Multiparametric MRI with MR elastography findings in patients with sinusoidal obstruction syndrome after oxaliplatin-based chemotherapy – Insights into Imaging

Study population

This retrospective study was approved by the Ethics Committee of our institute. Patients who underwent liver MRI with Gd-EOB-DTPA and MRE between January 2018 and June 2020 and had received at least three cycles of oxaliplatin for colorectal, gastric, or pancreatic cancer within the six months prior to MRI were included in this study. Patients with SOS and controls without SOS findings were evaluated retrospectively. The presence of SOS was diagnosed by liver MRI (including the 20-min hepatobiliary phase) in 18 patients, while 16 patients having no SOS finding formed the control group.

Of the patients in the SOS group, 14 were treated for colon/ rectal cancer, 3 for stomach cancer, and 1 for pancreatic cancer. Of the 16 patients taking oxaliplatin who had no SOS findings, 14 were treated for colon/rectum cancer and 2 for stomach cancer. A total of thirty-four patients (M/F:22/12) were enrolled in the study.

Laboratory findings

Alanine transaminase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma-glutamyltransferase (GGT), total bilirubin level, and platelet count were measured in each patient within one week of MRI. The reference ranges used by our institution for blood sample parameters were as follows: 0–33 IU/L for ALT, 0–32 IU/L for AST, 35–129 IU/L for ALP, 5–61 IU/L for GGT, 0.2–1.2 mg/dL for total bilirubin, and 150–400 × 103/μL for platelet count.

Imaging protocol

All patients included in the study underwent liver MRI with hepatobiliary phase using Gd-EOB-DTPA (Primovist; Bayer-Schering Pharma AG, Berlin, Germany) with a 1.5-T system (Siemens AERA, Germany and GE Signa, USA) with standard body and spine matrix coils. In addition, MRE examinations were performed with 1.5 T MRI (Siemens AERA) in the same session or within two weeks after the liver MRI examination. Also, MRE, T1 and T2 mapping, T2*, and high-speed T2-corrected multi-echo (HISTO) sequences were acquired in the same session.

With the active driver generating waves at 60 Hz, the MRE was performed with a 2D-GRE sequence modified with the following parameters: repetition time (TR)/echo time (TE), 50/21 ms; flip angle: 25 degrees, bandwidth: 31.25 kHz, matrix: 256 × 128, acquisition time: 2.5 min. Depending on the liver size, 2 or 3 slices of 10 mm thickness were obtained from the largest part of the liver by holding the patient’s breath.

T1 mapping was performed using the B1 inhomogeneity-corrected method with variable flip angle. Sequence parameters were as follows: repetition time (TR)/echo time (TE), 4.4/2.1 ms, rotation angle: 3 and 15 degrees, matrix: 256 × 156, FOV: 380 × 300 mm, slice thickness: 4 mm, acquisition time: 1.5 min.

T2 mapping was performed by calculating the T2 value from different TE’s using the SSFP-based TrueFISP sequence and the exponential signal decay model. The sequence parameters were as follows: TR: 166 ms, TE (0 ms, 25 ms, 55 ms), flip angle: 70 degrees, FOV: 420 × 260 mm, slice thickness 10 mm, matrix: 192 × 192, NEX: 1, acquisition time: 1.2 min.

T2* mapping was performed to assess liver iron load with the following parameters: TR: 200 ms, TE: 0.93/2.14/3.35/4.56/5.77/6.98/8.19/9.4/10.61/11.82/13.03/14.24 ms, rotation angle: 20 degrees, section thickness: 10 mm, FOV: 400 × 300 mm, matrix: 160 × 85.

The HISTO sequence to assess liver fat content was performed with the following parameters: TE 12/24/36/48/72 ms, TR 3000 ms, voxel size: 30 × 30 × 30 mm, acquisition time: 15 s.

Imaging analysis

MRI images were reviewed by two radiologists (4 and 5 years of experience) independently. First observer re-evaluated all images 6 months later for intra-reader evaluation. Discrepancies between readers were resolved by a senior radiologist with 16 years of experience. The final consensus reading was used for statistical analysis. All data were transferred to a workstation (Syngo.via, Siemens, Erlangen, Germany) for analysis. Right lobe, left lobe, and total liver stiffness scores were measured in all patients. ROIs were manually drawn during measurement, excluding lesions, large vessels, liver margins, and artifacts on the magnitude images generated with the MRE sequence. These ROIs were then copied to the stiffness maps, which provided liver stiffness values in kilopascals (kPa). The ROIs were drawn to encompass the parenchyma as much as possible.

T1 relaxation, T2 relaxation, and T2* values were measured over the corresponding sequences. Measurements were performed with a sufficiently large ROI (region of interest), excluding lesions, large vessels, liver margins, and artifacts from the right lobe of the liver (Figs. 4, 5). Corrected T1 (cT1) relaxation values were calculated using the formula “T1-420 + 20 × T2*” over T2* values [23]. Liver fat percentage was also noted.

In this study, patients who received oxaliplatin and developed SOS findings and those who received oxaliplatin but did not develop SOS findings (control group) were compared in terms of liver stiffness and T1–T2 relaxation time. Then, the patients we separated according to the severity of SOS within the SOS group were also compared with each other.

Within the SOS group, patients were classified according to the extent of reticular hypointensity in the hepatobiliary phase. The patients with reticular hypointensity less than 20% of the liver parenchyma were classified as mild SOS, between 20 and 50% as moderate, and more than 50% as advanced SOS (Fig. 1).

Classification of SOS severity according to the extent of reticular hypointensity on the hepatobiliary phase images; ‘mild’ (A) if reticular hypointensity are less than 20% of the liver parenchyma, ‘moderate’ (B) if 20–50%, and more than 50% ‘advance’ (C) grade SOS

Statistical analysis

Categorical measurements were summarized as numbers and percentages and continuous measurements as mean and (minimum–maximum). When comparing continuous measurements between groups, distributions were examined, Student’s t-test was used when variables met the parametric pretest assumption, and Mann–Whitney U test was used when they did not. One-way ANOVA test was used to compare continuous data by SOS severity. The accuracy of the methods was compared by calculating the areas under the curve (AUROC) from the ROC curves.

All inter- and intra-reader reliability was determined using the intra-class correlation coefficient (ICC) for continuous parameters and the kappa statistic (κ) and percentage agreement for categorical or binary parameters. ICC values range from 0 to 1, and values above 0.75 were considered to have excellent reliability. Kappa values range from − 1 to 1 and were categorized as poor (κ < 0), slight (κ = 0 to 0.20), fair (κ = 0.21 to 0.40), moderate (κ = 0.41 to 0.60), substantial (κ = 0.61 to 0.80), and almost perfect (κ = 0.81 to 1). The statistical significance level (p-value) in the tests was taken as 0.05. For statistical analysis of the data, IBM the SPSS 20.0 program was used.