Diagnostic performance of ultrasound-derived fat fraction for assessing steatotic liver disease with histopathology as the reference

Article information

Ultrasonography. 2026;45(1):48-58

Publication date (electronic) : 2025 October 21

doi : https://doi.org/10.14366/usg.25160

Yunlin Huang^1,*, Chao Sun^2,*, Xinliang Xu¹, Ying Wang¹, Li Wei¹, Jiaying Cao¹, Yijie Qiu¹, Juan Cheng¹, Shiwen Wang¹, Rui Cheng¹, Ming Wang¹, Jian-Gao Fan^,2, Yi Dong^,1

¹Department of Ultrasound, Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China

²Center for Fatty Liver Disease, Department of Gastroenterology, Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai Key Lab of Pediatric Gastroenterology and Nutrition, Shanghai, China

Correspondence to: Yi Dong, MD, PhD, Department of Ultrasound, Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, 1665 Kongjiang Road, Shanghai, 200092, China Tel, Fax: +86-21-25077258 E-mail: drdaisydong@hotmail.com

Jian-Gao Fan, MD, Department of Gastroenterology, Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai Key Lab of Pediatric Gastroenterology and Nutrition, 1665 Kong Jiang Road, Shanghai 200092, China Tel, Fax: +86-21-25077344 E-mail: fanjiangao@xinhuamed.com.cn

^*

These authors contributed equally to this work.

Received 2025 August 20; Revised 2025 October 21; Accepted 2025 October 21.

Abstract

Graphical abstract

Introduction

Steatotic liver disease (SLD) encompasses various etiologies of abnormal lipid accumulation in the liver (hepatic steatosis) and affects around 30% of the global population [1]. Metabolic dysfunction–associated steatotic liver disease (MASLD) is defined as SLD in patients with at least one component of metabolic syndrome [2]. Metabolic dysfunction–associated steatohepatitis, a subtype of MASLD, involves inflammation and hepatocellular injury, leading to liver fibrosis and ultimately cirrhosis [3]. Early identification and management of individuals with hepatic steatosis may help prevent adverse outcomes associated with cirrhosis [4].

Liver biopsy is considered the gold standard for assessing hepatic steatosis. Nevertheless, it has several limitations, including invasiveness, potential complications, sampling error, and limited reproducibility [5]. Magnetic resonance imaging proton density fat fraction (MRI-PDFF) is a noninvasive quantitative modality for evaluating hepatic steatosis [6]. However, this option is not widely accessible due to substantial costs and contraindications.

Conventional B-mode ultrasound (BMUS) is widely used in clinical practice owing to its availability and cost-effectiveness [7,8], but this approach is qualitative and subjective. Quantitative ultrasound techniques have been developed to evaluate hepatic steatosis, including the attenuation coefficient (AC), backscatter coefficient (BSC), and speed of sound [7]. The controlled attenuation parameter (CAP) is the most common method for measuring AC to evaluate hepatic steatosis. However, CAP is not recommended as a reference standard because of its limited capacity to stage hepatic steatosis, with areas under the receiver operating characteristic curve (AUCs) ranging from 0.69 to 0.95, and because optimal CAP thresholds for detecting hepatic steatosis remain uncertain [9,10].

Ultrasound-derived fat fraction (UDFF) is based on AC and BSC and reports hepatic fat fraction as a percentage [11]. In a prior prospective study, the present authors explored the feasibility of UDFF for detecting hepatic steatosis and found that UDFF values increased with higher hepatic steatosis grade (assessed by BMUS) [12]. Using MRI-PDFF as the reference standard, UDFF has shown good diagnostic performance for assessing hepatic steatosis [13–17]. However, limited research has explored the diagnostic performance of UDFF with liver biopsy and histopathology as the reference [18].

This study aims to evaluate the diagnostic performance of UDFF compared with CAP for detecting and grading hepatic steatosis in patients with SLD, using histopathology as the gold standard.

Materials and Methods

Results

Patient Characteristics

From February 2023 to May 2025, 72 patients were prospectively enrolled. Six patients were excluded due to refusal to undergo liver biopsy (n=5) or failure to obtain valid CAP values (n=1). Thus, 66 patients (37 female and 29 male) were included (Fig. 1). The median age was 46 years (IQR, 34 to 64), and the median body mass index (BMI) was 24 kg/m² (IQR, 22 to 27 kg/m²). Overall, 28.8% (19/66) had hypertension, 47.0% (31/66) had dyslipidemia, 22.7% (15/66) had type 2 diabetes mellitus, 45.5% (30/66) had overweight (BMI ≥25 kg/m² and <30 kg/m²), and 9.1% (6/66) had obesity (BMI ≥30 kg/m²). Demographic characteristics, biochemical test findings, and histopathological results are summarized in Table 1.

Fig. 1.

Flowchart of the study population.

UDFF, ultrasound-derived fat fraction; CAP, controlled attenuation parameter.

Table 1.

Patient characteristics

According to the histopathological findings, 45.5% (30/66) were diagnosed with steatohepatitis, 3.0% (2/66) with simple hepatic steatosis, 3.0% (2/66) with toxin-mediated liver injury, 13.6% (9/66) with viral hepatitis, 6.1% (4/66) with autoimmune hepatitis, and 19.7% (13/66) with primary biliary cholangitis, while 9.1% (6/66) were healthy. The distribution of hepatic steatosis grades was as follows: 50.0% (33/66) were S0, 27.3% (18/66) were S1, 12.1% (8/66) were S2, and 10.6% (7/66) were S3. The distribution of fibrosis within each steatosis grade is summarized in Supplementary Table 1. No biopsy-related adverse events occurred.

UDFF and CAP Values According to Histopathological Results

The median UDFF value was 5.8% (IQR, 3.4 to 11.3) across all patients. The median UDFF values for steatosis grades S0, S1, S2, and S3 were 3.5% (IQR, 3.0 to 7.0), 7.8% (IQR, 4.0 to 11.0), 15.3% (IQR, 5.0 to 17.8), and 20.0% (IQR, 19.0 to 33.0), respectively. UDFF values increased with higher grades of hepatic steatosis. The median UDFF value in patients with steatosis grade ≥S1 was higher than in those without steatosis (S0) (P<0.001). Specifically, significant differences in UDFF values were observed between grades S0 and S2 (P=0.008), S0 and S3 (P<0.001), and S1 and S3 (P=0.012). However, no significant differences were found between S0 and S1, S1 and S2, or S2 and S3 (all P>0.05) (Fig. 2A). Valid UDFF measurements were obtained in all patients. The median skin-to-capsule distance was 2.2 cm (IQR, 1.9 to 2.6).

Fig. 2.

Distributions of ultrasound-derived fat fraction (UDFF) (A) and controlled attenuation parameter (CAP) values (B) according to histopathological hepatic steatosis grade.

Boxes indicate the 25th–75th percentiles, whiskers indicate the minimum to maximum, and dots indicate outliers.

The median CAP value was 231 dB/m (IQR, 196 to 303) across all patients. The median CAP values for steatosis grades S0, S1, S2, and S3 were 198 dB/m (IQR, 173 to 225), 242 dB/m (IQR, 222 to 307), 290 dB/m (IQR, 238 to 332), and 329 dB/m (IQR, 315 to 368), respectively. CAP values increased with higher grades of hepatic steatosis. Significant differences in CAP values were found between S0 and S1 (P=0.023), S0 and S2 (P=0.010), and S0 and S3 (P<0.001). However, no significant differences were observed between S1 and S2, S1 and S3, or S2 and S3 (all P>0.05) (Fig. 2B). The CAP measurement success rate was 98.5% (65/66).

Representative UDFF and CAP images across different degrees of hepatic steatosis are shown in Fig. 3.

Fig. 3.

Representative images.

A 45-year-old man (body mass index [BMI], 21 kg/m²) with a controlled attenuation parameter (CAP) value of 216 dB/m (A) and an ultrasound-derived fat fraction (UDFF) value of 3% (B); histopathological findings indicated hepatic steatosis grade S0 (C). A 30-year-old woman (BMI, 26 kg/m²) with a CAP value of 236 dB/m (D) and a UDFF value of 10% (E); histopathology indicated steatosis grade S1 (F). A 29-year-old woman (BMI, 31 kg/m²) with a CAP value of 287 dB/m (G) and a UDFF value of 15% (H); histopathology indicated steatosis grade S2 (I). A 33-year-old man (BMI, 35 kg/m²) with a CAP value of 339 dB/m (J) and a UDFF value of 20% (K); histopathology indicated steatosis grade S3 (L) (C, F, I, L, x4).

Diagnostic Performance of UDFF for Assessing Hepatic Steatosis

Using histopathological results as the gold standard, the AUCs of UDFF and CAP measurements for detecting hepatic steatosis grade ≥S1 were 0.82 (95% CI, 0.72 to 0.92) and 0.84 (95% CI, 0.74 to 0.94), respectively (Fig. 4A). The AUCs of UDFF and CAP for diagnosing hepatic steatosis grade ≥S2 were 0.90 (95% CI, 0.80 to 1.01) and 0.86 (95% CI, 0.76 to 0.96), respectively (Fig. 4B). The AUCs of UDFF and CAP for detecting hepatic steatosis grade S3 were 0.99 (95% CI, 0.96 to 1.00) and 0.92 (95% CI, 0.85 to 0.99), respectively (Fig. 4C). The corresponding sensitivity and specificity of UDFF and CAP for detecting hepatic steatosis are shown in Table 2. Additionally, the respective UDFF cutoff values for detecting hepatic steatosis grades ≥S1, ≥S2, and S3 were 9.5%, 14.5%, and 16.3%.

Fig. 4.

Diagnostic performance of ultrasound-derived fat fraction (UDFF) and controlled attenuation parameter (CAP) for diagnosing hepatic steatosis grades ≥S1 (A), ≥S2 (B), and S3 (C).

AUC, area under the receiver operating characteristic curve; CI, confidence interval.

Table 2.

Diagnostic performance of UDFF and CAP for detecting and grading hepatic steatosis

No significant differences were found in diagnostic performance between UDFF and CAP for detecting hepatic steatosis grade ≥S1 (z=−0.33; 95% CI, −0.11 to 0.08; P=0.741) or ≥S2 (z=1.08; 95% CI, −0.04 to 0.13; P=0.280). However, UDFF displayed higher diagnostic performance than CAP for detecting hepatic steatosis grade S3 (z=2.09; 95% CI, 0.004 to 0.13; P=0.037).

In addition, using liver biopsy as the gold standard, the diagnostic performance of BMUS visual scoring for detecting hepatic steatosis (≥S1) was 0.70 (95% CI, 0.59 to 0.81). UDFF demonstrated significantly higher diagnostic performance than BMUS (z=2.12; 95% CI, 0.01 to 0.24; P=0.034).

Factors Associated with UDFF Measurements

Multivariate linear regression revealed that UDFF values were associated with the degree of hepatic steatosis (P<0.001), activity grade (P=0.021), CAP (P<0.001), aspartate aminotransferase level (P=0.045), and albumin level (P=0.004) (Table 3). Liver fibrosis was not significantly associated with UDFF.

Table 3.

Multivariable linear regression analysis of the associations between UDFF and patient characteristics

Spearman correlation analysis showed that UDFF values were positively and significantly correlated with CAP (r=0.73; 95% CI, 0.59 to 0.83; P<0.001), hepatic steatosis grade (r=0.66; 95% CI, 0.49 to 0.78; P<0.001), and activity grade (r=0.25; 95% CI, 0.001 to 0.47; P=0.043). UDFF was negatively correlated with high-density lipoprotein cholesterol (r=−0.28; 95% CI, −0.50 to −0.04; P=0.022). UDFF showed no significant correlation with liver fibrosis.

Discussion

In this prospective, biopsy-confirmed cohort, UDFF values were positively correlated with the extent of hepatic steatosis. The diagnostic performance of UDFF was effective for detecting and grading hepatic steatosis, and it exceeded that of CAP for grade S3 steatosis. Valid UDFF measurements were obtained in all patients. These findings indicate that UDFF can be feasibly implemented in clinical practice.

Considering the increasing prevalence of SLD and its adverse consequences, a noninvasive, quantitative tool is valuable for tracking disease progression. MRI-PDFF meets these criteria and is recommended as a reference standard for the diagnosis of SLD [23]. However, MRI-PDFF has several limitations in clinical practice, including high cost and limited accessibility. Prior studies have evaluated the diagnostic performance of UDFF for detecting hepatic steatosis in cohorts with SLD, using MRI-PDFF as the reference standard. These investigations reported positive correlations between UDFF and MRI-PDFF, with correlation coefficients ranging from 0.79 to 0.87. Furthermore, the findings indicated that UDFF exhibits good diagnostic accuracy in identifying hepatic steatosis (AUCs, 0.75–0.97) [13–16]. A recent multicenter study confirmed that UDFF is effective in diagnosing SLD [17]. However, those studies used MRI-PDFF rather than histopathology as the reference standard. A meta-analysis [24] reported that the histopathological hepatic steatosis fraction is often significantly higher than the MRI-PDFF measurement, and their correlation varies across SLD severity. Therefore, evaluating the correlation between UDFF and the histological proportion of hepatic steatosis is essential to minimize diagnostic inaccuracies.

To date, only limited data based on small cohorts have examined the diagnostic performance of UDFF for detecting hepatic steatosis using liver biopsy and histological results as the reference standard. A recent retrospective study included 73 patients with histologically confirmed liver disease who underwent UDFF [18]. The findings revealed that UDFF values were positively correlated with fat content (r=0.736), and UDFF showed good diagnostic accuracy for steatosis grades ≥S1, ≥S2, and S3 (AUCs, 0.926–0.971) [18]. Labyed and Milkowski [16] conducted a prospective study that enrolled 101 patients with diagnosed or suspected nonalcoholic fatty liver disease, 90 of whom underwent liver biopsy. The findings revealed that UDFF differed significantly across hepatic steatosis grades, with a Spearman correlation coefficient of 0.71 between UDFF and steatosis grade. The AUCs of UDFF for detecting hepatic steatosis grades ≥S1, ≥S2, and S3 were 0.94, 0.88, and 0.83, respectively, with corresponding cutoff values of 8.1%, 15.9%, and 16.1% [16]. The present study yielded comparable results: the AUCs of UDFF for detecting grades ≥S1, ≥S2, and S3 were 0.82, 0.90, and 0.99, respectively, with corresponding cutoff values of 9.5%, 14.5%, and 16.3%. Although the etiologies in the prior cohort differ from those in the present study population, these findings collectively indicate that UDFF has good diagnostic performance for diagnosing SLD.

CAP measurement is a commonly used tool for evaluating hepatic steatosis [25]. To compare the diagnostic accuracy of CAP with another quantitative ultrasound technique, Jung et al. [26] enrolled 123 patients with non-alcoholic fatty liver disease and utilized quantitative ultrasound fat fraction (QUS FF) to assess hepatic steatosis, with MRI-PDFF as the reference standard. Using cutoff values of QUS FF ≥5% and CAP ≥288 dB/m, QUS FF achieved higher diagnostic performance (AUC, 0.92) than CAP (AUC, 0.79) for detecting mild hepatic steatosis (defined by MRI-PDFF ≥5%; n=100). When using cutoff values of QUS FF ≥10% and CAP ≥306 dB/m, QUS FF also outperformed CAP (AUC, 0.90 vs. 0.75) for detecting moderate hepatic steatosis (defined by MRI-PDFF ≥10%; n=70). However, the study did not compare the diagnostic performance of QUS FF and CAP for severe hepatic steatosis. In a recent multicenter study [17], a subgroup analysis of 66 patients with suspected SLD was used to compare the diagnostic performance of UDFF and CAP. Using MRI-PDFF thresholds of ≥5%, ≥15%, and ≥25% for mild, moderate, and severe hepatic steatosis, no significant difference was found between UDFF and CAP [17]. Utilizing the same reference standard (MRI-PDFF) as Jung et al. [26], this multicenter study supplemented the existing literature by comparing the diagnostic accuracy of UDFF and CAP for severe hepatic steatosis. Nevertheless, varying MRI-PDFF cutoffs and differing etiologies limited direct comparison across studies. The present study enrolled 66 patients with abnormal liver function or clinically suspected SLD and compared the diagnostic performance of UDFF and CAP for detecting hepatic steatosis, using histopathology as the reference standard. UDFF and CAP cutoffs were determined by the Youden index. No significant differences were found between UDFF and CAP for detecting mild or moderate hepatic steatosis; however, UDFF showed superior diagnostic performance for severe hepatic steatosis. The discrepancy between the present findings and those of Jung et al. [26] may reflect differences in reference standards and disease etiologies.

In this study, the CAP measurement success rate was 98.5%, with no UDFF failures. These results likely reflect the use of real-time BMUS imaging to guide ROI placement during UDFF acquisition, enabling the operator to easily obtain high-quality images and valid UDFF results. UDFF values did not differ significantly between steatosis grades S0 and S1, whereas CAP values did show a significant difference. This finding contrasts with a previous investigation [27], which reported that UDFF values in grade S1 were significantly higher than in S0 (graded by MRI-PDFF) in a cohort of 115 patients with MASLD. Potential reasons for this discrepancy include limited sample size, differences in study cohorts, and the choice of diagnostic reference. Notably, in the present study, UDFF demonstrated a significantly higher AUC than BMUS for diagnosing hepatic steatosis (≥S1). Accordingly, UDFF may serve as a quantitative imaging option for diagnosing SLD, particularly when CAP measurements are unsuccessful.

The degree of hepatic steatosis has been reported to be associated with lobular inflammation, fibrosis, and steatohepatitis [28,29]. Prior studies have shown that liver stiffness assessed by ultrasonic shear wave elastography does not differ significantly across steatosis levels [30,31]. In a prospective study of 121 patients with MASLD, the diagnostic performance of UDFF and ultrasound shear wave elastography was evaluated in histologically confirmed hepatic steatosis and liver fibrosis. The findings indicated that UDFF values were inversely correlated with shear wave elastography measurements [32]. However, the correlation between UDFF and histologically confirmed fibrosis was not reported. In the present study, UDFF values did not correlate with histopathological fibrosis stage. Although ultrasonic beam attenuation theoretically depends on the degree of fibrosis [23], UDFF combines AC with BSC, which may help compensate for fibrosis-related attenuation. Inflammatory changes may also alter acoustic tissue properties (e.g., attenuation and backscatter) [33]. The present study indicated that UDFF was significantly and positively correlated with histopathological steatosis and activity grades.

Given the technical constraints of ultrasonic beam penetration in patients with elevated BMI, UDFF results should be interpreted with caution. In this study, BMI displayed no significant correlation with UDFF, corroborating prior findings [13,34].

The limitations of this study should be acknowledged. First, the sample size was relatively small, and the distribution of steatosis grades was uneven, with most patients classified as having S0 disease (33/66) and only a few cases as S3 (n=7). This imbalance may inflate the very high AUC and widen the CI observed for S3, limiting the generalizability of the findings to populations with a broader range of disease severity. Accordingly, these results, especially for patients with steatosis grade S3, should be considered preliminary and require confirmation in multicenter studies with larger, more balanced samples. Second, although UDFF measurements have been reported to achieve excellent interobserver agreement within an SLD cohort [17], all measurements in this study were performed by a single experienced radiologist, which may limit the applicability of the findings across broader populations and operators.

UDFF demonstrated good diagnostic performance for assessing hepatic steatosis, using histopathology as the reference standard. UDFF may serve as a useful quantitative imaging method to measure liver fat content in patients with SLD.

Supplementary Material

Article information Continued

Notes