A surgically unresectable, biopsy-proven hepatocellular carcinoma (HCC) developed in a 63-year-old man with cirrhosis. He survived 24 months without treatment. During that time the tumor burden decreased as demonstrated both radiologically and by the normalization of alpha-fetoprotein levels. The patient died of complications secondary to repeated esophageal variceal hemorrhage. Necropsy demonstrated prominent, ulcerated esophageal varices and liver cirrhosis without evidence of neoplasia either grossly or on a subsequent thorough microscopic examination. This case represents the first confirmation of HCC spontaneous regression in which a primary histologic diagnosis was confirmed by immunohistochemical and flow cytometric DNA analysis, and where tumor regression was proven by a thorough necropsy examination.

DNA ploidy and cell cycle analyses are being used with increasing frequency as additional, sometimes independent, biologic prognosticators of various malignancies. For obvious reasons, most of the large studies published are retrospective, using archival paraffin-embedded tissues likely to have been fixed and processed differently. To assess the effect of various fixation conditions on ploidy analyses, the authors performed a comprehensive study in which five contiguous aliquots from 26 tissue samples were fixed in formalin for 8. 24, and 48 hours; B5; and the ethanol-based fixative Omnifix (Omni, Xenetics Biomed Inc., Irvine, CA), respectively. The samples were prepared and stained with the use of Hedley's method; the DNA analyses were done with a Coulter EPICS V flow cytometer (Coulter Electronics, Hialeah, FL); and the resulting histograms were interpreted with Multicycle software (Phoenix Flow Systems, San Diego, CA). Mean channels and coefficients of variation (CVs) of the G0/G1 peaks were compared. Increasing fixation times in formalin produced slight shifting to the left on the G0/G1 peaks with differences up to 13 channels. Omni caused similar but more conspicuous changes, with maximal shifting of 51 channels. B5 created the opposite effect, with right-shifted G0/G1 peaks up to 22 channels. The CVs deteriorated progressively with increasing fixation times in formalin. B5 and Omni fixatives conferred the worse histograms, with 6 and 15 cases, respectively, displaying CVs greater than 8. Such significant differences preclude the use of any other internal standard but adjacent normal tissue that was processed exactly the same as the tumor. The validity of DNA results obtained otherwise should be considered questionable. These findings may explain some of the previously published controversial reports and emphasize the need to use an internal standard and more stringent criteria to define aneuploidy in paraffin-embedded tissue.