© 2002 American Thoracic Society
Sensitivity of Acid-Fast Staining for Mycobacterium tuberculosis in Formalin-fixed TissueFirst Department of Pathology, Yamaguchi University School of Medicine; and Department of Clinical Research, National Sanyo Hospital, Ube, Yamaguchi, Japan Correspondence and requests for reprints should be addressed to Hajime Fukunaga, 1-1-1 Minamikogushi, Ube, Yamaguchi 755-8505, Japan. E-mail: fukunaga{at}sanyou.hosp.go.jp
Microscopic examination of tissue sections of mycobacterial lesions frequently results in few or no bacilli seen, even if the lesions appear active histologically. This might be due to the effects of the fixative fluid and/or organic solvent, both of which are conventionally used to make tissue sections for histopathology, on the acid-fast staining of bacteria. The present study was performed to examine how formalin and xylene lower the sensitivity of acid-fast staining for Mycobacterium tuberculosis and to clarify the meaning of the staining result in tissue sections. Microscopic observation of mycobacteria smeared on glass slides revealed that both of these agents greatly reduced the sensitivity of acid-fast staining. Moreover, the number of bacilli was calculated in 30 samples of paraffin-embedded granulomatous lesions using acid-fast microscopy and real-time polymerase chain reaction. The numbers of bacilli present that were estimated by real-time polymerase chain reaction were considerably higher than those counted with a microscope. These results suggest that the bacilli are frequently missed or underestimated with acid-fast microscopy on formalin-fixed, paraffin-embedded tissue.
Key Words: formalin • xylene • real-time polymerase chain reaction
Despite longstanding intense efforts to conquer tuberculosis, this disease remains an expanding global health crisis with 1.86 billion infected people. To overcome tuberculosis, new diagnostic or therapeutic methods based on molecular biology are now being developed. Some are currently in clinical use; e.g., polymerase chain reaction (PCR) for identifying genes related to mycobacterial species or drug-resistant strains (1–3). On the other hand, classic laboratory tests, such as sputum smears, bacterial cultures, and histopathologic examination, still play a central role in the screening, diagnosis, and monitoring of this disease. For the time being, these remain the most widely used tests because they offer certain advantages. For example, the microscopic observation of tissue sections permits the distinction between granulomatous lesions and other clinically or radiographically similar lesions including cancers as well as the evaluation of the histologic type and stage of the tubercle lesions. However, some pathologists have become aware that the number of bacilli in tissue sections stained with acid-fast staining (acid-fast staining histopathology [ASH]) frequently seems to be much lower than that expected from the sputum smear data or the patient's condition. We also have believed that ASH has low sensitivity; therefore, we have re-evaluated its sensitivity. We postulated that the low detectability of ASH is due to formalin and/or xylene, which are used to make preparations of tissue sections but not of smears. All mycobacteria have a lipid-rich cell wall that avidly retains carbolfuchsin, rhodamine, or auramine dye even in the presence of acidic alcohol. Formalin and/or xylene might alter this property of the bacilli. The present study was performed to examine how formalin and xylene lower the sensitivity of acid-fast staining for Mycobacterium tuberculosis and to clarify the meaning of the staining result in tissue sections.
Preparation of Cell Smear and Acid-Fast Staining M. tuberculosis H37RV grown in a liquid medium (MYCO-broth; Kyokuto Pharmacy, Tokyo, Japan) were fixed with 10% neutral buffered formalin overnight. The bacillary density was adjusted to 108 CFU/ml, and the suspension was divided among 24 tubes by twofold serial dilution. Then, 250 µl from each tube was transferred into separate microtubes and centrifuged (6,000 x g, 15 minutes). After removing the supernatant, the cell pellets were dried, and each pellet was resuspended in xylene and incubated for 4 hours at 35°C. After centrifugation, the cell pellets were dried and resuspended in 250 µl of the medium. For the untreated control cells, twofold serial dilutions were used and the cells were resuspended in the medium to the same cell densities. Ten microliters from each tube was dropped onto a nonfluorescent and poly-L-lysine coated glass slide (Sigma, St. Louis, MO). The smears were stained by Ziehl–Neelsen staining (TB-color; Merck, Darmstadt, Germany) or rhodamine–auramine staining (TB Fluorescent Stain Kit; BD, Franklin Lakes, NJ) and were observed under a conventional light microscope or a fluorescent microscope with Nomarski optics, respectively. The sensitivity of acid-fast staining was evaluated as the percent of positive bacilli among the estimated number of total bacilli (calculated from the higher dilutions where the number of positive bacilli was small enough to be counted visually [< 200 cells/slide]). In the lower dilution series, where the number of positive bacilli was too large to count, the number of positive bacilli was expressed as +++, ++, or + (+++, heavy overlapping of bacilli that were spread all over the slide; ++, bacilli were densely spread all over the slide without heavy overlapping; +, spotty or sparse distribution of bacilli on the slide).
Real-Time PCR and ASH for Tissue Sections For DNA extraction, the paraffin was removed by soaking the slides in xylene. Then, the slides were transferred into Tris–EDTA buffer containing decreasing concentrations of ethanol. The tissue sections were scraped off the slides with a swab and were put into Tris–EDTA buffer with microfine glass beads in microtubes. The samples were then sonicated with an ultrasound lyser (Abbott, Abbott Park, IL). Two types of primers for Mycobacterium genus as well as two types of fluorescence-labeled M. tuberculosis–specific probes that specifically hybridized with the amplified products were designed. The sonicated sample, the primers, the probes, and hybridization buffer (LightCycler DNA Master Hybridization Probes; Roche Diagnostics, Mannheim, Germany) were mixed and were subjected to 60 cycles of amplification with a LightCycler (Roche Diagnostics). Total amount of M. tuberculosis DNA was calculated using the LightCycler software. The number of bacteria per tissue section was estimated by dividing the amount of DNA by the amount of DNA that was calculated to be present in a single cell (0.04 pg). Additional details on the method are provided in the online data supplement.
Influence of Formalin and Xylene on Acid-Fast Staining Microscopic observation of the smears revealed that the sensitivity of both types of acid-fast stains (Ziehl–Neelsen and rhodamine–auramine staining) for M. tuberculosis decreased drastically with formalin treatment or with the combined treatment of formalin and xylene (Figure 1 , Table 1) . As for the untreated bacilli, the rhodamine–auramine method stained all cells, whereas Ziehl–Neelsen staining stained only 21.8% of all cells. Among the formalin-treated samples, only 5.8% showed positive results with fluorescence staining, and 2.1% showed positive results with Ziehl–Neelsen staining. Moreover, for both formalin- and xylene-treated bacilli, the sensitivities fell to 0.5 and 0.2%, respectively.
Comparison between Real-Time PCR and ASH in Tissue Sections Table 2 shows the results of real-time PCR and ASH on 30 tissue sections of tuberculosis lesions (male/female, 19/11; mean age, 63.7 years). Using PCR, M. tuberculosis DNA was detected in cases 1–25 (83% of total cases), whereas using ASH (Ziehl–Neelsen staining), bacilli were detected in cases 14–25 (40% of total cases). All the PCR-negative samples were also ASH-negative (cases 26–30). However, the ASH-negative sections (18 cases) included 13 PCR-positive cases (cases 1–13). The largest number of bacilli observed with ASH was 240 (case 25), whereas the smallest was 1 (cases 13 and 14). On the other hand, the largest number of bacilli estimated from the PCR data was 27,100 x 105 (case 21) and the smallest number was 332 (case 17) among the ASH-positive cases. No significant correlation between the numbers of bacilli estimated by ASH and by PCR was observed (r = 0.055). The case with the highest number of bacilli with ASH was different from the case with the highest number of bacilli with PCR.
Searching for tubercle bacilli in tissue sections is laborious and time consuming because these bacilli are small and few. Most pathologists doubt the sensitivity of ASH especially when the number of bacilli observed in a tissue is low even when the lesion appears histologically active or many bacilli are present in the sputum. The reason for this low detectability by ASH has not been well studied. One might think that the number of bacilli in such specimens is low and that ASH is reliable. Indeed, mycobacterial lesions are caused by type IV hypersensitivity; therefore, the extent of inflammation might not correlate well with the amount of bacterial antigen. However, this hypothesis cannot explain the large discrepancy between the sputum smear data and the ASH data in a given patient. Some researchers have suspected that the organic solvents, which are used to make paraffin-embedded tissue samples, might affect the stainability of mycobacteria by acid-fast staining. This hypothesis seems reasonable because the molecular target of the acid-fast staining dyes (fuchsin, auramine, or rhodamine) is the mycolate on the bacterial surface (4, 5). Mycolate is soluble in organic agents (6) and might be more or less extracted from the cell surface into the liquid. Moreover, this hypothesis is indirectly supported by reports that the Fite method, which is an alternative acid-fast staining technique that uses less xylene in the staining and mounting process, has a higher sensitivity than Ziehl–Neelsen staining (7, 8). The present data clearly demonstrate that the combined use of formalin and xylene markedly reduces the sensitivity of acid-fast staining for mycobacteria, irrespective of the method used (TB-color or TB Fluorescent Stain Kit) (Figure 1, Table 1). Interestingly, treatment with 10% formalin alone also lowered the sensitivity of bacilli to the staining (Table 1). This is the first report on the effect of formalin on acid-fast staining. Because formalin is a water solution of formaldehyde and is not a lipid-soluble agent, the mechanism for its reduction in the sensitivity of acid-fast staining remains unknown. These data strongly suggest that the number of mycobacteria in tissue sections is significantly underestimated with conventional acid-fast staining (0.5% with the fluorescent method and 0.2% with Ziehl–Neelsen staining). Why a small fraction of the bacillary population retains its stainability after the treatment is not clear. There might be a large variation in the lipid content of the cell wall among cells. The high sensitivity of real-time PCR in detecting pathogens is generally well known (9–19), but the difference in the sensitivity between PCR and ASH observed in this study seems quite large. The estimated number of bacilli from the real-time PCR data was 103 to 109 times larger than the number counted with microscopy among the ASH-positive samples (Table 2). One of the reasons is the effect of formalin and xylene in ASH as presented in the smear experiments. Another reason for the difference between real-time PCR and ASH is that PCR detects not only nuclear DNA in live bacilli but also DNA molecules isolated from killed bacilli and deposited in the stroma or in the phagocytic cells (20). This suggests that PCR may be too sensitive to distinguish between active and inactive or infectious or noninfectious states of the lesion (21–34). The present study definitely and quantitatively demonstrated the effect of formalin and xylene on the stainability of mycobacteria by acid-fast staining and the extremely low sensitivity of ASH. However, this result should not discourage pathologists from using ASH to evaluate mycobacterial lesions. On the contrary, such knowledge may help explain any discrepancy between ASH and sputum smear test results and encourage pathologists to state the possibility of the presence of bacilli in their reports. The real-time PCR technique may not to be superior to ASH in the evaluation of disease activity as described previously. On the basis of the insight into the mechanisms responsible for the decreased sensitivity of acid-fast staining, a novel staining method with higher sensitivity for tissue sections needs to be developed to evaluate the presence of bacilli more correctly.
The authors thank Drs Takayuki Ezaki and Hiroyasu Takeyama for their general support of this research and Dr. Nagatoshi Fujiwara for his technical assistance. The authors also thankfully acknowledge Sadaki Endo and Kazumi Hori for managing this project as Chiefs of the Department of Clinical Laboratory, National Sanyo Hospital, Ube, Yamaguchi, Japan.
Supported by a grant from the Department of Clinical Research, National Sanyo Hospital. This article has an online data supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form November 13, 2001; accepted in final form July 2, 2002
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
2 x 103 cells on a slide). All the cells in the no-treatment group are positively stained (A and B). Formalin only treatment reduces the number of positive cells (C and D). Combined treatment with formalin and xylene causes few cells to be positive.
