Per-patient Negative Predictive Value of the CT and MRI Liver Imaging Reporting and Data System Version 2018 Treatment Response Algorithm for Hepatocellular Carcinoma
RADIOLOGY(2023)
摘要
Background: The Liver Imaging Reporting and Data System version 2018 (LI -RADS) treatment response algorithm (TRA) is a high-specificity, lower-sensitivity grading system to diagnose hepatocellular carcinoma (HCC) and recurrence after local-regional therapy. However, the emphasis on specificity can result in disease understaging, potentially leading to poorer posttransplant outcomes. Purpose: To determine the negative predictive value (NPV) of pretransplant CT and MRI assessment for viable HCC on a per -patient basis using the LI -RADS TRA, considering explant pathology as the reference standard. Materials and Methods: Patient records from 218 consecutive adult patients from a single institution with HCC who underwent liver transplant from January 2011 to November 2017 were retrospectively reviewed. Two readers blinded to the original report reviewed immediate (within 90 days) pretransplant imaging and characterized observations according to the LI -RADS TRA. Based on this, patients with LR-4, LR-5, or LR-TR (treatment response) viable tumors were designated as viable tumor; patients with solely LR-3 or LR-TR equivocal tumors were designated as equivocal; and patients with only LR-TR nonviable lesions were designated as no viable disease. Patients were designated as within or outside the Milan criteria. These per -patient designations were compared with the presence of viable disease at explant pathology. Fisher exact test was used to compare the differences between CT and MRI. Weighted kappa values were used to calculate interreader reliability. Results: Final study sample consisted of 206 patients (median age, 61 years [IQR, 57-65 years]; 157 male patients and 49 female patients). Per -patient LI -RADS TRA assessment of pretransplant imaging had an NPV of 32% (95% CI: 27, 38) and 26% (95% CI: 20, 33) (readers 1 and 2, respectively) for predicting viable disease. Seventy-five percent (reader 1) and 77% (reader 2) of patients deemed equivocal had residual tumors at explant pathology. Weighted interreader reliability was substantial (kappa = 0.62). Conclusion: Patient -based stratification of viable, equivocal, and nonviable disease at pretransplant CT or MRI, based on LI -RADS TRA, demonstrated low negative predictive value in excluding HCC at explant pathology. (c) RSNA, 2023
更多查看译文
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要