Despite the increasing development of techniques of rapid identification of mycobacteria by molecular genetic means, there is a definite need for a simple, sensitive, and specific test for tuberculosis (TB), which would improve or replace the sputum smear (1). The development of an easy and inexpensive diagnostic tool for this disease is an important goal, particularly in the view of a global increase in the number of cases of active TB, primarily affecting the developing world (1, 2). Accordingly, much effort has been put into the creation of a rapid and accurate test for the diagnosis of TB. However, methods based on molecular biology are costly and complicated, so that they are not useful for the routine diagnosis in low-income countries (3, 4, 5).

The demand for a more rapid diagnostic procedure has repeatedly focused on the possibility of demonstrating specific antibodies. Due to the often chronic character of TB, there is also a need for the test reflecting the activity and course of the disease. First attempts to establish serological methods for the diagnosis of TB were made about a hundred years ago, and since then a considerable number of such tests have been described (6). Those trials were unsuccessful due to low specificity of tests. The earliest assays were based on a crude antigenic preparation derived from the Mycobacterium Tuberculosis culture. The preparation consisted of a mixture of antigens common to all mycobacteria and also to other bacterial genera, such as Nocardia, Corynebacterium, etc. (6). Non-specific or false positive reactions in non-TB subjects resulted from cross-reactions with antigens present in the mycobacteria environment. Specificity of the test improved when highly purified and recombinant antigens specific for the TB complex have been introduced (7, 8, 9, 10). A relatively low sensitivity of the old tests used for the diagnosis of TB was also an important problem. This difficulty was overcome by the introduction of an Enzyme-linked immunosorbent assay (ELISA) that is a highly sensitive and reproducible serological technique. The method does not require sophisticated instrumentation, is simple, and the reagents that it employs are inexpensive, so that the ELISA can be performed in every laboratory. The use of the ELISA method with highly specific antigens is the best option for TB serodiagnosis (11, 12, 13). Serological tests may be especially useful for a rapid diagnosis of TB in low-income countries, which shoulder over 90% of the global burden of TB cases (1).

Poland is a country with moderate prevalence of TB (27.6 cases per 100000 persons) (1). The objective of this work was to assess the clinical usefulness of various commercially available immunoassays in the diagnosis of TB in Poland.


MATERIAL AND METHODS
Humoral immune responses to selected mycobacterial antigens were analyzed in a group of 319 TB patients (289 adults and 30 children) and in 254 control subjects (Table 1). All patients and control subjects gave informed consent and the study protocol was approved by the Ethics Committee of the National Institute of Tuberculosis and Lung Diseases in Warsaw, Poland. All subjects were Caucasian, HIV negative, and previously BCG vaccinated. Among the TB patients there were 267 cases of pulmonary TB and 52 cases of extrapulmonary TB (bone and joint, urinary tract, central nervous system, pericardial). In the TB group, there were 211 new cases and 108 chronic cases. The diagnosis of TB was proved by bacteriological examination in 223 cases and established by clinical and radiological criteria and by the response to treatment in 96 cases. Childhood TB was diagnosed on the basis of epidemiological considerations, clinical symptoms and signs, radiological examination, bronchoscopy, a tuberculin skin test, and a culture of bronchial or gastric lavage. Bacteriological confirmation was obtained in 24 cases. Samples from children without TB (asthma, pollinosis, nonspecific pulmonary infections, and healthy children) were used as control.

Table 1. Characteristics of subjects (age is given in years).

Serum samples were collected and stored at -40°C until further use. Immunoenzymatic tests measuring IgG against the 38kDa, IgA, IgG, and IgM against the 38kDa and lipoarabinomannan (LAM), IgG against the 38kDa and 16kDa, and IgG against the A60 were used. The antigens 38kDa and 16kDa were recombinant antigens specific for the Mycobacterium Tuberculosis complex. A60 was a composite molecule derived from tuberculin, containing lipids, polysaccharides, and proteins found in all mycobacteria species. LAM was a component of the bacterial wall present in all mycobacteria (12, 14, 15, 16). The tests were based on a solid double-antibody sandwich ELISA. Sera were diluted 1:50 or 1:100 and were added to micro wells precoated with antigens. All samples were assayed in duplicates. In the positive cases, an antigen-antibody complex reacted with peroxidase-labeled antihuman IgG conjugate. Using H2O2/TMB as substrate, the enzymatic activity of peroxidase was measured at 450 nm with the use of an automated reading system. All results were referred to a standard curve. The range of standards was provided for the generation of a semi-logarithmic reference curve. As the sera were diluted 1:50, the units extrapolated from the standard curve were multiplied by 50 (dilution 1:50) or by 100 (dilution 1:100) to obtain serounits for the interpretation of results.

Sensitivity, specificity, and negative and positive predictive values were calculated by standard methods (11). The cut off value was established by the receiver operating characteristic (ROC) curve method (17). To draw a ROC curve, the true positive fraction (sensitivity) and the false positive fraction (1 minus specificity) were calculated at different cut off values and plotted in a graph. A ROC curve describes probabilities of correct and incorrect results at different cut off values. The area under ROC curve reflects the accuracy of a test. The point that encloses the largest area of the graph is regarded as the most accurate cut off value. This point is the closest to the left-hand border and the top border of the ROC area (17).


RESULTS
The results showed a high specificity and a much lower sensitivity of serological tests in the adult population. Sensitivity, specificity, and the positive predictive value (PPV) and negative predictive value (NPV) of the examined tests were presented in Table 2. Tests using a mixture of antigens (A60 and IgA anti 38-kDa + LAM) were more sensitive but less specific than the tests based solely on recombinant proteins (38-kDa + 16-kDa). A good level of sensitivity was obtained for the IgG and IgA but not IgM class (Table 2).

Table 2. Characteristics of the examined tests, as combined for pulmonary and extrapulmonary tuberculosis in adults.

In pulmonary TB in adults, all the tests were more sensitive in culture positive (BK+) than in culture negative (BK-) TB and also in chronic cases (CHR) than in new cases (NEW) or in the patients with recurrent TB (REC) (Table 3). In extrapulmonary TB in adults, the sensitivity of IgG tests was comparable or slightly lower than that in pulmonary TB. The sensitivity of the IgA and IgM-based assays was very low (Table 4). IgG and IgA anti 38-kDa + LAM strongly correlated with each other in adults with TB (Fig. 1). There was no correlation between IgM and other immunoglobulin classes (data not shown).

Table 3. Sensitivity of the examined tests in adult patients with pulmonary TB.

Table 4. Sensitivity of the examined tests in adult patients with extrapulmonary TB.

TOTAL-all extrapulmonary patients, B & J-bone and joint TB, KIDN-urinary tract TB, NODES-lymphatic nodes TB, PLEURA-pleural TB.

Fig. 1. Correlation between IgG and IgA in adults with tuberculosis.

In children, the whole group sensitivity of the examined tests was significantly lower than that in adults, but the specificity was comparable (Table 5). Less than 15% of young children (0-9 years) had detectable IgG and IgA. The sensitivity of the IgM test in this group was 24%. In the adolescent group (11-18 years), the sensitivity was higher than that in the young children but still lower than in adults. The sensitivity of the examined tests in postprimary TB in children was similar to that in adults (Table 6).

Table 5. Characteristics of the examined tests in children under 18 years of age.

Table 6. Sensitivity of the examined tests in subgroups of children with tuberculosis.

TOTAL-all children with TB, <10YR-children under 10 years of age, >10YR-children over 10 years of age, NOD-hilar nodes TB, NOD/PUL-hilar nodes TB with pulmonary involvement, POSTP-postprimary TB in children.

DISCUSSION
The diagnosis of TB is based on clinical and radiological indices and on the identification of mycobacteria in secretions or in affected tissues (1). However, all these methods have limitations. Clinical and radiological symptoms and signs are unspecific and may mimic a large number of diseases like lung cancer, non-specific bacterial infections, fungal infections, interstitial lung diseases, vasculitis, and a large panel of extrapulmonary disorders.

The examination of a sputum smear is helpful and has a high positive predictive value. However, culture positive pulmonary TB is found in our population only in about 50% of patients (1). Furthermore, to achieve a good sensitivity of this method the examination of the sputum smear ought to be repeated several times, so that numerous visits of the patients are required. The critical point is also the correct reading of a slide depending on a high microscopic standard and experienced laboratory staff. Bacteriological culture is much more sensitive than microscopy and is also more specific, as it allows Mycobacterium TB to be distinguished from other mycobacteria that cause human disease. However, conventional culture methods are time-consuming and modern radiometric systems are expensive and require sophisticated laboratory equipment (1).

WHO estimates that about 1/3 of TB patients has extrapulmonary TB (2). In this form of the disease clinical findings and radiological presentations are very unspecific, smear and culture methods have low sensitivity, and invasive procedures are often required to obtain specimens for bacteriological and histological examinations (3, 4, 6, 18). The examination of a clinical sample by molecular biological means is a valuable tool but the expense of the equipment limits its usefulness (5). Although the outcome of the disease is related to the duration of illness before the initiation of therapy, rapid diagnosis is of great importance. Currently, in developing countries the diagnosis of extrapulmonary TB is based mainly on clinical and radiological findings. The literature about the usefulness of serological tests in extrapulmonary TB is scarce (12, 13).

The diagnosis of childhood TB is also difficult. It relies on the tuberculin skin test, chest radiographs, clinical symptoms and signs, and on epidemiological considerations (19, 20). Children rarely produce sputum and the rate of culture confirmation is very low. Better techniques would especially benefit young children and infants in whom early diagnosis is imperative for preventing progressive TB (19, 20, 21). There have been many attempts to develop reliable serodiagnostic tests for pulmonary TB, but their place in the routine diagnosis of TB in adults and children has not been established (7, 8, 9, 10).

The goal of the present work was to evaluate diagnostic usefulness of several serological tests in different forms of pulmonary and extrapulmonary TB in adults and children. Diagnostic accuracy of a test depended on the antigen used and on the population examined. For both adults and children, specificity of IgG assays based on recombinant antigens was very high (97-99%). The specificity of assays based on native antigens was lower, which may have resulted from the cross-reactivity of native antigens with environmental mycobacteria (6). At a chosen specificity, the sensitivity of IgG tests in adults was around 50%. It was slightly higher in native antigen tests then in recombinant antigen assays. However, taking into consideration a decreased specificity of LAM and A60-based tests, diagnostic accuracy measured by PPV and NPV was better for recombinant IgG based assays. According to the literature, the 38-kDa antigen provides the serodiagnostic test with the most favorable characteristics described to-date (14, 22). The combinatory use of the 38-kDa and 16-kDa may increase the test sensitivity compared with the 38-kDa alone (8). A high PPV indicates that positive results of tests can be used to confirm the diagnosis. Our results confirmed those observations in the Polish population. A significantly lower NPV of tests indicates that serology has no value in the exclusion of TB. The sensitivity of the IgA and particularly of the IgM-based assays was significantly lower than that of IgG assays. It means that the IgM and IgA-mediated humoral immune responses against the mycobacterial antigen may be influenced by factors (personal or environmental) that are not related to the presence of active disease (10, 13). The IgM and IgA tests have limited usefulness for the diagnosis of TB in our population. Variability of humoral immune responses to the mycobacterial antigen proves that the antibody response to a particular antigen may not be universal. The findings indicate that the person-to-person heterogeneity of antigen recognition, rather than the recognition of a particular antigen is a key attribute of the antibody response in TB.

The sensitivity of the tests depended also on the phase of the disease and on the presence of mycobacteria in sputum. It was much higher in culture positive and in chronic cases. That could be connected to a higher antigenic load and to the persistency of stimulation of the immune system (6, 23). In chronic and culture positive cases, antigenic stimulation persists and if the patient is genetically able to mount an immune response to the 38-kDa or 16-kDa antigen, antibody levels are expected to be elevated. Most of the serological studies have been done in patients with extensive pulmonary disease, usually smear positive (7, 8, 9). In this group of patients, antigenic load was the highest and the stimulation for antibody production was the strongest. However, the group that would most benefit from serological tests is the one with non-confirmed TB. Unfortunately, in this group of patients the sensitivity of different assays was reported to be significantly lower than in the confirmed cases (4, 23, 24).

Antibody response to the 38-kDa in pulmonary TB has been extensively studied, but there are only few reports about the utility of the 38-kDa-based serological tests in extrapulmonary TB (6, 12, 13, 14). In our study, the sensitivity of the examined tests in extrapulmonary TB was comparable with that in pulmonary TB, except for the IgM-based assay. The IgM-based immune response is usually connected with the early phase of pathogen invasion. In most cases, extrapulmonary TB represents chronic or recurrent form of the disease. That may explain the very low sensitivity of IgM based tests in this form of TB. The data on the use of serology in extrapulmonary TB diagnosis are scarce. Wilkins and Ivanyi (13) have described the competition ELISA assays, based on the TB72 38-kDa monoclonal antibody that displayed a high sensitivity and specificity. The authors detected antibodies to the 38-kDa antigen in 73% of cases of extrapulmonary TB at a chosen specificity of 98%. Stroebel et al. (12) have found that 15 of the 16 patients with active extrapulmonary TB had an elevated level of antibodies against the antigen 6. At a selected cut-off point, they reported the sensitivity of 94% and specificity of 100%. Alifano et al. (15) have detected IgG against the A60 mycobacterial antigen in 73.8% of extrapulmonary TB cases. Zou et al. (16) have reported a 69-86% positive yield for the IgG against A60 in 92 extrapulmonary TB patients. The difference between different reports could likely be explained by genetic and race factors, immunological status of the host, the phase and duration of the disease.

The sensitivity of all examined tests was much lower in the children group than in adults. It strongly depended on the children's age and was particularly low for children under 10 years of age. In adolescents, sensitivity increased and was comparable with that in adults in the postprimary TB. Except for the latter type of TB, the examined tests would have little value in the routine diagnosis of TB in children (19, 20, 21). The data from the literature indicate that none of the available serodiagnostic tests for TB have an adequate sensitivity and specificity under various clinical conditions to be useful for diagnosing TB in children. The reported efficiency of the serodiagnosis with the A60 antigen in children was variable, but rather in a low range (19, 21). Factors, such as the class of the immunoglobulin measured, the groups of patients and controls selected, and the children's age, could partly explain the varied sensitivities and specificities found by different authors. Rosen (21) has used mycobacterial sonicates in an ELISA in samples from 31 TB children and found the sensitivity of 26% and specificity of 40%. Delacourt et al. (19) have used an ELISA to detect the IgG and IgM antibodies directed against the mycobcaterial antigen A60 in children with TB. At a chosen specificity of 98%, IgG was detected in 68% of children with clinical disease; IgM detection had only 19% sensitivity. However, using the same anti-A60 ELISA at a specificity of 95%, Turneer et al. (20) have found the IgG sensitivity to be 26% for past TB, 6% for asymptomatic primary TB, and 14% for symptomatic TB.

The current resurgence of TB worldwide and particularly in developing countries has created a need for cheap and effective diagnostic technique. The X-ray and acid fast stain, the two most common diagnostic methods in the developing world are not specific and can lead to misdiagnosis. The addition of serological assays to the diagnostic arsenal could provide a more rapid and accurate diagnosis in active TB and may be a valuable guide for putting a patient on a correct drug regiment at the earliest possible date. Taking into account the high specificity, serology might be a suitable tool for the first-line testing in resource-poor countries, where access to diagnostic instruments is limited and cost efficiency has high priority.

In conclusion, our results demonstrated different antibody responses to each of the tested antigens and a distinct association of the response with the phase of the disease. The characteristics of the examined tests are better for culture positive and chronic TB cases. The IgG tests based on recombinant mycobacterial antigens (38-kDa and 16-kDa) present an acceptable level of sensitivity and a very good specificity in the diagnosis of TB and can be used in combination with other methods to increase diagnostic accuracy in this disease. The ELISA assays based on IgA and IgM are not helpful in clinical practice. Serological methods may be of value in the diagnosis of extrapulmonary TB that usually represents a chronic form of the disease and is especially difficult to be confirmed bacteriologically. In children, the sensitivity of the examined tests was insufficient to recommend them for the diagnosis of TB.


REFERENCES

1. Zwolska Z. TB in the world and in Poland at the 20th century. Bull Pol Acad Sci, Sec: Biol Sci 1996; 44: 261-271.
2. Farer S, Lowell AM, Meador MP. Extrapulmonary TB in the United States. Am J Epidemiol 1979; 109: 205
3. Ellis ME, el-Ramahi KM, al-Dalaan AN. TB of peripheral joints: a dilemma in diagnosis. Tubr Lung Dis 1993; 74: 399-404.
4. Mondal A. Cytological diagnosis of vertebral TB with fine-needle aspiration biopsy. Am J Bone Joint Surg 1994; 76: 181-184.
5. Kanoudhis M, van der Spel A, Shipley CA et al. Diagnosis of TB of skeletal TB by PCR. Br J Bone Joint Surg 1996; 78 (Suppl 2 and 3): 97.
6. Daniel TM. Immunodiagnosis of TB. In TB, WN Rom, SM Garay (eds). Little Brown, 1996; pp. 223-230.
7. Julian E, Matas L, Hernandez A, Alcaide J, Luquin M. Evaluation of a new serodiagnostic TB test based on immunoglobulin A detection against Kp-90 antigen. Int. J. Tuberc. Lung Dis 2000; 4: 1082-1085.
8. Chan ED, Reves R, Belisle JT, Brennan P, Hahn WE. Diagnosis by visually detectable immunoassay for lipoarabinomannan. Am J Respir Crit Care Med 2000; 161: 1713-1719.
9. Harrington JJ 3rd, Ho JL, Lapa e Silva JR et al. Mycobacterium TB lipid antigens: use of multi-antigen based enzyme immunoassay for free and complex dissociated antibodies. Int J Tuberc Lung Dis 2000; 4: 161-167.
10. Pottumarthy S, Wells VC, Morris AJ. A comparison of seven tests for serological diagnosis of TB. J Clin Microbiol 2000; 38: 2227-2231.
11. Rasolofo V, Rasolonavalona T, Ramarokoto H, Chanteau S. Predictive values of the ICT TB test for routine diagnosis of TB in Madagascar. Int J Tuberc Lung Dis 2000; 4: 184-185.
12. Stroebel AB, Daniel TM, Lau JHK, Leong JCY, Richardson H. Serologic diagnosis of bone and joint TB by an enzyme-linked immunosorbent assay. J Infect Dis 1982; 146: 280-283.
13. Wilkins EG, Ivanyi J. Potential value of serology for diagnosis of extrapulmonary TB. Lancet 1990; 336: 641-644.
14. Harboe M, Wiker HG. The 38-kDa protein of Mycobacterium TB: a review. J Infect Dis 1992; 166: 874-884.
15. Alifano M, Del Pezzo M, Lamberti C, Faraone S, Covelli I. Elisa method for evaluation of anti-A60 IgG in patients with pulmonary and extrapulmonary TB. Microbiologica 1994; 17: 37-44.
16. Zou YL, Zhang JD, Chen MH, Shi GQ, Pringnot J, Cocito C. Serological analysis of pulmonary and extrapulmonary TB with enzyme-linked immunosorbent assays for anti-A60 immunoglobulins. Clin Infect Dis 1994;19: 1084-1091.
17. Altman DG, Bland JM. Statistics notes: Diagnostic tests 3: Receiver operating characteristic plots. Br Med J 1994; 309:188-189.
18. Chierakul N, Damrongchokpipat P, Chaiprasert A, Arjratanakul W. Antibody detection for the diagnosis of tuberculous pleuritis. Int J Tuberc Lung Dis 2001; 5: 968-972.
19. Delacourt C, Gobin J, Gaillard JL et al. Value of ELISA using antigen 60 for the diagnosis of TB in children. Chest 1993; 104: 393-398.
20. Turneer M, Nerom EV, Nyabenda J et al. Determination of humoral immunoglobulin M and G directed against mycobacterial antigen 60 failed to diagnose primary TB and mycobacterial adenitis in children. Am J Respir Crit Care Med 1994; 150: 1508-1512.
21. Rosen EU. The diagnostic value of an enzyme-linked immunosorbent assay using adsorbed mycobacterial sonicates in children. Tubercle 1990; 71: 127-130.
22. Wilkinson RJ, Haslov K, Rappuoli R. et al. Evaluation of the recombinant 38-kilodalton antigen of Mycobacterium TB as a potential immunodiagnostic reagent. J Clin Microbiol 1997; 35: 553-557.
23. Daniel TM, Debanne SM. Serodiagnosis of TB and other mycobacterial diseases by enzyme-linked immunosorbent assay. Am Rev Respir Dis 1987; 135: 1137-1151.
24. Demkow U, Zielonka TM, Strzalkowski J, Michalowska-Mitczuk D, Augustynowicz-Kopec E, Bialas-Chromiec B, Kus J, Skopinska-Rozewska E, Zwolska Z. Diagnostic value of IgG serum level against 38-kDa mycobacterial antigen. Pneumonol Alergol Pol 1998; 66: 509-516.