Serodiagnosis of tuberculosis and leprosy by enzyme immunoassay

INTRODUCTION

Tuberculosis is currently very highly rated as an important infectious disease worldwide. In countries with a significant HIV problem it occurs in AIDS, posing a very serious medical problem [1]. There is therefore a great need for accurate and rapid laboratory diagnosis. The conventional methods of direct sputum examination and bacteriologic culture used for laboratory diagnosis of pulmonary tuberculosis are simple but insensitive or slow. They are generally unhelpful in the diagnosis of extrapulmonary tuberculosis when the focus of the disease is unknown or is difficult to access for the collection of laboratory specimens. Newer techniques for the rapid diagnosis and identification of mycobacterial species, such as polymerase chain reaction (PCR), ligase chain reaction (LCR) and several other molecular methods, have recently come into use and have greatly shortened the time needed for diagnosis of different types of pulmonary mycobacterioses. These techniques are, however, of little value in the diagnosis of extra-pulmonary disease, and serology would be considered a more appropriate approach to diagnosis.

However, as there is a great diversity of antigens and antibodies to mycobacteria, there has not been any acceptable standard serologic test for the diagnosis of tuberculosis. Recently, a specific antigen of 38 kDa has been identified [1] and has been expressed as a recombinant in Escherichia coli. Two kits are commercially available and utilize an enzyme immunoassay (EIA) for IgG antibodies [2]. One of the sets is Pathozyme TB-Complex for detection of antibodies to Mycobacterium tuberculosis complex, including M. tuberculosis, M. bovis and M. africanum, and the other is Pathozyme-Myco for detection of antibodies to various Mycobacterium species. This test has produced some promising results in a study in Pakistan of patients with smear- and culture-positive tuberculosis, yielding sensitivity and specificity rates of 79% and 86% respectively (Omega: unpublished, in-house data on serologic evaluation of the kit in Pakistan patients). This study was conducted to evaluate the usefulness of this diagnostic kit in various forms of tuberculosis and leprosy in the United Arab Emirates.

PATIENTS AND METHODS

The study was carried out by two laboratories in the cities of Dubai and Al Ain, the United Arab Emirates (UAE), between April 1995 and June 1996. The study was conducted on serum specimens received in the laboratory on the initial visits of the reported patients before they were fully diagnosed. Some of the patients had sputum or other laboratory specimens for mycobacterial smears and culture, but some patients (suspects) had only clinical and radiologic evidence for the suspected diagnosis.

Sera were obtained from a total of 247 persons, comprising 58 patients who were culture positive for M. tuberculosis complex or had acid- and alcohol-fast bacilli (AAFB) visualized in stained preparations, six proven but recently treated cases, 11 proven leprosy cases, 137 cases clinically or radiologically suspected but negative on culture and or microscopy, or from whom specimens for bacteriology were unavailable, and 35 controls who were either healthy individuals or patients with diseases not suspected to be tuberculosis. All patients and controls were adults, except for 13 children with proven extrapulmonary infections by M. tuberculosis (including BCG). Sera were tested in batches at convenient intervals. The EIA test was performed in parallel and in duplicate, according to the manufacturer's protocol, using the two antigens Pathozyme TB-Complex for M. tuberculosis complex IgG and Pathozyme-Myco for other Mycobacterium species. These kits were kindly supplied by Omega Diagnostics Ltd, Alloa, Scotland. Serum was diluted 1/100 with the serum diluent, and the kit components were brought to room temperature before the start of the procedure. Each serum and three controls were tested in duplicate. One hundred microliters of the diluted serum and control sera were dispensed into appropriate microtiter wells coated with mycobacterial antigen. The plates were sealed and incubated at 37°C for 60 min. The plates were then washed three times, automatically with the kit's diluted wash buffer. They were dried by shaking and inverting them on absorbent paper, taking biohazard precautions. To each well 100 μL of conjugate was added, and the plates were covered, shaken gently for 15 s and then reincubated for another 30 min. The wash procedure was repeated, and 100 μL of 3,3′,5,5′-tetramethylbenzidine (TMB) substrate was then added. After gentle shaking the plate was reincubated for 15 min. The reaction was stopped by addition of 100 μL stop solution to each well. A change from blue to yellow indicated the presence of antibodies.

The ELISA plate reader was blanked on air and the absorbence immediately read at 450 nm. The EIA absorbency was recorded and the absorbency figure of the test specimen was divided by that of the low positive control and the result expressed as an EIA index. An EIA index of <0.99 was considered negative, 1.0–1.5 as equivocal, and equal or >1.6 as positive. The two parallel results recorded for Pathozyme TB-Complex and Pathozyme-Myco were compared and the higher reading of the two was used as the final diagnostic value. Chi-squared analysis was performed to test the differences in the different groups and clinical situations.

RESULTS

Table 1 shows the EIA indices obtained in the two tests for the 247 subjects studied. A comparison of the two tests shows that the Pathozyme-Myco test produced more positives and showed higher levels of antibodies than the Pathozyme TB-Complex. Among the 58 patients with tuberculosis, 48 showed antibodies in both tests, and Pathozyme-Myco identified 47 of these, while Pathozyme TB-Complex identified only 27. Among the 83 people with an EIA index >1.6, Pathozyme-Myco was positive in 81, against only 20 with Pathozyme TB-Complex. All 11 leprosy cases were positive with Pathozyme-Myco, while only six were positive in Pathozyme TB-Complex.

Table 2 summarizes the differences between different conditions with positive results from either test. In the control group of 35 normal subjects or patients not suspected to have tuberculosis, there was none with an antibody level of 1.6 or more, while there were four equivocal positive results, three of which were in the Pathozyme-Myco test. On the other hand, all 23 patients with laboratory evidence of pulmonary tuberculosis had antibody levels of >1.0–1.5, while 22 (95.7%) had antibody levels of >1.6. The difference between the two groups is statistically significant at the two levels of EIA positivity. Twenty-five (71.4%) of 35 cases with laboratory evidence of extrapulmonary infection with M. tuberculosis complex were positive at one or both levels, but only 18 (51.4%) were unequivocal. Five of the seven equivocal results were in children, with tuberculous meningitis, cold abscess, cryptic marrow infection and lymphadenitis. There were 10 patients who were negative. These included mainly children with BCG abscesses and some adults with urinary tract problems and doubtful AAFBs seen on microscopy. Ten of 11 cases with leprosy showed antibody levels of 1.0–1.5, while nine had unequivocal results of >1.6. Four of the six previously treated tuberculosis cases had antibodies with EIA of >1.6, one was equivocal, and the other, who had been treated for renal tuberculosis, was negative. Of the 137 suspected tuberculosis cases (without laboratory evidence) tested, 30 (21.9%) had significant antibodies, while the rest were equivocal or negative.

DISCUSSION

The development of a reliable, sensitive and specific serologic test for tuberculosis and other mycobacterial infections has proved difficult. Recent evaluation of ELISA assays based on the A60 antigen has found these lacking [3]. The discovery of the 38-kDa lipoprotein antigen, which is more specific, has given new hope in serodiagnosis [4].

Using one of these 38-kDa antigens, the most recent commercially developed kits (Omega diagnostics), for the diagnosis of all types of mycobacterial diseases, the data obtained show that there is not, as yet, a very sensitive serologic test available. With either of the positive parallel tests and the lower cut-off point, 31 (88.7%) of the control group were negative and only four (11.3%) had equivocal results. On the other hand, all the patients (100%) with evidence of active pulmonary tuberculosis and leprosy were positive and the difference from the control group was highly significant (p<0.001). This type of high sensitivity for smear-positive patients has been demonstrated before [5]. Another recent evaluation reported a similar result in 36 cases of smear-proven or histologically proven extrapulmonary tuberculosis (Omega: unpublished in-house data on serologic evaluation of the kit in Pakistan patients).

The results are, however, different in the patients with laboratory evidence of extrapulmonary infection, where the sensitivity is 71.4% for the 35 cases at the higher cut-off point. The combined results of all patients with laboratory evidence of tuberculous infection give a sensitivity rate of 82.8%, including the equivocal diagnosis, and 62.5% for unequivocal diagnosis. The overall picture, however, is dependent on the proportion of pulmonary to other forms of tuberculosis.

Although antibody detection appears to be more frequent and of higher level in sputum-positive than in sputum-negative patients, factors other than bacterial load per se may play a role and could be of clinical relevance. In view of the intracellular nature of the mycobacteria, and because more seronegative patients are paucibacillary, it seems unlikely that antibody trapping in circulating immune complexes could explain this phenomenon [5].

In the suspected tuberculosis group, a lower positivity rate of nearly 20% and equivocal results of 47.3% were observed. This was significantly higher than in the control group. Although this group was not followed up for the final diagnosis, and response to anti-tuberculous therapy was not recorded, the results of the test might be useful as additional evidence for or against the diagnosis of tuberculosis.

The main serologic problem in other earlier studies was the use of crude reagents containing a great diversity of antigens, which detected antibodies at unacceptably high antibody frequency in healthy subjects. The recently introduced individual antigens, such as 38-kDa protein antigen, provide a contrast, and the main diagnostic obstacle is their lack of detection of low antibody levels in 20–30% of tuberculosis patients. However, this does not seem to be the problem in our study, as low antibody levels were detected in nearly all cases of disease. The 38-kDa antigen has been shown to be the most immunogenic, and now allows specific antibodies to mycobacteria to be assayed [6,7]. The serologic immunodominance of this antigen in tuberculosis was recently confirmed by Western blot and various monoclonal antibody tests [8].

The interpretation of the results and setting of the cut off-point are common difficulties in serodiagnosis. Serodiagnosis should be compared to standard chest radiography and sputum microscopy with bacteriologic culture. It is well known that immunoassays have not yet achieved higher levels of sensitivity under conditions of high specificity. We have suggested two levels of EIA index cut-off points, one for definite positive diagnosis, in which there are no false positives, and another level for equivocal results, which is very highly sensitive but may have false positives so that patients with this result need more evaluation. The value of parallel testing with the two antigens maximizes on positivity but might not be cost beneficial in view of the sensitivity of Pathozyme-Myco alone. This is surprising; there is no easy explanation why Pathozyme-Myco is more frequently positive in tuberculosis than Pathozyme TB-Complex.

In view of the epidemiologic importance of infectious pulmonary tuberculosis and its transmission to healthy susceptible contacts, serologic screening of populations at high risk in endemic areas should be considered. This test could also be of some additional value in the diagnosis of extrapulmonary tuberculosis.