In this prospective large multicenter cohort study based on over 500 examinations, ATI and CAP were highly accurate in diagnosing hepatic steatosis with MRI-PDFF as the reference standard. Both ATI and CAP increased significantly with the progression of steatosis grade independent of the degree of liver fibrosis. However, the CC for ATI and MRI-PDFF was significantly higher than the CC for CAP and MRI-PDFF. Moreover, ATI was not affected by BMI. In addition, AUROCs for ATI in detecting ≥ S1, S2, or S3 hepatic steatosis were significantly higher than those for CAP. ATI had significantly higher specificity, accuracy, PPV, and NPV for detecting ≥ S2 or S3 hepatic steatosis than CAP. As a result, the diagnostic performance of ATI is superior to that of CAP. Essentially the same results were obtained for the patients with MASLD.

MRI-PDFF is the non-invasive method for diagnosing hepatic steatosis grade with the highest diagnostic performance [20]. However, MRI-PDFF has several disadvantages including cost and lower patient acceptance. On the other hand, ATI and CAP are non-invasive methods for diagnosing hepatic steatosis that are cost-effective and have good patient acceptance. Moreover, the correlation between ATI and MRI-PDFF was strong. Therefore, ATI can be considered a non-invasive alternative to MRI-PDFF. Based on the perspective of outliners, patients with low SCD tended to have lower ATI values. CAP values of patients with high BMI, SCD, and SWE values tended to be higher and CAP values of patients with low BMI and SCD tended to be lower. This should be noted.

CAP has been widely used as an attenuation imaging method for diagnosing non-invasive hepatic steatosis, facilitating swift bedside assessments of tissue stiffness. Its usefulness has been widely reported. However, a study by Victor de Ledinghen et al. revealed a 7.7% failure rate among 5325 CAP examinations. Measurement failure was associated with female gender, higher BMI, and the presence of metabolic syndrome. Elevated CAP values were observed in patients with BMI ≥ 25 kg/m2, metabolic syndrome, alcohol consumption > 14 drinks per week, or liver stiffness > 6 kPa [29]. In this study, the CC for CAP and MRI-PDFF was significantly lower in patients with BMI ≥ 30 kg/m2 than in patients with BMI < 30 kg/m2. Moreover, despite using the XL probe for patients with SCD ≥ 25 mm, the CC for CAP and MRI-PDFF was significantly lower than that of ATI and the CC in patients with SCD ≥ 25 mm significantly lower than that in patients with SCD < 25 mm. This result suggested that CAP is affected by BMI or SCD despite using the XL probe in patients with SCD ≥ 25 mm. The American Association for The Study of Liver Disease (AASLD) Practice Guidance on the Clinical Assessment and Management of Nonalcoholic Fatty Liver Disease states that CAP does not accurately quantify or monitor changes in liver fat [30].

Regarding the influence of hepatic fibrosis on ATI and CAP, Yuri et al. reported that ATI is not affected by liver fibrosis [31]. On the other hand, the CCs between ATI and MRI-PDFF in patients with high 2D-SWE, high FIB-4 index, or high type IV collagen 7 s, which reflect advanced hepatic fibrosis, were significantly lower than those in patients with low SWE, low FIB-4 index, or low type IV collagen 7 s. However, the CCs for CAP and MRI-PDFF were not significantly different. Therefore, the influence of ATI on the diagnosis of hepatic steatosis in patients with advanced liver fibrosis should be kept in mind.

The AUROC for diagnosing ≥ S1 hepatic steatosis using ATI has been reported to be more than 0.85 [12, 32,33,34,35]. In this study, the AUROC for diagnosing ≥ S1 hepatic steatosis using ATI was 0.895, which was higher than using CAP. Ferraioli et al. have investigated the diagnostic ability of ATI and CAP [17]. In their study, ATI was more accurate than CAP for detecting and quantifying liver steatosis with a statistically significant difference for only ≥ S2. However, in our study, the AUROC for ATI in diagnosing all hepatic steatosis grades (≥ S1, ≥ S2, and S3) was significantly higher than for CAP. The specificity, accuracy, and PPV for ATI in detecting ≥ S2 or S3 hepatic steatosis were significantly higher than those for CAP. This result means that for detecting ≥ S2 or S3 hepatic steatosis grade, there are fewer false positives. Accordingly, ATI is considered more useful than CAP in diagnosing hepatic steatosis. Particularly, ATI had high accuracy for detecting ≥ S2 hepatic steatosis. The PPVs of ATI and CAP decreased 10–30% with the progression of steatosis grade. Of note, the PPVs of ATI and CAP in patients with S3 steatosis were lower than those in patients with S1 or S2 steatosis. Lower prevalence of subjects is known to be associated with lower PPV. The prevalence of patients with S3 steatosis was significantly lower than that in S1 and S2 steatosis. Moreover, the prevalence of patients with S2 steatosis was significantly lower than in S1 steatosis, which might be one reason why PPV was low in S3 steatosis.

In this prospective study, the CAP cutoff values for diagnosing ≥ S1, S2, or S3 hepatic steatosis were 258 dB/m, 272 dB/m, and 276 dB/m, respectively. The cutoff values for CAP were only slightly higher or slightly lower than the values in a meta-analysis [36]. The ATI cutoff values for diagnosing ≥ S1, S2, or S3 hepatic steatosis grade were 0.64 dB/cm/MHz, 0.72 dB/cm/MHz, and 0.75 dB/cm/MHz, respectively. The cutoff value for ≥ S1 in this study was almost the same as that in a pilot study; the cutoff value for ≥ S2 was the same value [32]. Therefore, the ATI cutoff values for detecting each steatosis grade in this prospective and multicenter study seem to be more applicable for use in daily practice.

In some of MASLD patients, ATI and CAP values have been less than the cutoff values for S1. These patients have been initially diagnosed MASLD by MRI-PDFF, liver biopsy and ultrasound including attenuation imaging before this research start. Therefore, some of the MASLD patients with the improvement of hepatic steatosis had the value less than cutoff values of S1. And SWE value in MASLD patients diagnosed as S0 based on ATI and CAP cutoff value was not higher than that of S1-3 (Supplementary Fig. 9). Therefore, ATI and CAP in the MASLD patients with S0 was not related with advance hepatic fibrosis.

Our study had several limitations. First, patients with various etiologies of liver disease were included. However, SLD consists of patients with not only MASLD but also other etiologies of hepatic steatosis. Therefore, the results of this study are useful in the clinical setting. Second, ATI, CAP and MRI-PDFF measurements were performed within 1 month. Therefore, steatosis grade could have decreased because of diet and exercise therapy. However, we think that there were few patients whose steatosis grade decreased over only 1 month in the clinical setting. Third, almost all subjects were Japanese. Further study of patients of various races is warranted.

In conclusion, ATI was strongly correlated with MRI-PDFF and had good diagnostic ability for each hepatic steatosis grade in this multicenter prospective study. In addition, ATI is a better non-invasive method for diagnosing hepatic steatosis than CAP. In particular, ATI is less affected by body shape. Further prospective studies should be performed throughout the world.