The contemporary view of cancer is that a malignancy arises from the transformation of the genetic material of a normal cell, followed by successive mutations, ultimately leading to the uncontrolled proliferation of progeny cells. This view can be traced to the pioneering work of Boveri [I 1, who in the early decades of this century formulated his" somatic mutation" hypothesis of the origin of cancer. Succeeding generations of investigators have added to and modified the original theory. Three contributions are particularly germane:(a) the work of Berenblum [2], who demonstrated the multistep, sequential nature of carcinogenesis;(b) the studies of Knudson [3], who showed that an inherited defect in one allele of a protective gene can predispose a person to cancer; and (c) the clonal-evolution model of tumor progression proposed by Nowell [4]. According to this hypothesis, a cell that acquires a specific genetic alteration may develop a proliferative advantage. Clonal expansion of this cell, driven by successive mutations, could lead to tumor progression. More recently, molecular studies have strongly supported the idea that multiple genetic changes are required for tumorigenesis. An archetypal example is that of colorectal malignancies, which involve genetic alterations on chromosomes 5q. 12p. 18q, and 17p and possibly other lesions as well [5, 6]. Moreover, recent evidence lends credence to Nowell's hypothesis: a specific mutation of thep53gene has been found to lead to selective growth advantage and thus to progression from low-grade to high-grade brain neoplasia [7].

However, despite the impressive amount of data that has been accumulated, a basic conundrum remains: precisely how many mutations are required for the pathogenesis of a specific tumor? Several attempts have been made to address this issue. In this Working Hypothesis some of these early models are modified and refined, based on current data, to provide fresh insights and to stimulate discussion.