JAIDS Journal of Acquired Immune Deficiency Syndromes:
Diagnostic Accuracy of a Rapid Urine Lipoarabinomannan Test for Tuberculosis in HIV-Infected Adults
Nakiyingi, Lydia MBChB*; Moodley, Vineshree Mischka MBChB†; Manabe, Yukari C. MD*,‡,§; Nicol, Mark P. PhD†; Holshouser, Molly MPH‡; Armstrong, Derek T. MHS‡; Zemanay, Widaad PhD†; Sikhondze, Welile MBChB†; Mbabazi, Olive BA*; Nonyane, Bareng A.S. PhD§; Shah, Maunank MD‡; Joloba, Moses L. PhD‖; Alland, David MD¶; Ellner, Jerrold J. MD#; Dorman, Susan E. MD‡,§
*Department of Medicine, Infectious Diseases Institute, College of Health Sciences, Makerere University, Kampala, Uganda;
†Division of Medical Microbiology, University of Cape Town, Cape Town, South Africa and National Health Laboratory Service, South Africa;
‡Johns Hopkins University School of Medicine, Baltimore, MD;
§Department of Medicine, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD;
‖Department of Medical Microbiology, College of Health Sciences, Makerere University, Kampala, Uganda;
¶Division of Infectious Diseases, Department of Medicine, New Jersey Medical School, Newark, NJ; and
#Department of Medicine, Boston Medical Center, Boston University, Boston, MA.
Correspondence to: Susan E. Dorman, MD, Center for Tuberculosis Research, Department of Medicine, Johns Hopkins University, CRB2 Room 1M-12, 1550 Orleans Street, Baltimore, MD 21231 (e-mail: email@example.com).
Supported by federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services, under contract No. HHSN2722000900050C, “TB Clinical Diagnostics Research Consortium.” Additional support was provided by NIH K23AI089259 to M.S.
ClinicalTrials.gov: Nos. NCT01525134 and NCT01693224.
Presented in part at the 19th Conference on Retroviruses and Opportunistic Infections, March 5–8, 2012, Seattle, WA.
The authors have no conflicts of interest to disclose.
L.N. and V.M.M. contributed equally to this work. L.N. and V.M.M.: data collection, data analysis, data interpretation, and report writing; Y.C.M., M.P.N., D.A., J.J.E., and S.E.D.: study design, data analysis, data interpretation, and report writing; M.H. and D.T.A.: study design, data collection, and data interpretation; W.Z., W.S., and O.M.: data collection, data analysis, and data interpretation; B.A.S.N. and M.S.: data analysis, data interpretation, and report writing; M.L.J.: data collection, data interpretation, and report writing.
Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (www.jaids.com).
Received November 22, 2013
Accepted January 31, 2014
Objective: In settings of high HIV prevalence, tuberculosis control and patient management are hindered by lack of accurate, rapid tuberculosis diagnostic tests that can be performed at point-of-care. The Determine TB LAM Ag (TB LAM) test is a lateral flow immunochromatographic test for detection of mycobacterial lipoarabinomannan (LAM) in urine. Our objective was to determine sensitivity and specificity of the TB LAM test for tuberculosis diagnosis.
Design: Prospective diagnostic accuracy study.
Setting: Hospital and outpatient settings in Uganda and South Africa.
Participants: HIV-infected adults with tuberculosis symptoms and/or signs.
Methods: Participants provided a fresh urine specimen for TB LAM testing, blood for mycobacterial culture, and 2 respiratory specimens for smear microscopy and mycobacterial culture.
Main Outcome Measures: For the TB LAM test, sensitivity in participants with culture-positive tuberculosis and specificity in participants without tuberculosis.
Results: A total of 1013 participants were enrolled. Among culture-positive tuberculosis patients, the TB LAM test identified 136/367 (37.1%) overall and 116/196 (59.2%) in the group with CD4 ≤100 cells per cubic millimeter. The test was specific in 559/573 (97.6%) patients without tuberculosis. Sensitivity of the urine TB LAM test plus sputum smear microscopy was 197/367 (53.7%) overall and 133/196 (67.9%) among those with CD4 ≤100. CD4 ≤50 [adjusted odds ratio (AOR), 6.2; P < 0.001] or 51–100 (AOR, 7.1; P < 0.001), mycobacteremia (AOR, 6.1; P < 0.01) and hospitalization (AOR, 2.6; P = 0.03) were independently associated with a positive TB LAM test.
Conclusions: In HIV-positive adults with CD4 ≤100, the TB LAM urine test detected over half of culture-positive tuberculosis patients, in <30 minutes and without the need for equipment or reagents.
The HIV pandemic has driven resurgence of tuberculosis, a leading cause of death in HIV-infected persons in sub-Saharan Africa.1,2 In individuals coinfected with HIV and Mycobacterium tuberculosis, the 2 pathogens are tragically synergistic. Among the clinical consequences are high mortality and “atypical” tuberculosis disease presentations such as disseminated tuberculosis and non-cavitary pulmonary tuberculosis, especially with advancing immunosuppression.3,4 These alterations in tuberculosis clinical manifestations impair the sensitivity and specificity of chest radiography, and the sensitivity of sputum smear microscopy, which are 2 of the most widely available diagnostic tools.5 Clinical sensitivities of sputum mycobacterial culture and nucleic acid amplification tests are less affected by HIV, but these tests are not widely available at point-of-care, and culture takes weeks for a result.6 Xpert MTB/RIF, a closed cartridge nucleic acid amplification TB test, is an advance with respect to ease of operator use but has limitations including requirements for electricity and climate/dust control, and a minimum testing time of about 2 hours.7 Furthermore, Xpert MTB/RIF is optimized for testing of sputum, a specimen type that is sometimes not obtainable or informative and is biohazardous. In HIV-infected individuals with advanced immunosuppression and tuberculosis, the clear survival benefit from prompt tuberculosis treatment plus early antiretroviral therapy (ART) makes imperative the timely, accurate diagnosis of both diseases.8–10
Lipoarabinomannan (LAM), a glycolipid component of the M. tuberculosis cell wall, is an attractive diagnostic target.11 As a bacterial product, LAM has the theoretical potential to discriminate active tuberculosis disease from latent tuberculosis infection independent of human immune responses. Diagnostic accuracy studies using a plate-based enzyme-linked immunosorbence assay for detection of LAM have shown sensitivities that are higher in HIV-infected versus HIV-uninfected tuberculosis patients, and that are inversely correlated with degree of HIV-associated immunosuppression.12–17 The Determine TB LAM Ag test (“TB LAM” test; Alere, Waltham, MA) is a lateral flow immunochromatographic assay for detection of urinary LAM. This point-of-care test requires a drop of unprocessed urine and no other equipment or supplies. Results are read by visual inspection 25 minutes after applying the urine to the test strip.
We reasoned that a simple, rapid, truly point-of-care test having increasing tuberculosis diagnostic sensitivity with increasing HIV-associated immunosuppression might be useful in diagnosing the subset of patients with the highest mortality from untreated tuberculosis and the greatest potential survival benefit from prompt treatment of both diseases. We performed a multicenter study to assess the accuracy of the TB LAM test, performed in real-time on fresh urine, for diagnosis of tuberculosis in HIV-infected adults.
Design and Settings
This diagnostic accuracy study was cross-sectional with limited longitudinal follow-up. Outpatient recruitment settings were the Infectious Diseases Institute clinic in Kampala, Uganda and Town Two clinic, Khayelitsha, near Cape Town, South Africa. Inpatient settings were Mulago Hospital in Kampala and G.F. Jooste Hospital in Cape Town. The enrollment period was January 2011 through November 2011. Data collection was planned before the investigational tests and reference standards were performed, and all study procedures were performed according to the study protocol and a written procedures manual.
The target population was HIV-positive adults suspected of having tuberculosis. Inclusion criteria were age ≥18 years; suspected to have active tuberculosis based on having at least one of cough, fever, night sweats, or weight loss; HIV positive; and informed consent. Exclusion criteria were administration of more than 2 days of anti-tuberculosis treatment within 60 days before enrollment and inability to provide a urine specimen.
Clinical and Laboratory Assessments
At enrollment, participants were interviewed for medical and demographic information, had a chest x-ray, and then provided a spontaneously voided urine specimen for LAM testing, blood for mycobacterial culture and CD4 T-cell count, and 2 respiratory specimens for smear microscopy and mycobacterial culture. If a participant could not spontaneously expectorate sputum, then sputum induction was performed using nebulized hypertonic saline. Participants who had a positive urine TB LAM test but no enrollment culture positive for M. tuberculosis underwent the same panel of tests and interview 2 months later, and medical records from the time of enrollment to the time of follow-up were reviewed. For all other participants, medical records from enrollment to 2 months postenrollment were reviewed or an interview was conducted.
All laboratory tests were performed on site in Uganda or South Africa in designated laboratories with existing external quality assurance programs, and testing was initiated immediately after specimen collection (ie, specimens were not stored before testing). Unprocessed sputum was smeared and stained using the Ziehl–Neelsen (ZN) method; smears were examined microscopically for acid-fast bacilli and graded.18 Sputa were decontaminated using N-acetyl-L-cysteine-NaOH19; a 0.5-mL portion of the sputum sediment was cultured using the BACTEC MGIT 960 system (Becton and Dickinson, Franklin Lakes, NJ), and 0.2 mL was inoculated onto Lowenstein–Jensen media. Cultures were incubated at 37° C for up to 8 weeks. Mycobacterial blood cultures were performed using the MYCO/F LYTIC (Becton and Dickinson) or BacT/ALERT MB (BioMerieux, Marcy-l'Etoile, France) systems, which previously have been shown to have similar diagnostic yields.20,21 Positive cultures were assessed for acid-fast bacilli using ZN staining and light microscopy and for M. tuberculosis complex using an anti-MPB64 antibody assay (Capilia TB-Neo; TAUNS Laboratories, Numazu, Japan) or the GenoType MTBDR plus assay (Hain Lifesciences GmbH, Nehren, Germany).
TB LAM tests were performed and graded by trained study staff according to manufacturer recommendations.22 Staff interpreting the TB LAM assays were blinded to clinical data and results of other diagnostic testing. Sixty microliters of fresh urine was applied to the test strip which was incubated at room temperature for 25 minutes. The strip was visually inspected, and the intensity of any visualized test band was graded by comparing test band color intensity to the color intensity of a series of bands on a manufacturer-supplied paper reference card. Test band intensity was graded as zero if no band was visualized, and grade 1 through 5 for visualized bands. A blinded second reader immediately conducted an independent reading. The initial reading was considered as the study result for that test, except for analyses of inter-reader agreement for which results of the first and second readers were considered. TB LAM results were neither provided to clinicians caring for enrolled participants nor used for clinical care decision making.
Final Diagnostic Classification of Participants
Participants were categorized into 1 of the 3 following diagnostic categories based on microbiological and clinical criteria.
M. tuberculosis cultured from any specimen.
No culture positive for M. tuberculosis, but at least one of the following: (1) sputum smear microscopy positive but no sputum culture positive for any mycobacteria, (2) started on anti-tuberculosis treatment with subsequent documented clinical improvement within 2 months of enrollment, (3) a diagnosis of active tuberculosis within 2 months of enrollment by a nonstudy clinician, and (4) death within 2 months of enrollment reported to be due to tuberculosis per a death certificate, autopsy report, or medical record.
No Evidence of Tuberculosis
Does not meet criteria for either “Culture-positive tuberculosis” or “Possible tuberculosis.”
We calculated that 100 culture-positive tuberculosis cases/site would provide, for each of the 2 study sites, 95% confidence intervals (CIs) for investigational test sensitivity of about ±10% over a range of investigational test sensitivities of 30%–70%. Assuming a 20% prevalence of culture-positive tuberculosis among participants, the enrollment target was 500 participants/site (1000 in total).
Descriptive statistics were used to characterize the study population. Results for 2 TB LAM readers were compared and proportionate agreement and κ statistics were calculated. Nonparametric receiver operator characteristic (ROC) analysis was performed to evaluate sensitivity and specificity based on different band intensity thresholds for TB LAM test positivity. To construct ROC curves, results of participants classified as “Culture-positive tuberculosis” or “No evidence of tuberculosis” were used; results of participants classified as “probable TB” were excluded. Sensitivities, specificities, and predictive values were calculated, with 95% CIs generated using the exact binomial distribution for the proportions. Unless stated otherwise, results shown for TB LAM testing are those using a test band positivity threshold of grade 2 (ie, any test band of intensity grade 2 or higher was considered positive, and any band of lesser intensity was considered negative). Sensitivities were calculated separately for participants with “Culture-positive tuberculosis” and “Possible tuberculosis.” Specificities were calculated for participants classified as “No evidence of tuberculosis.” A multivariable logistic regression model was used to investigate predictors of TB LAM test positivity. All statistical tests were 2 sided. Statistical calculations were performed using Stata 12 software (StataCorp, 2011, College Station, TX).
This study was approved by ethics committees of Johns Hopkins University School of Medicine, University of Cape Town, Joint Clinical Research Center (Kampala, Uganda), and Uganda National Council for Science and Technology. Participants provided written informed consent.
Role of the Funding Source
This study was designed and implemented by investigators of the TB Clinical Diagnostics Research Consortium (TB-CDRC). Neither the sponsor nor Alere was involved in study design, implementation, analysis, or decision to submit results for publication.
A total of 1037 individuals were screened, and 1013 (97.7%) were enrolled (Fig. 1). Sputum could not be obtained from 16/1013 (1.6%) participants, who were excluded from the analysis. Characteristics of 997 analyzed participants are presented in Table 1. Median CD4 (cells/mm3) was 152 [interquartile range (IQR), 41–337] and was lower in participants enrolled in Uganda (97; IQR, 21–290) versus South Africa (217; IQR, 87–389; P < 0.001 by nonparametric Wilcoxon rank-sum test).
Of note, 630/997 (63.2%) participants had no specimen (neither respiratory nor blood) that was culture positive for M. tuberculosis, and 367/997 (36.8%) had at least one specimen that was culture positive for M. tuberculosis. Among those 367 participants, in 243 (66.2%) only respiratory specimen(s) were culture positive, and in 108 (29.4%) respiratory and blood cultures were positive for M. tuberculosis [resulting in 351/997 (35.2%) participants with pulmonary TB based on respiratory cultures]; in 16 (4.4%) only the blood culture was positive. At least one respiratory specimen that was positive by smear microscopy was observed in 128/367 (34.9%) culture-positive participants. As shown in Figure 1, 367/997 (36.8%) participants were classified as “Culture-positive tuberculosis”; 57/997 (5.7%) as “Possible tuberculosis”; and 573/997 (57.5%) as “no evidence of tuberculosis” (Fig. 1).
TB LAM Test Operating Characteristics and Distribution of Results
A valid TB LAM result was obtained on the first attempt for all tests (997/997, 100.0%). The distribution of band intensity results was no band (grade 0), 657/997 (65.9%); grade 1, 183/997 (18.4%); grade 2, 42/997 (4.2%); grade 3, 38/997 (3.8%); grade 4, 55/997 (5.5%); grade 5 or greater, 22/997 (2.2%). Between 2 independent readers, agreement as to presence versus absence of a test band of intensity grade 2 or higher was 96.1% (kappa 0.92), agreement as to presence versus absence of a test band of any intensity was 95.7% (kappa 0.90), and agreement as to test band grade was 83.4% (kappa 0.77).
TB LAM Sensitivity Among Participants With Culture-Positive Tuberculosis, and Specificity Among Participants With No Evidence of Tuberculosis
ROC curve analysis showed that a test band intensity positivity threshold of grade 2 maximized sensitivity and specificity (Figure 2A) (see Table S1 and Figure S1, Supplemental Digital Content 1 and 2, http://links.lww.com/QAI/A512). Using a positivity threshold of grade 2, overall TB LAM sensitivity was 136/367 (37.1%) (Table 2). Sensitivity was higher in Uganda than in South Africa [83/182 (45.6%) versus 53/185 (28.7%); P = 0.0008]. Overall specificity was 559/573 (97.6%). Specificities were 268/282 (95.0%) and 291/291 (100%) in Uganda and South Africa, respectively (P = 0.0001). In a secondary analysis, we explored the potential impact of incomplete gold-standard mycobacteriologic assessment on TB LAM specificity by requiring that, to be classified as “no evidence of tuberculosis,” a participant have at least one evaluable (noncontaminated) respiratory culture and a noncontaminated blood culture in addition to not meeting the criteria for “Culture-positive tuberculosis” or “Possible tuberculosis.” In this secondary analysis, specificity was 530/541 (98.0%; 95% CI: 96.4 to 99.0). TB LAM accuracy results stratified by hospitalization status at the time of enrollment are shown in Table S2 (see Supplemental Digital Content 3, http://links.lww.com/QAI/A512).
TB LAM test sensitivity and specificity using a positivity threshold of grade 2, stratified by CD4 T-cell count, are shown in Table 2 and Figure 2B. Sensitivity was 116/196 (59.2%) among culture-positive participants with CD4 ≤100, and 20/169 (11.8%) among those with CD4 >100 (P < 0.0001). Sensitivity remained slightly higher in Uganda than in South Africa among participants with CD4 ≤100 [71/109 (65.1%) versus 45/87 (51.7%); P = 0.06], but sensitivities were similar between the 2 sites when only hospitalized participants with CD4 ≤100 were considered [Uganda 61/92 (66.3%) versus South Africa 38/61 (62.3%); P = 0.60]. Specificity was slightly lower in the group with CD4 ≤100 than in the group with CD4 >100 [156/165 (94.5%) versus 399/403 (99.0%); P = 0.001]. Among 57 participants with “possible tuberculosis,” the TB LAM test was positive in 7 (sensitivity 12.3%; 95% CI: 5.1 to 23.8). For reference, Table S3 shows TB LAM accuracy results if any visualized band was considered a positive test (see Supplemental Digital Content 4, http://links.lww.com/QAI/A512).
Factors Associated With a Positive TB LAM Test (Positivity Threshold Grade 2)
Among participants with “Culture-positive tuberculosis,” the adjusted odds ratio for TB LAM positivity among participants with CD4 ≤50 and CD4 51–100 were 6.2 (P < 0.001) and 7.1 (P < 0.001), respectively, versus CD4 >200 (Table 3); the adjusted odds ratio for mycobacteremia versus no mycobacteremia was 6.1 (P < 0.001), and for being hospitalized versus not hospitalized at the time of enrollment was 2.6 (P = 0.03). Among 124 participants with M. tuberculosis isolated from blood cultures, the TB LAM test was positive in 91 (73.4%). Within 2 months after enrollment, 128/997 (12.8%) participants died, including 42/157 (26.8%) of participants who were TB LAM positive and 86/840 (10.2%) of participants who were TB LAM negative (OR 3.2, 95% CI: 2.1 to 4.9).
TB LAM Test Sensitivity (Positivity Threshold Grade 2) Compared With That of Respiratory Specimen Direct Smear Microscopy
TB LAM test sensitivity was compared with that of respiratory specimen smear microscopy performed on 2 sputa at baseline. Among participants with culture-positive tuberculosis, 67/367 (18.3%) were positive by both TB LAM test and ZN smear microscopy, 170/367 (46.3%) were negative by both tests, 69/367 (18.8%) were positive by the TB LAM test only, and 61/367 (16.6%) were positive by ZN smear microscopy only. TB LAM sensitivity was similar to that of smear microscopy [136/367 (37.1%) versus 128/367 (34.9%); P = 0.53]. The combined sensitivity of TB LAM plus smear microscopy [either or both tests positive, 197/367 (53.7%)] was higher than the sensitivity of either test alone (see Table S4, Supplemental Digital Content 5, http://links.lww.com/QAI/A512). In participants with CD4 ≤100, the combined sensitivity of TB LAM plus smear microscopy was 133/196 (67.9%).
Our multicenter study shows that the TB LAM test detected well over half of tuberculosis patients with advanced HIV-related immunosuppression—the group in which tuberculosis mortality is highest and the benefits of early ART initiation are most marked. Test specificity exceeded the target of 95% proposed by an international expert committee for a point-of-care tuberculosis test.23 For the TB LAM test, agreement between 2 trained readers was high, and all tests yielded a valid result on the first attempt. TB LAM sensitivity correlated with advancing immunosuppression, and a positive TB LAM test was associated with mycobacteremia and hospitalization. These features suggest that the target population for testing is clinically very ill adults who have advanced HIV disease and CD4 ≤100. The availability of HIV rapid testing and expanding availability of point-of-care CD4 testing will facilitate identification of this target population, and the potential programmatic impact of TB LAM would be expected to be greatest in settings of high TB/HIV burden and low coverage and/or uptake of ART. The combination of TB LAM and direct smear microscopy had higher sensitivity than did either test used alone because these tests—one performed on urine and the other on sputum—detected tuberculosis in largely nonoverlapping subsets of patients.
TB LAM test sensitivity was higher at the Uganda site than at the South African site. This likely reflects differences between study sites with respect to tuberculosis and HIV disease severity; in support of this are our findings of similar sensitivities between sites among hospitalized participants with CD4 ≤100, and that enrollment site was not associated with TB LAM positivity when adjusted for other factors. However, we cannot fully exclude a role for potential differences between sites in M. tuberculosis bacterial characteristics, test interpretation after visualization, or test batch-to-batch variability. Test specificity was lower at the Uganda site than at the South African site. The specificity difference was most pronounced in the group with CD4 ≤50 and could be related to failure to make a “gold standard” culture diagnosis in some participants with tuberculosis and advanced immunosuppression.
Whether it is possible to improve the clinical sensitivity of urine LAM assays is an unresolved issue, and there are technical and biological aspects for consideration. For the TB LAM test, the lower limit of LAM detection is reported to be 0.25 ng/mL (Alere, personal communication, 2012) and is limited by the polyclonal antibodies and detection system used. Optimization of antibodies and use of detection systems with better analytical sensitivity might favorably decrease the lower limit of LAM detection in clinical specimens, and point to the need for continued research in these areas.24,25 From a biological standpoint, the pathophysiological processes that result in LAM antigenuria are unclear but have implications with respect to whether assays that can detect exquisitely low antigen concentrations in urine will be useful in nonimmunocompromised tuberculosis patients.26 For example, if LAM antigenuria is the consequence of glomerular filtration of circulating LAM, then assays with high analytical sensitivity may be useful in detecting physically “distant” pulmonary disease with relatively low total pathogen burden. However, if LAM antigenuria reflects clinically unrecognized renal tract tuberculosis, then application of more sensitive LAM assays to patients with only pulmonary tuberculosis will not be useful.27 In our study, urine cultures and/or urine Xpert MTB/RIF testing might have helped to resolve this issue but were not performed. Available evidence shows that a high proportion of urine specimens that are TB LAM positive is also positive by Xpert MTB/RIF.28–30 Because detection of M. tuberculosis by Xpert MTB/RIF requires intact bacilli, the available evidence, while not conclusive, supports the hypothesis that LAM antigenuria reflects renal tract TB.
We are aware of 3 other published studies using the TB LAM lateral flow test, all of which used urine that had been previously collected and frozen. Lawn et al31 tested urine specimens from 542 ambulatory HIV-infected adults being screened for tuberculosis using a rigorous protocol-specified battery of tests before initiation of ART in South Africa. Using a grade 1 positivity threshold, test sensitivity was 28.2% overall, but was 66.7% in the group with CD4 <50; specificity was 98.6% overall. Peter et al32,33 have reported 2 diagnostic accuracy assessments using frozen urine from hospitalized HIV-positive adults in South Africa whose tuberculosis diagnostic work-up had been directed by routine care clinicians. They found, as we did in our study, that using a grade 2 positivity threshold optimized TB LAM sensitivity and specificity. The first study included 242 participants, and among those with culture-positive tuberculosis test sensitivity (grade 2 cutoff) was 50% overall and 58% in the group with CD4 ≤200; specificity was 75% among participants whose cultures were negative.32 Specificity was higher (94%) when culture-negative participants who met clinical criteria for tuberculosis were omitted from the specificity calculation; it is worth pointing out that mycobacterial blood cultures typically were not performed and some culture-negative participants may have had tuberculosis.32 The second study reported by Peter et al33 included 281 participants (median CD4, 89 cells/mm3), and TB LAM sensitivity and specificity (grade 2 cutoff) were 46% and 96%, respectively. An unresolved issue is that of the optimal band positivity threshold, and our findings differ from those of Lawn et al who reported high specificity using a grade 1 positivity threshold. Possible explanations are batch-to-batch variation in TB LAM test strips, differences in visual interpretation of faint bands, use of frozen versus fresh urine, and/or bacterial contamination of urine samples causing weak false-positive bands.14
Our study has important strengths, including adherence to the recommendations of the Standards for Reporting of Diagnostic Accuracy Studies (STARD) Initiative and the QUADAS tool for quality assessment of diagnostic accuracy studies.34,35 To accurately classify participants, we cultured blood and sputum for mycobacteria to provide a rigorous gold standard including ability to detect disseminated nonpulmonary disease. In contrast to existing published studies, we used fresh rather than frozen urine.31–33 Performance of our study in 2 African settings and in hospitalized as well as ambulatory patients helps to clarify the assay's generalizability and to underscore that its primary role is likely to be in hospitalized patients and/or those with advanced immunosuppression. For clinical, laboratory, and data management procedures, we implemented rigorous internal quality management, performed the study according to Good Clinical Practice standards, and received periodic external quality monitoring. Although these practices maximized the validity of our study results, some may be challenging to implement in routine clinical practice.
A study limitation was that repeat mycobacteriology assessments were performed at 2 months after enrollment only for a subset of participants, namely those who had a positive LAM test but no culture positive for M. tuberculosis at baseline. This approach could cause bias in favor of the TB LAM assay. However, only 8 participants (2.2% of participants with microbiologically confirmed TB) with this pattern of baseline results had a follow-up culture positive for M. tuberculosis. Another limitation is that our testing algorithm, despite including sputum and blood cultures, may have failed to detect some participants with tuberculosis (especially solely extrapulmonary, nondisseminated tuberculosis) and therefore overestimated the true sensitivity or underestimated the true specificity of the TB LAM test. Our study did not incorporate sputum Xpert MTB/RIF testing, which may have additive yield when combined with urine TB LAM testing and which therefore is a promising companion test for optimizing the diagnostic yield of a “same-day” testing algorithm. We did not assess implementation feasibility or impact on clinical outcomes in the context of routine clinical care—important issues that warrant careful future assessment in light of the accuracy findings from our study and those of others and as a next step toward understanding whether and how the TB LAM test should be implemented.31–33 Cost will be an important consideration. A cost-utility analysis restricted to hospitalized, severely immunocompromised, HIV-infected African adults showed that the addition of lateral flow urine LAM testing to standard tuberculosis diagnostics detected additional TB cases at an incremental cost in US dollars per disability-adjusted life year averted of USD $86 in Uganda and USD $353 in South Africa and was therefore likely to be cost effective in those populations.36 Finally, although we purposefully sought diversity within our study population, our findings may not apply to all other populations, especially those with less advanced immunosuppression and/or less severe tuberculosis.
In summary, our multicenter study showed that, in tuberculosis suspects with advanced HIV-related immunosuppression, the TB LAM urine test detected over half of tuberculosis patients, and the TB LAM test used in combination with sputum smear microscopy detected two-thirds of tuberculosis patients. TB LAM specificity was high in this study that rigorously sought a gold-standard tuberculosis microbiological diagnosis. These features in a truly point-of-care urine test might allow rapid tuberculosis diagnosis and treatment in resource-limited settings and thereby reduce mortality and morbidity in a vulnerable patient population. Our results provide strong rationale for studies to assess the impact of urine TB LAM testing on patient-centered outcomes.
The authors thank the following individuals and organizations for their important contributions toward implementation of this study: Rosie Burton, Gaironesa Omar, Layla Hendricks, Melissa Jansen van Rensburg, Slee Mbhele, Meagan McMaster, the Provincial Government of the Western Cape, the City of Cape Town Health Department, clinical staff at Town 2 Clinic and G.F. Jooste Hospital, and staff of the C18 Medical Microbiology Laboratory of the Groote Schuur Hospital, National Health Laboratory Service in South Africa; Gloria Lubega, John Mark Bwanika, Teddy Nalwoga, Willy Ssengooba, Francis Mumbowa, Carolyn Namaganda, Allan Buzibye, and Henry Ojambo in Kampala, Uganda; Kathleen Robergeau Hunt and Nancy Dianis at Westat; David Hom, Rachel Kubiak, and Mary Gaeddert at the CDRC Data Coordinating Center at Boston University Medical Center; and Karen Lacourciere and Tena Knudsen at the National Institutes of Health. Catharina Boehme and Mark Perkins at the Foundation for Innovative New Diagnostics provided valuable input into the study design. The authors sincerely thank Professor Graeme Mentjies and Dr. David Meya, who served as independent safety monitors at the South Africa and Uganda sites, respectively. The authors and study site teams gratefully acknowledge and thank the study participants for their time and willingness to be involved in the study. The authors thank Alere for the generous donation of Determine TB LAM Ag test strips.
2. Getahun H, Gunneberg C, Granich R, et al.. HIV infection-associated tuberculosis: the epidemiology and the response. Clin Infect Dis. 2010;50(suppl 3):S201–S207.
3. Kyeyune R, den Boon S, Cattamanchi A, et al.. Causes of early mortality in HIV-infected TB suspects in an East African referral hospital. J Acquir Immune Defic Syndr. 2010;55:446–450.
4. Chamie G, Luetkemeyer A, Walusimbi-Nanteza M, et al.. Significant variation in presentation of pulmonary tuberculosis across a resolution of CD4 strata. Int J Tuberc Lung Dis. 2010;14:1295–1302.
5. Getahun H, Harrington M, O'Brien R, et al.. Diagnosis of smear-negative pulmonary tuberculosis in people with HIV infection or AIDS in resource-constrained settings: informing urgent policy changes. Lancet. 2007;369:2042–2049.
6. Boehme CC, Nabeta P, Hillemann D, et al.. Rapid molecular detection of tuberculosis and rifampin resistance. N Engl J Med. 2010;363:1005–1015.
7. Creswell J, Codlin AJ, Andre E, et al.. Results from early programmatic implementation of Xpert MTB/RIF testing in nine countries. BMC Infect Dis. 2014;14:2.
8. Blanc FX, Sok T, Laureillard D, et al.. Earlier versus later start of antiretroviral therapy in HIV-infected adults with tuberculosis. N Engl J Med. 2011;365:1471–1481.
9. Havlir DV, Kendall MA, Ive P, et al.. Timing of antiretroviral therapy for HIV-infection and tuberculosis. N Engl J Med. 2011;365:1482–1491.
10. Abdool Karim SS, Naidoo K, Grobler A, et al.. Integration of antiretroviral therapy with tuberculosis treatment. N Engl J Med. 2011;365:1492–1501.
11. Brennan PJ. Structure, function, and biogenesis of the cell wall of Mycobacterium tuberculosis. Tuberculosis (Edinb). 2003;83:91–97.
12. Shah M, Variava E, Holmes CB, et al.. Diagnostic accuracy of a urine lipoarabinomannan test for tuberculosis in hospitalized patients in a High HIV prevalence setting. J Acquir Immune Defic Syndr. 2009;52:145–151.
13. Lawn SD, Edwards DJ, Kranzer K, et al.. Urine lipoarabinomannan assay for tuberculosis screening before antiretroviral therapy diagnostic yield and association with immune reconstitution disease. AIDS. 2009;23:1875–1880.
14. Dheda K, Davids V, Lenders L, et al.. Clinical utility of a commercial LAM-ELISA assay for TB diagnosis in HIV-infected patients using urine and sputum samples. PLoS One. 2010;5:e9848.
15. Daley P, Michael JS, Hmar P, et al.. Blinded evaluation of commercial urinary lipoarabinomannan for active tuberculosis: a pilot study. Int J Tuberc Lung Dis. 2009;13:989–995.
16. Gounder CR, Kufa T, Wada NI, et al.. Diagnostic accuracy of a urine lipoarabinomannan enzyme-linked immunosorbent assay for screening ambulatory HIV-infected persons for tuberculosis. J Acquir Immune Defic Syndr. 2011;58:219–223.
17. Minion J, Leung E, Talbot E, et al.. Diagnosing tuberculosis with urine lipoarabinomannan: systematic review and meta-analysis. Eur Respir J. 2011;38:1398–405.
19. Kent PT, Kubica GP. Public Health Mycobacteriology: A Guide for the Level III Laboratory. Atlanta, GA: Centers for Disease Control; 1985.
20. Crump JA, Tanner DC, Mirrett S, et al.. Controlled comparison of BACTEC 13A, MYCO F/LYTIC, BacT/ALERT MB, and ISOLATOR 10 systems for detection of mycobacteremia. J Clin Microbiol. 2003;41:1987–1990.
21. Crump JA, Morrissey AB, Ramadhani HO, et al.. Controlled comparison of BacT/Alert MB system, manual Myco/F lytic procedure, and isolator 10 system for diagnosis of Mycobacterium tuberculosis bacteremia. J Clin Microbiol. 2011;49:3054–3057.
24. Mukundan H, Kumar S, Price DN, et al.. Rapid detection of Mycobacterium tuberculosis biomarkers in a sandwich immunoassay format using a waveguide-based optical biosensor. Tuberculosis (Edinb). 2012;92:407–416.
25. Ferrara F, Naranjo LA, Kumar S, et al.. Using phage and yeast display to select hundreds of monoclonal antibodies: application to antigen 85, a tuberculosis biomarker. PLoS One. 2012;7:e49535.
26. Lawn SD. Point-of-care detection of lipoarabinomannan (LAM) in urine for diagnosis of HIV-associated tuberculosis: a state of the art review. BMC Infect Dis. 2012;12:103.
27. Wood R, Racow K, Bekker LG, et al.. Lipoarabinomannan in urine during tuberculosis treatment: association with host and pathogen factors and mycobacteriuria. BMC Infect Dis. 2012;12:47.
28. Lawn SD, Kerkhoff AD, Vogt M, et al.. High diagnostic yield of tuberculosis from screening urine samples from HIV-infected patients with advanced immunodeficiency using the Xpert MTB/RIF assay. J Acquir Immune Defic Syndr. 2012;60:289–294.
29. Peter JG, Theron G, Muchinga TE, et al.. The diagnostic accuracy of urine-based Xpert MTB/RIF in HIV-infected hospitalized patients who are smear-negative or sputum scarce. PLoS One. 2012;7:e39966.
30. Lawn SD, Kerkhoff AD, Vogt M, et al.. HIV-associated tuberculosis: relationship between disease severity and the sensitivity of new sputum-based and urine-based diagnostic assays. BMC Med. 2013;11:231.
31. Lawn SD, Kerkhoff AD, Vogt M, et al.. Diagnostic accuracy of a low-cost, urine antigen, point-of-care screening assay for HIV-associated pulmonary tuberculosis before antiretroviral therapy: a descriptive study. Lancet Infect Dis. 2012;12:201–209.
32. Peter JG, Theron G, van Zyl-Smit R, et al.. Diagnostic accuracy of a urine lipoarabinomannan strip-test for TB detection in HIV-infected hospitalised patients. Eur Respir J. 2012;40:1211–1220.
33. Peter JG, Theron G, Dheda K. Can point-of-care urine LAM strip testing for tuberculosis add value to clinical decision making in hospitalized HIV-infected persons? PLoS One. 2013;8:e54875.
34. Bossuyt PM, Reitsma JB, Bruns DE, et al.. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD Initiative. Ann Intern Med. 2003;138:40–44.
35. Whiting P, Rutjes AWS, Westwood ME, et al.. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011;155:529–536.
36. Sun D, Dorman S, Shah M, et al.. Cost utility of lateral-flow urine lipoarabinomannan for tuberculosis diagnosis in HIV-infected African adults. Int J Tuberc Lung Dis. 2013;17:552–558.
sensitivity and specificity; tuberculosis; opportunistic infections; diagnosis; HIV
Supplemental Digital Content
© 2014 by Lippincott Williams & Wilkins
Highlight selected keywords in the article text.