Idiopathic portal hypertension associated withcytotoxic drugs

P Shepherd, D J Harrison

J Clin Pathol 1990;43:206-210

AbstractFour patients developed clinically impor-tant portal hypertension with histologicalfeatures of idiopathic portal hyperten-sion while they were receiving cytotoxicdrugs for chronic myeloid leukaemia andHodgkin's disease. Mild sclerosis of somesmall portal triads was the only abnor-mality seen at light microscopicalexamination in three of the four cases. Inthe remaining case light microscopicalfindings seemed to be normal. Two cases

examined by electron microscopy showedperisinusoidal fibrosis; in one case thiswas the only abnormality detected.There is an association between

idiopathic portal hypertension and theuse of chemotherapeutic agents, par-

ticularly thioguanine. Adequate his-tological examination of liver tissue,including electron microscopic studies, isrecommended for patients who develophepatic problems while receivingcytotoxic treatment to elucidate thisproblem.

Department ofHaematology, WesternGeneral Hospital,EdinburghP ShepherdDepartment ofPathology, Universityof Edinburgh, TeviotPlace, EdinburghD J HarrisonCorrespondence to:Dr P Shepherd,MRC Human Genetics Unit,Western General Hospital,Crewe Road, EdinburghEH4 2XU.

Accepted for publication12 October 1989

Idiopathic portal hypertension in the absenceof cirrhosis or other demonstrable causes ofportal hypertension is uncommon, accountingfor 3-40h of all cases in Western countries' but25-30% in India2 and Japan.3 Fibrosis of smallportal triads may be present and in most theaetiology is unknown: some have beenassociated with exposure to arsenic,4 5 vinylchloride,56 and vitamin A.7 Only rarely havechemotherapeutic agents been implicated.8Key et al recently attributed non-cirrhotic

portal hypertension to nodular regenerativehyperplasia (NRH) in four patients withchronic myeloid leukaemia who were treatedwith busulphan and thioguanine.9 NRH is rare

and may be difficult to diagnose on small needlebiopsy specimens.'0 In reported series ofpatients with NRH" 12 mild fibrosis of smallportal tracts was noted in most, and it has beensuggested that the regenerative nodules maydevelop as a response to obliterative portalvenopathy. "

We report the development of portal hyper-tension in four patients after treatment ofchronic myeloid leukaemia and Hodgkin's dis-ease with chemotherapeutic drugs. Mild portalsclerosis was present in three of four cases, andin one other the only apparent abnormality wasperisinusoidal fibrosis seen on electron micros-copical examination.

Case reportsCASE 1

A 55 year old woman developed chronicmyeloid leukaemia (CML) in July 1982.Chromosomal analysis showed the typical t9;22translocation. The spleen was palpable 7 cmbelow the costal margin, but the liver was notenlarged and liver function tests at diagnosiswere normal. She was entered into the MedicalResearch Council trial of treatment for chronicmyeloid leukamia (MRC CML II). Over thenext 17 months she received busulphan (totaldose 442 mg) and thioguanine (total dose17 5 g) with symptomatic control ofher diseaseand resolution of splenomegaly. The results ofthe liver function tests are shown in fig 1. InDecember 1983 busulphan and thioguaninewere discontinued because of possible pulmon-ary toxicity. In February 1984 21 months afterthe start of chemotherapy she developed gas-trointestinal haemorrhage secondary to oeso-

phageal varices. These were treated bysclerotherapy. Control of her CML was sub-sequently maintained with hydroxyurea. InMay 1984 she had further variceal haemorr-hage which was treated by oesophageal transec-tion. A wedge liver biopsy specimen was his-

x3 -

w x2-atclr Nq

Case l Vcirices

Bu/TG6 9 12 15Months from diagnosis

18 21

Varices

0 3

x3-

. x2-

0r -

a'

Case 2

N =0

x10-

c x5-

x2-Ni

Bu/TG Bu/TGI --r

3 6 9 12 15 18Months from diagnosis

21

Case 3 Varices

Ad' ,

0 34 36 42 48 54 60 66 70Months from diagnosis

Figure I Liver function tests showing degree ofabnormalities present before onset of varicealhaemorrhage. N= normal; O = alkaline phosphatase;0 = alanine transferase; a = bilirubin.

206

. .

LL.I

on June 14, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.43.3.206 on 1 M

arch 1990. Dow

nloaded from

Idiopathic portal hypertension associated with cytotoxic drugs

tologically normal. Angiography showed noevidence of hepatic or portal vein thrombosis.Over the next year she suffered recurrentvariceal bleeding and developed ascites. Threemonths later blast transformation occurred andshe died shortly after.

CASE 2A 55 year old man developed chronic myeloidleukaemia in March 1983. The Philadelphiachromosome, however, was not present. Thespleen was palpable 4 cm below the costalmargin, but the liver was not enlarged and liverfunction tests were normal. He was enteredinto the MRC CML II trial and receivedbusulphan (total dose 252 mg) and thioguanine(10- 1 g) with good control of his disease. Serialliver function tests are shown in figure 1. InDecember 1983 he developed haemoatemesisand malaena due to a bleeding gastric ulcer,seen on endoscopic examination. Chemoth-erapy was stopped and cimetidine started.Repeat endoscopy in March 1984 was normal.In September 1984, due to increasing leucocytecounts, busulphan and thioguanine were res-tarted. In December 1984 21 months afterstarting treatment haematemesis recurred andanother endoscopy showed bleeding oeso-phageal varices; these were treated withsclerotherapy. Chemotherapy was continueduntil September 1985 when recurrent varicealhaemorrhage uncontrolled by sclerotherapyrequired laparotomy and removal ofthe spleen.Portal venography showed increased pressurein the portal vein but no other abnormality. Inparticular, there was no evidence ofobstructionof the portal tract. Chemotherapy was changedto hydroxyurea and since then he has had nofurther problems.

CASE 3A 34 year old woman presented in 1981 withcervical lymphadenopathy. Biopsy and stagingshowed nodular sclerosing Hodgkin's disease(stage IIA). She was treated with radiotherapywith regression of the disease. In March 1985she developed recurrent Hodgkin's disease inthe lungs. Liver function tests and a bariumswallow at this time were normal. She wastreated with combination chemotherapy usingchlorambucil, vinblastine, procarbazine andprednisolone (ClVPP) for six courses withregression of her pulmonary disease. Shortlyafterwards she developed pronounced abnor-malities of liver function tests, particularlyalkaline phosphatase (figure 1). Her liver wasnot enlarged and an open liver biopsy specimenwas obtained. Because of the severely increasedalkaline phosphatase activity an endoscopicretrograde cholangiopancreatogram was per-formed to assess the biliary tree and pancreas.Mild deformity of the intrahepatic ducts in theleft lobe of the liver was seen but no otherabnormality. Spider naevi were noted and abarium swallow in February 1988 showedoesophageal varices.

CASE 4Chronic myeloid leukaemia was diagnosed in a68 year old woman in December 1986..She wasentered into the MRC CML II trial andreceived busulphan (total dose 550 mg) andthioguanine (total dose 23 6 g) until May 1988when thioguanine was stopped because of con-cern about potential hepatotoxicity associatedwith its use. At diagnosis slightly increasedtransaminase activities were present (< 2 x N)which persisted along with a mild increase inalkaline phosphatase during treatment. Busul-

Figure 2 Portalvenography necropsy: Dyewas injected into the portalvein filling the liver withretrograde filling of thespleen. The spleen showsnormal capillary fillingbut portal vein radicals arenipped off at the venularlevel.

207

on June 14, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.43.3.206 on 1 M

arch 1990. Dow

nloaded from

Shepherd, Harrison

Figure 3 Photomicrograph showing normal lobular architecture (MassonTrichrome).

phan was continued and she remained welluntil September 1988 when a bleedingduodenal ulcer, which was confirmed byendoscopy, developed. In January 1989 shedeveloped jaundice, ascites, and hepatosplen-omegaly. A bone marrow aspirate still showedchronic phase features but chromosomalexamination showed an isochromosome 17,suggesting transformation to a more acute

phase. Haematemesis developed and endos-copic examination showed bleeding oeso-

phageal varices and duodenal ulceration. Shedied in February 1989.

PathologyCASE 1

At necropsy a portal venogram showed delayedfilling with "cut off' of small portal veinradicles but not evidence of thrombosis (figure2). This appearance was not seen when controlliver tissue was studied. Histological examina-tion of the liver before death and at necropsy

had shown normal lobular architecture (figure3). The sinusoids were infiltrated by leukaemiccells and there was fibrosis of small portal triadswith narrowing of portal venules (figure 4).Electron microscopic examination showedfibrosis in the perisinusoidal space of Disse(figure 5).

Figure 4 Photomicrograph ofportal triad containing bile ducts and hepaticartery but no distinguishable portal vein. There is an increase in portalfibrosis(Orcein Shikata).

CASE 2The liver biopsy specimen at laparotomyshowed normal architecture with no evidenceof leukaemic infiltration of sinusoids. Therewas mild sclerosis of small portal triads similarto that seen in figure 4.

CASE 3A liver biopsy specimen seemed to be normalon light microscopical examination. Electronmicroscopic examination showed perisin-usoidal fibrosis in the space of Disse, similar tothat seen in figure 5.

CASE 4At necropsy prominent oesophageal variceswere present. No abnormalities were seen inthe portal vein, hepatic vein, or hepatic artery.Histological sections showed that liverarchitecture was well preserved without cirr-hosis, parenchymal fibrosis, or nodular regen-erative hyperplasia. Many portal triads werenormal but some showed increased fibrosiswith narrowing and occlusion of portal veinbranches. The appearances were those ofhepatoportal sclerosis.

DiscussionWe have described four patients whodeveloped non-cirrhotic portal hypertensionassociated with the use of cytotoxic agents.Studies of series of patients with idiopathicportal hypertension and comparison withsimilar entities designated as hepatoportalsclerosis and non-cirrhotic portal fibrosis haveshown that they describe similar conditions,characterised by variable degrees of portal tractfibrosis without cirrhosis on histologicalexamination.'1-351" 5 Angiographic abnor-malities with "cut off" of small portal veinswere seen without evidence of portal veinthrombosis.5 Portal venography carried out atnecropsy on case 1 showed delayed filling and"cut-oft' of portal vein radicles, apparentlyidentical with the findings of Villeneuve.' Thisis attributed to the perisinusoidal fibrosisshown on electron microscopical examinationand noted in other cases of idiopathic portalhypertension.28 16 The pathogenesis of thefibrosis is unknown-and in only a few caseshave specific aetiological agents been incrimin-ated-vinyl chloride,56 arsenic,45 vitamin A,7mercaptopurine and azathioprine.8Many chemotherapeutic agents are known to

be potentially hepatotoxic and may cause acuteor chronic hepatic lesions."7"8 The develop-ment of portal hypertension without lightmicroscopic evidence of liver injury such as

cirrhosis, necrosis, chronic active hepatitis orvascular obstruction (idiopathic portal hyper-tension), however, is very unusual. Nataf des-cribed two patients treated with continuousazathioprine and mercaptopurine after renaltransplantation who developed idiopathic por-tal hypertension without histological lesions at

light microscopy but with perisinusoidalfibrosis at electron microscopy.8 These cases

seem to be similar to two of ours. In theremaining two cases electron microscopicstudies were not performed. In one of Natafs

208

on June 14, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.43.3.206 on 1 M

arch 1990. Dow

nloaded from

209Idiopathic portal hypertension associated with cytotoxic drugs

Figure 5 Electronmicrograph showingbundles of striated collagenbetween endothelium andhepatocyte in the space ofDisse.

cases serial biopsy specimens showed progres-sion to hepatoportal sclerosis, suggesting thatthis may be a subsequent manifestation offibroblastic proliferation.Key et al have recently described five

patients with chronic myeloid leukaemia,treated with busulphan and thioguanine, whodeveloped portal hypertension with oeso-phageal varices.9 Histological findings for fourshowed small intrahepatic nodules, interpretedas nodular regenerative hyperplasia, a rarecause of portal hypertension. This is a difficultdiagnosis to make because of the limitedmaterial available: indeed it may be impossibleto establish without a wedge biopsy.'0 Evalua-tion of series of patients with nodular regen-erative hyperplasia has shown mild fibrosisaffecting the portal tracts in most,"1 12 and it hasbeen postulated that the hyperplastic nodulesarise as a secondary response to obliterativeportal venopathy." There may therefore be anassociation between this disorder and otherconditions associated with fibrosis of portaltracts.

In the Medical Research Council's secondtherapeutic trial in chronic myeloid leukaemia(MRC CML II trial) in which single agenttreatment (busulphan) was compared withcombination treatment (busulphan andthioguanine) 18 cases of oesophageal varicesoccurred during the chronic phase of the dis-ease (personal observation, CML Trial Office,Edinburgh). All 18 cases of varices occurred inpatients receiving combination treatment,strongly suggesting that thioguanine was ass-sociated with the liver injury. These includefour of the five patients reported by Key9 andcases 1, 2, and 4 described here. Thioguaninehas been implicated in hepatitis and veno-occlusive disease of the liver,' but in none of

these seven patients for whom biopsyspecimens were available were thest findingsseen. Thus seven out of these 18 MRC CMLtrial patients had non-cirrhotic portal hyper-tension. The association between the use ofcombination treatment and the development ofportal hypertension has led to its withdrawalfrom the MRC CML II trial.This association with chronic myeloid leuk-

aemia and the use of thioguanine is particularlyinteresting. A similar picture has not beendescribed after the use of thioguanine or arelated drug, mercaptopurine, in the treatmentof other leukaemias, such as acute myeloidleukaemia or acute lymphoblastic leukaemiawhere the drug can be used in a similar mannerduring maintenance treatment. Portal hyper-tension may be seen at diagnosis in chronicmyeloid leukaemia related to leukaemic infil-tration of sinusoids. 9 Perhaps residual infiltra-tion, despite apparent good control of disease,may render these patients more vulnerable tothis condition, or perhaps the combinationwith busulphan is uniquely dangerous?Another factor that may predispose to toxicityis the continuous use of chemotherapeuticagents, as is the case in CML. As shown in case3, however, this has also occurred with inter-mittent courses of treatment.

Sclerosis of portal tracts has also been foundat necropsy without evidence of liver disease.Its cause and potential clinical importance areunclear.20 What is clear from reported cases,however, is that clinically important portalhypertension can result from this kind oflesion. We believe, along with others, thatfibrosis is a reaction to injury by a variety ofidentified and unidentified toxins and is res-ponsible for the syndrome of "idiopathic portalhypertension"." Fibrosis may be minimal or

on June 14, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.43.3.206 on 1 M

arch 1990. Dow

nloaded from

Shepherd, Harrison

absent when assessed at light microscopy, butperisinusoidal fibrosis visible only by electronmicroscopy may be associated with clinicallyimportant portal hypertension.We draw attention to its association with

chemotherapeutic agents, particularlythioguanine, but also, as shown in case 3, othercytotoxic agents. Of the drugs given in case 3,chlorambucil may be the most likely offender.The paucity of previously reported cases of

idiopathic portal hypertension in associationwith chemotherapeutic agents suggests thatthis is the result of a rare idiosyncratic reactionto the drugs. Its incidence, however, may beunderestimated if liver biopsy or necropsymaterial is not obtained or adequatelyevaluated in any such patient who developsevidence of liver dysfunction. It is not knownwhether the lesions can regress with time. Inone case associated with exposure to vinylchloride a second biopsy specimen obtainedtwo years later showed persistent fibrosis.6Liver function tests are not particularly helpfulin screening for this disorder; they may benormal or only slightly abnormal as in two ofour cases. Portal venography showing thecharacteristic "cut off" may be useful.Adequate histological examination of liver tis-sue, including electron microscopy, is essentialfor evaluating possible injury in cases of portalhypertension or unexplained hepatic dysfunc-tion associated with the use of these agents, andwill be required to determine more clearly theincidence and pathogenesis of this disorder.

We are grateful to Drs N C Allan, H M Cameron, N CFinlayson, E J Gaminara, R D Hyde and A C Parker forpermission to report these cases, and to Anne Young for typing.

1 Iber FL. Portal hypertension in the presence ofnormal livermorphology. Ann N Y Acad Sci 1970;170:1 15-26.

2 Sama SK, Bhargava S, Gopi Nath N, et al. Non-cirrhoticportal fibrosis. Am J Med 1971;51:160-9.

3 Okuda K, Nakashima T, Okudaira M, et al. Liver pathologyof idiopathic portal hypertension. Comparison with non-cirrhotic portal fibrosis of India. The Japan idiopathicportal hypertension study. Liver 1982;2:176-92.

4 Morris JS, Schmid M, Newman S, et al. Arsenic and non-cirrhotic portal hypertension. Gastroenterology1974;64:86-94.

5 Villeneuve JP, Huet PM, Joly JG, et al. Idiopathic portalhypertension. Am J Med 1976;61:459-64.

6 Thomas LB, Popper H, Berk PD, et al. Vinyl chlorideinduced liver disease. From idiopathic portal hyper-tension (Banti's syndrome) to angiosarcomas. N Engl JMed 1975;292:17-22.

7 Russell RM, Boyer JL, Bagheri SA, Hruban Z. Hepaticinjury from chronic hypervitaminosis A resulting in portalhypertension and ascites. N EnglJMed 1974;291:435-40.

8 Nataf C, Feldmann G, Lebrec D, et al. Idiopathic portalhypertension (perisinusoidal fibrosis) after renal trans-plantation. Gut 1979;20:531-7.

9 Key NS, Emerson PM, Allan NC, et al. Oesophageal varicesassociated with busulphan-thioguanine combinationtherapy for chronic myeloid leukaemia. Lancet1987;ii: 1050-2.

10 Scheuer PJ. Liver biopsy interpretation. Third Edition.London: Bailliere Tindall, 1980.

11 Wanless IR, Godwin TA, Allen F, Feder A. Nodularregenerative hyperplasia of the' liver in haematologicdisorders: a possible response to obliterative portalvenopathy. Medicine 1980;59:367-79.

12 Stromeyer FW, Ishak KG. Nodular transformation(nodular 'regenerative' hyperplasia) of the liver. HumPathol 1981;12:60-71.

13 Aikat BK, Brunsnurmath SR, Chautiani PN, et al. Thepathology of non cirrhotic portal fibrosis. Hum Pathol1979;10:405-18.

14 Mikkelsen WP, Edmnondson HA, Peters RL, et al. Extra andintrahepatic pprtal hypertension without cirrhosis(hepatoportal sclerosis). Ann Surg 1965;162:602-20.

15 Benhamou JP, Lebrec D. Non-cirrhotic intrahepatic portalhypertension in adults. Clin Gastroenterol 1985-14:21-31.

16 Tandon BN, Lakshminarayanan R, Bhargava S, et al.Ultrastructure of the liver in non-cirrhotic portal fibrosiswith portal hypertension. Gut 1970;11:905.

17 Zimmerman HJ. Hepatotoxic effects of oncotherapeuticagents. Progr Liver Dis 1986;8:621-42.

18 Menard DB, Gisselbrecht MM, Marty M, et al. Antineo-plastic agents and the liver. Gastroenterology 1980;78:142-64.

19 Datta DV, Grover SL, Saini VK, et al. Portal hypertensionin chronic leukaemia. Br J Haematol 1975;31:279-85.

20 Wanless IR, Bernier V, Seger M. Intrahepatic portal veinsclerosis in patients without a history of liver disease. Anautopsy study. Am J Pathol 1982;106:63-70.

210

on June 14, 2020 by guest. Protected by copyright.

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.43.3.206 on 1 M

arch 1990. Dow

nloaded from