Adverse events weigh on colonoscopy mortality equation

Virtual colonoscopy's ultimate success or failure is tied in many ways to that of optical colonoscopy. Some factors, such as the relatively rare progression of polyps to colorectal cancer, are inseparably linked to both techniques. Others, such as the risk of adverse events, are more often associated with conventional colonoscopy.

At the same time, referrals from VC to colonoscopy expose patients unavoidably to the risks of the latter. So while negative screening patients may benefit from VC's better safety profile, those who need polypectomy will not. As a result, providers of both screening methods understand that the relationship between the two techniques is complex and multifactorial.

One expert with an eye on the big picture is Dr. David Lieberman, who spoke at the American Society of Clinical Oncology's Gastrointestinal Cancers Symposium in San Francisco in late January. Lieberman, who is chief of gastroenterology at Oregon Health and Science University in Portland, OR, took on the weighty question of whether colonoscopy produces a mortality benefit.

The short answer is that colonoscopy can potentially extend lives, when performed correctly and in the right circumstances. The larger question of whether colonoscopy produces a mortality benefit remains, for now, unanswerable. The relevance of the question to virtual colonoscopy, which may be safer but shares a similar cost and diagnostic accuracy profile with colonoscopy, is unavoidable.

Lieberman noted indirect evidence of a colonoscopy screening benefit from trials that look for fecal occult blood. Fecal occult blood testing (FOBT) is cheaper, safer, and far less accurate than colonoscopy, but has a lot of data on its side.

Large randomized FOBT trials have been conducted, including the Minnesota Colon Cancer Control Study of 45,000 subjects (New England Journal of Medicine, 1993, Vol. 328:19, pp. 1365-1371), the Nottingham Colorectal Cancer Screening trial (Gut, June 2002, Vol. 50:6, pp. 840-844), the Funen trial in Denmark (Scandinavian Journal of Gastroenterology, September 2004, Vol. 39:9, pp. 846-851, and June 1989, Vol. 24:5, pp. 599-606), and the French biannual screening trial of more than 20,000 subjects (Gastroenterology, June 2004, Vol. 126:7, pp. 1674-1680).

"The results were very consistent in each one of these studies -- cancers, when discovered, were discovered at an early stage in patients who were screened versus those who were not screened," Lieberman said. Over time it resulted in mortality reductions of 17% to 20%, and among patients who adhered to the screening regimen, the mortality reduction ranged from 30% to 39%. The results show that mortality can be achieved with population-based screening, and that the identification of early colorectal cancers is clearly beneficial, he said.

Similar benefits were demonstrated in flexible sigmoidoscopy studies, which were case-controlled studies but not randomized. They showed that mortality could be reduced through screening, but only in the portion of the colon that sigmoidoscopy could examine. The implications of these studies also weigh in colonoscopy's favor, Lieberman said.

The National Polyp Study of more than 1,400 patients followed over six years provides indirect evidence of a mortality reduction with colonoscopy. The colorectal cancer rates were lower than would have been expected in an unscreened population, and only five interval cancers were detected, compared to the 15-20 cancers that would have been expected, Lieberman said (Gastroenterology, February 1990, Vol. 28:2, pp. 371-379).

Colonoscopy no panacea

Still, the evidence suggests that colonoscopy "may not be as effective as we think it is," Lieberman said, when the weight of factors such as incomplete colonoscopy, adverse events, surveillance issues, and bias issues are included.

He cited a number of studies that looked for interval colorectal neoplasia after colonoscopy, and found interval cancers in 0.6% to 0.9% of the follow-up group over two to three years. (e.g., Schatzkin et al, 2000, 1,905 patients, 0.6% cancers; Alberts et al, 2000, 1,303 patients, 0.7% cancers; VA study, 2004, 872 patients, 0.9% cancers; Dartmouth study, 2005, 2,915 patients, 0.9% cancers).

"The point is that there is a significant rate of colon cancers even after experts have performed colonoscopy with complete polypectomy," Lieberman said. Another study, by Schoen et al, followed up sigmoidoscopy patients with the same exam at three years and found a number of advanced lesions in the distal colon.

In all, 3.1% (292/9317 subjects) were found to have an adenoma or cancer at follow-up, the authors wrote. There were 78 (0.8%) advanced adenomas (n = 72) or cancers (n = 6) in the distal colon, and 80.6% (58/72) had lesions found in a portion of the colon that had been adequately examined at the initial sigmoidoscopy, according to the authors (Journal of the American Medical Association, July 2003, Vol. 290:1, pp. 41-48).

There are several possible reasons for interval lesions after colonoscopy, including missed lesions, incomplete exams, and lesions not completely removed, he said.

First, the rate of incomplete colonoscopy is not insignificant. A large U.S. study reported complete colonoscopy to the cecum in 93% to 99% of patients, while a British study had a cecal intubation rate of just 57% in more than 9,000 patients (Gut, February 2004, Vol. 53:2, pp. 277-283). Exam technique and the quality of bowel prep clearly affect the completion rate, Lieberman said.

The three largest trials that compared virtual colonoscopy to conventional colonoscopy (Pickhardt et al, Rockey et al, Cotton et al) also demonstrated missed lesions at colonoscopy. "They found that 2% to 12% of patients that had polyps greater than 1 cm, which I think most of us would consider significant, were missed on optical colonoscopy," he said.

Pickhardt and his former colleagues at the U.S. Department of Defense examined 1,233 patients, and with the aid of segmental unblinding pegged optical colonoscopy's sensitivity at 87.5%. Lieberman noted that the group attributed missed lesions at colonoscopy to polyps that were located behind folds or otherwise difficult to examine with a colonoscope (New England Journal of Medicine, December 3, 2003, Vol. 349:23, pp. 2261-2264).

Also fueling doubt was a recent study in which colorectal lesions were found in the identical site in the colon where polyps had been removed three years earlier. The dietary intervention trial detected 13 cancers at follow-up among 2,079 elderly patents (2.2 cases per 1,000 person years observed). Seven of 13, (53.8%) of patients had either a potentially "avoidable" cancer or one detectable at an earlier time interval because of incomplete removal (4/13) or missed cancer (3/13), the study authors wrote (Gastrointestinal Endoscopy, March 2005, Vol. 61:3, pp. 385-391).

"Finally, it's possible that lesions grow faster than we think," Lieberman said. Biological deficits have been shown to influence disease incidence and progression. "We know that about 15% or so will have microsatellite instability leading to sporadic cancer," and that some lesion types are fast-growing, he said. Also, if lesion surveillance is inappropriately timed, the effectiveness of a colonoscopy program can be reduced, he said.

"There can be too much (surveillance) or too little -- there's evidence in the U.S. that we do too much," he said.

Contrary to published guidelines, a study that polled physician attitudes on surveillance found that 24% of GI physicians and 54% of surgeons recommended surveillance of hyperplastic polyps. Seventeen percent and 43%, respectively, recommended follow-up exams in less than three years, perhaps exposing patients to the risks of colonoscopy prematurely.

On the other hand, Lieberman said, too little surveillance can result in interim cancers. Recently published Medicare data found that the longer the screening interval, the higher the rate of interim polyps.

Colonoscopy risks

The VA cooperative study found a 0.3% rate of serious complications following colonoscopy (9/3, 196 procedures), including lower GI bleeding requiring intervention (n = 6), myocardial infarction and/or cerebrovascular accident (n = 2), and thrombophlebitis (n = 1). In subjects undergoing only diagnostic procedures, the major complication rate was 0.1%, according to the authors (Gastrointestinal Endoscopy, March 2002, Vol. 55:3, pp. 307-314).

Three deaths occurred over the 30-day period following the exams, Lieberman said. And although many of the reported complications seemed minor, "in older patients with heart disease or lung disease they might be somewhat more significant," Lieberman said of the study, which he co-authored.

A large U.S. study gave colonoscopy a good safety profile. Among 116,000 patients undergoing colonoscopy in ambulatory service centers across the U.S., it revealed a relatively low rate of 37 (0.3%) incidences of perforation and no deaths.

According to the authors, the most common site of perforation was the sigmoid colon (62%), followed by the ascending colon (16%); cecum, transverse colon, and splenic flexure (11%); and rectum, anastomotic, or unknown (11%). The time to diagnosis ranged from immediate to 72 hours (29 for less than one hour, eight for more than one hour). All patients were hospitalized; 35 (95%) underwent exploratory laparotomy and two (5%) were treated conservatively (Gastrointestinal Endoscopy, October 2003, Vol. 58:4, pp. 554-557).

A U.K. study of 9,223 colonoscopies produced far worse results, with an overall perforation rate of one for every 769 patients. Colonoscopy was considered a possible factor in six deaths that occurred within 30 days of the procedure, according to the authors. Training was lacking, however. "Only 17.0% of colonoscopists had received supervised training for their first 100 colonoscopies and only 39.3% had attended a training course," the authors wrote (Gut, February 2004, Vol. 53:2, pp. 277-283).

For now it appears that the overall risk of serious complications from colonoscopy is quite low, Lieberman said.

"But I think the other complications to remember are the possibility that exams may be incomplete, lesions may be missed, there may be incompletely removed lesions, and quite frankly, the rate of all these events is unknown in medical practice," he said. "The data I've shown ... so far comes from carefully controlled studies, and so in real-life practice we're really uncertain, and there have been no studies in clinical practice that have followed patients for 30 days the way a surgeon would."

The cumulative risk of multiple colonoscopies is also unknown and may increase with each exam, Lieberman said, citing an editorial by gastroenterologist Dr. David Ransohoff. The letter suggested that the odds could be against a hypothetical 50-year-old patient who entered a screening program and underwent three screening colonoscopies 10 years apart, with a cumulative risk of 1% of serious complications for each one. Under that scenario, one out of 10 patients could potentially die from colonoscopy screening (Gastroenterology, May 2005, Vol. 128:6, pp. 1685-1695).

The bigger picture

Finally, there is the idea of death from all causes, a steep hurdle for any diagnostic test to clear.

Is mortality reduced if we prevent a colorectal cancer death? Not if a patient dies from another cause -- say a heart attack -- around the same time colorectal cancer would have killed him, Lieberman said. Nor does the patient live longer if colonoscopy finds and removes the cancer but a repeat exam 10 years later leads to myocardial infarction and death -- or if colonoscopy removes an asymptomatic cancer and the patient dies from the procedure. For the time being, the net effect of colonoscopy on mortality of all causes cannot be accurately predicted.

More needs to be known about the harms, risks, and benefits of colonoscopy, he said, and even then, true randomized studies are impractical, necessitating the use of population models, which have a difficult time accounting for day-to-day practice realities such as poor preps and missed lesions.

A recent study by Ladabaum and Song from the University of California, San Francisco, used a Markov model to assess the impact of screening 75% of the U.S. population for colorectal cancer, Lieberman noted.

The hypothetical effort was expensive, but successful. According to the authors, screening decreased colorectal cancer incidence by 17% to 54% to as few as 66,000 cases per year, and reduced colorectal cancer mortality by 28% to 60% to as few as 23,000 deaths per year. Total annual national colorectal cancer-related expenditures were $8.4 billion without screening. Hypothetical expenditures dropped to between $1.5 billion and $4.4 billion with screening; however, total expenditures increased to between $9.2 billion and $15.4 billion dollars. Screening colonoscopy alone required 8.1 million colonoscopies per year (Gastroenterology, October 2005, Vol. 129:4, pp. 1151-1162).

"The potential benefits are clear that we might be able to prevent cancer by removing adenomas," Lieberman said, "and we certainly have evidence that we could reduce the incidence and should reduce the mortality risk."

By Eric Barnes staff writer
February 6, 2006

Related Reading

VC would raise screening costs, study concludes, October 31, 2005

Fecal occult blood screening cuts colorectal cancer mortality, June 25, 2004

Lower cardiovascular risk makes virtual colonoscopy safer, study finds, December 31, 2003

Screening colonoscopy worthwhile into the eighth and ninth decades, September 1, 2003

Copyright © 2006

Page 1 of 652
Next Page