CHEMOEMBOLIZATION-INDUCED TUMOR NECROSIS: CORRELATION OF MRI, PATHOLOGY AND CLINICAL OUTCOME IN CIRRHOTICS WITH HEPATOCELLULAR CARCINOMA.

Abstract

345 Purpose: We set out to determine how well MRI estimates of chemoembolization (CE) - induced tumor necrosis correlate with explant pathology and whether either is a predictor of outcome following OLT and chemo for the treatment of hepatocellular carcinoma (HCC). Methods: MRI is performed in all OLT candidates to estimate liver volume, to detect mass lesions, and to assess portal vein patency. Patients with known HCC undergo CE, OLT, and chemo. Biopsy for diagnosis of HCC was not performed routinely. Following each CE an MRI was done to determine percent necrosis of the targeted tumor(s), change in tumor size, and to check for additional lesions. Following OLT the explanted livers were examined and CE-induced tumor necrosis was quantified. Results: 360 OLT's were performed in 342 patients. CE was performed prior to OLT on 23 patients with known HCC. Three patients were excluded from analysis: two died of sepsis within 30 days of OLT and one patient had persistent disease after OLT. 16/20 patients are alive (median follow-up = 19 months). Six patients had recurrent HCC. MRI and pathologic data were complete for 13 patients with a total of 17 lesions. Pathologically nine tumors demonstrated >80% necrosis. Four tumors showed no necrosis and four demonstrated 20%, 30%, 70%, and 70% necrosis. In addition to necrosis it was noted that chemoembolized tumors were surrounded by a thick fibrous capsule. Estimates of tumor necrosis by MRI were within 10% of their pathologic assessment in 13/17 (76%) lesions. Percent necrosis by MRI and pathology in the 4 lesions that showed poor correlation were, respectively, 31% vs. 100%, 29% vs. 90%, 27% vs. 70%, and 87% vs. 70%. Cumulative adriamycin doses during CE of < 50 mg were used to treat 12 lesions whose mean necrosis was 14% whereas CE doses of >50 mg were used for 23 lesions whose mean necrosis was 74% (p<0.05). The number of CE treatments (1 vs. multiple) and the interval from CE to OLT (within 1 month vs. > 1 month prior to OLT) was not correlated with percent necrosis. There was no correlation between necrosis and clinical outcome. The six patients whose disease recurred had a mean tumor necrosis in 11 lesions of 44%. The fourteen patients who remain disease-free had a mean tumor necrosis in 24 lesions of 58%. Conclusions: The timing and frequency of CE did not have an impact on the degree of tumor necrosis but adriamycin doses > 50 mg administered during CE resulted in significantly more tumor necrosis than doses < 50 mg. MRI estimates of tumor necrosis compared very well with actual necrosis as determined by pathologic examination. There was no correlation between extent of tumor necrosis and recurrence rates. The presence of the fibrous capsule may be critical since the beneficial effect of CE in the treatment of HCC does not correlate with the degree of tumor necrosis induced by CE.