Sir

We are accustomed to think of programmed senescence and apoptosis as protections against cancer. A fine, recent review by Judith Campisi invites speculation that such protections also entail a cost in cancer susceptibility1. She discusses work supporting the view that certain tumour suppressors protect from cancer but accelerate ageing. However, there is another related factor involved in such 'antagonistic pleiotropy'. Papers by Frank and Nowak2 and others3,4 have shown that strategies for tissue production and maintenance that minimize total cell division, and lineage branch length in particular, engender a lower risk of tumorigenic mutations. The reverse is also true. Decades ago, cancer researchers began to characterize the so-called tumour promoters — agents that stimulate cell proliferation and increase the probability of cancer.

It is well known that aged animals have a reduced capacity for regeneration. Their stem cells are fewer in number and/or have diminished replicative properties5. Notwithstanding this fact, certain tissues — such as the gut — continue to create replacement cells at a constant rate throughout life. These seemingly contradictory observations can be reconciled by the hypothesis that as animals age, fewer of the original adult stem cells assume more of the burden for tissue production.

If we combine the theoretical work of Frank and Nowak with experimental observations in ageing tissues, we reach the paradoxical conclusion that programmed cellular senescence must be partly tumorigenic. In the ageing animal, replacement of equivalent numbers of cells after functional loss of stem cells requires that the remaining stem cells divide more. So, as attrition continues throughout life, surviving stem cells shoulder more and more of the load, increasing the branch length of their descendants. Indeed, the survivors might be those stem cells that, as Frank and Nowak propose, have sustained mutations early during development, which enhance their longevity. Studies of somatic evolution in the colon, blood and other tissues have yielded results consistent with genetic drift and an age-related decline in stem-cell populations, as well as a supralinear increase in somatic mutation number with age6.

Programmed senescence might explain all or part of the exponential rise in cancer incidence/mortality and age7. Depending on the original number of adult stem cells, their rate of loss, and the regenerative demands of the tissue, branch lengths might increase in different populations at different rates. For instance, some cell populations — such as lymphoid tissue — might experience a gradual decline in stem-cell function and a comparatively flat age-versus-incidence relationship. By contrast, prostate epithelium might have a sudden and relatively late crisis in stem-cell numbers, perhaps accounting for the steep dependence on age of prostate cancer. The basis for such tissue differences is not known, but might involve telomere shortening as well as activity of “gatekeeper” tumour suppressors. The observation that telomerase-deficient mice are prone to tumours has been attributed to genetic instability as a result of chromosome-end fusions8. But this heightened cancer susceptibility might also be a consequence of the increased branch length in those stem-cell clones that outlive their sisters.

Whether or not telomere shortening is involved, programmed senescence fosters an opportunity for the stem cells that endure to expand into the void, increasing generation number and the probability of tumorigenic mutation. These considerations have implications for the use of adult stem cells in therapy. To borrow from J. B. S. Haldane, high cancer mortality in old age might be the price we pay for the privilege of low cancer incidence in youth.