NAFLD is the most common liver disease in adults and children in the world. The pathophysiology of NAFLD is multifactorial, stemming from a complex interplay of ecological, genetic, and metabolic factors, such as high energy intake, limited physical activity, and malnutrition [27]. Combined with epigenetic factors, these promote insulin resistance and liver fat accumulation [13, 28].

The hypothesis of “two hits” is proposed for the development of NAFLD [29, 30], in which “the first hit” involves the accumulation of lipid in hepatocytes. In fact, the increase in lipid uptake, the increase in de novo synthesis of fat, the decrease in lipoprotein synthesis, or the decrease in fatty acid oxidation may all lead to lipid accumulation. After the “first hit”, the liver has a heightened sensitivity to many previously inconsequential factors, including gut derived endotoxin, endoplasmic reticulum (ER) stress, oxidative stress and subsequent lipid peroxidation, pro-inflammatory cytokines and adipocytokines signals, and mitochondrial dysfunction. These effects result in the “second hit”, which promotes liver cell damage and death, inflammation, activation of hepatic stellate cells (HSCs), and production of extracellular matrix, eventually leading to fibrosis [30, 31]. A fatty liver could develop into liver damage accompanied by fibrosis. The process of fibrosis begins with the activation of HSCs by cytokines and growth factors, while the subsequent acquisition of a new set of epigenetic markers maintains this new cell phenotype. Low levels of inflammatory signaling combined with epigenetic mechanisms usually keep HSCs latent, so a reduction in the availability of methyl donors can induce and maintain hepatic fibrosis [32].

Phospholipid balance, paramount for liver health, is maintained primarily by PEMT [33, 34, 35]. In the presence of PEMT, endogenous synthesis of phosphatidylcholine is possible, reducing the reliance on dietary choline [13]. When choline intake is insufficient, the PEMT pathway that synthesizes phosphatidylcholine becomes very important for maintaining the supply of phosphatidylcholine in the liver. Since the pathway occurs only in the presence of SAM, the regulation of SAM is essential in ensuring a sufficient phosphatidylcholine supply [1].

Any malfunctions in the metabolism of choline phospholipid could lead to hepatic steatosis, which can induce fatty liver. Hepatic steatosis occurs when fatty acid intake, de novo synthesis of fatty acid, or triglyceride synthesis in the liver exceeds the output or oxidation of lipids [1, 36]. Phosphatidylcholine is necessary for the assembly, secretion, and dissolution of cholesterol in bile. The homeostasis of phosphatidylcholine is achieved by balancing the uptake/synthesis of phosphatidylcholine, choline, and betaine, as well as the catabolism/secretion of phosphatidylcholine [1]. Phosphatidylcholine deficiency increases the de novo synthesis of saturated fatty acid in the liver. Compared with people with healthy liver, NAFLD patients exhibit an excessive level of triglycerides rich in saturated fatty acids, along with an increased risk of fatty liver disease, pancreatitis, cirrhosis, and liver cancer. This indicates that the increase in de novo synthesis of fat will lead to hepatic steatosis [1, 23]. Phosphatidylcholine is also involved in the assembly and secretion of very low density lipoprotein (VLDL) [4, 37]. It is necessary for the packaging and export of triglycerides in VLDL and the dissolution of secretory bile salts. VLDL granules transport fat and cholesterol from the liver to peripheral organs. Abnormal VLDL mediated triglyceride secretion is the main culprit of hepatic steatosis [13, 37, 38]. The assembly and secretion of liver VLDL is a complex process, involving apolipoprotein B (ApoB), microsomal triglyceride transfer protein (MTP), lipid mobilization, and synthesizing proteins. In fact, rare ApoB and MTP mutations are associated with progressive liver disease [23, 39].

The occurrence and development of NAFLD are closely related to obesity and insulin resistance. The accumulation of visceral adipose tissue is the main reason for an increase in fatty acids, proinflammatory regulators, and growth promoting regulators, it is closely related to the progress of nonalcoholic steatohepatitis (NASH) [40]. A low choline diet can lead to fatty liver and liver injury, possible due to the accumulation of fat and cholesterol in liver [4, 37]. This dietary requirement for choline is regulated by estrogen and single nucleotide polymorphisms (SNPs) in specific genes in choline and folate metabolism [3]. The classical function of estrogen is realized by its receptor ER$\alpha$ and ER$\beta$. ER$\alpha$ and ER$\beta$ bind to estrogen response elements (EREs) in the promoter regions of many estrogen response genes in the form of homodimer or heterodimer. Since there are several EREs in the promoter region of PEMT gene, estrogen can effectively regulate the expression of PEMT mRNA and the activity of liver cell enzyme when it is within the human physiological concentration range (0–100 nmol/L). Therefore, premenopausal women have the ability to enhance the de novo biosynthesis of choline and most postmenopausal women receiving estrogen therapy require less dietary choline [3, 26]. However, it was found that about 40% of premenopausal women have an SNP in their PEMT gene (rs12325817), thus inhibiting the induction of PEMT via estrogen and increasing their dietary demand for choline [5, 13].

Lipid homeostasis in the liver strikes a fine balance between uptake and synthesis, as well as secretion and catabolism. This homeostatic equilibrium is disturbed in Pemt${}^{\mathrm{-}\mathit{}\mathrm{/}\mathrm{-}}$ mice, leading to a decrease in their ability to secrete VLDL particles and severe accumulation of triglycerides in the liver [41]. Some studies have shown that Pemt${}^{\mathrm{-}\mathit{}\mathrm{/}\mathrm{-}}$ mice are more prone to hepatic steatosis and spontaneous NASH due to their damaged PEMT pathway and liver VLDL assembly ability [23]. In addition, the expression of PEMT mRNA in the liver tissue of NASH mice was significantly lower than that of simple steatosis mice [40]. Other factors, such as reduced fatty acid oxidation in the liver, or increased fatty acid flow from diet or white adipose tissue to the liver, may also contribute to the development of liver steatosis in Pemt${}^{\mathrm{-}\mathit{}\mathrm{/}\mathrm{-}}$ mice to some extent. Therefore, increasing fatty acid oxidation can compensate for low VLDL-triglyceride secretion and prevent, or even reverse, the fatty liver in PEMT-deficient mice [41].

Chronic impairment of lipid metabolism affects the balance of oxidants/antioxidants, influencing the functionality of metabolism-related organelles, leading to chronic endoplasmic reticulum stress (ERS), and mitochondrial dysfunction. The interaction between ERS and oxidative stress plays a crucial role in the pathogenesis of NAFLD [42]. Disruption of mitochondria-associated membrane (MAM) integrity results in an imbalance in Ca${}^{2+}$ homeostasis, increasing ERS and oxidative stress [43]. Increased oxidative stress also activates hepatic cell stress pathways, leading to inflammation and fibrogenesis, contributing to the progression of NASH. As the ER is responsible for the synthesis, folding, transportation, and maturation of proteins, it participates in the synthesis of molecules such as cholesterol, phosphatidylcholine, phosphatidylethanolamine, and triglyceride, as well as calcium homeostasis and drug metabolism. Under ER stress, cells activate the unfolded protein response (UPR) to restore the homeostasis equilibrium caused by abnormal protein metabolism [44]. It has been confirmed that the significant reduction of the ratio of hepatic phosphatidylcholine to phosphatidylethanolamine (PC/PE) and lysophosphatidylcholine to phosphatidylethanomine is associated with the occurrence of NAFLD [44, 45]. PEMT deficiency reduced the ratio of PC/PE in the ER, inducing ER stress and making the mice more prone to fatty hepatitis induced by a high fat diet [33, 46].

Hepatocytes are subjected to sustained stress or injury, and the activation of macrophages, leading to activation of resident hepatic stellate cells into myofibroblasts, which resulting in the accumulation of extracellular matrix. This process ultimately leads to fibrosis, cirrhosis, and even progression to hepatocellular carcinoma [47]. Low dietary choline result in steatohepatitis, fibrosis, cirrhosis. Studies have shown that Exosomes derived from human adipose mesenchymal stem cells (hADMSCs-Exo) can promote choline uptake, reduce lipid accumulation, and restore abnormal choline metabolism, thereby helping to alleviate liver cell damage and fibrosis. On this basis, choline supplementation is supposed to exert a synergistic effect with hADMSCs-Exo in anti-hepatic fibrosis [48].

In summary, NAFLD includes a series of pathologies, ranging from nonalcoholic fatty liver (NAFL, commonly referred to as simple steatosis, defined as steatosis without any histological or biochemical damage) to NASH, characterized by steatosis, inflammation and hepatocyte damage (swelling), with or without cirrhosis. NAFL is generally considered as a benign disease, and NASH is a progressive disease. Patients with NASH are likely to develop advanced fibrosis and cirrhosis, or even liver cancer (Fig. 4) [28, 49, 50].

NAFLD patients are at an increased risk of death from cardiovascular disease and diabetes, so there is a growing need for new and better diagnostic methods and treatments [29, 61]. Peroxisome proliferator-activated receptors (PPARs) are members of the nuclear hormone receptor superfamily, which can directly regulate target gene transcription by forming a heterodimer with the retinoid receptor X (RXR) and binding to the PPAR response element (PPRE) sequences in gene promoters [62]. There are three subtypes of PPAR: $\alpha$, $\delta$ and $\gamma$, which are differentially expressed in different tissues [31, 63]. PPAR$\alpha$ is widely expressed, mainly in cells that exhibit high levels of $\beta$-oxidation and mitochondrial activity, such as liver and cardiovascular tissue; PPAR$\delta$ is mainly expressed in skeletal muscle, less in adipose tissue and skin; PPAR$\gamma$ is highly expressed in adipose tissue [64, 65]. PPAR agonists can improve liver steatosis and reduce the incidence of NAFLD by improving fatty acid $\beta$-oxidation and reducing the release of inflammatory cytokines, so they may be considered as a drug for the treatment of NAFLD [65, 66]. In early clinical trials in rodents and humans, PPAR agonists have been shown to ameliorate NASH by inducing the expression of $\beta$-oxidation related genes, reducing inflammation and oxidative stress (Table 1, Ref. [41, 67, 68, 69, 70, 71, 72, 73, 74]) [31, 64].

PPAR$\alpha$ plays a key role in regulating fatty acid uptake, $\beta$-oxidation, ketogenesis, bile acid synthesis, and triglyceride conversion. PPAR$\alpha$ can induce the expression of many genes involved in fatty acid oxidation. PPAR$\alpha$ plays both a role in metabolic regulation and an anti-inflammatory role through the complex regulation of nuclear factor kappa-B (NF-$\kappa$B) [31, 41, 65]. Fibrates, which belong to phenoxyaryl acids, are synthetic PPAR$\alpha$ agonists for the treatment of dyslipidemia, which can reduce blood lipids, especially triglycerides. Fenofibrate and Ezetimibe, two types of fibrates, have shown drastically different results in studies. The former exhibited the ability to prevent and partially reverse NAFLD in Pemt${}^{\mathrm{-}\mathit{}\mathrm{/}\mathrm{-}}$ mice, while the latter can not reduce liver lipid accumulation, inflammation or fibrosis in these mice [41].

PPAR$\delta$ is expressed primarily in skeletal muscles, but trace amounts are also found in adipose tissue and skin. It is involved in the regulation of mitochondrial metabolism and fatty acid beta oxidation. PPAR$\delta$ is well expressed in hepatocytes as well as Kupffer cells and HSCs, suggesting that PPAR$\delta$ may play a potential role in inflammation and fibrosis. However, studies have shown that PPAR$\delta$ deficient mice are more susceptible to CCl4 induced liver fibrosis [31, 65].

PPAR$\gamma$ is highly expressed in adipose tissue and plays an important role in adipocyte differentiation, adipogenesis, and lipid metabolism [31, 65]. It has been reported that Nimesulide can inhibit obesity-related NAFLD and hepatic insulin resistance by regulating PPAR$\gamma$ [62]. Pioglitazone, a thiazolidinedione antidiabetic drug and insulin sensitizer, is a synthetic PPAR$\gamma$ agonist that can bind to PPAR$\gamma$ to form a heterodimer with Retinoid X Recepto (RXR). This heterodimer can affect the transcription of genes related to glucose, lipid metabolism, and inflammation reduction [31, 64, 75]. Some studies have shown that Pioglitazone may reduce hepatic steatosis by enhancing cytoplasmic lipolysis, $\beta$-oxidation, and autophagy [62]. Receptor-interacting protein kinase 3 (RIPK3) is an important regulator of necroptosis, a form of programmed cell death [76]. It regulates the production of reactive oxygen species within mitochondria, controls the sequestration of metabolic enzymes and resident mitochondrial proteins, modulates the activity of mitochondrial respiratory chain complexes, influences mitochondrial biogenesis, and impacts fatty acid oxidation [77, 78]. Emerging studies suggest that manipulating RIPK3 expression can have a significant impact on NAFLD. Specifically, recent studies have demonstrated that Ripk3${}^{\mathrm{-}\mathit{}\mathrm{/}\mathrm{-}}$ mice can effectively target and enhance the expression of PPAR$\gamma$, leading to a potential reversal of NAFLD progression [76, 78, 79]. These findings highlight the therapeutic potential of targeting RIPK3 in the treatment of NAFLD.

Although PPAR$\alpha$ agonists seem promising in animal studies, they have not been very effective in clinical trials. PPAR$\delta$ agonists may also be promising, but the data from human trials are very limited. PPAR$\gamma$ agonists are more successful in improving histological results, including fibrosis in clinical trials. However, the widespread use of thiazolidinediones is limited by its side effects such as weight gain, edema, bone loss, bladder cancer, and increased risk of cardiovascular complications. The utilization of agonists that activate multiple PPARs has also yielded promising outcomes, but it is important to note that the administration of dual or pan agonists has been associated with certain side effects [31, 50, 74, 75].

Farnesoid X receptor (FXR) is another kind of nuclear receptor and highly expressed in both hepatocytes and intestinal enterocytes where it controls all aspects of bile acid metabolism. It is endogenously activated by bile acids, which can improve insulin sensitivity and show direct anti-inflammatory and anti-fibrosis effects in NASH mouse model and human tissues [80]. These compounds are FXR agonists, whose prototype is Obeticholic Acid (OCA), which can improve insulin resistance [81]. The GLS1-subtype of glutaminase was enhanced in both NASH clinical biopsy and preclinical mouse models. Since GLS1 silencing significantly reduced steatosis and oxidative stress through complex metabolic reprogramming (including increased VLDL output), this suggests that GLS1 may be an effective target for the treatment of NASH [23]. Recently, a benzimidazole derivative has been designed that possess FXR/PPAR$\gamma$ dual-partial agonist activity and shares the ability to decrease the phosphorylation of PPAR$\gamma$-Ser273 mediated by cyclin-dependent kinase 5, as well as exhibiting metabolic stability in mouse liver microsomal assays [82]. Therefore, this analog would be a viable candidate drug for the treatment of NAFLD.

Since gut microbiota plays a key role in the occurrence and development of NAFLD, gut microbiota has the potential to be the target for the treatment of NAFLD. Probiotics, prebiotics, and synbiotics have demonstrated their therapeutic potential in modulating the microbiota composition and promoting microbial equilibrium [83]. Recent studies showed that gut microbiota also can be regulated by antibiotics such as Vancomycin [84] and some Traditional Chinese Medicine (TCM) formulas such as Qushi Huayu Decoction (QHD) [85].

Probiotics was defined as “living microorganism” by the World Health Organization in 2001. Bifidobacterium and Lactobacillus are the main probiotics, also called traditional probiotics, which are considered to be harmless to the human body [52, 53]. Only in recent times have researchers begun to comprehend the mechanisms of their functionality by employing rodent models. In animal models with diet-induced NAFLD, the administration of oral probiotics containing Lactobacillus fermentum has demonstrated the ability to decrease hepatic lipid accumulation, oxidative stress, and the expression of inflammatory markers (such as Tumor necrosis factor-$\alpha$ (TNF-$\alpha$) and interleukin-18 (IL-18)), thus provide protection against NAFLD-Hepatocellular carcinoma (HCC) by releasing the anti-tumor metabolite acetate through the gut-liver axis [86, 87]. Furthermore, the combination of Bifidobacterium infantis, Lactobacillus acidophilus, and Bacillus cereus has been shown to ameliorate hepatic lipid accumulation, inflammatory infiltration, liver enzyme levels, serum LPS, and inflammatory cytokine levels in an animal model of high-fat and high-glucose diet (HFD) [88]. Two meta-analyses, included patients with NAFLD, demonstrated that probiotics reduced body mass index, liver enzyme levels, inflammation, and improved symptoms of diabetes and dyslipidemia [89, 90]. They have revealed limited efficacy of probiotics in positively modulating human metabolism. The immunomodulatory effect of probiotics may be beneficial to the treatment of NAFLD, which can be used as an adjuvant therapy for NAFLD patients [91]. The mechanism of action is the increase of fatty acid formation, the change of colon pH, the induction of growth factors, and the change of gut microbiota, the inhibition of anti-colonization bacteria adhesion, immune regulation, the increase of phagocyte activity, the increase of IgA secretion, and the regulation of lymphocyte function [29, 52]. With the remarkable progress in genetic engineering technology, a novel generation of probiotics (NGPs) is currently under development and being applied in the field of NAFLD treatment research (Table 2, Ref. [92, 93, 94, 95, 96, 97, 98, 99, 100]) [101].

In addition, synergistic effects have been observed in the combination of probiotics and chemicals such as metformin in NASH treatment and statins in NAFLD treatment, which highlights the great potential of probiotics alone or in combination with other drugs [71]. However, the clinical efficacy of probiotics still needs to be further validated in well-designed and larger-scale studies.

Prebiotics are a class of non-digestible food ingredients that selectively alter the growth and/or activity of bacteria in the colon [102]. They selectively stimulate “good” bacteria in the colon and inhibit the growth and/or activity of “bad” bacteria [103]. Oligosaccharides such as fructans and galactans are as examples of prebiotics for humans, which promote the proliferation of Bifidobacteria [103]. There is evidence that prebiotic supplements can prevent the development of NAFLD in both experimental and clinical studies. The combination of prebiotics and natural ingredients will produce greater benefits than prebiotics themselves [30]. For example, Isomaltooligosaccharides (IMOs) combined with lycopene (an antioxidant) can prevent high-fat diet induced weight gain in NAFLD mice, enhance fat mobilization in adipose tissue, and improve insulin resistance and metabolic endotoxemia [52].

The combination of probiotics and prebiotics is called synbiotics [29, 52, 53]. Synbiotics usually benefit by selectively stimulating the growth of healthy bacteria and/or activating their metabolism. Synbiotics have shown a variety of benefits in metabolic diseases, such as improving insulin resistance, blood glucose control and inflammatory cytokine synthesis. Synbiotic therapy showed improvements in fasting blood glucose, triglycerides, and inflammatory cytokines in obese and lean NAFLD patients. Therefore, synbiotics is a promising method for the prevention or treatment of NAFLD targeted by gut microbiota. However, more clinical validation is needed [53].

Antibiotics are often used clinically, although their ability to disrupt the homeostasis of gut microbes is a double-edged sword. On the one hand, the short-term use of antibiotics will have a long-term impact on host metabolism; on the other hand, the use of some antibiotics can improve the disease. Compared with antibiotics, some components of Chinese herbal medicine have a greater prospect of regulating gut microbiota with less side effects. Berberine is a typical component of Chinese herbal medicine, which has strong antibacterial activity, especially for intestinal bacteria, because berberine is difficult to be absorbed in the intestinal tract. At present, more and more evidences confirm that berberine has therapeutic effects on metabolic diseases, including obesity, NAFLD, and type Ⅱ diabetes (T2D), by regulating gut microbiota [53].

As a polar compound, choline can only enter cells through transporters on the cell membrane. Many transporter systems and enzyme networks are involved in phosphatidylcholine metabolism, but they are often dysregulated in cancer cells. Increased transport of extracellular free choline into cancer cells is considered a major cause of choline phenotype changes in cancer cells, as this may be the limiting step in phosphatidylcholine formation in some cases. There are four different types of choline transporters related to cancer, classified according to their affinity: (a) high-affinity choline transporters (CHTs); (b) choline transporter-like proteins (CTLs); (c) organic cation transporter (OCTs); (d) organic cation/carnitine transporters (OCTNs). The expression levels of each transporter subtype are increased in cancer cells [9, 116]. Choline transport mediated by CTL1 is considered a fundamental function in phospholipid synthesis, and inhibiting this choline uptake function can induce cell apoptosis. CTL1 is upregulated in various malignant tumors, including central nervous system, breast cancer, prostate cancer, melanoma, kidney cancer, lung cancer, colon cancer, and pancreatic cancer. High expression of CTL1 and CTL2 mRNA was observed in the MIA PaCa-2 pancreatic cancer cell line, and inhibiting CTL1 function resulted in cell apoptosis. CTL1 inhibitors Amb4269951 and Amb4269675 significantly inhibited tumor growth in a mouse xenograft model, indicating that CTL1 is a target molecule for the treatment of pancreatic cancer [117]. Human lung adenocarcinoma tissue highly overexpresses CTL1, and partially overexpresses organic cation transporters 3 (OCT3). In the NCI-H69 cell line, the increase in choline uptake depends mainly on CTL1, partially on OCT3, OCTN1, and OCTN2 [118].

Choline kinase (Chk) is the first enzyme in the choline metabolism pathway, which converts choline to PC. Chk was first discovered in 1953. There are three subtypes of ChK in mammalian cells, Chk-$\alpha$1, Chk-$\alpha$2, and Chk-$\beta$, encoded by two different genes: choline kinase alpha (CHKA) and CHKB. However, only ChK-$\alpha$ plays an important role in maintaining the PC biosynthesis required for tumor cells, and ChK-$\beta$ alone cannot compensate for this activity [119]. In addition to its metabolic function, Chk-$\alpha$ has been shown to be highly expressed in many tumors and is closely related to the occurrence, development, and invasion and metastasis of tumors. It has become a promising molecular target for the diagnosis and treatment of cancer [9, 116]. In ovarian cancer, Dehydrogenase/reductase member 2 (DHRS2) downregulates CHK$\alpha$ protein expression, leading to the inhibition of the protein Kinase B, PKB (AKT) signaling pathway and choline metabolism pathway activity, ultimately inhibiting the proliferation and invasion and metastasis of ovarian cancer [120]. In breast cancer cells, small interfering RNA (siRNA) or short hairpin RNA (shRNA) targeting Chk-$\alpha$ can significantly reduce breast cancer cell proliferation by downregulating Chk-$\alpha$ mRNA expression [24]. In prostate cancer, high expression of CHK$\alpha$ not only plays a role in kinase function leading to abnormal choline metabolism, but also serves as a molecular partner of androgen receptor (AR), promoting the stability of AR, enhancing the AR signal in cells, and promoting prostate cancer proliferation. However, after knocking out CHK$\alpha$, AR signal transduction is inhibited, and prostate cancer proliferation and invasion are ultimately suppressed, suggesting that targeted therapy for CHK$\alpha$ may improve the treatment of advanced prostate cancer [121]. The expression of CHK$\alpha$ is upregulated in breast cancer cells, and inhibiting Chk has strong anti-tumor activity against human breast cancer xenografts in vivo [122].

Tumor development and metastasis involve epithelial mesenchymal transition (EMT). Some studies have shown that inhibition of phosphatidylcholine specific phospholipase C (PC-PLC) can lead to the loss of mesenchymal properties of metastatic breast cancer cells, indicating that PC-PLC plays an important role in the molecular pathway of maintaining mesenchymal like phenotype of metastatic breast cancer cells. Inactivation of PC-PLC may be a means to promote the differentiation of breast cancer cells and improve the effect of anti-tumor therapy [68]. Phospholipase D (PLD) was first discovered in plants in 1947 [123]. PLD is an enzyme belonging to the phospholipase superfamily, and the main subtypes in human cells are PLD1 and PLD2. PLD can hydrolyze phosphatidylcholine into phosphatidic acid and choline. The activity of PLD is significantly increased in tumor tissues and cells, including colorectal cancer, breast cancer, stomach cancer, thyroid cancer, brain cancer, and renal cancer, indicating that it may play a key role in the progression of cancer [124].

Positron emission tomography (PET) is a nuclear medicine modality that offers excellent sensitivity and the potential to provide quantitative information on biochemical processes within tissues or specific target macromolecules in vivo. PET in combination with computed tomography (PET/CT) radiocholine (labeled ${}^{11}$C or ${}^{18}$F) has emerged as a crucial tool in cancer imaging, allowing for the identification and characterization of areas with abnormal metabolic activity in cancer cells [131, 132, 133]. Radiocholine and f-18 Fluorodeoxyglucose (FDG) PET, play a crucial role in cancer diagnosis, staging, and treatment monitoring. For example, Radiocholine exploits the increased uptake of choline by malignant cells due to upregulated choline kinase, leading to elevated phosphatidylcholine formation [20, 24, 116]. This change in metabolism has been observed in various cancers, including colon [9], breast [68], liver [6], ovarian [130], and prostate cancers [134]. On the other hand, FDG PET, which visualizes glucose metabolism in cancer cells, is widely used for cancer detection and staging, aiding in determining the extent of spread in newly diagnosed or recurrent cancers [135]. In addition, clinical trials have shown that metabolic imaging has a significant impact on patient management, improving treatment planning, image-guided techniques such as radiotherapy and stereotactic surgery, and monitoring tumor response to treatment [132].

Currently, the evaluation of metabolic changes mainly relies on gas chromatography, liquid chromatography-mass spectrometry, and high-resolution multinuclear MRS technology [136]. Since the 1990s, MRS has made a great contribution to the characterization of phosphocholine metabolism in tumor cells and tissues, making it a potential indicator for tumor progression and therapeutic response [9, 137, 138]. Meanwhile, high resolution MRS can detect the homeostasis level and flux of metabolites in the biosynthesis and catabolism of phosphatidylcholine [139]. In the past 20 years, MRS has been widely used to study biochemical pathways in intact tissues and extracts from living tissues [8]. MRS can be used for noninvasive monitoring of various biologically important molecules in cells and tissues, highlighting important details critical to cell biochemistry. Therefore, MRS provides a noninvasive method to monitor the metabolites that play a central role in multiple pathways, measures the rate of reactions in these pathways in vivo, and studies how to control these rates in order to provide a given metabolic precursor and its required cofactors [8, 19, 139].

MRS uses different non-radioactive isotopes, such as ${}^1$H, ${}^{31}$P, ${}^{13}$C. Proton (${}^1$H) has the highest nuclear magnetic resonance (NMR) sensitivity, so its spectrum can reach the highest signal-to-noise ratio level [140]. The other core most widely used in metabolism research is ${}^{31}$P, which is a naturally occurring phosphorus isotope [141]. ${}^{31}$P MRS is a useful technique for quantitative analysis of phosphorus in cells, tissues, and their extracts [138, 141]. Although the sensitivity of ${}^{31}$P MRS is lower than that of ${}^1$H MRS, the chemical shift range of ${}^{31}$P MRS is larger, eliminating the need for pulse sequences to suppress strong solvent signal [19]. ${}^{13}$C labeling is particularly useful for studying specific metabolic pathways such as glycolysis and choline metabolism. However, the natural abundance of ${}^{13}$C isotope is only 1.1% and its intrinsic NMR sensitivity is much lower than that of ${}^1$H. Therefore, the ${}^{13}$C signal of biomolecules is very weak without isotope enrichment [8, 142]. One study used NMR techniques to analyze the concentrations of phosphatidylcholine metabolites, the activity of phosphocholine-producing enzymes and the uptake of methyl-${}^{14}$C-choline in human epithelial ovarian carcinoma cell lines (EOC) and compared them with normal or immortalized ovary epithelial cells (EONT). The quantitative analysis of phosphatidylcholine metabolites involved in ${}^1$H NMR total choline resonance (3.20–3.24 ppm) showed that the average value of phosphocholine in EOC cells was 3- to 8-fold higher than that of other phosphatidylcholine metabolites, acting as the main phosphatidylcholine metabolite, while the average value of glycerophosphocholine/phosphocholine decreased significantly. The activity of choline kinase in EOC cells was 12- to 24-fold higher than that in EONT cells. In EOC cells, the total activities of phosphatidylcholine specific phospholipase C and phospholipase D were also higher (about 5-fold), which indicated that the biosynthesis and catabolism of phosphatidylcholine might contribute to the accumulation of phosphocholine [143].

The advantage of MRS is that it does not need to separate and purify metabolites, thus avoiding the use of radioisotope substrates and fluorescence probe quenching [139]. Although it is limited by its low sensitivity, the application of ultra-high magnetic field intensity and the development of new technologies such as dynamic nuclear polarization (DNP) can reduce ameliorate this problem [144]. In addition, MRS has a unique advantage-high spectral resolution, and it can simultaneously detect and quantify the consumption of exogenously added substrates and the formation of products [137, 145]. Therefore, MRS can not only measure the basic level of intracellular metabolites in the samples, but also determine the activities of key enzymes involved in the metabolic pathways analyzed. In this way, MRS can simultaneously measure the activities of different enzymes in the same cell sample and quickly evaluate the rearrangement of metabolic flux under different experimental conditions [8].

In another study, ${}^1$H MRS and ${}^{13}$C MRS techniques were used to analyze the differences in choline membrane metabolism between ${}^{13}$C-choline labeled breast cancer cells and normal human breast epithelial cells (HMECs). Compared with HMECs, the expressions of choline kinase and phospholipase C were significantly overexpressed in breast cancer cells, while the expression of hemolytic phospholipase 1, phospholipase A2, and phospholipase D were significantly down regulated. The change in the expression levels of these enzymes led to the increase in phosphocholine level in breast cancer cells [146]. Deuterium metabolic imaging (DMI) is an emerging technique that utilizes deuterium magnetic resonance spectroscopy (${}^2$H-MRS) to measure the metabolic flux of deuterium-labeled substrates for in vivo imaging of tissue metabolism and cell death [147]. DMI employs various deuterium-labeled substrates such as glucose, acetate, choline, and fumarate to investigate metabolic pathways [148]. The most widely used substrate is [6,6${}^{\prime }$-${}^2$H${}_2$] glucose, which enables imaging of both glycolysis and oxidative metabolism. After injecting deuterium-labeled glucose into rats with a glioma model, radiation signals were first detected in the liver and then in the brain, showing higher lactate and lower Glx (the unresolved deuterium resonances of C4-labeled glutamate and glutamine) levels in the tumors than in the normal brain tissue [149]. This indicated that the tumors had more glycolysis and less oxidative metabolism. DMI can be utilized not only for MRS imaging but also for PET imaging. It offers several advantages over conventional labeling imaging, such as enabling simultaneous multi-substrate labeling [147, 149]. This capability greatly facilitates the investigation of metabolic, neurological, and cancer mechanisms.

Liver cancer (including hepatocellular carcinoma and intrahepatic cholangiocarcinoma) is one of the most common malignant tumors in the world [149, 150]. In contrast to virally driven HCC, up to 50% of NAFLD-HCC occurs in patients without cirrhosis [50].

As mentioned above, choline is very important for liver health. Many mechanisms by which choline deficiency exacerbates steatosis to liver cancer have been found: abnormal phospholipid synthesis, defective lipoprotein secretion, DNA oxidative damage and lipid peroxidation caused by mitochondrial dysfunction, ER stress, loss of hepatocyte apoptosis response, and alteration of protein kinase C signal transduction caused by diacylglycerol accumulation, etc. [9, 114, 116, 129]. Choline is an important part of mitochondrial membrane. Mitochondrial dysfunction is the central mechanism of NAFLD-HCC. When choline levels are low, mitochondrial bioenergy and fatty acid $\beta$-oxidation will be negatively affected. Choline deficiency can change the composition of mitochondrial membrane and cause mitochondrial leakage. The cardiolipin in these mitochondrial membranes is then oxidized, resulting in a decrease in the membrane concentration of phosphatidylethanolamine and phosphatidylcholine [22, 151]. This reduces mitochondrial membrane potential and, ultimately, leads to a decrease in respiratory chain complex Ⅰ activity [13, 32]. In addition, choline-deficient hepatocytes can produce excessive free radicals. The release of free radicals can cause the oxidation of nucleotides, form 8-hydroxydeoxyguanosine, inhibit the methylation of cytosine, and lead to DNA strand breakage and changes of epigenetic markers on DNA and histone [32, 152].

The death of hepatocytes in the absence of choline can also cause inflammatory responses, accompanied by neutrophil/macrophage mediated production of reactive oxygen species and nitrogen [13, 153]. Studies have found that there is a link between choline metabolism and macrophage inflammation [154, 155]. Pro-inflammatory polarization increases choline uptake and phosphatidylcholine biosynthesis, thus promoting macrophages to respond appropriately to immune stimulation [155]. The increase of choline uptake provides the necessary amount of phosphatidylcholine for the synthesis of cell membrane (organelle membrane and plasma membrane), which is conducive to the packaging and secretion of cytokines [115]. In addition, regulation of phosphatidylcholine biosynthesis can reprogram the immune response in primary macrophages [24, 151].

As mentioned above, choline is involved in methyl metabolism and can be used as a donor of methyl. Since there are epigenetic mechanisms that rely on DNA or histome methylation to alter gene expression, choline also plays a role in epigenetic regulation. A choline-methyl deficient diet resulted in a decrease in S-adenosylmethionine concentration and an increase in S-adenosylhomocysteine concentration in the liver of male rats and mice [3]. A simultaneously low concentration of S-adenosylmethionine and high concentration of S-adenosylhomocysteine inhibit the activity of methyltransferase. As S-adenosylhomocysteine has a high affinity for DNA methyltransferases (DNMTs), it competes with S-adenosylmethionine to inhibit DNMTs enzyme activity. Thus, choline-methyl deficiency leads to decreased DNA methylation at many sites, including oncogene c-myc, but also increases methylation of other cytokines (including some tumor suppressor genes, such as p53). Many tumor suppressor genes are epigenetically regulated, including cell cycle regulation genes (p15 and p16), apoptosis genes (DAPK and APAF-1), cell adhesion genes (CDH1 and CDH3), and DNA repair genes (BRCA1 and hMLH) [45, 152]. Histone methylation was also affected by a diet deficient in methyl. Choline-methyl deficiency diet resulted in a decrease in histone H3-lysine 9-trimethylation (H3K9me3) and histone H4-lysine 20 trimethylation (H4K20me3) in the liver [156].

In addition, gut microbiota dysregulation is also involved in the development of NAFLD-related liver cirrhosis and liver cancer. Dysregulation of gut microbiota leads to a surge in inflammatory cytokines, such as TNF-$\alpha$ and IL-8, and the activation of TLR-4 and TLR-9, which leads to the production of IL-1$\beta$ by Kupffer cells. IL-1$\beta$ promotes lipid accumulation and cell death in hepatocytes, causes steatosis and inflammation, and stimulates HSCs to produce fibrosis mediators, thus promoting liver fibrosis. Gut microbiota dysregulation can also promote the occurrence and development of NAFLD-related liver cancer by changing bile acid metabolism. Changes in gut microbiota can lead to high levels of deoxycholic acid (DCA), which leads to senescence-associated secretory phenotype (SASP) in HSCs and results in the secretion of various inflammatory and tumor promoting factors in the liver, thus promoting the development of liver cancer [28].

Compared with normal liver cells, hepatoma cells have drastically different epigenomes-hepatoma cells have gene specific DNA hypermethylation or hypomethylation, histone epigenetic markers changes, abnormal expression of DNA methyltransferase and histone modifying enzyme genes. Phosphatidylcholine is the precursor of lysophosphatidic acid (LPA), which regulates cell proliferation, differentiation, morphogenesis, and prevents apoptosis through G-protein-coupled transmembrane receptors. The genes encoding LPA signal receptors are regulated by epigenetic mechanisms. Abnormal methylation of LPA1 receptor gene was observed in rodents fed a choline-methionine deficient diet, with a pattern similar to that described in liver cancer [32].

Studies have shown that when the key gene in choline metabolism (betaine homocysteine methyltransferase gene BHMT) is deleted in mice, these mice will develop liver cancer [156]. It was also reported that the level of choline kinase-$\alpha$ (Chk-$\alpha$) mRNA in HCC tissues was higher than that in non-tumor tissues and an amplification of the Chk-$\alpha$ gene site increased the expression of Chk-$\alpha$. The tumor with high Chk-$\alpha$ expression has a higher invasive phenotype, resulting in a shorter survival time for the tumor patients. Moreover, Chk-$\alpha$ is essential for the physical interaction between epidermal growth factor receptor (EGFR) and mTOR complex 2 (mTORC2), which is necessary for hepatoma cells to form metastatic xenograft tumors in mice and to develop resistance to EGFR inhibitors. Therefore, stable and high expression of Chk-$\alpha$ in hepatoma cell lines is positively correlated with higher migration and invasion levels, as well as the ability to resist EGFR inhibitors and form metastatic tumors in mice [157] (Fig. 6).

In addition to liver cancer, abnormal choline metabolism has been found in a variety of cancers [9]. Choline phosphate is involved in the malignant transformation of breast cancer, lung cancer, colon cancer, prostate cancer, neuroblastoma, liver lymphoma, meningioma, and other tumors through the RAS oncogene [22]. Many studies have shown that the expression and activity of lipases (including fatty acid synthase and choline kinase) are up-regulated in all stages of prostate cancer, and are associated with poor prognosis and survival rate. The cell membrane composition of normal cells and malignant cells is different-the relative content of choline phospholipid, which is mainly distributed in the outer lobe of cell membrane, decreased in all malignant tumors [12].

Choline kinase inhibitors can be used as a new type of antiproliferative and anticancer drugs [158]. Several kinds of choline kinase inhibitors have been developed: Hemicholinium-3 (HC-3) is a competitive inhibitor with similar structure of Chk-$\alpha$, which can block the transport of choline and is very effective in vitro, but has high toxicity in vivo; MN-58b is a less toxic HC-3 derivative, which can inhibit cell growth in animal models; RSM-932A, a Chk-$\alpha$ inhibitor, has been selected for further clinical development due to its strong anticancer activity in vivo and no toxicity at effective dose; TCD-717, another Chk-$\alpha$ inhibitor, has recently completed phase Ⅰ clinical trials in patients with advanced solid tumors. In breast and liver cancer cell lines, the Chk-$\alpha$ inhibitor PL48 not only inhibits the activity and expression of Chk$\alpha$1, but also inhibits the uptake of choline by the CTL1 transporter, indicating that PL48 is a promising candidate drug that can be further clinically evaluated as a potential anti-cancer drug [159]. However, there are still no Chk-$\alpha$ inhibitors available for clinical use at the moment [130, 160].

Another study found that the expression of choline kinase is partly dependent on the function of cyclooxygenase-2 (COX-2). COX-2 silencing can significantly reduce the levels of phosphocholine and tCho, significantly increase the lipid levels, and form lipid droplets in human breast cancer cells. Moreover, COX-2 silencing can transform the parental cell metabolite mode into a less invasive cancer cell metabolite mode. Therefore, COX-2 targeted therapy can prevent tumor invasion and metastasis [160].

Tumor development and metastasis involve epithelial mesenchymal transition (EMT). Some studies have shown that inhibition of phosphatidylcholine specific phospholipase C (PC-PLC) can lead to the loss of mesenchymal properties of metastatic breast cancer cells, indicating that PC-PLC plays an important role in the molecular pathway of maintaining mesenchymal like phenotype of metastatic breast cancer cells. Inactivation of PC-PLC may be a means to promote the differentiation of breast cancer cells and improve the effect of anti-tumor therapy [68].