JOURNAL OF CLINICAL MICROBIOLOGY, Aug. 2010, p. 29722974 Vol. 48, No. 80095-1137/10/$12.00 doi:10.1128/JCM.00363-10Copyright 2010, American Society for Microbiology. All Rights Reserved.

Quantitative Analysis of a Urine-Based Assay for Detection ofLipoarabinomannan in Patients with Tuberculosis

Maunank Shah,1 Neil A. Martinson,1,2 Richard E. Chaisson,1 Desmond J. Martin,3Ebrahim Variava,4 and Susan E. Dorman1*

Johns Hopkins University School of Medicine, Baltimore, Maryland1; Perinatal HIV Research Unit, University of the Witwatersrand,Soweto, South Africa2; Toga Laboratories, Johannesburg, South Africa, and Department of Medical Virology, University of

Pretoria, South Africa3; and Tshepong Hospital, Northwest Department of Health, Klerksdorp, South Africa4

Received 22 February 2010/Returned for modification 18 May 2010/Accepted 2 June 2010

Urinary lipoarabinomannan (LAM) detection is a promising approach for rapid diagnosis of active tuber-culosis (TB). In microbiologically confirmed TB patients, quantitative LAM detection results increased pro-gressively with bacillary burden and immunosuppression. Patients with disseminated TB and/or advanced HIVare target populations for whom urine LAM detection may be particularly useful.

Effective control of tuberculosis (TB) is hindered by the lackof rapid, accurate diagnostic modalities. Current tools haveserious limitations when applied to HIV-infected patients, andclinical diagnosis can be challenging (2, 6, 8, 13). Sputumsmear microscopy is widely utilized to diagnose active TB buthas impaired sensitivity and detects less than half of HIV-TBcoinfection cases (2, 6, 8). Mycobacterial culture is the labo-ratory standard for diagnosis but is not widely available inresource-constrained settings and can take weeks to determinea positive result.

The detection of lipoarabinomannan (LAM), a 17.5-kDaglycolipid component of the mycobacterial cell wall, is an at-tractive approach to diagnosing active TB (1, 7, 9, 15, 16).LAM is released from metabolically active mycobacteria and isdetectable intact in urine (1, 7). The Clearview TB enzyme-linked immunosorbent assay (ELISA) (urine LAM test; In-verness Medical Innovations, Waltham, MA) is a direct anti-gen sandwich immunoassay in a 96-well-plate format thatprovides both quantitative and qualitative results. Our groupand others have recently reported on the qualitative diagnosticaccuracy of this urine LAM test (9, 10, 16). Urine LAM testsensitivity, while imperfect, appears to be higher than that ofsputum smear microscopy, and the test performs with a highpositive predictive value in populations with high HIV and TBprevalence (9, 12, 16). However, quantitative urine LAM testresults have not been studied fully. Quantitative analysis allowsa more complete understanding of test performance and mayoffer insight into optimal test usage. Preclinical data and lim-ited regression modeling from clinical studies have shown thatquantitative test results positively correlate with increasing ba-cillary burden (1, 16). We therefore examined the relationshipbetween quantitative urine LAM test results and TB charac-teristics among patients with culture-confirmed TB.

A full description of the study design has been publishedpreviously (16). Briefly, we conducted a nested prospective

cohort study at three hospitals in South Africa to evaluate thediagnostic accuracy of the urine LAM test. Hospitalized adultTB suspects were enrolled after informed consent. Study-di-rected testing included sputum smear microscopy for acid-fastbacilli (AFB), sputum mycobacterial culture, mycobacterialblood cultures, HIV and CD4 testing, urine LAM testing usingthe Clearview TB ELISA kit, and additional mycobacterialcultures (e.g., from enlarged lymph nodes) when clinically in-dicated. Participants had a follow-up study visit at 2 months.Participants with at least one positive AFB smear or mycobac-terial culture positive for Mycobacterium tuberculosis were de-termined to have confirmed TB and were included in thecurrent analysis. Individuals with M. tuberculosis isolated froma blood culture (mycobacteremia) were considered to havedisseminated disease. Pulmonary TB (PTB) was defined as thepresence of one or more sputa positive for AFB by microscopyand/or positive for M. tuberculosis by culture. Localized ex-trapulmonary TB was defined as isolation of M. tuberculosisfrom pleural culture or fine-needle aspiration of a lymph nodebut not from other respiratory specimens or blood. Quantita-tive urine LAM test results were expressed as optical density(OD) readings. The final sample OD was determined by sub-tracting the OD of the negative control from the sample read-ing, with a minimum value of 0 (i.e., a value of 0 was reportedif the OD of the negative control was greater than the OD ofthe sample). An OD above 0.1 was considered positive permanufacturer specifications. Multiple linear regression modelswere used to determine predictors of higher quantitative urineLAM test results. Wilcoxon rank sum and Kruskal-Wallis testswere used to compare nonnormal continuous outcomes; chi-square (2) tests were used to compare categorical outcomes.A P value of 0.05 was considered statistically significant, and95% confidence intervals (95% CI) were used. Statistical cal-culations were performed using Stata 10.1 (StataCorp). Thisstudy was approved by the institutional review boards of JohnsHopkins University School of Medicine and the University ofWitwatersrand.

Full characteristics of the study population are publishedelsewhere (16). Overall, 499 participants were enrolled, and193 (39%) had confirmed TB (185 with positive cultures for M.

* Corresponding author. Mailing address: Johns Hopkins UniversityCenter for Tuberculosis Research, CRB2, Room 1M-12, 1550 OrleansStreet, Baltimore, MD 21231. Phone: (410) 502-2717. Fax: (410) 955-0740. E-mail: dsusan1@jhmi.edu.

Published ahead of print on 9 June 2010.

2972

tuberculosis and 8 with positive AFB smears only); 6/193 (3%)confirmed TB cases were on recent TB treatment (less than 2months) at time of enrollment. Among the confirmed TBcases, the median age was 34 years, 65% were female, and 87%(167/193) had documented HIV infection, of which 9% (15/167) were on antiretroviral therapy (ART) at the time of en-rollment. TB manifestations ranged from localized to dissem-inated disease (Table 1). Localized extrapulmonary TB wasdiagnosed in 5 cases (3%), PTB without mycobacteremia (M.tuberculosis in sputum but not blood) in 145 cases (75%),mycobacteremia without PTB (M. tuberculosis in blood but notsputum) in 18 cases (9%), and combined PTB with mycobac-teremia (M. tuberculosis in blood and sputum) in 25 cases(13%). The overall LAM test sensitivity was 59% (95% CI, 52to 66) in participants with confirmed TB, and the specificitywas 96% (95% CI, 91 to 99) among individuals without TB.Among participants with confirmed TB, the range of quanti-tative LAM test results was 0 to 4.87 (median, 0.19; interquar-tile range [IQR], 0.03 to 1.3).

Among participants with confirmed TB, there was a signif-icant difference in median OD for those with more localizeddisease compared to those with disseminated disease (P 0.0001) (Table 1). The median OD increased progressivelyfrom 0.01 (IQR, 0.01 to 0.06) to 0.13 (IQR, 0.02 to 0.70) to 0.64(IQR, 0.14 to 1.43) to 1.4 (IQR, 0.43 to 2.8) in those withlocalized extrapulmonary TB, PTB alone, mycobacteremiawithout PTB, and combined PTB with mycobacteremia, re-spectively. Among 161 participants with sputum culture-posi-tive PTB, those that were AFB smear positive had higher ODsthan those that were smear negative (median ODs of 0.33versus 0.13, respectively; P 0.03) (Table 2). There was atrend toward higher OD with increasing grades of smear pos-itivity (median ODs of 0.13 [IQR, 0.02 to 0.62], 0.18 [IQR, 0.1to 1.3], 0.26 [IQR, 0.03 to 2.8], and 0.38 [IQR, 0.048 to 1.8] forsmear-negative, smear-positive grade 1, smear-positive grade2, and smear-positive grade 3 cases, respectively; P 0.18).Among confirmed TB patients with HIV infection, the urineLAM test OD increased with decreasing strata of CD4 count

(P 0.0001) (Table 3). The median OD increased progres-sively from 0.10 (IQR, 0.02 to 0.5) to 0.20 (IQR, 0.03 to 0.8) to0.42 (IQR, 0.04 to 1.0) to 1.3 (IQR, 0.2 to 2.8) in those withCD4 counts of 150, 101 to 150, 50 to 100, and 50, respec-tively (P 0.0001).

Multiple linear regression models were used to further ex-plore factors associated with higher quantitative urine LAMtest results. Covariates included in the models were age, sex,HIV status, mortality at 2 months, TB treatment, mycobacte-remia, sputum smear positivity, fever, Karnofsky score, chest Xray (CXR) cavitation, hypoxia, pulse, and CD4 strata. Adjust-ing for all other factors, individuals with mycobacteremia hadan average OD that was 0.64 OD units higher than that forindividuals with negative blood cultures (P 0.002); individ-uals with smear-positive pulmonary TB had an average ODthat was 0.33 OD units higher than that for individuals withsmear-negative pulmonary TB (P 0.05). Among HIV-in-fected patients, individuals with CD4 counts of 50 had anaverage OD that was 1.05 OD units higher than that for indi-viduals with CD4 counts of 150 (P 0.0001). No othercovariates were significantly associated with higher quantita-tive urine LAM test results.

Taken together, our findings show that quantitative urineLAM test results correlate with degree of bacillary burden inpatients with microbiologically confirmed TB. Further, the ab-solute urine LAM test OD increased dramatically as immuno-

TABLE 2. Optical density by sputum AFB smear microscopy statusamong participants with sputum culture-positive pulmonary TB

Sputum AFBsmear microscopy

status (no. ofcases)c

OD median(IQR)

OD mean(SD)

No. of cases withpositive urine

LAM test result(% sensitivity,

95% CI)

Negative (88) 0.13 (0.020.62)a,b 0.60 (0.93) 47 (53, 0.420.64)

PositiveAny grade (73) 0.33 (0.0521.71)b 0.94 (1.15) 47 (64, 0.520.75)Grade 1 (13) 0.18 (0.11.3)a 0.68 (0.89) 10 (77, 0.460.94)Grade 2 (15) 0.26 (0.032.8)a 1.13 (1.3) 10 (67, 0.380.88)Grade 3 (45) 0.38 (0.0481.8)a 0.95 (1.18) 27 (60, 0.440.74)

a P 0.18 for comparison of AFB-negative, AFB 1, AFB 2, and AFB 3results by the Kruskal-Wallis test.

b P 0.03 for comparison of AFB-negative and AFB-positive results by theWilcoxon rank sum test.

c Grading per WHO guidelines (http://wwwn.cdc.gov/dls/ila/documents/lstc2.pdf).

TABLE 1. Optical density by type of disease (determined by site)among participants with confirmed TB

Type of disease (no. ofcases)

OD median(IQR)

OD mean(SD)

No. of cases withpositive urine

LAM test result(% sensitivity,

95% CI)

Localized extrapulmonaryTB (5)a

0.01 (0.010.06) 0.09 (0.16) 1 (20, 172)

Pulmonary TB only (145)b 0.13 (0.020.70) 0.63 (0.97) 78 (54, 4562)Mycobacteremia only (18)c 0.64 (0.141.43) 0.99 (1.02) 14 (78, 5294)Pulmonary TB with

mycobacteremia (25)1.4 (0.432.8) 1.6 (1.1) 21 (84, 6496)

P value 0.0001d e 0.003f

a Four of five participants had M. tuberculosis isolated from pleural cultures,and 1 of 5 had M. tuberculosis isolated from a lymph node. The urine LAM testwas positive in the participant with M. tuberculosis isolated from the lymph nodebut negative in the 4 participants with localized pleural TB.

b M. tuberculosis in sputum but not blood.c M. tuberculosis in blood but not sputum.d Kruskal-Wallis test of equality between types of disease.e , analysis of variance (ANOVA) was not performed, due to nonnormal

distribution.f Chi-square (2) test comparing proportions of positive urine LAM test results

(sensitivities) between types of disease.

TABLE 3. Optical density by CD4 status among participants withHIV infection and confirmed TB

CD4 count(no. of cases)

OD median(IQR)

OD mean(SD)

No. of cases withpositive urine LAM

test results (%sensitivity, 95% CI)

150 (63) 0.10 (0.020.5) 0.42 (0.72) 32 (51, 0.380.64)101150 (16) 0.2 (0.030.8) 0.65 (0.96) 9 (56, 0.300.80)50100 (28) 0.42 (0.041.0) 0.72 (0.9) 20 (71, 0.510.87)50 (60) 1.3 (0.22.8) 1.4 (1.12) 51 (85, 0.730.93)P value 0.0001a b 0.00c

a Kruskal-Wallis test of equality between CD4 strata.b , ANOVA was not performed, due to nonnormal distribution.c Chi-square (2) test comparing proportions of positive urine LAM test re-

sults (sensitivities) between CD4 strata.

VOL. 48, 2010 NOTES 2973

suppression progressed among HIV-coinfected TB patients,probably due to higher mycobacterial burden and greater like-lihood of disseminated TB. Strikingly, the median quantitativeurine LAM test result was over 10-fold higher than the currentthreshold for test positivity in those with CD4 counts less than50 and 5-fold higher than this threshold in those with CD4counts between 51 and 100. Usage of quantitative urine LAMtest results may thus offer additional clinical insight into thedegree of TB disease severity that cannot be gleaned fromqualitative results alone.

To date, studies of the qualitative accuracy of the urineLAM test have yielded varied results. In heterogeneous pop-ulations with low HIV prevalence, the sensitivity has rangedfrom 13 to 51% (3, 4, 10, 12, 14). Alternatively, sensitivity risesamong HIV-positive patients and could reach 67 to 85% inthose with low CD4 counts (9, 16). This quantitative analysisprovides evidence that urine LAM test performance may bebest in patients with disseminated TB disease and high myco-bacterial disease burden. HIV-infected TB suspects with ad-vanced immunosuppression, a group in which sputum micros-copy is of low yield (5, 8, 11, 17), may be a target population forwhom the urine LAM test would be particularly useful. Ofnote, the combination of LAM testing and sputum smear mi-croscopy may have an additive benefit in this target group (16).While the urine LAM test is unlikely to stand alone for defin-itive TB diagnostic testing, its usage is attractive as a rapiddiagnostic modality that complements smear microscopy forHIV-prevalent, resource-constrained settings where mycobac-terial culture is unavailable.

This work was funded in part by the Doris Duke CharitableFoundation and the United Kingdom Government through a grantfrom the Department for International Development (DFID). Ad-ditional support was provided by NIH grants AI48526, AI01637,and T32AI007291-18.

Clearview TB ELISA kits were generously provided by InvernessMedical Innovations, which had no role in the study design, imple-mentation, or analysis, the manuscript writing, or the decision to sub-mit a manuscript for publication.

REFERENCES

1. Boehme, C., E. Molokova, F. Minja, S. Geis, T. Loscher, L. Maboko, V.Koulchin, and M. Hoelscher. 2005. Detection of mycobacterial lipoarabino-mannan with an antigen-capture ELISA in unprocessed urine of Tanzanian

patients with suspected tuberculosis. Trans. R. Soc. Trop. Med. Hyg. 99:893900.

2. Colebunders, R., and I. Bastian. 2000. A review of the diagnosis and treat-ment of smear-negative pulmonary tuberculosis. Int. J. Tuberc. Lung Dis.4:97.

3. Daley, P., J. S. Michael, P. Hmar, A. Latha, P. Chordia, D. Mathai, K. R.John, and M. Pai. 2009. Blinded evaluation of commercial urinary lipoarabi-nomannan for active tuberculosis: a pilot study. Int. J. Tuberc. Lung Dis.13:989995.

4. Dheda, K., V. Davids, L. Lenders, T. Roberts, R. Meldau, D. Ling, L. Brunet,R. van Zyl Smit, J. Peter, C. Green, M. Badri, L. Sechi, S. Sharma, M.Hoelscher, R. Dawson, A. Whitelaw, J. Blackburn, M. Pai, and A. Zumla.2010. Clinical utility of a commercial LAM-ELISA assay for TB diagnosis inHIV-infected patients using urine and sputum samples. PLoS One 5:e9848.

5. El-Sadr, W. M., and S. J. Tsiouris. 2008. HIV-associated tuberculosis: diag-nostic and treatment challenges. Semin. Respir. Crit. Care Med. 29:525531.

6. Getahun, H., M. Harrington, R. OBrien, and P. Nunn. 2007. Diagnosis ofsmear-negative pulmonary tuberculosis in people with HIV infection orAIDS in resource-constrained settings: informing urgent policy changes.Lancet 369:2042.

7. Hamasur, B., J. Bruchfeld, M. Haile, A. Pawlowski, B. Bjorvatn, G. Kalle-nius, and S. B. Svenson. 2001. Rapid diagnosis of tuberculosis by detectionof mycobacterial lipoarabinomannan in urine. J. Microbiol. Methods 45:4152.

8. Havlir, D. V., H. Getahun, I. Sanne, and P. Nunn. 2008. Opportunities andchallenges for HIV care in overlapping HIV and TB epidemics. JAMA300:423.

9. Lawn, S. D., D. J. Edwards, K. Kranzer, M. Vogt, L. G. Bekker, and R. Wood.2009. Urine lipoarabinomannan assay for tuberculosis screening before an-tiretroviral therapy diagnostic yield and association with immune reconsti-tution disease. AIDS 23:18751880.

10. Mutetwa, R., C. Boehme, M. Dimairo, T. Bandason, S. S. Munyati, D.Mangwanya, S. Mungofa, A. E. Butterworth, P. R. Mason, and E. L. Corbett.2009. Diagnostic accuracy of commercial urinary lipoarabinomannan detec-tion in African tuberculosis suspects and patients. Int. J. Tuberc. Lung Dis.13:12531259.

11. Perkins, M. D., and J. Cunningham. 2007. Facing the crisis: improving thediagnosis of tuberculosis in the HIV era. J. Infect. Dis. 196(Suppl. 1):S15.

12. Peter, J., C. Green, M. Hoelscher, P. Mwaba, A. Zumla, and K. Dheda. 2010.Urine for the diagnosis of tuberculosis: current approaches, clinical applica-bility, and new developments. Curr. Opin. Pulm. Med. 16:262270.

13. Reid, M. J., and N. S. Shah. 2009. Approaches to tuberculosis screening anddiagnosis in people with HIV in resource-limited settings. Lancet Infect. Dis.9:173.

14. Reither, K., E. Saathoff, J. Jung, L. T. Minja, I. Kroidl, E. Saad, J. F.Huggett, E. N. Ntinginya, L. Maganga, L. Maboko, and M. Hoelscher. 2009.Low sensitivity of a urine LAM-ELISA in the diagnosis of pulmonary tu-berculosis. BMC Infect. Dis. 9:141.

15. Sada, E., D. Aguilar, M. Torres, and T. Herrera. 1992. Detection of li-poarabinomannan as a diagnostic test for tuberculosis. J. Clin. Microbiol.30:24152418.

16. Shah, M., E. Variava, C. B. Holmes, A. Coppin, J. E. Golub, J. McCallum,M. Wong, B. Luke, D. J. Martin, R. E. Chaisson, S. E. Dorman, and N. A.Martinson. 2009. Diagnostic accuracy of a urine lipoarabinomannan test fortuberculosis in hospitalized patients in a high HIV prevalence setting. J.Acquir. Immune Defic. Syndr. 52:145151.

17. Swaminathan, S., C. Padmapriyadarsini, and G. Narendran. 2010. HIV-associated tuberculosis: clinical update. Clin. Infect. Dis. 50:13771386.

2974 NOTES J. CLIN. MICROBIOL.