1293. Correlation Between LI-RADS Treatment Response and mRECIST Classification Systems Among SBRT-Treated HCC Lesions
Authors * Denotes Presenting Author
  1. Diwash Thapa *; University of North Carolina School of Medicine
  2. Andrew Moon; University of North Carolina School of Medicine - Gastroenterology and Hepatology
  3. Thad Banefield; University of North Carolina School of Medicine - Radiology
  4. Hannah Kim; Vanderbilt University School of Medicine - Gastroenterology, Hepatology, Nutrition
  5. Kyle Wang; University of Cincinnati - Radiation Oncology
  6. Katrina McGinty; University of North Carolina School of Medicine - Radiology
  7. Lauren Burke; University of North Carolina School of Medicine - Radiology
Persistence of arterial phase hyperenhancement (APHE) immediately following stereotactic body radiation therapy (SBRT) for hepatocellular carcinoma (HCC) is known to occur in at least 50% of patients with pathologically nonviable tumors. While most SBRT-treated nodules in the acute phase would be classified as “stable” by modified Response Evaluation Criteria in Solid Tumor (mRECIST), Liver Reporting and Data System Treatment Response (LR-TR) permits an “equivocal” classification for expected post-treatment changes. Objective measures to predict the viability of LR-TR equivocal lesions have not been identified. Here we investigate changes in lesion size as such a measure and describe the correlation between LR-TR and mRECIST criteria among SBRT-treated lesions.

Materials and Methods:
A retrospective analysis of SBRT-treated lesions from 2012 to 2018 at UNC-CH was performed. Two board-certified body radiologists with >10 years of experience read MRI (1 read per scan) pre-procedure and post-procedure at 3, 6, 9, and 12 months. Readers assigned each nodule LR-TR and mRECIST categories. Nodules without any post-procedure MRI were excluded. Analyses were performed at the nodule level. Correlation between mRECIST and LR-TR (p<0.05 significant) and its change in time (p<0.05 significant) was studied using mixed effects logistic regression with predicted probability as a surrogate measure. Receiver operator curve (ROC) analyses were performed to determine if the difference in mRECIST size from pretreatment is predictive of disease recurrence for LR-TR equivocal lesions. The “gold standard” for recurrence was defined as need for retreatment of the same or other nodules. The area under the curve (AUC) was generated for analyses at 3 months; 3 and 6 months; and all time points.

53 lesions were included in this analysis (median mRECIST size at baseline 2.62 cm). While there was evidence of correlation between mRECIST and LR-TR (p=0.011), the correlation did not change over time (p=0.88). The predicted probability of mRECIST was higher for LR-TR non-viable, compared to LR-TR equivocal (p=0.0098) and viable (p=0.036). For LR-TR equivocal lesions, there was insufficient evidence that the pre- to post-treatment difference in mRECIST sizes was associated with recurrence at 3 months (N=14, p>0.99, AUC=0.57); 3 and 6 months (N=22, p=0.77, AUC=0.55); and all timepoints (N=30, p=0.94, AUC=0.48).

This study elucidates the temporal correlation of LR-TR and mRECIST for SBRT-treated lesions. An inflection point in time with a differential correlation between LR-TR and mRECIST classifications was not found. This might be because APHE can persist >12 months, beyond the study duration. The predicted probability of mRECIST assignment was highest for LR-TR non-viable lesions. Notably, these lesions lack APHE and could reflect reader confidence conferred by this imaging characteristic. No evidence was found that mRECIST size could be used to provide further guidance in predicting recurrence for LR-TR equivocal lesions and an alternate strategy needs to be explored.