Article Text

Original research
Caucasian lean subjects with non-alcoholic fatty liver disease share long-term prognosis of non-lean: time for reappraisal of BMI-driven approach?
  1. Ramy Younes1,2,3,
  2. Olivier Govaere1,
  3. Salvatore Petta4,
  4. Luca Miele5,6,
  5. Dina Tiniakos1,7,
  6. Alastair Burt1,
  7. Ezio David3,
  8. Fabio Maria Vecchio5,8,
  9. Marco Maggioni9,
  10. Daniela Cabibi10,
  11. Duncan McLeod11,
  12. Maria Jesus Pareja12,
  13. Anna Ludovica Fracanzani13,
  14. Rocio Aller14,
  15. Chiara Rosso3,
  16. Javier Ampuero15,
  17. Rocío Gallego-Durán15,
  18. Angelo Armandi3,
  19. Gian Paolo Caviglia3,
  20. Marco Y W Zaki1,16,
  21. Antonio Liguori5,
  22. Paolo Francione13,
  23. Grazia Pennisi4,
  24. Antonio Grieco5,6,
  25. Giovanni Birolo3,
  26. Piero Fariselli3,
  27. Mohammed Eslam17,
  28. Luca Valenti18,
  29. Jacob George17,
  30. Manuel Romero-Gómez15,
  31. Quentin Mark Anstee1,19,
  32. Elisabetta Bugianesi3
  1. 1 The Newcastle Liver Research Group, Translational & Clinical Research Institute, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne, UK
  2. 2 Boehringer Ingelheim International GmbH, Ingelheim, Germany
  3. 3 Department of Medical Sciences, Division of Gastroenterology and Hepatology, A.O. Città della Salute e della Scienza di Torino, Università degli Studi di Torino, Torino, Italy
  4. 4 Sezione di Gastroenterologia, PROMISE, Università di Palermo, Palermo, Italy
  5. 5 Dipartimento Universitario Medicina e Chirurgia Traslazionale, Università Cattolica del Sacro Cuore, Rome, Italy
  6. 6 Area Medicina Interna, Gastroenterologia e Oncologia Medica, Fondazione Policlinico A. Gemelli IRCCS, Rome, Italy
  7. 7 Dept of Pathology, Aretaieion Hospital, Medical School, National and Kapodistrian University of Athens, Athens, Greece
  8. 8 Area Anatomia Patologica, Fondazione Policlinico Gemelli IRCCS, Rome, Italy
  9. 9 Department of Pathology, Ca’ Granda IRCCS Foundation, Milan, Italy
  10. 10 Pathology Institute, PROMISE, University of Palermo, Palermo, Italy
  11. 11 Department of Anatomical Pathology, Institute of Clinical Pathology and Medical Research (ICPMR), Westmead Hospital, Sydney, New South Wales, Australia
  12. 12 Pathology Unit, Hospital Universitario de Valme, Sevilla, Spain
  13. 13 Unit of Medicine and Metabolic Disease Ca' Granda IRCCS Foundation, Policlinico Hospital, Department of Pathophysiology and Transplantation, University of Milan, Milan, Italy
  14. 14 Gastroenterology, Hospital Clínico Universitario de Valladolid, Centro de Investigación de Endocrinología y Nutrición, Universidad de Valladolid, Valladolid, Spain
  15. 15 UCM Digestive Diseases and SeLiver Group, Virgen del Rocio University Hospital, Institute of Biomedicine of Seville, University of Seville, Seville, Spain
  16. 16 Biochemistry Department, Faculty of Pharmacy, Minia University, El Minia, Egypt
  17. 17 Storr Liver Centre, Westmead Institute for Medical Research, Westmead Hospital and University of Sydney, Westmead, New South Wales, Australia
  18. 18 Translational Medicine, Department of Transfusion Medicine and Hematology, Fondazione IRCCS C'a Granda Ospedale Maggiore Policlinico, Department of Pathophysiology and Transplantation, University of Milan, Milan, Italy
  19. 19 Newcastle NIHR Biomedical Research Centre, Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle upon Tyne, UK
  1. Correspondence to Dr Elisabetta Bugianesi, Department of Medical Sciences, University of Turin, Corso Dogliotti 14, Torino 10126, Italy; elisabetta.bugianesi{at}; Dr Quentin Mark Anstee, Translational & Clinical Research Institute, Newcastle University, Newcastle upon Tyne, UK; quentin.anstee{at}


Objective The full phenotypic expression of non-alcoholic fatty liver disease (NAFLD) in lean subjects is incompletely characterised. We aimed to investigate prevalence, characteristics and long-term prognosis of Caucasian lean subjects with NAFLD.

Design The study cohort comprises 1339 biopsy-proven NAFLD subjects from four countries (Italy, UK, Spain and Australia), stratified into lean and non-lean (body mass index (BMI) </≥25 kg/m2). Liver/non-liver-related events and survival free of transplantation were recorded during the follow-up, compared by log-rank testing and reported by adjusted HR.

Results Lean patients represented 14.4% of the cohort and were predominantly of Italian origin (89%). They had less severe histological disease (lean vs non-lean: non-alcoholic steatohepatitis 54.1% vs 71.2% p<0.001; advanced fibrosis 10.1% vs 25.2% p<0.001), lower prevalence of diabetes (9.2% vs 31.4%, p<0.001), but no significant differences in the prevalence of the PNPLA3 I148M variant (p=0.57). During a median follow-up of 94 months (>10 483 person-years), 4.7% of lean vs 7.7% of non-lean patients reported liver-related events (p=0.37). No difference in survival was observed compared with non-lean NAFLD (p=0.069).

Conclusions Caucasian lean subjects with NAFLD may progress to advanced liver disease, develop metabolic comorbidities and experience cardiovascular disease (CVD) as well as liver-related mortality, independent of longitudinal progression to obesity and PNPLA3 genotype. These patients represent one end of a wide spectrum of phenotypic expression of NAFLD where the disease manifests at lower overall BMI thresholds.

Lay summary NAFLD may affect and progress in both obese and lean individuals. Lean subjects are predominantly males, have a younger age at diagnosis and are more prevalent in some geographic areas. During the follow-up, lean subjects can develop hepatic and extrahepatic disease, including metabolic comorbidities, in the absence of weight gain. These patients represent one end of a wide spectrum of phenotypic expression of NAFLD.

  • nonalcoholic steatohepatitis
  • fatty liver

Data availability statement

Data are available on reasonable request. N/A—No clinical trials.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Significance of this study

What is already known on this subject?

  • Non-alcoholic fatty liver disease (NAFLD) may occur in lean patients with a body mass index (BMI) <25 kg/m2 in both Asian and Caucasian ethnicities.

  • Although some longitudinal studies have been carried out in the Asian population, the natural history of NAFLD in lean Caucasian patients is still only partially explored.

What are the new findings?

  • NAFLD may progress in Caucasian patients with a normal BMI in the absence of longitudinal progression to obesity and independent of their PNPLA3 profile.

  • Despite a more favourable metabolic profile at baseline, lean NAFLD patients experienced both hepatic and extrahepatic complications, including hepatocellular carcinoma and CVD events.

  • Caucasian lean subjects with NAFLD have a predominant geographical localisation.

Significance of this study

How might it impact on clinical practice in the foreseeable future?

  • Lean subjects with NAFLD should not be overlooked in clinical practice, as they develop all disease outcomes in the long- term

  • Well-defined phenotyping strategies should be applied in clinical trials to separate the outcome in lean and non-lean NAFLD subjects.


The worldwide burden of non-alcoholic fatty liver disease (NAFLD) has been usually associated with the global widespread of obesity.1 However, there is growing interest towards understanding the development of NAFLD in specific subgroups of individuals. ‘Lean’ NAFLD patients have been identified as subjects with a body mass index (BMI) below the ethnic-specific cut-off of 25 kg/m2 in Caucasian and 23 kg/m2 in Asian subjects.2 Studies involving this subpopulation has been mainly carried out in Asia, where epidemiological studies consolidated a stable prevalence of this phenotype in many Asian countries and most, but not all, cross-sectional and longitudinal studies suggested less severe disease3–10

On the other hand, the prevalence and outcomes of lean subjects with NAFLD among Caucasian patients are controversial. In the US National Health and Nutrition Examination Survey (NHANES) III cohort, Younossi et al 11 reported a NAFLD prevalence of 7.4% in lean subjects. The phenotype was associated with younger age, greater insulin resistance and hypercholesterolaemia. However, NAFLD diagnosis was not based on histological criteria. Characteristics and outcomes of lean Caucasian patients with an histological diagnosis of NAFLD had been reported by Francanzani et al 12 and Hagström et al.13 In both cases, single-country experiences were described, in Italy12 and Sweden,13 respectively. Regretfully, a longitudinal study describing features and long-term outcomes of a multiethnic and internationally recruited cohort of biopsy-proven lean NAFLD was only published in its abstract form.14

In this study, we aimed to describe baseline characteristics and long-term outcomes of Caucasian lean subjects with biopsy-proven NAFLD, prospectively recruited in tertiary centres in Europe and Australia.


This is an observational, multicentre cohort study of well-characterised Caucasian patients with biopsy-confirmed NAFLD who had been prospectively enrolled and followed up in tertiary centres in Europe and Australia. The study has two components: the first, a cross-sectional analysis to characterise the clinical presentation of lean patients with NAFLD compared with non-lean, and the second, a longitudinal, follow-up analysis to determine the long-term morbidity and mortality of lean subjects with NAFLD compared with non-lean ones. Patients included were consecutively biopsied and were managed in academic medical centres in four different countries (Italy, UK, Spain and Australia). Patients had been prospectively enrolled between 1990 and 2016 and their data included in secured, local databases.15 Inclusion criteria were age ≥18 years and diagnosis of NAFLD confirmed by liver biopsy. Patients with clinically overt cirrhosis (significant thrombocytopenia, prolonged prothrombin time and/or US and CT imaging showing cirrhosis and/or splenomegaly and/or varices on endoscopy in the absence of decompensation) were excluded because liver biopsy is usually not performed as it is deemed not required for diagnosis confirmation. Exclusion criteria included a liver disease of other aetiology such as alcohol-induced or drug-induced liver disease, autoimmune or viral hepatitis, and cholestatic or metabolic/genetic liver disease including Wilson’s, haemochromatosis and alpha 1 antitrypsin deficiency. All the aforementioned liver diseases had been ruled out by specific clinical, laboratory or radiographic criteria and by a histological diagnosis other than NAFLD at liver biopsy. All patients had a negative history of alcohol abuse as indicated by a weekly ethanol consumption of <140 g in women, and <210 g in men. The history of alcohol consumption was investigated by interviewing the patients and in many cases by also interviewing close relatives during both the first and subsequent visits. In more than 90% of cases, the liver biopsy was performed because of persistent elevation of liver enzymes in patients with fatty infiltration of the liver detected on repeated imaging studies. In the remaining cases, biopsy was performed for repeatedly imaging evidence of severe hepatic steatosis. For the purpose of this study, an extensive review of all the data from clinic visits, laboratory and imaging data and liver biopsy reports was performed and only patients of Caucasian ancestry with an unequivocal diagnosis of NAFLD were included. Patients were excluded when data on alcohol consumption was not detailed in the medical records.

Extensive clinical and laboratory data were collected at the time of the liver biopsy. Laboratory evaluation included routine liver biochemistry; complete blood count; lipid profile; fasting glucose; iron parameters; viral serology for hepatitis B and C infection done either at the time of liver biopsy or during the follow-up visits; autoantibodies; alpha 1 antitrypsin levels and phenotype and ceruloplasmin levels. Patients with advanced fibrosis or newly diagnosed cirrhosis at liver biopsy underwent endoscopic screening for gastro-oesophageal varices and screening for hepatocellular carcinoma (HCC) at regular intervals following standard of care recommendations or guidelines in place at specific times as proposed by liver societies.

All patients had given written informed consent for participation in medical research.

Body mass index

The BMI (body weight in kilograms divided by the height in metres squared (kg/m2)) was calculated in every patient at the time of liver biopsy. Patients were categorised into those with normal BMI (<25 kg/m2), overweight (BMI between 25 and 29.9 kg/m2) and obese (BMI ≥30 kg/m2). Lean NAFLD in this Caucasian cohort was defined by BMI <25 kg/m2.

Other metabolic conditions were recorded, including (1) hyperglycaemia (fasting glucose ≥100 mg/dL) or previously diagnosed type 2 diabetes mellitus (fasting glucose ≥126 mg/dL or treatment with antidiabetic drugs), (2) hypertriglyceridaemia (fasting triglycerides ≥150 mg/dL), (3) low high density lipoprotein (HDL)-cholesterol (<40 mg/dL in men, <50 mg/dL in women) and (4) central adiposity (waist circumference ≥88 cm in women and ≥102 cm in men). When waist circumference was not recorded in the medical records, we used a BMI ≥30 kg/m2 to define obesity.

Liver histology

Liver biopsies were routinely stained with H&E and Masson’s trichrome. Liver biopsies were scored by experienced liver pathologists in each centre using the non-alcoholic steatohepatitis (NASH) Clinical Research Network (CRN) scoring system to grade NAFLD diagnostic histological features and stage fibrosis.16 Histological features analysed included steatosis grade (0–3), lobular inflammation grade (0–3), ballooning grade (0–2) and fibrosis stage (0–4) as recommended,16 based on a historical reading at the time of biopsy. There was no rereading in batch by the pathologist, which could have accounted for changes in reading of the biopsies over time. A total of eight experienced liver pathologists scored the liver biopsy features (AB and DT in the Newcastle centre, ED, FMV, MM and DC in the Italian centres, MJP in the Seville centre and DM in the Sydney centre). The majority of the hepatopathologists are members of the fatty liver inhibition of progression/Elucidating Pathways of Steatohepatitis Histopathology group, where interobserver agreement was previously reported to be overall ~77% (k=0.54) using the NASH-CRN histology criteria.17 This high level of interobserver agreement compares favourably to that reported by the NASH-CRN.16 18 The presence of NASH was recorded and categorised as NASH or non-NASH based exclusively on the pathologists’ opinion whether or not NASH was present (based on the pattern of injury and the combined presence of steatosis, ballooning and lobular inflammation).19 All liver biopsy samples were of appropriate quality and length, and had an appropriate number of portal tracts for a confident grading and staging of the histological features as judged by the pathologist. A threshold of 5% of hepatocytes showing steatosis was required for the diagnosis of NAFLD. An exception was made for patients who underwent liver biopsy for NAFLD suspicion and revealed cirrhosis without steatosis (n=8 patients) as it is well demonstrated that steatosis may disappear at advanced fibrosis stages.


Genomic DNA was isolated from peripheral blood using the EZ1 DNA Blood 350 µL kit (Qiagen) according to the manufacturer’s instructions. Genotyping of the PNPLA3 rs738409 C>G variant (I148M) was performed by the TaqMan Single Nucleotide Polymorphism allelic discrimination assay (Applied Biosystems).


Patients were followed by GI specialists or hepatologists at 6–12 month intervals after the diagnosis of NAFLD. At each visit, a complete medical history and physical examination was performed along with a routine laboratory workup to follow their liver disease and other medical conditions. In the long-term follow-up analysis, we included those patients who underwent the diagnostic liver biopsy before 2016; this date was chosen to have at least 3 years of follow-up for the last patient recruited. During the follow-up visits, staff researchers recorded liver events (end-stage cirrhosis by MELD score above 15, cirrhosis decompensation including ascites, hepatic encephalopathy and portal hypertensive bleeding), HCC occurrence (defined by imaging/histology criteria following current clinical guidelines),20 cardiovascular events (acute coronary syndrome (myocardial infarction, unstable angina, need for coronary revascularisation), peripheral arterial ischaemia, acute cerebrovascular event (transient ischaemic attack, acute ischaemic or haemorrhagic stroke)), autoimmune disease occurrence, non-liver related cancers (including breast, colorectal, lung, prostatic, haematological, melanoma, pancreatic and urinary tract cancers) and patient deaths. Giving the lack of approved therapies for NAFLD, treatment recommendations during the study period were similar in all centres and consisted of the standard recommendations to achieve and maintain appropriate body weight with increased physical activity and dietary changes. No specific dietary intervention or specific type of physical activity was used but subjects were encouraged to lifestyle change.21 Hypoglycaemic and lipid lowering therapy was prescribed according to international guidelines. Individuals who did not experience an outcome event and whose health status was unknown for more than 12 months from review in their medical records were considered lost to follow-up. No patients underwent bariatric surgery.

Statistical analysis

Patients were grouped into lean subjects with NAFLD (BMI <25 kg/m2) and non-lean with NAFLD (BMI ≥25 kg/m2). Data are presented as median (IQR) or otherwise specified, and number (percentage) of patients with a condition. Baseline characteristics were compared between lean and non-lean using standard non-parametric tests for continuous variables or the χ2 test for categorical variables. Cumulative overall mortality during follow-up was calculated using Kaplan-Meier analysis and compared by log-rank testing. Multivariate adjusted HR (aHR) ratio estimates (relative risk) for the outcome was calculated by Cox proportional hazard regression analysis to control for the effect of potential risk factors (confounders), while taking into consideration varying lengths of follow-up. The multivariate model included the variables ‘lean NAFLD’, plus variables that may potentially bias the outcome that is, age, gender, diabetes, advanced fibrosis, BMI and site. Variables independently associated with the outcome analysed were identified by a stepwise forward selection procedure using a threshold of p<0.1 for variable selection. Time at risk (T0) was from the date of liver biopsy to the date of outcome or last follow-up. Analyses were performed using IBM SPSS Statistics V.25.0 (SPSS, Chicago, Illinois) software.


A total of 1704 Caucasian patients were initially identified, but after revision 365 were excluded because the diagnosis of NAFLD was uncertain/associated with another liver disease or their health status was unknown for more than 12 months from review of their medical records. As per protocol, we only recruited patients with a baseline liver biopsy (excluding historical biopsies), thus cirrhosis was a histological finding in the absence of clinical signs/symptoms of liver failure or complications. We included 1339 patients in the present analysis; 195 (14.6%) of them were lean. Lean patients were significantly younger (median age 45 years. vs 49 years., p=0.03) and with a higher prevalence of male gender (147/195 (75.4%) vs 717/1144 (62.7%), respectively, p=0.001). Table 1 describes the distribution of NAFLD patients (both total and lean) from each centre. Noteworthy, the majority (89.7%) of lean patients with NAFLD were derived from Italian sites; more specifically, sites in Torino, Milano and Roma had a higher prevalence of lean subjects (n=552, lean=150) compared with sites in the UK, Spain, Sydney and Palermo (n=787, lean=45).

Table 1

NAFLD patients distribution among centres and countries

Cross-sectional analysis

Table 2 shows the baseline characteristics of the total patient population and the comparison between lean and non-lean NAFLD subjects. Waist circumference was measured in 986 patients, and the proportion of patients with central obesity was significantly lower in the lean NAFLD group as compared with the non-lean group (12.2% (20/164) vs 69.7% (572/822), respectively; p<0.001). On the metabolic side, lean patients with NAFLD had a significantly lower prevalence of diabetes and lower values of glucose and triglycerides. However, total cholesterol and HDL-cholesterol values did not significantly differ between the two groups and low density lipoprotein (LDL)-cholesterol was slightly higher in the lean NAFLD group (table 2). In order to avoid treatment bias for dyslipidaemia, we undertook a subanalysis focused on diabetic patients who typically receive statins; the results confirmed higher values of LDL in the lean group (median 116; IQR 101–159) vs the non-lean (median 108; IQR 81–147; p=0.04). Overall, PNPLA3 genotype was available for 799 patients; the two groups did not show significant differences when comparing the distribution of the three genotypes (CC, CG, GG).

Table 2

Clinical and demographic characteristics of the total patient population

Table 3 shows the histological features of the patient population and the comparison between lean and non-lean groups. Lean subjects with NAFLD had significantly less steatosis, less lobular inflammation, less ballooning and less advanced liver fibrosis as compared with the non-lean group. Accordingly, lean patients showed a significant lower prevalence of NASH (54.1% vs 71.2% in lean vs non-lean patients, respectively, p<0.001). Nonetheless, half of the lean patients displayed a mild/moderate fibrosis (F1-2) and 10.2% of them had advanced fibrosis or cirrhosis (F3-4). In lean patients, the prevalence of F3-4 fibrosis at liver biopsy was associated with the presence (p<0.001) but not with the number (p=0.614) of metabolic abnormalities. Conversely, lean NAFLD without metabolic abnormalities (26%) had a higher prevalence of F0-2 (p<0.001).

Table 3

Liver biopsy features of the total patient population

Long-term follow-up analysis

The long-term outcome analyses were performed only in patients for whom the occurrence/non-occurrence of the specific event was documented. Patients were followed up for a median period of 92 months (IQR 70–132 months) and the cohort had a total of 10 483.20 person-years of follow-up.

Long-term hepatic, extrahepatic events and mortality in lean versus non-lean patients with NAFLD

Figure 1 illustrates the frequency of liver and non-liver events recorded by physicians during nearly 8 years of median follow-up, comparing lean and non-lean patients with NAFLD; data are also reported as Kaplan-Meier curves in online supplemental figure 1. In particular, the most common complication was new-onset diabetes, which occurred in 90/785 of non-lean vs 11/177 of lean patients (aHR 1.55, 95% CI 0.83 to 2.9, p=0.171). It was followed by cardiovascular events, occurring in 122/1083 of non-lean vs 14/192 of lean (aHR 1.3, 95% CI 0.73 to 2.2, p=0.39) and extrahepatic cancers, which developed in 93/1076 subjects in the non-lean group vs 17/191 in the lean one (aHR 0.42, 95% CI 0.39 to 1.4, p=0.44). As expected, diabetes was an independent predictor of cardiovascular events (aHR 1.95, 95% CI 1.36 to 2.81, p<0.001) and extrahepatic cancers (aHR 1.53, 95% CI 1.01 to 2.31, p=0.042) in the entire cohort. Extrahepatic events did not show significant differences in terms of aHR in the two groups, despite a slightly higher incidence of diabetes and CVD complications in non-lean NAFLD. Liver-related events occurred in 88/1137 of non-lean patients compared with 9/193 in the lean group (aHR 1.4, 95% CI 0.7 to 2.7, p=0.39) and HCC developed in 29/1136 of non-lean subjects compared with 2/192 of lean patients (aHR 1.9, 95% CI 0.46 to 8.1, p=0.37).

Supplemental material

Figure 1

Incidence of clinical events recorded during follow-up in lean and non-lean patients with NAFLD. NAFLD, non-alcoholic fatty liver disease.

During the follow-up, 53 patients died (5 of them were lean). Causes of death were available for 45% of the cases (total n=24; 20 non-lean and 4 lean). In the non-lean NAFLD cohort, 10 patients died of liver-related conditions (specifically: 7 patients died of liver failure, 3 because of HCC); cardiovascular events were the cause of death for 2 patients, while 8 patients died of extrahepatic cancers. In the lean cohort, causes of death were available for 4/5 patients, and showed that two patients died of liver-related events (not-HCC), one patient of ischaemic stroke and one patient died of septic shock complications. Although the survival curve appeared to be more favourable in the lean NAFLD group, overall survival analysis did not show a significant difference in terms of mortality when comparing lean subjects to overweight and obese (log-rank 5.34, p=0.069) (figure 2A). On the contrary, comparing survival in non-obese and obese patients, the latter had a significantly worse prognosis (log-rank 5.34, p=0.021) (figure 2B).

Figure 2

Comparison of overall survival. (A) Lean patients with NALFD versus obese and overweight patients. (B) Non-obese NAFLD versus obese NAFLD patients. NAFLD, non-alcoholic fatty liver disease.

Of note, the age of NAFLD patients with BMI <25 kg/m2 was significantly lower than those with BMI ≥30 kg/m2 (p=0.003); the same holds true also for patients with BMI <30 kg/m2 compared with ≥30 kg/m2 (p<0.0001). This can lead to underestimation of the risk of events at follow-up in low-BMI classes. In fact, there is a significant correlation between age and BMI, but age is also strongly related to all events at follow-up and to death (online supplemental table 1). Since BMI is not a dichotomous variable, and the definition of lean/non-lean patients can be somewhat arbitrary, we tested the impact of the exact BMI value treated as a continuous variable. When the values of BMI were used for prediction of the clinical events, apparent correlations appeared with diabetes, HCC and cardiovascular events at follow-up. However, when we adjusted for age, the link between BMI and clinical events became negligible, indicating that the correlation only arises because age influences both BMI and the clinical events (online supplemental table 2). This suggests that a BMI-driven approach to risk stratification and detection of patients with NAFLD may not take into account important variables and it should be reconsidered.

Supplemental material

Long-term change in BMI in lean patients with NAFLD

A crucial point is whether lean patients who developed long-term events remained lean during follow-up or changed their BMI category. We retrieved the last available BMI of lean patients who had completed a follow-up visit in the last 12 months or the last available BMI before death. The analysis included 98% of lean patients and results are summarised in figure 3. In the 191 lean subjects with NAFLD, mean and median BMI were 23.3 kg/m2 and 23.6 kg/m2, respectively, at baseline and 23.7 kg/m2 and 23.8 kg/m2 (p=ns) at the end of follow-up. The vast majority (77.5%) of patients remained lean, with the following distribution: 40/57 (70.2%) in Turin, 39/48 (81.3%) in Milan, 20/24 (83.3%) in Palermo, 37/45 (82.2%) in Rome, 8/9 (88.9%) in Seville and 4/7 in Sydney (57.1%) (figure 3A). Of the 43 patients who progressed, only 10 had a significant increase in BMI, defined as BMI ≥27 kg/m2. Among these subjects only three patients increased their BMI falling in the obesity category (≥30 kg/m2). The frequency of long-term events did not differ significantly when analysed according to change in BMI category, confirming data reported on a larger scale in the entire cohort (figure 3B). Similarly, no differences were noted in aHR for hepatic and extrahepatic events and mortality (online supplemental table 3).

Figure 3

BMI category change in lean patients. (A) Lean patients whose BMI changed versus non-change in the analysed centres. (C) Frequency of long-term events in patients whose BMI changed versus non-change. BMI, body mass index.

Clinical and histological predictors of outcome

In the entire cohort, univariate analysis identified baseline diabetes, age and advanced fibrosis as predictors of mortality. Supporting previous reports that both NAFL and NASH have the capacity to progress,22 baseline NASH status was not significantly predictive of liver events (HR 1.24, 95% CI 0.8 to 1.9), HCC (HR 2.1, 95% CI 0.8 to 4.9), cardiovascular events (HR 1.4, 95% CI 0.9 to 1.9), or death (HR 0.9, 95% CI 0.5 to 1.6). At multivariate Cox regression analysis only advanced fibrosis (F3/F4) was independently associated with a higher risk of mortality in the whole cohort. In keeping with BMI not being an independent predictor of mortality, lean status did not significantly influence survival (table 4).

Table 4

Multivariate-adjusted HRs and 95% CIs of outcome mortality

Longitudinal analysis in the Italian population

Given the higher prevalence of lean patients in the Italian centres and to avoid the impact this heterogeneity may have on outcomes, we conducted a longitudinal sub-analysis focused on the Italian cohort including centres from Turin, Milan, Rome and Palermo. All analyses were adjusted for the principal confounders of outcomes (age, gender, BMI, diabetes, fibrosis status) and centre. The Italian cohort included 751 patients (lean=175, 23.3%) with a median follow-up of 84 months (IQR 70–115). At baseline, diabetes was diagnosed in 127/576 (22%) non-lean vs 14/175 (8%) lean patients (p<0.001). During follow-up 33/449 non-lean vs 10/161 lean developed diabetes, with no significant difference between groups (aHR 1.47, 95% CI 0.71 to 3.1, p=0.31). Cardiovascular events were reported by 46/569 (8.1%) non lean vs 11/173 (6.4%) lean patients (aHR 1.495% CI 0.7 to 2.7, p=0.35); while the number of patients who developed extrahepatic cancers were 52/562 (9.3%) and 16/172 (9.3%) in the non-lean and lean population respectively (aHR 1.1, 95% CI 0.6 to 1.9, p=0.83). Liver events were reported by 37/569 (6.5%) non-lean patients vs 8/173 (4.6%) lean (aHR 1.5, 95% CI 0.7 to 3.3, p=0.3); while HCC was diagnosed in 11/568 (1.9%) non-lean subjects vs 1/172 (0.6%) lean (aHR 2.9, 95% CI 0.4 to 23.6, p=0.31). Overall survival was not significantly different comparing lean and non-lean subjects (log-rank 0.46, p=0.51).


The major finding from this study is that NAFLD may develop and progress in Caucasian subjects with a normal BMI, predominantly males, in the absence of longitudinal progression to obesity and independent of their PNPLA3 genotype. Our findings highlight that these subjects are not healthy but are best considered ‘lean metabolically diseased’, being one end in the wide spectrum of phenotypic expression of NAFLD that share with the non-lean counterpart similar morbidity and mortality.

The histological features of Caucasian patients with NAFLD having a normal BMI confirmed that initially lean patients have milder liver damage (in term of steatosis, lobular inflammation, ballooning, NASH and advanced fibrosis) as compared with overweight and obese; nevertheless, they are not spared from the risk of progressive liver disease, as more than half of them had NASH and 1 in 10 had severe fibrosis at the time of diagnosis. The above mentioned findings are consistent with what has been previously reported by Hagström et al 13 and McPherson et al.22 The lower prevalence of advanced fibrosis and NASH at baseline in lean subjects does not invariably translate into better prognosis, as patients with NAFL who progress have a higher prevalence of baseline inflammation or ballooning and disease activity has been identified as an important determinant of progression.

An important confounding factor is the younger age of lean subjects with NAFLD at liver biopsy, extensively reported in previously published studies, leading to an underestimation of the risk of progression as older age is a main predictor of morbidity and mortality. Indeed, when BMI was used as a continuous variable and corrected for age, the link between BMI and clinical event became negligible. These observations, supported by the results of a recent and comprehensive meta-analysis,23 suggest that a BMI-driven approach to detecting patients with NAFLD can be misleading and should be reconsidered as (1) metabolic derangements can arise independent of obesity and (2) in the absence of intervention patients’ liver-related prognosis ultimately depends on the degree of liver damage.

Indeed, despite a more favourable metabolic profile at baseline, during a median follow of nearly 8 years lean NAFLD patients suffered both hepatic and extrahepatic complications, including onset of diabetes and CVD events, which was not explained by a concomitant significant weight gain. Lean subjects can be insulin resistant, although mechanisms may be different from those acting in obese patients.24 A study among lean Caucasians showed that NAFLD in lean individuals might have a distinct metabolomic profile, with lysophosphatidylcholine, phosphatidylcholine, tyrosine and valine levels being different from those in the obese NAFLD group.25 The skeletal muscle compartment is another important determinant of metabolic homoeostasis. Sarcopenia is commonly present along with increased adipose tissue mass in patients with NAFLD26 and peripheral insulin resistance might have a prominent role.

Another important finding of this study is the identification of a phenotype of Caucasian NAFLD patients with a predominant geographical localisation. Lean NAFLD subjects mostly occurred in the Italian centres, where they represented 22% of the patients, but long-term outcomes of lean versus non-lean NAFLD did not differ significantly within the cohorts ‘enriched’ of lean patients. In turn, this suggests real heterogeneity in Europe, as the frequency of lean patients in Italy was similar to that reported by Hagström et al in Sweden (~19%).13 This finding partially mirrors variation in the prevalence of obesity across Europe, ranging from around 20% of the general population in Italy and Sweden to approximately 24% in Spain, and 29% in UK (with peaks of 37% in NE England) and Australia.27 28 However, it might also suggest the presence of as-yet unidentified genetically determined ‘subclusters’ within NAFLD patients of Caucasian ethnicity that are most evident in countries with a lower prevalence of obesity where these effects are not overwhelmed by environmental factors. Although PNPLA3 genotype distribution did not significantly differ when comparing lean and non-lean patients, other variants, such those as TM6SF2, GCKR and MBOAT7 have not been investigated.29–31 However, data from public available databases show there are no great differences in the minor allele frequencies of the common PNPLA3 and TM6SF2 variants across European countries.32 33 A possible explanation is that NAFLD arises on a predisposed genetic background but liver damage is boosted by obesity; the latter however can blur the importance of genetic predisposition as well as of other cofactors encountered in a lifetime.

In addition, multiple local environmental factors may contribute to the development of this phenotype, influencing amount and distribution of body fat. Nutrition habits are quite different between Italy and Sweden. The Swedish traditional pattern is towards meat and meat products, sauce and potatoes and often bread and margarine,34 while specific aspects of the Italian food consumption pattern are a very large contribution from olive oil to fats and a large contribution from bread, pasta and pizza to cereals.35 In our cohort, cholesterol levels were no better, and LDL even worse, in lean compared with non-lean NAFLD even among diabetics, who commonly would be receiving statin treatment. This finding has also been observed in the NANHES III and in the Swedish cohorts13 and it might reflect dietary patterns enriched of cholesterol from high level consumption of red meat, previously reported in lean NAFLD36–38 and in the Italian population.35 It might also suggest an altered cholesterol metabolism; a different metabolic adaptation to the environment mediated by differences in FXR activity, bile acids (BAs) composition and gut microbiota has recently been suggested in lean NAFLD patients.39 In particular, increased BAs levels observed in lean NAFLD are reported to mediate resistance to diet-induced obesity, a phenomenon called ‘obesity resistance’40 41; further, at the microbiota level, lean NAFLD had an increased abundance of members belonging to the Clostridium genus as well as Ruminococcaceae, which are involved in the formation of BAs.42 43 Finally, discrepancies in lean patient prevalence across countries should take into account the different national health services and access modalities to tertiary care centres.

The main strengths of our study are the large number of patients included; the long-term follow-up averaging nearly a decade per patient; complete follow-up in the vast majority of patients; having a liver biopsy confirming the diagnosis of NAFLD in every case; and having experienced liver pathologists grading and staging the biopsy features.

Our study has some limitations, such as enrolment bias at tertiary centres or absence of specific treatment protocol with uniform diet and exercise across all centres. Results are mainly driven by the Italian population, but this may simply mean that this population is not so overwhelmed by obesity and a lean phenotype can be discerned, yet. Although the overall group results have been confirmed in a separate analysis of the Italian group, this potentially limits the generalisability of the results. Further, with about three decades of overall study duration, we were not able to adjust for specific treatment modalities. We acknowledge there was not an a-priori established protocol for follow-up data collection as this is a retrospective and real-life study of a prospectively recruited cohort. The lack of a central pathologist scoring the liver biopsies is another limitation of the study. However, all recruiting centres are well-known for their interest in NAFLD and involved in several scientific collaborations with homogeneous protocols for patient enrolment and NAFLD histological scoring through the years. The relative small sample of lean NAFLD and low event rate in the whole cohort can partially account for statistical differences, but this is expected in a population where cirrhosis was a histological finding and all patients were free of events at baseline. Further, PNPLA3 genotyping was performed in 799 patients and genotype distribution subgroups are small. On the other hand, this is the largest cohort of Caucasian lean NAFLD subjects with liver biopsy and longitudinal follow-up available so far.

In conclusion, lean subjects with NAFLD can develop the full spectrum of metabolic comorbidities and liver damage that occurs in non-lean patients, in the absence of longitudinal progression to obesity and independent of their PNPLA3 genotype. These individuals thus represent one end of the full phenotypic expression of NAFLD where the disease manifests at lower overall BMI thresholds and at a younger age. It is important to characterise this population by further ‘omics’ studies and to make a reappraisal of a biased BMI-driven approach to NAFLD. These findings may have implications for clinical management and for drug development: lean subjects should not be overlooked in clinical practice and for inclusion in clinical trials, and analysis of trial results should include a granular assessment of metabolic phenotypes beyond BMI-based approaches.

Data availability statement

Data are available on reasonable request. N/A—No clinical trials.

Ethics statements

Patient consent for publication

Ethics approval

The study was approved by appropriate regulatory bodies at all participating centres.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • QMA and EB are joint senior authors.

  • Twitter @RamyYounes9, @Dina Tiniakos, @lucavalenti75, @mromerogomez, @qanstee

  • Contributors Manuscript concept and design: RY, EB and QMA, writing: RY, data collection: RY, OG, SP, LM, DT, AB, ED, FMV, MM, DC, DM, MJP, ALF, RA, CR, JA, RG-D, AA, GPC, MYWZ, AL, PF, GP, AG, ME, LV, JG, MR-G, QMA and EB, statistical analyses: RY, EB, PF and QMA, revision and editing: EB, QMA, LV, JG and MR-G, acceptance of the final version: all authors.

  • Funding This study has been supported by the EPoS (Elucidating Pathways of Steatohepatitis) consortium funded by the Horizon 2020 Framework Program of the European Union under Grant Agreement 634413 and the Newcastle NIHR Biomedical Research Centre. The authors are contributing members of The European NAFLD Registry. The study was also supported by the Italian Ministry of Health, grant RF-2016-02364358 (Ricerca Finalizzata, Ministero della Salute). ME and JG are supported by the Robert W. Storr Bequest to the Sydney Medical Foundation, University of Sydney; a National Health and Medical Research Council of Australia (NHMRC) Program Grant (APP1053206, APP1149976) and Project grants (APP1107178 and APP1108422).

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Linked Articles