The Liver and Spleen

Subdivisions ( Fig. 23.2 )

Studies of the vasculature demonstrate an internal craniocaudal principal plane (dividing the liver into left and right) not usually visualised on imaging techniques. The principal plane is defined by three key landmarks: the IVC groove, the middle hepatic vein and the gallbladder fossa. The liver is further subdivided into Couinaud segments based on the vascular supply. The caudate lobe or segment I has an autonomous blood supply from both left and right branches of the portal vein and hepatic artery along with independent venous drainage directly into the IVC.

Liver parenchyma has a lobular structure, each lobe comprising a central draining vein surrounded by sinusoids bounded peripherally by portal tracts, a ‘triad’ of adjacent branches of the bile duct, portal vein and hepatic artery. At the cellular level, the liver is mainly composed of hepatocytes, stellate cells and Kupffer cells, a part of the RES. The liver receives approximately two-thirds of its blood supply from the portal vein and one-third from the hepatic artery. Blood drains via the hepatic veins to the IVC. During a meal, mesenteric blood flow volumes may double, increasing portal vein flow volumes correspondingly. The pressure difference between measurements in the wedged (occluded) hepatic vein and the IVC (the corrected sinusoidal pressure) is normally between 4 and 8 mm Hg.

Lobar Agenesis/Atrophy ( Figs 23.3 and 23.4 )

Right and left lobe agenesis has been reported but is controversial: the absence of supplying vasculature or dilated bile ducts is said to permit the diagnosis of true agenesis rather than early atrophy. Surgical hemihepatectomy or disease-related atrophy is more common. In normal livers, compensatory hypertrophy of the remaining lobe often occurs with corresponding displacement of the gallbladder.

Vascular Anatomy Variation ( Fig. 23.5 )

The common hepatic artery is one of the three major branches of the coeliac axis. After giving off the gastroduodenal artery, the main hepatic artery continues and divides into the right and left hepatic arteries. Variations of the hepatic arterial supply are important for radiologists and hepatic surgeons. The portal vein divides into right and left branches and variations are infrequent, although early branches arising from the main trunk or close to the main division may create problems during liver resection. Three major hepatic veins drain into the IVC in 70% of cases, but in the remaining 30% accessory veins occur (19% having two left hepatic veins, 8% two right hepatic veins and 2% two middle hepatic veins). Absence of the IVC is rare and associated with complete situs inversus but may occur with partial situs and a right-sided liver. In this circumstance the hepatic veins drain direct to one of the cardiac atria with the azygos vein replacing the IVC, passing posterior to the diaphragmatic crura into the chest. More commonly, aberrant gastric venous drainage of the posterior aspect of segment IV may occur and has been correlated with focal fat variation. Accurate definition of the vascular and biliary anatomy is particularly important before live donor liver transplantation.

Technique

Abdominal ultrasound (US) is routinely used with phased array transducers operating between 3 and 5 MHz, and Doppler capability, both spectral, colour and harmonic, is an integral part of the examination of the liver, allowing demonstration of hepatic blood flow and unequivocal bile duct identification. Contrast-enhanced US is variably used to add an arterial and portal phase study comparable with CT and MRI.

Normal ( Fig. 23.6 )

Normal liver parenchyma echo texture is homogeneous and slightly more reflective than adjacent renal cortex. Portal vein branches radiate from the hilum and have increased wall reflectivity. Hepatic veins converge on the IVC and right atrium and have walls indistinguishable from the adjacent parenchyma. At Doppler examination, the normal hepatic vein waveform reflects the transmitted right heart pressure changes with transient flow reversal flow during the cardiac cycle ( Fig. 23.7 ). The portal vein waveform is normally continuous antegrade (mean peak velocity approximately 15 to 25 cm/s) and may vary slightly with respiration and the cardiac cycle ( Fig. 23.8 ).

Technique ( Fig. 23.9 )

Current volumetric CT systems allow complete isotropic data acquisition of the upper abdomen in a few seconds and choice of section thickness post acquisition. Unenhanced imaging remains valuable for assessing diffuse hepatic changes, such as fat infiltration and iron deposition, and for evaluating focal changes, in particular subtle calcification and haemorrhage. Dual-energy CT systems may in future remove the need for a separate unenhanced acquisition and provide new characterisation methods ( Fig. 23.10 ). Multiphase contrast-enhanced imaging following intravenous (IV) administration of water-soluble iodinated contrast medium is routinely used for detection and characterisation of focal lesions. Many solid liver lesions have a predominantly arterial blood supply, whereas the liver parenchyma receives 75%–80% of its blood supply via the portal vein. During contrast enhancement, early and late arterial phase studies, along with portal and delayed phase imaging, may be obtained. Optimising protocols and phase timing to maximise lesion-to-liver contrast varies with individual CT system, but the minimum requirement for liver imaging is typically a relatively late arterial phase (e.g. centred 18 s post contrast medium arrival in the abdominal aorta) and a portal venous phase. Delayed CT imaging is used in selected cases (e.g. haemangiomas and cholangiocarcinoma). CT arteriography (CTA) and CT arterioportography (CTAP) using direct hepatic artery injection during CT examination and Lipiodol CT are now rarely used.

Normal

Liver parenchyma is homogeneous, with attenuation values of 54–60 Hounsfield units (HU), usually 8–10 HU greater than the spleen. Vascular structures can be identified by their location on the unenhanced images and confirmed by enhancement with IV contrast medium. The peripheral intrahepatic biliary tree is not normally visualised, although the main right and left hepatic ducts and the common hepatic and bile ducts are normally demonstrated.

Techniques ( Fig. 23.11 )

MRI has a wider range of contrast mechanisms than other imaging techniques and is increasingly used for lesion detection and characterisation. Biliary tract anatomy and hepatic vascular patency can be assessed during the same examination. A wide range of protocols is available because of the numerous combinations of field strength, pulse sequence implementation and interdependent sequence parameters, all of which can influence image quality. Multicoil surface arrays are essential, and most studies are mainly breath-hold examinations as rapid MRI sequences can rival CT, although they may have compromised contrast performance that may limit lesion detection sensitivity. This is traded off with improved anatomical definition of extrahepatic structures. A typical MRI protocol includes breath-hold T ₂ - and T ₁ weighted (T ₂ weighted and T ₁ weighted) imaging and chemical shift imaging for hepatic steatosis detection. High-quality T ₂ weighted imaging can be obtained with respiratory-triggered multishot RARE sequences and pre- and multiphase post-gadolinium imaging using rapid breath-hold three-dimensional (3D) T ₁ weighted volume imaging is now routine. Diffusion-weighted imaging (DWI) is increasingly used to improve liver lesion detection. MR-based quantification has been developed for the measurement of hepatic steatosis, iron and fibrosis using chemical shift imaging, T ₂ and T ₂ * relaxometry and elastography. These techniques are undergoing standardisation and validation but are starting to enter routine clinical practice.

Intravenous Contrast Agents ( Fig. 23.12 )

Both non-specific intravenous gadolinium agents and liver-specific agents are in routine clinical use. Gadolinium-based agents that equilibrate rapidly with extracellular fluid include gadolinium diethylenetriaminepentacetate (Gd-DTPA) and gadoterate meglumine (Gd-DOTA), as well as the more recent non-ionic agents, which include gadodiamide, gadobutrol and gadoteridol. These agents provide enhancement on T ₁ weighted images similar to the iodinated contrast media at CT examination. Breath-hold 3D T ₁ weighted sequences allow the acquisition of multiphasic (arterial, portal, delayed) examinations as for CT. The enhancement characteristics for many focal lesions are, not surprisingly, similar to those for CT.

Hepatobiliary specific agents have been developed, which target either the reticulo-endothelial system (RES) or hepatocytes. Iron oxide particles possess superparamagnetic properties that create susceptibility-induced dephasing of protons, thereby shortening T ₂ . A range of ultra-small paramagnetic iron oxide (USPIO) agents have been developed with varying sizes and properties targeting mainly the reticulo-endothelial cells but also capable of functioning as blood pool agents for vascular studies. The availability of the iron agents varies across the world and, in some regions, they have been withdrawn, probably because of declining utilisation.

Mn-DPDP (mangafodipir trisodium), Gd-BOPTA (gadobenate dimeglumine) and, most recently, Gd-EOB-DTPA (gadoxetate) are all hepatocyte-specific paramagnetic agents that accumulate in hepatocytes followed by biliary excretion. They cause enhancement of the normal liver parenchyma and biliary tree on T ₁ weighted imaging and indicate the presence of hepatocyte function. Mn-DPDP is no longer available but the other agents have been used for increasing the sensitivity of liver lesion detection, lesion characterisation and the study of the biliary tract.

Normal

The intensity of normal liver parenchyma is the same as, or slightly higher than, that of adjacent muscle. This holds for all sequence combinations except for inversion recovery techniques with inversion times that completely null liver signal. The appearance of vessels varies widely on MRI depending on pulse sequence, artefact suppression techniques and contrast media. In particular, intravascular signal on conventional spin-echo sequences may occur normally and should not be interpreted as thrombus without confirmation using a reliable time-of-flight or contrast-enhanced technique. In routine practice, liver-spleen differences are helpful as a simple guide to effective intrinsic T ₁ and T ₂ weighting. In general, the spleen should be lower signal than the liver on effectively T ₁ weighted images and higher signal than the liver on T ₂ weighted images.

Technique

Liver/spleen imaging is usually performed following injection of a colloid agent such as ^99m Tc-sulphur colloid, injected intravenously. Most colloid is taken up by the Kupffer cells in the liver and 5% to 10% is taken up by the spleen. A small portion is also absorbed by the bone marrow. Sulphur colloid is cleared rapidly from the bloodstream ( t _1/2 = 2 minutes) and in patients with normal liver function imaging may begin 5–10 minutes after injection but in those with compromised hepatic function and/or portal hypertension, optimal concentration of the sulphur colloid will take longer, and imaging can be delayed to account for this.

Gamma camera images are obtained in multiple projections, and liver/spleen angiographic and blood flow phases can also be obtained at the start of a study by acquiring rapid sequential images during the first 30 to 60 seconds. Single-photon emission computed tomography (SPECT) imaging can be employed to evaluate suspicious areas for focal or diffuse space-occupying disease. PET and PET/CT can provide both projection and tomographic images using a range of cyclotron-generated radionuclides with varying half-lives.

Normal

The evaluation of a sulphur colloid scintigram involves an assessment of liver size, shape, distribution of the radiopharmaceutical within the spleen, liver and bone marrow and the homogeneity of uptake within the liver and spleen. Peripheral indentations on the liver are normally produced by the lateral rib margins, xiphoid process, gallbladder, right kidney and heart. The hepatic veins make a triangular impression on the superior, central margin of the liver, and the porta hepatis makes an impression on the inferomedial segment of the right lobe.

Hepatic Steatosis

Diffuse steatosis is an increasingly common finding that reflects increased triglyceride loading of hepatocytes. It has a wide range of causes, including acute and chronic alcohol abuse, obesity, diabetes mellitus, insulin resistance, cystic fibrosis, malnourishment, total parenteral nutrition, tetracyclines, steroids and ileal bypass. Linkage to metabolic syndrome and cardiovascular disease make this formerly ignored condition the subject of much research interest. Focal fat variation is also common and is discussed later.

US detects hepatic steatosis through increased parenchymal reflectivity, which obscures the portal vein margins ( Fig. 23.13 ). Although this finding can be virtually diagnostic, further imaging may be required as fibrosis can also cause increased reflectivity. CT can demonstrate and quantify diffuse hepatic steatosis as the attenuation decreases by approximately 1.6 HU per mg of triglyceride increase per gram of liver substance. Confounding changes such as fibrosis, drug treatment and conditions such as haemochromatosis make this unreliable. The liver architecture is preserved, especially the vascular pattern and the liver enhances normally following IV contrast medium. With increasing fat infiltration the liver attenuation decreases, reversing, in turn, the normal liver-spleen difference and liver-blood difference ( Fig. 23.14 ). MRI is the most sensitive and specific technique for demonstrating hepatic steatosis. Dixon-based (see Fig. 23.11 ) ‘chemical shift’ or ‘in- and out-of-phase’ imaging ( Fig. 23.15 ) allows both an accurate diagnosis and, with appropriate T ₂ and other corrections, accurate quantification. MRI is also the most accurate test for diagnosis of focal fat variation.

Cirrhosis

Cirrhosis is the end stage of a wide variety of hepatic disease processes that cause hepatocellular inflammation and necrosis leading to hepatic fibrosis and nodular regeneration. Early changes may be detectable only on histological examination. As cirrhosis progresses, widespread fibrosis and nodular regeneration develop, along with macroscopic changes of liver morphology that can be detected on imaging. Liver stiffness also increases, but the commonest anatomical finding in advanced cirrhosis is atrophy of the posterior segments (VI, VII) of the right lobe. Hypertrophy of the caudate (I) lobe and of the lateral segments of the left lobe (II, III) is frequently seen. In primary sclerosing, cholangitis caudate lobe hypertrophy is found in virtually all cases, and the lateral segments of the left lobe (II, III) occasionally atrophy. The cause of these changes is uncertain but thought to be blood flow related. Hepatic and portal system dynamics may alter radically in cirrhosis, with both increased overall hepatic blood flow (through intrahepatic arteriovenous shunts) and decreased hepatic blood flow (resulting from increased intrahepatic vascular resistance) recognised in advanced disease.

US can demonstrate the nodularity of the liver margin in advanced cirrhosis, particularly when ascites is present and when using high-frequency transducers. Pure hepatic fibrosis increases reflectivity, resulting in loss of the margins of the portal vein branches, but is thought not to alter attenuation, a feature in the past used to discriminate steatosis from fibrosis. However, in practice both steatosis and fibrosis often coexist, making separation difficult. The liver texture becomes coarser or more heterogeneous as cirrhosis progresses, but this is difficult to quantify and subjective. As the liver atrophies in end-stage cirrhosis, the hepatic veins may become attenuated and difficult to visualise. Doppler US examination may reveal other non-specific features of cirrhosis: damping of the normal right heart waveforms in the hepatic veins reduced main portal vein blood flow (<10 cm/s mean peak) or hepatofugal flow. Occasionally, increased flow in a large recanalised paraumbilical vein will ‘steal’ blood from the right portal vein branch, leading to reversed flow in the right portal vein, but normal hepatopetal flow in the main and left portal veins. Hepatic arterial flow is usually increased in advanced cirrhosis as the portal contribution to hepatocyte perfusion decreases. This results in enlargement of the hepatic arterial system, which can be mistaken for enlarged bile ducts on US unless Doppler techniques are used to identify the vessels. Over the last decade several forms of US elastography have been developed that evaluate liver stiffness. These vary from a 1D non-imaging method ‘transient’ elastography to a pulsed shear wave method combined with 2D imaging ‘acoustic radiation force imaging’. Several of these methods provide absolute quantification of liver stiffness and large trials suggest that these techniques may have a role in the detection and quantification of liver fibrosis, although their exact role in patient management is not yet clear.

CT ( Fig. 23.16 ) is insensitive to early fibrosis changes but demonstrates the nodular margin and lobar atrophy/hypertrophy changes of advanced disease. On unenhanced examinations, regenerative areas have relatively normal attenuation but advanced fibrosis lowers attenuation, whereas the accumulation of iron in hepatocytes increases it. These features frequently coexist in many forms of cirrhosis, resulting in parenchymal heterogeneity both before and after enhancement with IV contrast medium. Portal phase imaging can be helpful in assessing portal vein patency, although flow volume and direction cannot be determined.

MRI is also insensitive to early fibrosis changes and there are no specific changes of parenchymal signal intensity on T ₁ weighted or T ₂ weighted imaging, although parenchymal heterogeneity ( Fig. 23.17 ) may occur on T ₂ weighted and delayed post-gadolinium T ₁ weighted imaging, but is difficult to quantify. MRI delineates the morphological changes of advanced cirrhosis but can also provide non-invasive assessment of portal vein patency, along with flow direction and bulk flow volume estimation, when other techniques have proved unhelpful. Studies using DWI and ³¹ P spectroscopy have given mixed results for trying to grade fibrosis. MR elastography can quantify liver stiffness in a similar fashion to US methods and can accurately detect advanced fibrosis. Further research is assessing its ability to detect relatively early stages of hepatic fibrosis.

Colloid scintigraphy is rarely used but in established cirrhosis demonstrates reduced, heterogeneous hepatic uptake and increased extrahepatic uptake. In advanced disease, morphological changes may be detected. Angiography may be used to assess vascular complications such as variceal bleeding and portal hypertensive changes. Hepatic arteriography in cirrhotic liver demonstrates increased tortuosity of intrahepatic branches, so-called ‘corkscrew vessels’, which reflect lobar shrinkage.

Viral Hepatitis

Viral hepatitis, including hepatitis B and hepatitis C, remains a major public health concern, as it may lead to liver failure and primary liver cancer, often detected late. Diagnosis and monitoring based on serological tests and imaging are relatively non-specific. In acute hepatitis, imaging excludes obstructive causes of jaundice. In chronic hepatitis with cirrhosis, imaging helps monitor disease progression, development of portal venous hypertension and complications such as hepatocellular carcinoma (HCC).

On US examination, non-specific decreased reflectivity occurs in acute viral hepatitis, although most cases have normal parenchyma. Gallbladder wall thickening is a common non-specific finding in acute hepatitis. On colloid scintigraphy the appearance of hepatitis is similar to the early stages of cirrhosis, with uneven and reduced uptake. In chronic hepatitis. CT, MRI and angiography are of limited value until cirrhotic changes develop.

Haemochromatosis and Iron Overload

Haemochromatosis and multiple transfusions may both result in iron deposition in the liver. Inherited genetic haemochromatosis causes hepatocyte iron accumulation (leading to subsequent cirrhosis) and iron accumulation in other organs, including myocardium, skin and endocrine glands. Affected individuals have an increased risk of developing malignancy in general and of HCC in particular. Initially the hepatic iron deposition is diffuse, but the development of cirrhosis and regenerative changes often results in uneven distribution. By comparison, hepatic iron overload from multiple transfusions (haemosiderosis) results in iron accumulation in the RES (Kupffer cells) in the liver, bone marrow and spleen. There is less risk of liver damage, and the pattern of organ involvement can aid diagnosis.

MRI ( Figs 23.18 and 23.19 ) is the most specific imaging technique, as intracellular iron exerts a local susceptibility effect, reducing parenchymal T ₂ and T ₂ *. This effect is most sensitively detected by T ₂ *w gradient-echo imaging, although with significant accumulation the effect is easily seen on T ₂ weighted spin-echo images, and when severe will affect T ₁ weighted images. The liver signal is abnormally reduced (to less than that of adjacent muscle). Several studies have demonstrated that hepatic iron concentration correlates strongly with both T ₂ * and T ₂ value, permitting accurate quantification. Abnormally reduced signal on T ₂ weighted imaging is the main feature in other affected organs such as spleen and pancreas. Unenhanced CT demonstrates hepatic iron deposition through an increase in HU value (>75 HU; Fig. 23.20 ), but this also occurs in amiodarone treatment and previous Thorotrast exposure. The changes are unreliable because of the confounding effect of steatosis. US may demonstrate increased parenchymal reflectivity, but there are no specific features that characterise iron deposition.

Wilson Disease

Wilson disease is an autosomal recessive disorder in which copper is deposited in the liver, as cornea and lenticular nucleus of the brain. Copper is hepatotoxic and triggers inflammation that progresses to cirrhosis. Imaging demonstrates the generalised cirrhotic changes, but the underlying cause is rarely evident. Copper accumulation rarely causes a detectable increase in hepatic attenuation on CT, and there is often coexistent steatosis counteracting the effect. On MRI, there may be a subtle increased signal on T ₁ weighted, with a decrease on T ₂ weighted images. There are no specific features on US studies.

Calcification

Benign parenchymal calcification may occur following focal insults such as tuberculosis, Pneumocystis infection, sarcoidosis, pyogenic abscess and parenchymal haematoma. The calcification is well demarcated and surrounded by otherwise normal parenchyma. In Pneumocystis carinii infection, widespread focal calcification may occur. Focal calcification also occurs within benign lesions (giant haemangioma) and malignant lesions, particularly mucin-secreting adenocarcinoma of the colon, where it is often relatively ill defined. Primary liver tumours such as hepatoblastoma and fibrolamellar hepatoma may also contain foci of calcification.

Plain radiographs demonstrate gross calcification, but unenhanced CT is more sensitive and detects subtle calcification: for example, metastases ( Fig. 23.21 ). US clearly demonstrates focal calcification, with increased reflectivity and a posterior acoustic corridor, but this feature alone does not always allow distinction from focal gas. Scintigraphy and MRI are insensitive to calcification.

Pneumobilia

Gas in the biliary tract may occur because of a sphincterotomy, or Roux loop procedure, allowing reflux of intestinal gas into the biliary tree. It can be identified by the linear distribution radiating from the hilum and gravity dependence with air predominantly in the non-dependent parts of the biliary tree.

Both US and CT ( Figs 23.22 and 23.23 ) demonstrate clearly pneumobilia and its distribution. On US the ducts are increased echo-reflective linear structures that may be differentiated from calcification by the pattern and movement of the gas related to respiration, bowel peristalsis or patient position. CT is extremely sensitive to the presence of gas, which is easily demonstrated and localised.

Portal Vein Gas

Portal vein gas is always abnormal and occurs when intestinal permeability increases and/or there is an increase in intestinal luminal pressure. These conditions are fulfilled in neonatal necrotising enterocolitis but also in adults with gastric emphysema, intestinal obstructions, infections and Crohn disease. It has also been described in blunt abdominal trauma, invasive abdominal malignancies (colon carcinoma, ovarian carcinoma), duodenal perforation at endoscopic retrograde cholangiopancreatography (ERCP) and in patients with colitis following a barium enema. The significance and outcome largely relate to the underlying aetiology. The gas typically radiates out from the hilum with less marked gravity dependence than pneumobilia and a more peripheral distribution (see Fig. 23.22 ).

US sensitively detects moving gas bubbles in the main portal vein, which can be visualised on B-mode images and detected by spectral Doppler as the gas bubbles reflect the sound beam overloading the system receivers, giving rise to a characteristic high-pitched random bubbling sound with focal aliasing artefacts on the spectral display. The phenomenon occurs with both portal vein gas bubbles and microemboli. If sufficient gas accumulates, it may become visible on CT peripherally in the portal vein branches and eventually becomes evident on plain radiographs.

Parenchymal Gas

This is abnormal and results from a gas-forming organism in an abscess or infarct, or occasionally following trauma or hepatic arterial thrombosis following liver transplantation. It may be seen after embolisation or thermal ablation of liver tumours.

CT ( Fig. 23.24 ) best delineates parenchymal gas collections and any related pathological changes. US will demonstrate gas collections, but defining their extent may be difficult when they are large or peripheral and may be confused with adjacent bowel.