WASHINGTON — A retrospective look at the Surveillance, Epidemiology, and End Results database indicates that too few nodes are being evaluated in some patients with pancreatic cancer, which may result in understaging, Dr. Mark Slidell said at a symposium sponsored by the Society of Surgical Oncology.
In the analysis of more than 4,000 patients, those who did not have any nodes examined were 44% more likely to die during follow-up, Dr. Slidell said. Several investigators have suggested that extended lymphadenectomy should be performed routinely, but that has not been conclusively shown to improve outcomes in previous studies, he said.
In addition, the data showed that the lymph node ratio—total number of positive nodes divided by the total number of examined nodes—is an important predictor of survival, said Dr. Slidell, a resident at Georgetown University, Washington. He and his colleagues at Johns Hopkins University in Baltimore had hypothesized that the ratio of metastatic to examined nodes may be more important for staging and survival than the number of nodes harvested alone.
But data on this measure, the lymph node (LN) ratio, are limited, with most previous studies performed at single institutions and academic centers.
He and his colleagues identified 4,005 patients (2,042 men and 1,963 women) in the National Cancer Institute's SEER database who had resection for pancreatic adenocarcinoma between 1988 and 2003. The patients' median age was 66, and most had tumors that were larger than 2 cm.
The database included standard demographic information and tumor size, grade, LN involvement, total number of LNs examined, and the number of positive nodes. The authors calculated the LN ratio by dividing the total number of positive nodes by the total number of examined nodes.
The median tumor size was 3 cm. Most patients who had surgery had a pancreatectomy.
The median number of nodes examined was seven. Most patients had fewer than 12 nodes examined, and 390 patients (10.1%) had 0 nodes examined. Of the 3,478 patients who did have nodes examined, 1,507 (43.3%) had no metastases and were classified as N0, and 56% (1,971) had metastatic disease (classified as N1).
The mean number of nodes examined in the negative group was 8, while the mean in the node-positive group was 11.
Overall, median survival was 13 months and the 5-year survival rate was 17%.
With multivariate analysis, the prognostic factors related to survival included tumor stage, tumor grade, a tumor size of greater than 2 cm, the number of nodes examined, the LN ratio, and N1 disease.
Patients with zero nodes examined were at highest risk of disease-specific death, as they were 44% more likely to die during follow-up, Dr. Slidell said.
Dr. Slidell and his colleagues also evaluated whether a greater number of positive nodes was associated with decreased survival. N0 patients had significantly better survival than N1 patients, but within the group of patients with N1 disease, an increasing number of positive nodes was not significantly associated with poorer survival; however, it was associated with a nonsignificant trend toward decreased survival, said Dr. Slidell.
N1 disease also increased the risk of disease-specific death, as did a higher LN ratio. Five-year survival for N1 disease was 7%, compared with 18% for N0 disease.
LN ratio proved to be even more important. For a ratio of 0, the median survival was 17 months, for ratios of 0–0.2 it was 15 months, and for 0.2–0.4 it was 12 months. Median survival declined to 10 months for ratios greater than 0.4.
The LN ratio appears to be a better predictor of survival, and should be considered for use as a potential stratification method in future clinical studies, he said.