Diagnosis of Tuberculosis by a Visually Detectable Immunoassay for Lipoarabinomannan

Transcription

1 Diagnosis of Tuberculosis by a Visually Detectable Immunoassay for Lipoarabinomannan EDWARD D. CHAN, RANDALL REVES, JOHN T. BELISLE, PATRICK J. BRENNAN, and WILLIAM E. HAHN Division of Pulmonary Sciences and Critical Care Medicine, and Department of Cellular and Structural Biology, University of Colorado Health Sciences Center, and National Jewish Medical and Research Center, Denver, Colorado; Denver Metro Tuberculosis Clinic, Denver Health Medical Center, Denver, Colorado; and Department of Microbiology, Colorado State University, Fort Collins, Colorado Recovery of tubercle bacilli from sputum, tissue, or body fluid is the standard for the diagnosis of tuberculosis (TB) although this process is technically demanding and relatively insensitive. We have developed a simplified, visually detectable, colloidal goldbased serological assay to qualitatively detect IgG directed against the mycobacterial cell wall component lipoarabinomannan (LAM). The objective of this investigation is to determine the accuracy of this assay in patients with active pulmonary TB and in control patients with or without latent infection. In patients with active TB, the sensitivity of anti-lam IgG was 85 to 93%. In five patients with active TB who were smear-negative, all tested positive for anti- LAM IgG. The specificity of the test depended on the presence of tuberculous infection. In U.S. citizens comprised of young healthy adults and rheumatology patients, the specificity was 100%. In an at-risk population for tuberculous infection who were either tuberculin skin test negative or positive, the specificity was 89%. The negative and positive predictive values of the test were 98% and 52%, respectively. We conclude that anti-lam IgG immunoassay is relatively sensitive and specific for active TB and thus, a potentially useful screening test for active TB. The diagnosis of active tuberculosis (TB) largely depends upon initial clinical suspicion and radiographic findings, with subsequent laboratory confirmation by sputum examination and culture, which remain the gold standard (1). Although acid-fast staining of bacilli in sputum smears is a simple and relatively fast means for detection of active TB, sensitivity is compromised because greater than 10 4 bacilli per ml of sputum are required for reliable detection. Thus, approximately one-half of the cases of active pulmonary TB are smear-negative, the failure rate being substantially greater in children, the elderly, and patients with acquired immunodeficiency syndrome (AIDS) (2). Although sputum digestion, decontamination, and concentration can increase the yield of Mycobacterium tuberculosis, organisms may be killed if the process is improperly performed, resulting in false-negative cultures. Other problematic issues with sputum analyses, albeit sporadic and laboratory-dependent, include doubtful labeling of acid-fast bacillus smear, overgrowth of cultures with nonmycobacterial microorganisms, and cross-contamination of specimens resulting in false-positive cultures or misprofiling of drug sensitivities. For example, in three separate studies from New York City and Denver, laboratory cross-contamination (Received in original form August 30, 1999 and in revised form November 18, 1999) Edward D. Chan is a recipient of the Clinical Investigator Development Award K08HL , the Bettina Garthwaite Lowerre Foundation for Tuberculosis Research Award, and the Parke-Davis Atorvastatin Research Award. LAM was produced under NIH, NIAID Contract NO1 AI Correspondence and requests for reprints should be addressed to Edward D. Chan, M.D., K613e, Goodman Building, National Jewish Medical and Research Center, 1400 Jackson Street, Denver, CO chane@njc.org Am J Respir Crit Care Med Vol 161. pp , 2000 Internet address: occurred in 3 to 12% of all sputum cultures for TB (3). In addition to these potential shortcomings associated with sputum examination, approximately 15% of verified TB cases annually in the U.S. lack bacteriologic evidence (smear-negative, culture-negative), and are diagnosed based on high clinical suspicion and favorable response to antituberculous therapy. Extrapulmonary and disseminated TB present unique diagnostic challenges including subtle and protean clinical manifestations, and the relative insensitivity of acid-fast smear of body fluid or tissues. In a World Health Organization (WHO)-sponsored TB Diagnostics Workshop held in 1997, it was determined that two of the leading priorities for the development of new diagnostics were tests to replace acid-fast microscopy for the diagnosis of smear-positive TB, and tests to improve the differential diagnosis of smear-negative TB (4). The search for a rapid and reliable diagnostic test for active TB based on examination of sputum, blood, or other clinical specimens has been the focus of a number of studies. Measurement of tuberculostearic acid in clinical specimens showed a high degree of accuracy, but the assay required considerable expertise and sophisticated equipment to perform (5). Detection of M. tuberculosis DNA by polymerase chain reaction (PCR) is highly sensitive, but the personnel training and costly equipment required would be prohibitive in many U.S. hospitals and certainly in most countries where TB is highly prevalent (6, 7). Although a recent analysis of molecular diagnostic techniques including PCR found it to be highly accurate on respiratory specimens (6), the sensitivity and specificity of PCR is reduced in serum specimens (8). The sensitivity of the molecular techniques is also diminished when testing smear-negative specimens (6). Furthermore, in many areas of the world where the rate of tuberculin positivity is high, the specificity of PCR for active cases would likely be reduced. In recent years, the detection of active TB by serologic means has been the subject of a number of investigations. Although serologic testing for TB is unnecessary in cases where sputum examination is diagnostic, its areas of potential use include patients who are unable to produce adequate sputum, who are smear-negative, or who are suspected to have extrapulmonary TB (9, 10). Previously, one of us showed that antibodies against lipoarabinomannan (LAM), a lipoglycan component of the mycobacterial cell wall, were present in 72% of patients with active TB but in only 9% of control patients (11). However, in the control group, individuals who were tuberculin skin test positive were not discriminated from those who were negative. The purpose of this report is to investigate the diagnostic accuracy of anti-lam IgG for pulmonary TB in a simplified and visually detectable dot-blot immunoassay. Because LAM is highly antigenic in vitro and is released into medium by growing mycobacteria, we hypothesized that detectable levels of anti-lam IgG will be present in the sera of patients with active TB but not in subjects who are not infected or have latent infection. Herein, we report that anti-lam IgG

2 1714 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL is linked to active disease with a high degree of sensitivity and specificity, and has a high negative predictive value. METHODS Study Groups We first compared patients with active TB from Bombay, India with low-risk residents from the United States. Sera from 29 patients with culture-confirmed active TB were obtained from a TB hospital in Bombay, India. Thirty-seven control sera were collected from United States Air Force Academy (USAFA) cadets and from patients evaluated at a rheumatology clinic at the University of Colorado Health Sciences Center in Denver, Colorado. These control samples were originally obtained for other unrelated studies. The cadets were all healthy and had negative responses to the purified protein derivative (PPD) skin test. The PPD status of the rheumatology patients was not known but they were all considered free of active TB by their physicians. To evaluate the test in a clinical setting in a U.S. population considered to be at increased risk for TB, 115 serum specimens were obtained from subjects who were being seen at the Denver Metropolitan Tuberculosis Clinic over a 3-mo period. This population was largely comprised of (1) immigrants or refugees from developing countries (Mexico, Southeast Asia, Eastern Europe) who were being screened for TB and chronic hepatitis B, and (2) patients with active TB during initial or follow-up laboratory monitoring for their treatment. These 115 specimens were labeled with sequential numbers and sent without patient identifiers or clinical information to the laboratory for testing. Thus, the laboratory personnel performing the seroassays were blinded to the clinical status of the patients. Subsequent record reviews were done to categorize the 115 individuals into four different classes of infection based on the results of their PPD skin test, chest radiographs, and when necessary, sputum examination for acid-fast bacilli and culture: class 0: no known recent exposure, no infection (PPD-negative), no disease; class II: latent infection as defined by a positive PPD ( 10 mm induration) and no clinical or radiographic evidence of active TB; class III: active TB confirmed by positive M. tuberculosis culture; class IV: PPD-positive ( 5 mm induration) with evidence of past disease by history or compatible chest radiograph (e.g., upper lobe fibronodular disease) but negative sputum smear and culture. Materials The LAM used for these studies was isolated from cells of M. tuberculosis H37Rv grown to midlog in glycerol alanine salts medium (12). To each 6-mm nitrocellulose disc, 50 ng of LAM dissolved in 0.4 l of water was applied on the center, then dried at 45 C for 30 min. Colloidal gold particles, 30 to 40 mm in diameter, were produced by the sodium citrate method (13). The colloid was loaded with recombinant protein G, that lacked the serum albumin binding domain (Genex Corp, Gaithersburg, MD), by adjusting the ph to 5.2, allowing the protein to adsorb onto the colloid (13). Figure 1. Method diagram of dot-blot assay to detect serum anti-lam IgG. Immuno-dot Blot Assay for Detection of IgG against LAM A diagram of the method is shown in Figure 1. Colloidal gold protein G was used to bind IgG present in serum specimens, which were kept frozen at 20 C until use. For each assay, 5 to 10 l of serum was thoroughly mixed with 0.4 ml of colloidal gold protein G suspended in phosphate-buffered saline (ph 7.2) at a concentration of 3 to 4 optical density units determined spectrophotometrically at a wavelength of 530 nm. IgG rapidly binds to the colloidal gold protein G owing to high affinity and extensive surface area. The LAM-containing nitrocellulose discs, blocked for nonspecific binding with bovine serum albumin, were incubated in the serum colloidal gold mixture for 40 to 60 min under continuous agitation at room temperature. After incubation, the discs were removed and washed free of unbound particles using the capillary action of an absorbent pad to draw buffer through the nitrocellulose. The air-dried discs were then examined for the presence of a light to deep purple spot indicative of the binding of the anti-lam IgG colloidal gold complex to the LAM blot. A small number of the serum specimens apparently contained components that caused the colloidal gold to bind nonspecifically over the entire surface of a nitrocellulose disc. This diffuse signal precluded a definitive interpretation and thus these specimens were labeled as indeterminate. Representative dot blots depicting positive, negative, and indeterminate results are shown in Figure 2. RESULTS Anti-LAM IgG Is Highly Sensitive for Active TB To test the accuracy of anti-lam IgG in the diagnosis of active TB, we compared the sera of patients with culture-confirmed active TB from India to sera from healthy U.S. controls (US- AFA cadets) and to patients with various collagen vascular diseases seen at the University of Colorado Health Sciences Center (Table 1). Twenty-seven of 29 patients with active TB were positive for anti-lam IgG (sensitivity of 93%; 95% confidence interval [CI] of 77 to 99%). In contrast, none of the 37 control sera from U.S. citizens were found to be positive for anti-lam IgG (specificity of 100%; 95% CI 91 to 100%). Therefore, based on these two populations, albeit widely disparate in the Figure 2. Representative nitrocellulose discs of dot-blot assay. The five illustrated discs represent a negative test for anti-lam IgG (-), positive tests for anti- LAM IgG ( to ), and an indeterminate test (ID). See text for explanation of an indeterminate blot.

3 Chan, Reves, Belisle, et al.: Serologic Diagnosis of TB 1715 TABLE 1 PRESENCE OF ANTI-LAM IgG IN SERA FROM PATIENTS WITH ACTIVE TB FROM BOMBAY, INDIA AND FROM U.S. CONTROL SUBJECTS* Classification Positive Negative Samples Sensitivity % Specificity % (95% CI) (95% CI) Active disease (77 99%) Controls (91 100%) Total * Specimens from patients with active disease (class III) were collected at a TB hospital in Bombay, India. These active cases were all verified by positive sputum culture. Control specimens were from either USAFA cadets or rheumatology patients at a university-based clinic in Denver, Colorado, all of whom were clinically free of active TB. 95% confidence interval. prevalence of tuberculous infection, the positive predictive value of the anti-lam IgG test for active TB is 100% (95% CI 87 to 100%) and the negative predictive value is 95% (95% CI 83 to 99%). The high positive predictive value is likely to be due, in part, to the high prevalence of active TB in the test population; i.e., 29 of the 66 subjects (44%) had active TB. Evaluation of Anti-LAM IgG among Patients in a TB Clinic The previous comparison between active TB patients from India and U.S. citizens without active TB is not particularly reflective of a particular population because the prevalences of tuberculous infection are widely divergent between India and the U.S. Thus, we examined the accuracy of anti-lam IgG in the context of various categories of tuberculous infections in a defined U.S. population known to be at increased risk for latent and active TB. Serum assays for anti-lam IgG were performed on 115 individuals, comprised largely of Southeast Asian, Mexican, and Eastern European immigrants, who were being screened for TB and hepatitis B (n 104) or were on directly observed therapy for active TB (n 11) at the Denver Metropolitan Tuberculosis Clinic (Table 2). One of these individuals had only a clinical diagnosis of TB (smear-negative, culture-negative) and thus was excluded from sample analyses. In the remaining 114 individuals, 43 were PPD-negative. Seventy-one individuals were either PPD-positive using a criterion of 5 mm (for class IV patients) or 10 mm induration with the standard 5 tuberculin units (TU) PPD given intradermally and/or had culture-proven TB. Although we have shown various degrees of positive blots (,, ) in Figure 2, the vast majority of the positive blots were by visual inspection. Densitometry was performed on the blots but there was no correlation between the density and true- or false-positives (data not shown). Of the 43 PPD-negative subjects (class 0), four had an indeterminate immunoblot, defined by a diffuse discoloration of the entire membrane rather than a distinct purple dot indicative of immunoreaction at the site of the adsorbed LAM (see Figure 2). Of the remaining 39 samples with definitive results, only two were positive, resulting in a specificity rate of 95% (95% CI 83 to 99%) (Table 2). Fifty skin test positive individuals were determined no to have active disease (class II) by chest radiographs and, when clinically indicated, by sputum smear and culture. Five of the 50 class II individuals had an indeterminate blot. Of the remaining 45 evaluable class II subjects, three were seropositive for LAM, resulting in a specificity rate of 93% (95% CI 82 to 99%) in this group of asymptomatic PPD-positive individuals (Table 2). A diagnosis of active TB (class III) was made in 13 persons based on positive sputum cultures. Ten of these individuals were on antituberculous treatment for various length of time at the time the serologic tests were performed. Eleven of the 13 patients had a positive anti-lam IgG dot blot (sensitivity 85%; 95% CI 55 to 98%) (Table 2). The two patients with active TB who tested negative for anti-lam IgG were on antituberculous chemotherapy for 178 and 108 d (mean, 143 d) at the time of the testing compared with a mean of 55 d for the 11 patients who were positive for anti-lam IgG (Figure 3). Six patients had smear-negative but culture-proven TB (sensitivity of sputum smear, 46%); five of the six tested positive for anti-lam IgG. Two anti-lam IgG positive individuals originally categorized as class II were subsequently diagnosed to have culture-confirmed active TB within a few weeks of the serologic testing. Neither patient had a recent contact with an active case of TB, and thus we believe that both of these patients had (subclinical) active TB at the time of the seroassay. These two individuals were reclassified to class III. A diagnosis of previous TB (class IV) was made in eight individuals on the basis of previous chest radiographs, historical information, and negative current sputum cultures. Most of these individuals were not previously treated for TB and had chest radiographs consistent with past TB, including fibronodular scarring and calcified granulomas. One of the patients had an indeterminate test result. Of the remaining seven subjects, anti-lam IgG was detected in five (specificity 29% CI 4 to 71%). TABLE 2 PRESENCE OF ANTI-LAM IgG IN A DENVER METROPOLITAN TB CLINIC POPULATION Class* Positive Negative Samples Sensitivity % (95% CI) Specificity % (95% CI) (83 99%) II (82 99%) III (55 98%) IV (4 71%) Total * Class 0: tuberculin skin test negative, defined as 10 mm induration. Class II: tuberculin skin test positive, defined as 10 mm induration. Class III: active tuberculosis. Class IV: tuberculin skin test positive, radiographic evidence of past disease. Samples excluding indeterminate results (indeterminate results were noted in four class 0, five class II, and one class IV subjects; see text and Figure 2. Figure 3. Anti-LAM IgG versus days of antituberculous treatment. In the two patients with active TB with false-negative anti-lam IgG, the days of treatment at the time of serologic testing (mean, 143 d) was longer than in patients with true-positive anti-lam IgG (mean, 55 d).

4 1716 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL Accuracy and Predictive Values of Anti-LAM IgG for Active TB in a High-risk Population To determine the positive and negative predictive values of anti-lam IgG for active TB in a high-risk population, we compared patients with active TB (class III) with those who did not have active disease (class 0, II, and IV) in the Denver Metropolitan Tuberculosis Clinic population. As shown in Table 3, the overall sensitivity of anti-lam IgG for active TB was 85% (11 of 13 patients) and the specificity was 89% (81 of 91 subjects). In this population where the prevalence of active disease was 13% (13 of 104 had active disease), the positive predictive value was 52% (i.e., of 21 subjects with a positive anti-lam IgG, 11 had active disease). In contrast, only two of 83 individuals with a negative anti-lam IgG test in fact had active disease, resulting in a negative predictive value of 98%. The 10 patients with indeterminate results were not included in the analysis of the sensitivity, specificity, and positive and negative predictive values because they could not be classified as either positive or negative with any degree of certainty. Instead, the accuracy of the test, defined as true positive true negative divided by all test results, was calculated to account for the indeterminate results. Thus, the accuracy of the application of the anti-lam IgG assay in the Denver Metropolitan Tuberculosis Clinic population was (11 81)/114 81% (see also Table 3). DISCUSSION TABLE 3 POSITIVE AND NEGATIVE PREDICTIVE VALUES OF ANTI-LAM IgG IN THE DENVER METROPOLITAN TUBERCULOSIS CLINIC POPULATION* Active TB ( ) ( ) Total Anti-LAM IgG ( ) ( ) Total * In the analysis of the sensitivity, specificity, positive and negative predictive values, the 10 indeterminate results were excluded (total, n 104). In the analysis of accuracy, all test results were included (total, n 114). One patient with only a clinical diagnosis (culture-negative) of TB was excluded from the study. Sensitivity 85%; specificity 89%; positive predictive value 52%; negative predictive value 98%; prevalence 13%; accuracy 81%. Accuracy [true ( ) true ( )]/all test [81 11]/114 81%. Class III patients. Class 0, II, and IV subjects. It is estimated that one-third of the world s population is infected with M. tuberculosis with approximately 8 million new cases of active TB and 3 million deaths per year. Such extensive infection of the human population requires a diagnostic test that is capable of discerning an active from a latent infection. The current screening test for TB with the tuberculin skin test is not able to distinguish these two states of infection. In light of the high prevalence and resurgence of TB worldwide, a serologic test that is both accurate and feasible in screening for active cases would be highly desirable, given that the overall sensitivity rate of sputum smear for active pulmonary TB is only about 50%. We have compared the IgG response to the mycobacterial cell wall component LAM in patients with active TB versus uninfected or latently infected subjects. In hospitalized patients with active TB from India, 93% of the patients tested positive for anti-lam IgG. Interestingly, the 7% false-negative rate of the anti-lam IgG is similar to the historically expected false-negative tuberculin skin test rate in patients who have untreated active disease. In contrast, none of the control sera (U.S. subjects comprised of young healthy adults and patients with chronic rheumatological diseases) tested positive, resulting in a specificity rate of 100% (Table 1). However, the positive and negative predictive values of a diagnostic test may be influenced greatly by the prevalence of the disease in the population being studied. We therefore examined a group of individuals, largely comprised of Southeast Asian, Mexican, and Eastern European immigrants, who were being screened or treated for latent and active tuberculous infection. Some of these individuals have had prior bacillus Calmette-Guérin (BCG) vaccination, which may further reduce the specificity of a serologic test for active disease (14). We found that the majority of the PPD-negative individuals (37 of 39 subjects) in this high-risk TB screening group were negative for anti-lam IgG (specificity 95%); i.e., two class 0 individuals had falsepositive anti-lam IgG tests. An alternative explanation is that the tuberculin skin tests of these two subjects may have been false-negatives because booster testing is not routinely performed with immigrant TB screening. A third possibility is that they may have been infected with mycobacteria other than M. tuberculosis (MOTT) although this would appear less likely given that there was no clinical evidence of disseminated or localized pulmonary disease. Among 45 class II persons (PPD-positive, no active disease) with an evaluable immunoblot, only three (7%) were positive for anti-lam IgG. The true prevalence of prior tuberculous infection in the class II individuals is not known because many of these individuals were previously immunized with BCG, although BCG does not usually result in a positive PPD. In addition, some of the individuals with false-positive anti-lam IgG may have early-stage pulmonary TB, as evidenced by the reclassification of two patients from class II to III in our series. Our experience with these two patients with subclinical active TB would suggest that further studies of individuals who are found to be anti-lam IgG positive should be done to determine whether such individuals have active pulmonary TB, even in the presence of a normal chest radiograph. The specificity of the test fell in class IV patients with a history of prior TB (29%) for reasons that are not certain. Perhaps the bacterial burden in these latently infected patients is higher than in those with normal chest X-rays. Most of these patients had radiographic evidence of prior TB (e.g., fibronodular scarring and calcified granulomas) and yet do not recall having been treated for active TB. Such patients are at higher risk of developing active TB. Because of the high incidence of false-positive test results in the class IV subjects, the positive predictive value of the test was low (52%). In other words, in patients with radiographic evidence of prior TB, a positive anti-lam IgG test is a poor predictor of active disease. Although detection of anti-lam IgM may increase the specificity of the test in class IV patients, the sensitivity is likely to decrease because in general, the primary IgM response to an antigen is approximately 10-fold less than the secondary IgG response. Thus, we have not used an IgM-based assay on account of lower levels and relatively short half-life of IgM. Overall, of the 83 total subjects in the Denver Metropolitan Tuberculosis Clinic who were negative for anti-lam IgG, only two had active TB. Thus, the high negative predictive value (98%) makes anti-lam IgG a potentially attractive screening test for active TB in a susceptible population. In class III (active disease) patients from the Denver Metropolitan Tuberculosis Clinic population, the sensitivity of anti-lam IgG was 85%. More importantly, five of the six patients with active TB who where smear-negative were positive

5 Chan, Reves, Belisle, et al.: Serologic Diagnosis of TB 1717 for anti-lam IgG, making this serologic test more sensitive than sputum smear in our sample population. Two subjects with active TB were negative for anti-lam IgG. It is known that humoral or cellular immune response to foreign antigens can vary among individuals and therefore some patients may not develop a detectable antibody titer to LAM. Furthermore, confounding variables such as an underlying illness (e.g., cancer, diabetes mellitus, malnutrition, AIDS), medication use (e.g., corticosteroids, cytotoxic agents), or even TB itself may impair the humoral and/or cellular immune response. A release of a relatively large antigenic load due to rapid antibiotic killing of bacilli, resulting in antigen antibody complexes, might also reduce the amount of free antibody available. In support of this hypothesis, when the two false-negative sera were subjected to low ph, which dissociates antigen antibody complexes, and then retested for anti-lam IgG, seropositivity against LAM was observed in both (S. Nochur, DynaGen Inc., Cambridge, MA, personal communication). A similar phenomenon was also observed by Gupta and colleagues (15), who showed that antibody levels to antigen 60 were lower in patients who were receiving treatment for TB versus those measured before treatment. Nevertheless, the overall detection of active TB based on the presence of anti-lam IgG was substantially more sensitive than results usually obtained from sputum smears. Although we did not examine patients coinfected with both human immunodeficiency virus (HIV) and TB with this dot-blot assay, we predict that the antibody response rate for LAM and other mycobacterial antigens is likely to be lower in patients co-infected with HIV (16 22). Of the 114 patients in Denver Metropolitan Health population, 10 had indeterminate results: 4/43 in class 0, 5/50 in class II, and 1/8 in class IV subjects. Despite these indeterminate results, we found the accuracy of the test to be 81%. Previously reported serologic studies have used either a mixture of tuberculous antigens such as purified extracted glycolipids (23), adsorbed mycobacterial sonicates (24), an antigen mixture from immunoabsorbent affinity chromatography (25), PPD (26), or more specific mycobacterial antigens such as antigen 5, antigen 60, a 30-kD antigen, P32 antigen (derived from PPD), trehalose-6,6 -dimycolate (cord factor), an 88-kD antigen, Kp90 antigen, and LAM (16 20, 22, 27 42). The sensitivities and specificities obtained from these assays have been highly variable. Mycobacterial antigen 5, also known as the 38-kD antigen, has been evaluated by ELISA with variable sensitivity (45 to 90%) and high specificity (90 to 100%) (25, 27, 29, 30, 43, 44). The sensitivity rate for antigen 5 appeared to be lowest in patients with smear-negative TB and highest in those who were smear-positive, suggesting that a higher burden of organisms is more associated with seropositivity (27, 32). Antigen 60, a thermostable component of PPD, has also been used in the serodiagnosis of TB although this protein is also present in Nocardia and Corynebacterium species (15, 31, 33 36). The sensitivity rate of antigen 60 for adults with active pulmonary TB is approximately 60 to 89%, with the rate lower for children and for patients with extrapulmonary TB (33, 35 37). The specificity of anti-antigen 60 for healthy individuals who had no or latent infection (classes 0 and II) was 82 to 100% (37). Testing against a combination of antigen 5, antigen 60, and another mycobacterial antigen Kp- 90 does not appear to increase the accuracy over each of the individual tests (30). The detection of anti-lam antibodies has been previously evaluated in various developing countries (Thailand, Mexico, Ghana, Tanzania) and in Europe (Italy, Spain, Switzerland) but to the best of our knowledge, not in the U.S. or Canada (11, 16 20, 42). In general, three important points can be made from these studies. First, the specificity of the test was excellent, ranging from 84 to 100%. Moreover, Julian and coworkers (19) tested 14 patients with MOTT and none had anti-lam IgG detectable although it is not clear how many of these patients were co-infected with HIV. Boggian and coworkers (17) also tested 104 patients with MOTT for anti-lam IgG and only two patients were weakly positive although all these patients were HIV-positive. Although the basic structure of LAM associated with M. tuberculosis is similar to that of M. avium complex (the predominant MOTT), the fine structural differences between the LAMs (e.g., degree and configuration of mannose-capping) have not been fully elucidated. Therefore, the degree of antibody cross-reactivity against LAM of M. tuberculosis versus that of MOTT is not known. Second, there was a wide range of sensitivity of anti-lam IgG among HIVnegative patients with active TB but the sensitivity was generally poor in HIV-positive patients. For example, in HIV-negative patients, the sensitivity ranged from 21.5 to 89% (16, 18 20, 42) but was between 7% and 40% in HIV-positive patients (16 20). In HIV-negative patients, the sensitivity of the MycoDot anti-lam IgG test in Tanzania, Ghana, and Thailand was 33%, 56%, and 63.2%, respectively (16, 18, 20). In contrast, the sensitivity was 89% in Italy (42) and in Spain, the overall sensitivity was only 21.5% although in relapse cases, the sensitivity was 100% (19). How these differences in the sensitivity of anti-lam IgG between the developing and European countries are related to the immune and nutritional status of the HIV-negative subjects is not known. Moreover, the variability in the sensitivity of the MycoDot test among these studies and in comparison to our assay may be due, in part, to differences in the technique because we have found that incubation of the sera for longer periods (40 to 60 min versus 20 min for the MycoDot) and with constant agitation increases the yield of the test. Third, as with most other types of serologic tests, the sensitivity of the test decreased with smear-negativity, a finding generally attributed to lower burden of organisms in smear-negative cases and/or to the greater incidence of smear-negativity in patients co-infected with HIV. However, in our HIV-negative population, five of the six smear-negative, culture-positive patients were positive for anti- LAM IgG. In summary, we have shown that anti-lam IgG is a sensitive marker of active TB in a U.S. population considered to be at increased risk for tuberculous infection. Using sputum culture as the gold standard, anti-lam IgG was positive in five of six patients with active TB who were smear-negative. Furthermore, in individuals with no or with latent infection, defined by negative and positive tuberculin skin tests, respectively, anti-lam IgG was quite specific for active disease. The high negative predictive value of anti-lam IgG in a population at risk for tuberculous infection makes it a potentially valuable screening test for active TB. However, in patients with radiographic evidence of prior TB but who were not treated, a positive anti-lam IgG appears to be a poor predictor of active disease. Although the cost of manufacturing such an assay would vary, we have estimated the cost (minus labor) of the dot-blot assay to be approximately 10 cents per blot. Lastly, the assay can be performed without sophisticated instrumentation with minimal training. Acknowledgment : The authors thank Gary Moore, Fazal Raheman, and Saraswarthy Nochur for expert technical assistance and Drs. Vincent Cottin, Carolyn Welsh, and Michael Iseman for critical review of the manuscript. References 1. Schluger, N. W., and W. N. Rom Current approaches to the diag-

6 1718 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL nosis of active pulmonary tuberculosis. Am. J. Respir. Crit. Care Med. 149: Kramer, F., T. Modilevsky, A. Waliany, J. Leedom, and P. Barnes Delayed diagnosis of tuberculosis in patients with human immunodeficiency virus infection. Am. J. Med. 89: Burman, W. J., B. L. Stone, R. R. Reves, M. L. Wilson, Z. Yang, H. El- Hajj, J. H. Bates, and M. D. Cave The incidence of false-positive cultures for Mycobacterium tuberculosis. Am. J. Respir. Crit. Care Med. 155: World Health Organization/Global Tuberculosis Programme: Product Development Guideline. WHO Tuberculosis Diagnostics Workshop. World Health Organization, Cleveland, OH Pang, J. A., H. S. Chan, C. Y. Chan, S. W. Cheung, and G. L. French A tuberculostearic acid assay in the diagnosis of sputum smearnegative pulmonary tuberculosis: a prospective study of bronchoscopic aspirate and lavage specimens. Ann. Intern. Med. 111: Dalovisio, J. R., S. Montenegro-James, S. A. Kemmerly, C. F. Genre, R. Chambers, D. Greer, G. A. Pankey, D. M. Failla, K. G. Haydel, L. Hutchinson, M. F. Lindley, B. M. Nunez, A. Praba, K. D. Eisenach, and E. S. Cooper Comparison of the amplified Mycobacterium tuberculosis (MTB) direct test, amplicor MTB PCR, and IS6110-PCR for detection of MTB in respiratory specimens. Clin. Infect. Dis. 23: Schluger, N. W., D. Kinney, T. J. Harkin, and W. N. Rom Clinical utility of the polymerase chain reaction in the diagnosis of infections due to Mycobacterium tuberculosis. Chest 105: Aguado, J. M., M. J. Rebollo, E. Palenque, and L. Folgueria Bloodbased PCR assay to detect pulmonary tuberculosis. Lancet 347: Steele, B. A., and T. M. Daniel Evaluation of the potential role of serodiagnosis of tuberculosis in a clinic in Bolivia by decision analysis. Am. Rev. Respir. Dis. 143: Wilkins, E. G. L., and J. Ivanyi Potential value of serology for diagnosis of extrapulmonary tuberculosis. Lancet 336: Sada, E., P. J. Brennan, T. Herrera, and M. Torres Evaluation of lipoarabinomannan for the serological diagnosis of tuberculosis. J. Clin. Microbiol. 28: Chatterjee, D., K. Lowell, B. Rivoire, M. R. McNeil, and P. J. Brennan Lipoarabinomannan of Mycobacterium tuberculosis. J. Biol. Chem. 267: Leunissen, S. L., and J. DeMey Preparation of gold probes. In A. J. Verkleij and J. L. Leunissen, editor. Immuno-gold Labeling in Cell Biology. CRC Press, Boca Raton, FL. 14. Turneer, M., J. P. Van Vooren, J. Nyabenda, F. Legros, A. Lecomte, J. Thiriaux, E. Serruys, and J. C. Yernault The humoral immune response after BCG vaccination in humans: consequences for the serodiagnosis of tuberculosis. Eur. Respir. J. 1: Gupta, S., S. Kumari, J. N. Banwalikar, and S. K. Gupta Diagnostic utility of the estimation of mycobacterial antigen A60 specific immunoglobulins IgM, IgA, and IgG in the sera of cases of adult human tuberculosis. Tuberc. Lung Dis. 76: Lawn, S. D., E. H. Frimpong, and E. Nyarko Evaluation of a commercial immunodiagnostic kit incorporating lipoarabinomannan in the serodiagnosis of pulmonary tuberculosis in Ghana. Trop. Med. Int. Health 2: Boggian, K., W. Fierz, and P. L. Vernazza Infrequent detection of lipoarabinomannan antibodies in human immunodeficiency virus associated mycobacterial disease. J. Clin. Microbiol. 34: Ratanasuwan, W., J. K. Kreiss, C. M. Nolan, B. A. Schaeffler, S. Suwanagool, S. Tunsupasawasdikul, C. Chuchottaworn, W. Dejsomritrutai, and H. M. Foy Evaluation of the MycoDot test for the diagnosis of tuberculosis in HIV seropositive and seronegative patients. Int. J. Tuberc. Lung Dis. 1: Julian, E., L. Matas, V. Ausina, and M. Luquin Detection of lipoarabinomannan antibodies in patients with newly acquired tuberculosis and patients with relapse tuberculosis. J. Clin. Microbiol. 35: Somi, G. R., R. J. O Brien, G. S. Mfinanga, and Y. A. Ipuge Evaluation of the MycoDot test in patients with suspected tuberculosis in a field setting in Tanzania. Int. J. Tuberc. Lung Dis. 3: Daniel, T. M., A. A. Sippola, A. Okwera, S. Kabengera, E. Hatanga, T. Aisu, S. Nyole, F. Byekwaso, M. Vjecha, L. E. Ferguson, P. Kataaha, and R. D. Mugerwa Reduced sensitivity of tuberculosis serodiagnosis in patients with AIDS in Uganda. Tuberc. Lung Dis. 75: Laal, S., K. M. Samanich, M. G. Sonnenberg, J. T. Belisle, J. O Leary, M. S. Simberkoff, and S. Zolla-Pazner Surrogate marker of preclinical tuberculosis in human immunodeficiency virus infection: antibodies to an 88-kDa secreted antigen of Mycobacterium tuberculosis. J. Infect. Dis. 176: Escamilla, L., R. Mancilla, W. Glender, and L.-M. Lopez-Marin Mycobacterium fortuitum glycolipids for the serodiagnosis of pulmonary tuberculosis. Am. J. Respir. Crit. Care Med. 154: Rosen, E. U The diagnostic value of an enzyme-linked immunosorbent assay using adsorbed mycobacterial sonicates in children. Tubercle 71: Daniel, T. M., G. L. de Murillo, J. A. Sawyer, A. M. Griffin, E. Pinto, S. M. Debanne, P. Espinosa, and E. Cespedes Field evaluation of enzyme-linked immunosorbent assay for the serodiagnosis of tuberculosis. Am. Rev. Respir. Dis. 134: Barrera, L., I. Miceli, V. Ritacco, G. Torrea, B. Broglia, R. Botta, C. P. Maldonado, N. Ferrero, A. Pinasco, I. Cutillo, M. Cornejo, E. Prokopio, and I. De Kantor Detection of circulating antibodies to purified protein derivative by enzyme-linked immunosorbent assay: its potential for the rapid diagnosis of tuberculosis. Pediatr. Infect. Dis. J. 8: Chan, S. L., Z. Reggiardo, T. M. Daniel, D. J. Girling, and D. A. Mitchison Serodiagnosis of tuberculosis using an ELISA with antigen 5 and a hemagglutination assay with glycolipid antigens. Am. Rev. Respir. Dis. 142: Jackett, P. S., G. H. Bothamley, H. V. Batra, A. Mistry, D. B. Young, and J. Ivanyi Specificity of antibodies to immunodominant mycobacterial antigens in pulmonary tuberculosis. J. Clin. Microbiol. 26: Alde, S. L. M., H. M. Pinasco, F. R. Pelosi, H. F. Budani, O. H. Palma- Beltran, and L. J. Gonzalez-Montaner Evaluation of an enzyme-linked immunosorbent assay (ELISA) using an IgG antibody to Mycobacterium tuberculosis antigen 5 in the diagnosis of active tuberculosis in children. Am. Rev. Respir. Dis. 139: Chiang, I.-H., J. Suo, K.-J. Bai, T.-P. Lin, K.-T. Luh, C.-J. Yu, and P.-C. Yang Serodiagnosis of tuberculosis: a study comparing three specific mycobacterial antigens. Am. J. Respir. Crit. Care Med. 156: Cocito, C. G Properties of the mycobacterial antigen complex A60 and its applications to the diagnosis and prognosis of tuberculosis. Chest 100: Cole, R. A., H. M. Lu, Y. A. Shi, J. Wang, T. De-Hua, and A. T. Zhou Clinical evaluation of a rapid immunochromatographic assay based on the 38 kda antigen of Mycobacterium tuberculosis on patients with pulmonary tuberculosis in China. Tuberc. Lung Dis. 77: Charpin, D., H. Herbault, M. J. Gevaudan, M. Saadjian, P. de Micco, A. Arnaud, D. Vervloet, and J. Charpin Value of ELISA using A60 antigen in the diagnosis of active pulmonary tuberculosis. Am. Rev. Respir. Dis. 142: Alifano, M., R. De Pascalis, M. Sofia, S. Faraone, M. Del Pezzo, and I. Covelli Evaluation of IgA-mediated humoral immune response against the mycobacterial antigen P-90 in diagnosis of pulmonary tuberculosis. Chest 111: Turneer, M., E. Van Nerom, J. Nyabenda, A. Waelbroeck, A. Duvivier, and M. Toppet Determination of humoral immunoglobulins M and G directed against mycobacterial antigen 60 failed to diagnose primary tuberculosis and mycobacterial adenitis in children. Am. J. Respir. Crit. Care Med. 150: Delacourt, C., J. Gobin, J.-L. Gaillard, J. de Blic, M. Veran, and P. Scheinmann Value of ELISA using antigen 60 for the diagnosis of tuberculosis in children. Chest 104: Zou, Y. L., J. D. Zhang, M. H. Chen, G. Q. Shi, J. Prignot, and C. Cocito Serological analysis of pulmonary and extrapulmonary tuberculosis with enzyme-linked immunosorbent assays for anti-a60 immunoglobulins. Clin. Infect. Dis. 19: Turneer, M., J.-P. Van Vooren, J. De Bruyn, E. Serruys, P. Dierckx, and J.-C. Yernault Humoral immune response in human tuberculosis: immunoglobulins G, A, and M directed against the purified P32 protein antigen of Mycobacterium bovis bacillus Calmette-Guérin. J. Clin. Microbiol. 26: Maekura, R., M. Nakagawa, Y. Nakamura, T. Hiraga, Y. Yamamura, M. Ito, E. Ueda, S. Yano, H. He, S. Oka, K. Kashima, and I. Yano Clinical evaluation of rapid serodiagnosis of pulmonary tuberculosis by ELISA with cord factor (trehalose-6,6 -dimycolate) as antigen purified from Mycobacterium tuberculosis. Am. Rev. Respir. Dis. 148: Huygen, K., J. P. Van Vooren, M. Turneer, R. Bosmans, P. Dierckx, and J. de Bruyn Specific lymphoproliferation, interferon production, and serum immunoglobulin G directed against a purified 32 kda

7 Chan, Reves, Belisle, et al.: Serologic Diagnosis of TB 1719 mycobacterial protein antigen (P32) in patients with active tuberculosis. Scand. J. Immunol. 27: Arikan, S., S. Tuncer, D. Us, S. Unal, and S. Ustacelebi Anti-Kp 90 IgA antibodies in the diagnosis of active tuberculosis. Chest 114: Del Prete, R., V. Picca, A. Mosca, M. D Alagni, and G. Miragliotta Detection of anti-lipoarabinomannan antibodies for the diagnosis of active tuberculosis. Int. J. Tuberc. Lung Dis. 2: Ma, Y., Y. M. Wang, and T. M. Daniel Enzyme-linked immunosorbent assay using Mycobacterium tuberculosis antigen 5 for the diagnosis of pulmonary tuberculosis in China. Am. Rev. Respir. Dis. 134: Daniel, T. M., S. M. Debanne, and F. van der Kuyp Enzymelinked immunosorbent assay using Mycobacterium tuberculosis antigen 5 and PPD for the serodiagnosis of tuberculosis. Chest 88: