SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

Background

Not all NAFLD patients are obese and many obese patients do not have NAFLD. Impaired peripheral fat storage may increase the delivery of lipids to the liver and facilitate NAFLD progression.

Aim

To assess the association of anthropometric measures of regional adiposity including arm fat index (AFI) (upper body fat), waist circumference (visceral fat) and body mass index (total body fat) on liver injury and fibrosis in NAFLD.

Methods

One hundred and forty-one patients with histological evidence of NAFLD were included in this study. Multivariate logistic regression models examined the contribution of age, sex, body mass index, AFI, triceps fold thickness (TST), waist and hip circumference to the odds of liver injury (NAS scores ≥3) and fibrosis (fibrosis scores ≥2) by liver biopsy.

Results

Arm fat index (OR: 0.82, 95% CI: 0.59–0.91) and TST (OR: 0.13, 95% CI: 0.04–0.42) were negatively correlated with NAFLD histological severity. In women, waist circumference was positively correlated with NAFLD severity (OR: 1.21(1.02–1.44). Age (OR: 1.05, 95% CI: 1.01–1.0) and waist circumference (OR: 1.07, 95% CI: 1.00–1.15) were significantly associated with fibrosis risk. In women, AFI (OR: 0.87, 95% CI: 0.76–0.99) and TST (OR: 0.22, 95% CI: 0.05–0.95) were negatively associated with fibrosis risk.

Conclusions

Regional anthropometric measures are associated with severity of NAFLD in a sex-specific manner. Men and women with lower arm fat depots and women with bigger waist circumference have a greater likelihood of liver injury. Age and waist circumference seem to be associated with liver fibrosis. Simple anthropometric measurements of peripheral fat deposits may help stratify significant liver injury risk.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

The rising incidence of non-alcoholic fatty liver disease (NAFLD) in both adults and children is believed to be due to the increased prevalence of obesity, type 2 diabetes and metabolic diseases, and is now considered the most frequent cause of elevated liver enzymes.[1] Non-alcoholic steatohepatitis (NASH) is considered the more severe spectrum of NAFLD and can progress on to cirrhosis and hepatocellular carcinoma.

Obesity is a risk factor for NAFLD, although there is overwhelming evidence that it is not so much how much adipose tissue one has, as where one has it. Lifestyle-induced weight loss results in improved cardio-metabolic and NASH status, whereas subcutaneous liposuction results in no such changes.[2, 3] It appears that it is visceral as opposed to subcutaneous or peripheral fat that is hepatotoxic.[4] Therefore, the ‘adequacy’ of peripheral adipose tissues and factors that determine whether lipids are stored preferentially in peripheries or viscerally is currently an area of intense interest. Indeed, it has recently been suggested that stimulating adipocyte differentiation to increase peripheral adipose tissue storage capacity may be a therapeutic option for obesity-related disorders.[5] Prior work by our group has shown the hepatic benefits of pioglitazone, which among other actions redistributes fat from central to peripheral stores.[6, 7]

Anthropometric assessments of lipid distribution are widely used to assess risk for a variety of cardiometabolic disorders. Waist circumference predicts cardiovascular risk[8] and has been shown to be more predictive of incident type 2 diabetes than waist-to-hip ratio[9] or BMI.[10] The role of peripheral fat deposition, however, has been less well studied. A large hip circumference has been shown to be an independent predictor of lower cardiovascular and diabetes-related mortality. Several studies have demonstrated that hip circumference and leg are negatively correlated with atherogenic lipid and glucose metabolites.[11]

Peripheral versus central lipid distribution patterns appear to be under hormonal control. The menopause is characterised by a central distribution, which is reversed by hormone replacement therapy,[12] while androgens appear to worsen adipose endocrine function by reducing circulating levels of adiponectin.[13] Converse to obesity, the lipodystrophies are a spectrum of disorders characterised by reduced or absent peripheral adipose tissue. Lipids are therefore ectopically stored in muscle and liver resulting in insulin resistance and NAFLD.[14] Therapeutic response is greatest in those with only partial lipodystrophy, presumably as they can repartition fat from ectopic to peripheral stores.[15] This reinforces the importance of adequate peripheral adipose stores.

In this study, we aimed to assess if upper arm adipose tissue deposition predicted histological severity of NAFLD. Assessments of upper limb adiposity were made as this is the most routinely and reliably assessed peripheral adipose store.[16, 17] The specific objectives of the study were to determine if upper arm fat deposit or central abdominal fat is correlated with severity of liver injury and liver fibrosis in patients with NAFLD.

Materials And Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

Study design and patients

We performed a hypothesis-driven cross-sectional analysis using data from well-characterised cohort of patients with NAFLD. This prospective cohort from Nottingham University Hospital has been the source of recruitment for previous studies.[6, 18] Patients were enrolled between November 2002 and September 2008. The inclusion criteria were as follows: (i) Age >18 years, (ii) Availability of liver biopsy slides, (iii) No significant alcohol consumption (defined as >14 units of alcohol per week in women and >21 units of alcohol per week in men on average over the past 2 years), (iv) Exclusion of other liver diseases made by detailed investigations including hepatitis B and C serology, iron studies (ferritin, transferrin saturation), α-1 antitrypsin, ceruloplasmin, auto-antibody profiles and immunoglobulins, (v) Availability of anthropometric data collected within 3 months of liver biopsy. The studies were approved by the Nottingham joint ethics committee.

Liver histology

Steatosis, inflammation (portal and lobular), hepatocyte ballooning and fibrosis were scored using the NASH Clinical Research Network criteria[19] by an experienced liver pathologist (P. K.) who was blinded to the clinical data of the patients. Features of steatosis, lobular inflammation and hepatocyte ballooning were combined to obtain the NAFLD activity score (NAS). A NAS of ≥3 and a fibrosis score of ≥2 were the outcome variables of interest.

Anthropometry

Height, weight, triceps skinfold thickness, mid-arm circumference, waist and hip circumference were measured within 3 months of liver biopsy. Body weight was measured to the nearest 0.1 kg on a clinical scale, and body height was measured to the nearest centimetre, from subjects without shoes. The midpoint of the outer extremity of the scapula to the olecranon process of the nondominant arm at a 90° angle was used as the mid-arm reference point. The mid-upper arm circumference (MUAC) was measured with a non-extensible, flexible tape to the nearest 0.1 cm. Triceps skinfold thickness (TST) was measured at the same arm with a skinfold caliper (Holtain LTD, Crymych, Wales, UK), with a constant pressure of 10 g/mm2 to the nearest 0.2 mm. Waist circumference was measured halfway between the lower border of the ribs and the iliac crest in a horizontal plane. Hip circumference was measured at the widest point over the buttocks. All anthropometric measurements were done in duplicate by the same trained nurse and the mean value documented. From these measurements, the BMI (weight in kilograms/height in metres2) was calculated.

The total upper arm area (TUA), upper arm muscle area (UMA), upper arm fat area (UFA) and arm fat index (AFI) were derived from the MUAC and TST using standard formulae.[20, 21] Using measurements made in centimetres, the following formulae were used: TUA = MUAC2/(4 × π); UMA = [MUAC − (TST × π)]2/(4 × π). To adjust for the area of the bone and to obtain the bone-free arm muscle area corrected for gender, the values were adjusted by subtracting 10.0 cm2 for males and 6.5 cm2 for females. UFA = TUA − UMA and AFI = (UFA/TUA) × 100.

Laboratory tests

Laboratory investigations performed in the clinical laboratory of Nottingham University Hospital with standardised methodology included liver biochemistry (alanine aminotransferase, bilirubin, alkaline phosphatase, gamma-glutamyl transferase), lipid profile, fasting glucose and Hepatitis B and C serology, iron studies (ferritin, transferrin saturation), alpha-1 antitrypsin, ceruloplasmin, auto-antibody profiles and immunoglobulins were done to rule out other causes of liver disease. The homeostasis model assessment for insulin resistance (HOMA-IR = insulin × glucose/22.5) was calculated when data were available for fasting glucose and insulin.

Statistical tests

Results are expressed as means ± standard deviations for continuous variables and as frequencies for categorical variables. P-values from anova or chi-squared tests were considered statistically significant if ≤0.05. Logistic regression analysis was performed to calculate odds ratios (OR) and their 95% confidence intervals. Age, sex, BMI, TST, waist circumference, hip circumference and AFI were included in the final multivariate model. Correlation matrices were used to identity collinearity. When collinearity was detected (rho >0.6), this was minimised by inputting the variable separately in the multivariate analysis. All statistical tests were done using pasw version 17 (IBM Corp, NY, USA). Correlation matrices were used to identity collinearity.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

Of 153 patients enrolled in the two studies, 3 had missing anthropometric data, 5 did not have liver biopsy slides for blinded re-evaluation and 5 were found to have steatosis scores of 0 on blinded re-evaluation and were all excluded from the final analysis. There were 141 patients (88 male and 53 female) with a mean age of 51.6 ± 11 years included in the final analysis. Table 1 gives the demographic and laboratory profile of the included patients distributed according to NAS scores and fibrosis scores. Significant (P < 0.05) differences were noted between the group of patients with NAS <3 and those with NAS scores ≥3 for triceps fold thickness and arm fat index, alanine aminotransferase, fasting glucose and HOMA-IR. Significant (P < 0.05) differences were noted between the group of patients with fibrosis scores <2 and those with fibrosis scores ≥2 for age, body mass index, waist circumference, total cholesterol, low density lipoprotein levels, fasting glucose and HOMA-IR. Table 2 gives the distribution of liver histology scores in the selected groups based on NAS scores and fibrosis scores. Fibrosis grades 1a, 1b and 1 c were clubbed together under grade1 for the purposes of this study (Figure 1).

Table 1. Demographic, anthropometric and laboratory profiles of the included patients according to NAS and fibrosis scores
VariablesNAS score <3 (n = 23)NAS score ≥3 (n = 118)P-valueFibrosis score <2 (n = 67)Fibrosis score ≥2 (n = 74)P-value
  1. All values reported as mean ± standard deviation.

  2. Bold values indicate statistically significant results.

Age (years)54.1 ± 9.851.1 ± 5.20.24848.2 ± 10.554.6 ± 10.8 0.001
Gender (F/M)10/1343/750.52420/4733/410.073
Body mass index28.8 ± 4.830.9 ± 5.20.07129.6 ± 4.931.5 ± 5.3 0.031
Waist circumference (cm)100.9 ± 12.5108.1 ± 14.50.028103.6 ± 12.1109.7 ± 15.7 0.018
Hip circumference (cm)110.3 ± 11.7112.6 ± 11.20.379110.6 ± 10.7113.7 ± 11.60.109
Triceps skinfold thickness (cm)2.8 ± 0.52.4 ± 0.6 0.010 2.5 ± 0.72.5 ± 0.60.959
Arm fat index (%)47 ± 640 ± 7 0.001 41.5 ± 9.141.1 ± 6.40.748
Alanine aminotransferase (U/L)53 ± 2183 ± 45 0.001 73.5 ± 36.136.1 ± 50.30.226
Bilirubin (μmol/L)10.6 ± 4.212.6 ± 6.00.06812.1 ± 6.312.4 ± 5.30.702
Gamma glutamyl-transpeptidase (U/L)118 ± 61128 ± 1220.542112 ± 14082 ± 1320.142
Albumin (g/L)43.2 + 3.043.7 ± 3.10.39643.9 ± 3.243.4 ± 2.90.362
Cholesterol (mmol/L)5.6 ± 1.15.6 ± 1.20.9125.6 ± 1.66.3 ± 2.3 0.016
Low density lipoprotein (mmol/L)3.3 ± 1.03.4 ± 1.00.8763.7 ± 0.93.1 ± 1.0 0.006
High density lipoprotein (mmol/L)1.3 ± 0.41.3 ± 0.40.8881.4 ± 0.41.3 ± 0.30.322
Triglycerides (mmol/L)2.2 ± 1.52.1 ± 1.50.7812.1 ± 1.22.2 ± 1.70.574
Fasting glucose (mmol/L)5.1 ± 1.26.2 ± 1.2 0.004 5.6 ± 1.65.4 ± 1.3 0.048
HOMA-IR (data on 111 patients only)2.8 ± 1.55.4 ± 4.0 0.001 3.1 ± 2.46.3 ± 4.5 0.001
Table 2. Distribution of steatosis, lobular inflammation, hepatocyte ballooning and fibrosis scores in the selected groups
VariableNAS score <3 (n = 23)NAS score ≥3 (n = 118)Fibrosis score <2 (n = 67)Fibrosis score ≥2 (n = 74)
  1. a

     Fibrosis grades 1b, 1b and 1c have been combined into grade 1.

Steatosis
121323320
2 2492031
3 0371423
Lobular inflammation
019 520 4
1 4844543
2 024 222
3 0 5 0 5
Hepatocyte ballooning
013 716 4
110584424
2 053 746
Fibrosisa
0142438 0
1 42529 0
2 440 044
3 019 019
4 110 011
image

Figure 1. (a) Distribution of NAS (non-alcoholic fatty liver disease activity scores) in the study population. (b) Distribution of fibrosis scores in the study population.

Download figure to PowerPoint

Table 3 provides details of the multivariate logistic regression analysis comparing patients with NAS scores <3 and those with NAS scores ≥3 taking into account age, gender, body mass index, waist circumference, hip circumference, triceps skinfold thickness and arm fat index, for all patients, male and female patients respectively. Waist circumference, triceps skinfold thickness and arm fat index were significant on univariate analysis, but on multivariate analysis adjusting for the other parameters of a priori interest, only triceps fold thickness and arm fat index remained significant. Triceps fold thickness and arm fat index were negatively associated with severity of NAFLD in both men and women, whereas waist circumference was positively associated with severity of NAFLD in women alone. Data on HOMA-IR were only available in 111 patients and thus it was left out of the primary logistic regression analysis. When the analysis was performed on the 111 patients on whom HOMA-IR was available, the only significant variables associated with histological severity of NAFLD were arm fat index (OR: 0.77 and 95% CI: 0.66–0.89, P = 0.001), triceps fold thickness (OR: 0.10, 95% CI: 0.03–0.40, P = 0.018) and HOMA-IR (OR: 1.48 and 95% CI: 1.01–2.19, P = 0.046). Hosmer-Lemeshow's test was used to test the null hypothesis that there is a linear relationship between predictor variable and the log odds of the outcome variable. The model fitted well according to the Hosmer-Lemeshow test. As there was significant correlation between AFI and TST (rho = 0.89) and WC and gender (rho = 0.74), these were inputted separately into the multivariate analysis (Figure 2).

Table 3. Association of measured and derived anthropometric measures, age and gender with histological severity of non-alcoholic liver disease (comparing NAS score of <3 with NAS score ≥3)
VariableUnivariate analysisMultivariate logistic regression analysis (all patients) (N = 141)Multivariate logistic regression analysis (male patients) (N = 88)Multivariate logistic regression analysis (female patients) (N = 53)
OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P
  1. BMI, body mass index; WC, waist circumference; HC, hip circumference; TST, triceps skinfold thickness; AFI, arm fat index; n/a, not applicable.

  2. Bold values indicate statistically significant results.

  3. a

     Multivariate model used included: Age, Gender, BMI, HC and AFI.

  4. b

     Multivariate model used included: Age, BMI, WC, HC and AFI.

  5. c

     Multivariate model used included: Age, Gender, BMI, HC and TST.

Agea0.98 (0.94–1.02)0.2480.95 (0.90–1.10)0.0630.94 (0.87–1.01)0.0791.04 (0.92–1.17)0.545
Gendera1.34 (0.54–3.32)0.5240.62 (0.21–1.88)0.398n/an/an/an/a
BMIa1.1 (0.99–1.22)0.0710.98 (0.78–1.23)0.8451.07 (0.68–1.35)0.6990.81 (0.56–1.74)0.264
WCb1.05 (1.01–1.1)0.0281.05 (0.97–1.13)0.211.09 (0.88–1.35)0.4141.21 (1.021.44) 0.031
HCa1.02 (0.98–1.06)0.3790.94 (0.86–1.03)0.2010.96 (0.81–1.13)0.6300.97 (0.87–1.08)0.581
TSTc0.35 (0.16–0.8)0.0100.13 (0.040.42) 0.006 0.11 (0.020.53) 0.006 0.13 (0.0020.91) 0.039
AFI0.85 (0.77–0.93)0.0010.82 (0.590.91) 0.001 0.81 (0.700.93) 0.003 0.82 (0.690.99) 0.045
image

Figure 2. (a) Scatter plot showing the distribution of the am fat index (AFI) values for men and women. (b) Box plot showing the mean (s.d.) and 95% CI of the arm fat index (AFI) in those with non-alcoholic fatty liver disease activity scores (NAS) of <3 and ≥3.

Download figure to PowerPoint

Table 4 provides details of the multivariate logistic regression analysis comparing patients with fibrosis scores <2 and those with fibrosis scores ≥2 taking into account age, gender, body mass index, waist circumference, hip circumference, triceps skinfold thickness and arm fat index, for all patients, male and female patients respectively. Age, body mass index and waist circumference were significantly associated with fibrosis on liver histology on univariate analysis, but on multivariate analysis adjusting for the other parameters of a priori interest, age and waist circumference were positively associated with increased fibrosis score. When analysed separately for male and female gender, age remained significant only for males and not in females and waist circumference almost reached statistical significance in females (P = 0.083). Both triceps fold thickness and arm fat index were negatively associated with fibrosis in females, but not in males. When the analysis was performed on the 111 patients on whom HOMA-IR was available, the only significant variables associated with histological severity of fibrosis were age (OR: 1.06, 95% CI: 1.01–1.11, P = 0.018) and HOMA-IR (OR: 1.22, 95% CI: 1.05–1.41, P = 0.006). Hosmer-Lemeshow's test was used to test the null hypothesis that there is a linear relationship between predictor variable and the log odds of the outcome variable. The model fitted well according to the Hosmer-Lemeshow test. As there was significant correlation between AFI and TST (rho = 0.94), these were inputted separately into the multivariate analysis (Figure 3).

Table 4. Association of measured and derived anthropometric measures, age and gender with histological severity of fibrosis in non-alcoholic liver disease (comparing fibrosis score of <2 with fibrosis score ≥2)
VariableUnivariate analysisMultivariate logistic regression analysis (all patients) (N = 141)Multivariate logistic regression analysis (male patients) (N = 88)Multivariate logistic regression analysis (female patients) (N = 53)
OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P
  1. BMI, body mass index; WC, waist circumference; HC, hip circumference; TST, triceps skinfold thickness; AFI, arm fat index; n/a, not applicable.

  2. Bold values indicate statistically significant results.

  3. a

     Multivariate model used included: Age, Gender, BMI, WC, HC and AFI.

  4. b

     Multivariate model used included: Age, Gender, BMI, WC, HC and TST.

Agea1.06 (1.02–1.09)0.0011.05 (1.011.09) 0.008 1.06 (1.011.11) 0.011 1.07 (0.99–1.17)0.103
Gendera1.89 (0.94–3.79)0.0732.51 (0.90–6.99)0.079n/an/an/an/a
BMIa1.08 (1.01–1.16)0.0311.05 (0.89–1.25)0.7401.14 (0.92–1.40)0.2390.86 (0.68–1.09)0.211
WCa1.03 (1.01–1.06)0.0181.07 (1.001.15) 0.048 1.05 (0.96–1.15)0.2771.10 (0.99–1.22)0.083
HCa1.03 (0.99–1.06)0.1090.95 (0.89–1.02)0.1380.93 (0.84–1.03)0.1430.99 (0.92–1.08)0.886
TSTb1.01 (0.59–1.74)0.9590.69 (0.35–1.38)0.2971.05 (0.96–1.15)0.9000.22 (0.050.95) 0.043
AFIa0.99 (0.95–1.04)0.7480.98 (0.93–1.03)0.3781.0 (0.95–1.06)0.9170.87 (0.760.99) 0.038
image

Figure 3. (a) Scatter plot showing the distribution of the am fat index (AFI) values for men and women. (b) Box plot showing the mean (s.d.) and 95% CI of the arm fat index (AFI) in those with non-alcoholic fatty liver disease activity scores (NAS) of <3 and ≥3.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

The population prevalence of NASH is greater in men, and this has been attributed to their greater visceral adipose storage tendency. In line with this, we also report a greater disease severity in men. The novel findings of this study are that histological features such as NAS was positively associated with waist circumference in women and negatively associated with arm fat index in both sexes. It therefore appears that peripheral adipose distribution reduces the possibility of NASH, whereas central distribution increases the likelihood in women. This is an intriguing finding. The reversal of gender-typical adipose tissue distribution patterns, with a peripheral distribution in men and a central (android) distribution in women, reversed the gender-ascribed NASH risk.

Clinical factors known to be independently predictive of fibrosis in NASH include age, obesity and diabetes.[22] In line with this, we have also found age to be associated with fibrosis. It has previously been shown in both children and adults that waist circumference is a stronger predictor for fibrosis than BMI.[23, 24] A BMI greater than 30 kg/m2 lacks sensitivity for total body adiposity,[25] and the BMI value has only a weak association with visceral and hepatic fat.[26, 27] Waist circumference is a better predictor of visceral fat than BMI, and a 1% increase in visceral fat increases fibrosis risk by more than three-fold.[27] This study further explores the differences between whole-body adiposity, as crudely assessed by BMI, and regional adiposity on risk of liver injury and fibrosis.

The current findings that adipose tissue distribution impacts on hepatic health are supported from prior work. A Korean group assessed the role of thigh and visceral fat on NAFLD as diagnosed by ultrasound.[28] They recruited from a mixed obesity and gastroenterology clinic setting. Increasing visceral fat on CT was associated with a greater likelihood of steatosis in both sexes. Conversely in women, a smaller thigh subcutaneous fat content increased the risk of steatosis. Data from the NASH Clinical Research Network database have shown similar findings. Android adipose distribution, as characterised by a greater waist circumference, conferred an increased likelihood of fibrosis in postmenopausal women, but not premenopausal women or men.[29] It was also reported that a greater peripheral adipose storage was associated with a reduced fibrotic likelihood in men and postmenopausal women. The magnitude of these observed effects was similar to a 10-year age gain. A large population-based study inferred an association between adipose tissue distribution and NAFLD and injury, although it lacked hepatic histological or radiological data.[30] The risk of an elevated serum alanine aminotransferase (ALT) was assessed in 11 821 male and female adults without viral hepatitis assessed as part of the US National Health and Nutrition Survey (NHANES). Trunk fat (quantified by DEXA) increased and extremity fat decreased the risk of an elevated ALT.

Other studies have failed to assess for the influence of sex on lipid distribution patterns. The visceral adiposity index (VAI), a composite score of waist circumference, triglycerides and HDL, has been shown by some groups to be independently predictive of liver fibrosis.[31] In morbidly obese adults, liver enzymes were negatively correlated with whole-leg fat mass, and positively correlated with trunk fat mass.[32] Cheung et al. made several observations.[33] First, they reported that waist circumference was correlated with lobular inflammation only. Secondly, upper limb adiposity (as measured by triceps and biceps skinfold thickness) was positively associated with histological features of liver injury, but not fibrosis. Finally, dorso-cervical lipohypertrophy (DCL), a fat pad found at the neck base similar to that in Cushing's syndrome or lipodystrophy, was associated with both injury and fibrosis. This is a curious finding, although the authors note that there is no objective method for quantifying DCL, and so they merely classified it as being present or absent. There are also no data on the inter-observer assessment of DCL and so it's utility in clinical assessment remains uncertain.

There are many techniques employable to determine regional adipose tissue distribution. The major advantages of waist circumferences and arm fat indexes are their speed and simplicity. Such data can be readily available during an initial clinical consultation. Controversies, however, do remain as to how to assess and interpret such measurements. Concerns arise over intra- and inter-observer variability and there is debate over the optimal site to measure waist circumference. Data to support such concerns, however, are scant. The absolute and inter-observer variability is small at less than 3%,[34, 35] and the prevalence of the metabolic syndrome is only modestly altered from measuring waist circumference at differing sites.[36] The clinical impact of such small degrees of variability is therefore likely to be small. Of greater concern is as how to interpret the measurements. Despite there being a continuous and linear association between waist circumference and cardiometabolic risk,[37] cut-off values are used to categorise risk. Although this clearly increases clinical simplicity, it may weaken some of its predictive power. The next issue is that although waist circumference correlates well with total abdominal fat, it fails to determine whether this is visceral, retroperitoneal or subcutaneous depots. Furthermore, fat deposition patterns are dependent on ethnicity as well as sex. Asians tend to have a smaller lean body mass than Caucasians and so have lower cut-off values for BMI and waist and have a greater rate of non-obese NAFLD.[38]

We acknowledge some limitations of this study. Regional adipose distribution patterns were assessed clinically (using anthropometry) rather than determined using measures like CT, MRI or impedance techniques. This approach aids the clinical reproducibility of our findings, although reduces the robustness of the findings. Data on insulin resistance were not available in all and hence it was left out of the primary logistic regression analysis. Finally, exercise, dietary and hormonal data may have facilitated a more in-depth analysis of the findings.

In conclusion, there is increasing evidence to suggest that regional fat distribution has a direct effect on hepatic fatty acid metabolism. Our data suggest that the central android distribution increases the likelihood of NASH in women and the peripheral distribution reduces the likelihood in men and women. It remains unclear as to why differing adipose distribution patterns develop, and greater understanding of this may aid the development of further treatment strategies. We envisage that adipose anatomical distribution patterns may play a pivotal role in the future clinical assessment of the risk of NASH and fibrosis in NAFLD.

Authorship

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

Guarantor of the article: Dr Venkat Subramanian.

Author contributions: VS designed the study, analysed the collated data and contributed to writing the paper. RDJ contributed to writing the paper. PK analysed the liver histology. GPA supervised the study and edited the paper. All authors approved the final version of the manuscript.

Acknowledgement

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References

We are grateful to Kathleen Barnard and Charlotte Davies for their help in recruiting patients for this study. We are grateful to Dr Stephen Ryder for his support in patient recruitment. This article presents independent research supported by the National Institute for Health Research (NIHR). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health. Declaration of personal and funding interests: None.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials And Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgement
  9. References