Noninvasive Diagnostic Methods in Liver Cirrhosis

Abstract

Liver cirrhosis is a condition characterized by the gradual development of liver fibrosis and the disruption of hepatic lobules. Patients who have decompensated cirrhosis face a significant risk of severe complications, including ascites, esophageal varices, liver failure, and hepatocellular carcinoma. Early diagnosis and timely intervention are crucial to preventing further liver damage, reducing morbidity and mortality associated with complications, and improving the prognosis. Additionally, timely diagnosis and accurate assessment of liver cirrhosis are critical for effective management and treatment. While liver biopsy has long been considered the gold standard for diagnosing cirrhosis, it has well-known limitations, including invasiveness, sampling error, and high expense. These limitations have restricted its widespread use in clinical practice. As a result, noninvasive diagnostic methods for liver cirrhosis have been proposed as alternatives to liver biopsy. Current noninvasive methods encompass liver and spleen stiffness measurements, ultrasound, computerized tomography, and magnetic resonance imaging, as well as serum biomarkers. Additionally, emerging technologies, such as omics, have led to the identification of novel biomarkers. However, the diagnostic performances of these methods vary among studies. Further, research and standardization of these methods are necessary to enhance their diagnostic accuracy and clinical utility in the evaluation of liver cirrhosis.

1. Introduction

Liver cirrhosis is a pathology characterized by progressive hepatic fibrosis and lobular disruption [1]. Decompensated cirrhosis patients are at heightened risk of critical complications such as ascites, esophageal varices, hepatic failure, and hepatocellular carcinoma. Prompt diagnosis and intervention in liver cirrhosis, including its attendant complications, are paramount in mitigating further hepatic deterioration, decreasing associated morbidity and mortality, and thus enhancing patients’ prognoses. Nevertheless, these imperatives face constraints in clinical application due to various limitations.

Liver biopsy has long been considered as the gold standard for diagnosing cirrhosis, whereas this method is beset with several potential limitations [2, 3]. Firstly, the biopsy sample’s length may not accurately represent the liver’s overall pathological state, with a longer biopsy sample potentially reducing the sampling error. It is generally accepted that a biopsy sample length of 15 mm suffices for stage assessment, though some reports advocate for a 25 mm length for more precise evaluation [3]. Additionally, a 16 gauge needle is deemed suitable for percutaneous liver biopsies. A second limitation involves variability in pathological assessments among pathologists, a challenge potentially mitigated by evaluation from specialized liver pathologists. Thirdly, liver biopsy, being invasive and costly, is unsuitable for widespread screening or longitudinal follow-up in the general population. Consequently, noninvasive diagnostic modalities have been proposed as pragmatic alternatives to liver biopsy, suitable for screening, monitoring, and follow-up. These encompass traditional techniques such as ultrasonography, computerized tomography, and magnetic resonance imaging, along with elastography for assessing liver and spleen stiffness, serum biomarkers and algorithm, and burgeoning methods including genetic models, omics, microbiome as well as artificial intelligence analyses.

Serum biomarkers	Variables
Patented biomarkers
FibroTest®; FibroMeter®; Flepascore®; FibroSpectll®; ELF®	age, gender, α-2-macroglobulin, γGT, apolipoprotein A1, haptoglobin, total bilirubin; age, platelet count, prothrombin index, AST, α-2-macroglobulin, hyaluronate, urea; bilirubin, γGT, hyaluronate, α-2-macroglobulin, age and gender; α-2-macroglobulin, hyaluronate and TIMP-1; age, hyaluronate, MMP-3 and TIMP-1
Non-patented biomarkers
FIB-4; APRI; AAR; Forns Index; Fibroindex; Lok index; ADAPT; NIS4; FibroScan-AST (FAST); NAFLD Fibrosis Score (NFS); BARD	age (yr) × AST [U/L]/(platelets [10^9/L] × (ALT [U/L])1/2; AST (/ULN)/platelet (10^9/L) × 100; AST/ALT ratio; 7.811–3.131 × ln(platelet count) + 0.781 × ln(GGT) + 3.467 × ln(age) - 0.014 × (cholesterol); 1.738–0.064 × (platelets [10^4/mm^3]) + 0.005 × (AST [IU/L]) + 0.463 × (gamma globulin [g/dl]); −5.56 - 0.0089 × platelet (10^3/mm^3) + 1.26 × AST/ALT ratio = 5.27 × INR; age, presence of diabetes, PRO-C3, and platelet count; microRNA-34a-5p, α2-macroglobulins, YKL-40, HbA1c; LSM, controlled attenuation parameter, AST; (−1.675 + 0.037 × age (yr) + 0.094 × BMI (kg/m^2) + 1.13 × IFG/diabetes (yes = 1, no = 0) + 0.99 × AST/ALT ratio - 0.013 × platelet count (×109/L) - 0.66 × albumin [g/dl]); BMI ≥28 = 1; AST/ALT ratio ≥ 0.8 = 2; diabetes = 1; score ≥ 2,
Emerging non-invasive tests based on serum biomarkers
LSM-spleen diameter to platelet ratio score (LSPS); Agile 4/Agile 3+	LSM, spleen diameter, and platelet count; LSM, platelet count, AAR, sex, diabetes

Table 1.

Noninvasive serum biomarkers in liver cirrhosis.

Abbreviation: AST, aspartate aminotransferase; ALT, alanine aminotransferase; TIMP-1, tissue inhibitor of metalloproteinases-1; GGT, gamma-glutamyl transferase; BMI, body mass index; IFG, impaired fasting glucose; MMP-3, matrix metalloproteinase-3; HbA1c, hemoglobin A1c; LSM, liver stiffness measurement; INR, international normalized ratio.

2. Conventional imaging methods

Conventional imaging modalities, including abdominal ultrasonography, computerized tomography (CT), and magnetic resonance imaging (MRI), play significant roles in the diagnostic landscape of liver cirrhosis. Standard ultrasonography, notable for its convenience, cost-effectiveness, and lack of radiation exposure, is aptly suited for routine surveillance of liver cirrhosis and its complications, such as ascites, portal hypertension, and hepatocellular carcinoma (HCC). However, its reliance on operator expertise introduces subjectivity and interobserver variability. Additional confounding factors impacting its accuracy include abdominal air, obesity, and hepatic inflammation [4]. CT imaging, in contrast, surpasses ultrasonography in delineating morphological alterations of smaller hepatic lesions and in detecting complications such as varices, vascular changes, and liver nodules. Moreover, CT quantifies the total area of spontaneous portosystemic shunts, serving as a prognostic indicator for hepatic encephalopathy (HE) and mortality [5]. Nonetheless, its routine use is limited due to the associated radiation exposure. MRI holds a superior edge over both abdominal ultrasonography and CT in the detection of smaller nodules. However, its application is constrained by contraindications in patients with metal implants and by its relatively higher cost, limiting its routine usage. Additionally, these conventional methods are typically inadequate for the identification of early-stage liver cirrhosis, thus underscoring the need for more sensitive diagnostic tools [6].

3. Serum biomarkers and algorithms

Serum biomarkers, encompassing both direct and indirect indicators, play a pivotal role in the noninvasive assessment of liver fibrosis and cirrhosis (Table 1). Direct biomarkers are those constituents released into the bloodstream during fibrogenesis or remodeling of the extracellular matrix. This category includes hyaluronic acid, the N-terminal peptide of pro-collagen III, pro-collagen 3 neoepitope (PRO-C3), and tissue inhibitor of metalloproteinase 1, among others [7]. Notably, algorithms derived from these direct markers, such as the enhanced liver fibrosis (ELF) test, PRO-C3, FibroTest, and FibroMeter, have been formulated and patented. These tests are known for their ability to predict decompensation linked to advanced fibrosis (defined as fibrosis stages 3 and 4). Conversely, indirect biomarkers typically reflect hepatocellular injury, inflammation, dysfunction, or portal hypertension [8]. Panels leveraging these indirect markers include fibrosis-4 index (FIB-4), aspartate aminotransferase-to-platelet ratio index (APRI) and aspartate aminotransferase-to-alanine aminotransferase ratio (AAR), Forns index, Lok index, or ADAPT. With the escalating global incidence of nonalcoholic fatty liver disease (NAFLD) worldwide, novel NITs have been proposed for the diagnosis of advanced fibrosis in these patients. The NAFLD fibrosis score (NFS), incorporating six elements, age, T2DM, BMI, AAR, albumin, and platelet count, has been developed from biopsy-proven NAFLD patients to identify liver fibrosis [9]. Additionally, the BARD score amalgamates AAR with BMI and the presence of T2DM. Its predictive value is constrained by the high prevalence of obesity and T2DM in the NAFLD cohort, leading to a majority of patients surpassing the defined threshold and resulting in a low positive predictive value (PPV) [10]. However, the diagnostic threshold of the NITs varies from different etiologies. For hepatitis B virus (HBV) or hepatitis C virus (HCV) infections, FIB-4 values below 1.45 and above 3.25 are interpreted as low and high stratification risks, respectively. In cases of NAFLD or alcohol-related liver disease (ALD), thresholds are set lower than 1.30 and higher than 2.67 for similar risk stratifications [11]. The accuracy of these serum biomarker panels in diagnostics is a subject of ongoing debate. Previous studies have indicated varying levels of efficacy, with some finding biomarkers, such as AAR predictive in certain contexts, such as chronic HCV infection, while others call for further validation and investigation, especially in diverse cohorts [12, 13]. In summary, while serum biomarker-based NITs exhibit high negative predictive values for excluding fibrosis or cirrhosis, their PPVs for diagnosis remain low, highlighting the need for continuous refinement and validation of these diagnostic tools [6].

4. Liver stiffness measurement (LSM) and spleen stiffness measurement (SSM)

Elastography methodologies have been advanced for LSM and SSM, encompassing techniques such as vibration-controlled transient elastography (VCTE), two-dimensional shear wave elastography (2D-SWE), point shear wave elastography (pSWE) (also named acoustic radiation force impulse imaging (ARFI)), and magnetic resonance elastography (MRE). The FibroScan® is capable of measuring stiffness values ranging from a minimum of 1.5 kPa to a maximum of 75.0 kPa. VCTE stands out for its convenience and cost-effectiveness in assessing fibrosis stages via LSM in patients. ARFI, in particular, offers insights into the relative structural stiffness of tissues juxtaposed with adjacent areas. This capacity to quantify absolute tissue modulus is pivotal for diverse clinical applications, facilitating the longitudinal monitoring of disease processes that manifest through tissue stiffening or softening, such as liver fibrosis and steatosis. Such capabilities are instrumental in determining the appropriate timing for initiating or ceasing treatment protocols and in staging disease progression and resolution [14]. Magnetic resonance elastography (MRE) has emerged as a significant advancement in this field [15, 16]. MRE’s capability to estimate LSM and SSM enables the evaluation of portal hypertension in cirrhotic patients, though it does not directly correlate with hepatic venous pressure gradient (HVPG) response [17]. However, limitations of MRE include its cost, time-intensive nature, infeasibility as a bedside procedure, and inapplicability in the presence of metal implants.

Baveno VI consensus recommended compensated advanced chronic liver disease (cACLD) patients with LSM < 40 kPa, and platelet count >150,000/μl could be waived from esophagogastroduodenoscopy (EGD). However, meta-analysis has indicated that the Baveno VI criteria, with a pooled sensitivity of 0.97 and specificity of 0.32, might misclassify high-risk varices (HRVs) [18]. To improve diagnostic accuracy, the Expanded Baveno VI criteria were proposed, which include a platelet count >110 × 10⁹ cells/μL and a TE < 25 kPa, potentially avoiding unnecessary EGD [19]. A recent study indicated that Baveno VI and Baveno VII criteria are not accurate for screening HRVs and CSPH in patients with HCC [20].

For patients with liver cirrhosis who develop clinically significant portal hypertension (CSPH), there is an elevated risk of decompensated events and liver-related mortality. The current gold standard for diagnosing CSPH is HVPG measurement [21]. LSM and SSM are valuable noninvasive alternatives for detecting and monitoring CSPH, correlating well with HVPG. LSM values >20–25 kPa and SSM values >40–45 kPa indicate a high likelihood of CSPH [22]. In a prospective study, Foucher et al. asserted the efficacy of LSM for diagnosing cirrhosis and its complications such as esophageal varices, HCC, and ascites in patients with chronic liver diseases, with values ranging from 12.5 to 75.5 kPa [23]. SSM serves as a direct and dynamic proxy for portal pressure, offering the potential to monitor the severity and improvement of portal hypertension and act as a surrogate marker for clinical outcomes. Notably, SSM appears more effective than LSM in monitoring treatment response in clinical trials aimed at reducing portal hypertension. The Baveno VII consensus recommended that an SSM ≤ 40 kPa could obviate the need for numerous unnecessary EGDs while also decreasing the incidence of diagnostic errors [22]. It has been reported that integrating SSM with the Baveno VI criteria resulted in increased specificity for excluding HRVs in HBV-related cirrhosis patients undergoing antiviral therapy [24]. Another research confirmed the efficacy of SSM in diagnosing CSPH and demonstrated that incorporating SSM could enhance the accuracy of the Baveno VII criteria in patients with metabolic-associated fatty liver disease (NAFLD) [25]. A meta-analysis showed that SSM in combination with Baveno VII can reduce the diagnostic gray zone for CSPH [26].

In clinical practice, the availability of NITs may guide patient evaluation approaches. LSM and SSM assessments using 2D-SWE have shown promising results for diagnosing cirrhosis in patients with autoimmune hepatitis, potentially less influenced by inflammation [27]. A single-center study from Japan indicated that SSM via ARFI can moderately accurately predict mortality in cirrhotic patients [28].

5. Emerging technologies

5.1 Imaging-based technologies

Recent advancements in imaging-based methodologies have emerged for the diagnosis of liver fibrosis and cirrhosis, leveraging ultrasonography, CT, MRI, and elastography. A noninvasive approach involving platelet count and spleen diameter has been proposed for predicting esophageal varices in patients with cirrhosis [29]. Subsequent research endeavors have sought to validate its effectiveness in both detecting and excluding esophageal varices, regardless of the underlying cause and severity of liver cirrhosis [30, 31]. Additionally, cutting-edge three-dimensional technologies are being explored. A noteworthy prospective head-to-head study by Loomba et al. demonstrated that 3D magnetic resonance elastography (3D-MRE) significantly outperforms its 2D counterpart in diagnosing advanced fibrosis, evidenced by an impressive AUC of 0.981 [32]. Despite its diagnostic superiority, the cost-effectiveness of 3D-MRE warrants further exploration to assess its viability for widespread clinical application. A novel ultrasound-based approach, three-dimensional multi-frequency shear wave absolute vibro-elastography (3D S-WAVE) with a matrix array, has been introduced. This technique has shown diagnostic capabilities comparable to MRE for liver fibrosis but offers advantages in terms of accessibility and cost. This innovation makes advanced diagnostic technology potentially available to a broader patient demographic, bridging the gap between high end, costly diagnostics and the practical needs of diverse healthcare settings [33]. Based on LSM, Kim et al. proposed a novel noninvasive model, the LSM-spleen diameter to platelet ratio score (LSPS), for predicting HRVs in HBV-related liver cirrhosis [34]. The variceal risk index (VRI), consisting of LSM, spleen diameter, and platelet count, has been effective in identifying CSPH and HRVs in patients with biopsy-proven compensated cirrhosis [35]. In a multicenter study, Sanyal et al. developed two novel scores, Agile 4 and Agile 3+, combining LSM with laboratory and demographic data to diagnose cirrhosis in NAFLD patients, demonstrating superior diagnostic capabilities compared to FIB-4 and LSM alone [36]. Another study also confirmed the superiority of Agile 3+ for the prediction of disease severity and liver-related events in NAFLD [37]. Several studies have compared the diagnostic capabilities of noninvasive models in chronic liver diseases, but the results remain controversial. A comparative analysis showed that a model combining LSM with SSM was more effective than other noninvasive models (LSPS, VRI, AAR, and Baveno VI criteria) in predicting HRVs in patients with viral cirrhosis [38]. Cho et al. assessed the diagnostic and prognostic values of various noninvasive fibrosis markers in patients with alcohol-related cirrhosis. Their findings indicated that LSM and LSPS outperformed other noninvasive fibrosis indices in predicting CSPH in patients with compensated cirrhosis [39]. However, the study by Colecchia et al. revealed that SSM was a valuable noninvasive tool for assessing varying degrees of portal hypertension and esophageal varices in HCV-related cirrhotic patients, outperforming the platelet count/spleen diameter and LSPS [40]. Notably, the Lok index was uniquely effective in enhancing the predictive accuracy of the model for end-stage liver disease (MELD) score for overall mortality in decompensated cirrhosis cases [39]. The LiverRisk score, which comprised age, sex, fasting glucose, cholesterol, AST, ALT, GGT, and platelet count, could predict liver fibrosis and liver-related outcomes in the general population [41].

5.2 Genetic prediction models

Prior research has suggested the potential application of gene signatures in the risk stratification for liver cirrhosis, particularly in chronic HBV and HCV [42, 43]. These studies have indicated that specific genetic markers could be instrumental in categorizing the risk levels associated with liver cirrhosis. One notable finding is the role of dynamic changes in the DNA methylation of the peroxisome proliferator-activated receptor gamma (PPARγ) in fibrogenesis. Such alterations in methylation patterns could serve as a promising biomarker for stratifying the risk of fibrosis progression in NAFLD [44]. Furthermore, the prevalence of a missense variant in the patatin-like phospholipase domain-containing protein 3 (PNPLA3), specifically the p.I148M variant, is significant (>20%) in the global population. This genetic variation has been associated with an increased risk of steatosis, cirrhosis, and HCC across various etiologies, emphasizing the genetic predisposition to these liver conditions [45]. Genetic risk scores, which amalgamate multiple risk polymorphisms, present a promising tool for identifying individuals predisposed to cirrhosis from an early age. A recent study incorporating 12 genetic risk alleles found that patients in the highest 20% bracket of polygenic risk were more than twice as likely to develop cirrhosis compared to those in the lowest 20% risk group. This finding underscores the potential of polygenic risk scores to quantify an individual’s comprehensive genetic risk for cirrhosis [46]. The integration of polygenic risk scores, encompassing hundreds or thousands of genetic risk alleles, represents a significant advance in understanding the complex genetic underpinnings of liver cirrhosis [47]. This approach aims to provide a holistic assessment of an individual’s genetic predisposition to cirrhosis, offering valuable insights for both preventive strategies and personalized medical interventions.

6. Microbiome

The alterations in the diversity and composition of the gut microbiome and its derivatives are increasingly recognized as crucial factors in the initiation and progression of liver diseases [48, 49]. In an innovative approach, Loomba et al. employed whole-genome shotgun sequencing of DNA from fecal samples, coupled with random forest analysis, to develop a predictive model. This model effectively identifies advanced fibrosis in patients with NAFLD, signifying a significant leap in noninvasive liver disease diagnostics [50]. Furthermore, the analysis of gut microbiome signatures has yielded promising results in the prediction of liver cirrhosis. These microbiome-based models exhibit considerable accuracy, underscoring the potential of gut microbiota as a biomarker for liver health and disease [51]. This emerging field, bridging gastroenterology and hepatology, opens new avenues for early detection and intervention strategies in liver cirrhosis, leveraging the intricate interplay between the gut microbiome and liver function.

6.1 Omics and artificial intelligence

The utilization of a combined noninvasive serum protein signature, employing SomaScan® technology, has demonstrated proficiency in accurately predicting HVPG response in HCV liver cirrhosis post-sustained virological response (SVR) [52]. Additionally, the serum caspase-cleaved fragment of keratin-18 has been identified as a marker correlated with the histological severity and liver-related mortality, not only in alcohol-related liver disease but also in NAFLD [53, 54]. Recent advancements in the field have seen the integration of mass spectrometry (MS)-based proteomics with machine learning (ML) methodologies to effectively detect significant fibrosis, showcasing robust performance [55]. In a pioneering study by Corey et al., the application of aptamer-based proteomics technology revealed that the ADAMTSL2 protein could serve as a biomarker for identifying individuals at risk of nonalcoholic steatohepatitis and fibrosis [56]. Additionally, a convolutional neural network model based on trichrome-stained liver biopsy slides has been developed to predict CSPH in patients with biopsy-proven NASH cirrhosis [57]. Another recent advancement involves an ML model based on histologic features, capable of accurately predicting HVPG, CSPH, the development of varices, and changes in HVPG in NASH cirrhosis [58]. Liu et al. innovatively utilized ML to accurately extract liver capsule features from high-frequency ultrasound images, aiding in the diagnosis of cirrhosis [59]. ML models performed better overall than FibroScan, FIB-4, FAST, and NFS. ML could be an effective tool for identifying clinically significant liver fibrosis and cirrhosis in patients with NAFLD. A recent study showed that LSM combined with an ML model could accurately identify MASLD-related cirrhosis and advanced fibrosis [60].

In conclusion, noninvasive tests offer a more affordable, convenient, and safer alternative for screening and monitoring liver conditions compared to liver biopsy. The accumulating body of evidence supports their efficacy in facilitating informed clinical decision-making. However, it is noteworthy that the diagnostic performance of these noninvasive tests exhibits variability across different studies. To optimize their diagnostic accuracy and clinical utility in assessing liver cirrhosis, further well-designed multicenter research studies for the validation of these methodologies are imperative.