Descriptive epidemiology of primary biliary cirrhosis in the province of Quebec

J-P VILLENEUVE, D FENYVES, C INFANTE-RIV ARD. Descriptive epidemiology of primary biliary cirrhosis in the province of Quebec. Can J Gastroenterol 1991;5(5):174-178. Primary biliary cirrhosis (PBC) is a ra re d isease, but is usually recognized because of the characteristic clinical picture and the diagnost ic specific ity of antimitochondrial antibody (AMA) determination . Information on the epitlemiology of PBC is limited. The autho rs have examined the incidence and prevalence of PBC in the province of Q uebec, where a ll short term hospitals are required to classify discharge summary diagno ·es accord ing to the International Classification of Diseases. Code 571.6 designates primary or secondary biliary c irrhosis. T he autho rs reviewed the charts of a ll patients to whom this cod .. was assigned during a six year period (1980-86). Two hundred and twen ty-e ight subjects satisfieJ predetermined diagnostic criteria for PBC. The mean annual inc idence rate was 3.9 per 106 popula tion, and the point prevalence in 1986 was 25.4 per 106 population . N inety-two patients were female, with a mean age a t the time of diagnosis of 55.7 years; 89.4% had positive AMA, and 10.5% were asymptomatic. As of January l , 1989, 126 patients were a live, 9 1 had d ied, and l L had undergone liver transplantation . C umulative five and 10 year survivals from the t ime of ini tial d iagnosis were 69% and 49%, re~pectivcly. In patien ts with serum bilirubins greater than JOO μmol/L (n=66), cumulative two year survival was 5.5%. T hese data indicate that the inciJence and prevalence of PBC in Q uebec arc similar to chose reported in O n tario and at the lower eml of the range of chose reported in western Europe. T he cl inical fea tures and evolut ion of PBCare also similar, and serum bilirubin is a major prognost ic facror.

P RIMARY BILIARY CIRRI lOSIS (P1 1C) IS characterized by progressive cholestasis and affects ch iefly middle agcJ or elderly women.The disease is rare, but is usually recognizeJ because of the characteristic clinical a nd histological picture, and the relative diagnostic specificity of a ntimitochondrial anti• body (AMA) d etermination .
Large se ri es of patients with PBC have bee n described from the Uni teJ States and Europe ( l -4) , whereas smaller series were reporred from Japan (5).However, populat ion -baseJ studies lll estimate inc idence or prevalence of the disease are sca rce.Most puhl ishe<l studies a rc from western Europe (6)(7)(8)(9)(10)(11)(12)(13)(14)(15).Some of these stud ie~ have shown regional d ifferences in the prevalence of PBC and apparent geographical clmtering of the disease.O nly one ~tu<ly has examined the epidemiology of PBC in North America ( 16).
The present study was unde rtaken to define the incidence and prevalence of PBC in the province of Q uebec (ropulation 6.5 mill ion).Cl inical, laboratory and survival Jara arc a lso presented.
En date du l er janvier 1989, 126 palicnts etaient en vie, 91 etaicnt <lece<lcs Cl 11 avaient subi une traru,planrarion hepariquc.Le pourcenrngc cumulatif de ,urvie a 5 ct a 10 ans a compLer de la dale du diagnostic atreignait 69 et 49 %, rcspccrivcmcnt.C hez lcs pat icnts Jonr le taux de bilirubine scrique etait ~upcrieur a 100 µm o l/L (n=66), la survie cumuative a Jcux ans ctait de 5,5 1 X,.PBC was based on one major anJ Lhree minor crite ri a.The major c riterion was apo irivc AMA deLcrmination.Minor criteria included: liver biopsy comrauble with a diagnosis of PRC (PBC Well> considered likely if a li ver biopsy was carried out and if the treating phys1c1an made a diagnosis of PRC on the discharge su mmary form; sl ides ~)( li ver histology were nol reviewed); n o evidence of biliary tract ohstrucuon assessed by ci Lh er cho la n g iogra m , ul1rnsounJ cxaminaLion, cnmputed tomography scan or operative finJmgs; and c ho lcstas1s, defined as ,1 scru m alkaline phl)sphmasc value greaL er than twice t he upper normal l1m1 t.
According LO these crite ria, diagnosis of PBC was cmcgmizeJ as 'certain' (one maior and Lwo n r Lhrcc 111111or crnena); 'likel y' (onl' maim and one minor criterion, or rhrcc mino r criteri a); 'possible' (no major and two minor criteria); anJ 'unlikely' (nl) major a nd one Ill' no minm cri1 cmin ).Data analysis: The survival status as of January l , 1989 and the unJcrlymg ca use of Jeath were obta ined from Lhe population registry of Quebec.Survival curves were estimated according tn che Kaplan-Meier method ( 18).The m11nbcr of mhabirnnts of the province of Quehcc was obtained from Canadian population census darn ( 19).

RESULTS
One hundred and four hospitals in Quebec reported a tmal of 719 cases for which co<le 571.6 appc;1red as a reported di agnosis on the discharge su mmary form.Patient informaLion was ohtmneJ for 698 of the 719 el1gihle CAl-.J GASTROENTFROL Vl)L 'i Nt) 5 SEJ'Tl,MRER/Cx Tt)I\ER 1991 Epid em iology of PBC in Quebec   O n Janua ry I, 1989, I 26 patie nts we re ali ve, 91 had died, and 11 h ad unde rgon e li ver t rn nsplan tatin n (fi ve a live and six dead).A mong the 9 1 who d ied, 60 deaths were re lated to li ve r disease, 28 were unre lated , and the ca use of death was unknown in three cases.
C um ulative five and 10 yea r surv iva ls were 69<)(, and 49<),(i, respec ti ve ly.To est imate the surviv,1 1 func tion for PRC, pa ti en ts who di ed of cause!> unre lated w live r disease were excluded , and patients wh o underwen t liver trans-176 The p rogn ostic value o f scrum bi liruhin for surv ival was also examined (T ab le 4 ).T he prohahili ry of su rviving two years was 13.3% in pa tien ts with a scru m h iliru hin va lue between 100 a nd 170 µm o l/ L, and 2.6% in chose with a va lue greater than 170 µrn ol/L.T hus, a mong 66 patients with a scrum b ili -100 80

DISCUSSION
To assess the inc id ence and prevalence of PBC in Q uebec, rh c authors took advantage of the mandatory rcpo n i ng o f d ischa rge summary diagnnm in Q ucbcc'tt hosp ita ls, and made the assumptio n that mm t r h ysic ia rn, would perform a li ver biop~y to establish a diagnosis of PBC and that pm icnts would be h ospita li zed for th is rrocedure, as li ver b io ps ies were nm done on an outpatient basis in th is area durmg  'Six patients were excluded from analysis because their serum btlirubln values were not ovollobte the sru<ly period .The re arc reason, w believe th.it the true prevalence and 111ci<lence of PRC may he undere:.timated.FiN, in 55 cases ( 20% of the database) th e diagnosi, of PBC was rejected hecau~e nf incomplete information, hut it is possihl e that some of these pa tients did h ave PBC.Second , patients who were diagnosed a~ havin g PBC prio r tn l 980, and who diJ nm require hospitalization during the fo llowing six years, would n ot have hee n detected by the survey method.This would result in a lower e~t imatc of prcvn lence, hut would no t Hffect the estimate of incidence.Howewr, the lmtcr e,timate wa:, affected hy p,1 ticnu, with : BC in whom a di agnosi~ was made wi thout liver hiopsy, si nce these patients were probably not h ospirnltzcd.T his may explain th e low proportion o f subjec t~ wnh asymptomatic PBC in chis study ( 10% ), as physicians may he less inclined to do I iver biopsies 1n such cases.Fina ll y, th e ex isten ce of cases of PBC that we re nm rccogni zeJ or <liagnoseJ woulJ lower the prevalence and inciJence.ln most previous stuJ ic~, unde rcsumation of disease frequency was also likely.Case-finding mc rhmb have included volunta ry reporting by physicians, exc1mination of hospirnl admissions, review of pathology reports, review of posi tive AMA tests, or a combination of the above.Studies relying on volun tary re porting by phys ic ians (10,12,16) arc partic ularly vulnerahle to an underestimation hias.S tudi es which counte<l cases from hospitalization events (6,15) arc potentia ll y less susccptih lc to suc h a bias, provided that citizens from the sruJi cd area:, maJe ~xclusive use of the regional or local hospitals from which cases were ascer-tamed.This was not a problem in the prese nt study since it includcJ a ll h ospitals in the prov ince.A mo re subtle proble m with this w urcc of darn is the c ha nge in regiona l or loca l hnspiraliza t ion rates over a period of tim e: , uch c hanges arc assoc1ntcJ with a numher of factor:, 1iften cx tran eou:, to the disease melf -for exa mpl e, t he rec ruitment o f ,1 :,pecia list in an area ( I 5 ).
The clinical and lahoratory featu res ()( PBC in the present st udy a rc comparahlc to those reported in 1Hher large stu<li cs ( 1-3 ).T he survival c urve of the present group is also almost identical to those of other large se ri es {3,20,2 1 ).
The sh ape of the c urve, sh ow ing a stcady J ecl inc in surviva l from the time of 111itial diagnos is, is nf partic ular interest.If PBC was Ji agnoseJ a l a uniform po int in time in th e nmural hiswry of the disease, ;mJ if mos1 patients were ~till a li ve six to e ight years after diagnosis, o n e wou ld expect th e survival c urve to he in itia lly flat, and then to dec rease fai rl y sh a rply after six co e ight yea rs.Instead , the steady decline in surviva l suggcsls that the natural history of PBC varies cons i<lerahly from pacicnt to patie nt , with some expe rienc ing rnpid evolution towards l ivc r failure anJ mh ers experiencing The prognostic value of sc rum hi liru h in in PBC h as heen demnnst rated hy severa l au dwrn (I, 3,22).In the prescm stud y, o nly th ree of 66 subj ects with sc rum hil iruhin values greater Lhan I 00 µm o l/L were ali ve two ycms later.The study was no t designed tn ,lsses, prognostic facto rs in PBC, and th e authors <lid not collect the data neccss,iry to va li<late mo re complex prognostic models such as the Mayo mode l (2), C hristensen 's mode l (20) or t he Yale model ( 3 ).Nevcn hel css, the prese nt darn inJicate that J ead1 was ne,irly certam a mo ng patients with scrum hiliruhin va lues grea ter than I 00 µmol/L.Thus, scrum bili rubin alone at the tested c ut-off point will distinguish pat ic ncs with poor prognoses from those with more favorable ones; PBC patients wi t h sc rum biliruhin va lues greater than 100 ~tmol/L should be considered candi<latcs for immed iate li ver trnnsplanrntion.A two year surviva l ra te of 74\X, has hcen reported in PBC patients fo llowing liver transplantation , clearly muc h bette r than t hat of untreated paucnts (23).
In .~ummary, the inc idence and preva lence of PRC in the province of Q uebec a rc si milar to those reported in O ntario, and at the lower end of the range of those reported in western Europe.Whethe r this imp lie~ a gen uine J1ffercnce in the epidemio logy o f PBC or a difference in the degree of awareness n( physicians, remains un<lc Lcrmin cd .The cli nical featu res a nd evoluuon of rBC in Q uebec appear lO he quite simil,1r to t hose rc po neJ fro m other countries, a nd sc rum hil iruhin is ,1 major rrognostic fac tor.ACKNOWLEDGEMENTS: Thi: aurhor, thank Drs A nnc Su1ton and Gac1 anm: Ro uthi er, M, EJit h Marcoux, C in c1ti: Ray mllnJ , Manon Bourcicr, and the med1rnl record, Jcpan ITil'll t ,taff for thei r tontril ,uti on rn th b st udy.l)r, J can-P1errc Villeneuve ,ind Churc lnfantc-Ri vmJ Mc 'cherchcur,-hour"crs' from the Fnnd, de la rcchcrchc en ,,mte du Q uchcL.

20 Not available 3 •
Serum bilirub in (°k of tota l c o ses) 0 to 34 ~tmol/L 51 35 to 100 µmol/L 17 101 to 170 µmol/L 9 G reate r than 170 µmol/L Percentage of symptomatic potlents was 29 µm o l/L.T h e distribution of scrum biliru bin values at Lhc last v isi t d uring LI ,c st udy period is shown in T a ble 2. Du ring Lhc study period (s ix yea rs), 159 new cases of PBC were d iagnosed, fo r a mean annua l in cidence rn te of 3.9 per I 0 6 popu laLion .On Marc h 3l, 1986, 166 of the 228 cases of PBC were ali ve, and the poim prevalence wa~ 2 5 .4per 10 6 popu lat inn.T h e incide nce and prevale nce o f PBC fo und in t he prescm sLudy a nd those reported from m her coun tries a rc shown in T able 3. C umul at ive surv iva l fro m th e t ime of in iLia l d iagnosis is shown in Figure I.

CAN J GASTROENTERUI VOi 5
Nu 5 S~i'lTM IWR/0: TO!ll;R 1991 Epidemiology of PBC in Quebec little progression .The possih il ity that the diagnos is is m,1d e at Jifferent t 1mes in the course of the d isease could also contribute m the sh ape n f the survival c urve.

TABLE 1
Diagnoses in 648 patients with code 571 .6 hospitalized in the province of Quebec between April 1, 1980 and March 31, 1986

TABLE 2
Clinical and laboratory data of 228 subjects with primary biliary cirrhosis

TABLE 3
Incidence and p revalence of primary biliary cirrhosis respecti vely.A mong patie n ts who di ed ofliverd iscasc, 37% we re younger than 60 yea rs of age.