Impact of caloric restriction on health and survival in rhesus monkeys from the NIA study

Abstract

Calorie restriction (CR), a reduction of 10–40% in intake of a nutritious diet, is often reported as the most robust non-genetic mechanism to extend lifespan and healthspan. CR is frequently used as a tool to understand mechanisms behind ageing and age-associated diseases. In addition to and independently of increasing lifespan, CR has been reported to delay or prevent the occurrence of many chronic diseases in a variety of animals. Beneficial effects of CR on outcomes such as immune function, motor coordination and resistance to sarcopenia in rhesus monkeys have recently been reported. We report here that a CR regimen implemented in young and older age rhesus monkeys at the National Institute on Aging (NIA) has not improved survival outcomes. Our findings contrast with an ongoing study at the Wisconsin National Primate Research Center (WNPRC), which reported improved survival associated with 30% CR initiated in adult rhesus monkeys (7–14 years) and a preliminary report with a small number of CR monkeys. Over the years, both NIA and WNPRC have extensively documented beneficial health effects of CR in these two apparently parallel studies. The implications of the WNPRC findings were important as they extended CR findings beyond the laboratory rodent and to a long-lived primate. Our study suggests a separation between health effects, morbidity and mortality, and similar to what has been shown in rodents, study design, husbandry and diet composition may strongly affect the life-prolonging effect of CR in a long-lived nonhuman primate.

Figures

Fig 1

Kaplan-Meier Survival Curves for All-Cause Mortality for Old-Onset Monkeys. All-cause mortality in the old-onset monkeys was analyzed using Cox regression with age of onset, sex, and diet as predictors. The effect of Diet was not significant (p = 0.934) and Sex was the only significant predictor (p =0.003), therefore Kaplan-Meier survival curves for the four (diet-by-sex) groups were plotted to display the results. Open circles represent monkeys that are still alive.

Fig 2A

Body weight (kg) predicted from age-dependent individual-specific trajectories for CR and CON monkeys (Old-onset). There were significant changes in body weight over time [F(20,311) = 2.52, p = 0.0004]. There was a significant main effect for Sex [F(1,29) = 21.11, p < 0.0001] with males being substantially heavier. Also there were significant Sex-by-Year [F(18,311) = 2.96, p < 0.0001] and Diet-Sex-Year interactions [F(17,311) = 2.37, p = 0.0019] with reductions in body weight due to CR being more prevalent for the male animals. Solid lines represent males; dashed lines represent females. Overall body weight trajectories were based 420 observations for 34 monkeys (84 observations for 8 CON-F; 145 for 10 CON-M; 57 for 7 CR-F; 134 for 9 CR-M). For ages 15–20 years, there were: 12 observations for 5 CON-F; 17 for 7 CON-M; 9 for 5 CR-F; 7 for 4 CR-M. For ages 20–25 years, there were: 38 observations for 8 CON-F; 42 for 10 CON-M; 21 for 5 CR-F; 34 for 8 CR-M. For ages 25–30 years, there were: 21 observations for 5 CON-F; 40 for 8 CON-M; 20 for 4 CR-F; and 41 for 8 CR-M. For ages > 30 years, there were: 46 observations for 8 CON-M; 52 for 6 CR-M and too few to plot for the females. One CR Male that weighed over 14 kg at the start of the experiment had undue influence on the average trend lines for CR Males and therefore the first 5 observations of his data were not used to construct the plot; however, the omission of these observations had virtually no influence on the results of the statistical analyses.

Fig 2B

Fasting serum triglycerides (mg/dL) predicted from age-dependent individual-specific trajectories for Old-onset CR and CON monkeys. Triglyceride levels increased with age [F(16,162) = 2.12, p = 0.0096] and CR monkeys had significantly lower levels than CON [F(1,21) = 5.76, p = 0.026]. Overall triglyceride trajectories were based 249 observations for 34 monkeys (50 observations for 6 CON-F; 87 for 8 CON-M; 32 for 4 CR-F; 80 for 8 CR-M. For ages 20–25 years, there were: 16 observations for 5 CON-F; 10 for 6 CON-M; 5 for 3 CR-F; 3 for 3 CR-M. For ages 25–30 years, there were: 21 observations for 5 CON-F; 32 for 7 CON-M; 20 for 4 CR-F; 27 for 7 CR-M. For ages > 30 years, there were: 45 observations for 7 CON-M; 50 for 6 CR-M and too few to plot for the females.

Fig 2C

Cholesterol predicted from age-dependent individual-specific trajectories for Old-onset CR and CON monkeys. Cholesterol levels increased with age [F(53,774) = 1.54, p = 0.009], and male monkeys had significantly lower levels than females [F(1,24) = 23.60, p < 0.0001]. A significant 3-way Diet-Sex-Age interaction [F(40,774) = 1.53, p = 0.02] indicated that cholesterol levels increased with age for CON males while CR males tended to have a slight reduction in cholesterol. Thus, at older ages (> 30 years), CR male monkeys have significantly lower cholesterol levels as compared to CONs. Overall cholesterol trajectories were based on 994 observations for 28 animals (204 observations for 7 CON-F; 301 for 7 CON-M; 134 observations for 5 CR-F; and 355 observations for 9 CR-Male). For ages 15–20 years, there were: 48 observations for 6 CON-F; 65 observations for 6 CON-M; 21 observations for 4 CR-F; and 32 observations for 4 CR-M. For ages 20–25 years, there were: 91 observations for 5 CON-F; 102 observations for 7 CON-M; 62 observations for 5 CR-F; and 118 observations for 8 CR-M. For ages 25–30 years, there were: 41 observations for 5 CON-F; 70 observations for 6 CON-M; 39 observations for 4 CR-F; and 101 observations for 8 CR-M. For ages > 30 years, there were: 64 observations for 6 CON-M; and 104 observations for 7 CRM; and too few to plot for the females.

Fig 2D

Fasting serum glucose (mg/dL) levels predicted from age-dependent individual-specific trajectories for Old-onset CR and CON monkeys. Five glucose measurements above 100mg/dL for one diabetic CON-M were omitted to remove the influence of these outliers on the analyses and graphs. There were significant changes in glucose over time [F(20,285) = 10.48, p < 0.0001] and males and females were significantly different in the trends over time [F(18,285) = 3.58, p < 0.0001] with males having increases in glucose levels over time whereas the glucose levels of the females slightly decreased. The overall CR difference was not significant, F(1,22) = 1.18, p = 0.288, and the CR differences in trend over time were not significant, F(20,285) = 1.23, = 0.2259. Additional analyses stratified by sex conditions showed that CON males had significantly higher glucose levels compared to CR males, F(1, 14) = 5.27, p = 0.04. Overall glucose trajectories were based 387 observations for 34 monkeys (79 observations for 7 CON-F; 131 for 8 CON-M; 48 for 4 CR-F; 129 for 8 CR-Male). For ages 15–20 years, there were: 12 observations for 5 CON-F; 17 for 7 CON-M; 2 for 2 CR-F; 6 for 3 CR-M. For ages 20–25 years, there were: 33 observations for 7 CON-F; 39 for 8 CON-M; 19 for 4 CR-F; 30 for 7 CR-M. For ages 25–30 years, there were: 21 observations for 5 CON-F; 35 for 7 CON-M; 20 for 4 CR-F; 41 for 8 CR-M. For ages > 30 years, there were: 45 observations for 7 CON-M; 52 for 6 CR-M; and too few to plot for the females.

Fig 3A

Kaplan-Meier Survival Curves for All-Cause Mortality for Young-Onset Monkeys. All-cause mortality in the young-onset monkeys was analyzed using Cox regression with age of onset, origin, sex, and diet (p = 0.255) as predictors with none of these factors being statistically significant. Therefore Kaplan-Meier survival curves for the two diet groups were plotted to display the results. Open circles represent monkeys that are still alive.

Fig 3B

Kaplan-Meier Survival Curves for Age-Related Mortality for Young-Onset Monkeys. Age-related mortality in the young-onset monkeys was analyzed using Cox regression with age of onset, origin, sex, and diet (p = 0.975) as predictors with none of these factors being statistically significant. Therefore Kaplan-Meier survival curves for the two diet groups were plotted to display the results. Open circles represent monkeys that were censored as non-age related deaths or are still alive.

Fig. 4A

Fasting serum glucose (mg/dL) levels predicted from age-dependent individual-specific trajectories for young-onset CR and CON monkeys. 14 glucose measurements above 100mg/dL in diabetic monkeys (7 observations for 3 CON-M; 5 for 2 CR-M; 2 for 1 CON-F) were omitted to remove the influence of these outliers on the analyses and graphs. There were significant changes in glucose over time [F(18,1112) = 11.24, p < 0.0001], and males and females were significantly different in the trends over time [F(18,1112) = 1.98, p = 0.0088] with males having a larger increase in glucose levels over time. There was no significant difference due to diet group. Solid lines represent males; dashed lines represent females. Overall glucose trajectories were based 1260 observations for 81 monkeys (346 observations for 23 CON-F; 350 for 20 CON-M; 281 for 20 CR-F; 283 for 19 CR-M). For ages < 5 years, there were: 23 observations for 8 CON-F and 22 observations for 9 CR-F; data for males < 5 years was not available. For ages 5–10 years, there were: 71 observations for 18 CON-F; 55 for 19 CON-M; 65 for 16 CR-F; 48 for 15 CR-M. For ages 10–15 years, there were: 93 observations for 20 CON-F; 100 for 20 CON-M; 79 for 17 CR-F; 85 for 18 CR-M. For ages 15–20 years, there were: 95 observations for 20 CON-F; 95 for 19 CON-M; 75 for 16 CR-F; 75 for15 CR-M. For ages 20–25 years, there were: 53 observations for 17 CON-F; 85 for 18 CON-M; 30 for 12 CR-F; 64 for 15 CR-M. For ages > 25 years, there were: 13 observations for 6 CON-F; 22 for 12 CON-M; 10 for 2 CR-F; 16 for 11 CR-M.

Fig 4B

Fasting serum triglycerides (mg/dL) predicted from age-dependent individual-specific trajectories for young-onset CR and CON monkeys. There were significant changes in triglycerides over time [F(14,843) = 17.59, p < 0.0001] and males and females were significantly different in the trends over time [F(14,843) = 5.36, p < 0.0001]. Furthermore, there was a Diet-by-Sex interaction indicating that the overall effect of CR on triglycerides was significantly different for male and female monkeys [F(1,68) = 5.07, p = 0.0276]. Specifically, CR males had lower triglycerides than CON males. By contrast, CR females had higher triglyceride levels than CON females. Overall triglyceride trajectories were based 973 observations for 81 monkeys (266 observations for 23 CON-F; 280 for 20 CON-M; 213 for 20 CR-F; 14 for 19 CR-M). For ages < 10 years, there were: 32 observations for 9 CON-F; 30 for 8 CR-F; data for males < 10 years was not available. For ages 10–15 years, there were: 76 observations for 18 CON-F; 77 for 20 CON-M; 71 for 16 CRF; 58 for 16 CR-M. For ages 15–20 years, there were: 91 observations for 20 CON-F; 95 for 19 CON-M; 72 for 16 CR-F; 75 observations for 15 CR-M. For ages 20–25 years, there were: 55 observations for 17 CON-F; 86 for 18 CON-M; 30 for 12 CR-F; 65 for 15 CR-M. For ages > 25 years, there were: 12 observations for 5 CON-F; 22 for 12 CON-M; 10 for 2 CR-F; 16 for 11 CR-M.

Fig 5A

Incidence of three major age-related conditions. “Y” represents an occurrence in a young-onset monkey and “O” indicates an old-onset monkey at the age of diagnosis. Animals may be represented more than once if multiple conditions existed. All cardiac conditions were diagnosed at necropsy and were either the cause of death or a significant pathological finding in addition to the immediate cause of death.

Fig 5B

Estimated proportions for the first occurrence of any age-related disease in each monkey from the young-onset age group (males and females combined) statistically controlling for sex of the animal and sex-by-CR interaction. These conditions included: cancer, diabetes, arthritis, diverticulosis, and cardiovascular disease. The difference between CON and CR is not statistically significant, p=0.06. Old-onset monkeys are not represented.