The median survival of all patients with hepatic metastases from colorectal cancer referred to the Sidney Farber Cancer Institute during a five-year period was 12.5 months. Two major factors influenced survival. The first was extent of disease at presentation. The second was the histologic grade of the cancer. The median survival of patients presenting with the least disease, characterized by less than four liver nodules visible on liver scan (n = 38), normal liver size on physical examination (n = 60), normal liver function test results (n = 30), and normal performance status (n = 91), was between 18 and 24 months, regardless of treatment. The median survival of those few patients (n = 13) who had objective responses to a variety of treatments, most of whom also had minimal disease at presentation, was also 24 months. Patients whose tumors were poorly differentiated or who had abnormal performance status or weight loss of greater than 10 per cent at presentation survived only six months (median). Those with four or more liver nodules, hepatomegaly (greater than 16-cm vertical span on physical examination), or abnormal liver function test results, survived ten, eight, and 12 months (median), respectively. It is concluded that a significant group of patients survived longer than would have been predicted by earlier literature surveys after the diagnosis of colorectal cancer metastatic to the liver. It is suggested that future therapeutic trials, using survival as a measure of response of patients with liver metastases from colorectal cancer, must be prospectively controlled before selection factors can be differentiated from significant therapy effect.

One hundred forty-five patients with colorectal cancer were analyzed in order to correlate the preoperative plasma carcinoembryonic antigen (CEA) levels with the sites and times of disease recurrence. The median follow-up periods of these patients was 50 months (range 36-72 months). Twenty-one patients were found to have metastases at the time of their operation. None of the seven patients whose primary tumors were classified as Dukes/Kirklin A have had tumor recurrence. Seventeen per cent of the patients with Dukes/Kirklin B tumors have had tumor recurrences, and 63% of the patients with Dukes/Kirklin C colorectal primary tumors have had tumor recurrence. No correlation was found between preoperative CEA values and subsequent risk of tumor recurrence or times to recurrence among the patients with Dukes/Kirklin B colorectal primary cancers. In Dukes/Kirklin C patients, however, elevated preoperative CEA values predicted a higher risk of tumor recurrence. Ninety per cent of the patients (19/21) with preoperative CEA levels greater than 5.0 ng/ml have had relapses, with a median time of 17 months before disease recurrence. Only 39% (9/23) of the patients with Dukes/Kirklin C lesions and CEA levels less than 5 ng/ml have had relapses and there is insufficient follow-up data as yet to determine the median survival time. If those patients whose Dukes/Kirklin C primary tumors were poorly differentiated on histologic examination are excluded, the contrast between patients having CEA levels greater than 5.0 ng/ml and those having CEA levels less than 5 ng/ml is even more marked. Sixteen of the 18 remaining patients whose CEA levels were greater than 5.0 ng/ml prior to curative resection have had relapses as compared with only three of 15 patients whose preoperative CEA values were less than 5. We conclude, therefore, that CEA is an important factor in stratifying patients after curative resection of their Dukes/Kirklin C colorectal tumors.