Context.—The incidence of obesity is increasing and contributes to the rising incidence of fatty liver. Body mass index (BMI) is used to assess the degree of obesity but does not take into account the pattern of body fat distribution.
Objectives.—To confirm the increasing incidence of fatty liver in an autopsy study. We hypothesized that a standardized measurement of abdominal subcutaneous adipose tissue (ASAT) might be a good noninvasive method for differentiating steatohepatitis from steatosis.
Design.—Consecutive complete adult postmortem cases were studied and liver sections were assessed with a steatohepatitis scoring system. Spleen weight, ASAT, and clinical information were obtained. Spleen histology was assessed in a subset of patients having splenomegaly in the absence of cirrhosis.
Results.—Patients with human immunodeficiency virus, hepatitis C virus, and appreciable alcohol use were excluded. Of 306 cases, the frequency of fatty liver was 51.6% with 33.3% having simple steatosis and 18.3%, having steatohepatitis. Mean ASAT was 3.7 cm in the steatohepatitis group versus 2.6 cm in the steatosis group (P < .001); this difference was greater in patients with a BMI less than 25 kg/m2 (P = .05). Fibrocongestive splenomegaly was noted in 9 of 38 patients with nonalcoholic steatohepatitis (24%) in the absence of cirrhosis.
Conclusions.—In this series of autopsy cases, a dramatic increase in the prevalence of fatty liver disease is demonstrated. Thicker ASAT is associated more with steatohepatitis than with simple steatosis, especially in patients with BMI below 25 kg/m2. Fibrocongestive splenomegaly may occur in the absence of cirrhosis in the presence of steatohepatitis.
The incidence of fatty liver disease, both nonalcoholic steatohepatitis (NASH) and simple steatosis, is increasing in the Western world.1,2 In a large autopsy study of obese patients and controls, Wanless and Lentz3 in 1990 found a 6.2% incidence of NASH. The need to distinguish between NASH and simple steatosis is essential, as patients with NASH can go on to develop portal hypertension and cirrhosis.4,5 Body mass index (BMI) is commonly used to define overweight and obesity but does not predict the presence of NASH or simple steatosis.6–9 Despite a normal BMI, patients may still develop nonalcoholic fatty liver disease (NAFLD), especially if they have a more prominent abdominal distribution of their adipose tissue.7,8
Kral et al9 showed a good correlation between waist to hip ratio and liver fat content, whereas body fat, overall weight, and BMI did not correlate with simple steatosis. Truncal fat is a good predictor of diabetes, hyperlipidemia, hypertension, and stroke.6 In 105 young Japanese patients, Nakao et al10 demonstrated that the abdominal fat index (the ratio of visceral to subcutaneous fat) was an independent risk factor for fatty liver. Other parameters that have been studied as predictors for fatty liver are waist circumference or waist to hip ratio, visceral fat distribution, and abdominal wall fat thickness as measured by ultrasonography.10,11 To date, there are little data regarding the actual measurement of abdominal subcutaneous adipose tissue (ASAT) and its relationship to NAFLD. By studying a cohort of consecutive autopsy cases, we sought to determine whether a thicker ASAT might be associated with NAFLD and whether it could differentiate NASH from simple steatosis. In this process, we noted an extremely increased incidence of NAFLD compared to that reported in other published autopsy studies, as well as the presence of splenomegaly occurring in a subset of patients with NASH not having advanced hepatic fibrosis.
MATERIALS AND METHODS
A retrospective review of consecutive complete autopsy cases of patients 18 years or older from 2001 to 2004 was undertaken. This study was approved by the Mount Sinai School of Medicine's Institutional Review Board in Manhattan, New York. These cases were identified through a Mount Sinai pathology database search. Autopsies that were limited to neuropathologic examination, chest examination or single organ, or other limited examinations, were excluded. Clinical information collected included patient demographics; human immunodeficiency virus (HIV), hepatitis C virus (HCV), and hepatitis B virus (HBV) status; history of diabetes, hypertension, hyperlipidemia, and alcohol use; and the cause of death. Patients having undergone liver transplant or previous splenectomy, and those with HIV, HCV, or significant alcohol use (more than 40 g/d and more than 20 g/d for men and women, respectively) were excluded from the final analysis. Patients were excluded if there was any mention whatsoever of alcohol use or dependency in the autopsy reports. Further review of records was undertaken to exclude the case if there was any suspicion of alcohol use. Human immunodeficiency virus cases were excluded to eliminate bias related to the confounding effects of antiretroviral therapy on fat distribution. Cases with metastatic liver disease, amyloidosis, cardiopulmonary failure, or hematologic malignancy were excluded from the analysis of liver and spleen weights.
As routine practice at our institution, the ASAT is determined in all postmortem examinations at the time of evisceration and is measured from skin to fascia. The ASAT is uniformly measured at the periumbilical region at an area where it is deemed thickest (Figure 1). All organ weights, including that of the spleen and liver, are also routinely recorded. In the current study, spleen weight above 180 g for females and 200 g for males was defined as splenomegaly.12 In an effort to better delineate the frequency of splenomegaly in noncirrhotic patients with steatosis and NASH, we arbitrarily chose 2 higher cutoff thresholds for spleen weight in both men and women (230 g and 250 g for women and men, respectively, and also 280 g and 300 g, respectively). Body mass index was calculated from the height and weight noted in the autopsy reports.
Representative sections of the spleen and of the right and left lobes of the liver are routinely obtained for histologic evaluation at autopsy. Hematoxylin-eosin– and trichrome-stained liver sections were assessed by a liver pathologist. Normal liver histology was defined as preservation of the hepatic architecture, absence of fibrosis, absence of steatosis, and no evidence of hepatitis. The study cohort was divided into 3 groups: group A consisted of cases without steatosis or NASH; group B, cases with simple steatosis; and group C, cases of NASH. Using the modified Brunt classification, the extent of steatosis was graded from 0 to 3: (0, <5%; 1, ≥5% and <33%; 2, ≥33% and ≤66%; and 3, >66%).13 Nonalcoholic steatohepatitis was graded by the presence of steatosis in addition to the following features: hepatocyte injury, such as ballooning degeneration, neutrophilic or lymphocytic infiltrates in the lobules and portal areas, and degree of fibrosis present in perivenular, perisinusoidal, and portal regions. Other significant pathologic changes in the liver were noted, including the presence of fibrosis or cirrhosis. In a subset of arbitrarily chosen cases, histologic assessment of the spleen was undertaken to examine the nature of the splenomegaly occurring in the absence of cirrhosis.
Statistical Analysis
Descriptive statistics were used and results were reported as mean with range, unless otherwise stated. Pearson χ2 analysis with cross tabs, was used to analyze variables in different stages. Likelihood ratio statistics were also used to determine differences in some cases. Continuous variables were compared by using the Mann-Whitney test. All statistical analysis was performed with the statistical software package SPSS for Windows (version 15.0, SPSS Inc, Chicago, Illinois). P value was set at .05 for statistical significance.
RESULTS
The study cohort consisted of 306 consecutive adult complete autopsy cases. These cases were selected from a total number of 737 autopsies performed during a period of 4 years (2001–2004). Cases excluded from the study included pediatric cases (n = 126), limited autopsies (n = 47), autolyzed liver tissue, significant or suspected alcohol use or incomplete clinical data (n = 186), HIV-positive cases (n = 72), and HCV-positive cases (n = 21). There were 148 males and 158 females (48.4%/51.6%) with median age of 71 years (range, 18–98 years). Table 1 shows the characteristics of all patient groups. There were 148 cases without histologic evidence of steatosis (group A). The overall prevalence of fatty liver disease, either simple steatosis (group B) or steatohepatitis (group C) was 51.6% (158 of 306). Steatosis ranging from mild to severe was present in 102 of 306 cases (33.3%) and steatohepatitis, in 56 of 306 cases (18.3%).
The mean ASAT was 3.7 cm in group C versus 2.6 cm in group B (P < .001). As compared with group B, a greater proportion of patients in group C had higher BMI (79.7% versus 20.3%; P = .003) and higher ASAT (P = .003). Although not statistically significant, patients with steatohepatitis died at a younger age than patients with simple steatosis (66 versus 70 years, P = .06). Compared to group C, there were more patients in group B with a BMI below 25 kg/m2 (51.6% versus 25%) than above 25 kg/m2 (48.4% versus 75%; P = .003). The difference in ASAT was greater between groups B and C in individuals with BMI below 25 kg/m2 (P = .05) as opposed to those with BMI above 25 kg/m2 (P = .59). There was a positive correlation between ASAT and BMI (R = .7, P < .001). Cases with BMI above 25 kg/m2 had a higher incidence of hepatomegaly (83.1% versus 54.7%; P < .001) and a higher incidence of diabetes (P = .03), but not a higher incidence of splenomegaly. Spleen weight was not increased in group C as compared to that in group B (mean = 209 g versus 171 g; P = .13). No infiltrative or neoplastic involvement of the spleen was noted in the 306 cases. Of the individual morphologic features of NASH, pericellular fibrosis and portal fibrosis were the only ones associated with splenomegaly, even when using higher thresholds for the definition of splenomegaly (P < .001, P = .001, and P = .001, respectively; and P <.001, P = .02, and P = .01, respectively). The presence of ballooned hepatocytes, Mallory hyaline, and lobular inflammation was not predictive of splenomegaly at all 3 thresholds.
In groups B and C, racial breakdown was similar: white, 50% versus 48%; African American, 23% versus 21%; Hispanic, 27% versus 31%; and Asian, 0% versus 0%, respectively (P = .14). There was no significant difference between the groups with regard to cause of death (Table 1).
Table 2 shows the fibrosis stage for patients having steatohepatitis and splenomegaly when using different cutoffs for normal spleen weight. In group B cases, no one had splenomegaly ascribable solely to simple steatosis; it was present in the setting of concurrent cirrhosis, extensive malignancy, or extrahepatic portal vein thrombosis. By using the cutoff for normal spleen weight (180 g and 200 g, for women and men, respectively), as expected, 13 of 17 steatohepatitis cases with cirrhosis had splenomegaly but, interestingly, 9 of 38 cases (24%) with stages 1 through 3 fibrosis also had splenomegaly, including 5 of 20 patients with stage 1 disease. None of these patients had other obvious conditions by history or histology that could account for the splenomegaly. None of the patients in either group had histologic evidence of portal venopathy such as hepatoportal sclerosis or nodular regenerative hyperplasia.4,14 Histologic analysis of the spleens uniformly revealed fibrocongestive splenomegaly without obvious differences in white pulp depletion, extent of sinusoidal congestion, or fibrosis. An example of this is shown in Figure 2. Figure 3 shows the average spleen weights as well as the range and standard deviations for the patients with steatohepatitis in the different stages of fibrosis. Mean spleen weight increases with increasing fibrosis stage (P = .007), even between stages 1 and 3.
COMMENT
Obesity may be a risk factor but not a prerequisite for NAFLD.2,10,11,15–17 Many patients with fatty liver are asymptomatic; a few may have complaints of fatigue and right upper quadrant discomfort.4,16 Physical examination in such patients is also usually unremarkable except for a palpable liver and, more infrequently, a palpable spleen.16 In a study of Italian patients with NASH,4 portal hypertension appeared to only occur in patients having established cirrhosis.
Several physical methods to assess body fat are available. The most commonly used method is BMI.7,9,11 Other methods include anthropometric measurements such as waist circumference or waist to hip ratio, visceral to subcutaneous adipose tissue ratio, visceral fat distribution, and abdominal wall fat thickness as measured by ultrasonography.7–9 According to some studies, waist circumference, visceral fat content, and abdominal sagittal diameter can predict the presence of NASH, which frequently occurs concurrently with the metabolic syndrome.10,18 Nakao et al10 previously proposed that abdominal wall fat index was an independent risk factor for the existence of fatty liver. Results of our current study confirm this and also demonstrate that increased ASAT may suggest the presence of NASH. To date, no single clinical parameter or imaging modality can readily discern steatosis from NASH.19
Obesity, along with diabetes and hyperlipidemia, increases the risk for developing NAFLD. This can histologically range from simple steatosis to steatohepatitis that may progress to cirrhosis. It is possible that people who have increased ASAT thickness also have a normal BMI, but as shown in this study, thicker subcutaneous fat tissue may suggest the presence of NASH even with a normal BMI. Studies that have looked at BMI as a risk factor for NASH may not have included a certain subset of patients with normal BMI who had thickened subcutaneous fat tissue. Yano et al20 analyzed the fat thickness patterns of abdominal sites that are used routinely for a transrectus abdominis musculocutaneous flap for breast reconstruction by using ultrasonography in 50 patients. They found that the average subcutaneous fat thickness over the abdomen was 24.0 ± 9.4 mm and that this thickness was highest at the center of the abdomen and lowest at the lateral areas. They noted that the average subcutaneous fat in the abdomen correlated with BMI. The mean ASAT in groups A and B was 2.2 and 2.6 cm, respectively, in the current study, which is similar to the results noted by Yano et al.20 Abdominal subcutaneous adipose tissue has been found to be statistically greater in patients with NASH (mean, 3.7 cm); a good correlation between BMI and ASAT was also noted in the current study.
Kral et al9 demonstrated that abdominal distribution of fat was a predictor of hepatic steatosis, independent of body weight or body fat. In a recent study, Jacobs et al7 showed that increased waist circumference is a risk factor for mortality regardless of BMI. Adams et al15 have shown a higher mortality among community-diagnosed patients with NAFLD than the general population. In the current study, we did not find that patients having steatohepatitis and thicker ASAT died at a younger mean age than patients with simple steatosis; however, there was a statistical trend toward this age effect. One might have expected that patients with NAFLD would have a higher incidence of cardiovascular deaths, but this was not shown in the current study.
In a previous Canadian autopsy study of obese individuals, Wanless and Lentz3 found a 6.2% incidence of NASH. In the current study of 306 consecutive autopsies performed at an urban tertiary medical center in New York City, NAFLD was found in 51.6% of cases, of which 18.3% showed NASH. Our results show a dramatic increase in NAFLD as compared to the Wanless and Lentz study,3 which is consistent with the clinical evidence from large epidemiologic studies, such as the National Health and Nutrition Examination Survey, which confirmed the increase in NAFLD.21,22 Most concerning in our cohort is the significant increase in NASH cases even for patients without increased BMI.
In the current study, we also determined that the presence of splenomegaly correlated better with NASH than with simple steatosis, with spleen weights increasing with greater degrees of hepatic fibrosis. This finding explains why portal hypertension may be present in certain patients with NASH, even in the absence of established cirrhosis. Although mean spleen weights were similar between groups B and C, this was accounted for by other reasons aside from the primary liver disease. No cases of bland steatosis alone were associated with the presence of splenomegaly. Histologic review of the enlarged spleen in cases of NASH demonstrated the typical histologic appearance of fibrocongestive splenomegaly. The typical histologic features of NASH—pericellular fibrosis and portal fibrosis—can explain this, as these features are lacking in patients with simple steatosis. Patients with alcoholic hepatitis23 and sinusoidal obstruction syndrome24 occurring in the setting of severe cellular rejection post liver transplant can have similar lesions and can also have noncirrhotic portal hypertension.
Strengths of the current study include a well-characterized, large consecutive autopsy study group with detailed histologic review of the liver as well as documented measurement of ASAT. Portal hypertension in the NASH cases was not due to portal venopathy or extrahepatic vascular thrombosis and appears clearly to be ascribable to the steatohepatitis. An inherent weakness in our study includes the fact that some patients with malignant neoplasms may have been receiving medications that could have induced steatohepatitis or steatosis and thus affected the histologic findings. In addition, BMI assessment of a hospitalized patient may be somewhat inaccurate, especially in the preterminal period. However, this limitation is present in all autopsy studies including the Wanless and Lentz study,3 which showed a much lower incidence of NAFLD than did the current study. Although the clinical history in the autopsy records was detailed, it is possible that some patients with undocumented alcohol use may have been included in the steatosis or steatohepatitis groups. It would be expected, however, that these patients would be equally distributed among groups A through C. Finally, autopsies from 2001–2004 were reviewed, and it is possible that the incidence of NAFLD is currently even higher.
In summary, this current US urban population autopsy study confirms a dramatic increase in the frequency of both simple steatosis and steatohepatitis. Thicker abdominal subcutaneous adipose tissue is associated with steatohepatitis, but not simple steatosis, also for patients with BMI below 25 kg/m2. There is good correlation between abdominal subcutaneous adipose tissue thickness and BMI. A trend noted in this study is that patients with NASH died at a younger age. Fibrocongestive splenomegaly, despite the absence of cirrhosis, may occur in some patients with steatohepatitis.
References
Author notes
The authors have no relevant financial interest in the products or companies described in this article.