Review

Non-alcoholic fatty liver disease: causes, diagnosis,

cardiometabolic consequences, and treatment strategies
Norbert Stefan, Hans-Ulrich Häring, Kenneth Cusi

The prevalence of non-alcoholic fatty liver disease (NAFLD) is increasing worldwide. In some patients with NAFLD, Lancet Diabetes Endocrinol 2018
isolated steatosis can progress to advanced stages with non-alcoholic steatohepatitis (NASH) and fibrosis, increasing Published Online
the risk of cirrhosis and hepatocellular carcinoma. Furthermore, NAFLD is believed to be involved in the pathogenesis August 30, 2018
http://dx.doi.org/10.1016/
of common disorders such as type 2 diabetes and cardiovascular disease. In this Review, we highlight novel concepts S2213-8587(18)30154-2
related to diagnosis, risk prediction, and treatment of NAFLD. First, because NAFLD is a heterogeneous disease, the
Department of Internal
advanced stages of which seem to be strongly affected by comorbidities such as insulin resistance and type 2 diabetes, Medicine IV, University
early use of reliable, non-invasive diagnostic tools is needed, particularly in patients with insulin resistance or Hospital Tübingen, Tübingen,
diabetes, to allow the identification of patients at different disease stages. Second, although the strongest genetic risk Germany (Prof N Stefan MD,
Prof H-U Häring MD); Institute
alleles for NAFLD (ie, the 148Met allele in PNPLA3 and the 167Lys allele in TM6SF2) are associated with increased of Diabetes Research and
liver fat content and progression to NASH and cirrhosis, these alleles are also unexpectedly associated with an Metabolic Diseases, Helmholtz
apparent protection from cardiovascular disease. If consistent across diverse populations, this discordance in NAFLD- Centre Munich, University of
related risk prediction between hepatic and extrahepatic disease might need to be accounted for in the management Tübingen, Tübingen, Germany
(Prof N Stefan, Prof H-U Häring);
of NAFLD. Third, drug treatments assessed in NAFLD seem to differ with respect to cardiometabolic and antifibrotic German Centre for Diabetes
efficacy, suggesting the need to better identify and tailor the most appropriate treatment approach, or to use a Research, Tübingen, Germany
combination of approaches. These emerging concepts could contribute to the development of a multidisciplinary (Prof N Stefan, Prof H-U Häring);
approach for endocrinologists and hepatologists working together in the management of NAFLD. Division of Endocrinology,
Diabetes and Metabolism,
University of Florida,
Introduction hepatocyte injury (eg, ballooning), with or without Gainesville, FL, USA
The prevalence of non-alcoholic fatty liver disease fibrosis. Because from an epidemiological perspective the (Prof K Cusi MD); and Division
(NAFLD) is increasing worldwide, and 25% of the global increased cardiometabolic risk in NAFLD seems to of Endocrinology, Malcom
Randall Veterans
adult population is potentially affected by the disease.1 depend strongly on the presence of advanced stages Administration, Medical
A total of 3% to 10% of all children and about 34% of of NAFLD, such as NASH with moderate-to-advanced Center, Gainesville, FL, USA
children who are obese in developed countries are fibrosis,12,13 improved implementation of existing, and (Prof K Cusi)
thought to have NAFLD.2,3 The increasing prevalence of further development of novel, non-invasive tools to Correspondence to:
NAFLD is accompanying the increasing prevalence diagnose these different stages of the disease, is necessary. Prof Norbert Stefan, Department
of Internal Medicine IV,
of other non-communicable diseases, including type 2 Furthermore, the extent to which the prediction of hepatic University Hospital Tübingen,
diabetes, cardiovascular disease, obesity-associated and and extrahepatic risk differs in NAFLD, and the Tübingen 72076, Germany
type 2 diabetes-associated cancer, and advanced liver mechanism explaining this phen­omenon, needs to be [email protected]
diseases such as hepatic cirrhosis and hepatic cancer.4−7 investigated. Finally, accumulating evidence from the tuebingen.de

The increasing prevalence of these diseases is related to most recent pharma­ cological clinical trials in patients
unhealthy lifestyles, particularly unhealthy diet, which with NAFLD suggests that each drug tested will have a
drives the increase in cardiometabolic diseases, cancers, broad spectrum of effects with respect to their anti-
and NAFLD.8−11 Cardiovascular disease is the leading inflammatory, antifibrotic, and cardiometabolic potential,
cause of death in people with NAFLD.12,13 However, making tailored and combin­ation therapeutic approaches
because people with NAFLD are often obese (or have the logical path in future management.
features of poor metabolic health even at normal weight)
and can have impaired glucose and lipid metabolism, Unhealthy lifestyle and NAFLD
insulin resistance, prediabetes, or type 2 diabetes,14−22 the Which parameters of an unhealthy lifestyle are the most
extent to which specifically the prevention and treatment important risk factors driving the epidemic of NAFLD?
of NAFLD could reduce morbidity and mortality in these Overnutrition and sedentarism often result in obesity
individuals remains unclear. and hepatic steatosis; however, these factors might not
Importantly, NAFLD is a heterogeneous disease, and necessarily result in hepatocyte necrosis, inflammation,
understanding the extent to which the liver phenotype and fibrosis. Furthermore, a subset individuals with
contributes to cardiometabolic risk in NAFLD is crucial to obesity (about 25–30%) might have metabolically healthy
provide clarity in this regard. NAFLD can be categorised obesity.23 Key lifestyle parameters, such as an increased
histologically into non-alcoholic fatty liver and non- intake of glucose, fructose, and saturated fat, induce
alcoholic steatohepatitis (NASH). Non-alcoholic fatty liver hepatic de-novo lipogenesis, subclinical inflammation in
is defined as the presence of at least 5% hepatic steatosis adipose tissue and liver, and insulin resistance in adipose
without evidence of hepatocellular injury in the form of tissue, the liver, and skeletal muscle. These lifestyle
hepatocyte ballooning. NASH is defined as the presence parameters are also accompanied by an increased risk of
of a least 5% hepatic steatosis and inflammation with type 2 diabetes, in which β-cell dysfunction-mediated

www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2 1

 Review

both aerobic and resistance exercise reduce liver fat

content by similar amounts. However, the beneficial
effects of exercise seem to depend substantially on
weight loss,28 although in short-term studies steatosis
Flour
Soda improved without a substantial decrease in bodyweight.29

Lifestyle-independent causes of NAFLD

Although the epidemic of NAFLD is believed to be driven
largely by unhealthy lifestyles, ageing and genetics
Unhealthy lifestyle
Glucose, fructose, saturated fat, and positive energy balance (which to date have not been shown to be considerably
Ageing influenced by lifestyle factors) might also have an
important role. Numerous inherited and acquired
genomic and epigenomic changes have a cumulative
Fatty acids, effect on the ageing phenotype.30 Additionally, age-
Visceral obesity and ceramides, De-novo lipogenesis, Diabetes
lipodystrophy-like cytokines, and triglyceride synthesis associated decline in skeletal muscle mass and functional
phenotype dysregulated 4.5
4.5
deterioration is believed to contribute to the pathogenesis
adipokines Insulin
Insuliin of many non-communicable chronic diseases, including
M

NAFLD.31 Finally, the age-related decline in sex hormones

and sex-hormone receptor expression, both in men and
Inflammation, women, not only results in a redistribution of adipose
fibrosis tissue from the lower body to the upper body, and from
Insulin resistance Gut
G t dysbiosis
dysbiosis
bi i
subcutaneous adipose depots to visceral depots, but also
in increased ectopic storage of lipids in the liver.16,32,33
Genome-wide association and exome-sequencing
DAMPs studies have provided robust evidence that genetic
PAMPs variability in PNPLA3, TM6SF2, MBOAT7, GCKR,
Glucose Genetics: SCFA, and
PNPLA3, ceramides HSD17B13, and other genes is associated with suscep­
TM6SF2,
MBOAT7,
tibility to and progression of NAFLD. These genes are
GCKR, and strongly involved in regulating the mobilisation of
HSD17B13 triglycerides from lipid droplets (PNPLA3), secretion
of VLDL (TM6SF2), remodelling of hepatic phosphati­dyl­
Figure 1: Causes of NAFLD
An unhealthy lifestyle involving a positive energy balance with increased caloric intake, particularly of glucose,
inositol acyl chain (MBOAT7), de-novo lipogenesis
fructose, and saturated fat, and sedentary behaviour results in an increased total and visceral fat mass, (GCKR), or bioactive lipid and oestradiol signalling
type 2 diabetes, gut dysbiosis, insulin resistance, and increased hepatic de-novo lipogenesis. Under these conditions, (HSD17B13; figure 1).34−36
diabetes-associated hyperinsulinaemia and hyperglycaemia exacerbate hepatic de-novo lipogenesis. Gut dysbiosis via
increased release of DAMPs and PAMPs, such as lipopolysaccharides, and dysregulated release of SCFAs and ceramides
induce hepatic inflammation and fibrosis. This process is amplified by a lipodystrophy-like phenotype with expansion
Diagnosis and staging of NAFLD
of visceral adipose tissue resulting in increased release of fatty acids and ceramides and a dysregulated pattern of For the diagnosis of hepatic steatosis, liver ultrasound is
cytokines and adipokines. A genetic predisposition for hepatic lipid accumulation, inflammation, and fibrosis also easily available in clinical practice, and with an overall
contributes to the pathogenesis of NAFLD. DAMPs=damage-associated molecular patterns. GCKR=glucokinase sensitivity of 85% and specificity of 94%, this method is
regulator. HSD17B13=hydroxysteroid 17-beta dehydrogenase 13. MBOAT7=membrane-bound O-acyltransferase
domain-containing 7. NAFLD=non-alcoholic fatty liver disease. PAMPs=pathogen-associated molecular patterns.
fairly accurate.37 However, the ability of liver ultrasound
PNPLA3=patatin-like phospholipase domain-containing protein 3. SCFAs=short-chain fatty acids. to detect hepatic steatosis at the ¹H-magnetic-resonance
TM6SF2=transmembrane-6 superfamily member 2. spectroscopy (¹H-MRS) cutoff of 5·6%38 is poor, and the
optimum sensitivity for liver ultrasound is achieved at a
hyperglycaemia promotes hepatic de-novo lipogenesis.24,25 liver fat content of at least 12·5% (sensitivity of about
Increased proinflammatory cytokine and dysregulated 80–85%).39 Among the proposed indices for the diagnosis
adipokine secretion from adipose tissue contributes to of hepatic steatosis, the widely used fatty liver index40 has
the process of increased lipid storage in the liver.24,25 a low sensitivity (73%) and specificity (74%) compared
Increased ceramide signalling also has an important role with ¹H-MRS-diagnosed hepatic steatosis.41 CT-based
in the pathogenesis of NAFLD.26,27 Additionally, gut dys­ diagnosis of hepatic steatosis is regarded as quite
biosis, induced by an unhealthy diet, is believed to accurate; however, because of the use of radiation and the
contribute to the accumulation of fat in the liver and the fact that this method of diagnosis was outperformed by
pathogenesis of NASH. In the presence of dysbiosis and dual-gradient echo MRI and ¹H-MRS,42 it cannot be
a leaky gut, bacteria-derived products can induce adipose recommended for the diagnosis of hepatic steatosis. The
tissue inflammation, hepatic steatosis, and hepatic controlled attenuation-parameter (CAP) feature, which
inflammation (figure 1).27 has been developed to quantify ultrasound attenuation
Finally, a sedentary lifestyle can also promote the during measurement of liver-stiffness vibration-con­
development of NAFLD. In this respect, increases in trolled elastography, has a sensitivity of 69% and a

2 www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2

 Review

specificity of 82% to detect biopsy-diagnosed steatosis.43

Parameters and biomarkers Cutoffs for
CAP (sensitivity 72%, specificity 86%) was outperformed advanced fibrosis*
by MRI-based proton-density fat fraction (PDFF; sen­
Non-invasive biomarker detection methods
sitivity 96%, specificity 100%) for the diagnosis of biopsy-
NAFLD fibrosis Age, BMI, IFG and diabetes, AST-to-ALT ratio, platelets, and ≤−1·455
proven steatosis.44 score50 albumin >0·676
Once hepatic steatosis is diagnosed by non-invasive FIB-4 index51 Age, AST, ALT, and platelet <1·3
methods, the true challenge is to measure disease >2·67
severity by establishing the presence of steatohepatitis, Enhanced liver Age, hyaluronic acid, aminoterminal propeptide of ≥9·8
and especially of moderate-to-severe fibrosis (fibrosis fibrosis test54 type III collagen, and tissue inhibitor of matrix metalloproteinase 1
stage of at least F2). Increased plasma alanine amino­ FibroTest Total bilirubin, γ-glutamyltransferase, α2-macroglobulin, >0·30
transferase (ALT) concentrations are not accurately (FibroSure)55 apolipoprotein A1, and haptoglobin, corrected for age and sex >0·70

predictive45 of disease severity, and even a plasma ALT Non-invasive imaging

concentration of two times the upper limit of normal VCTE56 Ultrasound-based measurement of low-frequency (50 Hz) >9·6
elastic shear-wave velocity
(>70 U/L) predicts NASH with a sensitivity of only
MRE57 MRI-based imaging of low-frequency mechanical waves >3·64
50% and a specificity of 61%.46 Thus, a liver biopsy
remains the most specific test to assess the nature and *Some indices have two cutoffs (to maximise sensitivity or specificity), which create grey zones of indeterminate
severity of liver diseases. However, the cost and potential values. For example, for the NAFLD fibrosis score, when applying the low cutoff score (−1·455) advanced fibrosis could
be excluded with high accuracy (negative predictive value of 93% in the estimation groups, and 88% in the validation
complications of biopsy have led to considerable interest groups). By applying the high cutoff score (0·676), the presence of advanced fibrosis could be diagnosed with high
in the development of novel, non-invasive methods for accuracy.50 NAFLD=non-alcoholic fatty liver disease. IFG=impaired fasting glucose. AST=aspartate aminotransferase.
use in clinical practice.47 Many non-invasive biomarkers ALT=alanine aminotransferase. FIB-4=fibrosis-4. VCTE=vibration-controlled transient elastography. MRE=magnetic
resonance elastography.
and radiological modalities have been proposed for
diagnosis of NASH and fibrosis.47,48 Among other bio­ Table 1: Non-invasive estimation of fibrosis
markers of NASH, circulating concentrations of
cytokeratin-18 fragments were proposed to be the most
reliable predictors of NASH in patients with NAFLD. endocrinology or primary care settings, because the
However, the clinical utility of cytokeratin 18 is limited by sensitivity and specificity of these tests are lower for
several issues, such as the relatively low power to earlier and more moderate stages of fibrosis than for
distinguish NAFLD from NASH or to determine the advanced fibrosis, and they have broad diagnostic grey
severity of NASH fibrosis.45,49 Blood lipidomics and zones (in which up to a third of patients cannot be
several scores based on anthropometrics, transaminases, classified).17 Vibration-controlled transient elastography
lipidomic parameters, cytokeratin-18 fragments, adipo­ (VCTE) and magnetic resonance elastography (MRE)
nectin, and resistin have been proposed;45 however, these have higher sensitivity and specificity for diagnosis of
methods are not widely used in routine clinical practice. advanced fibrosis than do the NFS and the FIB-4 index
Because the stage of fibrosis is considered to be the (table 1).47 Notably, VCTE is the best validated and most
most important determinant of mortality from liver commonly used type of elastography worldwide.56
disease and cardiovascular disease in patients with MRE is an MRI-based imaging method that involves
NAFLD,12,13 much interest surrounds the establishment low-frequency mechanical waves in the liver. MRE is the
of reliable, non-invasive tests for the diagnosis of the most expensive technique for non-invasive measurement
stage of fibrosis. In this respect, clinical parameters such of fibrosis; however, MRE is better than VCTE for the
as age, BMI, platelet count, liver transaminase detection of early fibrosis and of fibrosis stages F3–F4.
concentrations, or the diagnosis of impaired fasting Furthermore, the higher the BMI of the patient, the
glucose or diabetes are being used to generate non- better the performance of MRE compared with VCTE.57
invasive fibrosis scores.17,47,48 Among these scores, both Other methods such as multiparametric MRI, which
the NAFLD fibrosis score (NFS)50 and the fibrosis-4 uses T1, T2*, and PDFF to quantify iron deposition,
(FIB-4) index51 are based on routine clinical parameters hepatic fibroinflammation, and hepatic steatosis, also
and inexpensive biochemical measurements, and have a provides information about liver anatomy58 and might in
fairly high sensitivity and specificity for the diagnosis of the future also prove to be effective for the diagnosis of
advanced fibrosis (fibrosis stages F3–F4), with an overall advanced stages of liver disease.
positive predictive value of about 80% and negative
predictive value of about 90%.52,53 The most extensively Heterogeneity and dynamics of NAFLD
studied methods for the prediction of liver-related In their 2016 global meta-analysis, Younossi and
mortality, as per European guidelines,52 are the NFS,50 colleagues1 reported that the pooled overall prevalence of
FIB-4 index,51 enhanced liver fibrosis test,54 and FibroTest55 NASH in patients with NAFLD amounted to 59·10%.
(known as FibroSure in the USA; table 1). However, these Although this proportion is very high, the investigators1
methods might be more useful to guide disease also found that the prevalence of NASH in NAFLD
management in hepatology clinics in which patients strongly depended on a clinically based indication for
usually have more advanced disease than in biopsy. For example, in North America the prevalence of

www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2 3

 Review

NASH in patients with NAFLD was 60·64% in patients In most studies, these associations were independent of
with an indication for a biopsy, whereas the prevalence parameters commonly used to predict cardiometabolic
was only 29·85% in patients without such an indication. risk, such as age, sex, BMI, hypertension, and
From these data, Younossi and colleagues1 estimated that dyslipidaemia, although not independent of estimates of
the overall prevalence of NASH is between 1·50% and insulin resistance.65 Because insulin resistance is strongly
6·45%. involved in the pathogenesis of cardiovascular disease
In a meta-analysis of 11 cohort studies, Singh and and type 2 diabetes,66 and is strongly associated with
coworkers59 found that among patients with biopsy-proven NAFLD,14−22 the possibility that insulin resistance is a
NAFLD, 33·6% had fibrosis progression, 43·1% had major driver of increased risk of cardiometabolic disease
stable fibrosis, and 22·3% had improvement of fibrosis. in people with NAFLD cannot be excluded. However, the
Although most patients with NAFLD showed slow findings that hepatokines that are being secreted from
progression in their fibrosis stage (usually only by one or the fatty liver promote cardiometabolic diseases67 and
two stages), 17·2% of patients with non-alcoholic fatty that dysregulated hepatic lipid signalling results in
liver and baseline stage F0 fibrosis, and 18·2% of patients hyperglycaemia and dyslipidaemia62 supports a role of
with NASH and baseline stage F0 fibrosis, developed hepatic lipid overload in the pathogenesis of cardio­
progressive fibrosis.59 These findings suggest that, even metabolic diseases.
in patients with non-alcoholic fatty liver, or as often What are the relations between the different stages of
referred to, isolated steatosis, a subgroup of patients NAFLD, such as simple steatosis, NASH, and NASH
can rapidly progress to an advanced stage of fibrosis. In with fibrosis, and the incidence of cardiometabolic and
a study by McPherson and colleagues,60 a similar pro­ severe liver diseases and mortality? Although patients
portion of patients with non-alcoholic fatty liver (37%) and with isolated steatosis can develop NASH and progressive
NASH (43%) also had progression of fibrosis, and fibrosis, which puts them at an increased risk of
44% of the patients with non-alcoholic fatty liver morbidity and mortality, only fibrosis, but no other
progressed to NASH during a mean follow-up of 6·6 years. histological liver characteristics, was shown to indepen­
When the authors of both studies59,60 investigated dently predict increased all-cause and disease-specific
parameters predicting the progression of fibrosis, they mortality in patients with NAFLD.12,13,68 In the largest
found that diabetes60 and hypertension59 were the study ever done on liver biopsies, which included
strongest predictive clinical parameters. The relation 646 patients with biopsy-proven NAFLD with up to
between hypertension and increased progression of 40 years of follow-up, fibrosis (stages F2−F4), but not
fibrosis might predominantly be the result of insulin NASH (according to the NAFLD activity score and the
resistance and subclinical inflammation, which are often fatty liver inhibition of progression algorithm), strongly
seen in individuals with hypertension. Diabetes is quite and independently predicted increased liver-specific
prevalent in patients with NAFLD (23%), and more so in morbidity, and to a lesser extent, increased overall
those with NASH (47%); diabetes also represents a risk mortality.69 Although liver-related mortality differed
factor for progression to NASH, cirrhosis, and mortality.1 between patients with NAFLD (7·9%) and controls (1·4%),
Because the prevalence of NASH is much higher in cardio­ vascular disease mortality was not different
patients with type 2 diabetes than in the general between the two groups (36·9% vs 39·3%). However, in
population and NASH is believed to confer a greatly the above-mentioned studies, most patients with
increased risk of liver fibrosis, hepatocellular carcinoma, borderline or definite NASH had fibrosis stages F1−F4
and mortality,17,61 NASH is rightly being considered a new (eg, 84·9% of individuals in the study by Angulo and
complication of type 2 diabetes, and will probably be colleagues13 and 82·0% in the study by Hagström and
screened for in the future, in the same way as diabetic colleagues69). Therefore, the group of study participants
retinopathy and nephropathy. with NASH and stage F0 fibrosis was often too small to
allow definite conclusions about their risk of mortality or
Prediction of hepatic and extrahepatic diseases liver-related outcomes to be drawn. Because in these
and mortality in NAFLD studies ballooning grade, portal inflammation grade, and
Findings from research into the pathophysiology of NASH categories also predicted increased mortality and
hepatic and cardiometabolic diseases using different liver-related events in univariate analyses,13 and because
animal models62,63 highlighted the prominent role of stating when in the natural history of NASH fibrosis
NAFLD in the development of these diseases. Findings develops is difficult, NASH should still be considered a
from several epidemiological studies have suggested that condition that puts patients at an increased risk of
NAFLD is an independent risk factor for incident hepatic morbidity and mortality. Furthermore, the findings from
diseases, type 2 diabetes, cardiovascular disease, and Hagström and colleagues’ study69 might suggest that,
chronic kidney disease.64 Compared with healthy although cardiovascular disease-related mortality is high
controls, individuals with NAFLD had a 1·5-times to in NAFLD, advanced fibrosis in NAFLD can strongly
six times higher risk of cardiovascular disease, increase liver-related, and perhaps to a lesser extent,
cardiovascular disease-related mortality, and diabetes.64 cardiovascular disease-related, mortality.

4 www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2

 Review

Thus, the key questions concern whether NAFLD is an

A
independent risk factor for cardiometabolic disease, and 4 p=0·03
whether the diagnosis of NAFLD can improve the risk

Visceral fat (adjusted; kg)

prediction of cardiometabolic disease in a high-risk 3

population. 2

Cardiometabolic risk in NAFLD 1

Precise phenotyping of NAFLD and associated
0
comorbidities and statistical testing of the independent
relation between NAFLD and incidence of cardio­ B
metabolic diseases are necessary to address the link 9 p<0·0001
between NAFLD and cardiometabolic risk. So far, data 8
from precise cross-sectional phenotyping studies18 suggest 7
Liver fat (adjusted; %)
that there is a clear threshold of liver fat content 6
(roughly 6%), above which metabolic changes such as 5
muscle insulin resistance, hypertriglyceridaemia, and low 4
HDL-cholesterol concentrations are already fully 3
established. Further accumulation of liver fat content after 2
this threshold does not seem to be associated with 1
more severe metabolic or histological consequences.18
0
However, the fact that a positive linear correlation exists NGT IFG IGT IFG + IGT
between adipose tissue insulin resistance and liver fat
Figure 2: Relations of visceral fat mass and liver fat content with different
content supports the hypothesis that NAFLD is mainly
stages of glucose tolerance
driven by dysfunctional adipose tissue and the related Relations of adjusted (for age and sex) values of (A) visceral fat mass and
lipotoxic environment.18 Nevertheless, even if in such (B) liver fat content in individuals with NGT, IFG, IGT, and IFG plus IGT in
cardiometabolic outcome studies NAFLD might not be an 330 individuals in the Tübingen Diabetes Family Study. Reproduced from
Kantartzis and colleagues72 by permission of Springer Nature. Data are mean
independent marker of disease incidence, the possibility
values. Error bars show SE of the mean. NGT=normal glucose tolerance.
that NAFLD has a causative role cannot be excluded, IFG=impaired fasting glucose. IGT=impaired glucose tolerance.
because it could be involved in the pathogenesis of these
diseases via the promotion of dyslipidaemia, hyper­ dual-energy x-ray absorptiometry and MRI. The
glycaemia, insulin resistance, and a dysregulated hepato­ investigators also showed that, for a similar increase in
kine pattern.14−22,67,70 bodyweight and visceral fat mass, decrease in insulin
From a clinical perspective, in most epidemiological sensitivity was greater in individuals with NAFLD than in
studies NAFLD is associated with incident diabetes and those without NAFLD.74 Thus, increased fat mass
cardiovascular disease, independently of established and visceral obesity might not account for all of the
cardiometabolic risk factors, particularly high BMI and association between NAFLD and increased cardio­
waist circumference.64 However, because increased BMI metabolic risk, which has been observed in many studies,
and waist circumference are only weak estimates of and more research is warranted to identify the parameters
increased total body and visceral fat mass, both of which that could account for this risk.
are strongly associated with NAFLD,14−22 adjustment for
better measures of total body and visceral fat mass in Genetics of hepatic and extra-hepatic risk in
those studies may have provided null results for NAFLD. NAFLD
Findings from phenotyping studies with precise Research has allowed substantial progress in under­
measurements of liver fat content, total body fat mass and standing the role of genetics in the progression of non-
visceral fat mass, prediabetes, and insulin resistance alcoholic fatty liver to NASH, fibrosis, and hepato­cellular
could help to resolve this issue. In the Tübingen Diabetes carcinoma. Using a Mendelian randomisation approach,
Family Study,71 increased liver fat content, as measured by Lauridsen and colleagues75 showed that fatty liver
¹H-MRS, was a stronger determinant of insulin resistance associated with the 148Met allele in PNPLA3 was not
and increased intima-media thickness of the common causally linked to ischaemic heart disease.75 Furthermore,
carotid artery than total body fat mass or visceral fat mass, in a large exome-wide association study of plasma lipids
both measured by MRI.71 In the same study population, in more than 300 000 individuals, genetic variants located
liver fat content was more strongly associated with at the 148Met allele in PNPLA3 and 167Lys allele in
different stages of prediabetes than was visceral fat mass TM6SF2 were strongly associated with increased liver fat
(figure 2).72 Furthermore, Fabbrini and colleagues73 content and progression to NASH, cirrhosis, and
showed that increased ¹H-MRS-derived liver fat content hepatocellular carcinoma, but also with diabetes, lower
was a stronger determinant of insulin resistance than was blood tri­glycerides, lower LDL-cholesterol concentrations,
total body or visceral fat mass, which was measured by and protection from coronary artery disease.76 By what

www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2 5

 Review

mechanisms might these two genetic variants induce disease severity.84 However, in the same study,84 the
more severe liver disease and type 2 diabetes, but lower investigators found no evidence of strong interaction
blood triglycerides and LDL cholesterol? In a 2017 study, between adiposity and the same sequence variants to
BasuRay and coworkers77 showed that the 148Met variant influence other adiposity-associated traits. These findings
in PNPLA3 might disrupt ubiquitination and proteasomal support the hypothesis that with a high genetic risk of
degradation of PNPLA3, resulting in the accumulation of NAFLD there might be a dissociation between hepatic
PNPLA3−148Met and impaired mobilisation of trigly­ and extrahepatic complications. The exact role that
cerides from hepatic lipid droplets. These changes might genetic testing ought to have to diagnose NAFLD
in turn result in lower blood triglyceride concentrations in remains uncertain, and it is not currently recommended
individuals with the 148Met allele in PNPLA3 than in by European52 or US53 clinical practice guidelines.
those who do not carry the allele, a hypothesis that is Nevertheless, genetic testing will probably become a
supported by data from a study by Liu and colleagues.76 more important tool for risk prediction in the future
The association between the 148Met allele in PNPLA3 because of the apparent relation between genetics and
and lower LDL-cholesterol con­ centrations could be disease severity.
explained by a possible role of PNPLA3 in the metabolism
of apoB-containing lipo­proteins in the liver;78 however, no Cardiometabolic versus antifibrotic treatment
evidence directly supports this hypothesis. By contrast, of NAFLD
TM6SF2 is thought to be involved in the regulation of Overview
cholesterol biosynthesis and triglyceride and apoB The prediction of cardiometabolic risk and hepatic
secretion.79 Additionally, in a small case-control study by disease progression might differ, at least for some
Musso and coworkers80 involving 60 patients with biopsy- genetic determinants of NAFLD, but differences are also
proven NAFLD who were not obese, did not have diabetes, apparent in the effects of treatment on these outcomes.
and were normolipidaemic and 60 matched controls Guidelines recommend a holistic approach, tackling
genotyped for the rs58542926 C→T (Glu167Lys) both resolution of NASH to halt fibrosis progression and
polymorphism in TM6SF2, carriers of the 167Lys allele in mitigation of cardiovascular risk.52,53 Meaningful weight
TM6SF2 had lower postprandial lipaemia, a less loss would be a good example of such an integrated
atherogenic lipoprotein, and postprandial cholesterol approach, achieving both goals; but weight loss alone is
redistribution from smaller atherogenic lipoprotein rarely successful, and pharmacological therapy has often
subfractions to larger intestinal and hepatic VLDL1 had mixed results. Targeting liver fibrosis through bile
subfractions. Furthermore, postprandial plasma VLDL1 acid (farnesoid-X-receptor) pathways by use of obeticholic
cholesterol response independently predicted the severity acid results in an unfavourable cardiometabolic profile,85
of liver histology. Because triglyceride-rich lipoprotein whereas treating fibrosis by modulating glucose and
uptake promotes high-fat-induced liver injury,81 and lipid metabolism86 or inflammation87 has had modest
because triglyceride-rich lipoproteins link cholesterol success. These results have led to a debate as to which
concentration in VLDL subclasses to hepatic cholesterol approach to follow and the most appropriate outcomes
content, inflammation, and fibrosis,82 Musso and to target for NASH trials,88 and whether the best strategy
colleagues80 concluded that the 167Lys allele in TM6SF2 would be one centred on so-called antifibrotic or
might divert toxic cholesterol away from the vessel walls cardiometabolic approaches.
into the liver and adipose tissue, thereby promoting liver
injury and adipose dysfunction and protecting from Antifibrotic therapies for NAFLD
cardiovascular disease. Similar mechanisms could also Generally, if approved in clinical practice, the use of
account for the relation between the 148Met allele in treatments intended to prevent adverse liver outcomes
PNPLA3 and a reduced risk of cardiovascular disease. The should be focused on patients with NASH, and within
increased risk of diabetes in carriers of the 167Lys allele in this group on patients with at least moderate fibrosis (at
TM6SF2 and the 148Met allele in PNPLA376 might be least stage F2), because such patients have higher
explained by a lower concentration of LDL cholesterol in cardiovascular and liver-related morbidity and mortality
the blood. However, because genetic variants with a than patients with milder disease.12,13,68,69 Although an
similarly strong ability to lower LDL-cholesterol ideal inter­vention stage (ie, where the benefit of inter­
concentrations display different effects on incident vention to reverse NASH-fibrosis will be the greatest) has
diabetes,83 other mechanisms of these alleles, other than not been established, these recom­mendations help to
those related to lowering LDL cholesterol, could induce define a patient group in which liver disease is likely to
diabetes. progress to meaningful outcomes (ie, cirrhosis or even
With respect to genetic determinants of NAFLD, hepato­cellular carcinoma) in a substantial number of
Stender and colleagues84 showed that adiposity augments individuals unless active intervention takes place. In
the genetic risk of NAFLD from steatosis to hepatic other words, treating the many patients with mild
inflammation to cirrhosis, which supports the hypothesis fibrosis (stage F1) could lead to unnecessary treatment
that genetics might predominantly be a modifier of and expenses, and in many of these people liver disease

6 www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2

 Review

will not improve with treatment. These assumptions designed, 48-week randomised controlled trial, lifestyle
might change in the future as we learn more about intervention in patients with biopsy-proven NASH that
disease progression in individuals at the highest risk for resulted in a mean weight loss of 9·3% (vs 0·2% in the
liver disease, such as patients with obesity and control group) significantly improved the NAFLD activity
type 2 diabetes. At the other end of the disease spectrum, score compared with the control group, but fibrosis was
a strong case can be made for treating patients with not affected (p=0·62).97 Thus, in NASH-fibrosis, pharma­
cirrhosis (fibrosis stage F4), especially in the early stages; cological treatment might be necessary in addition to
although, equally, if cirrhosis is reached, it might be too ongoing lifestyle intervention. In another large study
late for therapies to have a beneficial effect. assessing a lifestyle intervention, fibrosis improved when
No therapies that primarily target fibrosis in NASH weight loss of at least 10% was achieved, but the results
have been approved, but many are under investigation.89−92 were variable, and less than 10% of study participants
From available phase 2b and 3 trials, antifibrotics have lost this amount of weight.98 Most bariatric surgery
had minimal effects on the unfavourable cardiometabolic studies, which also included patients with NASH, and in
milieu of NASH. As such, combination therapy might which weight loss is often higher than 30%, have shown
have greater success if underlying metabolic insults, improvements in NAFLD activity score, but with less
believed to be driving necroinflammation and eventually conclusive or mixed results for liver fibrosis.52,53 Notably,
fibrosis (ie, insulin resistance, lipotoxicity, glucotoxicity, after bariatric surgery, patients with NASH still have an
and subclinical inflammation), can be mitigated—a increased risk of death compared with bariatric patients
hypothesis that has yet to be fully tested but that underlies without NASH, although patients with NASH still
the use of cardiometabolic therapies for treatment of benefit from bariatric surgery in so far as they have a
NASH. decreased risk of mortality.99
Metformin is an insulin sensitiser, but is not
Cardiometabolic therapies for NAFLD believed to offer unique benefits for steatohepatitis. The
In NAFLD, the liver becomes an indicator of cardio­ glucose-lowering effect of pioglitazone is similar to that
metabolic health, a barometer that allows the of metformin (HbA1c reduction of about 1·0−1·2%), but
identification of individuals (both with and without by contrast with metformin, the thiazolidinediones
obesity),14,16 who are insulin resistant and in a metabolic have been shown to improve insulin sensitivity and
state of severe adipose-tissue dysfunction and lipo­ liver histology (ie, steatosis, hepatocyte necrosis, and
toxicity.19,26,27,93 Patients with prediabetes or type 2 diabetes inflammation) in patients both with100,101 and without102,103
and NAFLD have the highest cardiometabolic disease diabetes. A fairly short-term (6 month) study by Belfort
risk,13,15 with a liver-imaging ¹H-MRS threshold of at least and colleagues100 established that NASH could be
5·5% clearly establishing a point at which dyslipidaemia, reversed within a relatively short period with pioglitazone.
insulin resistance, and lipotoxicity are fully established The effect of thiazolidinediones on fibrosis is more
to drive steatohepatitis.18 However, from a liver-centric modest than their effect on NASH. However, pioglitazone
perspective, to avoid overdiagnosis and overtreatment,94 was shown to reduce liver fibrosis and increase adipose-
general practitioners should focus on the diagnosis and tissue insulin sensitivity (highly linked to the
treatment of patients with NASH who have moderate-to- pathophysiology of the disease) more substantially in
severe fibrosis. patients with type 2 diabetes than in patients with
A cardiometabolic intervention for NASH relies on the prediabetes;104 additionally, findings from a meta-analysis
central hypothesis that weight loss, either alone or of all available randomised trials100−103 showed a benefit
combined with the use of drugs developed for the of pioglitazone in patients with advanced fibrosis
treatment of diabetes, by reversing insulin resistance and (stages F3–F4).105 These findings lend credence to the
hyperglycaemia, will result in decreased cardiometabolic hypothesis that amelioration of the metabolic insult
risk while slowing or halting steatohepatitis disease might help to mitigate fibrosis progression.
activity, and eventually, fibrosis.15,17,93 Treatment options Pioglitazone is effective for the prevention of diabetes
include lifestyle modification, bariatric surgery, and in people with prediabetes,106 and reduces cardiovascular
pharma­ cological treatments (table 2), as reviewed events in patients with metabolic syndrome or pre­
elsewhere.89−92,94 Lifestyle modification should always diabetes with a history of a stroke.107 The drug also
be the first-line treatment for NASH. Histological improves atherogenic dyslipidaemia, an important
improvement in NASH is usually proportional to the cardio­­vascular target.108 Notably, steatosis and insulin
amount of weight loss.95 Although a bodyweight loss of resistance, but not steatohepatitis per se, promote
about 5% is associated with a reduction in liver fat atherogenic dyslipidaemia in NAFLD.109 Pioglitazone
content of about 30% and an improvement of metabolic ameliorates atherogenesis and cardiovascular events in
abnormalities, a weight loss of about 7–10% might be patients with type 2 diabetes,110−114 and was associated
needed to substantially reduce hepatocyte necrosis and with lower mortality compared with other diabetes
inflammation.29,53,96 But how effective is a large amount of drugs in a large European multicohort study.115 However,
weight loss on the regression of fibrosis? In a well no reduction in cardiovascular mortality was shown

www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2 7

 Review

Cardiometabolic effects Hepatic effects

Insulin resistance Major cardiometabolic effects Cardiovascular Steatosis NAS Fibrosis
disease benefit
Lifestyle modification Moderate decrease Weight loss, and mild decrease in dyslipidaemia Yes Moderate decrease Moderate decrease Small decrease
and blood pressure or no effect
Bariatric surgery Substantial decrease Weight loss, and mild decrease in dyslipidaemia Yes Substantial decrease Substantial decrease Small decrease
and blood pressure
Thiazolidinediones
Pioglitazone Substantial decrease Mild decrease in dyslipidaemia and blood pressure Yes Substantial decrease Substantial decrease Small decrease
or no effect
Glucagon-like peptide-1 receptor agonists
Liraglutide Small decrease Weight loss Yes Moderate decrease Moderate decrease No effect
Exenatide Small decrease Weight loss No Moderate decrease NA NA
Dipeptidyl peptidase-4 inhibitors
Sitagliptin No effect No effect No No effect NA NA
Vildagliptin No effect No effect No Small decrease NA NA
Sodium-glucose co-transporter-2 inhibitors
Canagliflozin Small decrease Weight loss and small decrease in blood pressure Yes Small decrease NA NA
Empagliflozin Small decrease Weight loss and small decrease in blood pressure Yes NA NA NA
Dapagliflozin Small decrease Weight loss and small decrease in blood pressure Unknown No effect NA NA
Luseogliflozin* NA Weight loss and small decrease in blood pressure Unknown Small decrease NA NA
Ipragliflozin* NA Weight loss and small decrease in blood pressure Unknown Small decrease NA NA
Antioxidants
Vitamin E No effect Small decrease in oxidative stress and potential No (potentially Moderate decrease Moderate decrease† No effect
small decrease in inflammation harmful)
Phosphodiesterase inhibitors
Pentoxifylline No effect Small decrease in oxidative stress and potential Unknown Small decrease Small decrease No effect
small decrease in inflammation

Data are from randomised controlled trials only. NAS=non-alcoholic steatohepatitis activity score. NA=no data available. *Randomised open-label trials. †No significant effect on NAS in patients with
type 2 diabetes.

Table 2: Lifestyle, weight loss, and pharmacological interventions for the treatment of non-alcoholic fatty liver disease89-92,94

with pioglitazone compared with sulfonylureas in studies, no association was identified between
patients with type 2 diabetes in a large clinical trial done pioglitazone and bladder cancer and when patients who
in Italy.116 Weight gain (less than 2 kg on average), heart had ever used the drug were compared with those who
failure, bladder cancer, and fractures were not had not.121 However, there was a significant association
significantly different between treatment groups. For a between exposure to pioglitazone for 1–2 years (hazard
population at high cardiovascular risk, the metabolic ratio 1·28, 95% CI 1·08–1·55]) and more than 2 years
benefits of pioglitazone are frequently overlooked (1·42, 1·14–1·77) with bladder cancer. On an absolute
because of fear of long-term weight gain (3−5% of scale, development of bladder cancer in patients both
patients),117 oedema in the lower extremities (about 5% of exposed and not exposed to pioglitazone was lower than
patients), and potential bone loss.118 Notably, weight gain 0·3%. The numbers needed to treat for one additional
with pioglitazone makes patients more metabolically case of bladder cancer ranged from 899 to 6380 (median
healthy.100,101 This observation could help to account for 2540), whereas the numbers needed to treat for
the cardiovascular disease reduction associated with preventing one cardiovascular disease event ranged
pioglitazone use,103,110−114 when compared with the from four to 256 and for one resolution of NASH ranged
epidemiological evidence for increased cardiovascular from two to 12. In summary, having been incorporated
disease with obesity from overeating. Heart failure as a treatment option for patients with and without
has been reported in patients with undiagnosed pre- diabetes in several guidelines,52,53 pioglitazone could be a
existing heart disease when associated with crucial treatment option for patients with NASH,
thiazolidinedione-induced fluid retention,108 but pioglita­ similar to primacy given to metformin in initial treatment
zone has been shown to improve cardiac function.119 of patients with type 2 diabetes—ie, pioglitazone could
Finally, controversy surrounding pioglitazone and act as an initial low-cost therapy that provides hepatic
bladder cancer diminished after a 10-year prospective and cardiometabolic benefits, with the potential to be
study did not show an association.120 In a meta-analysis combined with future antifibrotic therapies under
of 11 observational cohort studies and seven case-control development.

8 www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2

 Review

Several other glucose-lowering drugs are also under could help to predict the development of hepatic and
intense investigation for NASH. Dipeptidyl peptidase-4 extrahepatic disease, and perhaps even guide the choice
inhibitors are neutral with respect to cardiovascular of the pharmacological intervention with emphasis on
outcomes in type 2 diabetes, and are not believed to the metabolic or antifibrotic efficacy of drug treatments,
have a substantial beneficial effect on NAFLD.96 By
contrast, the results of the landmark proof-of-concept Risk factors
LEAN trial of the glucagon-like peptide-1 (GLP-1) Obesity, glucose, fructose, saturated fat, genetics, and diabetes
receptor agonist liraglutide122 suggested the potential for
Imaging
use of this drug class in patients with NASH. Findings biomarkers
from several studies support the ability of GLP-1
receptor agonists to reduce plasma aminotransferases
and hepatic steatosis, as reviewed elsewhere,17,89−91 and
newer GLP-1 receptor agonists such as semaglutide are
being assessed in large clinical trials in obesity and in Lifestyle and
NASH.123 Liraglutide and semaglutide have been shown pharmacological
Imaging
intervention:
to affect cardiometabolic risk and reduce cardiovascular biomarkers Cirrhosis and HCC
>10% weight loss†,
events in patients with type 2 diabetes,124 at least in part healthy diet‡,
metabolic and
due to weight loss, amelioration of hyperglycaemia, and antifibrotic drugs
enhanced insulin action. Finally, sodium-glucose
co-transporter-2 (SGLT2) inhibitors are being used
Imaging Lifestyle and
increasingly frequently in the treatment of type 2 pharmacological
biomarkers NASH and fibrosis
diabetes.125 In animal studies of NASH, SGLT2 inhibitors intervention:
>10% weight loss†,
have shown metabolic and liver histological benefits— healthy diet‡, CVD and diabetes
these findings are also supported by results of small, and metabolic drugs
Fat and
uncontrolled clinical trials in which plasma amino­ Genetics? fibrosis
brro
rosis
o
transferases and steatosis were reduced.96 SGLT2 Lifestyle NASH CVD
VD
inhibitors promote weight loss (probably the main Intervention:
5–8% weight loss*
mechanism of action in NAFLD) and have been reported and healthy diet Fat and Isolated steatosis
Genetics? fibrosis
to reduce cardiovascular events in patients with
CVD
type 2 diabetes,124 both of which are attractive properties
Isolated steatosis
for patients with NAFLD. Fat
at and
at and
Genetics? fibrosis
bbrro
ros
osis

Conclusions CVD
VD
VD

The ongoing worldwide epidemic of NAFLD and the

fact that a subgroup of individuals with the disease is at
an increased risk of developing advanced stages of liver
disease and cardiometabolic disease means that NAFLD Lifetime
is an important priority for health care and research.
However, NAFLD is a complex and heterogeneous Figure 3: Natural history of NAFLD, risk of liver and cardiometabolic diseases, and treatment approaches
disease, in which the stage of liver damage and the CVD=cardiovascular disease. HCC=hepatocellular carcinoma. NASH=non-alcoholic steatohepatitis. *Weight loss of
5−8% is recommended for the prevention of cardiometabolic diseases in individuals who are overweight and
cardiometabolic risk parameters strongly affect risk obese with an increased cardiometabolic risk.23 †Weight loss of more than 10% was associated with a reduced risk
prediction, disease progression, and liver-related mor­ of cardiovascular morbidity and mortality among patients who were overweight or obese with type 2 diabetes in
bidity and mortality. Thus, a close collaboration between the Look AHEAD study,23 and is considered necessary to improve cardiometabolic risk in individuals who are
metabolically unhealthy.23 ‡A healthy diet was effective at reducing increased liver fat content and protecting from
hepatologists and endocrinologists is necessary to
cardiovascular events and cardiovascular mortality, independent of weight loss.23
provide the best medical support for patients. A
multifactorial approach is required, involving a
combination of diagnostic tools, stratification of risk Search strategy and selection criteria
prediction of hepatic and cardiometabolic diseases, and We searched PubMed for full-text original studies and review articles in English
tailored treatment of hepatic and cardiometabolic published from Jan 1, 1990, to March 31, 2018, to identify reports about the causes and
complications. Although in the early stages of NAFLD, consequences of non-alcoholic fatty liver disease. The search terms used were
during which the prevalence of NASH and advanced “nonalcoholic fatty liver disease”, “nonalcoholic steatohepatitis” and “liver fibrosis”,
fibrosis is low, a weight loss of 5−8% and a healthy diet together with “hepatocellular carcinoma”, “mortality”, “cardiovascular mortality”,
might be sufficient treatment (figure 3), in more “liver-related mortality”, “type 2 diabetes”, “insulin resistance”, “cardiovascular disease”,
advanced stages of liver disease—associated with genetic “prediction”, “prevention”, “lifestyle intervention”, and “treatment”. The reference lists
risk and the presence of diabetes—intensified lifestyle of the identified papers were also used to identify further papers of interest. The final
intervention supported by pharma­ cological treatment reference list was selected on the basis of relevance to the topic of this Review.
might be necessary. In the near future, genetic testing

www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2 9

 Review

including combination therapy. Much progress has been 15 Stefan N, Fritsche A, Schick F, Häring HU. Phenotypes of
made in the scientific understanding of the natural prediabetes and stratification of cardiometabolic risk.
Lancet Diabetes Endocrinol 2016; 4: 789−98.
history of NAFLD in the past decade. In the future, 16 Stefan N, Schick F, Häring HU. Causes, characteristics, and
personalised risk prediction and individualised treatment consequences of metabolically unhealthy normal weight in
could become a reality in the management of this disease. humans. Cell Metab 2017; 26: 292−300.
17 Bril F, Cusi K. Management of nonalcoholic fatty liver disease in
Contributors patients with type 2 diabetes: a call to action. Diabetes Care 2017;
NS and KC reviewed the scientific literature and wrote the Review. 40: 419−30.
H-UH critically reviewed the initial draft and contributed to the 18 Bril F, Barb D, Portillo-Sanchez P, et al. Metabolic and histological
development of the submitted version. implications of intrahepatic triglyceride content in nonalcoholic
fatty liver disease. Hepatology 2017; 65: 1132−44.
Declaration of interests
We declare no competing interests. 19 Gastaldelli A. Insulin resistance and reduced metabolic flexibility:
cause or consequence of NAFLD? Clin Sci (Lond) 2017;
Acknowledgments 131: 2701−04.
This work was supported in part by funding from the German Research 20 Byrne CD, Targher G. NAFLD: a multisystem disease. J Hepatol
Foundation (KFO 114 and STE 1096/1−3) and the German Federal 2015; 62: S47−64.
Ministry of Education and Research to the German Centre of Diabetes 21 Younossi Z, Anstee QM, Marietti M, et al. Global burden of NAFLD
Research. and NASH: trends, predictions, risk factors and prevention.
Nat Rev Gastroenterol Hepatol 2018; 15: 11−20.
References
22 Rinella ME. Nonalcoholic fatty liver disease: a systematic review.
1 Younossi ZM, Koenig AB, Abdelatif D, Fazel Y, Henry L, Wymer M.
JAMA 2015; 313: 2263−73.
Global epidemiology of nonalcoholic fatty liver disease—
meta-analytic assessment of prevalence, incidence, and outcomes. 23 Stefan N, Häring HU, Schulze MB. Metabolically healthy obesity:
Hepatology 2016; 64: 73−84. the low-hanging fruit in obesity treatment?
Lancet Diabetes Endocrinol 2018; 6: 249–58.
2 Anderson EL, Howe LD, Jones HE, Higgins JPT, Lawlor DA,
Fraser A. The prevalence of non-alcoholic fatty liver disease in 24 Donnelly KL, Smith CI, Schwarzenberg SJ, Jessurun J, Boldt MD,
children and adolescents: a systematic review and meta-analysis. Parks EJ. Sources of fatty acids stored in liver and secreted via
PLoS One 2015; 10: e0140908. lipoproteins in patients with nonalcoholic fatty liver disease.
J Clin Invest 2005; 115: 1343–51.
3 Alisi A, Feldstein AE, Villani A, Raponi M, Nobili V.
Pediatric nonalcoholic fatty liver disease: a multidisciplinary 25 Stefan N, Kantartzis K, Häring HU. Causes and metabolic
approach. Nat Rev Gastroenterol Hepatol 2012; 9: 152−61. consequences of fatty liver. Endocr Rev 2008; 29: 939−60.
4 GBD 2016 Disease and Injury Incidence and Prevalence 26 Petersen MC, Shulman GI. Roles of diacylglycerols and ceramides
Collaborators. Global, regional, and national incidence, prevalence, in hepatic insulin resistance. Trends Pharmacol Sci 2017; 38: 649−65.
and years lived with disability for 328 diseases and injuries for 27 Marra F, Svegliati-Baroni G. Lipotoxicity and the gut-liver axis in
195 countries, 1990−2016: a systematic analysis for the Global NASH pathogenesis. J Hepatol 2018; 68: 280−95.
Burden of Disease Study 2016. Lancet 2017; 390: 1211−59. 28 Bacchi E, Negri C, Targher G, et al. Both resistance training and
5 GBD 2016 Risk Factors Collaborators. Global, regional, and national aerobic training reduce hepatic fat content in type 2 diabetic
comparative risk assessment of 84 behavioural, environmental and subjects with nonalcoholic fatty liver disease (the RAED2
occupational, and metabolic risks or clusters of risks, 1990−2016: randomized trial). Hepatology 2013; 58: 1287−95.
a systematic analysis for the Global Burden of Disease Study 2016. 29 Lomonaco R, Sunny NE, Bril F, Cusi K. Nonalcoholic fatty liver
Lancet 2017; 390: 1345−422. disease: current issues and novel treatment approaches. Drugs 2013;
6 Pearson-Stuttard J, Zhou B, Kontis V, Bentham J, Gunter MJ, 73: 1−14.
Ezzati M. Worldwide burden of cancer attributable to diabetes and 30 Kennedy BK, Berger SL, Brunet A, et al. Geroscience: linking aging
high body-mass index: a comparative risk assessment. to chronic disease. Cell 2014; 159: 709−13.
Lancet Diabetes Endocrinol 2018; 6: e6–15. 31 Kalinkovich A, Livshits G. Sarcopenic obesity or obese sarcopenia:
7 Younossi ZM, Otgonsuren M, Henry L, et al. Association of a cross talk between age-associated adipose tissue and skeletal
nonalcoholic fatty liver disease (NAFLD) with hepatocellular muscle inflammation as a main mechanism of the pathogenesis.
carcinoma (HCC) in the United States from 2004 to 2009. Ageing Res Rev 2017; 35: 200−21.
Hepatology 2015; 62: 1723−30. 32 Tchkonia T, Thomou T, Zhu Y, et al. Mechanisms and metabolic
8 Micha R, Peñalvo JL, Cudhea F, Imamura F, Rehm CD, implications of regional differences among fat depots. Cell Metab
Mozaffarian D. Association between dietary factors and mortality 2013; 17: 644−56.
from heart disease, stroke, and type 2 diabetes in the United States. 33 Palmer BF, Clegg DJ. The sexual dimorphism of obesity.
JAMA 2017; 317: 912−24. Mol Cell Endocrinol 2015; 402: 113−19.
9 Khera AV, Emdin CA, Drake I, et al. Genetic risk, adherence to a 34 Eslam M, Valenti L, Romeo S. Genetics and epigenetics of NAFLD
healthy lifestyle, and coronary disease. N Engl J Med 2016; and NASH: clinical impact. J Hepatol 2018; 68: 268−79.
375: 2349−58. 35 Anstee QM, Seth D, Day CP. Genetic factors that affect risk of
10 Stefan N, Häring HU, Hu FB, Schulze MB. Divergent associations alcoholic and nonalcoholic fatty liver disease. Gastroenterology 2016;
of height with cardiometabolic disease and cancer: epidemiology, 150: 1728−44.
pathophysiology, and global implications. 36 Abul-Husn NS, Cheng X, Li AH, et al. A protein-truncating
Lancet Diabetes Endocrinol 2016, 4: 457−67. HSD17B13 variant and protection from chronic liver disease.
11 Eslamparast T, Tandon P, Raman M. Dietary composition N Engl J Med 2018; 378: 1096−106.
independent of weight loss in the management of non-alcoholic 37 Hernaez R, Lazo M, Bonekamp S, et al. Diagnostic accuracy and
fatty liver disease. Nutrients 2017; 9: 800. reliability of ultrasonography for the detection of fatty liver:
12 Ekstedt M, Hagström H, Nasr P, et al. Fibrosis stage is the strongest a meta-analysis. Hepatology 2011; 54: 1082−90.
predictor for disease-specific mortality in NAFLD after up to 38 Szczepaniak LS, Nurenberg P, Leonard D, et al. Magnetic resonance
33 years of follow-up. Hepatology 2015; 61: 1547−54. spectroscopy to measure hepatic triglyceride content: prevalence of
13 Angulo P, Kleiner DE, Dam-Larsen S, et al. Liver fibrosis, but no hepatic steatosis in the general population.
other histologic features, is associated with long-term outcomes of Am J Physiol Endocrinol Metab 2005; 288: E462–68.
patients with nonalcoholic fatty liver disease. Gastroenterology 2015; 39 Bril F, Ortiz-Lopez C, Lomonaco R, et al. Clinical value of liver
149: 389−97. ultrasound for the diagnosis of nonalcoholic fatty liver disease in
14 Yki-Järvinen H. Non-alcoholic fatty liver disease as a cause and a overweight and obese patients. Liver Int 2015; 35: 2139−46.
consequence of metabolic syndrome. Lancet Diabetes Endocrinol 40 Bedogni G, Bellentani S, Miglioli L, et al. The Fatty Liver Index:
2014; 2: 901−10. a simple and accurate predictor of hepatic steatosis in the general
population. BMC Gatroenterol 2006; 6: 33.

10 www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2

 Review

41 Kantartzis K, Rettig I, Staiger H, et al. An extended fatty liver index 63 Santhekadur PK, Kumar DP, Sanyal AJ. Preclinical models of
to predict non-alcoholic fatty liver disease. Diabetes Metab 2017; non-alcoholic fatty liver disease. J Hepatol 2018; 68: 230−37.
43: 229−39. 64 Adams LA, Anstee QM, Tilg H, Targher G. Non-alcoholic fatty liver
42 Lee SS, Park SH, Kim HJ, et al. Non-invasive assessment of hepatic disease and its relationship with cardiovascular disease and other
steatosis: prospective comparison of the accuracy of imaging extrahepatic diseases. Gut 2017; 66: 1138−53.
examinations. J Hepatol 2010; 52: 579−85. 65 Pisto P, Santaniemi M, Bloigu R, Ukkola O, Kesäniemi YA.
43 Karlas T, Petroff D, Sasso M, et al. Individual patient data Fatty liver predicts the risk for cardiovascular events in middle-aged
meta-analysis of controlled attenuation parameter (CAP) technology population: a population-based cohort study. BMJ Open 2014;
for assessing steatosis. J Hepatol 2017; 66: 1022−30. 4: e004973.
44 Park CC, Nguyen P, Hernandez C, et al. Magnetic resonance 66 Saltiel AR, Kahn CR. Insulin signalling and the regulation of
elastography vs transient elastography in detection of fibrosis glucose and lipid metabolism. Nature 2001; 414: 799–806.
and noninvasive measurement of steatosis in patients with 67 Stefan N, Häring HU. The role of hepatokines in metabolism.
biopsy-proven nonalcoholic fatty liver disease. Gastroenterology 2017; Nat Rev Endocrinol 2013; 9: 144−52.
152: 598−607. 68 Dulai PS, Singh S, Patel J, et al. Increased risk of mortality by
45 Younossi ZM, Loomba R, Anstee QM, et al. Diagnostic modalities fibrosis stage in nonalcoholic fatty liver disease: systematic review
for non-alcoholic fatty liver disease (NAFLD), non-alcoholic and meta-analysis. Hepatology 2017; 65: 1557−65.
steatohepatitis (NASH) and associated fibrosis. Hepatology 2017; 69 Hagström H, Nasr P, Ekstedt M, et al. Fibrosis stage but not NASH
published online Dec 9. DOI:10.1002/hep.29721. predicts mortality and time to development of severe liver disease in
46 Verma S, Jensen D, Hart J, Mohanty SR. Predictive value of ALT biopsy-proven NAFLD. J Hepatol 2017; 67: 1265−73.
levels for non-alcoholic steatohepatitis (NASH) and advanced 70 Tilg H, Moschen AR, Roden M. NAFLD and diabetes mellitus.
fibrosis in non-alcoholic fatty liver disease (NAFLD). Liver Int 2013; Nat Rev Gastroenterol Hepatol 2017; 14: 32−42.
33: 1398−405.
71 Stefan N, Kantartzis K, Machann J, et al. Identification and
47 Tapper EB, Lok AS. Use of liver imaging and biopsy in clinical characterization of metabolically benign obesity in humans.
practice. N Engl J Med 2017; 377: 756−68. Arch Intern Med 2008; 168: 1609−16.
48 Castera L. Noninvasive assessment of liver fibrosis. Dig Dis 2015; 72 Kantartzis K, Machann J, Schick F, Fritsche A, Häring HU,
33: 498−503. Stefan N. The impact of liver fat vs visceral fat in determining
49 Cusi K, Chang Z, Harrison S, et al. Limited value of plasma categories of prediabetes. Diabetologia 2010; 53: 882−89.
cytokeratin-18 as a biomarker for NASH and fibrosis in patients 73 Fabbrini E, Magkos F, Mohammed BS, et al. Intrahepatic fat, not
with non-alcoholic fatty liver disease. J Hepatol 2014; 60: 167−74. visceral fat, is linked with metabolic complications of obesity.
50 Angulo P, Hui JM, Marchesini G, et al. The NAFLD fibrosis score: Proc Natl Acad Sci USA 2009; 106: 15430−35.
a noninvasive system that identifies liver fibrosis in patients with 74 Fabbrini E, Yoshino J, Yoshino M, et al. Metabolically normal obese
NAFLD. Hepatology 2007; 45: 846−54. people are protected from adverse effects following weight gain.
51 Shah AG, Lydecker A, Murray K, et al. Comparison of noninvasive J Clin Invest 2015; 125: 787−95.
markers of fibrosis in patients with nonalcoholic fatty liver disease. 75 Lauridsen BK, Stender S, Kristensen TS, et al. Liver fat content,
Clin Gastroenterol Hepatol 2009; 7: 1104−12. non-alcoholic fatty liver disease, and ischaemic heart disease:
52 European Association for the Study of the Liver (EASL), European mendelian randomization and meta-analysis of 279 013 individuals.
Association for the Study of Diabetes (EASD), European Eur Heart J 2018; 39: 385−93.
Association for the Study of Obesity (EASO). EASL-EASD-EASO 76 Liu DJ, Peloso GM, Yu H, et al. Exome-wide association study of
clinical practice guidelines for the management of non-alcoholic plasma lipids in >300,000 individuals. Nat Genet 2017; 49: 1758−66.
fatty liver disease. Diabetologia 2016; 59: 1121−40.
77 BasuRay S, Smagris E, Cohen JC, Hobbs HH. The PNPLA3 variant
53 Chalasani N, Younossi Z, Lavine JE, et al. The diagnosis and associated with fatty liver disease (I148M) accumulates on lipid
management of nonalcoholic fatty liver disease: practice guidance droplets by evading ubiquitylation. Hepatology 2017; 66: 1111−24.
from the American Association for the Study of Liver Diseases.
78 Kollerits B, Coassin S, Beckmann ND, et al. Genetic evidence for a
Hepatology 2018; 67: 328−57.
role of adiponutrin in the metabolism of apolipoprotein
54 Rosenberg WM, Voelker M, Thiel R, et al. Serum markers detect B-containing lipoproteins. Hum Mol Genet 2009; 18: 4669−76.
the presence of liver fibrosis: a cohort study. Gastroenterology 2004;
79 Fan Y, Lu H, Guo Y, et al. Hepatic transmembrane 6 superfamily
127: 1704−13.
member 2 regulates cholesterol metabolism in mice.
55 Ratziu V, Massard J, Charlotte F, et al. Diagnostic value of Gastroenterology 2016; 150: 1208−18.
biochemical markers (FibroTest-FibroSURE) for the prediction of
80 Musso G, Cipolla U, Cassader M, et al. TM6SF2 rs58542926 variant
liver fibrosis in patients with non-alcoholic fatty liver disease.
affects postprandial lipoprotein metabolism and glucose
BMC Gastroenterol 2006; 6: 6.
homeostasis in NAFLD. J Lipid Res 2017; 58: 1221−29.
56 Afdhal NH, Bacon BR, Patel K, et al. Accuracy of fibroscan,
81 Jo H, Choe SS, Shin KC, et al. Endoplasmic reticulum stress
compared with histology, in analysis of liver fibrosis in patients with
induces hepatic steatosis via increased expression of the hepatic
hepatitis B or C: a United States multicenter study.
very low-density lipoprotein receptor. Hepatology 2013; 57: 1366–77.
Clin Gastroenterol Hepatol 2015; 13: 772−79.
82 Männistö VT, Simonen M, Soininen P, et al. Lipoprotein subclass
57 Loomba R. Role of imaging-based biomarkers in NAFLD:
metabolism in nonalcoholic steatohepatitis. J Lipid Res 2014;
recent advances in clinical application and future research
55: 2676–84.
directions. J Hepatol 2018; 68: 296−304.
83 Lotta LA, Sharp SJ, Burgess S, et al. Association between
58 Banerjee R, Pavlides M, Tunnicliffe EM, et al. Multiparametric
low-density lipoprotein cholesterol lowering genetic variants and
magnetic resonance for the non-invasive diagnosis of liver disease.
risk of type 2 diabetes: a meta-analysis. JAMA 2016; 316: 1383–91.
J Hepatol 2014; 60: 69−77.
84 Stender S, Kozlitina J, Nordestgaard BG, Tybjærg-Hansen A,
59 Singh S, Allen AM, Wang Z, Prokop LJ, Murad MH, Loomba R.
Hobbs HH, Cohen JC. Adiposity amplifies the genetic risk of fatty
Fibrosis progression in nonalcoholic fatty liver vs nonalcoholic
liver disease conferred by multiple loci. Nat Genet 2017; 49: 842−47.
steatohepatitis: a systematic review and meta-analysis of
paired-biopsy studies. Clin Gastroenterol Hepatol 2015; 13: 643−54. 85 Neuschwander-Tetri BA, Loomba R, Sanyal AJ, et al.
Farnesoid X nuclear receptor ligand obeticholic acid for
60 McPherson S, Hardy T, Henderson E, Burt AD, Day CP,
non-cirrhotic, non-alcoholic steatohepatitis (FLINT): a multicentre,
Anstee QM. Evidence of NAFLD progression from steatosis to
randomised, placebo-controlled trial. Lancet 2015; 385: 956−65.
fibrosing-steatohepatitis using paired biopsies: implications for
prognosis and clinical management. J Hepatol 2015; 62: 1148−55. 86 Ratziu V, Harrison S, Francque SM, et al. Elafibranor, an agonist of
the peroxisome proliferator-activated receptor-α and -δ, induces
61 Bril F, Cusi K. Nonalcoholic fatty liver disease: the new
resolution of nonalcoholic steatohepatitis without fibrosis
complication of type 2 diabetes mellitus.
worsening. Gastroenterology 2016, 150: 1147−59.
Endocrinol Metab Clin North Am 2016; 45: 765−81.
87 Friedman SL, Ratziu V, Harrison SA, et al. A randomized,
62 Samuel VT, Shulman GI. Nonalcoholic fatty liver disease as a nexus
placebo-controlled trial of cenicriviroc for treatment of nonalcoholic
of metabolic and hepatic diseases. Cell Metab 2018; 27: 22−41.
steatohepatitis with fibrosis. Hepatology 2018; 67: 1754−67.

www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2 11

 Review

88 Ratziu V. A critical review of endpoints for non-cirrhotic NASH 109 Bril F, Sninsky JJ, Baca AM, et al. Hepatic steatosis and insulin
therapeutic trials. J Hepatol 2018; 68: 353−61. resistance, but not steatohepatitis, promote atherogenic
89 Cusi K. Treatment of patients with type 2 diabetes and non-alcoholic dyslipidemia in NAFLD. J Clin Endocrinol Metab 2016; 101: 644–52.
fatty liver disease: current approaches and future directions. 110 Lincoff AM, Wolski K, Nicholls SJ, Nissen SE. Pioglitazone and risk
Diabetologia 2016; 59: 1112–20. of cardiovascular events in patients with type 2 diabetes mellitus:
90 Younossi ZM, Loomba R, Rinella ME, et al. Current and future a meta-analysis of randomized trials. JAMA 2007; 298: 1180−88.
therapeutic regimens for non-alcoholic fatty liver disease (NAFLD) 111 Mazzone T, Meyer PM, Feinstein SB, et al. Effect of pioglitazone
and non-alcoholic steatohepatitis (NASH). Hepatology 2017; compared with glimepiride on carotid intima-media thickness in
published online Dec 9. DOI:10.1002/hep.29724. type 2 diabetes: a randomized trial. JAMA 2006; 296: 2572−81.
91 Townsend SA, Newsome PN. Review article: new treatments in 112 Nissen SE, Nicholls SJ, Wolski K, et al. Comparison of pioglitazone
non-alcoholic fatty liver disease. Aliment Pharmacol Ther 2017; vs glimepiride on progression of coronary atherosclerosis in
46: 494−507. patients with type 2 diabetes: the PERISCOPE randomized
92 Diehl AM, Day C. Cause, pathogenesis, and treatment of controlled trial. JAMA 2008; 299: 1561−73.
nonalcoholic steatohepatitis. N Engl J Med 2017; 377: 2063−72. 113 Dormandy JA, Charbonnel B, Eckland DJ, et al. Secondary
93 Cusi K. Role of obesity and lipotoxicity in the development of prevention of macrovascular events in patients with type 2 diabetes
nonalcoholic steatohepatitis: pathophysiology and clinical in the PROactive Study (PROspective pioglitAzone Clinical Trial In
implications. Gastroenterology 2012; 142: 711–25. macroVascular Events): a randomised controlled trial. Lancet 2005;
94 Rowe IA. Too much medicine: overdiagnosis and overtreatment of 366: 1279−89.
non-alcoholic fatty liver disease. Lancet Gastroenterol Hepatol 2018; 114 Zghebi SS, Steinke DT, Rutter MK, Emsley RA, Ashcroft DM.
3: 66−72. Comparative risk of major cardiovascular events associated with
95 Marchesini G, Petta S, Dalle Grave R. Diet, weight loss, and liver second-line antidiabetic treatments: a retrospective cohort study
health in nonalcoholic fatty liver disease: pathophysiology, evidence, using UK primary care data linked to hospitalization and mortality
and practice. Hepatology 2016; 63: 2032−43. records. Diabetes Obes Metab 2016; 18: 916−24.
96 Barb D, Portillo-Sanchez P, Cusi K. Pharmacological management 115 Strongman H, Christopher S, Majak M, et al. Pioglitazone and
of nonalcoholic fatty liver disease. Metabolism 2016; 65: 1183−95. cause-specific risk of mortality in patients with type 2 diabetes:
extended analysis from a European multidatabase cohort study.
97 Promrat K, Kleiner DE, Niemeier HM, et al. Randomized controlled
BMJ Open Diabetes Res Care 2018; 6: e000481.
trial testing the effects of weight loss on nonalcoholic
steatohepatitis. Hepatology 2010; 51: 121–29. 116 Vaccaro O, Masulli M, Nicolucci A, et al. Effects on the incidence of
cardiovascular events of the addition of pioglitazone versus
98 Vilar-Gomez E, Martinez-Perez Y, Calzadilla-Bertot L, et al.
sulfonylureas in patients with type 2 diabetes inadequately
Weight loss through lifestyle modification significantly reduces
controlled with metformin (TOSCA.IT): a randomised, multicentre
features of nonalcoholic steatohepatitis. Gastroenterology 2015;
trial. Lancet Diabetes Endocrinol 2017; 5: 887−97.
149: 367–78.
117 Balas B, Belfort R, Harrison SA, et al. Pioglitazone treatment
99 Goossens N, Hoshida Y, Song WM, et al. Nonalcoholic
increases whole body fat but not total body water in patients with
steatohepatitis is associated with increased mortality in obese
non-alcoholic steatohepatitis. J Hepatol 2007; 47: 565−70.
patients undergoing bariatric surgery. Clin Gastroenterol Hepatol
2016; 14: 1619−28. 118 Viscoli CM, Inzucchi SE, Young LH, et al. Pioglitazone and risk for
bone fracture: safety data from a randomized clinical trial.
100 Belfort R, Harrison SA, Brown K, et al. A placebo-controlled trial of
J Clin Endocrinol Metab 2017; 102: 914−22.
pioglitazone in subjects with nonalcoholic steatohepatitis.
N Engl J Med 2006; 355: 2297–307. 119 Clarke GD, Solis-Herrera C, Molina-Wilkins M, et al. Pioglitazone
improves left ventricular diastolic function in subjects with
101 Cusi K, Orsak B, Bril F, et al. Long-term pioglitazone treatment for
diabetes. Diabetes Care 2017; 40: 1530−36.
patients with nonalcoholic steatohepatitis and prediabetes or
type 2 diabetes mellitus: a randomized trial. Ann Intern Med 2016; 120 Lewis JD, Habel LA, Quesenberry CP, et al. Pioglitazone use and
165: 305–15. risk of bladder cancer and other common cancers in persons with
diabetes. JAMA 2015; 314: 265−77.
102 Aithal GP, Thomas JA, Kaye PV, et al. Randomized,
placebo-controlled trial of pioglitazone in nondiabetic subjects with 121 Davidson MB, Pan D. An updated meta-analysis of pioglitazone
nonalcoholic steatohepatitis. Gastroenterology 2008; 135: 1176–84. exposure and bladder cancer and comparison to the drug’s effect on
cardiovascular disease and non-alcoholic steatohepatitis.
103 Sanyal AJ, Chalasani N, Kowdley KV, et al. Pioglitazone, vitamin E,
Diabetes Res Clin Pract 2018; 135: 102−10.
or placebo for nonalcoholic steatohepatitis. N Engl J Med 2010;
362: 1675–85. 122 Armstrong MJ, Gaunt P, Aithal GP, et al. Liraglutide safety and
efficacy in patients with non-alcoholic steatohepatitis (LEAN):
104 Bril F, Kalavalapalli S, Clark VC, et al. Response to pioglitazone in
a multicentre, double-blind, randomised, placebo controlled
patients with nonalcoholic steatohepatitis with vs without
phase 2 study. Lancet 2016; 387: 679–90.
type 2 diabetes. Clin Gastroenterol Hepatol 2018; 16: 558–66.
123 Dhir G, Cusi K. Glucagon like peptide-1 receptor agonists for the
105 Musso G, Cassader M, Paschetta E, Gambino R. Thiazolidinediones
management of obesity and non-alcoholic fatty liver disease: a novel
and advanced liver fibrosis in nonalcoholic steatohepatitis:
therapeutic option. J Investig Med 2018; 66: 7−10.
a meta-analysis. JAMA Intern Med 2017; 177: 633−40.
124 Cefalu WT, Kaul S, Gerstein HC, et al. Cardiovascular outcomes
106 DeFronzo RA, Tripathy D, Schwenke DC, et al. Pioglitazone for
trials in type 2 diabetes: where do we go from here? Reflections
diabetes prevention in impaired glucose tolerance. N Engl J Med
from a Diabetes Care editors’ expert forum. Diabetes Care 2018;
2011; 364: 1104–15.
41: 14−31.
107 Kernan WN, Viscoli CM, Furie KL, et al. Pioglitazone after ischemic
125 DeFronzo RA, Norton L, Abdul-Ghani M. Renal, metabolic and
stroke or transient ischemic attack. N Engl J Med 2016; 374: 1321−31.
cardiovascular considerations of SGLT2 inhibition.
108 Yau H, Rivera K, Lomonaco R, Cusi K. The future of Nat Rev Nephrol 2017; 13: 11−26.
thiazolidinedione therapy in the management of type 2 diabetes
mellitus. Curr Diab Rep 2013; 13: 329−41. © 2018 Elsevier Ltd. All rights reserved.

12 www.thelancet.com/diabetes-endocrinology Published online August 30, 2018 http://dx.doi.org/10.1016/S2213-8587(18)30154-2