Histopathology 1977,
I,
277-288
A trichrome stain for the intrahepatic localization of the hepatitis B surface antigen (HBsAg)
L . G U B E T T A , * M . R l Z Z E T T 0 , O.CRIVELL1, G.VERME & S.ARICO Departments of Pathology and Gastroenterology Ospedale Mauriziano Umberto I, Torino, Italy Accepted for publication
14
April 1977
GUBETTA L., RIZZETTO M., CRIVELLI O., VERMEG . & ARIC~) S. (1977) Histopathology I, 277-288
A trichrome stain for the intrahepatic localization of the hepatitis B surface antigen (HBsAg) A modified trichrome stain is described for the intrahepatic localization of the hepatitis B surface antigen; HBsAg containing cells exhibit specific green metachromasia contrasting with the granular brown colour of non infected hepatocytes and with the deep eosinophilic colour of ground glass cells of HBsAg-negative alcoholic or drug hepatitis. The technique is simple and reliable for routine screening of HBsAg positive material; its sensitivity is greater than H & E, similar orcein and inferior to immunohistochemistry as performed on frozen sections. Histological diagnosis can be made on the same slide, since several other morphological details are provided in the trichrome stained preparations. With this technique 387 biopsies from HBsAg seronegative individuals were negative; full cytoplasm metachromasia was mostly seen in asymptomatic HBsAg carriers, focal or partial staining in patients with histological evidence of liver cell necrosis. The presence and the staining pattern of HBsAg were of no help in predicting transition to chronicity or a transition from chronic persistent to chronic active hepatitis. Keywords : HBsAg, liver disease, trichrome stain, orcein stain, immunofluorescence
Introduction The localization of the hepatitis B surface antigen (HBsAg) in the liver provides valuable information to the histopathologist (Deodhar, Tapp & Scheuer 1975, Portmann, Galbraith, Eddleston, Zuckerman & Williams 1976), the viral origin of
* Address for correspondence : Dr L.Gubetta, Department of Pathology, Ospedale Mauriziano Umberto I, Cs. Turati 46, 10128 Turin, Italy. 277
278
L.Gubetta et ul.
the disease may become apparent only after immunochemistry (Ray, Desmet, Fevery, De Groote, Bradburn & Desmyter 1976a) and distinctive patterns of distribution of HBsAg are helpful in the diagnosis of different types of hepatitis (Gudat, Bianchi, Sonnabend, Thiel, Aenishaenslin & Stalder 1975) and in the recognition of the healthy chronic carrier status (Hadziyannis, Vissoulis, Moussouros & Afroudakis 1972, Rizzetto, Bonino, Verme, Gubetta, Toselli, Canese & Borgialli 1976a). Different methods are at present available to detect the antigen in the liver, each of them presenting a few disadvantages when used routinely in busy hospital laboratories. Sensitive and specific immunochemical techniques are demanding and require expensive equipment. Storing of positive fluorescent specimens and their retrospective examination is impossible for the transient nature of the phenomenon and endogenous background enzymatic activity may be troublesome after immunoperoxidase when the tissue is inflamed with many white blood cells. The recognition of HBsAg containing hepatocytes by their ground glass appearence after haematoxylin and eosin (H & E), (Hadziyannis, Gerber, Vissoulis & Popper 1973) depends on minimal variations in the tonality of the stain; similar appearances are sometimes due to fixation and staining artefacts and the identification of scattered or partially infected cells may be difficult (Portmrnn et al. 1976, Rizzetto et al. 1976a). The Shikata stains are more selective (Shikata, Uzawa, Yoshiwara, Akatsuka & Yamazaki 1974); artefacts, however, have been noted with orcein stain depending on the quality and age of the reagent (Thomsen, Poulsen & Petersen 1976, Rizzetto et al. 1976a) and aldehyde fucsin was noted to stain non-specifically any material containing disulphide bridges, like lipofuscin and glycoproteins taken up by Kupffer cells (Shikata et al. 1974, Thomsen et al. 1976). Both procedures provide no other histological information, apart from the localization of HBsAg, and additional slides of the same specimen must be stained with other dyes for diagnostic purposes. While studying liver collagen with several stains, a peculiar green metachromasia was noted in liver biopsies from HBsAg seropositive patients after the EA 31 Papanicolau solution. Further studies showed that this appearance was specifically related to the presence of intrahepatic HBsAg (Gubetta, Rizzetto & Verme 1976). During the last 12 months we have used routinely this modified trichrome stain on all the liver specimens sent to the pathology department. The technique, besides permitting the recognition of HBsAg containing hepatocytes, provides several other morphological details and enables the pathologist to make diagnostic histology on the same slide. Tn the present communication, this method is compared with the other techniques at present available to detect HBsAg; the results obtained are presented and their diagnostic significance is discussed.
Materials and methods Five hundred and forty-two biopsies were included in this study, 387 from HBsAg
Trichrome stain for HBsAg
279
seronegative patients with a variety of liver disorders and I 55 from HBsAg seropositive individuals; in the latter group histology showed acute hepatitis in 27 cases, a chronic persistent hepatitis (CPH) in 21, chronic active liver disease with or without nodular transformation (CALD) in 45, inactive cirrhosis (IC) in I 0, hepatocellular carcinoma in four, a normal liver or minimal abnormalities in 22. In a further group of 26 patients, the picture was compatible either with an acute hepatitis running a prolonged course (APH) or an early active chronic disorder. The first diagnosis was provisionally advanced in each case for the lack of anamnestic or biochemical evidence of chronic disease. A second biopsy after 6-12 months was obtained from the patients with APH, those with CPH and those with CALD without nodular transformation. Immediately after liver puncture each biopsy was cut in two portions, the longer (at least two-thirds of the whole specimen) was fixed in 4% formol-saline, the shorter was frozen in the cryostat.
HISTOLOGY
Formalin fixed specimens were stained with H & E, with the modified orcein stain of Shikata (Or.) (Shikata et al. 1974) and with a trichrome stain (Tr.) whose details are appended (Gubetta et al. 1976). IMMUNOCHEMISTRY
The frozen and formalin fixed sections of each biopsy were investigated by direct and indirect immunofluorescence (IFL) and by direct immunoperoxidase (IPS) for the presence of HBsAg. A fluorescein (FITC) and a peroxidase conjugated antiserum monospecific against HBsAg were prepared from the Beheringwerke rabbit precipitating antiserum RBB/r5; the preparations details and the specificity of the conjugates have been previously reported (Rizzetto et a/. 1976b). Standard indirect IFL was carried out using the anti HBsAg serum diluted 1/6 and a FITC conjugated antiserum against rabbit Ig obtained from goat (Beheringwerke TKF/o5). Direct IPS was performed as previously described (Rizzetto et al. 1976b). To assess the specificity and sensitivity of the trichrome stain serial formalin fixed sections of HBsAg positive biopsies were stained with Tr. and alternatively with Or., H & E and the fluorescent or peroxidase conjugated antisera.
Results SPECIFICITY AND SENSITIVITY OF TRICHROME STAIN
After trichrome stain, HBsAg containing hepatocytes displayed a homogeneous cytoplasmic metachromasia contrasting with the brownish colour of non infected
280
L.Gubetta et al.
Figure I. Ground glass hepatocytes. a Green cytoplasmic metachromasia after trichrome stain. Liver biopsy from an healthy HBsAg chronic carrier. Trichrome. x 400. b Deep eosinophilic colour after trichrome. Liver biopsy from an HBsAg-negative alcoholic patient. Trichrome. x 400.
Figure 2. Hepatocytes with green metachromasia. a Ground glass appearance after trichrome stain. Liver biopsy from a patient with chronic persistent hepatitis. Trichrome. x 400. b Section serial to a; the same hepatocytes with green metachromasia stained with a peridoxase conjugated antiserum against HBsAg. x 400.
Figure 3. Hepatocytes with green metachromasia. a Ground glass appearance after trichrome stain. Liver biopsy from a patient with chronic active hepatitis. Trichrome. x 400. b Section serial to a; the same hepatocytes with green metachromasia stained with a fluorescein conjugated antiserum against HBsAg. x 400.
Trichrome stain for HBsAg
283
cells (Figure la). Cytoplasmic inclusions of alcoholic patients and deposits of lipofuscin, biliary pigments, haemosiderin and glycogen did not stain green. Cytoplasmic inclusions of a,-antitrypsin deficiency, lipoidoses, mucopolyssccharidoses, Wilson’s disease and primary oxalosis were not investigated. Ground glass cells of alcoholic or drug hepatitis stained deeply red (Figure rb). The specificity of green metachromasia for HBsAg was confirmed by the positive reactions observed in corresponding cells in serial sections stained with orcein or with fluorescein or peroxidase conjugated monospecific antisera against HBsAg (Figure 2a & b, Figure 3a & b). Comparison of the sensitivity of Tr. with that of immunochemistry, Or. and H & E is reported in Table I. Table I. Number and percentage of specimens positive for HBsAg after trichrome, orcein, haematoxylin and eosin and immunofluorescence in I 55 biopsies from HBsAg seropositive patients, related to disease category
Diagnosis Asymptomatic chronic carriers (HCC) Acute hepatitis Acute prolonged hepatitis (AW Chronic persistent hepatitis (CPW Chronic active liver disease (CALD) Inactive cirrhosis (IC) Hepatocellular carcinoma
No. of No. and percentage of biopsies positive for HBsAg after biopsies IFL examined Trichrome Orcein H&E
22
16 (70%)
16 (70%)
6 (27.2%)
16 (70%)
27
26
I1
(42%)
I 1 (42%)
4 (15%)
16 (61%)
21
16 (76%)
16 (76%)
8 (38%)
14 (67%)
45
I0
(22%) 5 (50%)
10 (22%)
I (2%)
I7 (37%) 5 (50%)
I0
4
2
5 (50%) 2
2
Comparison of Tr. with Or. and H & E was performed on serial sections of the fixed part of the biopsy. Evaluation of Tr. versus immunochemistry was performed by comparing Tr. results in the fixed part with those in the frozen part of the same biopsy after IFL. Trichrome staining was not compared with immunochemistry on fixed material because in preliminary experiments occasional irregular and random denaturation of the antigen was noted after formalin fixation. Since indirect 1FL is considered the most sensitive method available for the recognition of HBsAg in the liver (Ray & Desmet 1975) and the fixed part of each biopsy was at least double the length of the frozen, negative Tr. and positive immunochemistry were regarded as being caused by the higher sensitivity of the latter wheress positive Tr. and negative immunochemistry was assumed to represent genuine absence of HBsAg in the smaller part of the biopsy.
284
L.Gubetta et al.
Trichrome sensitivity was lower than IFL, equal to Orcein and greater than H &E. Kupffer cells containing HBsAg were identified best after Tr. while they could be easily confused with freezing artefacts in immunofluorescence. Membrane staining was never observed after Tr. and only occasionally after iminunochemistry. Two biopsies were negative in IFL and positive after Tr.; they showed a persistent chronic hepatitis and a scattered distribution of HBsAg positive cells. Contrast was greatest after orcein and IFL; with these techniques, however, non-specific freezing or fixation artefacts could not be easily distinguished from genuine H BsAg, while after Tr., artefacts never exhibited green metachromasia and could be promptly recognized from deposits of the antigen. HBsAg
LOCALIZATION IN LIVER BIOPSIES
Metachromasia was never observed in the biopsies from 387 HBsAg-negative patients with a variety of liver disorders. No staining was seen in any of the 27 HBsAg seropositive patients whose biopsy showed acute hepatitis at its peak; none had circulating antigen after 6 months or developed chronic disease.
Table 2. Mean percentage of cells in the section exhibiting green metachromasia after Tr. and specific HBsAg staining by immunofluorescence in the fixed and frozen parts of the same positive biopsy; patients divided according to histological diagnosis
HCC
APH
CPH
24.6
10.5
19.2
44.2
25.5
30
CALD
IC
Hepatocellular carcinoma
20.5
15
37.5
30
~
Mean percentage of metachromatic cells in the formalin fixed part of the biopsy Mean percentage of fluorescent cells in the frozen part of the biopsy
5.2
18
HCC - healthy chronic carriers APH - acute prolonged hepatitis CPH - chronic persistent hepatitis CALD chronic active liver disease IC - inactive cirrhosis ~
The number of biopsies studied and the percentage of specimens positive with Tr., immunochemistry and other histological techniques are reported in Table I; the mean proportion of metachromatic and immunofluorescence positive cells observed in the fixed and in the frozen part of positive biopsies in different forms of hepatitis and in healthy asymptomatic HBsAg carriers is reported in Table 2 .
Trichrome stain for HBsAg
285
Kupffer cell staining was observed only in patients with chronic disease, with a greater incidence in active forms in which positive hepatocytes were fewer; in one biopsy from a patient with CALD only some of the macrophages exhibited a green colour while liver cells were negative. The intracellular pattern of HBsAg staining varied between the different groups ; in asymptomatic carriers staining was predominantly diffuse in the whole cytoplasm, in CALD it was usually limited to part of the hepatocyte with no constant topographical distribution within the cell, whilst in inactive disease both patterns were seen in variable combination. Focal HBsAg staining was also the predominant pattern in I I out of a group of 26 biopsies whose features of piecemeal necrosis, lobular inflammation and early fibrosis were compatible either with a late stage of an acute hepatitis or with early chronic disease ; an acute hepatitis running a protracted course was provisionally diagnosed for lack of anamnestic or biochemical evidence of chronic disease. After one year of follow-up, all Tr. positive patients with APH were still circulating the antigen; four developed chronic active hepatitis, four developed CPH, and three became chronic carriers with a normal liver on histology. The majority of Tr. negative patients with APH became HBsAg seronegative and asymptomatic within 6 months; five of them however, developed a chronic HBsAg seropositive liver disease. None of the Tr. positive or negative patients with chronic liver disorders became HBsAg seronegative after the year of follow-up. A second biopsy usually showed progression of the disease in patients with CALD while different pictures were observed in 16 patients with Tr. positive CPH: four of them developed CALD, eight still showed persistent hepatitis, two had liver fibrosis with portal scarring and two had histologically normal liver.
Discussion
Green metachromasia after trichrome staining was selectively confined to the cytoplasm of liver cells containing HBsAg, as shown by the correspondence of positive cells in serial section stained by immunochemistry (Figure 2a & b, Figure 3a & b). Metachromasia appeared to be specific for HBsAg ; cytoplasmic inclusions of the alcoholic, lipofuscin, haemosiderin, biliary pigments and glycogen did not stain green. Specimens from rare metabolic disorders were not available; even if positive, however, their frequency would be insignificant and in our experience green metachromasia is assumed to indicate viral disease. Ground glass cells of HBsAg-positive hepatitis could be easily distinguished from those of drug or alcoholic hepatitis by staining difference, the latter exhibiting a deep eosinophilic colour which contrasted sharply with green metachromasia (Figure Ia & b). Sensitivity of trichrome was higher than H & E, similar to orcein and lower than direct and indirect immunochemistry on frozen specimens; the use of frozen substrates, however, is unpractical, requiring splitting of the biopsy with a smaller amount available for histological diagnosis. Immunohistochemistry on fixed material appeared unreliable to us. Though HBsAg is reported to be unaffected (Ray, Van
286
L.Gubetta et al.
Damme & Desmet 1974) or only slightly diminished (Portmann et al. 1976) after fixation, and indirect imniunohistochemical techniques on fixed material were considered to be far more sensitive than orcein (Ray & Desmet i975), we found a poor correlation between fluorescence in frozen and in fixed parts of the same biopsy because of random and irregular denaturation of the antigen by formalin ; moreover, non specific staining was often increased by using both fixed tissue and indirect techniques as noted also by Huang (r975) and by Lamothe (Lamothe, LaurencinPiche, CGtC, Guerin, Viallet & Richer 1976). The contrast between positive and negative hepatocytes was similar after Tr. and Or. and greatest after fluorescence; with the latter technique however, freezing or fixation artefacts were difficult to distinguish from genuine HBsAg deposits, while they could be easily recognized after trichrome by their different staining. Contrast was less after peroxidase; variability of results and a strong background were often observed with this technique, which in our experience did not appear suitable for routine histochemistry. The trichrome has several advantages over the other methods employed to detect HBsAg in the liver. Preparations can be stored for later reviews, whereas IFL declines in a few days impeding retrospective studies. The technique is less demanding than immunohistochemistry and more reliable than orcein which can produce non-specific granular cytoplasmic and nuclear staining, depending on the quality and age of the dye (Rizzetto et al. 1976b). Histological diagnosis is not possible on IFL or orcein preparations and additional slides must be processed with other conventional stains; the trichrome, on the other hand, provides good morphological detail of parenchymal and connective tissue structures allowing simultaneous histological interpretation of the biopsy and also localization of HBsAg. An inverse relationship between the number of HBsAg positive cells and the amount of necrosis has been repeatedly noted in chronic diseases (Deodhar et al. I 975, Gudat et al. I 975, Hadziyannis et a/. I 972). Similarly, in this study, the percentage of positive biopsies and the proportion of cells exhibiting specific HBsAg staining in positive biopsies were lower in active than in inactive forms suggesting that the cellular expression of HBsAg is somehow related to protection from viral damage. The frequent overlap observed between the CPH and CALD groups made the finding of scanty intrahepatic HBsAg of limited value in differential diagnosis, only extensive involvement of whole liver lobules being reliably diagnostic of inactive or absent disease. Distinctive patterns of antigen distribution within the hepatocyte have also been correlated with various forms of hepatitis ; membrane (Alberti, Realdi, Tremolada & Cadrobbi 1975)and peripheral staining after IFL were considered typical of chronic active hepatitis and active cirrhosis, while focal cytoplasmic localization was found mostly in acute disease (Ray et al. 1976a). We were not able to identify patterns of metachromasia or fluorescence characteristic of any form of hepatitis. Membrane staining was not observed after trichrome, but was sometimes seen after TFL in the frozen part of the biopsy, implying that sensitivity of histological techniques is probably too low to detect a linear distribution of the antigen or that it might not
Trichrome stain for HBsAg
287
be stainable when present on the liver cell surface. Focal or partial staining of the cytoplasm without any constant topographical distribution within the hepatocytes was seen mostly in biopsies with histological evidence of disease. This pattern, however, was not useful in differentiating chronic active disease from persistent or acute, self-limiting, prolonged hepatitis. In our experience, the absence or the presence of HBsAg in the liver and its distribution patterns are only of limited value to the histopathologist, nonetheless, they provide useful indications complementary to other serological and histological data in the management of chronic HBsAg carriers. The intrahepatic localization of the HB core antigen is certainly a more sensitive and reliable indicator of disease than the surface antigen (Gudat et al. 1975, Ray et al. 1976b); its detection, however, is still limited at present to a few research centres. In most other laboratories, recognition of viral infection at a cellular level is possible only through the staining of the surface antigen. The trichrome stain described in this paper appears to be useful because of its simplicity and reliability and for the information it can provide on HBsAg in liver biopsies in busy hospital laboratories, often overrun by daily diagnostic routine. Appendix: method for trichrome stain
Paraffin sections cut at 4-5 pm. Reagents : I Harris’ Haematoxylin z Trichrome stain Light green (yellowish) Bismark brown Eosin (yellowish) Lithium carbonate Phosphotungstic acid
0.5 gin x IOO ml alcohol 95%, 45 ml 0.5 gm x roo ml alcohol 95%, 45 ml 0.5 gm x roo ml alcohol 95%, 45 ml 0.2 rnl saturated solution, 0.02 gm
Procedure : T Deparaffinize as usual to distilled water through two changes of xylene plus absolute and 95% alcohol. 2 Place in Harris’ Haematoxylin for 5 minutes. 3 Wash in distilled water. 4 Wash well in running water for 15 minutes. 5 Rinse in 75% alcohol for I minute. 6 Rinse in 95% alcohol for I minute. 7 Stain in trichrome solution for 5 minutes. 8 Rinse in 95% alcohol followed by two changes of absolute alcohol. 9 Clear in two changes of xylene and mount in Canada balsam. References ALBERTI A,, REALDI G . , TREMOLADA F. & CADROBBI P. (1975) HBsAg o n liver cell surface in viral hepatitis. Lancet i, 346 E
288
L.Gubetta et al.
DEODHAR K.P., TAPPE. & SCHEUER P.J. (1975) Orcein staining of hepatitis B antigen in paraffin sections of liver biopsies. Journal of Clinical Pathology 28, 66-70 GUBETTA L., RIZZETTO M. & VERMEG. (1976) A trichrome stain for the detection of hepatitis B surface antigen in liver biopsies. Pathologica 78, 413-416 GUDATF., BIANCHIL., SONNABEND W., THIELG., AENISHAENSLIN W. & STALDER G.A. (1975) Pattern of core and surface expression in liver tissue reflects state of specific immune response in hepatitis B. Laboratory Investigation 32, 1-9 HADZIYANNIS S., VISSOULISC., MOUSSOUROS A. & AFROUDAKIS A. (1972) Cytoplasmic localization of Australia antigen in the liver. Lancet i, 976-979 HADZIYANNIS S., GERBER M.A., VISSOULIS C. & POPPERH. (1973) Cytoplasmic hepatitis B antigen in 'ground-glass' hepatocytes of carriers. Archives of Pathology 96, 327-330 HUANGS.N. (1975) Immunohistochemical demonstration of hepatitis B core and surface antigens in paraffin sections. Laboratory Investigation 33, 88-95 LAMOTHE F., LAURENCIN-PICH~ J., CBTEJ., GUERINR., VIALLETA. & RICHER G. (1976) Detection of surface and core antigens of hepatitis B virus in the liver of 164 human subjects. Gastroenterology 71, 102-108 B., GALBRAITH R.M., EDDLESTON A.L.F., ZUCKERMAN A.J. & WILLIAMS R. (1976) PORTMANN Detection of HBsAg in fixed liver tissue-use of a modified immunofluorescent technique and comparison with histochemical methods. Gut 17, 1-9 RAY M.B., VAN DAMME B. & DESMET V.J. (1974) Evaluation of a modified fluorescent technique for the detection of Australia antigen in liver tissue. Journal of Immunological Methods 4,47-52 RAYM.B. & DESMET V.J. (1975) Inimunofluorescent detection of hepatitis B antigen in paraffin embedded liver tissue. Journal of lmmunological Methods 6, 283-289 RAYM.B., DESMET V.J., FEVERY J., DE GROOTEJ., BRADBURNE A.F. & DESMYTER J. (1976a) Distribution patterns of hepatitis B surface antigen (HBsAg) in the liver of hepatitis patients. Journal of Clinical Pathology 29, 94-100 RAYM.B., DESMET V.J., BRADBURNE A.F., DESMYTER J., FEVERY J. & DE GROOTE J. (1976b) Differential distribution of hepatitis B surface antigen and hepatitis B core antigen in the liver of hepatitis B patients. Gastroenterology 71,462-469 RIZZETTO M., BONINOF., VERME G., GUBETTA L., TOSELLIM., CANESEM.G. & BORGIALLI R. (1976a) The hepatitis B antigen asymptomatic chronic carriers. An immunological and histological study. Rendiconti di Gastroenterologia 8, 83-91 RIZZETTO M.,CRIVELLI O., SURIANI R. & VERME G. (rg76b) Intrahepatic localization of the surface (HBsAg) and core (HBcAg) antigenic determinants associated with hepatitis B virus in biopsy samples from patients with liver disease. La Ricerca in Clinice ed in Laboratoria 6, 41-54 SHIKATA T., UZAWA T., YOSHIWARA N., AKATSUKA T. & YAMAZAKI S. (1974) Staining methods of Australia antigen in paraffin section-detection of cytoplasmis inclusion bodies. Japanese Journal of Experimental Medicine 44, 25-36 THOMSEN P., POULSEN H. & PETERSEN P. ( I 976) Different types of ground glass hepatocytes in human liver biopsies; morphology, occurrence and diagnostic significance. Scandinavian Journal of Gastroenterology 11, I 13-1 19
I,
277-288
A trichrome stain for the intrahepatic localization of the hepatitis B surface antigen (HBsAg)
L . G U B E T T A , * M . R l Z Z E T T 0 , O.CRIVELL1, G.VERME & S.ARICO Departments of Pathology and Gastroenterology Ospedale Mauriziano Umberto I, Torino, Italy Accepted for publication
14
April 1977
GUBETTA L., RIZZETTO M., CRIVELLI O., VERMEG . & ARIC~) S. (1977) Histopathology I, 277-288
A trichrome stain for the intrahepatic localization of the hepatitis B surface antigen (HBsAg) A modified trichrome stain is described for the intrahepatic localization of the hepatitis B surface antigen; HBsAg containing cells exhibit specific green metachromasia contrasting with the granular brown colour of non infected hepatocytes and with the deep eosinophilic colour of ground glass cells of HBsAg-negative alcoholic or drug hepatitis. The technique is simple and reliable for routine screening of HBsAg positive material; its sensitivity is greater than H & E, similar orcein and inferior to immunohistochemistry as performed on frozen sections. Histological diagnosis can be made on the same slide, since several other morphological details are provided in the trichrome stained preparations. With this technique 387 biopsies from HBsAg seronegative individuals were negative; full cytoplasm metachromasia was mostly seen in asymptomatic HBsAg carriers, focal or partial staining in patients with histological evidence of liver cell necrosis. The presence and the staining pattern of HBsAg were of no help in predicting transition to chronicity or a transition from chronic persistent to chronic active hepatitis. Keywords : HBsAg, liver disease, trichrome stain, orcein stain, immunofluorescence
Introduction The localization of the hepatitis B surface antigen (HBsAg) in the liver provides valuable information to the histopathologist (Deodhar, Tapp & Scheuer 1975, Portmann, Galbraith, Eddleston, Zuckerman & Williams 1976), the viral origin of
* Address for correspondence : Dr L.Gubetta, Department of Pathology, Ospedale Mauriziano Umberto I, Cs. Turati 46, 10128 Turin, Italy. 277
278
L.Gubetta et ul.
the disease may become apparent only after immunochemistry (Ray, Desmet, Fevery, De Groote, Bradburn & Desmyter 1976a) and distinctive patterns of distribution of HBsAg are helpful in the diagnosis of different types of hepatitis (Gudat, Bianchi, Sonnabend, Thiel, Aenishaenslin & Stalder 1975) and in the recognition of the healthy chronic carrier status (Hadziyannis, Vissoulis, Moussouros & Afroudakis 1972, Rizzetto, Bonino, Verme, Gubetta, Toselli, Canese & Borgialli 1976a). Different methods are at present available to detect the antigen in the liver, each of them presenting a few disadvantages when used routinely in busy hospital laboratories. Sensitive and specific immunochemical techniques are demanding and require expensive equipment. Storing of positive fluorescent specimens and their retrospective examination is impossible for the transient nature of the phenomenon and endogenous background enzymatic activity may be troublesome after immunoperoxidase when the tissue is inflamed with many white blood cells. The recognition of HBsAg containing hepatocytes by their ground glass appearence after haematoxylin and eosin (H & E), (Hadziyannis, Gerber, Vissoulis & Popper 1973) depends on minimal variations in the tonality of the stain; similar appearances are sometimes due to fixation and staining artefacts and the identification of scattered or partially infected cells may be difficult (Portmrnn et al. 1976, Rizzetto et al. 1976a). The Shikata stains are more selective (Shikata, Uzawa, Yoshiwara, Akatsuka & Yamazaki 1974); artefacts, however, have been noted with orcein stain depending on the quality and age of the reagent (Thomsen, Poulsen & Petersen 1976, Rizzetto et al. 1976a) and aldehyde fucsin was noted to stain non-specifically any material containing disulphide bridges, like lipofuscin and glycoproteins taken up by Kupffer cells (Shikata et al. 1974, Thomsen et al. 1976). Both procedures provide no other histological information, apart from the localization of HBsAg, and additional slides of the same specimen must be stained with other dyes for diagnostic purposes. While studying liver collagen with several stains, a peculiar green metachromasia was noted in liver biopsies from HBsAg seropositive patients after the EA 31 Papanicolau solution. Further studies showed that this appearance was specifically related to the presence of intrahepatic HBsAg (Gubetta, Rizzetto & Verme 1976). During the last 12 months we have used routinely this modified trichrome stain on all the liver specimens sent to the pathology department. The technique, besides permitting the recognition of HBsAg containing hepatocytes, provides several other morphological details and enables the pathologist to make diagnostic histology on the same slide. Tn the present communication, this method is compared with the other techniques at present available to detect HBsAg; the results obtained are presented and their diagnostic significance is discussed.
Materials and methods Five hundred and forty-two biopsies were included in this study, 387 from HBsAg
Trichrome stain for HBsAg
279
seronegative patients with a variety of liver disorders and I 55 from HBsAg seropositive individuals; in the latter group histology showed acute hepatitis in 27 cases, a chronic persistent hepatitis (CPH) in 21, chronic active liver disease with or without nodular transformation (CALD) in 45, inactive cirrhosis (IC) in I 0, hepatocellular carcinoma in four, a normal liver or minimal abnormalities in 22. In a further group of 26 patients, the picture was compatible either with an acute hepatitis running a prolonged course (APH) or an early active chronic disorder. The first diagnosis was provisionally advanced in each case for the lack of anamnestic or biochemical evidence of chronic disease. A second biopsy after 6-12 months was obtained from the patients with APH, those with CPH and those with CALD without nodular transformation. Immediately after liver puncture each biopsy was cut in two portions, the longer (at least two-thirds of the whole specimen) was fixed in 4% formol-saline, the shorter was frozen in the cryostat.
HISTOLOGY
Formalin fixed specimens were stained with H & E, with the modified orcein stain of Shikata (Or.) (Shikata et al. 1974) and with a trichrome stain (Tr.) whose details are appended (Gubetta et al. 1976). IMMUNOCHEMISTRY
The frozen and formalin fixed sections of each biopsy were investigated by direct and indirect immunofluorescence (IFL) and by direct immunoperoxidase (IPS) for the presence of HBsAg. A fluorescein (FITC) and a peroxidase conjugated antiserum monospecific against HBsAg were prepared from the Beheringwerke rabbit precipitating antiserum RBB/r5; the preparations details and the specificity of the conjugates have been previously reported (Rizzetto et a/. 1976b). Standard indirect IFL was carried out using the anti HBsAg serum diluted 1/6 and a FITC conjugated antiserum against rabbit Ig obtained from goat (Beheringwerke TKF/o5). Direct IPS was performed as previously described (Rizzetto et al. 1976b). To assess the specificity and sensitivity of the trichrome stain serial formalin fixed sections of HBsAg positive biopsies were stained with Tr. and alternatively with Or., H & E and the fluorescent or peroxidase conjugated antisera.
Results SPECIFICITY AND SENSITIVITY OF TRICHROME STAIN
After trichrome stain, HBsAg containing hepatocytes displayed a homogeneous cytoplasmic metachromasia contrasting with the brownish colour of non infected
280
L.Gubetta et al.
Figure I. Ground glass hepatocytes. a Green cytoplasmic metachromasia after trichrome stain. Liver biopsy from an healthy HBsAg chronic carrier. Trichrome. x 400. b Deep eosinophilic colour after trichrome. Liver biopsy from an HBsAg-negative alcoholic patient. Trichrome. x 400.
Figure 2. Hepatocytes with green metachromasia. a Ground glass appearance after trichrome stain. Liver biopsy from a patient with chronic persistent hepatitis. Trichrome. x 400. b Section serial to a; the same hepatocytes with green metachromasia stained with a peridoxase conjugated antiserum against HBsAg. x 400.
Figure 3. Hepatocytes with green metachromasia. a Ground glass appearance after trichrome stain. Liver biopsy from a patient with chronic active hepatitis. Trichrome. x 400. b Section serial to a; the same hepatocytes with green metachromasia stained with a fluorescein conjugated antiserum against HBsAg. x 400.
Trichrome stain for HBsAg
283
cells (Figure la). Cytoplasmic inclusions of alcoholic patients and deposits of lipofuscin, biliary pigments, haemosiderin and glycogen did not stain green. Cytoplasmic inclusions of a,-antitrypsin deficiency, lipoidoses, mucopolyssccharidoses, Wilson’s disease and primary oxalosis were not investigated. Ground glass cells of alcoholic or drug hepatitis stained deeply red (Figure rb). The specificity of green metachromasia for HBsAg was confirmed by the positive reactions observed in corresponding cells in serial sections stained with orcein or with fluorescein or peroxidase conjugated monospecific antisera against HBsAg (Figure 2a & b, Figure 3a & b). Comparison of the sensitivity of Tr. with that of immunochemistry, Or. and H & E is reported in Table I. Table I. Number and percentage of specimens positive for HBsAg after trichrome, orcein, haematoxylin and eosin and immunofluorescence in I 55 biopsies from HBsAg seropositive patients, related to disease category
Diagnosis Asymptomatic chronic carriers (HCC) Acute hepatitis Acute prolonged hepatitis (AW Chronic persistent hepatitis (CPW Chronic active liver disease (CALD) Inactive cirrhosis (IC) Hepatocellular carcinoma
No. of No. and percentage of biopsies positive for HBsAg after biopsies IFL examined Trichrome Orcein H&E
22
16 (70%)
16 (70%)
6 (27.2%)
16 (70%)
27
26
I1
(42%)
I 1 (42%)
4 (15%)
16 (61%)
21
16 (76%)
16 (76%)
8 (38%)
14 (67%)
45
I0
(22%) 5 (50%)
10 (22%)
I (2%)
I7 (37%) 5 (50%)
I0
4
2
5 (50%) 2
2
Comparison of Tr. with Or. and H & E was performed on serial sections of the fixed part of the biopsy. Evaluation of Tr. versus immunochemistry was performed by comparing Tr. results in the fixed part with those in the frozen part of the same biopsy after IFL. Trichrome staining was not compared with immunochemistry on fixed material because in preliminary experiments occasional irregular and random denaturation of the antigen was noted after formalin fixation. Since indirect 1FL is considered the most sensitive method available for the recognition of HBsAg in the liver (Ray & Desmet 1975) and the fixed part of each biopsy was at least double the length of the frozen, negative Tr. and positive immunochemistry were regarded as being caused by the higher sensitivity of the latter wheress positive Tr. and negative immunochemistry was assumed to represent genuine absence of HBsAg in the smaller part of the biopsy.
284
L.Gubetta et al.
Trichrome sensitivity was lower than IFL, equal to Orcein and greater than H &E. Kupffer cells containing HBsAg were identified best after Tr. while they could be easily confused with freezing artefacts in immunofluorescence. Membrane staining was never observed after Tr. and only occasionally after iminunochemistry. Two biopsies were negative in IFL and positive after Tr.; they showed a persistent chronic hepatitis and a scattered distribution of HBsAg positive cells. Contrast was greatest after orcein and IFL; with these techniques, however, non-specific freezing or fixation artefacts could not be easily distinguished from genuine H BsAg, while after Tr., artefacts never exhibited green metachromasia and could be promptly recognized from deposits of the antigen. HBsAg
LOCALIZATION IN LIVER BIOPSIES
Metachromasia was never observed in the biopsies from 387 HBsAg-negative patients with a variety of liver disorders. No staining was seen in any of the 27 HBsAg seropositive patients whose biopsy showed acute hepatitis at its peak; none had circulating antigen after 6 months or developed chronic disease.
Table 2. Mean percentage of cells in the section exhibiting green metachromasia after Tr. and specific HBsAg staining by immunofluorescence in the fixed and frozen parts of the same positive biopsy; patients divided according to histological diagnosis
HCC
APH
CPH
24.6
10.5
19.2
44.2
25.5
30
CALD
IC
Hepatocellular carcinoma
20.5
15
37.5
30
~
Mean percentage of metachromatic cells in the formalin fixed part of the biopsy Mean percentage of fluorescent cells in the frozen part of the biopsy
5.2
18
HCC - healthy chronic carriers APH - acute prolonged hepatitis CPH - chronic persistent hepatitis CALD chronic active liver disease IC - inactive cirrhosis ~
The number of biopsies studied and the percentage of specimens positive with Tr., immunochemistry and other histological techniques are reported in Table I; the mean proportion of metachromatic and immunofluorescence positive cells observed in the fixed and in the frozen part of positive biopsies in different forms of hepatitis and in healthy asymptomatic HBsAg carriers is reported in Table 2 .
Trichrome stain for HBsAg
285
Kupffer cell staining was observed only in patients with chronic disease, with a greater incidence in active forms in which positive hepatocytes were fewer; in one biopsy from a patient with CALD only some of the macrophages exhibited a green colour while liver cells were negative. The intracellular pattern of HBsAg staining varied between the different groups ; in asymptomatic carriers staining was predominantly diffuse in the whole cytoplasm, in CALD it was usually limited to part of the hepatocyte with no constant topographical distribution within the cell, whilst in inactive disease both patterns were seen in variable combination. Focal HBsAg staining was also the predominant pattern in I I out of a group of 26 biopsies whose features of piecemeal necrosis, lobular inflammation and early fibrosis were compatible either with a late stage of an acute hepatitis or with early chronic disease ; an acute hepatitis running a protracted course was provisionally diagnosed for lack of anamnestic or biochemical evidence of chronic disease. After one year of follow-up, all Tr. positive patients with APH were still circulating the antigen; four developed chronic active hepatitis, four developed CPH, and three became chronic carriers with a normal liver on histology. The majority of Tr. negative patients with APH became HBsAg seronegative and asymptomatic within 6 months; five of them however, developed a chronic HBsAg seropositive liver disease. None of the Tr. positive or negative patients with chronic liver disorders became HBsAg seronegative after the year of follow-up. A second biopsy usually showed progression of the disease in patients with CALD while different pictures were observed in 16 patients with Tr. positive CPH: four of them developed CALD, eight still showed persistent hepatitis, two had liver fibrosis with portal scarring and two had histologically normal liver.
Discussion
Green metachromasia after trichrome staining was selectively confined to the cytoplasm of liver cells containing HBsAg, as shown by the correspondence of positive cells in serial section stained by immunochemistry (Figure 2a & b, Figure 3a & b). Metachromasia appeared to be specific for HBsAg ; cytoplasmic inclusions of the alcoholic, lipofuscin, haemosiderin, biliary pigments and glycogen did not stain green. Specimens from rare metabolic disorders were not available; even if positive, however, their frequency would be insignificant and in our experience green metachromasia is assumed to indicate viral disease. Ground glass cells of HBsAg-positive hepatitis could be easily distinguished from those of drug or alcoholic hepatitis by staining difference, the latter exhibiting a deep eosinophilic colour which contrasted sharply with green metachromasia (Figure Ia & b). Sensitivity of trichrome was higher than H & E, similar to orcein and lower than direct and indirect immunochemistry on frozen specimens; the use of frozen substrates, however, is unpractical, requiring splitting of the biopsy with a smaller amount available for histological diagnosis. Immunohistochemistry on fixed material appeared unreliable to us. Though HBsAg is reported to be unaffected (Ray, Van
286
L.Gubetta et al.
Damme & Desmet 1974) or only slightly diminished (Portmann et al. 1976) after fixation, and indirect imniunohistochemical techniques on fixed material were considered to be far more sensitive than orcein (Ray & Desmet i975), we found a poor correlation between fluorescence in frozen and in fixed parts of the same biopsy because of random and irregular denaturation of the antigen by formalin ; moreover, non specific staining was often increased by using both fixed tissue and indirect techniques as noted also by Huang (r975) and by Lamothe (Lamothe, LaurencinPiche, CGtC, Guerin, Viallet & Richer 1976). The contrast between positive and negative hepatocytes was similar after Tr. and Or. and greatest after fluorescence; with the latter technique however, freezing or fixation artefacts were difficult to distinguish from genuine HBsAg deposits, while they could be easily recognized after trichrome by their different staining. Contrast was less after peroxidase; variability of results and a strong background were often observed with this technique, which in our experience did not appear suitable for routine histochemistry. The trichrome has several advantages over the other methods employed to detect HBsAg in the liver. Preparations can be stored for later reviews, whereas IFL declines in a few days impeding retrospective studies. The technique is less demanding than immunohistochemistry and more reliable than orcein which can produce non-specific granular cytoplasmic and nuclear staining, depending on the quality and age of the dye (Rizzetto et al. 1976b). Histological diagnosis is not possible on IFL or orcein preparations and additional slides must be processed with other conventional stains; the trichrome, on the other hand, provides good morphological detail of parenchymal and connective tissue structures allowing simultaneous histological interpretation of the biopsy and also localization of HBsAg. An inverse relationship between the number of HBsAg positive cells and the amount of necrosis has been repeatedly noted in chronic diseases (Deodhar et al. I 975, Gudat et al. I 975, Hadziyannis et a/. I 972). Similarly, in this study, the percentage of positive biopsies and the proportion of cells exhibiting specific HBsAg staining in positive biopsies were lower in active than in inactive forms suggesting that the cellular expression of HBsAg is somehow related to protection from viral damage. The frequent overlap observed between the CPH and CALD groups made the finding of scanty intrahepatic HBsAg of limited value in differential diagnosis, only extensive involvement of whole liver lobules being reliably diagnostic of inactive or absent disease. Distinctive patterns of antigen distribution within the hepatocyte have also been correlated with various forms of hepatitis ; membrane (Alberti, Realdi, Tremolada & Cadrobbi 1975)and peripheral staining after IFL were considered typical of chronic active hepatitis and active cirrhosis, while focal cytoplasmic localization was found mostly in acute disease (Ray et al. 1976a). We were not able to identify patterns of metachromasia or fluorescence characteristic of any form of hepatitis. Membrane staining was not observed after trichrome, but was sometimes seen after TFL in the frozen part of the biopsy, implying that sensitivity of histological techniques is probably too low to detect a linear distribution of the antigen or that it might not
Trichrome stain for HBsAg
287
be stainable when present on the liver cell surface. Focal or partial staining of the cytoplasm without any constant topographical distribution within the hepatocytes was seen mostly in biopsies with histological evidence of disease. This pattern, however, was not useful in differentiating chronic active disease from persistent or acute, self-limiting, prolonged hepatitis. In our experience, the absence or the presence of HBsAg in the liver and its distribution patterns are only of limited value to the histopathologist, nonetheless, they provide useful indications complementary to other serological and histological data in the management of chronic HBsAg carriers. The intrahepatic localization of the HB core antigen is certainly a more sensitive and reliable indicator of disease than the surface antigen (Gudat et al. 1975, Ray et al. 1976b); its detection, however, is still limited at present to a few research centres. In most other laboratories, recognition of viral infection at a cellular level is possible only through the staining of the surface antigen. The trichrome stain described in this paper appears to be useful because of its simplicity and reliability and for the information it can provide on HBsAg in liver biopsies in busy hospital laboratories, often overrun by daily diagnostic routine. Appendix: method for trichrome stain
Paraffin sections cut at 4-5 pm. Reagents : I Harris’ Haematoxylin z Trichrome stain Light green (yellowish) Bismark brown Eosin (yellowish) Lithium carbonate Phosphotungstic acid
0.5 gin x IOO ml alcohol 95%, 45 ml 0.5 gm x roo ml alcohol 95%, 45 ml 0.5 gm x roo ml alcohol 95%, 45 ml 0.2 rnl saturated solution, 0.02 gm
Procedure : T Deparaffinize as usual to distilled water through two changes of xylene plus absolute and 95% alcohol. 2 Place in Harris’ Haematoxylin for 5 minutes. 3 Wash in distilled water. 4 Wash well in running water for 15 minutes. 5 Rinse in 75% alcohol for I minute. 6 Rinse in 95% alcohol for I minute. 7 Stain in trichrome solution for 5 minutes. 8 Rinse in 95% alcohol followed by two changes of absolute alcohol. 9 Clear in two changes of xylene and mount in Canada balsam. References ALBERTI A,, REALDI G . , TREMOLADA F. & CADROBBI P. (1975) HBsAg o n liver cell surface in viral hepatitis. Lancet i, 346 E
288
L.Gubetta et al.
DEODHAR K.P., TAPPE. & SCHEUER P.J. (1975) Orcein staining of hepatitis B antigen in paraffin sections of liver biopsies. Journal of Clinical Pathology 28, 66-70 GUBETTA L., RIZZETTO M. & VERMEG. (1976) A trichrome stain for the detection of hepatitis B surface antigen in liver biopsies. Pathologica 78, 413-416 GUDATF., BIANCHIL., SONNABEND W., THIELG., AENISHAENSLIN W. & STALDER G.A. (1975) Pattern of core and surface expression in liver tissue reflects state of specific immune response in hepatitis B. Laboratory Investigation 32, 1-9 HADZIYANNIS S., VISSOULISC., MOUSSOUROS A. & AFROUDAKIS A. (1972) Cytoplasmic localization of Australia antigen in the liver. Lancet i, 976-979 HADZIYANNIS S., GERBER M.A., VISSOULIS C. & POPPERH. (1973) Cytoplasmic hepatitis B antigen in 'ground-glass' hepatocytes of carriers. Archives of Pathology 96, 327-330 HUANGS.N. (1975) Immunohistochemical demonstration of hepatitis B core and surface antigens in paraffin sections. Laboratory Investigation 33, 88-95 LAMOTHE F., LAURENCIN-PICH~ J., CBTEJ., GUERINR., VIALLETA. & RICHER G. (1976) Detection of surface and core antigens of hepatitis B virus in the liver of 164 human subjects. Gastroenterology 71, 102-108 B., GALBRAITH R.M., EDDLESTON A.L.F., ZUCKERMAN A.J. & WILLIAMS R. (1976) PORTMANN Detection of HBsAg in fixed liver tissue-use of a modified immunofluorescent technique and comparison with histochemical methods. Gut 17, 1-9 RAY M.B., VAN DAMME B. & DESMET V.J. (1974) Evaluation of a modified fluorescent technique for the detection of Australia antigen in liver tissue. Journal of Immunological Methods 4,47-52 RAYM.B. & DESMET V.J. (1975) Inimunofluorescent detection of hepatitis B antigen in paraffin embedded liver tissue. Journal of lmmunological Methods 6, 283-289 RAYM.B., DESMET V.J., FEVERY J., DE GROOTEJ., BRADBURNE A.F. & DESMYTER J. (1976a) Distribution patterns of hepatitis B surface antigen (HBsAg) in the liver of hepatitis patients. Journal of Clinical Pathology 29, 94-100 RAYM.B., DESMET V.J., BRADBURNE A.F., DESMYTER J., FEVERY J. & DE GROOTE J. (1976b) Differential distribution of hepatitis B surface antigen and hepatitis B core antigen in the liver of hepatitis B patients. Gastroenterology 71,462-469 RIZZETTO M., BONINOF., VERME G., GUBETTA L., TOSELLIM., CANESEM.G. & BORGIALLI R. (1976a) The hepatitis B antigen asymptomatic chronic carriers. An immunological and histological study. Rendiconti di Gastroenterologia 8, 83-91 RIZZETTO M.,CRIVELLI O., SURIANI R. & VERME G. (rg76b) Intrahepatic localization of the surface (HBsAg) and core (HBcAg) antigenic determinants associated with hepatitis B virus in biopsy samples from patients with liver disease. La Ricerca in Clinice ed in Laboratoria 6, 41-54 SHIKATA T., UZAWA T., YOSHIWARA N., AKATSUKA T. & YAMAZAKI S. (1974) Staining methods of Australia antigen in paraffin section-detection of cytoplasmis inclusion bodies. Japanese Journal of Experimental Medicine 44, 25-36 THOMSEN P., POULSEN H. & PETERSEN P. ( I 976) Different types of ground glass hepatocytes in human liver biopsies; morphology, occurrence and diagnostic significance. Scandinavian Journal of Gastroenterology 11, I 13-1 19
