serological and histological diagnosis of primary biliary

J. clin. Path. (1966), 19, 527

Serological and histological diagnosis of primarybiliary cirrhosis

R. B. GOUDIE1, R. N. M. MACSWEEN, AND D. M. GOLDBERG

From the Western Infirmary, Glasgow

SYNOPSIS A simple immunofluorescence test for antibody to a mitochondrial antigen present inmany tissues is a reliable method of distinguishing most cases of primary biliary cirrhosis fromjaundice due to extrahepatic biliary tract obstruction. Of 30 cases diagnosed as primary biliarycirrhosis, 26 had antimitochondrial antibody whereas none of 77 cases with jaundice due to extra-hepatic bile duct obstruction showed this serological abnormality. The antibody was also found inthe serum of three of 42 patients who had other forms of cirrhosis and in two of 266 patients withno evidence of liver disease.

Clinical, biochemical, and serological findings favour the view that primary biliary cirrhosis is areal entity which, in our present state of knowledge, cannot be defined clearly by any single methodof investigation. In particular, the liver may show a variety of histological appearances which,interpreted without regard to the other features of the case, may lead to errors in diagnosis.

Primary biliary cirrhosis is a rare condition of un-known cause characterized clinically by pruritus,hepatomegaly, andjaundice of obstructive type in theabsence of extrahepatic biliary tract obstruction andwithout pain or fever. Pathologically the best knownfeatures are a fine 'monolobular' cirrhosis of theliver, pericholangitis affecting the medium-gized bileducts in the portal tracts, and disappearance ofmanyof the small portal bile ducts; in the early stages truecirrhosis may not have developed, and in the latestages the appearances of the liver may be indi-stinguishable from those ofportalcirrhosis (Sherlock,1963).In 1965, Walker, Doniach, Roitt, and Sherlock

reported that antimitochondrial antibody wasdemonstrable by means of an indirect immuno-fluorescence technique in the serum of all of 32patients with primary biliary cirrhosis. The antibodywas not found in patients with extrahepatic biliarytract obstruction. These findings support the conceptthat primary biliary cirrhosis as described above is anosological entity, and they may shed some light onthe causation of this obscure condition; they alsoprovide the basis for a simple test to differentiateprimary biliary cirrhosis from other forms ofobstructive jaundice, and so may prevent undesirablesurgical exploration of the extrahepatic biliaryISupported by a grant from the ScottishHome and HealthDepartmentReceived for publication 13 July 1966

passages. We have repeated the studies of Walkeret al. (1965) and in this paper report partial confirma-tion of their findings. In addition we draw attentionto the frequency of misleading reports on liverbiopsy material in primary biliary cirrhosis andemphasize some of the difficulties in making anaccurate histological diagnosis.

CASES STUDIED

Fluorescent antibody tests were carried out on theserum of 204 patients with jaundice due to a varietyof causes, on five patients with primary biliarycirrhosis who did not have jaundice, and on 266patients without jaundice or other evidence of liverdisease.

PRIMARY BILIARY CIRRHOSIS (30 CASES) In view of thedifficulty in defining rigid diagnostic criteria, themain clinical and biochemical features of the 30cases are given in Tables I and II. In 24 cases thediagnosis of primary biliary cirrhosis had been madebefore serological testing. The diagnosis of threepatients thought to have portal cirrhosis (cases 2, 24,and 25) and of three patients thought to have lupoidhepatitis (cases 1, 9, and 26) was later changed afterserological testing and review of the clinical andbiochemical findings.

527

copyright. on January 6, 2022 by guest. P

rotected byhttp://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.19.6.527 on 1 Novem

ber 1966. Dow

nloaded from

http://jcp.bmj.com/

R. B. Goudie, R. N. M. Macsween, and D. M. Goldberg

TABLE ICLINICAL DETAILS OF PATIENTS WITH PRIMARY BILIARY CIRRHOSIS OR MITOCHONDRIAL ANTIBODY

Age Sex Duration Pruritus Xanthoma Titre ofAntibody Titre of AntibodyComplement-fixation ImmunofluorescenceTest

Evidence againstExtrahepaticBiliary Obstruction

Primary Biliary Cirrhosis1 54 F2 67 F3 54 F

4S67891011121314151617181920

6764433745605143555967494272385644

FFFFFFMFFFFFFFFFF

5 yr.3 yr.

10 yr.

3 yr.1 yr.4 yr.3 yr.1 yr.6 yr.10 yr.6 mth.6 mth.3 yr.9 mth.7 yr.2 yr.3 yr.1 yr.1 yr.14 yr.

21 50 F 2 yr.22 41 F 9 yr.

232425262728

504651545165

FFFFFF

11 yr.5 yr.3 yr.5 yr.2 yr.2 yr.

29 58 F 6 yr.30 56 F 2 yr.

Portal and Postnecrotic Cirrhosis31 48 F 3 yr.32 56 F 8 yr.33 39 F 15 yr.

Controls34 67 M -35 48 F -

PORTAL AND POSTNECROTIC CIRRHOSIS (42 CASES)Histological evidence of portal or postnecroticcirrhosis was obtained in 29 cases; the remaining13 subjects had (1) a clinical history of chronic liverdisease with recurrent jaundice of at least one year'sduration; (2) portal hypertension, ascites, orepisodes of hepatic coma; (3) abnormal serumturbidity and flocculation tests; (4) serum alkalinephosphatase below 35 King-Armstrong (K.A.)units/100 ml.; (5) serum transaminase activitiesconsistent with cirrhosis as described by Wr6blewski(1958).

EXTRAHEPATIC BILLARY TRACT OBSTRUCTION (72CASES) AND SECONDARY BILIARY CnuRosis (5 CASES)Fifty-eight patients had operative or necropsyevidence of extrahepatic biliary tract obstruction dueeither to stone or tumour. In five there was histo-

logical evidence of secondary biliary cirrhosis. Thebiliary tract was not explored in 19 who had clinicaland biochemical evidence of obstructive jaundice; ofthese, 13 were known to have carcinoma, and theremaining six had gallstones or a non-functioninggallbladder demonstrated radiologically.

INFECTIVE HEPATmS (25 CASES) All had typicalfeatures of the disease, including a prodromal illnessand an acute attack ofjaundice lasting not more thansix weeks. In 20 the serum glutamic pyruvate trans-aminase (S.G.P.T.) was at least 1,000 units/ml.; inthe other five, who were all under 23 years old, thetransaminase levels estimated late in the illness werenot diagnostic.

DRUG-INDUCED JAUNDICE (10 CASES) This consistedof transient jaundice following administration of

528

CaseNo.

+

+++±+

+±++

+

Xanthela

Xanthelasma+

Xanthelasma++

+

XantheLasma

2,0482,048256

256256128512256

1282,0481,024

>4,0961,0001,0001,000256512512

1,000

2,048500

5004,096

<816

<8<8

2,0482,0482,048

1281,024128512128

1,024128

>8,192128512

2,048128

> 8,192512512512512

1,0242,048

128128

1,024256

Neg.Neg.

Neg.Neg.

1282,048512

Needle biopsyNecropsyLaparotomy,cholangiogramNecropsyNeedle biopsyLaparotomyLaparotomyNeedle biopsyNeedle biopsyLaparotomyNeedle biopsy

Needle biopsyLaparotomyNeedle biopsyNeedle biopsyLaparotomyCholangiogramLaparotomyLaparotomy,cholangiogramCholangiogramLaparotomy,cholangiogram

NecropsyNeedle biopsyNeedle biopsyLaparotomyCholangiogramLaparotomy,cholangiogramLaparotomy

NecropsyLaparotomy

+

642,048

>4,096

500128

32512



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/

529

TABLE IIBIOCHEMICAL FINDINGS IN PATIENTS WITH PRIMARY BILIARY CIRRHOSIS OR MITOCHONDRIAL ANTIBODY1

Case No. Serum Bilirubin Serum Alkaline Thymol Turbidity SGOT SGPT Serum(mg./100 ml.) Phosphatase (units) (units/ml.) (units/ml.) Cholesterol

(KA. units/100 ml.) (mg./100 ml.)

Primary Biliary Cirrhosis1 3-82 5-43 8-14 13-45 3-06 7-67 20-08 2-89 1-610 5-711 50-0212 2-513 1-714 0-915 5-216 0-717 15-618 0-719 24-020 1-121 2-822 26-023 22-824 6-225 0-826 3-127 11-228 13-629 16-030 1-6

Portal and Postnecrotic Cirrhosis31 32-832 2-233 2-3

Controls3435

040-5

120626296699222080857010011543867076809715410850

20890902746100140136105

203035

1612

181115912485941417101062446

119122510121314233

26

13S

8

4

125117867038

140260302619618013898496036104115200807298

2151031321961681044146

49628124

1415

786741404660

2206025112202655014584274186250150234180155705682

24822025022

4021245

1213

233410684170130

1,1901,435250355540936

1,694370323285360480505

1,650415300675

1,650473240264630885

1,250347

157195165

310310

'Maximum recorded values.'Isolated finding.

drugs known to be hepatotoxic: para-aminosali-cylic acid (4 cases), chlorpromazine (3), imipramine(1), phenindione (1), and alcohol (1). In eight, liverbiopsy showed appropriate histological changes.

HAEMOLYTIC JAUNDICE (4 CASES) The patients hadtypical haematological and biochemical features.

MISCELLANEOUS FORMS OF JAUNDICE (21 CASES) Theseincluded recurrent intrahepatic cholestasis (1 case),primary shunt hyper-bilirubinaemia (1), Gilbert'sdisease (1), jaundice associated with perniciousanaemia (1), cirrhosis complicating ulcerative colitis(1), jaundice associated with muscular dystrophy (1),jaundice due to chronic venous congestion of theliver (4), lupoid hepatitis (4), and seven cases in whicha definitive diagnosis has not yet been made.

PATIENTS WITHOUT JAUNDICE General hospital

patients included 183 patients with a variety ofdiseases routinely admitted to a general medical unit,and one selected patient (case 34) noted to givegranular staining of thyroid epithelium (see below)in a test for thyroid microsomal antibody. In 20patients with severe chronic thyroiditis the diagnosiswas confirmed by a precipitin reaction with thyro-globulin. Nineteen patients with pernicious anaemia,22 with rheumatoid arthritis, 18 with systemic lupuserythematosus, and three with idiopathic hyper-lipaemia and xanthomatosis were also tested.

METHODS

IMMUNOFLUORESCENCE TECHNIQUE Blocks of renal cortexfrom freshly killed rats were frozen on to metal chuckswith CO2 snow, and 6,u sections were cut in a cryostat.Serum to be tested was diluted 1 in 8 in normal salinebuffered with veronal (pH 7 2) and applied to the unfixed

Serological and histological diagnosis ofprimary biliary cirrhosis



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


sections for 30 minutes at 20°C. After washing withbuffered saline for 10 minutes fluorescein-conjugatedantihuman globulin serum (Burroughs Welcome andCompany) was applied for 30 minutes and after a final10-minute wash in buffered saline the sections weremounted in buffered glycerol and examined with a Gillettand Sibert conference microscope using blue light. Toreduce non-specific fluorescent staining, the fluorescein-conjugated antihuman globulin serum had been previous-ly absorbed with cell fragments (obtained by centrifuga-tion at 35,000 g for one hour at 4°C.) of rat kidneyhomogenized in buffered saline. Sera which gave positiveresults when diluted 1 in 8 were titrated in fourfolddilutions in parallel with similar dilutions of normalcontrol serum. All tests were read independently by twoobservers who were unaware of the nature of the serabeing tested.

COMPLEMENT-FIXATION TESTS These were performed onall but three of the sera giving immunofluorescent stainingof rat kidney epithelium. Guinea-pig complementpreserved in Richardson's fluid, sensitized sheep red cells.standardized by the method of Osler, Strauss, and Mayer(1952) and complement-fixation buffer (Oxoid) were used.The antigen was prepared by homogenizing rat liver in0-25 M-sucrose at 4°C. with a teflon pestle; after remov-ing nuclei and cell debris by centrifugation at 1,500 g for15 minutes, the crude antigen was made by suspending incomplement-fixation buffer the deposit obtained bycentrifugation at 35,000 g for 60 minutes at 4°C.

Tests were performed in perspex trays using 0-1 ml. ofdoubling dilutions of serum inactivated by heating at56°C. for 30 minutes, 0-2 ml. of complement containing5 HD50, and 0-1 ml. of a suitable dilution of antigen(usually the button obtained from 1 g. of liver in 40 ml. ofbuffer). After one hour at 37°C. with frequent shaking,0-1 ml. of 2% sensitized sheep cells was added and thetrays incubated for a further hour at 37°C., then left over-night 4°C. at before reading. Anticomplementary activityof serum was assessed by testing serum dilutions in thepresence of 3HD50 of complement with 0-1 ml. of bufferinstead of antigen; 50% haemolysis judged visually wastaken as the end point.

RESULTS

Figure 1 shows positive and negative immuno-fluorescence tests on rat kidney. In the positive testwith serum from a patient with primary biliarycirrhosis the cytoplasm of the renal tubular epithelialcells shows brilliant fluorescence and the glomeruliappear as dark areas in the section; with normalserum there is faint general staining of all the renalstructures, sometimes with brighter staining of theluminal margins of the tubular cells and of castswithin the tubules. These tests have been performedwith sera diluted 1 in 8, since undiluted normalserum causes considerable immunofluorescent stain-ing which is difficult to distinguish from a true posi-

FIG. 1. Frozen section of rat kidney stained by indirectimmunofluorescence. (a) With primary biliary cirrhosisserum. There is brilliant granular fluorescence (white) ofthe cytoplasm of renal tubular epithelium. The two largedark spaces are unstained glomeruli. (b) With normalserum. There is diffuse faint non-specific staining oftubules and glomeruli (top right and bottom lzft).

530

or



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


-a4s*.,. > >

b b __

O * 1 x

^.l=3

_h

!- _a

F-| 9E _:s.. S.' Es

_~~~~~~~%oh".:ii/£om-. * ~

FIG. 2a FIG. 2b FIG. 2cFIG. 2. Frozen section of human thyroid stained by indirect immunofluorescence. (a) With primary biliary cirrhosisserum. There is distinctly granular staining of the epithelium lining the follicles. (b) With thyroiditis serum. The epithelialcytoplasm is uniformly stained. (c) With normal serum. The epithelium is virtually unstained apartfrom very weak nuclearfluorescence.

tive reaction. (In case 23, due to unusually strongnon-specific staining at a dilution of 1 in 8, anobviously positive result was obtained only afterdiluting the serum 1 in 32.)

Sera reacting positively with rat kidney were alsoshown to give positive results with similarly pre-pared sections of rat liver, brain and adrenal, humangastric mucosa, adrenal cortex and medulla, liver,and thyroid. As noted by Walker et al. (1965), thestaining of human gastric mucosa closely resemblesthat seen with the majority of pernicious anaemiasera, whereas the staining of thyroid epithelium has amuch coarser granular appearance (Fig. 2) than thatobtained with the thyroid-specific microsomalantibody found in thyroiditis serum.

Details of antibody titres in patients with positiveimmunofluorescence tests are shown in Table I andthe results of all immunofluorescence tests are sum-marized in Table III. Antibody was demonstrated in26 of 30 (87 %) cases with primary biliary cirrhosis;in the positive cases titres obtained by immunofluor-escence were not less than 1/128 and most of thetitres were much higher. Antibody was found in threepatients (7°%) with portal or postnecrotic cirrhosis.Case 31 had postnecrotic cirrhosis demonstrated atnecropsy three years after a clinically typical attackof infective hepatitis, case 32 had histological evidenceof portal cirrhosis, and case 33, with hepato-splenomegaly, bled from oesophageal varices 15years after an attack ofjaundice. The sera of cases 34and 35 gave a granular staining pattern of thyroidepithelium during tests for thyroid microsomal anti-

TABLE IIIIMMUNOFLUORESCENCE WIrH RAT KIDNEY IN JAUNDICED

PATIENTS AND CONTROLS

No. Tested No. Positive

Jaundiced PatientsPrimary biliary cirrhosisPortal and postnecrotic cirrhosisExtrahepatic biliary tract obstructionSecondary biliary cirrhosisInfective hepatitisDrug-induced jaundiceHaemolytic jaundiceMiscellaneous

Patients without jaundiceGeneral hospital patientsChronic thyroiditisPernicious anaemiaRheumatoid arthritisSystemic lupus erythematosusIdiopathic hyperlipaemia andxanthomatosis

3042725

25104

21

18420192218

26 (87%)3 (7%)000000

1 (05%)001 (<5%)0

0

body, and were subsequently tested with rat kidney.Case 34 had osteoarthritis, chronic bronchitis, and anE.S.R. of 48 mm. in one hour, and case 35 hadclassical rheumatoid arthritis. Both have slightlyraised serum cholesterol levels. These two speciallyselected cases have been included in the appropriatecontrol groups in Table III, and give an exaggeratedincidence of positive tests in general hospital patients(0 5 %) and in cases of rheumatoid arthritis (5%).Positive tests were not found in the other jaundicedand control groups studied.

Figure 3 illustrates positive and negative comple-ment-fixation reactions using the 35,000 g sediment

531



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


*m-.... e .......... + .e'^^<

>e ::

A.

,zI At e

,;......R

*:.*. 4 * .t 4: M.; ';

*~~~~~ ~~~~ ~~96 .: :a a-

0...

1-i 0: JOY

a a * *wr -t. :

i* * * * * * A3) .

HIG. 3. Complement-fixation test with sera from primary biliary cirrhosis. The control rows contain serum, 3HD50 ofcomplement and no antigen (test for serum anticomplementary activity). The test rows contain serum, 5HD50 of comple-ment and antigen preparedfrom rat liver. Tests 21 and 22 are negative, and 23 and 24 are positive (titre 256).

of rat liver homogenate as antigen. Details of titresobtained are shown in Table I. Positive complement-fixation reactions were found in those patients whoseserum gave immunofluorescent staining of ratkidney epithelium. The correlation coefficient (r)between log2 immunofluorescence and complement-fixation titres is 0 72 (t =- 804, p < 0 001).

HISTOLOGICAL STUDIES

The following description is based on the appearanceof paraffin sections stained by haematoxylin andeosin of the eight wedge biopsies of liver obtained atlaparotomy (cases 3, 7, 10, 14, 17, 20, 22, and 28) andon the four necropsy cases (cases 2, 4, 17, and 23)which were available to us for study.

EVIDENCE OF CIRRHOSIS Well-developed mono-lobular cirrhosis with fibrous bands regularly extend-ing round liver lobules to link adjacent portal tracts

was seen in one of the wedge biopsies and in two ofthe four necropsy cases, while monolobular cirrhosiswith linking of only some of the portal tracts wasseen in three wedge biopsies. In three biopsies therewere fibrous bands extending out from the portaltracts (portal tract fibrosis) but continuity betweenadjacent tracts was not established. The cirrhosis wasof distinctly multilobular type in three cases, all ofthree years' duration or longer.

CHOLESTASIS Cholestasis, as shown by bile thrombiand/or intracellular granules of bile pigment, waspresent in only five wedge biopsies but was a con-sistent feature in the necropsy material.

CELLULAR INFILTRATION OF PORTAL TRACTS This

was present in every case, and consisted mainly oflymphocytes with a variable number of plasma cells.The infiltrate was diffuse, frequently with nodules ofdensely aggregated cells. The infiltrating cells,

r

532



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


together with fibroblasts, frequently invaded thelimiting plate of liver cells round the portal tractsgiving an eroded appearance. A few eosinophil poly-morphs were present in the portal infiltrate in mostcases, and in a third of the cases small numbers ofneutrophil polymorphs were noted.

SMALL INTRAHEPATIC BILE DUCTS Those less than50 ,u in diameter were markedly reduced in numberin 10 of our 12 cases (Fig. 4). In four of these therewas focal proliferation of tubular structures simulat-ing bile duct proliferation and composed of flateosinophilic cells lying near the hepatic lobules andpossibly derived therefrom. Marked proliferation ofbile ducts was observed in case 23, which was of 11years' duration (Fig. 5).

MEDIUM SIZED INTRAHEPATIC BILE DUCTS Thosebetween 50 and 100 p. in diameter showed peri-cholangitis in six of the seven wedge biopsies. Thiswas characterized by accumulation of lymphocytesand plasma cells around these structures and byinfiltration of the bile duct epithelium by lympho-

cytes, with swelling, eosinophilia, and heaping-up ofthe epithelium and absence of small branches (Figs.6 and 7). This feature was identified in only one of thefour necropsy cases possibly due to difficulties im-posed by post-mortem autolysis.

LIVER CELL DEGENERATION Liver cell degenerationof feathery type was common, especially in areasshowing erosion of the limiting plate. Areas of focalliver cell necrosis within the lobules, and usuallysurrounded by polymorphs, were seen in six of thewedge biopsies. The frequency of this change in thenecropsy cases is uncertain because of post-mortemautolysis. Mild fatty change was seen in half the totalcases.

GRANULOMATA Localized aggregates of endothe-lioid cells resembling sarcoid follicles were found inthe portal tracts and in the hepatic parenchyma in twoof the wedge biopsies (Figs. 8 and 9). In two otherwedge biopsies less clearly demarcated clusters ofendothelioid cells as described by Greville Williams

b~~~~~~~~~~~~~~fi A-L ' I tAt'Lk q4 ~

FIG.4.Portal tract in primaiy~~biliArycrhssMhwigfboi,fca n ifs ymhctcadpasacl nta

don,4erosionao thecliiingpiate ofilivery celloss,andoabsnceofibroile ducts.aneathfue liverhobules are pafwsmacll tubultarstructures, probably derived from liver cells (haemzatoxylin and eosin x 135).

533



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


FIG. 6

FIG. 5

FIG. 5. Proliferation of snmall bile ducts in fibrous tissuein primary biliary cirrhosis of long duration (haematoxylinand eosin x 135).

FIG. 6. Medium sized bile duct in primary biliarycirrhosis. There is massive lymphocytic and plasma cellinfiltration of the portal tract and these cells have invadedthe bile duct. There is epithelial degeneration in the leftlower part of the duct. Note the absence of small branches(haematoxylin and eosin x 150).

FIG. 7. Medium sized bile duct in primary biliarycirrhosis. The heaped-up epithelium is infiltrated bylymphocytes and plasma cells as is the periductal tissue.There is an ill-defined cluster of endothelioid cells at thetop right hand corner (Masson x 370).

wt.. ,P % *ft

&I..% qmb.a 'lb No0

OF .,* 0404 -, t.

ir - lb Al.t 0

FIG. 7

534



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


.FIGis.9

FIG. 9

FIG. 8. Primary biliary cirrhosis. Sarcoid-like folliclecomposed of endothelioid and small giant cells in a portaltract (haematoxylin and eosin x 205).

FIG. 9. Primary biliary cirrhosis. Sarcoid-like folliclewithin liver lobule (haematoxylin and eosin x 205).

FIG. 10. Primary biliary cirrhosis. Necrotic focus withLanghans giant cell and endothelioid cells, resemblingtuberculous follicle (haematoxvlin and eosin x 205).

FIG. 10

FIG. 8

535



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


(1965) were seen in the portal tracts (see Fig. 7). Inone necropsy case necrotic follicles with Langhanstype giant cells were noted (Fig. 10); no tuberclebacilli were demonstrated in sections from this liver,which was unfortunately the only organ available forexamination.

DISCUSSION

Our finding that 87% of patients diagnosed as casesof primary biliary cirrhosis and a small proportion ofpatients with other types of cirrhosis have circulatingantibody which is readily demonstrable by granularimmunofluorescent staining of rat kidney epitheliumlargely confirms the observations of Walker et al.(1965).Complement fixation between primary biliary

cirrhosis sera and tissue extracts has previously beendescribed (Gajdusek, 1958; Deicher, Holman, andKunkel, 1960; Walker et al., 1965) and our demon-stration of a significant correlation between comple-ment-fixation titres with rat liver extract andimmunofluorescent staining of rat kidney suggeststhat both methods may broadly reflect the sameantigen-antibody systems. The occasional finding ofdisparity between titres by the two methods may be aconsequence of the indefinite end point observed withtitrations, especially by the immunofluorescencetechnique, or may indicate some variation in thespecificity of the antibody in individual patients.Beyond confirming absence of organ and speciesspecificity, we have nothing to add concerning thenature of the antigen concerned in these reactions.

Following the studies of Ahrens, Payne, Kunkel,Eisenmenger, and Blondheim (1950) primary biliarycirrhosis has been established as a clinical entitythough it is difficult to define it precisely. Probablythe lack of a precise clinical definition accounts forour finding of antibody in only 870% of cases ofprimary biliary cirrhosis in contrast to the 10000reported by Walker et al. (1965). Our sero-negativecases are indistinguishable from those with positiveantibody tests on the clinical, biochemical (Tables Iand 11), and histological evidence available to us. Itremains to be seen whether the sero-negative casesare fundamentally different from those with anti-body, or are suffering from essentially the samedisease in which antibody formation is merely acommon epiphenomenon.The finding of antibody in occasional cases of

portal or postnecrotic cirrhosis may indicate thatthese cases are atypical variants of primary biliarycirrhosis (e.g., case 33 which has a relatively highserum alkaline phosphatase of 35 K.A. units/100 ml.)or may be taken as further evidence against thefundamental importance of antibody formation inthe pathogenesis of primary biliary cirrhosis.

Experimental findings favour the latter view sinceliver damage by carbon tetrachloride causes anti-mitochrondrial antibody formation in rats, but notthe development of biliary cirrhosis (Weir, 1963); andin diseases in which autoimmunity is believed to havea pathogenic role, e.g., experimental allergicencephalomyelitis or thyroiditis, sensitized lympho-cytes (delayed hypersensitivity) and not circulatingantibody are thought to be the harmful agents(Paterson, 1960; Felix-Davies and Waksman, 1961).Whether or not antibody is of importance in the

pathogenesis of primary biliary cirrhosis, it is clearthat the immunological test is an important newparameter in the study of human liver disease. Itstrongly supports the view that, by and large, primarybiliary cirrhosis as described in clinical and bio-chemical terms is a real entity. Furthermore itprovides a useful confirmatory test for distinguishingthe majority of cases from conditions, such as extra-hepatic biliary tract obstruction in which surgicalexploration is called for, and from drug-inducedcholestatic jaundice.The biochemical features observed in the present

subjects conform to the pattern usually associatedwith primary biliary cirrhosis (Ahrens et al., 1950;Sherlock, 1959). This includes disproportionatelyhigh levels of serum alkaline phosphatase relative tothe frequently mild degree ofjaundice. The very highbilirubin values recorded for cases 4 and 11 weresporadic or terminal phenomena, whereas the alka-line phosphatase was never very much lower than thehighest observed levels documented in Table II.Thymol turbidity was usually abnormal inestablishedcases, though not infrequently normal on firstpresentation. In general, values for this test werehigher than those encountered in extrahepaticobstruction, though it was uncommon to find suchdramatic elevations as are seen in a high proportionof cases with portal cirrhosis. On the other hand,transaminase activities were usually higher than thosemet with in portal cirrhosis with comparable bili-rubinaemia. Unlike the pattern most often en-countered in non-malignant extrahepatic obstruction,activity of serum glutamic oxalacetic transaminase(SGOT) was frequently higher than that of SGPT inthe same specimen. In many cases there was a slightor moderate elevation in serum globulin, but com-pared with typical cases of portal cirrhosis, thealbumin was only slightly depressed. The serumcholesterol level was raised in 25 cases and, wheremeasured, serum phospholipids were also elevated.These abnormalities need not be associated withdermal xanthomata and should always be sought inpatients with long-standing jaundice since they aremost useful indices for distinguishing primarybiliary from other forms of cirrhosis.

536



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/


The pathological changes in the liver in primarybiliary cirrhosis seem to be less widely known thanthe clinical features. In the cases described in thispaper several incorrect histological diagnoses wereinitially made which misled the clinicians; theseincluded portal cirrhosis, cryptogenic cirrhosis,continuing hepatitis, lupoid hepatitis, and cholan-gitis. These errors undoubtedly arise from un-familiarity with this rather uncommon disease, fromattempted diagnosis on biopsies too small to revealthe characteristic changes in the medium-sized andsmall bile ducts, from variation in liver histologydepending on the duration of the disease, and fromthe conflicting accounts of the disease in the litera-ture. We can confirm Sherlock's statements (1963)that needle biopsy is useful mainly for the exclusionof extrahepatic biliary tract obstruction in whichthere is marked proliferation of the intrahepatic bileducts (Fig. 11), and that instead of the well-knownmonolobular cirrhosis there may be merely portaltract fibrosis or frank portal cirrhosis in patients withprimary biliary cirrhosis. The statement by Cameronand Hou (1962) that bile thrombi are invariablypresent is apparently based on necropsy material; asemphasized by Popper, Rubin, and Schaffner (1962)they may be absent, particularly in the early stage ofthe disease. Undoubtedly the most characteristichistological lesion of primary biliary cirrhosis is thereduction in number of the small intrahepatic bileducts, but focal proliferation of structures resemblingvery small bile ducts is not uncommon, and in thelate stages, when portal cirrhosis is established, bileduct proliferation may be widespread. The otherfindings favouring a positive diagnosis of primarybiliary cirrhosis is the pericholangitis of medium-sized intrahepatic bile ducts described by GrevilleWilliams (1965), though this lesion is often difficultto find and is sometimes absent. The presence ofgranulomata resemblingsarcoid follicles has previous-lv been noted by Rubin (1963), and Scheuer (1966)has drawn our attention to his finding of theselesions in about one third of cases of primary biliarycirrhosis.The infiltration of portal tracts by lymphocytes

and plasma cells, the proliferation of fibrous connec-tive tissue, and the alteration of epithelial structuresfound in the liver in primary biliary cirrhosis may beregarded as analogous to the lesions of autoimmunethyroiditis and gastritis. Localized granulomaformation is not, however, a feature of the lesionsassociated with organ-specific autoimmunity. Granu-lomata composed of endothelioid cells are sometimesfound in the lesions of ulcerative colitis (Ackerman,1959) and of systemic lupus erythematosus (Teilum,1946) and on morphological grounds there might be

FIG. 11. Secondary biliary cirrhosis showing floridproliferation of small bile ducts (haematoxylin and eosinx 135).

reason to suspect that primary biliary cirrhosis has apeculiarity of immunological reactivity in commonwith these diseases.

We wish to record our thanks to the following physicianswho supplied us with sera: Drs. H. T. Howat, W. J.Cunliffe, and J. P. O'Loughlin, Manchester RoyalInfirmary; Dr. G. P. Lewis, Stobhill Hospital, Glasgow;Drs. A. G. Melrose and R. Hume, Southern GeneralHospital, Glasgow; Drs. A. Lyell and J. A. Boyle, RoyalInfirmary, Glasgow. Dr. I. Roitt of the MiddlesexHospital, London, supplied us with our original positivecontrol sera. Mr. E. MacDonald and Miss M. Barr gavevaluable technical assistance.

REFERENCES

Ackerman, L. V. (1959). Surgical Pathology, 2nd ed., p. 383. Mosby,St. Louis.

Ahrens, E. H., Jr., Payne, M. A., Kunkel, H. G., Eisenmenger, W. J.,and Blondheim, S. H. (1950). Medicine (Baltimore), 29, 299.

Cameron, R., and Hou, P. C. (1962). Biliary Cirrhosis, p. 149. Oliverand Boyd, Edinburgh and London.

Deicher, H. R. G., Holman, H. R., and Kunkel, H. G. (1960). Arthr.and Rheum., 3, 1.

537



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/

R. B. Goudie, R. N. M. Macsweein, and D. M. Goldberg

Felix-Davies, D., and Waksman, B. H. (1961). Ibid., 4, 416.Gajdusek, D. C. (1958). Arch. intern. Med., 101, 9.Osler, A. G., Strauss, J. H., and Mayer, M. M. (1952). Anmer. J.

Syph., 36, 140.Paterson, P. Y. (1960). J. exp. Med., 111, 119.Popper, H., Rubin, E., and Schaffner, F. (1962). Amer. J. Med., 33,

807.Rubin, E. (1963). Gastroenterology, 45, 400.Scheuer, P. J. (1966). Personal communication.

Sherlock, S. (1959). Gastroenterology, 37, 574.Sherlock, S. (1963). Diseases of the Liver and Biliary System, 3rd ed.,

p. 415. Blackwell, Oxford.Teilum, G. (1946). Acta med. scand., 123, 126.Walker, J. G., Doniach, D., Roitt, 1. M., and Sherlock, S. (1965).

Lancet, 1, 827.Weir, D. M. (1963). Immunology, 6, 581.Williams, G. E. G. (1965). J. Path. Bact., 89, 23.Wr6blewski, F. (1958). Advanc. clin. Chem., 1, 314.

538



j.com/

J Clin P


ber 1966. Dow

nloaded from

http://jcp.bmj.com/

serological and histological diagnosis of primary biliary

Documents