1975, British Journal of Radiology, 48,190-199

Liver and pancreas scanning in extrahepatic obstructive jaundice (with special reference to tumours of the bile and hepatic ducts) By J. E. Agnew, B.A., M.Sc, O. James, M.A., M.R.C.P.,* and I. A. D. Bouchier, M.D., F.R.C.P., F.R.C.P.(Edinburgh)f Departments of Physics and Medicine, Royal Free Hospital, London NW3 2QG (Received January, 1974 and in revised form May, 1974)

198

ABSTRACT

Au-gold colloid liver scans and 75Se-selenomethionine pancreas scans in 72 patients with extrahepatic obstructive jaundice were assessed by blind marking. They were compared with liver and pancreas scans from 20 control patients and liver scans from 33 patients with diffuse liver disease. 56 per cent of the liver scans in extrahepatic obstructive jaundice showed a filling defect in the hilar region of the liver. This was most frequently seen in the most deeply jaundiced patients, and was reported in 80 per cent of patients with a serum bilirubin greater than 15 mg/100 ml. The liver scan alone cannot distinguish between different forms of extrahepatic obstructive jaundice although severe loss of left lobe uptake appeared to favour a diagnosis of carcinoma of the bile or hepatic ducts. A normal pancreas scan virtually excludes a pancreatic carcinoma as the cause of obstructive jaundice. A pancreas scan showing severely reduced uptake suggests a carcinoma of the pancreas or of the lower end of the common bile duct.

Accurate diagnosis in patients with obstructive liver function tests is very important but can be extremely difficult. Pre-operative percutaneous cholangiography, and recently endoscopy with retrograde cholangiography, have made the management of such patients much easier. However, these procedures have drawbacks. Percutaneous cholangiography is not without risk. If successful it virtually commits the surgeon to laparotomy within two hours. If unsuccessful, the diagnosis remains in doubt. Endoscopy can be difficult to perform and requires special training and technique. X rays obtained from endoscopic cannulation of the bile ducts may be difficult to interpret. Liver and pancreas scans are safe, non-invasive investigations. The liver scan may help to distinguish extrahepatic obstruction from other causes of apparently obstructive liver function tests. In particular the presence on the scan of a filling defect in the hilum of the liver favours a diagnosis of extrahepatic obstruction (Morris et al., 1965). The role of the pancreas scan lies in attempting to distinguish between pancreatic and extrapancreatic causes *Present address: Department of Medicine, University of Newcastle upon Tyne. fPresent address: Department of Medicine, University of Dundee.

for obstructive jaundice. Several reports have stressed that a normal pancreas scan virtually excludes pancreatic disease as the cause of obstruction (Doutre et al, 1970; Heslip and Overton, 1970; Agnew, Youngs and Bouchier, 1973). The converse does not apply. A significant incidence of scan abnormality occurs in patients found subsequently to have a normal pancreas. In addition an abnormal scan may result from pancreatitis secondary to blockage of the pancreatic duct associated with obstruction of the lower end of the bile duct. In order to define more fully the role of combined liver and pancreas scanning, two questions arise: (1) Is there a recognizable liver scan appearance in extrahepatic obstruction, whatever the cause? (2) Can combined liver and pancreas scanning help to distinguish between: (a) carcinoma of the bile and hepatic ducts, (b) pancreatic carcinoma, (c) non-malignant extrahepatic biliary obstruction? This paper attempts to provide some answers to these questions. It is based on radioactive colloid liver scans and 75Se-selenomethionine pancreas scans performed over a seven-year period and subsequently re-assessed by blind reporting. PATIENTS STUDIED

Carcinoma of the bile or hepatic ducts {referred to hereafter as duct carcinoma): 25patients

Diagnosis was established by laparotomy, or post-mortem, between one and 25 days after the scan except for one patient in whom the interval was four months. Details of the tumour site and of the surgical assessment of the normality or otherwise of the pancreas are given in Table I; the classification of tumour sites is shown in Fig. 1. In classifying surgical reports phraseology such as "slight enlargement of the pancreas" or "slight thickening" was interpreted as "normal". All patients had obstructive jaundice at the time of scanning. (Mean serum bilirubin 18 mg/100 ml.,

190

MARCH 1975

Liver and pancreas scanning in extrahepatic obstructive jaundice TABLE I OPERATION, BIOCHEMICAL AND SCAN FINDINGS IN DUCT CARCINOMA

Serumf

Results of blind reporting

til lculinf*

No.

Age Sex

phosOperation Serum* pha(or aubilitase topsy) rubin (KA Site of report on (mg/ units/ Scan tumour pancreas 100 ml.) 100 ml.) grade

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

65 64 62 62 54 70 48 66 39 62 57 66 61 48 50 67 71 44 73 58 74 56 36 48 56

A A A A A,B,C B B B B,C B,C B,C,D B,C,D C C C C C,D C,D D D D D D D D

M F M M F M M M M M F F F M F M M M M F F F M F M

Abnml. Abnml. Abnml. Normal Abnml. — Abnml. Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Normal Abnml.

46 >10 M0 4 24 10 17 38 6 28 7 8 22 22 6 32 6 9 14 25 21 14 7 26 31

95 88 93 157 31 28 40 51 45 32 21 44 40 125 76 74 56 48 28 64 45 70 166 80 50

•Normal range 0-1-0-8 mg/100 ml.

4 4 4 4 4 4 3 4 4 4 2 4 4 4 4 4 4 4 4 4 4 4 4 4 4

Enlarg- Patchy ement uptake 4—

+ — + —

+• +

—

++ — — + + ++ _ ++ ++ + +

— —

+ + — — — — ++ — _ _ ++ + + + -\— ++

Pancreas

Hilar filling defect

++ ++ + + + + + _ ++ _ + ++ + + + — — _ + 4++

Other Abfilling normal defect splenic (s) uptake

+ — + _ — —

— —

— _ — — _ + — — —

-L -f— _ — — _ — — —

—

-"

— _ __ _ _

-J__

"T-

Scan grade 4 4 4 1 4 4 4 1 3 2 1 2 4 4 3 1 1 2 2 2 2 2 3 2 2

Generalized in uptake

-f +

+ + — + + _ — — — 4-

-f— — — _ — — — — —

t Normal range 3-13 KA units/100 ml.

range 4-46, mean serum alkaline phosphatase 65 KA units/100 ml., range 21-166). The mean duration of jaundice prior to the scan was six weeks, excluding one patient with a ten-month history.

RIGHT & LEFT HEPATIC DUCTS

LIVER COMMON HEPATIC

Liver

/ t^mS^

CYSTIC DUCT

PANCREATIC DUCT FIG. 1. Schematic anatomy of the bile and hepatic ducts. (The letters A, B, C and D mark the four regions of the biliary tree referred to under "site of tumour" in Table I: A— lower end of common bile duct, B—upper end of common bile duct, C—common hepatic duct, and D—right and left hepatic ducts and their junction.)

191

Carcinoma of the pancreas with obstructive jaundice: 30 patients Diagnosis was established by laparotomy (or post-mortem) within 25 days of the scan. The tumour appeared to affect only the head of the gland in 24 cases and the whole gland in six. All patients were jaundiced when scanned (mean serum bilirubin 14 mg/100 ml., range 2-29; mean serum alkaline phosphatase 62 KA units/100 ml., range 26-250). The mean duration of jaundice was similar to that in duct carcinoma. Non-malignant obstructive jaundice with normal pancreas: 17 patients Jaundice was due to a stone, or stones, in the common bile duct (15 patients) or to stricture of

VOL.

48, No. 567 J. E. Agnew, O. James and A. D. Bouchier

the common bile duct (two patients). All patients underwent laparotomy within 70 days of the scan. The group was restricted to patients with a normal pancreas (as judged at laparotomy) to ensure that no patient with a pancreatic cause for obstructive jaundice was included. Patients with gallstone pancreatitis have been discussed elsewhere (James, Agnew and Bouchier, 1974). Although the mean duration of jaundice was similar, obstruction to the common bile duct was often incomplete and jaundice less deep than in the preceding groups (mean serum bilirubin 8 mg/100 ml., mean serum alkaline phosphatase 50 KA units/100 ml.). Control subjects: 20 patients Patients were included in this group only if laparotomy within 100 days of the scan showed no abnormality of the liver, pancreas or bile duct. Patients with previous gastric surgery, cirrhosis, duodenal ulceration, diabetes mellitus or gross obesity were excluded as all these conditions have been reported to be associated with pancreas scan abnormalities. The majority of the patients had initially been referred for pancreatic scanning on account of abdominal pain. Patients with diffuse liver disease: 33patients This group comprised 11 patients with haemochromatosis, nine with cryptogenic cirrhosis, six with alcoholic cirrhosis or fibrosis and seven with primary biliary cirrhosis. Diagnosis was established by conventional means including liver biopsy in almost all cases. However, very few of these patients underwent laparotomy. Only their liver scans are reported in this paper; their pancreatic scans and function tests are to be reported elsewhere (James et al., 1975). METHODS

Scans were performed with a five-inch crystal rectilinear scanner. All patients had a conventional pancreas scan started between five minutes and one hour after intravenous injection of 3 /v,Ci 75Seselenomethionine per kilogram body weight. This scan was performed at 24 cm/minute which in normal subjects corresponds to between 60 and 105 counts per cm over the region of the pancreatic head (Agnew et al, 1969); constant speed scanning facilitates the assessment of diminished uptake in the abnormal scan. Approximately 70 per cent of the patients also had a subtraction scan (Agnew et al., 1973). This was performed with a three-channel analogue unit—two channels being set to bracket the principal peaks of the 75Se y-ray spectrum and the

third the 412 keV peak of I98Au used as the liver marker ("Tc m was rejected as the liver marker because its 140 keV y emission coincides with one of the major 75Se emissions and H31 n m was rejected because of its appreciable decay during the time of a scan). All patients had a liver scan with 150 /xCi 198 Au-gold colloid (scanning speed 60 cm/minute, giving approximately 150 counts per cm over the right lobe of the liver). The scans were intermixed with scans from 128 other patients with a variety of diagnoses scanned over the same period and were then assessed by blind marking by three observers. Both liver and pancreas scans were classified in four grades (Agnew et al, 1973). Grade 1—reported normal by all three observers Grade 2—reported normal by two observers Grade 3—reported normal by one observer Grade 4—not reported normal by any observer. The observers reported on five specific liver scan abnormalities: enlargement, patchiness of uptake, filling defect in the hilar region, filling defect elsewhere and abnormal splenic uptake. They also reported whether each abnormal pancreas scan showed a moderate ( + ) or a severe ( + + ) generalized loss of uptake, or a localized filling defect. Except where otherwise specified, results quoted are based on the majority verdict of the observers. This blind marking showed that the five specific liver scan defects listed did not adequately specify all the patterns of uptake encountered. Accordingly the liver scans alone were re-submitted to a fourth blind reporter who was asked whether any hilar filling defect showed a "branching" pattern and whether left lobe uptake was severely impaired. Because indentation of the lower border of the liver, possibly attributable to pressure from the gall bladder, may resemble a filling defect (Freeman et al., 1969) and interpretation of the feature as an abnormality can constitute an important cause of false positive reports (Covington, 1973), this feature was not considered as an abnormality. In many patients the count rate at points over the head of the pancreas, the right lobe of the liver and a background area had been recorded on the conventional pancreas scan. These figures were used to calculate a pancreatic :liver count rate ratio (Agnew et al, 1969). RESULTS

Liver scans in duct carcinoma (Table I) Twenty-three of the 25 liver scans were unanimously reported abnormal (Grade 4). None was unanimously reported normal.

192

MARCH

1975

Liver and pancreas scanning in extrahepatic obstructive jaundice

FIG. 2. Liver scans in duct carcinoma showing hilar defect: (a) and (b) were reported by all three observers to show a moderate defect ( : ); (c) and (d) were reported by all three observers to show a severe defect (— L ).

The specific defect most commonly recorded was a localized failure of uptake in the hilar region (Fig. 2). Subsequent blind review of the scans by a single observer suggested that 13 of the 16 hilar defects seen showed some degree of "branching". Taking together reports of "hilar" and "other" filling defects, a localized defect was reported by at least two observers in 22 cases and by all three in 16 cases. Four patients (Nos. 4, 7, 8 and 16) had tumours less than 2 cm in diameter. All had abnormal scans; three showed a hilar defect. In contrast, a patient (Xo. 11) with a massive tumour replacing the entire left lobe had a grade 2 normal scan. Distant liver metastases were demonstrated in only three patients. Few reports of "other" filling defects can therefore be attributed to such metastases. However, several patients showed very thin uptake in the left

lobe. On review this was thought by the three observers to have hampered distinction between "hilar" and "other" defects. Later reporting by the fourth blind observer suggested that six patients showed severe loss of left lobe uptake. Three showed no uptake to the left of mid-line (Fig. 3). No relationship could be detected between the site of the tumour and the incidence of reports of a hilar defect. Pancreas scans in duct carcinoma (Table I)

Only five scans were unanimously termed normal (grade 1). All the abnormal scans (grades 3 and 4) were thought to show a generalized loss of isotope uptake throughout the gland. No filling defects were reported. Five patients (Nos. 1-5) had tumours of the lower end of the common bile duct (region A, Fig. 1); the scan was abnormal in four and normal in one.

193

VOL.

48, No. 567 J. E. Agnew, O. James and A. D. Bouchier Scans in carcinoma of the pancreas and in non-malignant obstructive jaundice (Tables II and III) Only one patient with obstructive jaundice due to a pancreatic carcinoma had a normal liver scan. The incidence of hilar defects in carcinoma of the pancreas was similar to that in duct carcinoma (60 per cent). Fewer hilar defects were reported in non-malignant obstructive jaundice than in the other obstructed patients but the difference was not statistically significant. Nine hilar defects in carcinoma of the pancreas and three in non-malignant obstructive jaundice showed "branching". Three patients with pancreatic carcinoma and two with non-malignant obstructive jaundice showed poor uptake of colloid in the left lobe of the liver but none showed total loss of left lobe uptake. Sixty per cent of the pancreas scans in pancreatic carcinoma showed severe generalized loss of uptake. This constitutes a marked contrast to the duct carcinoma patients among whom only two showed severe loss of uptake. Selection of patients for the non-malignant obstructive jaundice group was restricted to patients with pancreases surgically proven as normal. Five, however, had abnormal "false positive" scans—each showing a moderate loss of uptake.

FIG. 3. Liver scans in duct carcinoma showing severe loss of left lobe uptake.

Scans in control patients (Tables II and III) The incidence of false positive liver scans was high (25 per cent), but no defect was considered severe by any observer. Two patients appeared to have hilar filling defects. In one, the "defect" was possibly just an exaggeration of the normal interlobar notch; but in the other there was a marked defect although surgery one month later revealed no abnormality to account for this appearance. No control patient showed impaired left lobe uptake. The incidence of false positive pancreas scan reports was slightly lower. In no instance was a severe loss of uptake reported.

At operation only the patient with the normal scan had a normal pancreas. Seven patients (Nos. 6-12) had tumours involving the extrapancreatic portion of the common bile duct (region B, Fig. 1). The Liver scans in diffuse liver disease Twenty-six (79 per cent) of the liver scans in this pancreas was considered normal at surgery in five group were abnormal. Twenty-four (75 per cent) cases of whom four had normal scans. Twelve of the thirteen patients with tumours were classed as grade 4 and, of these, all but one predominantly above the level of the cystic duct showed an enlarged liver, a patchiness of uptake (regions C and D, Fig. 1) had a pancreas considered or both. Abnormal splenic uptake was reported in normal at surgery. Nine had normal scans; one 16 cases. One patient with haemochromatosis (No. 23) had a technically very poor scan. The re- showed a moderate hilar filling defect. Two patients maining three scans must be considered "false with haemochromatosis showed severely reduced positives". In each case the abnormality was a left lobe uptake. moderate loss of uptake throughout the pancreas and Relationship between hilar defect and depth of in only one was interpretation made difficult by a jaundice (Fig. 4) liver markedly overlapping the pancreas. Taking together all patients with extrahepatic 194

MARCH 1975

Liver and pancreas scanning in extrahepatic obstructive jaundice TABLE II DISTRIBUTION OF SCAN GRADES IN THE FOUR DIAGNOSTIC GROUPS (NO. OF PATIENTS IN EACH GRADE)

Pancreas scan grades

Liver scan grades Diagnosis

No. of patients

1

Duct carcinoma

25

—

1

1

23

Carcinoma of pancreas (with obstructive jaundice

30

1

—

5

Non-malignant obstructive jaundice (with normal pancreas)

17

2

—

Controls

20

10

5

2

2

3

5

8

4

8

24

1

—

1

28

3

12

7

5

4

1

3

2

8

8

4

—

3

4

1

4

TABLE III OF LIVER SCAN DEFECTS (NO. OF PATIENTS WITH SCAN REPORTED BY AT LEAST TWO OBSERVERS TO SHOW THE SPECIFIC DEFECT)

Defect in liver scan Diagnosis

No. of patients Enlarged liver

Patchy uptake

Hilar filling defect

Other filling defect

Abnormal splenic uptake

Duct carcinoma

25

13

9

16

5

3

Carcinoma of pancreas (with obstructive jaundice)

30

17

10

18

5

7

Non-malignant obstructive jaundice (with normal pancreas)

17

8

5

6

2

4

Controls

20

1

2

2

—

—

50-, 40oo a

30-

obstructive jaundice, the mean serum bilirubin level in the patients with hilar defects ( + or + + ) was very significantly higher than that in the patients without a hilar defect (p < 0-001 by Student's ttest). There was also a significant difference between the mean serum bilirubin level in the patients with a moderate ( + ) defect and those with a severe ( + + ) defect (p