Definitions and history

Nonalcoholic fatty liver disease (NAFLD) is a clinico-pathological syndrome which covers a large gamut of hepatic and extra-hepatic manifestations and complications associated with and predisposing to either individual features of the metabolic syndrome (MetS) [typically type 2 diabetes (T2D) and obesity] or to the full-blown MetS [1–7]. The epidemiological surge of NAFLD is, at least in part, spurious, [i.e. fuelled by our ever increasing ability to discover this common condition in asymptomatic individuals as well as in cases of otherwise “cryptogenic” cirrhosis and hepatocellular carcinoma (HCC)] [8, 9] and, to a larger extent, due to (western-type) lifestyle and lifestyle changes (in developing countries) [10]. Irrespective of this academic contention (spurious phenomenon versus true increase), hard data support the notion that NAFLD accounts for > 30% of deaths due to liver disease and diabetes and ~8% of all-cause mortality in the USA [11].

In 1980, Ludwig from the Mayo Clinic, USA, in reporting on 20 patients whose liver biopsies exhibited fatty and necro-inflammatory changes, Mallory bodies, fibrosis and cirrhosis, coined a novel definition consisting of a double composite word: “nonalcoholic steatohepatitis” (NASH) [12]. However, liver disease in the obese had been addressed by Zelman in 1952; in 1977, Haller had already understood that “hepatic steatosis” was a feature of the MetS; what we now call “NASH” was first reported by Thaler from Wien in 1962; what we would now call “NASH-cirrhosis” was first reported in 1979 by Itoh in Japan; in 1988, Diehl defined NAFLD as an “alcohol-like liver disease in the non-alcoholic” and, in 1995, Lonardo et al, in reporting on what they had defined the “bright liver syndrome” highlighted the associations with gallstones and atherosclerosis [13, 14]. Analysis of the history of NAFLD research has shown that early progress in our understanding of the disease was fostered by individual researchers/groups. However, from 2000 onward, contributions of symposia and scientific societies became prominent [13]. Additional aspects of NAFLD’s history have been focused in their excellent article by Fouad et al. [15].

Aims

The idea behind this narrative review article, based on a 25-year experience in the study of this disease, is to provide a personal point of view regarding some selected topics of research on NAFLD. To this end, rather than following the “canonical pathway” in writing, which leads to diagnosis and management starting from epidemiology and pathogenesis, specific research questions will be addressed here. This article is not intended to be exhaustive of all the possible topics: to this end, reference should be made to excellent, more systematic reviews [16–19] and scientific society guidelines, which have been compared and annotated elsewhere [20, 21]. The present article is principally addressed to clinical and translational investigators but Health Authorities, Ph.D. students, nurses and pharmaceutical industries involved in the active research of innovative drug therapies may also find hints of potential interest here.

Pre-test probability of disease

Novel genetic risk scores (GRS) are increasingly being proposed and validated [26–28] to gauge the pre-test probability of NAFLD. However, the use of GRS has a much wider scope spanning further and beyond the simple pre-test assessment of disease [26–31] (Table 1).

This research pathway is predicted to gain increasing popularity, not only in paediatric practice, but also in adults, thanks to a growing availability, at cheaper costs, of commercial kits including genetic polymorphisms useful in evaluating the predisposition to developing NAFLD. Besides these tests, which will contribute to expanding our knowledge of the so called “genetic NAFLD”, other algorithms will have to be developed and validated to identify the common forms of “metabolic NAFLD”. The prototype of such tests, the fatty liver index, originally proposed by Bedogni et al. [32], should probably be reworked through inclusion of at least sex and reproductive status, which are two key modifiers of the risk of NAFLD [33]. In any case, direct confirmation of steatosis through imaging techniques, together with the exclusion of competing causes of (steatogenic) liver disease remain a key step in the diagnosis of NAFLD. The type of imaging technique [for example conventional ultrasonography with semi-quantitative indices versus controlled attenuation parameter (CAP) versus magnetic resonance (MR)] will vary as a function of the resources allocated with the more expensive (and accurate) MR imaging predicted to be used more in affluent countries than the cheaper and globally more available ultrasonographic techniques [34].

In conclusion, the most rational (non-invasive) diagnostic approach to NAFLD and its accurate classification in the individual patient are a definite research priority.

Risk of advanced/progressive disease

Collectively, evidence suggests that NASH, rather than simple steatosis, carries an increased risk of more rapidly progressing to advanced fibrosis/cirrhosis and developing HCC [19, 35]. However, some longitudinal studies have proven that NAFLD is not simply a dichotomous disease (NASH vs. non-NASH). For example, compared to simple steatosis at baseline, which has the lowest risk of disease progression, the presence of even minor degrees of liver injury and inflammation (and perhaps severe steatosis, although in the absence of NASH), is associated with an increased risk of developing significant fibrosis, often in parallel with the deterioration of metabolic comorbidities [36].

A systematic review of ten longitudinal histological studies showed, as early as 2009, that inflammation was the key predictor of eventual histological fibrosis progression and a rational therapeutic target in NASH [37]. More recently, a European study based on an exploration cohort of 140 individuals with suspected NAFLD and a separate validation cohort of 78 patients with biopsy-proven NASH found that liver fibrosis and NASH were strongly associated with increasing BMI and HOMA-IR [38]. A recent study, by reporting that obesity, either alone or added to alcohol consumption and T2D, is strongly associated with fibrosing liver disease, extends the importance of BMI beyond NAFLD [39]. Together, these studies suggest that fibrosis is the end-result of repeated bouts of hepatic necro-inflammation which, in their turn, are fuelled by metabolic derangements principally resulting from obesity.

Consistently, data from treatment trials support a strong association between histological resolution of steatohepatitis and improvement of fibrosis in NASH patients [40]. Histopathological classification of NASH has traditionally been based on Brunt’s criteria and the NAS by Kleiner et al. [41, 42]. The NAS, being quantifiable and relatively more reproducible than the diagnosis of definite NASH by Brunt, has largely been used as an endpoint in clinical trials, allowing to semi-quantitatively monitor the severity of NAFLD over time. However, studies suggest that these histopathological classification systems may have distinct clinic-pathological correlations.

Diagnosis of definite NASH according to Brunt’s histological criteria has been reported as a stronger predictor of metabolic abnormalities than the NAS score by Kleiner [43–45]. Moreover, studies have shown that–given that it was unable to predict fibrosis progression–NAS was not associated with adverse clinical outcomes in these patients [36].

The new and recently validated histopathological ‘Fatty Liver Inhibition of Progression’ algorithm and Steatosis, Activity, and Fibrosis (SAF) score have proven to be able to identify NASH patients with distinct clinical (higher prevalence of metabolic risk factors, more severe IR, and higher levels of aminotransferases) and biological (liver fibrosis) profiles of disease severity [38].

Indeed, solid evidence suggests that liver fibrosis dictates the long-term prognosis of NAFLD patients being significantly associated with overall, liver-related and also cardiovascular mortality [46–48].

The identification of subjects with NASH and/or fibrosis is a key topic in the management of NAFLD patients allowing for adequate treatment schedules and follow-up of these patients [49]. Liver biopsy is the reference standard for the diagnosis of liver steatosis with grading/staging of concurrent necro-inflammatory/ fibrotic changes; however, this invasive procedure may be associated with major complications and patient discomfort [50, 51]. In addition, it is exposed to the risk of sampling errors and diagnostic inaccuracies as a result of NASH being unevenly distributed throughout the liver tissue [52]. Therefore, it should be reserved only to selected patients at risk of progressive liver disease and, of course, cannot be adopted in large epidemiological or interventional studies. On these grounds the research on diagnostic tools for non-invasive and accurate identification of NASH and fibrosis staging is actively ongoing [50, 51, 53, 54].

Clinical scores used for detecting NASH have shown inconsistent results [55]. Biomarkers of fibrosis, such as the NAFLD fibrosis score (NFS), may be adopted in clinical practice to non-invasively screen subjects for the absence or presence of advanced fibrosis, but all of these perform sub-optimally [36, 51]. Subjects with suspected advanced fibrosis or indeterminate scoring should either undergo non-invasive assessment of liver fibrosis with sonoelastographic techniques or be evaluated for liver biopsy [36, 49, 51].

Ultrasound semi-quantitative scores, such as ultrasonographic fatty liver indicator (US-FLI), have shown to be sensitive for liver steatosis as low as 10% and correlated with anthropometric, metabolic and histological parameters of NAFLD [53, 56, 57]. A US-FLI score ≤ 4 is quite accurate in ruling out NASH but higher cut-offs performed sub-optimally in differentiating steatosis from NASH [56, 58]. Therefore, this cannot substitute liver biopsy but rather assist in identifying subjects at risk of progressive NASH in whom liver fibrosis severity must be assessed non-invasively by ultrasound-based elastography first and with liver biopsy, when indicated [53]. Studies evaluating the diagnostic performance of ultrasound semi-quantitative scores or new ultrasound quantitative techniques, including CAP and those more recently developed (e.g., backscatter analysis, acoustic structure quantification and speed of sound), coupled with clinical laboratory parameters for detecting NASH are eagerly awaited. MR elastography has shown good accuracy for detecting NASH but it is expensive and its availability is limited [50].

Sonoelastographic techniques, such as transient elastography performed by Fibroscan or acoustic radiation force impulse (ARFI), have shown good accuracy for estimating advanced fibrosis and cirrhosis (F3 and F4 stages), especially if coupled with clinical data, but their accuracy in detecting F1-F2 fibrosis is sub-optimal [50, 51, 59].

Emerging evidence suggests that certain patterns of expression of microRNA (e.g., miR-122, miR-30c, miR-27b/30) are significantly associated with NASH and/or fibrosis, representing ideal candidates to be integrated into prediction algorithms of NAFLD severity [60–62].

In conclusion, future studies should be aimed at evaluating cost-effectiveness and cost-utility ratios of genetic tests and GRSs that also interact with endocrine and metabolic features of the individual patient. Non-invasive diagnostic algorithms based on integrated imaging, liver biomarkers and genetic signature should be devised to accurately detect NASH and, especially, hepatic fibrosis and monitor its changes over time.

Ruling out competing causes of liver disease: how far should we go?

It is obviously important to carefully exclude those causes of steatosis which are amenable to specific treatment. This is the case, for example, of alcoholic liver disease, hepatitis C virus (HCV) infection and hypothyroidism. Abstinence remains the backbone of treatment of alcoholic liver disease [63]. Effective antiviral cures are available to eradicate HCV [64]. Thyroid replacement therapy has the potential to reverse the pathogenic mechanisms of NAFLD [65]. However, complicating both research and clinical practice, multi-causality may intervene in the development of NAFLD in the individual patient, either at baseline or in the course of follow-up.

Over time, NAFLD has variably been named diabetic hepatitis, alcohol-like liver disease in the non-alcoholic, bright liver syndrome, and insulin-resistance-associated hepatic iron overload [13]. However, given that consensus must be reached on how the diagnosis of NAFLD should be pursued in the individual patient, the very name of NAFLD should give way to more precise definitions, such as proposed by multiple Authors [66–68]. Additionally, in clinical practice and in clinical trials, primary NAFLD (i.e. associated with MetS) should be differentiated from NAFLD deemed to predispose to incident MetS and, ideally, from specific forms of genetic NAFLD and from forms owing to composite etiology (i.e. genetic-and-metabolic) as well as from all the innumerable forms of secondary NAFLD [69].

The rationale for distinguishing alcoholic fatty liver disease (AFLD) from NAFLD was first challenged by Völzke in 2012 [70]. For example, the liver in the obese closely mimics alcoholic liver disease histologically [71, 72]. Moreover, the distinction of AFLD from NAFLD is mainly based on specific pathogenic mechanisms. However, alcohol abuse and metabolic disorders associated with obesity often overlap in the same individual and deceptive differences in clinical presentation and outcome are possibly accounted for by selection bias [70]. AFLD and NAFLD also share genetic factors, behavioural/social underpinnings, as well as certain contributors to mortality [73, 74]. For example, in both conditions, disease is more prevalent in males [74] and both NAFLD [2, 75] and AFLD [76] exhibit prominent extra-hepatic/systemic features. As for the differences between AFLD and NAFLD it has been pinpointed that the advanced stages NAFLD usually present more in older age than AFLD [74]. Regarding laboratory features, the aspartate aminotransferase (AST)/alanine aminotransferase (ALT) < 1 ratio is more common in NAFLD than AFLD, provided that cirrhosis is absent [74]. To sum up, commonalities between AFLD and NAFLD outnumber distinctions, which supports a shift from artificial categories to a more general approach to fatty liver syndromes as “multi-factorial disorders” [70, 74]. This mutated paradigm is predicted to contribute to improving prevention schedules and assist in exploring more updated nomenclatures and innovative treatment strategies [74].

Academic attention has focused on the risk of NAFLD occurring secondary to rare diseases such as LAL but findings have been inconclusive so far [77].

Further to its etiologic definition, histological characterization of NAFLD should be specified whenever liver biopsy is deemed to be clinically indicated. If liver biopsy is not indicated, non-invasive indices e.g., NFS and fibrosis-4 [78] should be performed, alone or, preferentially, combined with the evaluation of liver stiffness obtained through elastographic (either sono- or MR-based) techniques.

Moreover, a complete work-up of concurrent diseases should be performed based on our present understanding of extra-hepatic manifestations and complications of NAFLD [2]. A proposal for a rational classification of NAFLD based on an extensive diagnostic evaluation of hepatic and extra-hepatic disease is illustrated as follows (Figure 1).

Display full size

Figure 1. 

Articulating the taxonomy of MAFLD. The “LDE” system. Based on archaic suggestions [66] and earlier attempts [67], an international panel including eminent world authorities in NAFLD, none excluded, renamed NAFLD as metabolic associated fatty liver disease (MAFLD) [79]. This proposal now needs to be further articulated in order to fully explore the complexity of disease in the individual patient. The LDE system has a simple structure consisting of a prefix (L for liver), a central section (D for determinants) and a suffix (E for extra-hepatic); Liver (L) - We propose that MAFLD should include a prefix illustrating what, in the individual patient, is known about liver involvement. Information regarding this may be achieved, based on clinical judgement, either with non-invasive or biopsy evaluation [36]. Steatosis and inflammation may be classified into mild, moderate or severe and fibrosis as absent (F0), mild (F1-F2) ore advanced (F3-F4). Individuals who have overt cirrhosis and/or HCC should be declared so; Determinants (D) - The importance of sex and reproductive status cannot be overemphasized and genetic determinants, although not requested in all patients with NAFLD (reviewed in 49), must be acknowledged whenever data are available (Table 1). The minimal endocrine assessment in those with NAFLD is to ascertain whether they have either T2D or IR (assessed with HOMA-IR) and whether T2D is associated or not with obesity (i.e. so called “diabesity” [4]. Given its distinct pathogenic features and the possibility to receive specific treatment, hypothyroidism should also be evaluated in all cases [65]. Other types of NAFLD secondary to other endocrine derangements, such as polycystic ovary syndrome (PCOS), growth hormone (GH) deficiency and hypogonadism, should always be declared when the diagnosis is clinically overt [80]; Extra-hepatic (E) - Finally, the systemic nature of NAFLD must be fully illustrated by the suffix “E”, which explores and synthesizes data on extra-hepatic manifestations of disease [2]; As an example, illustrating this proposed taxonomy, patient Ms. X.Y. might be declared to have MAFLD (L steatosis mild; inflammation absent; fibrosis absent; D post-menopausal; no SNP identified; associated with hypothyroidism the full-blown MetS; E arterial hypertension; medio-intimal carotid thickening; previous colon cancer). This taxonomy lends itself to being expressed as either a descriptive diagnosis or a score

Intermittent fasting (IF)

Conflicting with the large availability of foodstuffs in many countries worldwide, evolution has selected our ability to survive and to adapt to a food-deprived/fasted state [91]. IF defines recurring periods (e.g. 16h to 48h) with little or no energy intake, with intervening periods of normal food intake [91]. IF, together with periodic fasting (i.e. IF with periods of fasting or fasting-mimicking diets from 2 to ≥ 21 days), beneficially affect health and disease processes/states both in experimental conditions and in humans. These favorable effects, which are mediated by activation of adaptive cellular stress response, improved mitochondrial health, DNA repair, autophagy, stem cell-based regeneration as well as enduring metabolic effects, translate into increased longevity and protection from degenerative disease, cardiovascular disease and cancer [91]. IF also protects from the MetS and associated cardio-metabolic disorders [91]. The only study conducted in patients with NAFLD has shown that an alternate-day calorie restriction diet followed for 8 weeks–as compared to usual habit diet–was associated with beneficial effects on BMI, ALT, grade of hepatic steatosis at ultrasonography and fibrosis scores (assessed with ARFI shear wave elastography) with a good adherence level [92]. Further studies are needed to confirm and expand these results.

Endocrine NAFLD/NASH

Other examples of personalized management protocols may regard those forms of NAFLD which occur secondary to specific endocrine derangements [80]. Furthermore, a deeper understanding of NAFLD pathophysiology in these forms of NAFLD/NASH secondary to endocrine disorders may help in identifying novel therapeutic approaches [65]. For example, a key question is how we may succeed in uncoupling NAFLD from IR [93].

Anticoagulation

A growing number of patients with NAFLD and progressive liver disease (i.e. NASH-cirrhosis) will be candidates for anticoagulant therapy in the forthcoming years. This prediction is based on the notions that NAFLD/NASH, a globally epidemic condition, is independently associated with an increased risk of abnormalities of cardiac structure/function, including cardiac rhythm disorders such as atrial fibrillation (AF), and also with idiopathic venous thromboembolism (VTE) [94–100]. Accumulating “real world” data suggest that direct oral anticoagulants can be used in AF or VTE patients with advanced chronic liver disease showing similar efficacy and reduced bleeding complications compared to warfarin [99]. An association between inflammation, activation of the coagulation system and the development of hepatic fibrosis and portal hypertension in chronic liver disease has been disclosed by experimental and clinical studies [99, 101]. A protective effect of anticoagulant therapy on the complications of portal hypertension and fibrogenesis has been observed in small sample studies performed in patients with cirrhosis mainly of viral origin [99, 101].

By highlighting that the imbalance between pro-coagulant and anticoagulant activities can also lead to thrombosis, recent evidence has subverted the antique dogma that chronic liver disease represents an acquired bleeding disorder [102]. Stine et al. [103], were first in reporting on the increased risk of portal vein thrombosis in patients with cirrhosis owing to NASH. This is a population in which a more liberal use of oral anticoagulation for preventive and therapeutic purposes will have to be considered. However, given the complex interaction between NAFLD, obesity and body composition; and based on the potential for exercise to beneficially affect all three, additional research is needed to better define the causal role of each in contributing to the pro-thrombotic state of NAFLD in order to improve patient-oriented outcomes [104, 105].

In conclusion, future studies should better define NAFLD in the individual patient (and the taxonomy proposed in Figure 1 may be an example of how this aim may be achieved) in order to guide personalized management aimed at impacting on the liver itself, the disease complications as well as on cardiovascular outcome are eagerly awaited.

The efficacy and long-term feasibility of intermittent fasting need to be evaluated. Although maintenance of life-style changes remains a challenge over the long run, it is worth remembering that even transient remissions of steatosis may metabolically be beneficial [106].

Anticoagulation strategy in patients with cirrhosis is expected to reduce the risk of portal vein thrombosis but, among the bleeding risks, it will obviously be important considering variceal and non-variceal haemorrhage. Cost-utility and risk-benefit ratios of this strategy remain to be evaluated.

Chronic pulmonary disease

Epidemiological evidence suggests that chronic lung conditions, such as chronic obstructive pulmonary disease (COPD) and sleep apnoea syndrome, may bi-directionally be associated with either the MetS or its individual components (i.e. diabesity) [128–132].

Mantovani et al. [132], by meta-analyzing six observational studies globally enrolling more than 133, 000 individuals, three quarters of whom were Asians, more than a quarter with NAFLD, observed that NAFLD was significantly associated with reductions of lung volumes at baseline and that such reductions may also be a risk factor for incident NAFLD over follow-up. Not unexpectedly, NAFLD is a risk factor for the development of long-term cardiovascular events and mortality in patients with COPD [133].

As summarized elsewhere [134], studies have illustrated the range of similarities linking COPD and NAFLD which are common non-communicable diseases affected by a genetic predisposition and associated with an unhealthy lifestyle. Moreover, the liver and the lung also have similar vascular anatomy and immune-metabolic functions, such as antigen processing and regulation of energy homeostasis [134]. First-line treatment of COPD and NAFLD includes cessation of smoking, diet and physical activity [134].

In conclusion, COPD and NAFLD are prevalent systemic diseases with a high rate of cardio-nephro-metabolic and cancer co-morbidities and outcomes which, collectively, exact a high health and economic toll. How is it possible that NAFLD and COPD live together? Which is the egg and which is the chicken? Additional studies are warranted in individuals with COPD and NAFLD to address these research questions.

NAFLD and COVID-19

Clinical vignette

A 59-year old Caucasian man was admitted to our Medical COVID Unit owing to severe acute respiratory syndrome coronavirus 2 (SARS-Cov-2) pneumonia. Fever, dysgeusia, nausea, fatigue and polyarthralgia had begun on the 16th of March, a few days following a contact with a definite case of COVID-19. This patient’s laboratory profile exhibited IL-6 240.73 pg/mL (normal range 0-10), PCR 13.9 mg/dL (0–0.7), AST 357 U/L (1–37), ALT 354 U/L (1–40), GGT 210 U/L, (18–136), lactate dehydrogenase (LDH) 1242 U/L (208–378), creatine kinase (CK) 400 U/L (10–170), total bilirubin 0.96 mg/dL (0.16–1.10), alkaline phosphatase 123 U/L (38–126), procalcitonin 0.5 ng/mL (< 0.5), prothrombin time/international normalized ratio 1.28 (0.80–1.20). Competing causes of liver disease and bacterial pneumonia were ruled out. Chest X-rays showed bilateral interstitial pneumonia (Figure 2) and hemogasanalysis disclosed partial pressure of oxygen/fraction of inspired oxygen ratio = 266.

Display full size

Figure 2. 

Chest X-Ray in SARS-CoV-2. Chest X-Ray displays a typical bilateral ground glass pattern, which tends to affect the posterior segments of lower lobes, with a peripheral and sub-pleural distribution

On the 30th of March, owing to worsening respiratory insufficiency, the patient received non-invasive ventilatory support and was administered oral Hydroxychloroquine and intravenous Tocilizumab at standard doses. Concurrently, laboratory data as well as dyspnea and subjective general conditions improved. The patient was discharged from hospital on the 8th of April.