NUCLEAR MAGNETIC RESONANCE TOMOGRAPHIC IMAGING IN LIVER DISEASE

963

NUCLEAR MAGNETIC RESONANCE TOMOGRAPHIC IMAGING IN LIVER DISEASE F. W. SMITH

Department of Nuclear Medicine, Aberdeen Royal Infirmary ANNE REID J. R. MALLARD J. M. S. HUTCHISON

Department of Bio-medical Physics and Bio-engineering, University

of Aberdeen The non-invasive diagnostic technique of whole-body nuclear magnetic resonance was (NMR) imaging evaluated in 30 patients with established liver disease and 20 patients without liver disease. Comparison with diagnostic ultrasound and radionuclide liver scan shows that NMR easily differentiates malignant tumours from benign cystic lesions and provides useful information in patients with cirrhosis and metastatic deposits. In the demonstration of space-occupying lesions in the liver, NMR is as sensitive as ultrasound and more so than radionuclide liver scans when the metastases are less than 1. 5 cm in diameter. In the demonstration of cirrhosis it is more sensitive than both ultrasound and radionuclide liver scan. The specificity of NMR is superior to both ultrasound and radionuclide liver scan, both of which only demonstrate the presence of lesions, whereas NMR tomographic imaging based on the proton spin-lattice time (T1) of tissue accurately indicates the nature of the lesion.

Summary

Introduction WHOLE-BODY nuclear magnetic resonance (NMR) imaging has been shown to be of value in the demonstration of benign cystic lesions of the liver,’in the diagnosis of both primary and secondary tumours,2 and in the differentiation of cystic and solid tumours of the kidney.3 These early reports gave no indication of the sensitivity or specificity of this new non-invasive diagnostic technique. Ideally, NMR imaging should be compared with X-ray computerised tomographic (CT) scanning. Unfortunately, no whole-body X-ray CT is available in our region and we have therefore compared the NMR images with diagnostic ultrasound, radionuclide liver scan, and, wherever possible, histology of needle-biopsy specimens or direct inspection of the liver at laparotomy. NMR imaging, as used here, produces an excellent anatomical display based on the differences in proton spinlattice relaxation times (T 1) of different tissues.4 The NMR images are of transverse tomographic sections 18-5mm thick, displayed either in monochrome in a similar way to X-ray CT displays or in colour with a range ofT values (or proton concentration values) coded with 16 colours.

The NMR imaging technique is based on measurement of the response of the hydrogen protons within tissues, mostly in tissue water, to an applied radiofrequency radiation. When a patient is placed in a magnetic field, all the protons align with the magnetic field and precess around it at a definite resonant frequency. If the resonant radiofrequency pulse is beamed into the patient, the protons absorb energy and change their alignment relative to the applied magnetic field: a 90° radiofrequency pulse sets them precessing at 90° to the field. When the irradiating pulse is ended, the protons one by one relax back to their original alignment and re-radiate the absorbed energy. This radiated resonant frequency is detected and measured, the size of signal being related to the number of protons involved. After a 180° radiofrequency pulse the proton spins are inverted and the length of time needed for their relaxation is associated with the environment of the protons, the intensity of the emitted radiation falling exponentially at a rate characteristic for their environment or the lattice of which they are a part-a measurement known as proton spin-lattice relaxation time (T1). The easier it is for them to pass energy to neighbouring molecules, the more quickly they can relax to their original state and the faster is the fall ofintensity of the re-radiated signal. Thus tissues with a large amount of bound water with protons closely bound to proteins, such as muscle and liver, have a short T1, whereas fluids such as urine, cerebrospinal fluid, and ascitic fluid have a

long T 1 . Patients and Methods

30 patients with established liver disease and 20 with normal livers who were being investigated for other abdominal complaints were examined with the Aberdeen NMR imager, with radionuclide liver scanning, and with abdominal ultrasound. All three investigations were done within three days of each other. The Aberdeen NMR imager, which has been described in detail elsewhere, 5-7 is based on a four-coil air-cored magnet producing a static field of 0 - 04 tesla, giving a proton NMR frequency of 1 - 7 MHz for the hydrogen protons of body tissues. It can accept the whole human body. Each section of 18’ 5 mm equivalent thickness requires data to be collected from 128 electrical signals, each signal being collected during a 1 s interval. Thus, all the data for each section are collected in just over 2 min. The radionuclide liver scans were carried out after the intravenous injection of 2 mCi 99mTcsulphur colloid. Four standard views of the liver were taken with an Ohio-Nuclear 410 gamma camera, 300 K counts for each view. A supplementary tomographic view of each liver was obtained with an

orbiting gamma camera based on a Nuclear Enterprises large-field gamma camera head linked with a DEC Gamma 11 data-processing unit. A Nuclear Enterprises Diasonagraph/EMIsonic 4200 scanner used for the ultrasonic liver scans. The diagnosis in all 30 patients with liver disease was established either histologically by needle biopsy or by direct inspection at laparotomy, apart from 1 patient with polycystic disease, in whom was

A, Gordon A. The origin of bone that forms in association with cancellous chips transplanted into muscle. Br J Plast Surg 1952; 5: 154-60.

18. Ham

Bassett CAL. Clinical

implications of cell function in bone grafting. Clin Orthop 1973;

87: 49-59. Abbott LC, Schottstaedt ER, Saunders JBCM, Bost FC. The evaluation of cortical and cancellous bone as grafting material J Bone Joint Surg 1974; 29: 381 - 14. GrayJC, Elves MW. Early osteogenesis in compact bone isografts: A quantitative study of the contribution of the different graft cells Calcif Tissue Int 1979; 29: 225 - 37. Daniel RK Mandibular reconstruction with free tissue transfers. Ann Plast Surg 1978; 1: 346-71

Glowacki J, Altobelli D, Mulliken JB The fate of mineralized and demineralized osseous implants in cranial defects. Calcif Tissue Int 1981, 33: 71-76 Marble HB.Homografts of freeze-dried bone in cystic defects of the jaws. Oral Surg Oral Wed Oral Path 1968; 26: 118-23.

19.

Senci N. On the healing of aseptic bone cavities by implantation of antiseptic decalcifed

bone. Am JJ Med Sci 1889; 98: 219-43. 20. Sharrard WJW, Collins DH The fate of human decalcified bone grafts. Proc Roy Soc Med 1961; 54: 1101-02. 21. Urist MR, Dawson E. Intertransverse process fusion with the aid of chemosterilized autolysed allogeneic (AAA) bone. Clin Orthop (in press). 22. Reddi AH, Huggins CB. Cyclic electrochemical inactivation and restoration of competence of bone matrix to transform fibroblasts Proc Nat Acad Sci U.S.A. 1974; 71: 1648-52. 23. Urist MR, Mikulski A, Boyd SD. A chemosterilized antigen-extracted bone morphogenetic alloimplant for bone banks. Arch Surg 1975 110: 416-28.

964

NMR

E.

tomographic

sections in liver disease.

A. Normal liver and spleen B Ear)B C Advanced cmhosis with ascites D. S1ultrpk snmi: :v. _ ;’,,:1": 1’" la,!aSèS tram Polycysnc liver F Large round metastases from an as yo unidentified pnmary G. Large hepatoma 11

a carcmoma

.:..’

965

diagnosis was made ultrasonically as an incidental finding during the investigation of Cushing’s disease. 20 patients who underwent abdominal imaging during the investigation of nonhepatic disease and who had normal liver-function tests at that time the

were used as

controls to assess the normal appearances of the liver.

Results

The NMR tomographic section (see accompanying figure) is viewed in the same convention as X-ray CT images-i.e., from below with the patient’s back at the bottom of the image and his right side therefore appearing on the left. The ranges of proton spin-lattice relaxation times in normal and diseased liver are shown in table I. The normal liver (figure, A) appears as an irregular display of T1 values, which fall in the range 140-170 ms and are colour-coded on our display in the blue range. The normal spleen is seen in the brown-orange range and normally has a T in the range 250-290 ms. In front of the spleen, the stomach is seen filled with fluid, on top of which is a bubble of air (black). The gastric fluid has a long Tj, which shows as white. When greater than 1000 ms it becomes black but surrounded by white: this accounts for the irregularity in the centre of the gastric fluid. Between the stomach and the vertebral body the aorta is clearly seen, with the inferior vena cava (IVC) to its right, and the common bileduct anterolateral to the IVC. In the normal liver intrahepatic bileducts are not normally visualised and when seen probably indicate obstructive jaundice. The T images do not display bone, presumably because of the low water content of bone and its similarity to the adjacent connective tissues. The spinal canal, containing cerebrospinal fluid of long T, (400-450 ms), is clearly seen as a round white structure behind the vertebral body. Blood vessels such as the coeliac artery, splenic artery, and supermesenteric artery have occasionally been imaged. The reason for the nonregular visualisation of these vessels is the section thickness (18.5mm) and the fact that we are currently making sections intervals. patients with varying degrees of cirrhosis and 1 with chronic active hepatitis have been studied. In the patient with chronic active hepatitis there was a slight increase in T1 radiating from the porta hepatis along the biliary tree. The TI in this case was in the range 170-180 ms. The figure (B) shows the liver of a 62-year-old man with alcoholic liver disease, in whom biopsy showed extensive fatty change and chronic inflammatory infiltrate in the portal tracts with prominent fibrosis consistent with early cirrhosis. The liver shows a generalised increase in TI in the range 180-220 ms. The spleen is slightly enlarged and has a longer T than normal. In advanced cirrhosis the liver T1 may be as long as 300 ms as it was in the patient shown in the figure (C). In this patient the liver is small and, like the spleen, is surrounded by ascitic fluid, which has a long T and therefore appears white and black within white, the black indicating "over range"

Metastatic tumours in the liver appear as well-demarcated of increased T in the range 280-450 ms. Although there is some overlap with cirrhosis in T! at the lower end of the range, the appearances of the metastases are quite characteristic (figure, D) and are easily differentiated from cirrhosis (B and C). Large metastases similarly provide no real problems in diagnosis because of their characteristic appearance and T (figure, F). 2 patients with primary hepatoma have been examined: in both there was a large infiltrating lesion of the liver with a T! in the tumour range of 280-450 ms. The tomogram shown (figure, G) is a low one through the right lobe of the liver, the upper pole of the left kidney being seen medial to the normal spleen. 3 patients with cholangiocarcinoma have been examined. In each the tumour was clearly demonstrated and the T was found to vary within a lower range than either hepatoma or metastases (range 200-350 ms). The case shown in the figure (H) demonstrates clearly the presence of tumour in the anterolateral aspect of the right lobe and also a smaller tumour mass in the posteromedial aspect. (The patient also had haemochromatosis, which does not give a characteristic appearance on the NMR image, unlike that seen on X-ray CT.) The spleen is slightly enlarged with the stomach (white) lying in front of it. The aorta and IVC are both easily areas

recognised. Benign cystic or blood-filled lesions of the liver are easily recognised. Haemangiomas and haemangiofibromas have a characteristic smooth lobulated appearance with a T which varies according to the constituents of the fluid they contain. The NMR appearances of a haemangiofibroma have been described! in which the T in the cystic mass was the same as that of blood (table I). A patient with a haemorrhagic serous TABLE I-PROTON SPIN-LATTICE RELAXATION TIMES

(TIN LIVER

DISEASE

at 20 mm

5

(T1>1000 ms). Two high sections through the liver (figure, D and E) show of black around either the spleen or the liver. In these instances the black represents the air-filled lung which, because of the very low proton density, does not give off a significant signal. This doubling up of colours for "over

a rim

range" and "under range" is only a minor problem in interpretation of images because the anatomical situation of air-filled structures is well recognised and because very fluid lesions usually have an irregular white margin (figure, C), .,indicating their fluid nature.

cyst was found to have a lobulated lesion with septae running through it. Its T was in the range 650-800 ms, which is

longer than pure blood but shorter than serum. In polycystic disease, where the simple cysts contain serous fluid, the Tof the cysts is greater than 1000 ms. A high tomographic section through the liver of a patient with polycystic disease (figure, E) shows two round black cysts in the liver. In this section airfilled lung is seen in front of and behind the liver. The right and left ventricles of the heart, the aorta, and the IVC

clearly

are

seen.

In obstructive jaundice the appearances are characteristic. The intrahepatic bileducts, which are not normally visualised, appear, and the common bileduct, which is seen in normal

subjects with

as a narrow

structure, is

seen as a

larger

structure

characteristic T of between 300 and 400 ms. Cholecystitis has been demonstrated in 3 patients. The gallbladder, which is not always seen in normal subjects, is a

966

visualised surrounded by an irregular margin of increased T presumably due to the associated oedema.

1,

Discussion Careful evaluation of 20 control patients with normal liverfunction tests indicates that normal liver is clearly demonstrated by NMR tomography and that the normal T1 for liver falls into the narrow range of 140-170 ms. 5 patients with histologically confirmed cirrhosis were readily diagnosed from the NMR tomogram, whereas utrasound diagnosed only 3 of them and radionuclide liver scan diagnosed 4. The degree of correlation between NMR, ultrasound, and radionuclide liver scan was good in the detection of malignant tumours of the liver (table II), only 1 TABLE 11-NMR TOMOGRAPHY IN LIVER DISEASE

that there is

large potential for ionising technique. a

this non-invasive,

non-

J. R. M. thanks the Medical Research Council for support to the over NMR imaging programme in Aberdeen and Johnson and Johnson/Technicare for the gift of the computer and display system. We thank Mr R. Selb:e :0: carrying out the imaging on each patient. Requests for reprints should be addressed Infirmary, Foresterhill, Aberdeen AB9 2ZB.

to

F. W. S., Aberdeen

Rew

REFERENCES 1. Smith FW, Mallard JRM, Hutchison JMS, Reid A, Johnson G, Redpath TW, Selbie RD. Clinical application of nuclear magnetic resonance. Lancet 1981; i. 78-79 2. Smith FW, Hutchison JMS, Mallard JR, et al. Oesophageal carcinoma demonstrated by whole-body nuclear magnetic resonance imaging. Br Med J 1981, 282: 510-12 3. Smith FW, Hutchison JMS, Mallard JR, Reid A, Johnson G, Redpath TW, Selbie RD Renal cyst or tumour - Differentiation by whole-body nuclear magnetic resonance

imaging. Diagnostic Imaging (in press). 4. Edelstein WA, Hutchison JMS, Smith FW, Mallard J, Johnson G, Redpath TW Human whole-body NMR tomographic imaging: normal sections. Br J Radiol 1981; 54: 149-51. 5. Mallard JR, Hutchison JMS, Edelstein WA, Ling CR, Foster MA, Johnson G In-vivo NMR imaging in medicine: The Aberdeen approach, both physical and biological Phil Trans R Soc B 1980; 289: 519-30. 6. Hutchison JMS, Edelstein WA, Johnson G. A whole body NMR imaging machine J Phys E: Scient Instruments 1980; 13: 947-55. 7. Edelstein WA, Hutchison JMS, Johnson G, Redpath TW. Spin warp NMR imaging and applications to human whole body imaging. Phys Med Biol 1980, 25: 751-56

RANDOMISED TRIAL OF PENTOXIFYLLINE VERSUS ACETYLSALICYLIC ACID PLUS DIPYRIDAMOLE IN PREVENTING TRANSIENT ISCHAEMIC ATTACKS

* Large space-occupying lesions were demonstrated by both these techniques, true nature was not, since they were described as tumours.

but their

with small metastases being underdiagnosed by radionuclide liver scan. All the other 13 cases were demonstrated by all three techniques. Benign cystic lesions of the liver, such as haemangiomata and simple cysts, are readily demonstrated by ultrasound, which shows them as cystic tumours but cannot exclude their being malignant with a necrotic cystic centre, and by radionuclide liver scan which demonstrates them as large space-occupying lesions. NMR tomography, by virtue of being able to measure accurately the T of the lesion as well as displaying it pictorially, gives an accurate indication of the composition and is therefore superior to the other two methods. Similarly, the diagnosis of cholecystitis is simplified by NMR tomography. In the 3 cases seen in this series both the ultrasound examination and radionuclide liver scan showed no abnormality, whereas NMR showed the enlarged gallbladder with surrounding oedema. In our experience NMR tomography does not demonstrate the presence of biliary calculi, an area where ultrasound is very sensitive, but it shows very clearly the presence of dilated bileducts and indicated the level of obstruction in each of the 4 patients with obstructive jaundice examined.

patient

Conclusion NMR tomographic imaging appears to be an excellent noninvasive technique for the demonstration of various liver conditions. In demonstrating the presence of malignant tumours it is as accurate as ultrasound and radionuclide liver scan and appears to be superior to them in the diagnosis of benign masses, obstructive jaundice, and inflammatory conditions such as cholecystitis. This short series suggests

ERNESTO HERSKOVITS ALBERTO VAZQUEZ ARTURO FAMULARI ROBERTO SMUD LEOPOLDO TAMAROFF HECTOR FRAIMAN ANA MARIA GONZALEZ JOSÉ VILA VICENTE MATERA

Hospital Juan A. Fernándes, Hospital Nacional Prof. A. Posadas, Hospital Sirio-Libanés, and Hospital Italiano, Buenos Aires,

Argentina In a multicentre trial to compare the ability of a combination of acetylsalicylic acid and dipyridamole (1050 mg + 150 mg/day, group A) to prevent recurrence of transient ischaemic attacks (TIA) with that of pentoxifylline (1200 mg/day, group B), 36 patients received the combination and 30 pentoxifylline. There was no statistically significant difference between the groups as regards age, sex, blood pressure, site of origin of TIA, and incidence of other risk factors. The incidence of recurrent TIAs during 1 year of follow-up was 28% in group A and 10% in group B; this difference was significant (p<0.05). The incidence of permanent strokes was similar in the two groups but distinctly lower (4.5%) than that usually reported after untreated TIA.

Summary

Introduction determining potential benefits of a therapeutic intervention in patients with acute cerebral ischaemia one should remember that a transient ischaemic attack (TIA) is a warning sign of an impending stroke in a third to a half of these patients within the first year. 1,2 Anticoagulation. antiaggregation, and surgery have all been said to modifv the natural course of the disease. However, none of these therapeutic measures has been generally accepted so far Several studies3-5 seem to show the favourable effect of anticoagulants against recurrences and even permar.en: strokes in patients after their first ischaemic episode. Some of IN

the