Keywords
Bariatric surgery - disease classification - lipodystrophy - liver biopsy - nonalcoholic
fatty liver disease - pediatric fatty liver disease - steatohepatitis - steatosis
Definition and Classification of Fatty Liver Disease
Definition and Classification of Fatty Liver Disease
If one examines the introductory text from almost any article written about nonalcoholic
fatty liver disease (NAFLD) or nonalcoholic steatohepatitis (NASH), one is almost
certain to find a phrase that describes NAFLD as both “identical to alcoholic liver
disease” and “a spectrum of changes from steatosis to NASH to cirrhosis.” Although
there is some truth to these phrases, both are problematic. Although there may be
significant overlap with features of alcoholic liver disease, there are also distinctions
between the two.[1] Second, the concept of the spectrum oversimplifies a complex disease, and implies
that there is a continuous and unbroken movement from steatosis on one end to cirrhosis
on the other. Further, it suggests a disease categorization that is one-dimensional
and limited to a few disease states. The natural history of NAFLD on an individual
patient level is largely unknown and it is not clear that all patients with steatosis
are at risk for steatohepatitis or advanced fibrosis. Patients with steatohepatitis
may cycle in and out of states in which that diagnosis can be recognized histologically,
depending on therapeutic interventions and/or lifestyle modifications. There may be
more than one path to cirrhosis depending on a variety of factors including the age
of the patient. Disease classification in NAFLD must be robust enough to account for
these possibilities.
In addition to these disease classification and natural history issues, NAFLD itself
is a histologically complex disease. Unlike chronic hepatitis, which can be adequately
graded and staged by characterizing the inflammation and the fibrosis, the histologic
characterization of NAFLD and NASH by definition includes description of steatosis
and cell injury in addition to inflammation and fibrosis. These histologic features
are not specific by themselves and may be found as components in other liver diseases.
When they combine in specific patterns, we can recognize specific disease states such
as steatohepatitis, even in conjunction with other diseases. At times, however, the
pathologist may be limited to describing the degree of necroinflammatory injury and
carefully cataloguing the changes seen. We propose that, for the purposes of clinical
investigation, pathologists and clinicians categorize the histologic changes from
biopsies done to evaluate NAFLD into sets of diagnoses as outlined in [Tables 1] and [2] and described in detail below. Because many of the individual histologic features
of NAFLD may occur in other liver diseases and there remains no positive serologic
test for NAFLD, there should be adequate clinical exclusion of other forms of liver
diseases prior to categorizing a biopsy as a type of NAFLD.
Table 1
Histologic Categorization of Disease States in Patients at Risk for Nonalcoholic
Fatty Liver Disease (NAFLD)
|
Category
|
Definition
|
|
No significant evidence of fatty liver disease (Not NAFLD)
|
Insufficient steatosis for a diagnosis of steatosis (the NASH CRN uses a threshold
of 5% of hepatocytes showing steatosis), without other changes (ballooning, fibrosis)
that would suggest steatohepatitis.
|
|
Steatosis
• Steatosis with inflammation
• Steatosis with nonspecific fibrosis
|
Steatosis without specific changes to suggest a form of steatohepatitis. This category
may include spotty lobular inflammation and/or mild degrees of fibrosis of uncertain
significance.
|
|
Steatohepatitis
• Zone 3 borderline steatohepatitis
|
Form of steatohepatitis most common in adults; defined as a zone 3 centered injury
pattern that includes steatosis, inflammation, ballooning injury (often with Mallory-Denk
bodies) with or without fibrosis. Borderline steatohepatitis has some, but not all
of the features that would allow a diagnosis of steatohepatitis.
|
|
Zone 1, Borderline pattern
|
Form of steatohepatitis that occurs mainly in young children, characterized by zone
1-centered injury (portal inflammation, portal-based fibrosis, zone 1 steatosis, ballooning
injury in zone 1 if present).
|
|
Cryptogenic fibrosis/cirrhosis
|
Presence of fibrosis (usually advanced) or cirrhosis, with little to no steatosis
and no changes (ballooning, Mallory-Denk bodies) that would suggest borderline or
definite steatohepatitis. Other explanations for fibrosis (besides steatohepatitis)
should be considered.
|
NASH CRN, Non-alcoholic Steatohepatitis Clinical Research Network.
Table 2
Comparison of Features between Categories of Nonalcoholic Fatty Liver Disease (NAFLD)
|
Lesion
|
Not NAFLD
|
Steatosis
|
Steatohepatitis
|
Zone 1 Borderline Pattern
|
Cryptogenic Fibrosis/Cirrhosis
|
|
Steatosis
|
None to minimal (<5%)
|
Any degree >5%, often in zone 3 or panacinar
|
Any degree, often in zone 3 or panacinar when fibrosis is mild
|
Any degree, often in zone 1 or panacinar
|
None to minimal
|
|
Inflammation
|
Usually mild
|
Any degree, usually mainly lobular
|
Any degree, usually more lobular when fibrosis is mild
|
Any degree, portal inflammation may be more prominent
|
Any degree, portal inflammation may be more prominent
|
|
Ballooning
|
None
|
None
|
Present when pattern is definite, may be mild or absent in borderline cases when characteristic
fibrosis is seen
|
Usually absent, when present usually in zone 1
|
None
|
|
Fibrosis
|
Usually none, but may have stage 1–2 fibrosis
|
Usually none, but may have stage 1–2 fibrosis
|
Sometimes none, but usually fibrosis present, often perisinusoidal in zone 3
|
Usually present as periportal fibrosis (stage 1c) or portal-portal bridging fibrosis
(stage 3)
|
More than stage 2 fibrosis
|
|
NAS
(typical range)
|
0–1
|
1–5
|
3–8 (Definite)
2–6 (Borderline)
|
2–6
|
0–3
|
NAS, NAFLD Activity Score.
No Significant Evidence of Fatty Liver Disease (Not NAFLD)
When patients with suspected fatty liver disease are biopsied, one possible outcome
is that there is no evidence of fatty liver disease. This is known to occur in up
to 10% of bariatric biopsies and has been reported in nonmorbidly obese individuals
as well.[2]
[3]
[4] In addition, it might be the result of effective therapy or might simply be the
patient's true baseline state. Minimal steatosis might be present and for the purposes
of clinical scientific study, the NASH CRN (Non-alcoholic Steatohepatitis Clinical
Research Network) has used a literature based cutoff of 5% steatosis (meaning that
less than 5% of hepatocytes show steatosis) as the threshold.[5]
[6] Other mild, but nonspecific changes (including minimal inflammation or fibrosis)
could be present without affecting the categorization. These findings may represent
prior injury and resolution and may warrant a descriptive diagnosis and comment in
the pathology report. Diffuse glycogenosis might be present and may provide an explanation
for imaging findings. In routine clinical practice in which patients are screened
by current generation imaging techniques prior to biopsy, this diagnosis should be
unusual. In the setting of clinical research, patients who are at risk for NAFLD,
but have no histologic evidence of disease would make an excellent control group for
understanding the pathophysiology of steatosis.
Steatosis
It is important to remember that steatosis is not a specific feature of liver disease,
but a histologic finding that can be identified in various other liver biopsies for
elevated liver tests. In fact, zone 1 steatosis is a common distribution in chronic
hepatitis C. The pathologist should always be alert to other possibilities, particularly
when the clinical evaluation is incomplete.
Many patients biopsied to evaluate NAFLD will have a significant degree of steatosis,
but will not have evidence of steatohepatitis. There is macrovesicular steatosis that
is usually present in a zone 3 or panacinar distribution ([Fig. 1]).[7] Steatosis in a zone 1 distribution should suggest the zone 1, borderline pattern.
Azonal steatosis is most often seen in biopsies with advanced fibrosis, but the zonal
distribution may also be difficult to discern when the steatosis is either very mild
or the biopsy is fragmented or inadequate.[7] Spotty lobular inflammation is often present and can be in the form of small collections
of macrophages (microgranulomas) or lymphocytes, similar to the spotty inflammation
in chronic viral hepatitis. Portal inflammation is usually mild or absent; no ballooning
injury suggests steatohepatitis. Fibrosis, if present at all, should be limited to
mild periportal or perisinusoidal fibrosis. In the Central Pathology Committee of
the NASH CRN, this category has been used for cases in which there is clear zone 3
steatosis with or without periportal fibrosis, but without perisinusoidal fibrosis.
In such cases, the criteria for steatohepatitis are incomplete, and the temporal relationship
of steatosis and the fibrosis is not known. Because a biopsy with findings in this
category shows generally nonspecific findings, a descriptive diagnosis and comment
would be appropriate for the pathologist's report.
Figure 1 Distribution patterns of steatosis in nonalcoholic fatty liver disease. Shown are
the four distribution patterns of steatosis defined in the Non-alcoholic Steatohepatitis
Clinical Research Network feature scoring system. The red rectangles represent stylized
liver biopsies with triangular portal areas (P) and round central veins (V). The small
white circles denote steatotic hepatocytes. The amount of steatosis is the same in
each pattern and is relatively mild for demonstration purposes. (A) Zone 3 steatosis.
(B) Zone 1 steatosis. (C) Panacinar steatosis. (D) Azonal steatosis.
Steatohepatitis
Steatohepatitis, as a name for a unique diagnostic pattern that had been observed
in nonalcoholic, overweight adult patients, was first proposed in a landmark study
by the Mayo Clinic.[8] However, the alcoholic-like pathology of steatohepatitis was described in overweight
and/or diabetic patients long before the study that popularized the term, steatohepatitis
(see, e.g., early studies by Zimmerman[9]
[10]). Until recently, it was the only pattern within NAFLD that was recognized to be
associated with the development of advanced fibrosis. Unlike the steatosis category
described above, steatohepatitis is a specific pattern of liver injury that may be
recognized even though a second liver disease may be present. The presence of a characteristic
ballooning injury is the key to the diagnosis. Ballooning injury is a type of cellular
change in which the cells become enlarged and the cytoplasm becomes irregularly clumped
with optically clear, nonvesiculated areas. Steatotic vacuoles may be seen in ballooned
cells, but they should not fill the cytoplasm. Early in the disease, ballooned cells
are seen most frequently in zone 3 near the hepatic veins, but later in the disease
or in very severe cases, they may lose this clear zonal distribution. The most classic
appearing balloon cells will contain Mallory-Denk bodies (MDBs), which are eosinophilic,
ropey cytoplasmic inclusions often found near the nucleus. MDBs are composed of hyperphosphorylated,
misfolded cytokeratin 8 and 18 filaments.[11] Normally, such aggregates would be degraded in the proteosome, but in steatohepatitis
this process is blocked. MDBs stain positively using antibodies against ubiquitin
and p63, both proteins involved in degradation pathways, as well as with antibodies
against cytokeratins 8 and 18.[12] Using the latter stain, the remaining cytoplasm in the cell will be negative, whereas
nonballooned hepatocytes will remain diffusely stained.
Although the term steatohepatitis implies that steatosis and inflammation are important
parts of the injury, these components may vary considerably in degree. A biopsy that
shows characteristic ballooning injury, but has little steatosis or inflammation,
may still be confidently classified as steatohepatitis. Early in the disease, when
fibrosis is mild or absent, the steatosis typically shows a zone 3 distribution; the
inflammation is predominantly lobular and associated with the steatosis and ballooning.[7] As the disease progresses, the steatosis may lose this classical zonal distribution
and the portal inflammation may become more prominent. The lobular inflammation may
become so severe in zone 3 that the pathologist may mistake inflamed perivenular areas
for inflamed portal areas and classify the biopsy as “chronic hepatitis.” This pitfall
can be avoided by looking for the balloon cells, which are usually present in such
cases as well as the absence of the interlobular bile duct. As has been recently documented,
in severe inflammation, an arteriole may be prominent in zone 3.[13]
The pattern of fibrosis is also distinctive in steatohepatitis. Although not required
to make a diagnosis of steatohepatitis, fibrosis is usually present and its presence
can be very helpful. Fibrosis begins in zone 3 as delicate strands of collagen that
isolate one or more hepatocytes. A good quality Masson trichrome stain (or equivalent)
is critical to find the earliest evidence of fibrosis. Later, periportal fibrosis
will develop, with strands of collagen entrapping periportal hepatocytes and extending
into the surrounding parenchyma. At this stage, a component of ductular reaction may
be observed. Rarely, a case with characteristic ballooning injury will show only periportal
fibrosis—these cases should still be classified as steatohepatitis. As the disease
progresses, bridging fibrosis will form. Unlike viral hepatitis, in which the fibrotic
bridges are usually single, thick, collagen septations of collapsed parenchyma and
ductular reaction, the bridges that form first in steatohepatitis may be extensions
of the perisinusoidal fibrosis that forms networks between adjacent vascular structures.
Bridging fibrosis may be central–central, with retention of hepatocytes within the
complex network of fibrosis; may be central–portal, with eventual loss of hepatocytes,
or may be portal–portal in which parenchymal loss and ductular reaction comprise the
septal structures. Eventually, the trapped hepatocytes will be lost, and regeneration
will create solid nodules of hepatocytes, so that the end stage may resemble cirrhosis
resulting from other forms of nonbiliary liver disease. Because the fibrotic progression
in steatohepatitis involves both portal areas and hepatic veins ([Fig. 2]), fibrosis staging systems used for viral hepatitis are inadequate and fibrosis
modeling is nonlinear.
Figure 2 Possible fibrosis progression pathways in steatohepatitis. Fibrosis progression in
steatohepatitis is potentially more complicated than in the chronic viral hepatitides.
Solid arrows depict possible paths of progression, whereas hatched arrows indicate
possible paths of regression. Stages of fibrosis in the Non-alcoholic Steatohepatitis
Clinical Research Network system are shown beneath the corresponding texts.
When the histologic pattern is not distinct enough to make an unequivocal diagnosis
of steatohepatitis, but there are still suggestive changes, the term, borderline steatohepatitis
has been utilized.[14]
[15] In the NASH CRN it has been the practice of the Central Pathology Committee to use
this category for cases with characteristic fibrosis, but in which no characteristic
ballooning injury is found. There is some variation in practice on this point and
many pathologists feel comfortable making the diagnosis of steatohepatitis if characteristic
fibrosis is present in the absence of ballooning. In other cases, in which the changes
of ballooning and/or fibrosis are mild or equivocal, borderline steatohepatitis may
be a better diagnosis than either definite steatohepatitis or steatosis alone. Studies
from the NASH CRN have demonstrated that patients with biopsies that show borderline
steatohepatitis have clinical characteristics that fall between those with definite
steatohepatitis and those classified as not having steatohepatitis.[16]
Other histologic features may be seen in biopsies of patients with NAFLD, including
nonzonal small patches of hepatocytes with distinctly microvesicular steatosis, megamitochondria,
glycogenated nuclei, large lipogranulomas, and hepatocellular glycogenosis, but these
findings have little bearing on the categorization of NAFLD or the diagnosis of steatohepatitis.
The cytoplasmic changes of microvesicular steatosis and glycogenosis deserve special
mention. Microvesicular steatosis results in a distinct alteration in hepatocytes:
under low power magnification, the hepatocytes are pale, but on higher magnification,
an almost foamy appearance is noted. Although tiny vesicles may be appreciated, it
is a challenge to actually discern individual fat vesicles, and one would not be able
to count them. The nucleus is commonly central in the cell. This form of steatosis
is quite distinct from the small droplets that can be seen coalescing around the single
large vesicle in macrovesicular steatosis, or the occasional droplets within hepatocytes
that do not alter the appearance of the cytoplasm otherwise. Glycogenotic hepatocytes
may mimic ballooned hepatocytes. The glycogen lakes do not stain well with hematoxylin
or eosin and may cause the cytoplasm to have the clumped appearance of true ballooning
injury. Apoptotic hepatocytes are not included in the classification of steatohepatitis,
but have been shown in several studies and by a variety of methods to correlate with
severity of steatohepatitis.
Notably absent in noncirrhotic NAFLD biopsies are the characteristic histologic features
associated with alcoholic hepatitis. In alcoholic hepatitis, steatosis is not necessarily
present, MDB-containing hepatocytes may not be ballooned and may be surrounded by
polymorphonuclear leukocytes (satellitosis); dense perisinusoidal fibrosis and not
uncommonly, bilirubinostasis are also present. The prognostic lesion of sclerosing
hyaline necrosis of alcoholic hepatitis is not seen in NAFLD. However, in most cases
it will not be possible to distinguish alcoholic from nonalcoholic steatohepatitis
based on the liver biopsy alone.[1]
[17]
Zone 1, Borderline Pattern
In 2005, Schwimmer and colleagues published a series of cases of NAFLD in children.[18] Although some of the children had steatosis alone and some had steatohepatitis similar
to that seen in adults, some had a previously unrecognized pattern of fibrotic fatty
liver disease in which the injury seemed to involve acinar zone 1 rather than acinar
zone 3. Since that time, these observations have been extended by the NASH CRN and
others with an incidence varying from 8 to 22% in the populations studied.[19]
[20] Within the NASH CRN, the pathology committee separated out these cases as “borderline,
zone 1, steatohepatitis.”[19] The borderline term has been used to indicate evidence of a potentially fibrotically
progressive NAFLD that lacked the full spectrum of features of typical definite steatohepatitis.
This pattern is seen most often in younger, usually pre- or peripubertal children
and rarely in older teens and adults. Biopsies show steatosis that is either zone
1 predominant or panacinar; in either case, the steatosis is closely associated with
portal areas and tends to diminish around the hepatic veins ([Fig. 3]). Ballooning injury is unusual and MDB are essentially never seen in this pattern.
Portal inflammation is present, but usually mild, with little or no interface hepatitis.
Because of the lack of ballooning, the presence of fibrosis is the only feature that
sets these cases apart from the nonspecific pattern of steatosis described above.
The fibrosis is initially periportal and progresses to portal–portal bridging fibrosis,
with distinct sparing of the hepatic veins. The natural history of the zone 1, borderline
pattern is unclear due to the lack of longitudinal studies that include patients with
this pattern of injury. It should be noted that none of the existing imaging methods
that can identify steatosis or fibrosis in the liver would be able to distinguish
this zone 1, borderline pattern from classical steatohepatitis.
Figure 3 Comparison of steatohepatitis to the zone 1 borderline pattern. In both photomicrographs,
portal areas are indicated as “P” and central veins are indicated by “V.” (A) Steatohepatitis.
There is clear ballooning injury near the central vein, with inflammation and steatosis
nearby. The region near the portal area is devoid of these findings. (B) Zone 1 borderline
pattern. The portal area is expanded by fibrosis and chronic inflammation. Steatosis
completely surrounds the portal area. In contrast, the central vein region shows none
of these findings.
Cryptogenic Fibrosis and Cirrhosis
Because of the temporally variable nature of steatosis and the tendency of the characteristic
lesions of ballooning, MDB and perisinusoidal fibrosis to disappear along with acinar
architecture as the disease progresses, there will be some cases of advanced liver
disease that lack sufficient changes to be classified into one of the four groups
outlined above. Histologically, there is advanced fibrosis present, usually cirrhosis
but sometimes only bridging fibrosis, minimal steatosis and no features that would
suggest steatohepatitis, such as ballooning injury or MDBs. When this occurs because
of therapy and a preceding biopsy shows steatohepatitis, a descriptive diagnosis that
includes the phrase “consistent with resolved steatohepatitis” could be used. When
either cryptogenic fibrosis or cirrhosis is present at baseline, a diligent search
should be made to exclude other causes of advanced liver disease before ascribing
the case to previous fatty liver disease that has now disappeared, as it is well-recognized
that other forms of chronic liver disease, such as alcoholic cirrhosis and autoimmune
hepatitis, may also “burn out.”
Evaluation of Liver Biopsies in NAFLD
Evaluation of Liver Biopsies in NAFLD
Adequacy
Prior to evaluation of liver biopsies in any form of medical liver disease, the pathologist
must make a determination of adequacy of sampling. This continues to be most relevant
in NAFLD due to the fact that clinical testing can only result in exclusion of other
liver disease and a strong likelihood of NAFLD, but not a definitive diagnosis. For
adequacy, therefore, the biopsy must penetrate sufficiently below the capsule and
into hepatic parenchyma to avoid a purely subcapsular sample. Length of the sample
can be considered a surrogate parameter of adequacy; 1.5 cm is generally considered
appropriate.[21] The width of the biopsy is also important, as narrow bore biopsies may not show
complete acini and may transect portal tracts, leaving the pathologist without a firm
assessment of architectural integrity.
Just as with other forms of chronic liver disease, the process of NAFLD may not necessarily
affect the liver parenchyma uniformly.[22] However, a variety of techniques can be utilized to minimize sampling “error.” These
are particularly important in clinical studies that include serial biopsies. The biopsies
should be obtained from subjects in a similar fashion; otherwise, a certain degree
of variation may be due to the difference in technique. These include the use of large
bore needles (16 gauge or wider) and a needle biopsy length of at least 1.5 cm to
2.0 cm.[23]
[24]Recognition of the differences of anatomy between the lobes highlights the potential
for greater fibrosis in left lobe biopsies, due to larger portal structures near the
capsule and higher capsule : parenchyma ratio compared with a right lobe biopsy. Biopsies
obtained during ratio surgical procedure will result in clusters of polymorphonuclear
leukocytes (known to pathologists as surgical hepatitis) if care is not taken to perform
the biopsy as quickly as possible after induction of anesthesia.[25] This may result in inability to appropriately judge inflammation. Failure to place
the biopsy immediately in fixative may result in autolysis, which will lower estimates
of inflammation due to lysis of lymphocytes. Finally, the significance of pathologist
expertise has been documented in this disease,[26] just as in other forms of chronic liver disease.[27]
Grading and Staging
The concept of grading and staging the histologic features in NAFLD are theoretically
similar to those for chronic hepatitis: grade is a global measure of hepatocellular
and necroinflammatory injury and reflects features that are potentially reversible.
Stage is an assessment of fibrosis location (i.e., scar) and architectural remodeling;
thus, it is potentially irreversible. A subtle, but important distinction between
grade and stage, is the former describes amount, whereas the latter does not. Stage,
in all scoring systems to date, describes only parenchymal location of collagen and
matrix deposition, and vascular/architectural alterations, but not absolute quantity.[28]
[29 ]Morphometric measurements of fibrosis in liver disease require specialized techniques,
which yield important, but different information than staging. Descriptors such as
mild, moderate, and severe can be utilized for grade, whereas these should not be
applied to stage because of the inclusion of fibrosis location (zone 3 perisinusoidal,
for example), and architectural alteration(s) when present (i.e., cirrhosis).
In 1999, a classification of NAFLD was proposed[30 ]that combined the necroinflammatory lesions and fibrosis. That same year, a method
of grading and staging NASH was proposed.[31] The two studies served separate purposes; the latter was modeled upon the contemporaneous
paradigms in chronic hepatitis that separately assessed features of ongoing injury
(i.e., activity) and scarring (i.e., fibrosis; reviewed in[32]). The 1999 proposal specifically focused on lesions that differed from those of
portal-based chronic hepatitis scoring systems. The NASH CRN further developed a semiquantitative
scoring system for evaluation of serial biopsies by routine hematoxylin and eosin
and Masson trichrome stains from patients in clinical trials.[5] In this system, three features were identified in a blinded, validated study that
correlated with the separately derived diagnosis of steatohepatitis/steatosis amount,
lobular inflammation amount, and ballooning injury severity. The lesions are individually
scored and the unweighted sum is the NAFLD Activity Score (NAS). The fibrosis score,
as discussed and diagrammed in [Fig. 2], reflects the unique patterns of fibrosis that may occur in the variety of conditions
of NAFLD. [Table 3] summarizes the NAS and fibrosis scores.
Table 3
Grading and Staging
|
|
Grade
|
|
Components to score and add
|
Grade 0
|
Grade 1
|
Grade 2
|
Grade 3
|
Sum
|
|
Steatosis
|
<5% (of hepatocytes)
|
5–33%
|
34–66%
|
>67%
|
0–3
|
|
Lobular inflammation
|
0/20x
|
<2 foci/20x
|
2–4 foci/20x
|
>4 foci/20x
|
0–3
|
|
Ballooning
|
0
|
Few; inapparent
|
Easily noted; many
|
N/A
|
0–2
|
|
NAFLD activity score (NAS)
|
|
|
|
|
0–8
|
|
Stage
|
Fibrosis location
|
Architectural alterations
|
|
1A
|
Zone 3, perisinusoidal, delicate
|
No
|
|
1B
|
Zone 3, perisinusoidal, dense
|
No
|
|
1C
|
Portal, periportal only
|
May see ductular reaction
|
|
2
|
Zone 3 perisinusoidal + portal/periportal
|
May see ductular reaction; may have foci concerning for bridging
|
|
3
|
Bridging fibrosis: C-C; C-P; P-P
|
Yes, see text
|
|
4
|
Cirrhosis
|
Yes, see text
|
NAFLD, Nonalcoholic fatty liver disease; NAS, NAFLD Activity Score.
NAS vs NASH: Or…NAS: Can It Be Considered an Appropriate Replacement for the Diagnosis
of NASH?
Prior to the discovery of the hepatitis C virus and the widespread epidemic of obesity
and associated dysregulated lipids, liver test elevations could not always be explained
by serologic assays, however, there was a recognition that “chronic hepatitis” existed
and had prototypic characteristic histopathologic features regardless of specific
etiology. For example, regardless of pathogenesis, “chronic hepatitis” is characterized
by portal chronic inflammation that breeches the limiting plate (i.e., interface activity).
One of the first methods of semiquantitative evaluation was created to assess treatment
efficacy; this system was popularized as the Knodell Histologic Activity Index.[33] Scores for activity could range between 0 to 14, and those for fibrosis, 0 to 4.
Interestingly, no single HAI score was chosen as a diagnostic score of chronic hepatitis;
the diagnosis remains up to the pathologist after clinical inclusion/exclusion by
serologic testing.
When the NAS (0–8) was proposed, the study was based on 32 adult biopsies read blinded
twice by nine pathologists, and clearly documented that biopsies with higher scores
(≥ 5) tended to be diagnosed as steatohepatitis, whereas biopsies with lower scores
(≤3) were generally not considered to be diagnostic of steatohepatitis.[5] However, there were several biopsies in the NAS 3 to 5 range that overlapped all
diagnostic criteria. The relationship between the score and the diagnosis is a consequence
of how the score was created because the score incorporated major histologic features
(aside from fibrosis) that pathologists use to make the diagnosis of steatohepatitis.
The NASH CRN has published a more recent study of 934 noncirrhotic adult biopsies
scored by the Central Pathology Committee and contemporaneous demographic and laboratory
features. The findings support the concepts that the NAS and the diagnosis are interrelated,
but also do in fact have distinct clinicopathologic meanings; thus, they should not
be considered interchangeable.[34] Among cases with NAS ≥ 5, 7% were diagnosed as “not steatohepatitis” and 38% as
“borderline.” In the definite steatohepatitis category, only 75% were NAS ≥ 5, thus
the remaining 25% had scores ≤ 4. The sensitivity and specificity of NAS ≥ 5 for definite
steatohepatitis were 0.75 and 0.83, respectively. By multivariable regression, it
was shown that although both NAS and the diagnosis correlated with elevations of aspartate
aminotransferase and alanine aminotransferase, only the diagnosis of steatohepatitis
did so with metabolic syndrome, diabetes, and insulin resistance as measured by either
homeostasis model assessment–insulin resistance (HOMA-IR) or the Quantitative Insulin
Sensitivity Check Index (QUICKI). These findings have important implications for studies
that base entry criteria solely on the NAS, rather than on the pathologist's separately
derived diagnosis. Such studies risk exclusion of cases with mild but definitive findings
of steatohepatitis, and inclusion of cases with marked steatosis and inflammation
but lack ballooning.
Other Lesions of Interest in NAFLD/NASH
Other Lesions of Interest in NAFLD/NASH
Apoptosis
Apoptotic hepatocytes are readily noted in liver biopsies as rounded, eosinophilic
cytoplasmic fragments, which appear to be free within the sinusoids or surrounded
by Kupffer or other inflammatory cells. In some instances, the apoptotic hepatocytes
may contain pyknotic nuclear material; however, most frequently they do not. Regardless
of the method of ascertainment – the terminal deoxynucleotidyltransferase dUTP nick
end labeling (TUNEL) assay, serum keratin 18 fragments, or simply recording apoptotic
hepatocytes by light microscopy – the severity of apoptosis has been shown to correlate
with other markers of severity in steatohepatitis.[35]
[36]
[37]
[38]
Inflammation
Inflammatory infiltrates can be noted in the hepatic acini/lobules or the portal tracts.
Specifically, the infiltrates are predominantly of the chronic inflammatory cell phenotypes,
such as lymphocytes, monocytes, occasional eosinophils, and macrophages/Kupffer cells.
Rare polymorphs are noted; however, a lesion common to many forms of alcoholic hepatitis,
referred to as satellitosis because of polymorphs encircling MDB-containing hepatocytes,
is rare in NASH. Likewise, the portal-based lesion of ductular reaction accompanied
by fibrosis and periductular polymorphs, pericholangitis, which frequently occurs
in alcoholic hepatitis,[1] is a feature that would raise a concern of pancreatitis or even alcoholic injury
in NASH. On the other hand, lipogranulomas of varying sizes and distribution are common
to both NAFLD and alcoholic liver disease. Lipogranulomas not only consist of varying
amounts of vacuolated fat droplets, but also of admixtures of mononuclear or eosinophilic
white cells and varying degrees of collagen. Lipogranulomas near a terminal hepatic
venule may initially raise a consideration of perisinusoidal fibrosis without careful
evaluation.[6]
[39]
Inflammation in portal tracts may or may not appear to correlate with that in the
lobules. In fact, chronic inflammation that is disproportionate to the above-described
lesions of activity deserves evaluation for a concurrent form of chronic liver disease,[40]
[41]
[42] as the recognized prevalence of NAFLD[43] does not preclude that of the other liver diseases, such as hepatitis C virus (HCV),
autoimmune hepatitis, hemochromatosis, and even ALD, to name a few. There are various
proposals in the pathology literature as to how to approach a liver biopsy with a
documented form of chronic liver disease and NAFLD. Documentation of amount and location
of steatosis may or may not serve as a clue to the “origin” of the steatosis, as steatosis
in HCV gt non3 is commonly periportal, in contrast to perivenular in NAFLD. One group
has suggested that ballooning is sufficient to differentiate NASH from HCV.[44] Another group has suggested stricter criteria, i.e., the presence of zone 3 pericellular
fibrosis as the lesion of distinction of steatohepatitis from portal-based fibrosis
of other forms of chronic liver disease.[40] The former will result in higher incidence of steatohepatitis than the latter, as
hepatocellular ballooning is a lesion known to occur in viral hepatitis.[45]
Residual portal chronic inflammation was noted to be greater in relative proportion
to lobular inflammation in liver biopsies in otherwise successfully treated subjects
in a small medical treatment trial.[46] This finding, although not directly noted, can be found in other reports of surgical[47] and medical treatment trials in adults.[48] Recently, the large database of biopsies and matching demographic material from
the NASH CRN was studied for the significance of greater than mild portal chronic
inflammation in baseline biopsies of 728 adults and 205 children. In both groups,
the lesion was associated with clinical and histologic features of severity and advanced
disease.[49]
Vascular Alterations
Recent work has highlighted the microscopic observation of the intraacinar branch
of the hepatic artery in the perivenular region in active steatohepatitis ([Fig. 4]).[13] The finding correlates with severity of disease. The presence of this vessel alongside
the outflow vein may lead to confusion for a portal tract, particularly when perivenular
inflammation is marked. Thus, careful study to document the presence or absence of
an accompanying interlobular bile duct becomes crucial.
Figure 4 Perivenular fibrosis with a neo-artery. In this case of steatohepatitis, there is
a perivenular cuff of inflammation and fibrosis. A small arteriole is present (arrow).
No bile duct is seen. This appearance of the central vein may cause an observer to
mistakenly conclude that this is a portal area, leading to a diagnosis of chronic
hepatitis rather than steatohepatitis.
Implications of Histologic Evaluation for Designing Treatment Trials
This topic was the focus of a recent AASLD Endpoints Workshop that was subsequently
published.[15] The definitions as detailed above have been proposed in the article for steatosis
and steatohepatitis (definite and borderline). The use of the NASH CRN scoring system
has been discussed as a validated method of objective semiquantitative assessment
of histologic features that can be compared before and after treatment intervention.
The NAS itself has applied in a long-term outcome study of 52 Asian subjects with
repeated biopsies at 36 months. The authors found moderate correlation of change in
activity score and serum keratin 18 levels, and independent correlation with increase
in body mass index in subjects who showed progression over a 3-year period.[50] Weight reduction was the most important variable in improvement or lack of worsening
in both activity and fibrosis.
NAFLD in Lipodystrophy
Lipodystrophy encompasses a heterogeneous group of diseases with abnormal subcutaneous
adipose tissue distribution. The changes in subcutaneous fat may be localized, partial
(e.g., affecting major portions of the body), or general. Partial and general forms
of lipodystrophy may be the result of congenital genetic mutations or may be acquired,
often as a medication side effect. One of the consequences of the loss of adipose
tissue is peripheral insulin resistance, as well as decreased serum levels of adiponectin
and leptin. Fat may be stored in other organs, including the liver, skeletal muscle,
and pancreas.[51]
[52] Patients with lipodystrophy often have type 2 diabetes mellitus and hypertriglyceridemia.
Although hepatic steatosis and steatohepatitis have been recognized in patients with
lipodystrophy, most of the data are in the form of case reports and small series.
In recent a series of 10 lipodystrophic patients undergoing leptin therapy, liver
biopsies were performed before and during therapy to assess whether the systemic metabolic
improvements were having the expected effect on the liver.[53] Of the 10 patients, eight patients had steatohepatitis at baseline, one had only
mild steatosis, and one did not have NAFLD at all. Fibrosis was common, with some
degree of fibrosis in eight patients, whereas three had bridging fibrosis and one
had cirrhosis. The pattern of injury was that of the classical steatohepatitis described
above, although only three had MDBs. Treatment with leptin was associated with resolution
of histologic and biochemical features of liver disease. Other investigators have
also reported steatosis and steatohepatitis on biopsy in patients with lipodystrophy.[54]
[55] Ultrastructural studies have demonstrated alterations in mitochondria including
abnormal cristae, matrix rarefication, and crystalloids, as well as increased numbers
of peroxisomes.[56]
[57] Certain antiretroviral therapies used to treat human immunodeficiency virus (HIV)
infection have been associated with acquired forms of lipodystrophy and also with
NAFLD, and NAFLD is a common finding on liver biopsy in this population.[58] Because NAFLD and NASH are common liver diseases and HIV-infected patients may also
have other predisposing factors, a causal association between individual drugs and
liver disease is difficult to establish.
NAFLD in Bariatric Surgery Patients
Patients undergoing bariatric surgery for weight loss are at risk for NAFLD and often
have other comorbidities related to severe obesity, including diabetes mellitus, hypertension,
sleep apnea, and coronary artery disease. A high percentage will have metabolic syndrome.
Because the left lobe of the liver is accessible to the surgeon during the procedure,
biopsy surveys have provided a window into the liver disease of these patients. A
meta-analysis of such surveys that included 1620 patients from 12 studies noted a
high prevalence of steatosis and steatohepatitis (91% and 37%, respectively).[3] Although the proportion of patients with steatosis was high across all of the studies
(85 to 99%), the incidence of steatohepatitis varied from 24 to 98%. The incidence
of individual findings beyond steatosis varied considerably, with inflammation seen
in 24 to 98% and fibrosis in 13 to 97% across studies. These differences could represent
differences in operational histologic definitions, but may also represent differences
in populations undergoing bariatric surgery at the various institutions. Despite the
high prevalence of liver disease, biopsy of the liver during the procedure is not
standard practice among surgeons. Visible inspection of the liver by the surgeon is
not sufficient to exclude liver disease. In a study of 100 patients in which the external
appearance of the liver was systematically recorded by the surgeon, there was no relationship
between the gross appearance and the presence or severity of liver disease on biopsy.[59]
Several studies have reported an increased prevalence of periportal fibrosis (NASH
CRN stage 1C) in morbidly obese patients. Abrams and coworkers first brought attention
to this finding. In their cohort of 195 bariatric surgery patients, a third had stage
1C fibrosis, most with some degree of steatosis and inflammation.[60] The proportion of early stage fibrosis that is portal as opposed to perisinusoidal
is ∼50%,[61]
[62] which is a larger proportion than in nonbariatric populations undergoing biopsy
to evaluate NAFLD.[14] The clinical significance of this finding and how it relates to the other features
of NAFLD is unclear.
Summary
When liver biopsies are performed as part of clinical studies of NAFLD, attention
should be paid to the overall disease classification according to the pattern of injury
as well as to the grade and stage of the disease. Individual lesions can be evaluated
for severity and distribution within the acinar architecture of the liver. There may
be differences in the histologic features of various populations such as bariatric
surgery patients or pediatric patients as compared with adult populations derived
from hepatology clinics or diabetes clinics. It is important to note that the process
of grading and staging is related to, yet separate from the process of assigning a
diagnostic pattern. Both are important in the pathologic evaluation of NAFLD liver
biopsies.
Acknowledgment
This work was supported in part by the Intramural Program of the NIH, National Cancer
Institute.
Abbreviations
AASLD:
American Association for the Study of Liver Diseaseflow
HCV:
hepatitis C virus
HIV:
human immunodeficiency virus
HOMA-IR:
homeostatic model assessment-insulin resistance
MDB:
Mallory-Denk body
NAFLD:
nonalcoholic fatty liver disease
NASH:
nonalcoholic steatohepatitis
NASH CRN:
Non-alcoholic Steatohepatitis Clinical Research Network
