Can reference images improve interobserver agreement in reporting liver fibrosis?

Can Automated 3-Dimensional Dixon-Based Methods Be Used in Patients With Liver Iron Overload?

PURPOSE

Accurate quantification of liver iron concentration (LIC) can be achieved via magnetic resonance imaging (MRI). Maps of liver T2*/R2* are provided by commercially available, vendor-provided, 3-dimensional (3D) multiecho Dixon sequences and allow automated, inline postprocessing, which removes the need for manual curve fitting associated with conventional 2-dimensional (2D) gradient echo (GRE)-based postprocessing. The main goal of our study was to investigate the relationship among LIC estimates generated by 3D multiecho Dixon sequence to values generated by 2D GRE-based R2* relaxometry as the reference standard.

METHODS

A retrospective review of patients who had undergone MRI scans for estimation of LIC with conventional T2* relaxometry and 3D multiecho Dixon sequences was performed. A 1.5 T scanner was used to acquire the magnetic resonance studies. Acquisition of standard multislice multiecho T2*-based sequences was performed, and R2* values with corresponding LIC were estimated. The comparison between R2* and corresponding LIC estimates obtained by the 2 methods was analyzed via the correlation coefficients and Bland-Altman difference plots.

RESULTS

This study included 104 patients (51 male and 53 female patients) with 158 MRI scans. The mean age of the patients at the time of scan was 15.2 (SD, 8.8) years. There was a very strong correlation between the 2 LIC estimation methods for LIC values up to 3.2 mg/g (LIC quantitative multiecho Dixon [qDixon; from region of interest R2*] vs LIC GRE [in-house]: r = 0.83, P < 0.01; LIC qDixon [from segmentation volume R2*] vs LIC GRE [in-house]: r = 0.92, P < 0.01); and very weak correlation between the 2 methods at liver iron levels >7 mg/g.

CONCLUSION

Three-dimensional-based multiecho Dixon technique can accurately measure LIC up to 7 mg/g and has the potential to replace 2D GRE-based relaxometry methods.

Prognostic Role of Liver Biopsy in Patients With Severe Indeterminate Acute Hepatitis

BACKGROUND AND AIMS

Severe indeterminate acute hepatitis (sIAH) is a poorly understood rare disease with no specific therapy. This study aims to define the clinicopathological characteristics of sIAH and the role of liver biopsy in determining prognosis.

METHODS

Patients with sIAH admitted to a single center between 2010 and 2019 were included. Histopathological patterns of liver biopsies were reviewed by 2 histopathologists, and key findings further were specified by multiplex immunofluorescence. Patients that died or underwent liver transplantation were analyzed as nonsurvivors.

RESULTS

Of 294 patients with acute hepatitis, 43 with sIAH were included. Seventeen (39.5%) underwent liver transplantation and 7 (16.2%) died within 3 months. Multilobular necrosis was the predominant histopathological feature, being significantly more frequent in nonsurvivors (62.5% vs 21.1%; P = .016). Necrotic areas showed low HNF4α and Ki67 expression but high expression of CK19 and cell death markers identifying areas of severe tissue injury and inadequate regenerative response. Patients with multilobular necrosis had higher international normalized ratio, Model for End-Stage Liver Disease, and Model for End-Stage Liver Disease-Sodium scores compared with those without (P values for all markers <.05). Multivariate Cox analysis revealed that multilobular necrosis (hazard ratio, 3.675; 95% confidence interval, 1.322-10.211) and lower body mass index (hazard ratio, 0.916; 95% confidence interval, 0.848-0.991) independently predicted death or transplantation.

CONCLUSIONS

The results of this study provide novel insights into the important role of liver biopsy in sIAH patients, suggesting that the presence of multilobular necrosis is an early indicator of poor prognosis.

R2 relaxometry based MR imaging for estimation of liver iron content: A comparison between two methods

PURPOSE

To compare the reproducibility and accuracy of R2-relaxometry MRI for estimation of liver iron concentration (LIC) between in-house analysis and FDA-approved commercially available third party results.

METHODS

All MR studies were performed on a 1.5T scanner. Multi-echo spin-echo scans with a fixed TR and increasing TE values of 6 ms, 9 ms, 12 ms, 15 ms, and 18 ms (spaced at 3 ms intervals) were used. Post-processing of the images to calculate mean relaxivity, R2, included drawing of regions of interest to include the whole liver on mid-slice. The relationship between liver R2 values and estimated LIC calculated with in-house analysis and values reported by an external company (FerriScan^®, Resonance Health, Australia) were assessed with correlation coefficients and Bland-Altman difference plots. Continuous variables are presented as mean ± standard deviation. Significance was set at p value < 0.05.

RESULTS

474 studies from 175 patients were included in the study (mean age 10.4 ± 4.2 years (range 1-18 years); 254 studies from girls, 220 studies from boys). LIC ranged from 0.6 to 43 mg/g dry tissue, covering a broad range from normal levels to extremely high iron levels. Linearity between proprietary and in-house methods was excellent across the observed range for R2 (31.5 to 334.8 s^-1); showing a correlation coefficient of r = 0.87, p < 0.001. Bland-Altman R2 difference plot between the two methods shows a mean bias of + 21.5 s^-1 (range - 47.0 to + 90.0 s^-1 between two standard deviations). LIC reported by FerriScan^® compared with LIC estimated in-house with R2 as reported by FerriScan^® agreed strongly, (r = 1.0, p < 0.001).

CONCLUSION

R2 relaxometry MR imaging for liver iron concentration estimation is reproducible between proprietary FDA-approved commercial software and in-house analysis methods.

Magnetic resonance elastography SE-EPI vs GRE sequences at 3T in a pediatric population with liver disease

PURPOSE

The goal of our study is to compare hepatic stiffness measures using gradient-recalled echo (GRE) versus spin-echo echo planar imaging (SE-EPI)-based MR Elastography (MRE) at 3T used to measure hepatic stiffness in a patients with suspected liver diseases.

MATERIALS AND METHODS

This retrospective study included 52 patients with liver disease who underwent a 3T MRE exam including both an investigational SE-EPI-based technique and a product GRE-based technique. Regions of interest (ROI) were placed on the elastograms to measure elastography-derived liver stiffness as well as the area included within the ROIs. The mean liver stiffness values and area of ROIs were compared.

RESULTS

The mean liver stiffness was 3.72 kilopascal (kPa) ± 1.29 using GRE MRE and 3.78 kPa ± 1.13 using SE-EPI MRE. Measurement of liver stiffness showed excellent agreement between the two pulse sequences with a mean bias of - 0.1 kPa (range - 1.8 to 1.7 kPa) between sequences. The mean measurable ROI area was higher with SE-EPI (313.8 cm² ± 213.8) than with the GRE technique (208.6 cm² ± 114.8), and the difference was statistically significant (P < 0.05).

CONCLUSIONS

Our data shows excellent agreement of measured liver stiffness between GRE and SE-EPI-based sequences at 3T. Our results show the advantage of a SE-EPI MRE sequence in terms of image quality, ROI size and acquisition time with equivalent liver stiffness measurements as compared to GRE-MRE sequence.

Feasibility and agreement of stiffness measurements using gradient-echo and spin-echo MR elastography sequences in unselected patients undergoing liver MRI

OBJECTIVE

  To evaluate the agreement of three MR elastography (MRE) sequences in patients undergoing liver MRI for clinical care.  Methods:  A cross-sectional retrospective study was performed with 223 patients referred for liver MRI, including 12 patients with liver iron overload. Data obtained with spin-echo (SE) and gradient-echo (GRE) MRE sequences were compared. Multiple linear regression adjusted for the presence of liver fat was also performed to assess the correlation between fat infiltration and stiffness measurements results. Agreement between two SE sequences was assessed in patients with liver iron overload.  Results: We found strong correlation between the GRE sequence and two SE sequences. Spearman's correlation coefficients between the GRE, SE, and SE-EPI MRE sequences in patients with liver R2* <75Hz were 0.74, 0.81, and 0.80. GRE-MRE failed in patients with liver R2* > 75 Hz. In this subgroup, the correlation coefficient between both SE-MRE sequences was 0.97. Liver fat did not interfere with the results.

CONCLUSION

In clinical setting, there is a high correlation between the GRE and SE MRE stiffness measurements, independently of the degree of liver fat infiltration measured by PDFF. A strong correlation between SE-MRE sequences is found even in patients with iron overload.  Advances in knowledge: Our study addresses liver iron and fat content simultaneously to describing the technical feasibility and correlation between different MRE sequences in consecutive unselected patients refereed for liver MRI. EPI SE-MRE should be considered an optimal alternative to assess liver fibrosis in patients in whom GRE-MRE failures, such as iron-overloaded, in pediatric, elderly, or severely ill populations.