ORLANDO—Survival and staging in colon cancer are related to the number of lymph nodes containing metastasis. But according to several reports from the 2008 Gastrointestinal Cancers Symposium, there may be a better way to determine prognosis. The “index of metastasis,” also called the “lymph node ratio,” may be a more accurate predictor of cancer-related survival than the number of positive nodes, according to investigators who presented data at the meeting.
Index of metastasis
George J. Chang, MD, assistant professor of surgical oncology, M.D. Anderson Cancer Center, presented an analysis of 57,410 patients identified from the 2007 SEER registry with stage III colon cancer diagnosed between 1998 and 2004 (abstract 336).
Five-year cancer-related survival rates by stage at diagnosis were determined for these patients. This was correlated with the index of metastasis (IM), which is the number of positive lymph nodes divided by the total number of lymph nodes examined.
This is the first study to quantify the advantages of IM over the American Joint Committee on Cancer (AJCC) model and to delineate the range over which IM remains accurately prognostic, Dr. Chang noted.
The median number of lymph nodes evaluated was 10 for T1-2 tumors, 12 for T3 tumors, and 11 for T4 tumors. The median IM differed based on T stage (see Table 1).
Within each T-stage group, the IM more accurately predicted survival than nodal (N) stage.
Depending on IM quartile, 5-year cancer-related survival ranged from approximately 77% to 88% for T1-2 tumors, from 42% to 76% for T3 tumors, and from 20% to 60% for T4 tumors.
By nodal stage, 5-year cancer-related survival for T1-2 tumors was 85% for N1 disease and 73% for N2 disease; for T3 tumors, 68% and 47%, respectively; and for T4 tumors, 50% and 28%, respectively.
In an interview, Dr. Chang explained the benefit of this approach, noting that AJCC staging does not account for the cases in which only a few lymph nodes are retrieved. Survival within an AJCC/TNM stage can range widely, due to the heterogeneous nature of the substage cohort and variations in the quality of the surgery and pathological assessment.
“AJCC staging is extremely sensitive to the number of nodes,” Dr. Chang said. “The IM provides a much more accurate estimation of survival and is stable to variations in total nodes evaluated.”
For example, for patients with T3N1 disease, there is a 22% difference in survival between patients with less than 8 nodes examined (56%) vs those with more than 18 nodes examined (78%) (ie, lowest vs highest quartiles of nodes). For T4N1 disease, the difference is 24%. Using the IM, however, the difference is just 3% to 7% between IM quartiles 1 and 3, for each tumor stage, he pointed out, and only 12% to 15% for IM quartile 4.
“By using this model, we can do a survival analysis irrespective of the total number of nodes recovered,” he said. “If you recover 4 nodes and 2 are positive, the IM is 0.50. If you recover 8 nodes and 2 are positive, the index is 0.25. The outcomes will be very different for these two situations.”
M.D. Anderson investigators plan to develop a nomogram or calculator that will make use of the IM practical for the clinician, he said.
Strong correlation between IM and survival was also reported by investigators from the University of Arkansas for Medical Science (abstract 488), who examined 381 patients with colon or rectal cancer and grouped them by lymph node ratio (LNR) (ratio of positive to total nodes) and by number of nodes resected.
They found that patients with lower LNRs had improved survival (P = .028), irrespective of the number of nodes harvested (see Table 2). The number of nodes removed did not affect survival in either node-positive or node-negative patients, reported Rhonda W. Gentry, MD, a third-year fellow in the Division of Hematology/Oncology.
Do surgical variables matter?
Pathological examination of 12 or more lymph nodes has been advocated as a quality measure (see ONI, Jan. 2007, page 2) and is the metric for which most surgeons and pathologists currently strive. It is assumed that the quality of resection and pathologic assessment influences the number of nodes recovered, but nodal number is also affected by patient and tumor variability.
In a multicenter study (abstract 453), researchers asked whether there are surgical variables that can predict lymph node number in patients undergoing standardized resection. The study included 787 patients with stage I-III colon cancer from clinical trials. The median number of nodes recovered was 12 (range, 0 to 72).
Kellie Mathis, MD, a surgery resident at the Mayo Clinic, Rochester, Minnesota, reported substantial variability in lymph node harvests. “Even patients with very large specimens sometimes had very few nodes,” she said.
Lymph node number was only modestly influenced by surgical variables, such as length of proximal margin, distal margin, total bowel, and tumor location. Tumors on the right side vs left side of the colon were more likely to be associated with the finding of 12 or more nodes.
“We found there were no specific measurements that guaranteed the harvesting of 12 or more nodes,” Dr. Mathis said.
Can something be done to improve the odds? “Surgeons need to do anatomic resections properly and consistently, and pathologists just have to look diligently. That’s all they can do,” Dr. Mathis said.
Node counts rising
By national data, recovery of 12 or more nodes is achieved in only about 50% of cases, but Dr. Chang maintained it is possible to recover 12 or more nodes in almost all cases, when surgeons and pathologists truly feel the need to be thorough.
At the GI meeting, Dr. Chang and his colleagues presented a retrospective review of 480 consecutive patients with colon cancer surgically resected at M.D. Anderson between 1998 and 2005 (abstract 448). Over the 7 years, the proportion of patients who had 12 or more nodes examined increased progressively from 62.3% in 1998 to 94.3% in 2005, and the median number of lymph nodes increased from 14 to 22, he reported.
These improvements occurred as a result of an increased multidisciplinary awareness and greater emphasis on adequate evaluation, Dr. Chang said.
“We have learned that the more lymph nodes examined, the better the outlook for stage II and III patients. Our study suggests that we can change the practice of surgeons and pathologists by making them aware of the importance of increasing the number of lymph nodes examined and by emphasizing this in tumor board discussions,” he told ONI.
Fewer than 12 nodes was most likely when tumors were located in the transverse or left colon vs the right (P = .0003). The number of lymph nodes was independent of sex, race, age, or tumor grade.
Dr. Chang said that both surgeons and pathologists are responsible for improving the lymph node yield. When there is a low yield on pathological exam, it is wise to request another review of the specimen. While high-volume centers are meeting the recommendations, oncologists at smaller centers may need to be proactive with their surgeons and pathologists to assure this is done, he added.