Gut 1992; 33: 1671-1674 1671 Non-invasive diagnosis of portal occlusion by radionuclide Gut: first published as 10.1136/gut.33.12.1671 on 1 December 1992. Downloaded from

P MacMathuna, M K O'Connor, D G Weir, P W N Keeling

Abstract TABLE I Clinical details ofthe three patient groups The accuracy of non-invasive in detecting portal vein occlusion PVO Cirrhosis Control* was assessed in 61 patients - 10 with portal vein No 10 25 26 Mean age (years) 47 52 4 occlusion confirmed by conventional porto- Male/female 5/5 10/15 14/12 graphy, 25 with chronic disease and a Liver disease 3/10 25/25 0/26 patent portal vein (mild=12, severe=13), and PVO=portal vein occlusion. 26 with normal liver function, who served as *Controls=.patients undergoing bone, cardiac, or cerebral controls. The median percentage portal scanning. venous flow for the portal vein occlusion group was 8% (range 1-30) (consistent with negligible and in post-transplant graft dysfunction.4 flow) compared with 78% (52-87) for control At present, the accurate diagnosis of PVO subjects (p<0.005) and 68% (61-80) and 49% requires either arterial portography or spleno- (23-59) respectively for patients with mild and portography, both of which are invasive severe liver disease (p<0.001 and p<0.005). radiological proceedures that may have compli- At a portal venous inflow of <20%, the cations, particularly in patients with a coagulo- proceedure had a specificity of 100% and pathy.56 A reliable non-invasive diagnostic sensitivity of 90% in diagnosing portal vein technique would represent a considerable occlusion. Non-invasive radionuclide angio- advance. Recent reports have advocated either graphy provides a safe and accurate screening ultrasonography78 or computer tomography.9 method for evaluating portal vein patency or The application of radionuclide angiography in occlusion in the investigation of portal hyper- assessing portal vein patency or occlusion has tension or before . received limited attention. We have previously (Gut 1992; 33: 1671-1674) described a new non-invasive dynamic hepatic scintigraphic technique to define the portal

venous and hepatic arterial contributions to total http://gut.bmj.com/ Complete or partial occlusion ofthe portal vein is liver perfusion.'0 More 'recent in vivo animal a common cause of in child- studies provide experimental evidence to hood.'2 In adults, portal vein occlusion (PVO) validate this technique." Preliminary use of this may complicate chronic liver disease but more method in a small number of patients success- frequently results from intra-abdominal sepsis, fully detected PVO.'° The present study aimed neoplasia, or a thrombotic disorder.3 In many to assess the accuracy of this technique in the

cases, however, no aetiological factor is found. diagnosis of PVO in a larger cohort of patients. on September 25, 2021 by guest. Protected copyright. Departments ofClinical Gastrointestinal haemorrhage is the commonest and Nuclear Medicine, Trinity clinical presentation, while the presence of College Medical School, ascites and encephalopathy usually indicate Patients and methods St James' Hospital, associated liver dysfunction. Early recognition of Dublin P MacMathuna the condition is important since management PATIENTS M K O'Connor may differ from that of cirrhosis, particularly if Three groups of patients were included in this D G Weir portacaval surgery is considered.3 PVO is a investigation (Table I). P W N Keeling contraindication to liver transplantation and Correspondence to: Dr Padraic MacMathuna, after transplantation, thrombosis of the portal Gastrointestinal Unit, Mater vein (or hepatic ) is a recognised cause of Group 1: patients with PVO (Table II) Misericordiae Hospital, Dublin 7, Ireland. graft dysfunction.4 Consequently, defining the Ten patients (5 men and 5 women) with a mean venous Accepted for publication patency or occlusion of the portal system age of 47 years (range 21-72) were assessed. The 20 May 1992 is equally important in pretransplant assessment diagnosis of PVO was confirmed in each case by TABLE II Individual clinical and radiological details concerning the 10 patients with portal vein occlusion Patient Age Cavernous no (years) Aetiology Cirrhosis Clinicalfeatures Diagnostic method transformation 1 54 Idiopathic No Oesophageal varices Angiography No 2 21 Idiopathic No Oesophageal varices Angiography No 3 55 Pancreatitis No Oesophageal varices Angiographv No 4 35 Idiopathic/? trauma Yes Oesophageal varices/hepatic Angiography No encephalopathy 5 63 Cirrhosis Yes Oesophageal varices/hepatic Angiographv No encephalopathy 6 40 Polycylhaemia rubra vera No Oesophageal varices/ascites Angiographv No 7 69 Idiopathic No Oesophageal sarices Angiographx No 8 38 Idiopathic No Asymptomatic +angiographv Yes 9 72 Cirrhosis/hepatocellular Yes Oesophageal varices Angio+post mortem Yes carcinoma? 10 51 Idiopathic No Oesophageal varices Angiographv Yes 1672 MacMathuna, O'Connor, Weir, Keeling

, Lung score`2 into two groups: mild disease with a Tc99 Child-Pugh score of <8 (n= 12) and severe Gut: first published as 10.1136/gut.33.12.1671 on 1 December 1992. Downloaded from ROI disease with a score of >8 (n= 13).

Group 3: normal control subjects There were 26 control subjects (14 men and 12 I women) with a mean age of 54 years (range 25- 78) none ofwhom had any clinical, biochemical, or radiological evidence of liver disease or extra- hepatic portal hypertension. These patients were undergoing cardiac, bone, or cerebral isotope scanning and dynamic liver scintigraphy could be incorporated without unnecessary additional exposure to radionuclides. Written informed consent was obtained from each patient before the study.

METHODS The technique of dynamic hepatic radionuclide angiography has been described previously in detail.'0 In brief, fasting patients were positioned supine beneath a large field of view gamma camera (General Electric, Maxi II). After a 5 minute resting period, in patients with PVO or liver disease, 370 MBq (10 mCi) of 99mTc Figure 1: Schematic outline illustrating thefield ofview pertechnetate were administered by bolus peri- ofthe gamma camera conventional arterial portography or spleno- pheral intravenous injection followed rapidly by including regions ofinterest portography with additional conclusive evidence a 30 ml saline flush to ensure rapid transit of the (ROIs) over the liver, spleen, lung, and left kidney. from laparotomy and post mortem examination radionuclide into the heart. In control subjects, in two cases (see Table II). In three patients there 555-740 MBq (15-20 mCi) of 99mTc labelled was coexisting liver disease and in one of these human albumin, methylene diphosphonate, or patients (patient 9), the PVO was associated with glucoheptonate were administered in a similar the development of hepatocellular carcinoma. manner. Computer image acquisition began at Pancreatitis (patient 3) and polycythaemia rubra the onset of the injection. Images were acquired http://gut.bmj.com/ vera (PRV) (patient 6), accounted for two other onto a 64x64 word matrix at a rate of 0 5 cases while no cause was evident in the remaining seconds/image for 100 seconds and stored on an five patients (idiopathic). A2 computer system for subsequent analysis (Medical Data Systems, Ann Arbor, MI, USA). The procedure takes approximately 3 minutes to Group 2: patients with chronic liver disease without complete. PVO Preliminary data analysis involved a summa- on September 25, 2021 by guest. Protected copyright. This group comprised 25 patients (12 men and tion of the initial 30 images to form a composite 13 women) with a mean age of 51 years (range image outlining the arterial phase of tracer 30-74), all of whom had documented chronic transit through the hepatobiliary system. liver disease (clinical, biochemical, and histo- Regions of interest (ROIs) were drawn over the logical criteria) with patent portal vein as shown liver, spleen, lungs, and left kidney ensuring that radiologically. The aetiological factors for the the liver ROI excluded the descending aorta and liver disease included alcohol (n=19), chronic right kidney (Fig 1). The ROIs were then used to active hepatitis (n=3), Wilson's disease (n=2), generate time-activity curves (TACs). Because of and haemochromatosis (n=l). Twenty of the interference from the right lung into the liver patients had portal hypertension, shown by the ROI, a correction factor was applied to take presence of endoscopically proved oesophageal account of this 'cross-talk', resulting in a varices, with or without splenomegaly. The corrected liver TAC (Fig 1). Deconvolution patients were classified according to the severity analysis of the TACs was used to improve of liver disease using the modified Child's-Pugh separation of the arterial and portal venous

3.Or 10.0r A 60 Figure 2: Left: corrected liver and spleen curves. x 50 Centre: curves following 30 deconvolution analysis. Modified cJ 1*0 Right: magnitude ofthe 0 spleen modified so that the spleen upslope matches that ofthe liver curve. AS and AL 0 0 0 represent the areas under the 50 100 so 100 50 100 spleen and liver curves respectively. Time (sec) Non-invasive diagnosis ofportal vein occlusion by radionuclide angiography 1673

phases of total liver blood flow. In addition, this TABLE IH Individual results ofthe hepatic arterial (% HA) for the 10 patients with portal method takes account of tracer recirculation and and portal venous (% PV)flow venous occlusion Gut: first published as 10.1136/gut.33.12.1671 on 1 December 1992. Downloaded from therefore allows analysis of the deconvoluted curves by area under the curve as a measure of Patient no % HA % PV blood flow. The liver, spleen, and left kidney 1 98 2 TACs were deconvoluted with the lung curve 2 92 8 3 98 2 using a modified fast Fourier transform tech- 4 97 3 nique as described by Juni et al.'3 If the assump- 5 99 1 6 90 10 tion is made on an anatomical basis, that blood 7 99 1 flow through the splenic artery has a similar Mean (patients 1-7) 96.4 3-6 8* 70 30 pattern to hepatic arterial flow, then the splenic 9* 89 1 1 TAC can be used as a model for hepatic arterial 10* 82 18 flow. In order to compensate for differences in Mean (patients 1-10) 91-4 8-6 absolute flow and in attenuation by overlying *These patients also had cavernous transformation of the portal tissues, the splenic TAC was multiplied by a vein. constant so that the upslope of the splenic TAC overlies the early segment of the total liver patients with coexisting cavernous transforma- TAC (Fig 2). Thus, the modified splenic TAC tion of the portal vein were excluded, the calcu- approximates to the hepatic arterial TAC. lated % PV was further reduced to 3% (range The liver and modified splenic TAC are then 1-10). Only the three patients with cavernous integrated to yield the areas under the curve, AL transformation had a % PV flow of more than and As, respectively (Fig 2). The hepatic arterial 10% (11%, 18%, and 30% respectively) (Table component (% HA) was calculated from the III). formula: Figure 3 illustrates the differences in % PV flow between the patients with PVO and the two % HA=As/ALX 100 control groups (2 and 3). The % PV flow for and the portal venous component (% PV) control subjects (group 3) was 78% (range, 53- determined as: 86) which is consistent with known physiological estimates.'4 In group 2, a progressive reduction % PV= 100-% HA. in the % PV contribution with increasing Recent in vivo experimental animal studies have severity of liver disease was observed, from 68% validated this deconvolutional technique to in patients with mild disease (range 61-80) define % HA and % PV by dynamic radiouclide (p>0-10 v controls) to 49% in the group with angiography. severe disease (range 22-61) (p<0001 v con- The results are expressed as % HA and % PV trols). In two patients with decompensated http://gut.bmj.com/ and are shown as median (range). Specificity and cirrhosis (Child-Pugh score C12&13), the % PV sensitivity values were calculated and statistical flow was 23% and 28%, respectively. The % PV analysis was performed using the Mann- flow was significantly reduced in the PVO group Whitney U test for non-parametric data. p<005 when compared with both control groups 2 and 3 was taken as the level of significance. (p<0-0001 v controls, p<0.001 v mild, and p<0 005 v severe liver disease (Fig 3)). For a % PV flow of 20% or less, the specificity on September 25, 2021 by guest. Protected copyright. Results and sensitivity of this proceedure in accurately Figure 3: Results ofthe Table III outlines the individual values for % HA detecting PVO were 100% and 90%, respect- calculated percentage portal ively. At a % PV flow of 30% or less, the venous flowfor control flow and % PV flow in each of the 10 patients subjects, patients with mild with PVO (group 1). The % PV flow in group 1 specificity was reduced to 85% while the sensi- and severe liver disease, and which reflects tivity was increased to 100% patients with portal vein was 8-6% (range 1-30), negliable occlusion (PVO). The cross portal venous inflow to the liver. When the three bars represent medians. 100 r Discussion Confirmation of portal vein patency or occlusion relies upon venous phase arterial *I. predominantly 80 0.0 portography or splenoportography, both

jaL invasive and time consuming radiological pro- S. 0 0 cedures that may have complications. Several 0 0 ._ 60 0 investigators have attempted to define the portal venous and hepatic arterial contributions to total 0@0 flow non-invasive means including 40 0 liver blood by CLJu ultrasound78 and radionuclide angiography. "" -18 1 ~~~~~~~~0WThe degree of portasystemic shunting in cirrho- 0 20 0 sis has also been examined by scintigraphy, however per rectal administration of the radio- a nuclide is required. 2 21 The present investigation 0 is the first report of radionuclide angiography Controls Liver disease PVO specifically used to assess the patency/occlusion Mild Severe venous = (n= 10) of the portal system. (n 26) (n= 12) (n= 13) We have shown that the % PV flow in all the L j p< 0.001 patients with PVO was 8.6% with a range of 1 to p

hepatic artery. Furthermore, when the patients proceedure is safe and inexpensive. Further

with cavernous transformation are excluded, evaluation of radionuclide angiography in Gut: first published as 10.1136/gut.33.12.1671 on 1 December 1992. Downloaded from the observed portal venous input to the liver comparison with other non-invasive methods becomes negligible (<3%). Defining occlusion (ultrasound, computed tomography, and nuclear of the portal vein as a % PV flow of 20% or less, magnetic resonance) and conventional arterial the technique is highly specific and sensitive. portography merits consideration. The results indicate therefore that this radio- Preliminary results of this study were presented to British Society nuclide technique accurately detects occlusion to of Gastroenterology meeting, September 1988 and published in portal venous inflow to the liver, particularly in abstract form in Gut 1988; 29: A1461. patients without coexisting cavernous transfor- mation. Another report also demonstrated the 1 Webb LJ, Sherlock S. The aetiology, presentation and natural absence of any significent portal venous flow in history of extrahepatic portal venous obstruction. Q J Med 1979; 48:627-39. two patients in whom portal vein thrombosis was 2 Alvarez F, Bernard 0, Brunelle F, et al. Portal obstruction in subsequently confirmed at necropsy. 16 children 1: clinical investigation and haemorrhage risk. J Pediatr 1983; 103: 696-702. There are two circumstances that may poten- 3 Triger DR. Extrahepatic portal venous obstruction. Gut 1987; tially reduce the diagnostic accuracy of this 28: 1193-7. 4 Dixon A, Sherwood T. Radiological assessment. In: Calne R, technique. Firstly, the presence of a cavernous ed. Liver transplantation: The Cambridge-King's College transformation or 'cavernoma' represents the Hospital experience. London: Grune & Stratton, 1987: 91-3. 5 Bergstrand I. In: Abrams B, ed. Angiology. Boston, Mass: development of multiple periportal collateral 1971; ii: 1030. channels in order to bypass the obstruction 6 Burcharth F. Safe splenoportography: objective confirmation is still required. Hepatology 1988; 8: 425-6. within the portal vein. Consequently, this pro- 7 Raby N, Karani J, Powell-Jackson P, Meire H, Williams R. vides a potential route for portal venous blood to Assessment of portal vein patency: comparison of arterial portography and ultrasound scanning. Clin Radiol 1988; 39: reach the liver. The % PV flow observed in the 381-5. three patients with cavernous transformation 8 Webb LJ, Berger LA, Sherlock S. Grey-scale ultrasonography of portal vein. Lancet 1977; ii: 675-7. included in the present study ranged from 11% 9 Mathieu D, Vasile N, Grenier P. Portal thrombosis: dynamic up to 30%. In these circumstances, the sensi- CT features and course. Radiology 1984; 154: 737-41. 10 O'Connor MK, MacMathuna P, Keeling PWN. Hepatic tivity of this method of detecting portal vein arterial and portal venous components of liver blood flow: a occlusion may be diminished. Nevertheless, dynamic scintigraphic study. J Nucl Med 1988; 29: 466-72. 11 O'Connor MK, Krom RF, Carton EG, et al. Ratio of hepatic only one patient with documented PVO in the arterial to portal venous blood flow - validation of radio- present study had a % PV flow of greater than nuclide techniques in an animal model.J NuclMed 1992; 33: 239-45. 20%. 12 Pugh RNH, Murray-Lyon IM, Dawson JL, Pietroni MC, Secondly, a major reduction in portal venous Williams R. Transection of the oesophagus for bleeding oesophageal varices. BrJ Surg 1973; 60: 646-9. inflow to the liver associated with reversed 13 Juni JE, Thrall JH, Froelich J, Hickwa RD, Clinthorne NH. (caudal) blood flow, may occur in the absence of A simple technique for reducing deconvolutional artifact in scintigraphic studies. IEEE Trans 1982; 4: 174-7. obstruction within the portal venous system,22 as 14 Campra JL, Reynolds TB. The hepatic circulation. In: Arias I, http://gut.bmj.com/ observed in two of the patients with decompen- Popper H, eds. The Liver: biology and pathobiology. New York: Raven Press, 1982: 627-45. sated alcoholic cirrhosis with % PV flows of 28% 15 Biersack HJ, Torres J, Thelan M, Monzon 0, Winkler C. and 23% respectively (Fig 3). Consequently, % Determination of liver and spleen perfusion by quantitative sequential scintigraphy: results in normal subjects and PV flow values ofbetween 20 and 30% fall within patients with portal hypertension. Clin Nucl Med 1981; 6: what might be termed a diagnostic 'grey area'. It 218-20. 16 Wraight EP, Barber RW, Ritson A. Relative hepatic arterial seems appropriate at the present time, therefore, and portal flow in liver scintigraphy. Nucl Med Commun

that the suspicion ofPVO, particularly in patients 1982; 3: 273-9. on September 25, 2021 by guest. Protected copyright. 17 Fleming JS, Ackery DM, Walmsley BH, Karran SJ. Scinti- with decompensated liver disease as indicated by graphy of arterial and portal blood supplies to the liver. the presence of an appreciably reduced portal J NuclMed 1983; 24: 1108-13. 18 Parkin A, Robinson PJ, Baxter P, Leveson SH, Wiggins PA, blood flow, needs to be confirmed by conven- Giles GR. Liver perfusion scintigraphy - method, normal tional arterial portography. Nevertheless, at a range and laparotomy correlation in 100 patients. Nucl Med Commun 1983; 4: 395-402. % PV flow of below 20%, radionuclide angio- 19 Sarpar R, Tarcan YA. An improved method of estimating the graphy allowed a clear distinction to be made portal venous fraction of total hepatic blood flow from computerised radionuclide angiography. Radiology 1983; between patients with a patent and obstructed 147: 559-62. portal venous system. 20 Yen CK, Koblik P, Breznock B, et al. Portasystemic shunt fraction quantification using transrectal administration of In conclusion, this study shows that non- iodine-123 iodoamphetamine in dogs with chronic invasive radionuclide angiography is ofbenefit in ligation and after propranolol administration. J Nucl Med 1989; 30:1702-7. screening for obstruction within the portal vein 21 Kashiwagi T, Azuma M, Ikawa T, et al. Portasystemic as part of the preliminary investigation of portal shunting in portal hypertension: evaluation with portal scintigraphy using transrectal administered I-123 IMP. hypertension and chronic liver disease or during Radiology 1988; 169: 134-40. the assessment for liver transplantation. This 22 Rector WG, Hoefs JC, Hossack KF, Everson GT. Hepatofugal portal flow in cirrhosis: observations on hepatic method may also be helpful in confirming the haemodynamics and the nature of the arterioportal patency/occlusion of a portacaval shunt. The communications. Hepatology 1988; 8: 16-20.