Immunocytochemistry versus nucleic acid amplification in fine needle aspirates and tissues of extrapulmonary tuberculosis.

BACKGROUND: Immunocytochemistry (ICC) is an established routine diagnostic adjunct to cytology and histology for tumor diagnosis but has received little attention for diagnosis of tuberculosis. AIMS: To have an objective method of direct visualization of mycobacteria or their products in clinical extrapulmonary tuberculosis (EPTB) specimens, immunocytochemical localization of M. tuberculosis antigen by staining with species specific monoclonal antibody to 38-kDa antigen of Mycobacterium tuberculosis complex.
MATERIALS AND METHODS: Immunostaining with specific monoclonal antibody to 38-kDa antigen of Mycobacterium tuberculosis complex was done in fresh and archival fine needle aspirates and tissue granulomata of 302 cases of extrapulmonary tuberculosis and was compared with the molecular diagnostic i.e., nucleic amplification and conventional [Cytomorphology, Ziehl Neelsen (ZN) staining and culture] tests and 386 controls.
RESULTS: Diagnostic indices by Bayesian analysis for all types of archival and fresh material varied from 64 to 76% in nucleic acid amplification (NAA) and 96 to 98% in ICC. There was no significant difference in the diagnostic indices of ZN staining and/ or ICC in fresh or archival material whereas the sensitivity of NAA differed significantly in fresh versus archival material both in cytology (71.4% vs 52.1%) and histology (51.1% vs 38.8%). ICC can be easily used on archival smears and formalin-fixed paraffin-embedded tissue sections with almost equal sensitivity and specificity as with fresh material, in contrast to NAA which showed significant difference in test results on archival and fresh material.
CONCLUSIONS: Low detection sensitivity of MTB DNA in archival material from known tuberculous cases showed the limitation of in-house NAA-based molecular diagnosis. ICC was found to be sensitive, specific and a better technique than NAA and can be used as an adjunct to conventional morphology and ZN staining for the diagnosis of EPTB in tissue granulomas.

Introduction

World over the emphasis has been on treatment of tuberculosis than on diagnosis. Diagnostic tests in pipeline have been hi-tech that have not taken into consideration the setting where the test is to be applied. As a result expected case detection rate is low even in pulmonary tuberculosis that is comparatively easier to diagnose as compared to extrapulmonary tuberculosis (EPTB). Various existing nucleic acid amplification (NAA) assays have shown excellent sensitivity and specificity in respiratory specimens, but their use in clinical settings for non-respiratory specimens still requires validation. Moreover they are still out of reach in routine diagnosis in resource poor countries.[12]

Of the two forms of tuberculosis, EPTB is often difficult to diagnose because of its diverse clinical presentation and uncertain clinical evaluation. Lymph node is the most common type of EPTB and fine needle aspiration cytology (FNAC) is the first investigation advised. Immunocytochemistry (ICC) is an established routine diagnostic procedure in cytology and histology for tumor diagnosis but has received little attention for diagnosis of tuberculosis.[3]

Present study was aimed at immunocytochemical localization of Mycobacterium tuberculosis antigen by staining with species specific monoclonal antibody to 38-kDa antigen (MTSS) of Mycobacterium tuberculosis complex in fresh and archival fine needle aspirates and tissue granulomata of extrapulmonary tuberculosis to have an objective method of direct visualization of mycobacteria or their products in clinical EPTB specimens in comparison with conventional and NAA tests.

Materials and Methods

Study design

It was a case-control study for diagnostic test evaluation in a setting of a tertiary care teaching hospital. The index test to be evaluated was ICC applied to FNAC smears and histological sections of EPTB.

Samples

A total of 302 extrapulmonary specimens and 386 controls were taken for detailed study. These included fresh fine needle aspirates of patients suspected of EPTB, archival FNA smears and both fresh and archival formalin-fixed paraffin-embedded (FFPE) tissue sections of both cytologically and histologically confirmed cases of EPTB of all age groups and both sexes. The samples in cytology were lymph node fine needle aspirates whereas in histology the samples included tissues from tuberculosis of other extrapulmonary sites also in addition to tuberculosis of lymph nodes. Of the 302 cases, ZN and immunostaining was done in all the cases, culture in 178 cases (in fresh FNA material only) and NAA in 193 cases only [Table 1] .

Table 1

Detailed results of cases and controls subjected to Ziehl-Neeslen, immunohisto(cyto)chemical staining and NAA (PCR)

Cyto and histological diagnosis

Prospectively enrolled patients suspected of EPTB were subjected to FNAC using 23-25 gauge needle fitted to a 20 mL disposable syringe. Multiple smears were made from the aspirated material for hematoxylin and eosin, May-Grünwald-Giemsa stain for cytomorphologic study, ZN staining and ICC. The cases finally confirmed as EPTB were included in the study. The left over aspirated material (approximately 5-10 μL) in the needle was expelled in 1.5 mL of balanced salt solution (BSS) for M. tuberculosis culture. Repeat aspiration was done for MTB DNA extraction and NAA. Cultures were done only on aspirated material and not on biopsy tissues.

The diverse cytomorphological spectra [Plate 1] were noted in cytological smears of suspected cases of tuberculosis which were categorized in to seven groups.[134] Histological diagnosis of granulomatous lesions was made according to diagnostic criteria described in standard text.

Plate 1

Cytomorphological spectrum from group I to group VII

Inclusion/ exclusion criteria

Inclusion criteria– All confirmed cases of EPTB from archival group and suspected cases from prospective group were included.

Exclusion criteria– Inadequate and unsatisfactory smears or tissue blocks were excluded from the study.

Gold standard (reference standard) for diagnosis of EPTB

An extended gold standard was used as reference standard, according to which true cases of tuberculous lymphadenitis comprised of AFB-positive smears (irrespective of group I –VII morphology), cytomorphological groups I and II, and AFB-negative groups III to VII that subsequently were confirmed on histology and /or on response to anti-tuberculous treatment (ATT). Thus the final diagnosis was based on one or more conventional diagnostic tests inclusive of FNAC and/ or biopsy, ZN staining, culture supported with clinico-radiological picture and positive montoux test. Instances where results of conventional diagnostic tests were negative or equivocal, response to treatment (4 weeks) was taken as final arbiter for diagnosis. This reference standard was independent of the index test (ICC with 38-kDa MTSS antibody) i.e., the index test did not form the part of reference standard.

Immunocytochemistry

Both archival FNA smears and archival FFPE tissue sections (2-3 years old) were retrieved from the records of cytology and histology laboratories of the Department of Pathology. After a thorough reappraisal of final diagnosis of EPTB on FNA smears and tissues sections, the workable smears and paraffin blocks were taken out for study. The smears were destained with 1% acid alcohol while 5-6 sections of 3-5-μm thickness were cut from tissue blocks for ZN and ICC. Immunostaining in smears and sections was done by streptavidin-biotin method using species specific monoclonal antibody against 38-kDa antigen of MTB complex already standardized in our laboratory.[3] The observations were interpreted as follows:

Brown-colored deposits in and around the areas of granulomas indicating positive immunostaining [Figures 1a, b].

Figure 1a

Microphotograph of immunostained FNA smear of group I cytomorphology showing dark brown granular deposits of positive immunostaining in the area of caseous necrosis (ICC, ×400)

Figure 1b

Microphotograph of FNA smear of Group II cytomorphology showing positive dark brown immunostaining in the area of epithelioid cell cluster. Individual bacterial rods are not seen in the smear (ICC, ×400)

Areas of caseous necrosis showing staining of extracellular bacilli, large and small fragments of bacilli and antigenic dust [Figures 2 and 3].

Figure 2

Immunostained FNA smear of group IV morphology showing numerous bacilli (black arrows) in the area of caseous necrosis. Some brown staining is also seen in the background may be due to staining of mycobacterial products (ICC, ×400)

Figure 3

Microphotograph showing intense positive immunostaining of antigenic dust in the centre in a lymph node aspirate smear of Group VII cytomorphology that had polymorphous cell population suggesting reactive lymphadenitis (ICC, ×400)

Noncaseating areas showing staining of extracellular and intracellular bacilli as well as the mycobacterial antigens characteristically located within the cytoplasm of macrophages, giant cells and as diffusely stained granular brownish material [Figures 4a, b].

Figure 4a

Immunostained histological section from a case of tuberculous granuloma showing brown stained intracellular and extracellular bacilli (black arrows ) (ICC, ×400)

Figure 4b

Immunostained histological section of tuberculous lymphadenitis showing dark brown mycobacterial products (ICC, ×400)

Nucleic acid amplification

In house NAA for MTB complex DNA from fresh FNA material, archival and fresh FNA smears and tissue sections was done in three steps as standardized[1] in our laboratory:

Mycobacterial DNA extraction by phenol: chloroform method.

MTB complex DNA amplification by using IS6110 primer sequence.

The amplified mixture was analyzed by 3% agarose gel electrophoresis (0.5% agarose in trisborate EDTA buffer pH 8.0) and visualized on 360 nm wavelength UV transilluminator [Figure 5].

Figure 5

Nucleic acid amplification (PCR) from fresh fine needle aspirates

NAA based PCR using primers to IS6110 insertion sequence of M. tuberculosis DNA, was also done on FNA smear scrapings and tissue sections. The stained archival smears irrespective of the type of staining were destained before the DNA extraction step.

Controls

Controls comprised 386 samples inclusive of negative, positive, random and blind controls which were run with every batch of ICC staining and NAA [Table 1].

Statistical Analysis: Bayesian analysis on 193 cases was done to evaluate accuracy, positive predictive value and negative predictive value of ZN and Immunostaining, Culture and NAA (PCR). The results of immunostaining were compared with conventional (ZN staining and culture) and molecular (NAA) methods.

Results

The overall comparative detailed results of ZN staining, immunostaining and NAA for MTB complex from fine needle aspirates and tissue granulomata of both fresh and archival material from cases of EPTB is shown in Table 1. The cytomorphological spectrum classified in to seven groups as seen in tuberculous lymphadenitis (TBLN) is shown in plate 1. The diagnostic algorithm in each group is based on cytomorphological features and ZN staining.

Diagnostic accuracy values of NAA and ICC were calculated for all types of archival and fresh material as shown in Table 2. Diagnostic accuracy rate for NAA ranged from 64 to 76%, while it was 96 to 98% for ICC.

Table 2

Diagnostic accuracy of NAA and ICC

The diagnostic indices (by Bayesian analysis) of immunostaining for fresh and archival fine needle aspirates and tissue granulomata of EPTB were compared with those of molecular (NAA) and conventional tests [Table 3]. On comparison of diagnostic indices of various tests in cytological and histological material, it was observed that there was no significant difference in the values of diagnostic indices of ZN staining and/ or ICC in fresh or archival material whereas the sensitivity of NAA differed significantly in fresh versus archival material both in cytology (71.4% vs 52.1%) and histology (51.1% vs 38.8%). Comparison of ICC versus NAA and their relative costs compared with conventional cytodiagnosis, ZN staining and culture are shown in Tables 4 and 5, respectively.

Table 3

Shows comparative details of diagnostic indices of ZN staining, culture, CR and ICC/ IHC-staining techniques done on prospective (fresh) and archival FNA material and FFPE tissue sections of both cytologically and histologically diagnosed cases of EPTB

Table 4

Immunocytochemistry versus nucleic acid amplification

Table 5

Cost of conventional tests, NAA and immunocytochemistry

Discussion

The objective of using fresh as well as archived FNA and histological materials was to assess the reproducibility of the diagnostic test in our setting where it is expected to be applied under different situations. The diagnostic indices for ICC/ IHC for fresh FNA smears were quite close to that observed from archival material implying that positive immunostaining does not depend on the type (fresh/ archival) or storage period of the material while the indices for NAA that were found to be higher for fresh FNA and FFPE tissue specimens indicating that quality of DNA deteriorates with time (age of material).

FNAC is an established, safe, quick and inexpensive procedure for detecting tuberculosis with a diagnostic accuracy ranging from 71.3% to 97%.[4-11] As well known, the diagnosis is based on the features such as epithelioid granulomas with / without caseous necrosis and Langhan's giant cells.[7-9] In the present study, diagnosis of tuberculosis was rendered by FNAC alone in 76.2% cases; diagnostic accuracy went up to 82.7% when it was combined with ZN staining. Groups I-IV cytomorphology in majority of the instances conform to the standard text book description whereas findings in Group V-VII are nondesciptive and ultimately turn out to be tuberculous. The overall AFB positivity in FNA smears and FFPE tissue sections including both fresh and archival, ranged from 32.40% to 60.05%. Purulent aspirates showed the highest number of AFB (58.44%) and lowest number of granulomas whereas aspirates of blood-mixed particles showed the highest number of granulomas and lowest number of AFB (45.27%). AFB positivity in smears and sections depends upon the bacillary load in the specimen.[810] However, ZN staining cannot differentiate between tuberculous and nontuberculous mycobacteria. In our study, two of the aspirates (lymph node and pleura) with strong AFB positivity were shown to contain nontuberculous mycobacteria, which was evident by the negative ICC and NAA. Both these cases were HIV positive. Culture could not be done in these patients because both the patients were lost to follow-up.

Low sensitivity of NAA from archival material could be attributed to possible factors such as low mycobacterial DNA content, inefficient DNA extraction, or DNA purification, or incomplete lysis (the complex wall of mycobacteria may require several enzymic steps for lysis presence of inhibitors, amount of material scraped, effects of fixation, staining and destaining steps involved, and other technical errors.)[11-13] It is also known that DNA quality deteriorates during the storage of slides; that might have been an additional factor responsible for low NAA positivity in archival material.[41214-16] An important parameter that may have an impact on the outcome of genomic amplification is the nature of the tissue tested, tissue type, tissue processing (intense mixing, high speed centrifugation and heat shock) and the number of copies of target sequence.[14] False positivity was observed only in three cases of negative controls (a case of lymphoma, a case of nonspecific lymphadenitis and a case of metastatic carcinoma – nontuberculous controls) could be explained by the theoretical possibility of laboratory contamination. Low PCR positivity from both archival FNA smear scrapings and FFPE tissue sections of known tuberculous cases and high cost limits the use of PCR for retrospective diagnosis of EPTB. Therefore, it should be reserved for cases in which there is strong clinical suspicion of tuberculosis with equivocal results of conventional diagnostic methods. The NAA test may not be able to replace microscopy or culture. It should be used only in conjunction with conventional tests and clinical data. While some NAA assays reportedly seem to work quite well, there is very wide variability even from very resource-rich laboratories, making their use in the field uncertain. Their difficulty of use and quality control issues combined with the variability and lack of sensitivity in sputum negative and extrapulmonary TB do not support their use.[17]

Review of published English literature shows reports of occasional application of immunohistochemistry using polyclonal and monoclonal antibodies raised against 35 kDa, 65 kDa, MPT64 proteins (antigens) and BCG strains of MTB for detection of mycobacterial antigens in various clinical and experimental specimens and that too mainly in tissue sections.[318-20] To the best of our knowledge, detection of mycobacterial antigen by ICC using monoclonal antibody against 38-kDa antigen of MTB complex in fine needle aspirate smears and tissue sections of EPTB has not been reported so far by other workers except us in our previous work.[319] The 38-kDa protein is one of the most potent cell surface immunogen of MTB solely detected in the virulent M. tuberculosis and in lesser amounts in the vaccine strain M. bovis BCG.[2021] Hence the study was taken up with the assumption that IHC with monoclonal antibody to 38-kDa antigen of MTB complex would be able to specifically locate the MTB complex antigen in the tuberculous granulomas. With immunostaining technique we were able to demonstrate MTB antigen in the areas where acid fast bacilli were absent or scarce. This indicated that even debris that were derived from mycobacteria retained its antigenic property although it had lost its AFB-staining property. The findings are in close conformity with previous workers.[22-24] Of the 25 AFB-negative sputum smears, immunostaining was positive in 10 cases i.e., 40% of AFB-negative smears were positive by ICC. Therefore, it has to be seen further in a well designed study whether the ICC can be used as an adjunct to the diagnosis of pulmonary TB also, specially in AFB-negative cases. This can have an added advantage over conventional ZN staining. Immunostaining was also performed in smears obtained from stock cultures of bacteria, mycobacteria other than tuberculosis (M. avium, M. smegmatis, M. fortuitum and M. kansasii) and fungal species which was negative in all these smears. The findings in smears of atypical mycobateria are supported by the observations by Thangaraj et al, 1996 that 38-kDa antigen was not found in M. avium, M. paratuberculosis, M. smegmatis, M. fortuitum, M. chelonei, M. leprae, M. kansasii, M. marinum and M. vaccae, indicating that detection of 38-kDa antigen in clinical specimens could be a direct and specific evidence of infection with MTB complex.[25]

The classical histological picture of tuberculous granulomatous inflammation is not a diagnostic problem in a tissue biopsy. However, when the sections show noncaseous epithelioid granulomas mimicking tuberculosis, a common occurrence in our biopsies posing a diagnostic dilemma, the positive immunostaining with species specific antibodies in these cases will also rule out the possibility of sarcoidosis or other nonspecific tuberculoid granulomas.[3] Comparison of ICC and NAA showed that ICC is a sensitive and specific technique and could be used as an adjunct to conventional ZN staining for the diagnosis of EPTB in tissue granulomas. Since the first specific investigation advised in these cases is FNAC, ICC can be done on the same smears prepared for cytology reporting. Additional smears or extra efforts are not required for obtaining samples for ICC while for NAA (PCR) each time fresh FNA material is required. The sensitivity and diagnostic accuracy of ICC is almost equal in fresh or archival material whereas NAA on the same archival material has considerably low detection sensitivity and diagnostic accuracy. In addition, storage is not a problem for ICC from archival FNA and tissue material since it does not require special facility and can be stored at room temperature. ICC can also be performed on routinely stained cytosmear after destaining the same smears and subjecting them to immunostaining. ICC was found to be cost effective as compared to NAA. FNAC combined with ICC will obviate the need of tissue biopsy. Role of FNAC as a sample collection modality for diagnostic workup of extrapulmonary tuberculosis therefore can not be undervalued.[32627] The index test may be applicable as a diagnostic adjunct to a major proportion of cases of EPTB with good sensitivity, specificity and easy interpretation where the primary diagnosis is based on cytological or histological material. Since it is a microscopy-based test, this can be done practically only in those laboratories which are geared for histology and cytology reporting. ICC does not require a sophisticated molecular laboratory. It requires only the laboratory facility with training of the staff, standardization of technique and quality control. It has the ability to distinguish between MTB complex and nontubercular mycobacteria. Further, there is no spectrum bias because the spectrum of patients included in the study were representative of the patients who received the test.[28]

The major constraint of ICC is that due to economical reasons and the lack of trained technicians/anatomic pathologist, as well as, due to lack of stringent quality control, it cannot be made a routine procedure at peripheral laboratories. Moreover, in places such as DOTS centres, PHCs and CHCs, only sputum samples are examined for AFB. The other constraint is that a quick report is not possible.

Conclusions

To conclude, ICC is a more sensitive, specific and cost-effective method of detecting mycobacterial antigen and therefore has a potential for its validation and generalization on large scale in field setting for its use on FNA smears of suspected cases of extrapulmonary tuberculosis. An anatomical pathologist plays a significant role in this regard and his or her role should not be underestimated at the time of policy making for diagnostic laboratory services of tuberculosis.[29]