50 percent of ad libitum intake slows many age-related changes (Weindruch, 1996; Weindruch and Walford, 1988; Finch, 1990:506-536; Hopkin, 1995; Roth et al., 1995)—e.g., the mortality-rate doubling time, mean life spans, and maximum life spans of rodents are extended by 30 percent or more. Genotype-specific tumors are delayed, e.g., in the p53 "knock-out" mouse strain, which has a very high incidence of malignancy (Hursting et al., 1994). At necropsy, some very old food-restricted rats (nonmutant strains) lacked any display of gross pathology that could be assigned as a cause of morbidity or mortality (Maeda et al., 1985; Weindruch and Walford, 1988; Finch, 1990:518-519). The causes of death in these rats could include transients in blood pressure, cardiac rhythms, and blood glucose, which might not lead rapidly to gross organ pathology.
Food restriction also delays puberty and slows the age-related loss of ovarian oocytes in laboratory mice (Nelson et al., 1985). When mice are fed ad libitum, fertility recovers and is maintained to later ages than in fed controls. The effects of food restriction extend to delaying age-related impairments in key neuroendocrine functions, such as the preovulatory surge of gonadotrophins (McShane and Wise, 1996) and the pulsatile release of growth hormone (Sonntag et al., 1995). The ad libitum food intake of laboratory rodents may be atypical, because food resources are subject to seasonal and other fluctuations in natural populations. This plasticity in the reproductive schedule is hypothesized to be adaptive, because it would coordinate reproduction with food availability (Holliday, 1989; Harrison and Archer, 1989; Graves, 1993). In particular, Graves (1993) argues that these responses to food restriction demonstrate tradeoffs between energy expended on survival versus reproduction.
The generalizability of these effects of food restriction to humans is not clear. In ongoing studies with rhesus monkeys, chronic food restriction had effects on metabolism that, in part, resembled those observed in rodents. Food restriction decreased both blood glucose and insulin, (integrated 24-hour levels) and increased insulin sensitivity (Roth et al., 1995). This change might be expected to oppose the age-related trend in humans for subtle, but persistent, decreases in insulin sensitivity during aging, in association with age-related trends for increased levels of fasting blood glucose (Rowe et al., 1983; Harris et al., 1987).
I suspect that gene-environment interactions will eventually be identified that lead to vulnerability of particular organ systems for degeneration during aging, but that a great deal of plasticity will be observed in the age when a particular degree of impairment is manifested. A potential example of this is the strong statistical association of higher education with a lower incidence of Alzheimer disease at later ages, which is observed in five human populations on three continents (Katzman, 1993). Education, of course, may be a proxy (surrogate) variable for other primary environmental factors. The existence of at least four genetic risk factors for Alzheimer disease gives a basis for evaluating how