The Elucidating Pathways of Steatohepatitis (EPoS) staging system: A reproducible 7-tiered fibrosis scoring system in MASLD

Abstract

Background & Aims

Fibrosis in metabolic dysfunction-associated steatotic liver disease (MASLD) is histologically staged using the 5-tiered NASH CRN system, which has limited resolution for advanced fibrosis stages that are strongly associated with liver-related outcomes. We developed and validated an expanded histological staging system designed to more accurately reflect the spectrum of MASLD fibrosis.

Methods

Nine expert hepatopathologists from the Elucidating Pathways of Steatohepatitis (EPoS) Histopathology Group reviewed MASLD liver biopsies to define, by consensus, the fibrosis stages of a new histological scoring system. To assess validity, correlations with quantitative collagen proportionate area (CPA), enhanced liver fibrosis (ELF) score, and liver stiffness measurement (LSM) were evaluated. To assess inter-observer variability and intra-observer agreement, each pathologist independently reviewed 45 cases (glass slides) using the EPoS staging system. After three months, the same slides were digitized and re-evaluated by each pathologist, blinded to their previous assessments.

Results

The EPoS staging system comprises seven stages (0–6), covering the full spectrum of MASLD fibrosis. Compared with the NASH CRN system, stages 1a, 1b, and 1c were combined into a single stage, while stages 3 and 4 were each subdivided into two additional stages. The EPoS system showed significant correlations with CPA, ELF score, and LSM values (all p <0.001). Mean inter-observer agreement for EPoS staging was excellent (κ = 0.84). Agreement remained high on digital evaluation (κ = 0.80), while intra-observer agreement was similarly strong (mean κ = 0.85; range 0.79–0.89).

Conclusions

The EPoS fibrosis staging system is a robust expanded scoring method that accurately reflects histological CPA, LSM, and ELF values. EPoS staging may be useful for assessing fibrosis in MASLD natural history studies and therapeutic trials. Digital pathology provides results comparable to glass slides for reproducible fibrosis staging.

Impact and implications

The traditional 5-tiered NASH CRN system for staging metabolic dysfunction-associated steatotic liver disease (MASLD)-related fibrosis has limited resolution for advanced stages, which are most strongly associated with liver-related outcomes. The newly developed and validated EPoS expanded 7-tiered staging system provides greater granularity for advanced fibrosis and more accurately reflects fibrosis progression in MASLD. The EPoS system is expected to improve the accuracy of fibrosis staging for patient selection in MASH (metabolic dysfunction-associated steatohepatitis) clinical trials and for assessing changes in fibrosis during both natural disease progression and therapeutic interventions, as well as in routine pathology practice. In addition, we demonstrate that digital pathology slides can be reliably used for reproducible fibrosis staging in MASLD, yielding results comparable to those obtained from glass slides, with direct practical implications for liver biopsy interpretation using scanned images.

Graphical abstract

Highlights

• •EPoS is a 7-tiered (0-6) fibrosis scoring system for MASLD.
• •EPoS has excellent inter-observer variability and intra-observer agreement.
• •EPoS stages accurately reflect liver tissue collagen quantitation in MASLD biopsies.
• •EPoS stages correlate with non-invasive liver stiffness measurements and enhanced liver fibrosis scores.
• •Digital slides are comparable to glass slides for reproducible fibrosis staging in MASLD.

Introduction

Liver fibrosis, a complex and dynamic tissue remodeling process, is the common hallmark of chronic liver diseases, among which metabolic dysfunction-associated steatotic liver disease (MASLD), previously known as non-alcoholic fatty liver disease (NAFLD),¹ is becoming the most prevalent worldwide.² Progression of MASLD leads to cirrhosis and its main complications including portal hypertension, liver failure and malignant transformation.³ In this clinical setting, liver fibrosis exhibits distinctive features, also observed in cases of excessive alcohol consumption, including early development of centrilobular sinusoidal/pericellular fibrosis.⁴

In MASLD, histological liver fibrosis stage is a major prognostic factor^5,6 and is a regulatory-accepted, reasonably likely surrogate endpoint that is widely adopted in therapeutic trials.⁷ Accordingly, its detection, accurate evaluation and ability to appreciate progression/regression is critical in routine practice but also in research studies and drug development.⁸ Traditionally, and independent of the risk factors associated with chronic liver diseases, assessment of liver fibrosis has relied on semi-quantitative scoring systems applied on liver tissue sections stained mainly with Masson trichrome (MT) or Sirius red histochemical stains.⁹ The Non-alcoholic Steatohepatitis Clinical Research Network (NASH CRN) scoring system, based on a 5-tiered scale from F0 (normal) to F4 (cirrhosis) has been used for staging fibrosis in MASLD natural history studies and in clinical trials.¹⁰ Although widely accepted and commonly used, this staging scoring system has shown its limitations, mostly lacking granularity, especially in advanced stages (F3 and F4) which are most strongly linked to liver-related outcomes.^5,6 In paired liver biopsies in clinical trials and natural history studies, subtle changes in fibrosis severity within stages F3 and F4 are not detected using the NASH CRN scoring system.¹¹ Alternative quantitative systems based on computer-assisted morphometric analysis of whole slide images (digital histological slides), such as measurement of collagen proportionate area (CPA), have demonstrated their added value, providing an objective, accurate and continuous fibrosis assessment. Studies evaluating CPA in MASLD liver biopsies have shown that it correlates well with histological stage and liver stiffness and is an independent predictor of long-term outcomes in MASLD.^12,13 Digital technology-based approaches, some of them requiring complex methods of artificial intelligence tools, have been applied to liver pathology, especially for improving the accuracy of diagnosis of steatohepatitis and reducing intra- and interobserver variability.[14], [15], [16]

Despite the remarkable advances in the development of non-invasive tests able to stage fibrosis,^17,18 liver biopsy remains the reference standard for accurately grading and staging MASLD as it provides a comprehensive picture of tissue changes, assessing necroinflammatory activity grade, fibrosis stage, and potential concurrent disease(s).^19,20 In addition, regulatory agencies, such as the EMA and the US FDA, grant conditional approval for new therapies for metabolic dysfunction-associated steatohepatitis (MASH) based on histological outcomes.^21,22 Accordingly, we aimed to develop and validate an expanded staging system (EPoS staging system) that more accurately reflects the different stages of fibrosis in liver biopsies from patients with MASLD. Given the growing interest in digital pathology, we also aimed to compare fibrosis assessment based on the EPoS system using both glass (conventional pathology) and whole slide images (WSIs) (digital pathology) to evaluate the reliability of digital histology slides for staging fibrosis.

Materials and methods

In a first step, nine expert European liver pathologists (the EPoS Histopathology Group), most of whom had previously collaborated to validate the SAF histological scoring system for MASLD²³ and were members of the LITMUS (Liver Investigation: Testing Marker Utility in Steatohepatitis) Histopathology Group,^18,24 physically met at the Headquarters of the European Society of Pathology, Brussels, Belgium and jointly reviewed a set of 45 anonymized MASLD liver biopsies to define in consensus the various stages of the new EPoS fibrosis staging system. The liver biopsies were selected to cover the whole spectrum of MASLD, ranging from simple MASL to MASH and from no fibrosis to advanced fibrosis and cirrhosis, and were adequate in size and portal tract/terminal hepatic vein numbers. For each biopsy, one glass slide stained with MT was used. All glass histological slides were scanned at x20 magnification to produce a digital histological image of each liver biopsy. The pathologists were blinded to the clinical data related to the selected liver biopsies.

Following the development of the EPoS fibrosis staging system, its reliability was first assessed by evaluating its correlation with the quantitative collagen proportionate area (CPA) in a set of 2,968 liver biopsies from patients with MASH enrolled in the phase III RESOLVE-IT clinical trial of elafibranor in MASH.²⁵ All biopsies had been assessed as adequate for size, staining and scanning quality by a liver pathologist (PB) and an imaging engineer. Fibrosis stage was evaluated on glass slides by a single expert liver pathologist (PB) using both the 5-tiered NASH CRN (stages 0-5)¹⁰ and the new EPoS histological staging systems.

CPA was quantified on WSI from Sirius red-stained sections from each biopsy scanned at x20 magnification, as previously described.¹³ Briefly, a specific software (Halo® Module for area quantification, Indica Labs, Albuquerque, NM, USA) selected the overall tissue area on the digital slide and calculated this in pixels. Subsequently, the areas of stained collagen, avoiding areas of structural collagen (i.e. portal tracts, large vessels, liver capsule) were also expressed in pixels. The “fibrosis ratio” between the two areas was expressed as the relative proportion (%) of collagen in the liver tissue or CPA. The slides were processed in batches to allow normalization of the data for each batch according to the merged values of the previous batch(es).

To further assess the reliability of the EPoS fibrosis staging system, we evaluated its correlation with liver stiffness measurement (LSM) by vibration-controlled transient elastography (FibroScan, Echosens, Paris, France) and with the non-invasive fibrosis multimarker test enhanced liver fibrosis (ELF). The ELF test is based on hyaluronic acid, tissue inhibitor of matrix metalloproteinase-1, and amino-terminal propeptide of procollagen type III serum values. LSM and ELF correlation data were collected from the LITMUS metacohort section of the European MASLD Registry, a cohort of prospectively recruited individuals with biopsy-proven MASLD. Details of the cohort and methodology for vibration-controlled transient elastography and ELF measurements have been published previously.^18,26

To evaluate inter-observer and intra-observer variability for the newly developed EPoS staging system and to compare those with the NASH CRN staging system, each pathologist independently reviewed the initial set of 45 MT-stained MASLD biopsies on glass slides.

To assess the reliability of digital slides for accurate fibrosis staging, WSI (digital slides) from each of the 45 MT-stained MASLD liver biopsies were evaluated electronically by each pathologist, 3 months after reviewing the corresponding glass slides, and blinded to their first assessment of those.

The Student-Newman-Keuls test was applied for all pairwise statistical comparisons used to assess the correlation between CPA and histological staging. Correlations between EPoS or NASH CRN fibrosis stages with LSM and ELF scores were assessed using Spearman's rank correlation coefficient. The level of significance was set at p = 0.05. The inter-observer and intra-observer variability was evaluated using the kappa score.

Results

Development of the EPoS histological staging system

The EPoS staging system, developed to cover the whole spectrum of fibrosis in MASLD, included seven stages (0 to 6), where 0 corresponds to normal or nearly normal liver and 6 corresponds to cirrhosis. Compared to the NASH CRN scoring system, stages 1a, b and c were grouped together, while stages 3 and 4 were each subdivided into two additional stages to increase granularity in more advanced stages. Schematically, the new 7-tier fibrosis staging system is shown in Fig. 1.

Fig. 1.

Schematic representation of the EPoS fibrosis staging system and its differences from the NASH CRN staging system for MASLD.

In the first column, the NASH CRN stages are shown on the left and the equivalent EPOS stages on the right (red front). In the second column (Comments), details for the evaluation of fibrosis are described and rationale for grouping together the NASH CRN stage 1 sub-stages a, b, c is given. EPoS, Expanded Prognostic Staging; MASLD, metabolic dysfunction–associated steatotic liver disease; NASH CRN, Non-alcoholic Steatohepatitis Clinical Research Network.

In more detail, EPoS stage 0 is normal liver or nearly normal liver, the latter including minimal zone 3 sinusoidal fibrosis (affecting ≤2 zone 3 areas per biopsy with only delicate sinusoidal fibrosis but without pericellular fibrosis) or minimal portal fibrosis without periportal expansion (one focus per biopsy). Minimal fibrosis would have been scored as stage 1 using the NASH CRN system.

EPoS stage 1 is zone 3 (centrilobular) sinusoidal/pericellular fibrosis or isolated portal/periportal fibrosis. Compared to NASH CRN, stages 1a, b and c were grouped together because their histological evaluation may be affected by sampling error, have poor reproducibility and their clinical relevance is not clear.

EPoS stage 2 includes extensive sinusoidal or periportal fibrosis reaching acinar zone 2 or portal/periportal and zone 3 fibrosis. EPoS stage 2, according to the first definition, would have been scored as stage 1 using the NASH CRN system, while the second definition is similar to NASH CRN stage 2.

EPoS stages 3 and 4 are defined by few septa (≤2/10 mm length of biopsy) and many septa (>2/10 mm length of biopsy) without hepatocellular nodule formation, respectively. All types of septa are considered (portal-portal, portal-central, central-central). EPoS stages 3 and 4 would have been scored as stage 3 using the NASH CRN system.

EPoS stage 5 is defined by many septa with an occasional (up to 2) completely circumscribed hepatocellular nodule.

EPoS stage 6 is cirrhosis with diffuse nodular architecture and is further subdivided into three substages (6a, 6b, 6c) according to the Laennec staging system.^27,28

Representative histological images from each one of the 7 stages (0-6) of fibrosis according to the EPoS staging system for MASLD are shown in Fig. 2.

Fig. 2.

Representative histological images from each one of the 7 stages (0-6) of fibrosis according to the EPoS staging system for MASLD (Masson trichrome stain: collagen fibres are shown in blue, hepatocytes in red, fat droplets within hepatocytes are white.

EPoS, Elucidating Pathways of Steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease).

Correlation of the 7-tiered EPoS staging system with CPA in MASLD biopsies

Table 1 summarises the results of normalised CPA measurements by EPoS fibrosis stage in 2,968 digitised MASLD liver biopsies and includes corresponding results stratified by NASH CRN fibrosis stage from the same set of biopsies. CPA measurements in MASLD biopsies were significantly correlated with EPoS fibrosis stage (F ratio 110.29, p <0.001) (Fig. 3A,B). Similarly, CPA measurements in MASLD biopsies were significantly correlated with NASH CRN fibrosis stage (F ratio 132.24, p <0.001) (Fig. 3C,D). Based on the lower F ratio value, the correlation between CPA and EPoS stages appears better than the correlation between CPA and NASH CRN stages.

Table 1.

Normalised CPA according to EPOS and NASH CRN fibrosis staging.

EPOS stage	0	1	2	3	4	5	6
N (2,968)	62	362	1,067	624	614	196	43
Mean CPA	3.709	4.555	5.426	7.09	9.828	12.552	17.683
Median CPA	3.389	3.844	4.687	6.17	8.816	11.569	18.628
NASH CRN stage	0	1	2	3		4
N (2,968)	62	374	1,138	1332		62
Mean CPA	3.709	4.567	5.493	9.109		16.529
Median CPA	3.389	3.844	4.796	7.927		15.023

CPA, collagen proportionate area.

Fig. 3.

Correlation of CPA with fibrosis stage in MASLD biopsies using the EPoS and NASH CRN staging systems.

(A) Scatterplot diagram showing CPA measurements according to EPoS stage; (B) Line diagram with whiskers showing mean CPA measurements according to EPoS stage. F ratio 110.29, p <0.001; (C) Scatterplot diagram showing CPA measurements according to NASH CRN stage; (D) Line diagram with whiskers showing mean CPA measurements according to NASH CRN stage. F ratio 132.24, p <0.001. EPoS, Elucidating Pathways of Steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease; NASH CRN, Non-alcoholic Steatohepatitis Clinical Research Network.

Tables S1 and S2 show Student-Newman-Keuls test results for pairwise comparisons between CPA and EPoS stage and between CPA and NASH CRN stage, accordingly.

Correlation of the 7-tiered EPoS staging system with LSM and ELF scores in patients with MASLD

The LITMUS metacohort consisted of 946 individuals with biopsy-proven MASLD, including 58.4% (547/937) with MASH (histology data for 9 patients incomplete). ELF and LSM values were available in 384 and 498 individuals, respectively. Table S3 summarises the baseline characteristics of the cohort according to EPoS fibrosis stage. Baseline characteristics of the LITMUS metacohort according to NASH CRN fibrosis stage are summarised in Table S4.

The EPoS staging system correlated significantly with ELF scoring (rho 0.52, p <0.001) (Fig. 4A) and LSM values (rho 0.60, p <0.001) (Fig. 4B). Correlations of the NASH CRN staging with ELF and LSM were also significant in the LITMUS metacohort (rho 0.52, p <0.001 and rho 0.59, p <0.001, respectively, Fig. S1).

Fig. 4.

Boxplots showing the relationship between EPoS fibrosis stage and non-invasive fibrosis markers in patients with biopsy-proven MASLD.

Boxplot diagrams of (A) ELF scores stratified by EPoS fibrosis stage (0-6) in 384 patients with biopsy-proven MASLD. Spearman rho 0.52, p <0.001; (B) LSM in kPa stratified by EPoS fibrosis stage (0-6) in 498 patients with biopsy-proven MASLD. Spearman rho 0.60, p <0.001. Boxes indicate median and interquartile range; whiskers represent minimum and maximum values, excluding outliers.

EPoS fibrosis stages strongly correlated with NASH CRN fibrosis stages (rho 0.98, p <0.001) (Fig. S2).

Inter-observer variation for EPoS fibrosis staging system using glass slides and digital slides (e-slides)

Table 2 shows the inter-observer variation of the nine expert liver pathologists when using the EPoS staging system on glass slides of 45 MASLD cases. The mean kappa score (7-tier) was 0.84 (very good, almost perfect agreement). The mean kappa score (5-tier) when applying the NASH CRN staging system was lower at 0.78 (good, substantial agreement).

Table 2.

Inter-observer variation between nine expert liver pathologists (OBS) applying the EPoS 7-tier staging system using glass slides (mean kappa score = 0.84).

	OBS1	OBS2	OBS3	OBS4	OBS5	OBS6	OBS7	OBS8	OBS9
OBS1		0.88	0.86	0.85	0.87	0.87	0.77	0.78	0.89
OBS2	0.88		0.89	0.85	0.89	0.92	0.79	0.83	0.92
OBS3	0.86	0.89		0.86	0.92	0.90	0.81	0.80	0.91
OBS4	0.85	0.85	0.86		0.82	0.91	0.87	0.83	0.89
OBS5	0.87	0.89	0.92	0.82		0.89	0.75	0.75	0.88
OBS6	0.87	0.92	0.90	0.91	0.89		0.84	0.88	0.93
OBS7	0.77	0.79	0.81	0.87	0.75	0.84		0.78	0.82
OBS8	0.78	0.83	0.80	0.83	0.75	0.88	0.78		0.86
OBS9	0.89	0.92	0.91	0.89	0.88	0.93	0.82	0.86

Table 3 shows the inter-observer variation of the expert liver pathologists when using the EPoS staging system using e-slides of 45 MASLD cases. The mean kappa score (7-tier) was 0.80 (very good, almost perfect agreement).

Table 3.

Inter-observer variation between nine expert liver pathologists applying the EPoS staging system using digital slides (e-slides) (mean kappa score = 0.80).

	OBS1	OBS2	OBS3	OBS4	OBS5	OBS6	OBS7	OBS8	OBS9
OBS1		0.83	0.80	0.68	0.80	0.84	0.73	0.81	0.84
OBS2	0.83		0.81	0.76	0.91	0.84	0.84	0.88	0.88
OBS3	0.80	0.81		0.76	0.77	0.91	0.80	0.86	0.84
OBS4	0.68	0.76	0.76		0.79	0.75	0.83	0.79	0.82
OBS5	0.80	0.91	0.77	0.79		0.84	0.81	0.88	0.89
OBS6	0.84	0.84	0.91	0.75	0.84		0.82	0.91	0.88
OBS7	0.73	0.84	0.80	0.83	0.81	0.82		0.83	0.83
OBS8	0.81	0.88	0.86	0.79	0.88	0.91	0.83		0.85
OBS9	0.84	0.88	0.84	0.82	0.89	0.88	0.83	0.85

Intra-observer variation for EPoS fibrosis staging system: comparison between glass slides and digital slides (e-slides)

The intra-observer variation was minimal using the EPoS 7-tier staging system on glass slides and e-slides, with kappa values of agreement ranging from 0.79 to 0.89 (mean k coefficient 0.85, very good, almost perfect agreement), as shown in Table 4. The intra-observer agreement when using the 5-tier NASH CRN system on glass and e-slides was lower with a mean kappa score of 0.70 (good/substantial agreement).

Table 4.

Intra-observer variation for each of the nine expert liver pathologists (OBS) comparing evaluation of 45 MASLD biopsies with the EPoS staging system on glass slides versus digital slides (e-slides).

	Mean Kappa co-efficient = 0.85 (range 0.79-0.89)
	OBS1	OBS2	OBS3	OBS4	OBS5	OBS6	OBS7	OBS8	OBS9
k score	0.80	0.85	0.83	0.79	0.85	0.86	0.82	0.86	0.89

Discussion

In MASLD, histological analysis of liver biopsies addresses several important issues, including the assessment of fibrosis severity (staging), which is a key independent prognostic factor and a crucial endpoint in clinical trials.⁷ Accurately staging fibrosis for patient eligibility in clinical trials and evaluating any changes in fibrosis – however subtle – during natural progression of the disease or therapeutic interventions is essential. Staging fibrosis in chronic liver diseases typically employs semiquantitative scoring systems that assess the degree of liver fibrosis and changes in parenchymal and vascular architecture. For patients with MASLD, the NASH CRN fibrosis staging system, which uses a 5-tiered score, is commonly utilized, though interobserver concordance varies across studies.¹⁰

To improve semiquantitative liver fibrosis assessment, we developed and validated an expanded staging system (EPoS) based on a 7-tiered score. This situation is reminiscent of developments in histological evaluation of liver fibrosis in the 1990s for patients with chronic viral hepatitis, which led to more granular staging systems.²⁹ The EPoS system notably includes sub-staging of advanced liver fibrosis (F3 and F4 according to NASH CRN), which is closely linked to prognosis.⁵ In a prospective multicenter study of 150 liver biopsies, Kuwashiro et al.³⁰ assessed histological scoring concordance between local and central pathologists specializing in liver diseases. They found a 43.3% concordance rate for fibrosis staging, with local pathologists diagnosing stages 3 or 4 in 25.3% of cases compared to 13.3% by experienced pathologists. This underscores the need to optimize evaluation criteria with clear and accurate stage definitions.

In EPoS stages 3 and 4 (equivalent to F3 in NASH CRN), the presence of fibrous septa is evaluated according to biopsy length (10 mm), allowing for an objective quantitative assessment, which also helps gauge sample adequacy. For EPoS stages 5 and 6 (corresponding to F4 in NASH CRN), the number of completely circumscribed hepatocellular nodules (with a cut-off of 2) is considered. In addition, to reflect the diverse nature of cirrhosis, EPoS stage 6, defined by diffuse nodular architecture, is further subdivided into three substages (6a, 6b, 6c) according to the Laennec staging system.^27,28 This histological subclassification of cirrhosis, based on fibrous band thickness and parenchymal nodule size, applied to the METAVIR scoring system, has been shown to improve the prediction of prognosis and portal hypertension-related complications.³¹

For EPoS stage 1, stages 1a, b, and c according to NASH CRN are grouped together, as distinguishing between 1a (mild centrilobular sinusoidal fibrosis only seen with MT stain) and 1b (moderate centrilobular sinusoidal fibrosis detected on routine H&E stain) may depend on technical factors like staining intensity, which can affect reproducibility. Notably, stage 1c was included in the NASH CRN scoring system primarily to allow for more accurate histological evaluation of pediatric MASLD biopsies, since children with MASLD are more susceptible to early portal/periportal fibrosis.³²

Alternative quantitative fibrosis measurement methods have already demonstrated their effectiveness in classifying liver biopsies and predicting long-term outcomes for patients with MASLD.^12,13 To assess the EPoS staging system as a quantitative measure of fibrosis extent we measured CPA in a large number of MASH biopsies from the phase III RESOLVE-IT clinical trial and found excellent concordance between the new expanded 7-tiered EPoS fibrosis staging system and the amount of fibrosis, thus confirming its enhanced semiquantitative value. The correlation between CPA and NASH CRN staging was also excellent albeit weaker compared to that of EPoS. Previous comparison of the 5-tiered NASH CRN system with CPA in MASLD cases has shown good to excellent correlation based on results from two liver pathologists.³³

The EPoS staging system is a clinically relevant histological method for fibrosis evaluation in MASLD given that EPoS fibrosis stages correlate significantly with ELF scores and LSM in patients with biopsy-proven MASLD. The EPoS staging showed slightly stronger correlation with LSM compared to NASH CRN, while there was no difference between the two staging systems regarding ELF score. In the same patient cohort, it had previously been shown that LSM had acceptable accuracy (area under the curve 0.83 [0.80-0.86]) for detection of advanced fibrosis based on NASH CRN staging.¹⁸

The interobserver reproducibility of the EPoS staging system was excellent between expert liver pathologists (kappa = 0.84), while the mean kappa value for the NASH CRN fibrosis staging system (5-tier) was only good (kappa = 0.78), indicating that the applicability of an expanded staging system for MASLD does not pose additional difficulty in reaching agreement among observers. Inter-observer agreement was comparable to that of the NASH CRN staging system (kappa = 0.83) in its original publication¹⁰ and other expanded staging systems that have been used in assessing fibrosis in MASLD, such as the Ishak fibrosis staging system (index intraclass correlation coefficient 0.87) and the alternate NAS fibrosis staging system that expands bridging fibrosis (intraclass correlation coefficient 0.83).¹¹

The development of scanning systems generating digital slides (e-slides) has enabled the transition from conventional to digital pathology, facilitating easy access, remote histological interpretation, particularly for consult review and clinical trials, and increasing use of computational analyses, including automated quantitative assessment.³⁴ Despite increasing interest in digital analysis to quantify collagen deposition in the liver with various tools, their integration into routine practice will require time and dedicated software, and their impact on pathological workflows needs careful evaluation.¹⁶ We have shown that digital slides provide material as suitable as glass slides for staging fibrosis in MASLD, based on the high reproducibility of the 7-tiered EPoS staging system scores for each of the nine observers. These results are in keeping with recently published data comparing agreement between three liver pathologists scoring digital slides on the AISight™ whole slide image management system and on glass slides of the same cases using the NASH CRN system that showed non-inferior average agreement of digital scoring compared to glass scoring (weighted kappa 0.64 vs. 0.604, respectively). Furthermore, intra-rater agreement kappa values between digital and glass scoring ranged from 0.837 to 0.901 for fibrosis.³⁵

Owing to increased granularity for advanced stages of fibrosis, the new EPoS classification should allow for detection of fibrosis changes with higher sensitivity than the NASH CRN staging system, although this should be specifically tested. Comparative studies in the field of chronic hepatitis between 5-tiered (i.e. Metavir) and 7-tiered (i.e. Ishak) fibrosis staging systems have shown that the latter are more suitable for clinical trials due to their higher granularity and that they are equally good for staging with excellent correlation between them.[35], [36], [37], [38] The use of a histological scale with higher sensitivity to change is of critical importance when testing antifibrotic medications, as phase II trials with liver biopsy are of limited size (and often duration) thus carrying the risk of falsely negative results.³⁹ Data on the ability of the EPoS 7-tiered staging system to detect fibrosis improvement after non-pharmacological interventions are available from a study on bariatric surgery.⁴⁰ In that report, 45% of patients with advanced fibrosis (stages 3 and 4 according to NASH CRN) still had advanced fibrosis a mean of 5.5 years after a successful surgical procedure. However, when the same slides were assessed using the EPOS 7-tier classification, half of the patients had a one stage reduction in fibrosis, thus confirming fibrosis regression undetectable by the NASH CRN staging system.⁴⁰ The prognostic value of the EPoS 7-tiered classification will need to be carefully investigated in future studies. A large multicenter study has shown the ability of this new fibrosis staging system to predict liver related complications and liver-related death.⁴¹

The limitations of our study include the relatively small number of MASLD liver biopsies on which the EPoS staging system was developed (n = 45). However, this number exceeds that used in the design and validation of the NASH CRN staging system in adults with MASLD (n = 32)¹⁰ and is comparable to the sample size in studies assessing inter-observer variability of the SAF histological scoring system (n = 40).²¹ The EPoS staging system has been developed on a set of liver biopsies from adults with MASLD and its applicability in pediatric MASLD remains to be assessed.

Further validation in follow-up or post-treatment MASH biopsies, to account for possible fibrosis regression, is needed. Nevertheless, this is the first 7-tiered staging system in MASLD designed and validated for use by expert liver pathologists, most with experience in MASH clinical trials, that correlates well with objective quantitative measurements of fibrosis and can be used with excellent inter-observer agreement on both glass and digital pathology slides.

The EPoS fibrosis staging system is an expanded histological scoring system which may prove useful to assess fibrosis changes in natural history studies or in short-term therapeutic trials. Its usefulness in routine practice, its generalizability and its prognostic significance in MASLD remain to be evaluated.

Abbreviations

CPA, collagen proportionate area; ELF, enhanced liver fibrosis; EPoS, Expanded Prognostic Staging; LSM, liver stiffness measurement; MASH, metabolic dysfunction–associated steatohepatitis; MASLD, metabolic dysfunction–associated steatotic liver disease; MT, Masson trichrome; NASH CRN, Non-alcoholic Steatohepatitis Clinical Research Network; WSI, whole slide image.

Authors’ contributions

PB, DGT, QMA and VR conceptualized the study; DGT, PB, JA, SD, ASHG, MG, CL, PS, LT, AB, KW, QMA, VR and VP assisted in data collection and data analysis; DGT, VP, QMA and VR drafted the manuscript; all authors critically revised the manuscript and approved the final version for publication.

Data availability

The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request.

Financial support

The project leading to these results has been supported by the EPOS (Elucidating Pathways of Steatohepatitis) consortium funded by the Horizon 2020 Framework Program of the European Union under Grant Agreement 634413 and the LITMUS (Liver Investigation: Testing Marker Utility in Steatohepatitis) consortium funded by the Innovative Medicines Initiative 2 Joint Undertaking under grant agreement No. 777377. This Joint Undertaking receives support from the European Union's Horizon 2020 research and innovation programme and EFPIA.

Conflicts of interest

DGT reports consultancy on behalf of the National and Kapodistrian University of Athens for ICON, Inventiva, CymaBay, Clinnovate and is Chair, Advisory Board, European Society of Pathology. QMA reports grants from AstraZeneca, Boehringer Ingelheim, Intercept; consultancy on behalf of Newcastle University for Abbvie, Alimentiv, Akero, AstraZeneca, Axcella, 89Bio, Boehringer Ingelheim, Boston Pharmaceuticals, Corcept Therapeutics, Echosens, Eli Lilly, Gilead, GlaxoSmithKline, HistoIndex, Intercept, Inventiva, Madrigal, Medpace, Merck, Metadeq, Novo Nordisk, PathAI, Pfizer, Pharmanest, Prosciento, Regeneron, Resolution Therapeutics, Roche, Terns; honoraria from Avalere, Catalyst Medical Education, Integritas Communications, Medscape, Novo Nordisk, Springer Healthcare; royalties from Elsevier Ltd; Advisory board on behalf of Newcastle University for Medpace, Gilead. VR reports grants from MSD (to institution) and consultancy for Boehringer-Ingelheim, Akero, Novo-Nordisk, 89Bio, Sagimet, GSK. PB, JA, SD, ASHG, MG, CL, PS, LT, AB, KW, VP report no conflicts of interest.

Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the Supplementary data to this article: