Nonalcoholic fatty liver disease (NAFLD) has become the most prevalent liver disease in the United States, affecting all age groups. It encompasses a spectrum of disease, ranging from simple steatosis, steatohepatitis (NASH), fibrosis and cirrhosis, to hepatocellular carcinoma. (1) This issue of The WCC Note examines the imaging avenues to diagnosis and quantify hepatic steatosis. [...]
Nonalcoholic fatty liver disease (NAFLD) has become the most prevalent liver disease in the United States, affecting all age groups. It encompasses a spectrum of disease, ranging from simple steatosis, steatohepatitis (NASH), fibrosis and cirrhosis, to hepatocellular carcinoma. (1) This issue of The WCC Note examines the imaging avenues to diagnosis and quantify hepatic steatosis.
What does recent literature report about ultrasound imaging of hepatic steatosis?
1. The gray scale findings include the following: 
a. The diagnosis of fatty liver can be made if:
i. The liver is more echogenic than
the renal cortex and spleen.
ii. Ultrasound wave attenuation is present.
iii. The diaphragm loses definition.
iv. The intrahepatic architecture has poor
delineation.
v. There should be more than just one or
two of the above present. (2)
b. A 2008 review reported ultrasound sensitivity
ranged 67-84% and specificity 77-100% for
severe fatty liver (more than 30% fat by weight).
It has been reported as poor at diagnosing
lesser degrees of steatosis. (3)
c. Subjective visual assessment of fatty liver at
ultrasound has marked observer variability. (4)
d. A 2009 study reported that a hepatorenal sono-
graphic index of 1.49 (the ratio between the
mean brightness levels in a region of interest in
the liver and spleen) predicted steatosis of >5%
with a sensitivity of 90%, specificity 90%. Steato-
sis of >60%; specificity 93%). (5)
2. Elastography
a. A significant positive correlation was reported between
median acoustic radiation force
impulse elastography (ARFI) and liver fibrosis in patients
with NAFLD. (6)
What are some updates on CT of hepatic steatosis? 
1. Hepatic steatosis can be diagnosed on
CT if:
a. Noncontract
i. Liver attenuation is at least
10 Hounsfield units (HU) less
than the spleen.
ii. The liver attenuation is<48
HU (7, 8, 9); or <40 HU when
lipid is about 30% . (9, 10)
b. With contrast:
i. The comparison of the liver
and spleen HU is not as reliable. (2)
ii. Fatty liver can be diagnosed if liver
attenuation is less than 40 HU. (2)
2. Then if the liver is <40 HU, is that spe-
cific for liver steatosis?
a. No. Ischemic or mucinous metasta-
ses, or abscesses can have this
attenustion. Clinical laboratory, and
other imaging features need consid-
eration. (2)
3. Lipid quantification can be preformed by
the following methods:
a. Hepatic attenuation measurement
i. A value of 40 HU is reported
to represent fetty change of
approximately 30%. (9, 10)
b. Hepatic attenuation index
i. A ratio of hepatic HU to splenic
HU less than 0.8 is reported as
highly specific for moderate to
severe (>30%) macrovesicular
steatosis. (11, 9)
c. Hepatic attenuation difference at
dual-energy CT
i. Ma et al note, in review, that
while there is a paucity of literature
to validate its use, an increase in fatty
content associates with desreased HU at low
energy; when the energy level increases, the
fat attenuation increases. (9)
4. Unenhanced CT studies have reported:
a. Visual grading and liver attentuation index were shown reliable and similarly accurate for
diagnosis of 30% or higher macrovesicular steatosis in living hepatic donor candidates. (12)
b. Moderate to severe macrovesicular steatosis (i.e. >30%) can be accurately diagnosed in
the living hepatic donor, avoiding biopsy, but biopsy is still needed if the CT calculates
<30% fat. Coexistent fatty liver and hemosiderin or occult liver disease would be possible.
(13, 14, 15)
c. Low dose unenhanced CT detected hepatic steatosis in asymptomatic patients, while
clinical risk factor profiles proved unreliable. (16)
What MRI methods are used to evaluate hepatic steatosis?
1. Spectroscopy
a. This technique uses the frquency position along the x-axis to separate and character-
ize chemicals within voxels. (17)
b. Localized or single-voxel MRI. Sequences include:
i. Point-resolved spectroscopy (PRESS)
ii. Stimulated echo acquisition mode (STEAM)
iii. A reconfigured STEAM sequence has been reported with breath-hold acquisition of
TZ-corrected lipid measurement. (18)
iv. A disadvantage is that a large, single voxel is studied. (19)
c. The summation of individual lipid peaks calculates the total liver triglyceride content. (9)
2. Chemical shift imaging: Fat and water protons precess at different frequencies in a magnetic
field. Exploiting this allows for detection and quantification of fatty infiltration. Multiple
sequences have been developed on this basis. These are:
a. Two-point Dixion MRI
i. This techniques offsets the rephasing pulse in a spin echo (SE) sequence to create
out -of-phase images, with the unmodified SE images used as in-phase. Summation
and subtration of these images yields water-only and fat-only images to quantify fat,
but magnetic field inhomogeneity and longer scan times limit its use (9)
ii. A recent study at 3T reported a 2D decomposition technique to identify distinct
in-phase/opposed and fat/water ratios for in vitro steatosis, iron overload, and
combined disease. (20)
b. Three-point Dixon MRI
i. Developed to overcome the field inhomogeneity, it uses a third image with phase
correction but increases scan time. (9)
c. Modified Dixon
i. When faast gradient echo (GRE) was developed, this methodused shorter TEs and
TRs to decrease scan time and allow breath hold images.
ii. As reviewed by Ma et al, it can detect mild hepatic fat of 10% or more. (9)
d. Triple-echo chemical shift GRE
i. This breath-hold low flip angle technique with correction for T2* was reported to
accurately quantify hepatic fat. (21)
e. Opposed-phase T1
i. When fat and water proton magnetization are in phase, their signal is additive.
when out-of-phase, signal intensity decreases.
ii. Dual echo fast GRE sequences decrease scan time, allow breath hold imaging,
and minimize T2* when shorter TEs are used.
iii. Opposed-phase T1-weighted images showed signal intensity loss that could be
used to grade the severity of liver steatosis. (22)
iv. A relative signal decrease of less than 20% allowed correst prediction of liver
donation appropriateness in 53 of 57 patients. (23)
v. Using MR spectroscopy as the referenceee standard, in-and out-of-phase imaging
rapidly estimated liver fat content. A cutoff value of 5.1% discriminated between
normal and elevated liver fat. (24)
vi. Potential pitfall include:
1. The presence of liver iron, which can cause signal intensity loss on in-
phase images. (25)
2. Fat fractions>50%, which cannot be reliably assessed (26)
3. Fat is spectrally complex. (26)
3. MR elastography
a. This technique employs three phases. Mechanical waves are generated in tissue. The
micron-level displacements are imaged using motion-sensitizing gradients. Wave images
are processed to generate quantitative maps. (26)
b. MR 7T elastography detected early steatohepatitis in rats by showing increased elasticity.
(27, 28)
4. Low-flip-angle multiecho GRE
a. This is reported to provide high diagnostic and fat-grading accuracy in NAFLD. (29)
b. According to O’Regan et al., it can provide fat measurement without acquiring a separate
T2* map (unlike dual echo) and correlates highly with T2-corrected proton MR spectro-
scopy. (30)
5. Fast spin echo (FSE)
a. T2-weighted fat saturated FSE images are compared to T2-weighted non-fat-saturated
FSE images. A decrease in signal intensity on the fat-saturated images suggests fatty
infiltration.
b. This method avoids the T2* effect signal loss of liver iron in the cirrhotic patient, which
can be problematic in GRE sequences. (31, 9)
What are the patterns of hepatic fat deposition?
1. Diffuse fatty infiltration is most common. (2)
2. Focal depostion or diffuse fatty infiltration with focal sparing shows:
a. No mass effect,
b. Geographic shape,
c. Poorly defined margins,
d. Positioning adjacent to the porta hepatis, gallbladder fossa, ligamentum venosum, or
falciformm ligament (perhaps because of variant venous circulation), and
e. Contrast enhancement similar to or less than normal liver. (2)
3. Multifocal depostition:
a. Is an uncommon pattern with multiple fat foci an atypical locations,
b. May be round or oval,
c. Is a difficult diagnosis,
d. Must have microscopic fat,
e. Chemical shift GRE may be helpful, and
d. May be seen with regenerative nodules in cirrhosis. (2)
4. Perivascular:
a. Has fat halos around hepatic and/or portal veins;
b. Has an unknown pathogenesis. (2)
5. Subcapsular:
a. This distribution occurs in insulin-dependent diabetics on peritoneal dialysis who get
insulin added to the peritoneal dialysate.
b. The etiology is thought to be due to direct exposure of that region to a higher
concentration of insulin. (2)
6. Patients with fatty liver and concomitant focal liver lesions may display peritumoral sparing of
the fat, leading to atypical imaging appearances. (32)
What tumors are pitfalls and can contain microscopic fat?
1. Hepatic adenomas may contain microscopic fat.
Hepatocellular carcinoooooomas, angiomyolipoma, and nodular hyperplasia may contain
microscopic fat and soft tissue. (2, 33)
Conclusion: Noncontrast CT can accurately diagnose moderate to severe hepatic steatosis (>30%)
but is not accurate at lower levels. MRI techniques to detect and quantify hepatic steatosis currently
emphasize chemical shift imaging, with spectroscopy as the gold standard. Ultrasound suffers from
subjectivity and inability to diagnose lesser degrees of hepatic fat, though a recent study of hepatorenal index was encouraging.
Research and reporting by Margaret D. Phillips, M.D.
Reviewer and publisher: Stephen J. Pomeranz, M.D.
For full sources and credit, please download the PDF copy of the newsletter here
