Short reports

behaviour is unpredictable; tumours withincreased malignant potential, by virtue of sizeand cytology, being termed atypical carcinoids.In addition to positive staining with routineepithelial and neuroendocrine markers, theyusually express CEA but not S100. Althoughcommonly presenting as a primary lesion,metastases are not unusual and silent primarytumours have been described.' Neurologicalcomplications are present in 40% of metastaticcarcinoids and true central nervous systemmetastases have been documented both withinthe brain and spinal cord.2Primary paragangliomas of the cauda equina

are well described' and, although malignantbehaviour is rare, metastatic spread has beendocumented.4 Nuclear atypia and necrosis areunusual features. Positive staining for cytokera-tin is uncommon and has not been reported forCEA to our knowledge. The supportingsustentacular cells are classically S 100 pcsitive,although they are not always present.5

Pituitary carcinomas are a well recognisedalbeit rare entity. Generalised spread has beenreported6 although central nervous systemspread is more common. Metastasis to thecauda equina has been documented.7 Mostcarcinomas are histologically identical to ad-enomas; however, nuclear atypia has beenrecorded and said to indicate increased malig-nant potential.8 Cytokeratin positivity has beenreported9 but is uncommon; however, thepituitary is a relatively common site formetastatic carcinoma from other sites.'0

ConclusionThis widespread tumour illustrates the prob-lem of classifying neuroendocrine tumourswhen there is no obvious primary site. Thediagnoses of atypical carcinoid, malignantparaganglioma, and pituitary carcinoma wereall considered; however, our favoured diagnosiswas atypical carcinoid.

1 Nida TY, Hall WA, Glantz MJ, Clark HB. Metastatic carci-noid tumour to the orbit and brain. Neurosurgery1992:31:949-52.

2 Patchell RA, Posner JB. Neurologic complications ofcarcinoid. Neurology 1986;36:745-9.

3 Sonneland PRL, Scheithauer BW, LeChago J, CrawfordBG, Onofrio BM. Paraganglioma of the cauda equinaregion. Clinicopathologic study of 31 cases with specialreference to immunocytology and ultrastructure. Cancer1986;58: 1720-35.

4 Lack EE, Cubilla AL, Woodruff JM. Paragangliomas of thehead and neck region. A pathologic study of tumours from71 patients. Hum Pathol 1979;10: 191-218.

5 Brodkey JA, Brodkey JS, Watridge CB. Metastatic paragan-glioma causing spinal cord compression. Spine 1995;20:367-72.

6 Scheithauer BW, Randall RV, Laws ER, Kovacs KT,Horvath E, Whitaker MD. Prolactin cell carcinoma of thepituitary. Clinicopathologic, immunohistochemical, andultrastructural study of a case with cranial and extracranialmetastases. Cancerl985;55:598-604.

7 Asai A, Matsutani M, Funada N, Takakura K. Malignantgrowth hormone-secreting pituitary adenoma with haema-togenous dural metastasis: case report. Neurosurgery 1988;22:1091-4.

8 Challa VR, Marshall RB, Hopkins MB, Kelly DL, CivantosF. Pathobiologic study of pituitary tumours: report of 62cases with a review of the recent literature. Hum Pathol1985;16:873-84.

9 Frost AR, Tenner S, Tenner M, Rollhauser C, Tabbara SO.ACTH-producing pituitary carcinoma presenting as thecauda equina syndrome. Arch Pathol Lab Med 1995;119:93-6.

10 Nudleman KL, Choi B, Kusske JA. Primary pituitarycarcinoma: a clinical pathological study. Neurosurgery1985;16:90-5.

Effusion cytology of hepatocellular carcinomawith in situ hybridisation for human albumin

M R Stephen, K Oien, R K Ferrier, R A Burnett

Department ofPathology, WesternInfirmary, GlasgowGIl 6NT, UnitedKingdomM R StephenK OienR K FerrierR A Burnett

Correspondence to:Dr M R Stephen.

Accepted for publication26 February 1997

AbstractWhile the cytological features of hepato-cellular carcinoma on fine needle aspira-tion cytology are well described, cases ofhepatocellular carcinoma with malignantcells in ascitic fluid and their characteris-tics are not. A patient is described withcirrhosis resulting from chronic hepatitisB virus infection, ascites, and hepatocel-lular carcinoma diagnosed by effusioncytology. The malignant cells in the effu-sion were shown to be positive for afetoprotein using immunocytochemistry,and for human albumin using in situhybridisation, confirming the diagnosis ofhepatocellular carcinoma. Further inves-tigations in a terminally rn patient werethus avoided.(C Clin Pathol 1997;50:442-444)

Keywords: ascitic fluid; hepatocellular carcinoma;albumin; in situ hybridisation

Ascites is a common complication of bothcirrhosis and hepatocellular carcinoma. Theyield of malignant cells from hepatocellularcarcinoma is generally low (approximately10% in the series described by Falconieri etal'), and they may be more often identified incases of hepatocellular carcinoma withoutcirrhosis.' There may be difficulty distinguish-ing the malignant cells from reactive andatypical mesothelial cells which are commonlyseen in effusions from cirrhotic patients. Theimmunocytochemical staining profile ofhepatocellular carcinoma expressing positivityfor keratins and negativity for CEA2 does nothelp to distinguish hepatocellular carcinomacells from mesothelial cells. Positive staining fora fetoprotein, while helpful, is not specific.'Albumin gene detection by in situ hybridisationis a highly specific aid to the confirmation of theorigin of the malignant cells in an effusion.'

442

on August 26, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.50.5.442 on 1 May 1997. D

ownloaded from

Short reports

--.M.. : . :. ^::i_

.... :t:s_

,.,,.. ,,j :'-sl-v X,Tw<''

,,s, *.,

.E........

..w...s~~~~~~~~~~~~~~~~.

b~~~~~~~~~~~~~~~~~.... .......... ....

Figure 1 A group of dispersed malignant cells with mildly pleomorphic nuclei andgranular chromatin surrounded by moderate amounts ofgranular cytoplasm contaznznintracytoplasmic vacuoles some of which were bile droplets. (Cytospin haematoxylin aneosin, x600.)

A

B

Figure 2 (A) Two malignant cells strongly reacting to a fetoprotein antibody (cytospiAP-Vector SK-5100, x600). (B) Strong diffuse cytoplasmic staining for human albu,mRNA in a malignant cell. (The hybridised probe was visualised using nitrobluetetrazoliumlbromo-chloro-indolyl-phosphate (NBTIBCIP) substrate. Cytospin x600.)

* Case reportThe patient was a 60 year old woman ofChinese extraction who had made many tripsto Hong Kong. She had chronic hepatitis Bvirus infection (HbsAg and HbeAg positive),which had been diagnosed 18 months beforethe most recent admission and had beenconfirmed by liver biopsy, the report of which

' concluded that although there was somefibrous scarring, cirrhosis was not established.Focal areas of hepatocyte dysplasia, however,were noted. She presented terminally withrapid clinical deterioration with malaise,anorexia, deepening jaundice, and diureticresistant ascites. Ultrasound and computerisedtomography suggested a multicentric hepato-cellular carcinoma. Her a fetoprotein serumconcentrations were grossly raised at2070 KU/I (normal < 5 KU/1). In view of hergross ascites and disturbed coagulation, arepeat liver biopsy was not performed. Para-

1g centesis was done, largely for symptomaticid relief. Confirmation of the hepatocellular

carcinoma was made from the effusion cytol-ogy. In view of the poor prognosis, sympto-matic treatment only was continued. Permis-sion for necropsy was refused.

Cytology findingsCytospin preparations were made and weremoderately cellular. Small lymphocytes andreactive mesothelial cells were present in thebackground. The malignant cells lay in smallgroups and as single cells. The nuclei tended tobe large and hyperchromatic without promi-nent nucleoli. The cytoplasm was faintlygranular. A striking feature was the presence ofsmall round cytoplasmic vacuoles (fig 1). Someof the vacuoles contained PAS positive materialwhile others stained positively for iron and bile.Many apoptotic cells were noted. The abnor-mal cells stained strongly on immunocyto-chemistry with antibodies for a fetoprotein (fig2A), which is a hepatocellular product but isnot absolutely specific for malignant cells froma hepatocellular carcinoma. Albumin is aspecific product of normal and transformedhepatocytes. In situ hybridisation for the detec-tion of the albumin gene was strongly positivein this case (fig 2B).DiscussionThe diagnosis of hepatocellular carcinoma isimportant because many cases present late andthe patients may have systemic problems suchas coagulopathy that prevent invasive proce-dures to obtain diagnostic material. Thecytological features on fine needle aspiration ofhepatocellular carcinoma are well described,but are of little use when applied to effusioncytology for example, endothelial rimming ofneoplastic cell groups and intranuclear inclu-sions are not notable features in an ascitic fluid.

Metastatic malignancy-for example fromlung or gastrointestinal tract in peritonealeffusions, shows a relatively limited range ofcytological expression, and confusion with reac-tive mesothelial cells can be a disturbing prob-

in lem. The application of a technique which iswin specific to the hepatocyte is clearly a great ben-

efit in defining the site of origin of a neoplasm.

443

on August 26, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.50.5.442 on 1 May 1997. D

ownloaded from

Short reports

An in situ hybridisation procedure to revealalbumin mRNA on formalin fixed hepatic tissueusing a digoxigenin labelled oligonucleo-tide probe has been developed by Murray et al.4In our case a 2 kb cDNA sequence correspond-ing to a coding sequence of human albuminprovided by ATCC (Rockville, Maryland, USA)was used to produce a SP6 transcribed,digoxigenin 11 UTP labelled anti-sense ribo-probe. The cytospin preparations were formalinfixed and the method of in situ hybridisationdeveloped by Stewart et al5 was employed. Thepatient had clinical signs strongly suggestive ofhepatocellular carcinoma, supported by sero-logical, ultrasound, and computerised tomogra-phy findings. The peritoneal effusion containedsingle neoplastic cells and groups of neoplasticcells showing nuclear pleomorphism and granu-lar chromatin surrounded by moderate amountsof rather granular cytoplasm. These cytologicalfeatures were certainly in keeping with theappearances of hepatocellular carcinoma asdescribed in fine needle aspirate,6 and afterdemonstrating human albumin mRNA in these

cells we felt that the diagnosis of hepatocellularcarcinoma in the ascitic fluid was certain.We feel that application of this rapid,

reliable, and specific technique of in situhybridisation for human albumin can confirmthe diagnosis of hepatocellular carcinoma oncytospin preparations of ascitic fluid, thusmaking further invasive diagnostic proceduresunnecessary.

1 Falconieri G, Zanconati F, Colautti I, Dudine S, Bonifacio-Gori D, Di Bonito L. Effusion cytology of hepatocellularcarcinoma. Acta Cytol 1995;39:893-7.

2 Ma C-K, Zarbo RJ, Frierson HF, Lee MW Comparativeimmunohistochemical study of primary and metastaticcarcinoma of the liver. Am J Clin Pathol 1993;99:551-7.

3 Papotti M, Pacchioni D, Negro F, Bonino F, Bussolati G.Albumin gene expression in liver tumours: diagnosticinterest in fine needle aspiration biopsies. Mod Pathol 1994;7:271-5.

4 Murray GI, Paterson PJ, Ewen SWB, Melvin WT. In situhybridisation of albumin mRNA in normal liver and hepa-tocellular carcinoma with a digoxigenin labelled oligonu-cleotide probe. J Clin Pathol 1992;45:2 1.

5 Stewart CJR, Farquharson MA, Kerr T, McCorriston J.Immunoglobulin light chain mRNA detected by in situhybridisation in diagnostic fine needle aspiration cytologyspecimens. J Clin Pathol 1996;49:749-53.

6 Pilotti S, Rilke F, Claren R, Milellan M, Lombardi L. Con-clusive diagnosis of hepatic and pancreatic malignancies byfine needle aspiration. Acta Cytol 1988;32:27-38.

Dilutional hyponatraemia: a cause of massive fatalintraoperative cerebral oedema in a childundergoing renal transplantationA Armour

State PathologistsDepartment, Instituteof Forensic Medicine,Grosvenor Road,Belfast, UnitedKingdomA Armour

Correspondence to:Dr Alison Armour,Consultant Pathologist,Directorate of Pathology,PO Box 202, Royal PrestonHospital, Sharoe Green LaneNorth, Fulwood,Preston PR2 4HG,United Kingdom.

Accepted for publication12 February 1997

AbstractA four year old boy with polyuric renalfailure resulting from recurrent urinarytract infections and vesicoureteric refluxfrom birth underwent renal transplanta-tion. In the past he had had five uretericreimplant operations and a gastrostomy,as he ate nothing by mouth. He requiredperitoneal dialysis 13 hours a night, sixnights a week. His fluid requirements were2100 ml per day. This included a night feedof 1.5 litres Nutrizon. Before operation hereceived 900 ml of Dioralyte instead of theNutrizon feed, and peritoneal dialysis wasperformed as usual. The operation itselfwas technically difficult and there wasmore blood loss than anticipated, requir-ing intravenous fluids and blood. Theoperation ended about four hours later buthe did not wake up. Urgent computed tom-ography revealed gross cerebral oedema.He died the next day. At necropsy the brainwas massively oedematous and weighed1680 g.(7 Clin Pathol 1997;50:444-446)

Keywords: cerebral oedema; operation; intravenousfluids

There are various causes of cerebral oedemaincluding inflammatory conditions, ischaemia,trauma, space occupying lesions, anoxia, tox-

ins, and metabolic disorders-in particularhyponatraemia' and water intoxication.2 Cere-bral oedema has been defined as an increase inbrain volume due to an increase in its watercontent.3 It can be localised or generalised. Inthe conscious patient it produces symptoms ofraised intracranial pressure, but in the uncon-scious the symptoms are masked. Cerebraloedema developing as a result of hyponatrae-mia is well documented4'6 but most of thesecases have developed postoperatively or follow-ing intravenous administration of fluids in aconscious patient. The event described hereoccurred during anaesthesia, and at the end ofthe operation-about four hours later-thepatient, a child, did not wake up and haddeveloped papilloedema. Urgent computerisedtomography showed gross cerebral oedemawith slit-like ventricles. Brain stem tests werecarried out and he was declared dead the nextday, about 26 hours from the start of theoperation. This case illustrates the complexityof fluid management in an intraoperative fatal-ity. To pathologists carrying out thesenecropsies-most probably at the behest of thecoroner-it is important to realise that asymp-tomatic dilutional hyponatraemia can occurintraoperatively when the symptoms of hypo-natraemia and cerebral oedema are masked dueto anaesthesia and unconsciousness. Arieff et al4studied 16 cases of symptomatic postoperative

444

on August 26, 2021 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.50.5.442 on 1 May 1997. D

ownloaded from