whether a patient actually receives the test result. The third of patients in our cohort (and more than half of those attending the peripheral health centers) did not have an Xpert result, largely because of difficulties producing a sputum sample, issues related to the availability of staff to collect samples and challenging sample transport (common in resource-limited settings). These patients could have avoided having their TB diagnosis based entirely on clinical judgment had TB-LAM been used for clinical decision-making. Systematic Xpert testing remains important for diagnosing TB and for testing rifampicin resistance. However, in settings where Xpert results are not available or are frequently delayed, POC-LAM in all TB-symptomatic patients regardless of CD4 count can be a useful parallel testing option to Xpert.
Although some clinical signs (higher WHO clinical stage, low BMI, tachycardia, and lower blood pressure) have previously been reported to be associated with higher LAM yields,11,12,15,16 in this study, “seriously ill” patients and those with low BMI did not have higher LAM positivity. Moreover, despite the fact that patients with reported weight loss and >38°C temperature were more frequently LAM positive, weight loss was so common in the cohort that this finding would be less meaningful, and testing only patients with >38°C temperature would miss a high proportion of those who could benefit from LAM.
Our findings also indicate that it is important to take the degree of LAM positivity (grade) into consideration when managing patients, as patients with stronger positivity (higher LAM grades) had substantially higher mortality than those with lower LAM-grade or LAM-negative, perhaps explained by a higher bacillary load. Others studies in South Africa have also found a higher risk of mortality associated with higher LAM grades in patients newly diagnosed with HIV.11,17,18 Although no death occurred in LAM-positive patients less immunocompromised (CD4 > 200), LAM grade may be important when CD4 count is unknown because patients with higher grade were also more immunocompromised, making them a population where LAM grade would be useful to identify their increased risk of death. In addition, in patients with CD4 <200 cells/µL, higher grades were also associated with a higher risk of mortality and LAM testing may potentially decrease mortality if patients are treated rapidly.
This study has some limitations: bacteriologically confirmed TB was defined using only Xpert in sputum because of the limitations of the study context, creating an imperfect reference standard for LAM since other specimens (blood, urine, and tissue) increase TB detection.19 Yet, the real-world conditions (despite MSF support) in which the study was conducted should also be considered a strength, producing results that reflect on-the-ground realities and the value of LAM.
In Malawi and similar contexts where TB diagnostics tools are limited, urine LAM can be useful to diagnose TB regardless of whether CD4 counts are available. Incorporating LAM into the standard care could diagnose more patients with signs and symptoms of TB. In addition, LAM grade is useful to identify patients at higher risk of death when CD4 count is unknown. Our findings do not support using specific clinical signs as a proxy for CD4 when prioritizing urine-LAM eligibility.
The authors thank all patients, care providers, and data managers at the study sites. The authors are grateful to the Médecins Sans Frontières staff team and the Ministry of Health and National TB Control Program in Malawi for their support of this study. The authors thank Janet Ousley for editing the manuscript.
1. World Health Organization. Global Tuberculosis Report 2018. Geneva, Switzerland: World Health Organization; 2018.
2. Peter JG, Cashmore TJ, Meldau R, et al. Diagnostic accuracy of induced sputum LAM ELISA for tuberculosis diagnosis in sputum-scarce patients. Int J Tuberc Lung Dis. 2012;16:1108–1112.
3. Drain PK, Gounder L, Sahid F, et al. Rapid urine LAM testing improves diagnosis of expectorated smear-negative pulmonary tuberculosis in an HIV-endemic region. Sci Rep. 2016;6:19992.
4. Shah M, Hanrahan C, Wang ZY, et al. Lateral flow urine lipoarabinomannan assay for detecting active tuberculosis in HIV-positive adults. Cochrane Database Syst Rev. 2016:CD011420.
5. Peter JG, Zijenah LS, Chanda D, et al. Effect on mortality of point-of-care, urine-based lipoarabinomannan testing to guide tuberculosis treatment initiation in HIV-positive hospital inpatients: a pragmatic, parallel-group, multicountry, open-label, randomised controlled trial. Lancet. 2016;387:1187–1197.
6. Gupta-Wright A, Corbett EL, van Oosterhout JJ, et al. Rapid urine-based screening for tuberculosis in HIV-positive patients admitted to hospital in Africa (STAMP): a pragmatic, multicentre, parallel-group, double-blind, randomised controlled trial. Lancet. 2018;392:292–301.
7. World Health Organisation. The Use of Lateral Flow Urine Lipoarabinomannan Assay (LF-LAM) for the Diagnosis and Screening of Active Tuberculosis in People Living With HIV. Policy Guidance. Geneva, Switzerland; 2015:74. Available at: http://apps.who.int/iris/bitstream/10665/193633/1/9789241509633_eng.pdf?ua=1&ua=1
8. Huerga H, Mathabire Rucker SC, Cossa L, et al. Diagnostic value of the urine lipoarabinomannan assay in HIV-positive, ambulatory patients with CD4 below 200 cells/μl in 2 low-resource settings: a prospective observational study. PLoS Med. 2019;16:e1002792.
9. Bjerrum S, Kenu E, Lartey M, et al. Diagnostic accuracy of the rapid urine lipoarabinomannan test for pulmonary tuberculosis among HIV-infected adults in Ghana-findings from the DETECT HIV-TB study. BMC Infect Dis. 2015;15:407.
10. Peter J, Theron G, Chanda D, et al. Test characteristics and potential impact of the urine LAM lateral flow assay in HIV-infected outpatients under investigation for TB and able to self-expectorate sputum for diagnostic testing. BMC Infect Dis. 2015;15:262.
11. Drain PK, Losina E, Coleman SM, et al. Clinic-based urinary lipoarabinomannan as a biomarker of clinical disease severity and mortality among antiretroviral therapy-naive human immunodeficiency virus-infected adults in South Africa. Open Forum Infect Dis. 2017;4:ofx167.
12. Thit SS, Aung NM, Htet ZW, et al. The clinical utility of the urine-based lateral flow lipoarabinomannan assay in HIV-infected adults in Myanmar: an observational study. BMC Med. 2017;15:145.
13. Nakiyingi L, Moodley VM, Manabe YC, et al. Diagnostic accuracy of a rapid urine lipoarabinomannan test for tuberculosis in HIV-infected adults. J Acquir Immune Defic Syndr. 2014;66:270–279.
14. Drain PK, Losina E, Coleman SM, et al. Value of urine lipoarabinomannan grade and second test for optimizing clinic-based screening for HIV-associated pulmonary tuberculosis. J Acquir Immune Defic Syndr. 2015;68:274–280.
15. Lawn SD, Kerkhoff AD, Burton R, et al. Diagnostic accuracy, incremental yield and prognostic value of Determine TB-LAM for routine diagnostic testing for tuberculosis in HIV-infected patients requiring acute hospital admission in South Africa: a prospective cohort. BMC Med. 2017;15:67.
16. Balcha TT, Winqvist N, Sturegård E, et al. Detection of lipoarabinomannan in urine for identification of active tuberculosis among HIV-positive adults in Ethiopian health centres. Trop Med Int Heal. 2014;19:734–742.
17. Kubiak RW, Herbeck JT, Coleman SM, et al. Urinary LAM grade, culture positivity, and mortality among HIV-infected South African out-patients. Int J Tuberc Lung Dis. 2018;22:1366–1373.
18. Kerkhoff AD, Wood R, Vogt M, et al. Prognostic value of a quantitative analysis of lipoarabinomannan in urine from patients with HIV-associated tuberculosis. PLoS One. 2014;9:e103285.
19. Lawn SD, Kerkhoff AD, Nicol MP, et al. Underestimation of the true specificity of the urine lipoarabinomannan point-of-care diagnostic assay for HIV-associated tuberculosis. J Acquir Immune Defic Syndr. 2015;69:e144–e146.