乌鸦传媒

SAN FRANCISCO -- Failure to have surveillance colonoscopy after treatment for stage I colon cancer tripled the relative risk of cancer-specific death, according to data from a national cancer registry.

By two different statistical methods, patients who did not have surveillance colonoscopy had more than a three-fold increase in cancer-specific death during follow-up. Another analysis showed that patients who had two surveillance colonoscopies had almost a 50% reduction in the hazard for cancer-specific death compared with those that had one.

Although the absolute risk of cancer-specific death was small, the results suggested that surveillance colonoscopy might have helped prevent a majority of the deaths, said Robert B. Hines, PhD, of the University of Central Florida in Orlando, at the Gastrointestinal Cancers Symposium.

"We performed two analyses, both of which used valid methods, and we found the same thing: Patients who did not have an initial colonoscopy had over a three-fold greater risk of cancer-specific death," Hines told 乌鸦传媒. "In the second analysis, we could also look at the effect of a second surveillance colonoscopy, and we found a 47% reduction in the relative risk of cancer-specific death, although the absolute difference was very small."

The data are eye-opening, but more studies are needed to clarify the factors involved in the observed cancer-specific survival differences, said Marcia Cruz-Correa, MD, PhD, of the University Puerto Rico Comprehensive Cancer Center in San Juan.

"This study is well done, the data are clear, and there is definitely an association, but we don't know why," said Cruz-Correa. "We don't think it is the colonoscopy. In patients with stage I disease, it's gone after surgery. We don't think there is residual disease."

"This is a great data set to open our eyes," she added. "It says there is something going on, but we don't know yet if the colonoscopy is the driver. We need more good data like this to help us understand what is going on."

Guidelines for surveillance colonoscopy for colon cancer differ with regard to recommendations for stage I disease, which has a favorable prognosis. Several organizations, including the National Comprehensive Cancer Network and the American Society of Colon and Rectal Surgeons, recommend an initial colonoscopy within the first year of treatment of stage I disease and a second surveillance procedure 3 to 5 years after the first one, said Hines.

The American Society of Clinical Oncology has no colonoscopy recommendation for patients with stage I disease because of a lack of evidence, and no studies had examined the impact of colonoscopy on colon cancer-specific survival in patients with stage I disease, he continued. In an effort to provide some evidence, investigators analyzed data from the NCI Surveillance, Epidemiology, and End Results registry program linked to Medicare claims and they identified patients, ages 66-84, with newly diagnosed stage I colon cancer.

Hines and colleagues categorized the patients according to when they had an initial surveillance colonoscopy: 1, 2, or 3 years after surgery, or no surveillance. Propensity-score weighting was used to balance covariates. The study population comprised 8,783 patients, more than one-fourth with no surveillance colonoscopy.

Investigators performed two statistical analyses. The first analysis included only patients with ≥3 years of follow-up, which coincided with the time frame for initial colonoscopy. The second analysis included all follow-up data and treated first and second surveillance colonoscopy as time-dependent exposures. They also calculated the number needed to treat (NNT) to prevent one colon cancer-specific death.

The first analysis yielded a 5-year survival of 99.11% for patients who had their initial surveillance colonoscopy within a year of surgery. The estimated survival dipped slightly to 98.82% and 98.55% for patients who had their initial surveillance colonoscopy within 2 or 3 years after surgery. Neither value differed significantly from the patients who had colonoscopy within a year of surgery.

Patients with no surveillance colonoscopy within 3 years had an estimated 5-year survival of 97.26% as compared with the reference group that had a first surveillance colonoscopy during the first year after surgery. Though small, the difference translated into a hazard ratio of 3.11 (P<0.001).

The second analysis showed an estimated 5-year survival of 97.53% for patients who had an initial surveillance colonoscopy and 98.69% for the patients who had a first and second colonoscopy after surgery, a small but statistically significant improvement that represented a 47% reduction in the hazard ratio (P=0.022).

For patients who did not have surveillance colonoscopy, the estimated 5-year survival declined to 92.14%, associated with a hazard ratio of 3.27 versus the patients who had an initial colonoscopy (P<0.001).

Both of the NNT analyses suggested that most colon cancer deaths might have been prevented with surveillance colonoscopy. The first analysis showed 60 cancer-specific deaths among 2,138 patients and an NNT of 54.1, which would have prevented an estimated 39.6 deaths (66.9%). In the second analysis, 169 colon cancer deaths occurred in 2,425 patients who did not have colonoscopy. The analysis suggested an NNT of 18.6 to prevent 130.6 of the deaths (77.3%).

Comment