The concept that a benign adenoma can become malignant is widely believed, as the letters attest.

Dr. Neugut and Ms. Jacobson's observations that the prevalence of adenomas correlates with cancer incidence, that patients who underwent polypectomy have a higher cumulative risk for developing colon cancer, that adenomas and cancer are associated with similar risk factors, and that 5% of patients with untreated polyps developed cancer at another site are all manifestations of an epidemiologic association between adenomas and cancer. However, association does not prove causation. These observations are consistent with the hypothesis that affected patients have some underlying mucosal defect that results in both benign and malignant cellular proliferation. If this hypothesis is correct, removing the benign lesions will not prevent the malignant ones from appearing.

If such a defect is present, it is not surprising that benign adenomas appear at an earlier age. Because most neoplastic growths (98.6% by my model) are benign and develop over time (rather than sprouting at the same moment), the first growth to appear would almost always be predictably benign. The presence of a preexisting mucosal defect is just as parsimonious an explanation as postulating an adenoma-carcinoma sequence.

Dr. Urbanski expands on the histologic arguments that benign adenomas "transform" (or degenerate, defined by Webster as "descend to a markedly worse condition in kind or degree") into malignancy.

He notes that different histologic subtypes of adenomas have different risks for malignant transformation, citing as evidence the observation that flat or serrated adenomas are more likely to contain dysplastic elements. However, although dysplasia is a marker of a proliferative mucosa, one cannot assume that a malignancy must have an earlier dysplastic phase. We cannot know with certainty the past and future histologic or clinical behavior of any such lesion based on a single sample in time. To prove that a particular benign lesion becomes malignant, we would have to remove it, cut it up, establish total histologic benignity, reassemble it, reimplant it, and monitor its natural course. Clearly impossible. Nonetheless, the foundation of Dr. Urbanski's belief in committed adenomas is circumstantial.

The purpose of my model was not to identify any heretofore unknown etiologic factors for colon cancer, as Urbanski's analogy to lung cancer implies. Rather, it was to examine the most compelling argument presented by Neugut and Jacobson, namely, that when some patients had their polypoid lesions histologically examined after many years of observation, cancer was found. How long can a malignancy remain confined to one place? I estimate that at any one moment, 725 000 people in the United States have malignant colonic polyps. This is not consistent with the number of symptomatic colon cancers that develop, unless the progression to symptoms takes many years.

A latency lasting perhaps decades belies the interpretation by Neugut and Jacobson that the patients with polyps "develop cancer at the same site". These polypoid lesions could have been malignant (histologically speaking) from the beginning, rather than originating from a subpopulation of adenomas committed to transformation.

Whether adenomas become cancer is not the clinical issue. What we can prove by randomly assigning patients to either screening or "no screening" (with polypectomy when lesions are found) is whether screening and polypectomy constitute a reasonable policy. At this time, we do not know if polypectomy will reduce the risk for cancer. The annual cost of such a national policy (after a $17 to $46 billion startup) would be over $6 billion and would require more endoscopists than currently exist [1].

We seem to have bicoastal (New York and California), if not international (Canada), agreement that the evidence regarding the adenoma-carcinoma sequence is circumstantial, that malignant polyps can have long latency periods, and that some colon cancer arises de novo. Until one or more well-designed and -executed prospective, randomized controlled trials show that screening and polypectomy reduce the morbidity and mortality of colon cancer in an economically feasible way, we should not undertake such efforts.