Mitochondria Matter: Systemic Aspects of Nonalcoholic Fatty Liver Disease (NAFLD) and Diagnostic Assessment of Liver Function by Stable Isotope Dynamic Breath Tests

The liver plays a key role in systemic metabolic processes, which include detoxification, synthesis, storage, and export of carbohydrates, lipids, and proteins. The raising trends of obesity and metabolic disorders worldwide is often associated with the nonalcoholic fatty liver disease (NAFLD), which has become the most frequent type of chronic liver disorder with risk of progression to cirrhosis and hepatocellular carcinoma. Liver mitochondria play a key role in degrading the pathways of carbohydrates, proteins, lipids, and xenobiotics, and to provide energy for the body cells. The morphological and functional integrity of mitochondria guarantee the proper functioning of β-oxidation of free fatty acids and of the tricarboxylic acid cycle. Evaluation of the liver in clinical medicine needs to be accurate in NAFLD patients and includes history, physical exam, imaging, and laboratory assays. Evaluation of mitochondrial function in chronic liver disease and NAFLD is now possible by novel diagnostic tools. “Dynamic” liver function tests include the breath test (BT) based on the use of substrates marked with the non-radioactive, naturally occurring stable isotope 13C. Hepatocellular metabolization of the substrate will generate 13CO2, which is excreted in breath and measured by mass spectrometry or infrared spectroscopy. Breath levels of 13CO2 are biomarkers of specific metabolic processes occurring in the hepatocyte cytosol, microsomes, and mitochondria. 13C-BTs explore distinct chronic liver diseases including simple liver steatosis, non-alcoholic steatohepatitis, liver fibrosis, cirrhosis, hepatocellular carcinoma, drug, and alcohol effects. In NAFLD, 13C-BT use substrates such as α-ketoisocaproic acid, methionine, and octanoic acid to assess mitochondrial oxidation capacity which can be impaired at an early stage of disease. 13C-BTs represent an indirect, cost-effective, and easy method to evaluate dynamic liver function. Further applications are expected in clinical medicine. In this review, we discuss the involvement of liver mitochondria in the progression of NAFLD, together with the role of 13C-BT in assessing mitochondrial function and its potential use in the prevention and management of NAFLD.


Introduction
The liver plays a key role in lipid homeostasis, with steps including the synthesis, oxidation, and transport of free fatty acids (FFA), triglycerides (TG), cholesterol, and bile acids (BA). Chronic liver diseases encompass a spectrum of conditions ranging from metabolic to viral, alcohol-related diseases, drug-related diseases, autoimmune diseases, and tumours. The hepatocyte can be damaged by various hits and intracellular organelles can be part of the dysfunctional cell, with changes including microsomal hypertrophy, mitochondrial damage by free fatty acids overload and insufficient β-oxidation and activation of peroxisomal metabolism.
Growing evidence points to dysfunctional mitochondria as key contributors in the pathogenesis of the chronic metabolic conditions (i.e., obesity, metabolic syndrome and type 2 diabetes mellitus) frequently linked to liver disease. These processes act through pathways leading to oxidative stress, chronic inflammation, and insulin resistance. Thus, diagnostic techniques able to early detect and monitor mitochondrial dysfunction have great relevance in terms of possible primary/secondary prevention measures and of therapies specifically targeting liver mitochondria [1][2][3].
In this paper we will focus on nonalcoholic fatty liver disease (NAFLD) with emphasis on mitochondrial dysfunction and the role of novel "dynamic" noninvasive breath test (BT) to assess mitochondrial function. We will also point to current potential therapeutic approaches targeting mitochondria in NAFLD.

General Features of Mitochondria
Mitochondria are intracellular organelles that provide energy for the body cells. In the liver there are about 500-4000 mitochondria per cell [4] equalling about 18% of the entire cell volume [5]. Mitochondria play a key role in the metabolic pathways and signalling networks [6]. They participate in degrading pathways of carbohydrates, proteins, lipids, and xenobiotics [7,8], and ultimately generate ATP as energetic source [6][7][8]. The morphological and functional integrity of mitochondria maintain functioning networks and pathways inside the mitochondria and in the cell. The fat and energy balance in hepatocytes is regulated by mitochondrial activities, including FFA β-oxidation, electron transfer and production of ATP and ROS [9]. Essential elements include mitochondrial DNA (mtDNA), membrane constituents, lipoprotein trafficking, pro-and anti-oxidant balance, and metabolic demand and supply [10].

The Fate of Free Fatty Acids
The routes providing the circulating (long-chain) FFA to the liver and their mitochondrial fate is of relevance for ATP production and hepatocyte health maintenance. About 60% of circulating FFA derive from lipolysis of TG in adipose tissue [11], and enter the hepatocyte by using the transporters FFA translocase/CD36, fatty acid binding protein (FABP), and caveolin-1. About 15% FFA are of dietary origin and contained in TG within ApoEenriched chylomicrons. Chylomicrons are assembled in the enterocyte following dietary fat digestion in the intestinal lumen. This step occurs during emulsion and micellization of fat by the bile acid (BA) pool made of both primary and secondary tauro-, glycol-conjugated BA [12,13]. In the hepatocyte chylomicron remnants bind specific membrane receptors taht have a high affinity for the surface protein ApoE. Lastly, about 35% of FFA in the hepatocyte originate from de novo lipogenesis (DNL) from dietary carbohydrates (glucose converted to pyruvate during glycolysis). The FFA pool in the hepatocyte provides the substrate for re-esterification with glycerol to form TG via the key enzymes, diglyceride acyltransferase (DGAT)1 and DGAT2. This amount of TG serves as stored fat as lipid droplets in small amounts, i.e., less than 5% of cell content. When needed, TG can be hydrolysed by hydrolases, e.g., the patatin-like phospholipase domain-containing protein 3-PNPLA3 (adiponutrin)-to release FFA in the cytosol [14,15]. The TG pool also provides FFA for two major routes of elimination [16]. (a) export to blood as TG within very-low density lipoproteins (VLDL) assembled in the endoplasmic reticulum [17]; (b) β-oxidation of FFA in mitochondria. In the Golgi apparatus the apolipoprotein B (ApoB) undergoes disulphide bond formation and association with TG (by protein disulphide isomerase and microsome triglyceride transfer protein (MTP)) [18]. Of note, increased intake of sucrose in the mice model leads to rapid development of hyperinsulinemia, hepatosteatosis, and insulin resistance. Furthermore, insulin enhances hepatic expression of the FA transporter CD36 involving a PPAR-γ-dependent mechanism. In the general scenario, these results indicate that hyperinsulinemia is an early and potent inducer of hepatosteatosis, insulin resistance, and dysglycaemia. A further step is the progression to type 2 diabetes and NAFLD. In addition, during conditions of hyperinsulinemia, dysfunctional insulin clearance becomes evident, due to abnormal insulin degrading enzyme regulation. This step, in turn, directly impairs postprandial hepatic glucose disposal and increases susceptibility to dysmetabolic conditions, including fatty liver, mitochondrial dysfunction, especially in the setting of Western diet/lifestyle.

β-Oxidation of FFA in Mitochondria
This important mitochondrial pathway includes FFA β-oxidation, the tricarboxylic acid cycle (TCA), electron flow along the electron transport chain, electrochemical proton gradient generation, and ATP synthesis. In starvation, ketone bodies are produced due to absence of oxaloacetate used in gluconeogenesis. Pyruvate can enter the mitochondrion via the mitochondrial pyruvate carrier (MPC) as well as be synthesised from L-lactate after transport of L-Lactate in the matrix, via its own carrier, and oxidation via the mitochondrial L-lactate dehydrogenase [19,20]. In the matrix, pyruvate can provide Acetyl-CoA via the pyruvate dehydrogenase complex and oxaloacetate (OAA) via the pyruvate carboxylase. Due to citrate synthase, pyruvate and oxaloacetate give citrate which is exported for FFA synthesis is the cytoplasm during DNL [21]. For fatty acid catabolism, the acyl-CoA synthase transforms the cytosolic FFA into fatty acyl-CoA ( Figure 1). Acyl-CoA+ carnitine are catalysed to CoA and acylcarnitine by the carnitine palmitoyl-transferase 1 (CPT-1) which is in the outer side of the inner mitochondrial membrane. Acylcarnitine can enter the mitochondria across the inner membrane in exchange with L-carnitine. This step requires the acylcarnitine/L-carnitine antiporter. The carnitine palmitoyl-transferase 2 (CPT-2), localized at the matrix side of the inner membrane, will process the acyl-carnitine to Acyl-CoA+ L-carnitine (the latter ready to be exchanged with new incoming Acyl-carnitine). The resulting acyl-CoA in the mitochondrial matrix is ultimately oxidised via the β-oxidation to acetyl-CoA which then enters the tricarboxylic acid (TCA) cycle with production of carbon dioxide and water. A further step implies the activation of the electron transport chain and ATP production. Oxidative metabolism and hepatocyte energy homeostasis depend on FFA βoxidation, the tricarboxylic acid cycle (TCA), electron flow along the electron transport chain, electrochemical proton gradient generation, and ATP synthesis. Ketone bodies are produced due to an absence of oxaloacetate used in gluconeogenesis (e.g., starvation and diabetes). Starting from blood, dietary glucose, dietary FFA (as TG within chylomicrons) and FFA circulating after TG lipolysis, enrich the FFA pool in the hepatocyte. During fatty acid synthesis, glucose from dietary sources during glycolysis is converted to pyruvate which can enter the mitochondrion via the mitochondrial pyruvate carrier (MPC). Pyruvate also can be synthesised from L-lactate after transport of L-lactate in the matrix, via its own carrier, and oxidation via the mitochondrial L-lactate dehydrogenase [19,20]. In the mitochondrial matrix, pyruvate provides acetyl-CoA via the pyruvate dehydrogenase complex and oxaloacetate (OAA) involving the pyruvate carboxylase. Due to citrate synthase, pyruvate and oxaloacetate give citrate which can be exported to allow for FFA synthesis is the cytoplasm in the de novo lipogenesis (DNL). Abbreviations: ACC, acetyl-CoA carboxylase (ACC); ANT, adenine nucleotide translocator; CACT, carnitine-Acylcarnitine Transferase; CPT-1, carnitine palmitoyltransferase-1; CPT-2, carnitine palmitoyltransferase-2; DNL, de novo lipogenesis; electron transfer flavoprotein (ETF); ETFDH, ETF dehydrogenase; FFA, free fatty acids; β-HB, β-hydroxybutyrate; MPC, mitochondrial pyruvate carrier; OAA, oxaloacetate; PEP, phosphoenolpyruvate; TG, triglycerides; VLDL, very low-density lipoprotein [16,21].

Definition
The term nonalcoholic fatty liver disease (NAFLD) points to the deposition of excess TG as lipid droplets in the cytoplasm of hepatocytes. Steatosis is defined as a hepatic TG level exceeding the 95th percentile for lean, healthy individuals (i.e., >55 mg per g of liver), histologically defined when 5% or more of the hepatocytes contain visible intracellular triglycerides [22,23] or the estimated liver fat content is ≥5% by a magnetic resonance imaging proton density fat fraction (MRI-PDFF) or ≥5.56% by magnetic resonance spectroscopy [24].
NAFLD has become the leading liver disease worldwide with an estimated 2 billion individuals affected [25]. NAFLD represents a spectrum of disease that may develop in individuals without significant alcohol consumption [26] and ranges from steatosis to steatohepatitis. Nonalcoholic fatty liver (NAFL), featuring simple steatosis, with little or no inflammation and no evidence of hepatocellular injury, affects about 80% of NAFLD subjects and is the non-progressive form since the risk of progression to liver cirrhosis is minimal [27]. About 20% of NAFLD manifests as nonalcoholic steatohepatitis (NASH), featuring steatosis, inflammation, and hepatocellular injury with ballooning and apoptosis.
Although NAFLD is by far the most prevalent cause of liver steatosis, ectopic fat accumulation may occur in the liver for a variety of reasons in viral hepatitis B and C (in particular genotype 3), lipodystrophy, Wilson's disease, starvation, parenteral nutrition, abetalipoproteinemia, hepatotoxic drugs (e.g., methotrexate, tamoxifen, glucocorticoids, amiodarone, valproate, and anti-retroviral agents for HIV), pregnancy, HELLP (hemolytic anemia, elevated liver enzymes, low platelet count) syndrome, Reye syndrome, and inborn errors of metabolism (i.e., lecithin-cholesterol acyltransferase deficiency, cholesterol ester storage disease, and Wolman disease). However, alcohol-associated liver injury remains the second most frequent aetiology of steatosis. The similarity and overlap between alcoholassociated liver disease and NAFLD has been the source of confusion and the subject of academic debate. The term "non-alcoholic" indeed overemphasizes "alcohol" and underemphasizes the role of metabolic risk factors, since NAFLD is commonly associated with obesity, hypertension, dyslipidaemia, and diabetes [17,26]. To acknowledge that NAFLD is no longer a diagnosis of exclusion, and it represents a continuum of liver disease caused by metabolic derangements, a change in terminology from NAFLD to metabolic dysfunction-associated fatty liver disease (MAFLD) has been recently proposed. Accordingly, hepatic steatosis is associated with at least one of the following three comorbidities: overweight/obesity (especially expansion of visceral fat), presence of type 2 diabetes mellitus, or evidence of metabolic dysregulation [35]. Nevertheless, some authors warned that understanding of the molecular basis of the disease entity, new insights in risk stratification, and other important aspects of NAFLD may be more urgent than nosology itself [36]. Indeed, there remains much to learn about the contribution of environment, comorbidities and the gut microbiome to the pathogenesis and natural history of NAFLD [16,[37][38][39].

Prevalence and Natural History
NAFLD has become the most frequent liver disorder of our times [22,[40][41][42]. The median prevalence of NAFLD is about 25% worldwide and trends are increasing [36,43,44]. This is likely due to the increasing prevalence of obesity, type 2 diabetes mellitus, sedentary lifestyles, dyslipidemia, and metabolic syndrome, mainly in North America and Europe [43,[45][46][47]. However, the burden of NAFLD has also become evident in non-obese individuals ('lean NAFLD'), with a prevalence of about 10%-30% in both Western and Eastern countries [48], typically associated with metabolic dysfunction and a comparatively increased cardiovascular risk [46,49]. NAFLD puts the population at increased risk for liver-related mortality as well as all-cause-mortality due to increased risk of cardiovascular disease and extrahepatic malignancies [50][51][52]. Liver fibrosis is currently the strongest known predictor of poor clinical outcomes in NAFLD. The time sequence of fibrosis progression in NAFL is significantly slower (average 14 years) than in NASH (about 7 years) and even less in a subgroup of 'rapid progressors encompassing 10% to 20% of patients with NAFLD [27]. Thus, much attention has been devoted to the identification of predictors of rapid progression (i.e., higher serum ALT, morbid obesity, diabetes, and possibly genetic susceptibility with family history of cirrhosis in first-degree relatives) [53][54][55]. Once cirrhosis has developed in NAFLD, the incident risk of developing HCC is about 1.5%-2% per year. Therefore, HCC screening in NASH-related cirrhosis is recommended [56]. NAFLD is now the second leading indication for liver transplantation in the US, including a growing number of cases with NASH-related HCC [44].

Diagnosis
Liver biopsy followed by liver histology is the gold standard for diagnosing NAFLD. The procedure is usually echo-assisted and performed by transcutaneous puncture of the liver after local anaesthesia. A cylindric liver fragment is promptly placed in a solution containing formalin. The procedure, however, is invasive, and exposes patients to the risk of potential complications. The compliance of the patients is therefore very low. Liver biopsy should be reserved to subgroup of patients with suggestive signs/symptoms/evidence of steatohepatitis or early cirrhosis and when careful histological assessment is required to quantify the degree and stage of liver damage as fibrosis, inflammation, and necrosis, or during research protocols looking at the progression of liver fibrosis and efficacy of specific therapies.
Therefore, in clinical practice, the diagnosis of liver diseases relies on a history, physical exam and tests that investigate morphological and functional aspects. The liver is essential for many metabolic and energetic processes in the body and there is no single test that could assess liver function in a comprehensive way. Each test provides a specific set of information focusing on various mechanisms involved in liver function. A major challenge in clinical hepatology is therefore to appropriately combine the results of diagnostic tests in an accurate and complementary way to achieve the final diagnosis.
By history, NAFLD patients often carry one or more components of metabolic syndrome or "fellow travellers", such as cholesterol cholelithiasis [57,58]. Other causes of liver steatosis and chronic liver diseases must be therefore excluded [59]. In NAFLD, alcohol consumption should be absent or very limited. This includes not more than three standard drinks/day (i.e., 21 drinks/week) in men or not more than 2 drinks/day in women (i.e., 14 drinks/week, equal to 14 g of pure alcohol/standard drink = 98 kcal), as indicated by the American Association for the Study of Liver Diseases [26]. In fact, alcohol consumption greater than the threshold puts individuals at risk of alcoholic liver disease eventually associated with coexisting NAFLD. This situation makes the diagnosis of NAFLD and risk assessment even more difficult.
Laboratory tests include serum aminotransferase levels as markers of hepatocyte cytolysis, but they are not sufficient for making the diagnosis, as laboratory tests may be normal in patients with NAFLD and may be abnormal in patients with many other conditions. Serum alanine aminotransferase (ALT) in NAFLD is typically higher than serum aspartate aminotransferase (AST) unless the disease has already progressed [60]. Serum tests to assess for other disorders include viral hepatitis serology, iron studies, and autoimmune antibody assays. Some of these 'static' tests measure serum parameters of synthesis (prothrombin, cholesterol, albumin), hepatocellular injury (transaminases), detoxification (ammonium), excretion and cholestasis (bilirubin, alkaline phosphatase, GGT) [61].
Imaging techniques in the evaluation of NAFLD include abdominal ultrasonography, computerized tomography (CT), and magnetic resonance.
Abdominal ultrasonography can easily detect a hyperechoic texture in the liver ("bright liver") due to diffuse fatty infiltration. The main advantages of ultrasound include wide availability, safety, and low-cost. This non-invasive technique can easily allow a screening of patients at risk and is a useful tool for monitoring treated patients. However, liver ultrasound is not able to distinguish the necro-inflammatory changes typical of steatohepatitis, and has a poor accuracy in diagnosing the presence of a mild steatosis (i.e., <30%) [62]. Therefore, the ultimate diagnosis of both NASH and NAFLD can be underestimated.
Computed tomography can assess the liver brightness, measuring pixel values in Hounsfield Unit with quantitative determination of attenuation in comparison with the fat-free spleen [63,64]. The possibility of quantitative results is the main advantage of this imaging technique. However, as for ultrasound, the diagnostic accuracy of liver CT decreases with lesser severity of steatosis, with a sensitivity of 52-62% in case of mild steatosis (i.e., fat fraction of 10-20%) [65].
MR-based methods including proton spectroscopy and calculation of the protondensity fat fraction (PDFF) are far superior to ultrasound or CT in measuring intrahepatic fat content but, as for ultrasound and computed tomography, cannot distinguish between simple steatosis and steatohepatitis. PDFF measure, however, represents an advantage, as compared with CT, since it requires no internal calibration or reference standard. Advanced MR techniques can also consider confounders as iron overload, and can easily and rapidly allow a volumetric assessment of NAFLD [66].
By contrast, there has been significant progress in the non-invasive assessment of fibrosis in NAFLD. Vibration-controlled transient elastography is increasingly used as a point-of-care method to assess and regularly monitor fibrosis based on the liver stiffness and can also be utilized to grade hepatic steatosis. While there are additional ultrasoundbased liver stiffness measurement techniques, MR elastography has proven more accurate although this method currently remains primarily in the realm of research and clinical trials due to its significant cost.
Of note, none of the imaging techniques employed to diagnose NAFLD will explore the true "dynamic" liver function and need to be integrated with further "functional" examination techniques, such as breath test.
The management of NAFLD is still a matter of debate. According to AASLD guidelines, systematic screening for NAFLD is not advisable at this time, since there is no consensus about the true cost-effectiveness of the screening [16,26]. In addition, there is no licensed or registered pharmacotherapy for NAFLD and management remains focused on healthy lifestyles as previously discussed by our group [45,46,67]. Early identification of risk factors associated with NAFLD progression is therefore paramount to delay or prevent the consequences related to advanced liver disease. However, reliable, and sensitive non-invasive diagnostic tests are still lacking in NAFLD and are actively being investigated. In this respect, diagnostic tests focusing on mitochondrial function may provide novel diagnostic and prognostic possibilities both during the evolution of disease and in therapeutical trials. These aspects are discussed in the following sections.

Mitochondrial Dysfunction in the Liver
Mitochondrial dysfunction is one of the most distinctive characteristics of NAFLD [68]. In NAFLD patients, increased plasma levels of FFA are firstly associated with increased intrahepatic inflow [24] and early mitochondrial biogenesis through peroxisome proliferatoractivated receptor-α (PGC1-α) activation. This step, in turn, leads to increased FFA oxidation rates and increased or unchanged mitochondrial function [69]. Coupling of FFA oxidation to ATP generation might be dysfunctional already, because of emerging ultrastructural changes and increased expression of uncoupling proteins. With progression of NAFLD, however, mitochondrial ATP generation is further impaired resulting in defective cellular energy charge [70][71][72][73]. The precise pathways governing such changes of mitochondrial performance are still unknown.
The increased accumulation of FA in the hepatocytes (neutral lipid droplets) during insulin-resistance-associated NAFLD, which is pathologically defined as hepatic steatosis, lead to a series of mitochondrial alterations ranging between mitochondrial DNA (mtDNA) damage to sirtuin alteration. The mtDNA, a circular double-stranded molecule located in the mitochondrial matrix, encodes about the 10% of mitochondrial proteins, the others being encoded by the nuclear DNA. mtDNA encodes proteins necessary for the assembly and activity of mitochondrial respiratory complexes [74]. Ongoing oxidative stress during steatosis can severely impair mtDNA function [5] with further amplification of oxidative stress, mitochondrial biogenesis, and ultimately NAFLD severity and inflammation [75][76][77][78].
Alteration of the mitochondrial function compromises also the prooxidant/antioxidant balance, with an increase in non-metabolized fatty acids (FA) in the cytosol as a consequence of the blockade of FFA β-oxidation and the resulting stimulation of ROS production [79,80]. Mitochondrial dysfunctions are often accompanied by considerable ultrastructural changes such as megamitochondria, loss of cristae, and formation of paracrystalline inclusion bodies in the organelle matrix [81].
In addition, in NAFLD, the excessive accumulation of lipotoxic lipids in the hepatocyte generates a dysfunctional electron transfer chain with generation of abnormal levels of ROS via involvement of glycerol 3-phosphate dehydrogenase (GPDH), α-ketoglutarate dehydrogenase (AKGDH), and pyruvate dehydrogenase (PDH). Besides, the excessive accumulation of FFA into mitochondria, subsequent to an increased uptake or an insulinresistance situation, may elicit an increase of the inner mitochondrial membrane permeability. Mitochondrial cytochrome P450 2E1 (CYP2E1), a potential direct source of ROS, has been shown to have an increased activity in a rodent model of NASH as well as in NASH patients [82,83]. CYP2E1, a cytochrome responsible for long-chain fatty acid metabolism, produces oxidative radicals and could also act as a part of the "second hit" of the pathophysiological mechanism of NAFLD [84]. In addition to the pro-oxidant mechanism, a decreased activity of several detoxifying enzymes was seen using an experimental model of NASH. Glutathione peroxidase (GPx) activity is reduced likely due to GSH depletion and impaired transport of cytosolic GSH into the mitochondrial matrix [85]. The initial mitochondrial dysfunction can be further exacerbated by the production of mtDNA mutation by ROS and highly reactive aldehydes, such as malondialdehyde (MDA) and 4-hydroxy-2-nonenal (4-HNE), through lipid peroxidation following the interaction between ROS and PUFA. Cytochrome C oxidase may be directly blocked by MDA while 4-HNE may contribute to "electron leakage" uncoupling complex 2 of the ECT whose oxidative capacity may be also diminished by derivative damage by interaction between mitochondrial membranes and both MDA and 4-HNE [86].
Aquaporin-8 (AQP8), a pleiotropic aquaporin channel [87][88][89] allowing movement of hydrogen peroxide in addition to water and ammonia, localized at multiple subcellular levels in hepatocytes [90], is also present in mitochondria, where it has been suggested to facilitate the release of hydrogen peroxide across the inner mitochondrial membrane following ROS production [91].
Mitochondrial redox imbalance and high Ca 2+ uptake have been shown to induce the opening of the permeability transition pore (PTP) with consequent disruption of energylinked mitochondrial functions and triggering of cell death in many disease states including non-alcoholic fatty liver disorders [92].
In previous studies, we used the rat model of a choline-deprived diet for 30 days inducing simple liver steatosis. In particular, peroxidation of the membrane lipid components participates in mechanisms of oxygen-free radical toxicity [93]. Cardiolipin is a phospholipid localized almost exclusively within the inner mitochondrial membrane close complexes I and III of the mitochondrial respiratory chain. Notably, cardiolipin becomes an early target of oxygen-free radical attack, a step leading to deranged mitochondrial bioenergetics. In a first study, we assessed various parameters related to mitochondrial function such as complex I activity, oxygen consumption, reactive oxygen species (ROS) generation and cardiolipin content and oxidation. Complex I decreased by 35% in mitochondria isolated from steatotic livers, compared with the controls, and changes were associated with parallel changes in state 3 respiration. At the same time, hydrogen peroxide (H2O2) generation increased significantly in mitochondria. The mitochondrial content of cardiolipin, a phospholipid required for optimal activity of complex I, decreased by 38% in parallel with an increase in the level of peroxidised cardiolipin. Data confirm that dietary steatosis induces mitochondrial dysfunction revealed by deranged complex I function attributed to ROS-induced cardiolipin oxidation and function [94]. A putative scenario of damage is depicted in Figure 2. Putative mechanisms of damage involving cardiolipin in liver mitochondria during liver steatosis. Following several predisposing factors, liver steatosis develops. The increased production of ROS is associated with mitochondrial cardiolipin oxidation, defective complex 1, and furthers mitochondrial and hepatocyte dysfunction. Cardiolipin is a phospholipid localized almost exclusively within the inner mitochondrial membrane close to complexes I and III of the mitochondrial respiratory chain. Mechanisms of damage have been elucidated in the study by Petrosillo et al., using the rodent model fed a choline-deficient diet to induce simple liver steatosis [94].
In addition, using the choline-deficient steatogenic diet in the rat model, we measured the circulating and hepatic redox active and nitrogen-regulating molecules thioredoxin, glutathione, protein thiols (PSH), mixed disulphides (PSSG), NO metabolites nitrosothiols, nitrite plus nitrate (NOx), and lipid peroxides (TBARs). The histologically proven hepatocellular steatosis (75% of liver weight at day 30) was paralleled by increased serum and hepatic TBARs (r = 0.87, p < 0.001) and lipid content (r = 0.90, p < 0.001). Liver glutathione and thioredoxin 1 initially increased and then decreased, while, from Day 14, PSH decreased, and NO derivatives increased. Mitochondrial nitrosothiols were inversely related to thioredoxin 2. These results suggest that adipocytic transformation of hepatocytes is accompanied by major interrelated modifications of redox parameters and NO metabolism, especially at the mitochondrial level, suggesting an early adaptive protective response but also an increased predisposition towards pro-oxidant insults [95].
The combination of these events explains how mitochondrial dysfunction becomes a key step paving the way to cells and organ damage. The main events in this scenario include the lack of energy supply by ATP and excessive generation of ROS. In case of prolonged starvation or diabetes, for example, ketone body synthesis occurs, when oxaloacetate is depleted due to its involvement in gluconeogenesis. In this scenario in the mitochondria, the acetyl-CoA does not enter the TCA cycle, and is converted to ketone bodies (i.e., acetone, acetoacetate, and β-hydroxybutyrate (β-HB)). Some metabolic markers might appear in the systemic circulation, for example with an abnormal acetoacetate/β-OHbutyrate ratio [9,96], but they cannot be easily monitored and are rather unspecific. In type 2 diabetes mellitus and NAFLD, the hepatic mitochondrial metabolism is impaired [97,98], associated with remodelling of mitochondrial lipids [99], and increased mitochondrial mass and respiratory capacity [100]. Lipotoxicity can influence acetyl-CoA metabolism [101] with excessive turnover of the tricarboxylic acid cycle [97]. In the steatogenic model of cultured hepatocytes, the combination of fructose and FFA caused profound effects on the lipogenic pathways. We noticed increased steatosis and reduced cell viability, increased apoptosis, oxidative stress and, mitochondrial respiration in the Seahorse system. Hepatic cell abnormalities can be prevented, and in this model, the damage improved by treating the cells with the nutraceutical silybin [102].
Mitochondrial dysfunction is associated, in NASH, with the ongoing oxidative state of hepatocytes, and is able to affect intracellular signalling pathways by generation of DAMPs and to activate stellate cell [103].
A recent in vitro study demonstrated that circulating factors contained in plasma samples from NAFLD patients were able to generate a NAFLD-like phenotype in isolated hepatocytes, with effects mediated by NLRP3-inflammasome pathways and by the activation of intracellular signalling related to SREBP-1c, PPAR-γ, NF-kB and NOX2 [104].
Besides external conditions affecting the metabolic homeostasis and mitochondrial function, genetically driven conditions can also lead to altered hepatic mitochondrial activity and peroxisomal β-oxidation [105][106][107]. In particular, the membrane bound Oacyltransferase domain containing 7-trans-membrane channel-like 4 (MBOAT7-TMC4) is localized to the intracellular membranes of mitochondria, endoplasmic reticulum, and lipid droplets. MBOAT7-TMC4 acts as lysophospholipid acyltransferase, and regulates the incorporation of arachidonic acid into phosphatidylinositol [108]. This pathway, due to its key role, might be considered a promising therapeutic target. The expression of MBOAT7-TMC4 is decreased in the rs641738 polymorphism, and this leads to the onset of liver steatosis and to an altered liver histology [109,110], to fibrosis in alcoholic liver disease [111] and in chronic hepatitis C [112]. In a mice model of NASH, the deletion of hepatocyte Mboat7 is linked to with increased fibrosis, with no effects on inflammation [113]. Mboat7 also promotes the degradation of lysophosphatidylinositol, and the accumulation of this molecule in Mboat7 KO mice generates NASH [114], also through the activation of the G-protein coupled receptor GPR55 [115]. Aging processes is associated with altered subcutaneous adipose tissue function, with mechanisms that involve a reduced mitochondrial activity [116,117], the accumulation of senescent adipocytes, and impaired development of pre-adipocytes [118]. Liver mitochondria also play a relevant role in lipidinduced hepatic insulin resistance, through mechanisms linking specific lipid metabolites and cellular compartments and leading to subcellular dysfunctions [119]. In particular, the quantitative assessment of DAG stereoisomers (sn-1,2-DAGs, sn-2,3-DAGs, and sn-1,3-DAGs) and ceramides in the endoplasmic reticulum, mitochondria, plasma membrane, lipid droplets, and cytosol showed, using an antisense oligonucleotide, the onset of hepatic insulin resistance in rats, which was associated with the acute liver-specific knockdown of diacylglycerol acyltransferase-2. The dysregulation of peroxisome proliferator-activated receptor-gamma co-activator-1α (PGC-1α) contributes to the pathogenesis and to the sequence of NASH-HCC, with metabolic pathways involving gluconeogenesis, fatty acid oxidation, antioxidant response, DNL, and mitochondrial biogenesis [120].
Experimental data indicate that mitochondrial dysfunction is also a specific target for toxic chemicals of environmental origin mainly introduced bycontaminated food and water and leading to NAFLD.
A recent study in 2446 young adults showed that toenail cadmium concentration, a marker of long-term exposure, was associated with higher odds of prevalent NAFLD independently from race, sex, BMI or smoking status [121]. In a mouse model of chronic cadmium exposure, hepatic Cd concentrations ranging from 0.95 to 6.04 µg/g wet weight were able to induce, following a 20-week exposure, NAFLD and NASH like phenotypes linked with mitochondrial dysfunction, fatty acid oxidation deficiency and a significant suppression of sirtuin 1 signalling pathway [122]. Epidemiologic studies point to a positive association between arsenic exposure (i.e., urinary arsenic concentrations) and risk of NAFLD [123]. This evidence is paralleled by experimental findings showing, in isolated rat liver mitochondria exposed to arsenic, a marked decrease in total mitochondrial dehydrogenase activity with increased ROS generation, MMP, and MDA levels, and decreased activity of mitochondrial catalase and GSH [124].
In a cohort of 6389 adolescents from the NHANES survey, blood mercury levels were linked with the risk of NAFLD, with the most evident association in underweight or normal weight subjects [125]. In a recent animal model, exposure to methylmercury during 12 weeks induced mitochondrial swelling, ROS overproduction, increased gluthatione oxidation, and reduced protein thiol content [126].
There are few ways to investigate mitochondrial metabolic processes, i.e., using isolated organelles, mitochondrial fractions, and cell culture [136]. Few studies explored the impaired mitochondrial function in NAFLD. Protocols investigating the effects of xenobiotics and drugs on mitochondrial function can provide some information [137,138]. Metabolomics can also explore specific mitochondrial functions [139] by studying genetic perturbations [140]. The measure of circulating mitochondrial DNA (mtDNA) is another biomarker of mitochondrial dysfunction. Changes to liver mitochondrial DNA (mtDNA) can precede mitochondrial dysfunction and irreversible liver damage. Malik et al. [141] by using a rodent dietary approach, demonstrated that a high-fat or a high-fat/high-sugar diet for 16 weeks was associated with fast alterations in mtDNA. Thus, dietary changes in liver mtDNA can occur in a relatively short time. Mouse liver contained a high mtDNA content (3617 +/− 233 copies per cell), which significantly increased when the mice were fed an HFD diet. This increase, however, was not functional; i.e., it was not translated into an increased expression of mitochondrial proteins. Furthermore, liver dysfunction was accelerated alongside the downregulation of mitochondrial oxidative phosphorylation (OXPHOS) and mtDNA replication machinery as well as upregulation of the mtDNAinduced inflammatory pathways.

Studying Liver Mitochondrial Function at a Translational Level
Strategies to diagnose mitochondrial damage and afterwards to prevent progression by therapy in NAFLD are actively being investigated. All the above-mentioned procedures focusing on mitochondrial function, however, lack true translational value, can be complex, expensive, and make the comparison between different models and in vivo studies somewhat difficult. A major problem is that mitochondria function at the crossroads of several complex metabolic processes, which can be influenced by unknown precursors affecting metabolic pathways.
Thus, no specific, easily available test provides information on liver mitochondrial status in clinical medicine, and we miss reliable biomarkers that inform on hepatocyte mitochondrial function and fitness in the NAFLD. Further studies and novel diagnostic tools are required in this field. The following paragraphs will focus on the use of breath tests by stable isotopes to explore liver mitochondrial function.

General Features of BT
BT represent an expanding field in diagnosis of liver function and are "dynamic" tools dealing with distinct functional aspects of liver metabolism. BTs can be employed for the follow-up of liver disease, including mitochondrial function in NAFLD [96]. Liver BT can provide information about the complex metabolic function of the liver by marking a given substrate with the stable isotope 13 C, measured in breath as 13 CO 2 . Our group has provided studies dealing with evaluation of both mitochondrial and microsomal activity in the liver [36,96,[142][143][144][145][146][147][148][149]. The rationale of BT depends on a given substrate that is metabolized at different levels in the body. The metabolized substrate produces gases (e.g., CO 2 , H2) transferred to blood, excreted, and quantified in expired air. Various sensors can detect the end-product in breath. The measured metabolite becomes the biomarker of a specific metabolic process [150]. Examples of BTs include the urea BT for the diagnosis of infection by H. pylori in the stomach, the hydrogen breath test for the diagnosis of lactose intolerance, or the study of small intestinal bacterial overgrowth [151][152][153][154]. A few BTs are relatively simple to perform, safe, and non-invasive, with potential applications in several conditions. Liver BTs are used to assess the hepatocyte capacity to metabolize a substrate in a time-dependent way [144,155]. Few BT have been developed and employ substrates labelled with the stable isotope (non-radioactive), naturally occurring 13 C marking one specific carbon atom in the substratum. The essential characteristics of 13 C-BT when assessing liver function are depicted in Table 1 [143,156]. Table 1. General characteristics of an ideal substrate for studying dynamic liver function [143].

Pharmacokinetic and metabolic aspects
Rapidly and consistently absorbed by oral route Primary liver metabolization Low hepatic extraction ratio (20-30%) (i.e., metabolism independent from liver blood flow) Clear metabolic pathway Simple pharmacokinetic Short elimination half-life Minimal compartmentalization of generated 13 CO 2 Early appearance of 13 CO 2 in breath Methodological aspects Safe, without side effects Test simple to prepare and administer No (or minimal) interaction with extra-hepatic tissues (i.e., adipose tissue, muscle) Reproducible over time Repeatable (useful for follow-up) Costs Affordable A single functional test cannot explore the whole liver function, since liver metabolic pathways are characterized by intrinsic complexity in terms of uptake, site of metabolization, and pathways involved during the hepatic phase of the substrate.
After proximal intestinal absorption, the substrate reaches the liver via the portal vein, and undergoes metabolism in the hepatocyte with ultimate production of 13 CO 2 , which appears quickly in expired air [96,143,157]. For liver function, several substrates have been developed and marked at one carbon site with the natural stable isotope carbon 13 C. Substrates are designed to target liver microsomes (i.e., methacetin, aminopyrine, phenacetin, caffeine, lidocaine, and erythromycin), cytosolic enzymatic activity (i.e., phenylalanine and galactose), and mitochondria (methionine, KICA, and octanoic acid) ( Figure 3).
Stable isotopes BT can be used to assess the course of a disease or the effect of therapies and has no restrictions with respect to infants and pregnant women. Few drawbacks are the limited availability and costs of the equipment, costs of substrates, and need for experienced operators. To date, the application of stable isotope breath testing has remained rather experimental in few referral centres and currently no substrate (apart from urea for the diagnosis of H. pylori infection) is officially approved for clinical use [96,147,158]. Approval from local ethical boards is recommended [96]. The main nonparenchymal cells populating the sinusoid are the fenestrated sinusoidal endothelial cells in contact with the blood, the phagocytic Kupffer cells, which adhere on the luminal aspect, and the hepatic stellate cells which are specialized pericytes that extend processes throughout the space of Disse. They serve as myofibroblasts during times of hepatic injury and repair, liver-associated lymphocytes (Pit cells). This mass of sinusoidal nonparenchymal cells account for approximately 6% of the total liver volume, and about 30% of the total number of liver cells. At the liver sinusoid, several substrates are used to explore liver function by a breath test (see text for details) [143]. Figure 4 depicts the general methodology of 13 C-BT. Briefly, the substrate labelled with 13 C is dissolved in tap water and administered by oral route [96,158]. Subjects are required to be fast overnight, i.e., at least 8-12 h. No special diet is required the day before the test, which is generally performed in the morning.

Methodology of 13 C-BT
The test is performed in a quiet room and the subject should not exercise and should refrain from smoking for at least 30 min before and during the test. This approach will minimize variations in endogenous CO 2 production due to physical activity or combustion. A first breath collection is performed at baseline into plastic bags (250-500 mL) or special glass tubes with rubber caps (exetainer) that are properly labelled. Afterwards, the subject drinks the solution with substrate within 1-3 min. Samples of expired air are then collected at different time points, usually every 15 min up to 30-120 min, depending on the protocol. As an example, a total of 9 samples are taken for a 15-min sampling for 2 h (i.e., at time 0, 15, 30, 45, 60, 75, 90, 105, and 120 min). Plastic bags are preferred if few samples/subjects are necessary to study, while exetainers are preferred if several samples are planned on the same day or the automatic sampling system is available. Breath test can be also defined as "field tests" since with the subject appropriately instructed, samples can be collected at home, in the ward or in the outpatient clinics simultaneously and centralized in the referral lab. Bags and exetainers are tightly closed and samples are measured within 24-48 h, where the equipment is available. Figure 4. The ten-step protocol for the breath analysis using 13 C-substrates for the study of liver function. Starting from the clinical question (1), the fasting subject collects breath samples (2), and, after ingestion of the solution containing the substrate (3), further steps include gastric emptying, duodenal absorption, and portal transfer to the liver (4). The substrate undergoes metabolization at different levels (i.e., microsomes, cytosol, and mitochondria) with production of 13 CO 2 (5). Afterwards, 13 CO 2 undergoes incorporation into the bicarbonate pool and quick transport to the lung (6) where appears in exhaled air ready for collection (7). Breath samples are measured by infrared spectroscopy or mass spectrometry (8) with software constructing time-dependent curves of the metabolic process (9). The information serves to elaborate the specific clinical interpretation of liver function (10). An important assumption is that gastric emptying is not severely delayed and that the lung function is not severely impaired [96,143,145,158]. Cartoon adapted from Di Ciaula et al., Eur. J. Internal Medicine 2021 [39].
The enrichment of expired 13 CO 2 is then analysed by isotope ratio mass spectrometry or by infrared spectroscopy (e.g, IR-300 plus, Beijing Richen-Force Science & Technology Co., Ltd., Bejing, China or Helifan Plus, Fischer ANalysen Instrumente GmbH, Leipzig, Germany). Devices based on molecular correlation spectroscopy can detect variations less than 1:1000 in the 13 CO 2 / 12 CO 2 ratio.
The equipment is set to calculate the rate of exhalation of 13 CO 2 at each time point from the measured increment in the isotopic abundance of 13 CO 2 (δ 13 CPDB). The algorithm takes into account the purity of the labelled compound with a constant endogenous production of CO 2 of 300 mmol/m 2 /h. The results are expressed as a percentage of the administered dose recovered per hour. The cumulative percentage of 13 CO 2 in breath is calculated as the area under curve (AUC) of the 13 CO 2 exhalation rate compared with the time curve determined by linear interpolation using the trapezoidal rule [159,160]. For the 13 C-methacetin breath test, the authors reported that the inter-test coefficient of variation of the 13 C-MBT in healthy volunteers and liver disease patients is 13.2% for the PDR peak, with a coefficient of variation of 23.9% for the cumulative PDR at 20 min cPDR20.

Factors Potentially Affecting the Use of 13 C Breath Tests for the Assessment of Liver Function
The assessment of liver function by 13 C-breath test can be affected by physiologic or pathologic conditions acting on the individual baseline CO 2 production or on the perfusion of the liver by blood [161]. Furthermore, the liver metabolism of oral-ingested substrates can be either decreased or increased by concomitant treatments with drugs influencing the activity of the cytochrome P450 (Table 2). A severely delayed gastric emptying (e.g., diabetic, or idiopathic gastroparesis, severe motility defects, inflammation, and malignancies) might interfere with the delivery of the substrate to the duodenum. Thus, before the examination, the operator should investigate the clinical history to detect symptoms potentially related with altered gastric emptying or conditions affecting the absorption of the substrate from the gastrointestinal tract [161,162]. In this case, the intravenous administration of the substrate (as the LiMAx ® test) can be useful to obtain accurate results, also limiting inter-individual variations in substrate absorption after oral administration [163][164][165][166].

Assessing Liver Mitochondrial Function by BT
The assessment of liver function and responses to therapy is often required at an early stage in clinical medicine. The assessment of early involvement of mitochondrial function could be part of this procedure adding prognostic information [167]. There are substrates that target mitochondria and can be marked with 13 C, which becomes detectable in expired air. Substrates are alpha-ketoisocaproic acid (KICA), methionine, and octanoic acid [147]. Benzoic acid undergoes glycine conjugation, but was used only in the animal model of liver cirrhosis [168]. Following the mitochondrial catabolism of a certain substrate, CO 2 is produced. The labelled carbon (*C) will appear in breath as *CO 2 , as a marker of mitochondrial clearance of the substrate [96]. The radioactive isotope 14C had limitations because of radiation exposure, and thus could not be used in pregnant women and children. 13 C, by contrast, is a stable, nonradioactive, naturally occurring isotope. It accounts for about 1.1% of all the natural carbon on Earth and in plants and the food chain. The following assumption define this type of breath test, or a given exogenous substrate, i.e., Hepatic clearance = Hepatic Perfusion x Hepatic Extraction (where Hepatic Extraction is the ratio of the difference between inflow and outflow concentration ÷ by inflow concentration of the probe) [169]. The hepatic clearance is either flow-limited (range 0.7-1.0) or enzymelimited (<0.3) [170]. Few 13 C-labeled substrates are suitable to explore mitochondrial liver function, and include alpha-ketoisocaproic acid, methionine, and octanoic acid (Figure 3). The mitochondrial metabolism of these three substrates is depicted in Figure 5.
KICA is a substrate for branched chain alpha ketoacid dehydrogenase, located in the hepatic mitochondrial matrix [171]. Methionine is a substrate for protein synthesis or alternatively enters the methionine cycle, with subsequent transformation in Sadenosylmethionine, a methyl donor. S-adenosylhomocysteine, resulting from donation of the methyl group, is hydrolyzed to homocysteine, which, in turn, is destinated to transsulfuration or to remethylation to methionine. The trans-sulfuration step, in particular, generates α-ketobutyrate, which enters mitochondria with subsequent decarboxylatation [172]. Octanoic acid, on the other hand, is β-oxidized with generation of acetyl coenzyme A, which enters the Krebs cycle and undergoes oxidation to CO 2 [172].
All these three substrates have been used to accurately assess hepatic mitochondrial function in NAFLD patients. By different substrates, BT can also allow to explore the hepatocyte at a cytosolic ( 13 C-phenylalanine and galactose) or microsomal level (aminopyrine, phenacetin, caffeine, lidocaine, methacetin, and erythromycin) [172]. From a clinical point of view, these substrates provide information in different clinical settings (Table 3).  Table 3. Liver mitochondrial breath tests: substrates and evidence for potential clinical applications.

ALD
Effect of acute alcohol consumption (even low-moderate doses) [171] ALD Discrimination between chronic alcohol consumption and nonalcoholic chronic liver disease [173] ALD Monitoring and ascertaining of alcohol withdrawal [173] NAFLD Discrimination between simple steatosis and steatohepatitis (NASH) and between low-grade and high-grade fibrosis [147] (

Healthy subjects
Validation studies with 2 mg/kg body weight (methyl-13 C)-methionine. Breath 13 CO 2 enrichment measured at base line and every 15 min thereafter for 180 min. [174] Healthy subjects Effect of alcohol consumption (30 min after the ingestion of ethanol 0.3 g/kg body weight). Decreased excretion with ethanol, due to impaired mitochondrial oxidation [175].

Potential Clinical Application
Each substrate has possible clinical applications (Table 3), as shown in patients with alcohol-related liver disease, NAFLD, viral hepatitis, liver cirrhosis, hepatocellular carcinoma, and evaluation of drug or alcohol toxicity.

13 C-KICA BT
KICA has a molecular weight of 130.141800 g/mol (MF: C6H10O3, IUPAC name: 4methyl-2-oxopentanoic acid), and is part of the metabolic pathway of the amino acid leucine ( Figure 5A). The biotransformation of KICA to isovaleryl-CoA depends on the alphaketoacid dehydrogenase in the liver. The decarboxylation of KICA is a specific function of mitochondria, as confirmed in isolated mitochondria [182], in experimental models [157], and in humans. No gender difference exists when interpreting the results according to body composition [183]. The major competing pathway for KICA elimination is the transamination of KICA to leucine. The 13 C-BT requires the concomitant administration of unlabelled leucine which suppresses this pathway, and this step makes the KICA BT highly sensitive for mitochondrial function. In addition, KICA decarboxylation is depends on the availability of NADH. Thus, ethanol at 0.5 g/kg body weight increases the availability of NADH (leading to decreased KICA decarboxylation), while 1 g of aspirin decreases liver NADH (leading to a higher KICA decarboxylation rate) [171]. These aspects need to be considered during the test performance and the evaluation of results. 13 C-KICA is given at a dose of 1 mg/kg body weight plus 1 g unlabeled L-leucine while 13 C-methacetin is given at a dose of 1.5 mg/kg body weight (generally 75 mg). The substrate is generally flavorless and dissolved in 100 mL of tap water. Such a small volume shortens the drinking time and will allow the prompt initiation of the gastric emptying process. Figure 6. Metabolism of 13 C-ketoisocaproic acid (KICA) as a breath biomarker of mitochondrial function. This study was conducted on 39 (20 lean and 19 obese) hypertransaminasemic patients with histologically-proven NAFLD ranging from simple steatosis (NASH 0-I) to severe steatohepatitis and fibrosis (NASH II-IV). Control subjects were 20 lean and 8 overweight healthy individuals. Compared with healthy subjects and patients with simple steatosis, NASH patients had enhanced methacetin demethylation (p = 0.001), but ketoisocaproate decarboxylation was mildly or greatly decreased and delayed in simple steatosis and steatohepatitis, respectively. Ketoisocaproate decarboxylation was impaired further in obese patients with NASH, but not in patients with simple steatosis and in overweight controls. Symbols (*, #) indicate significant differences compared to the controls (0.006 < p < 0.001). From Portincasa et al. [147].
Several studies are available with KICA. The decarboxylation decreases in alcoholics compared with patients with NAFLD and controls [173,184]. In another study 13 male patients with heavy intake of alcohol during the last month were compared with 10 healthy volunteers. Abnormal liver status was confirmed by abnormal aspartate aminotransferase, alanine aminotransferase, or gamma-glutamyltransferase analyses. Healthy women had a higher percentage exhalation of 13 CO 2 than both healthy males and alcoholic males. Surprisingly, there was no significant impairment of KICA decarboxylation as an effect of chronic intake of alcohol or alcohol-induced steatosis. The information obtained by 13 C-KICA are limited in case of excessive alcohol consumption likely due to a fast normalization of the values [185]. We found that KICA decarboxylation was defective in patients with histologically-proven advanced nonalcoholic steatohepatitis (NASH) but not in patients with simple steatosis [147] (Figure 6). Data were inversely related to the extent of fibrosis especially in obese patients.
In a subsequent study, we found that KICA decarboxylation was decreased in cirrhotic patients with HCC compared with cirrhotic patients without HCC and identical Child-Pugh scores [158], i.e., a classification ranging from score 5 to 15 which incorporates five variables (serum albumin and bilirubin, ascites, encephalopathy, and coagulation as prothrombin time) for assessing the prognosis of liver cirrhosis [186]. The mitochondrial function was further impaired during with radiofrequency ablation (RFA) and trans-arterial chemoembolization (TACE). The application of the 13 C-KICA BT might extend to other conditions, since we described a slight mitochondrial malfunction in a young patient diagnosed with massive liver echinococcosis occupying most of the liver. We detected a cumulative dose recovery (CPDR) of 22% (normal: CPDR within 120 min ≥ 23% normal).
The 13 C-methacetin breath test investigating the liver microsomal function was normal. Notably, mitochondrial liver function improved the following pericystectomy and limited hepatectomy. Other applications of KICA BT are possible. Several drugs can enter the mitochondria and accumulate, a step often interfering with respiratory complexes or electron transfer [187]. For example, aspirin, ibuprofen (nonsteroidal anti-inflammatory drugs), amiodarone (antiarrhythmic agent), and valproate (an anticonvulsant, histone deacetylase inhibitor) inhibit mitochondrial fatty acid β-oxidation [187,188]. The nucleoside analogues are widely used in transplanted patients or in HIV-and HBV-infected subjects. Drugs can incorporate into mitochondrial DNA, inhibit c-DNA polymerase, and hinder the replication process [189]. Nevertheless, viral infection itself (HIV and HCV) may impair mitochondrial function, as confirmed in HCV-infected cells [190] and in patients [177]. In addition, xenobiotics can cause excessive activation of the mitochondrial permeability of the transition pores and alter mitochondrial function. Drugs include acetaminophen [191], N-nitrosofenfluramine [192], salicylate [193], and nimesulide (in vitro) [194]. The KICA BT might provide helpful results to study the integrity of liver organelles before the administration of potentially toxic drugs and to detect drug-induced mitochondrial damage before the appearance of symptoms to timely manage patients and prevent adverse effects. Examples are tacrolimus, aspirin [171], and ergot alkaloids [195].

13 C-Methionine BT
Methionine has a molecular weight of 149.21134 g/mol, MF: C5H11NO2S, IUPAC name: (2S)-2-amino-4-methylsulfanylbutanoic acid. It is an essential amino acid that is involved in metabolic processes. Exogenous methionine contributes to protein synthesis [196]. Methionine can be transformed into S-adenosylmethionine, the main biological methyl donor, which is hydrolysed to homocysteine, either undergoing trans-sulfuration to α-Ketobutyrate or remethylation to methionine. α-Ketobutyrate enters the mitochondria and undergoes decarboxylation. The administration of oral L-(1-13 C)-methionine will then be associated with production of labelled 13 CO 2 . Transmethylation to S-adenosylmethionine also provides the substrate for the synthesis of sarcosine, which is oxidized to formaldehyde and production of CO 2 in mitochondria. Methionine differentially labelled in the methyl group and in position 1 can be used to study the complex metabolism of methionine [175]. For mitochondrial function studies, suitable substrates are either L-(1-13 C) methionine or (methyl-13 C)-methionine. There are some limitations, e.g., comparisons of different studies can be quite difficult since more than one 13 CO 2 is formed according to the L-(1-13 C) or (methyl-13 C)-labelled methionine [197]. Some studies used intravenous rather than oral methionine and in this case the comparisons become even more difficult [198]. Methionine BT can provide information about mitochondrial function during acute intoxication and in chronic liver diseases. However, acute ethanol consumption impairs 13 C-methionine decarboxylation in normal liver [174]. The metabolism of methionine decreases in patients with liver cirrhosis and especially in those with an aethanol etiology [175], in patients with biopsy-proven severe NAFLD in relation to the extent of steatosis [178], and in patients taking high-dose valproic acid [178] or nucleoside analogues for the treatment of HIV [189]. Methionine BT is reported in hepatitis C-infected cells [190], and in patients with Friedreich ataxia [179] an autosomal recessive degenerative disorder caused by loss of function mutations in the frataxin gene (FXN gene), located on chromosome 9q13 [199].

13 C-octanoate BT
Octanoic acid (OA) has a molecular weight of 144.21144 g/mol, MF: C8H16O2, IUPAC name: octanoic acid, caprylic acid. OA is a straight medium chain saturated fatty acid with an 8-carbon backbone, it is found in the milk of various mammals and is a minor component of coconut oil and palm kernel oil. OA enters mitochondria independently of the carnitine transport system and is β-oxidized to acetyl coenzyme A (AcCoA) [200]. AcCoA enters the Krebs cycle and is oxidized to CO 2 unless utilized for the synthesis of other energy-rich compounds ( Figure 5C). The 13 C-octanoate should also reflect hepatic mitochondrial function (β-oxidation capacity). For use in humans the test requires informed consent. In the animal models 13-octanoate BT was informative about liver function in rat models of acute hepatitis and thioacetamide-induced liver cirrhosis, but not in cholestatic liver injury [201]. In NASH patients, the oxidation of octanoate was either unchanged, although greater in women than men [181] or increased [180], and unchanged in those with early stage and advanced cirrhosis with and without a porto-systemic shunt [202]. Such apparently discrepant results with octanoate might be due to subtle differences in the metabolic pathways, the substrates employed, or by extra-hepatic mitochondrial oxidation of octanoate. Gender differences should be also taken into account, when considering the study of Schneider et al. [181], for example. Unfortunately, a comparison of different substrates and BT in the same group of subjects/patients has not been performed, so far. If liver damage is absent, the 13 C-octanoate BT is a useful diagnostic test to measure the rate of gastric emptying to solids, i.e., a muffin enriched with the labelled substrate [146,203].

Why Studying Liver Mitochondrial Function in NAFLD
There is no established therapy not as a monotherapy nor in combination with NAFLD. The complexity and the number of pathogenic mechanisms involved in the full spectrum of NAFLD, makes this goal difficult to achieve and experiment with, so far [16]. Nevertheless, there might be some arguments for studying mitochondrial function in NAFLD patients (Table 4). Table 4. Arguments for assessing liver mitochondrial function in NAFLD patients.
Increasing Scientific Interest About the Role of Mitochondria in NAFLD Impaired liver mitochondrial function may occur early during the onset and progression of NAFLD. General measures for NAFLD can be beneficial to liver mitochondria as well.
Few medications show some beneficial effects on liver mitochondria. Improved mitochondrial function can contribute to ameliorate other liver dysfunctions in NAFLD patients.
In general, a modification in lifestyle (i.e., diet and regular physical exercise [45,46]) and other general measures serve to maintain body weight or reduce body weight in overweigh/obese subjects. Ideally, weight loss should be in the range of 5-7% and 7-10% in NAFLD and NASH, respectively, in both overweight and obese patients [45]. This approach can improve liver biochemical tests, liver histology, serum insulin levels, and quality of life [204][205][206][207][208][209]. We have learnt that in NASH, liver fibrosis can improve after at least 10% weight reduction, although this goal is difficult to achieve in the majority of patients and to maintain for a long time [142,207]. To improve insulin sensitivity and reduce body weight, the diet must be based on long-term caloric restriction rather than intermittent fasting [210]. This approach will prevent oxidative damage [211,212]. Bariatric surgery is indicated in the subgroup of morbid obese patients or obese patients with increased cardiovascular risk. This choice can reduce the prevalence of NASH [213,214]. Specific risk factors for cardiovascular disease, diabetes mellitus must be screened and appropriately treated in NAFLD patients (e.g., antidiabetic agents, lipid-lowering therapy). Alcohol consumption, even in small amounts, is not recommended, since it is associated with progression of liver fibrosis [215].
When considering the aspects related to liver mitochondria in NAFLD, potential targets include nuclear receptors and compounds involved in different signaling pathways, mitochondrial transporters, enzymes playing a major role in mitochondrial metabolism, biomolecules involved in pathways controlling reactive oxygen species (ROS) and oxidative stress. Therapeutic strategies, however, are highly experimental and in several cases tried in animal or in vitro models. Although potentially able to ameliorate mitochondrial function in NAFLD, these agents require further evidence about use in humans, safety, efficacy, duration of treatment, type of steatosis, etc.

Future Perspectives and Conclusions
The interpretation of the results investigating the mitochondrial function by 13 C-BT in vivo requires some considerations. Studies show that both marked steatosis and NASH, and ethanol consumption, cause mitochondrial dysfunction, which becomes detectable by KICA and methionine BT. This aspect should be particularly useful in the case of early treatments possibly limiting the progression of the disease, and in terms of secondary prevention measures. On the other hand, experimental studies indicate that mitochondrial dysfunction can precede the onset and progression of NAFLD, and these findings also pave the way to possible primary prevention measures. Several factors can influence results from BT, affecting liver perfusion and/or mitochondrial performances. Thus, particular attention is required in the selection of subjects undergoing 13 C-BT and in confounding factors, also considering possible inter-individual differences. 'Competing' mitochondria may be active in extra-hepatic tissues (e.g., muscle). The production of CO 2 can vary substantially among subjects [268] and therefore the labelled CO 2 in breath might become independent of circulating or renally excreted bicarbonate and the endogenous production of unlabeled CO 2 . Thus, adequately validated BT will have clinical utility for diagnosis, prognosis, or efficacy of treatments. Remaining limitations of BT applications in clinical practice should be overcome by further translational studies and clinical trials, in parallel with complementary diagnostic techniques. The implementation of novel substrates to investigate additional mitochondrial pathways is greatly warranted.