Intentional Weight Loss Reduces Mortality Rate ... - Wiley Online Library

Intentional Weight Loss Reduces Mortality Rate in a Rodent Model of Dietary Obesity Joseph R. Vasselli,* Richard Weindruch,† Steven B. Heymsfield,†† F. Xavier Pi-Sunyer,* Carol N. Boozer,* Nengjun Yi,‡ Chenxi Wang,‡§¶ Angelo Pietrobelli,* ** and David B. Allison‡§

Abstract VASSELLI, JOSEPH R., RICHARD WEINDRUCH, STEVEN B. HEYMSFIELD, F. XAVIER PI-SUNYER, CAROL N. BOOZER, NENGJUN YI, CHENXI WANG, ANGELO PIETROBELLI, AND DAVID B. ALLISON. Intentional weight loss reduces mortality rate in a rodent model of dietary obesity. Obes Res. 2005;13:693–702. Objective: We used a rodent model of dietary obesity to evaluate effects of caloric restriction-induced weight loss on mortality rate. Research Measures and Procedures: In a randomized parallel-groups design, 312 outbred Sprague-Dawley rats (onehalf males) were assigned at age 10 weeks to one of three diets: low fat (LF; 18.7% calories as fat) with caloric intake adjusted to maintain body weight 10% below that for ad libitum (AL)-fed rat food, high fat (HF; 45% calories as fat) fed at the same level, or HF fed AL. At age 46 weeks, the lightest one-third of the AL group was discarded to ensure a more obese group; the remaining animals were randomly assigned to one of three diets: HF-AL, HF with energy restricted to produce body weights of animals restricted on the HF diet throughout life, or LF with energy restricted to produce the body weights of animals restricted on the LF diet throughout life. Life span, body weight, and leptin levels were measured. Results: Animals restricted throughout life lived the longest

Received for review April 15, 2004. Accepted in final form February 10, 2005. The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. *New York Obesity Research Center, Columbia University, New York, New York; †University of Wisconsin, Madison, Wisconsin; ‡Department of Biostatistics, Section on Statistical Genetics, University of Alabama at Birmingham, Birmingham, Alabama; §Clinical Nutrition Research Center, University of Alabama at Birmingham, Birmingham, Alabama; ¶Department of Epidemiology and Clinical Investigation Sciences, University of Louisville, Louisville, Kentucky; **Pediatric Unit, Verona University Medical School, Verona, Italy; and ††Merck Research Laboratories, Rahway, NJ. Address correspondence to David Allison, Department of Biostatistics, University of Alabama, Ryals Public Health Building, Room 327, 1665 University Boulevard, Birmingham, Alabama 35294-0022. E-mail: [email protected] Copyright © 2005 NAASO

(p ⬍ 0.001). Life span was not different among animals that had been obese and then lost weight and animals that had been nonobese throughout life (p ⫽ 0.18). Animals that were obese and lost weight lived substantially longer than animals that remained obese throughout life (p ⫽ 0.002). Diet composition had no effect on life span (p ⫽ 0.52). Discussion: Weight loss after the onset of obesity during adulthood leads to a substantial increase in longevity in rats. Key words: intentional weight loss, mortality rate, longevity, leptin, lifespan

Introduction Several well-established findings stand in curious juxtaposition. First, obesity in humans leads to an increased mortality rate and reduced life span (1,2). Second, caloric restriction in a number of experimental model organisms including protozoa, water fleas, Drosophila, rodents, and dogs induces marked reductions in mortality rate and increases in longevity (3,4). Third, among obese humans who are trying to lose weight, weight loss (WL)1 is associated with marked reductions in cardiovascular disease risk factors (5,6), decreased rates of incident type 2 diabetes (7), and increased quality of life (8). Given the above, one might reasonably expect that WL among obese individuals who are trying to lose weight would result in a reduced mortality rate. However, this has not yet been demonstrated to be the case. The literature on effects of WL on mortality rate among humans is complex and subject to multiple interpretations. Across the overwhelming majority of studies, WL was associated with an increased rather than a decreased mortality rate (9). However, it has been noted that much of the WL observed in these observational epidemiological studies may be unintentional, resulting from occult disease. Thus, occult disease incidence may confound the relationship

1 Nonstandard abbreviations: WL, weight loss; EL, ever lean; LF, low fat; HF, high fat; AL, ad libitum; EO, ever obese.

OBESITY RESEARCH Vol. 13 No. 4 April 2005

693

WL Reduces Mortality Rate, Vasselli et al.

between WL and mortality rate, leading to the appearance of a deleterious association. In response, a number of investigators have tried to conduct studies in which they isolate the effects of intentional from unintentional WL, usually by restricting the subject pool in various ways. These ways include using only subjects who report at the start of some period that they are trying to lose weight and eliminating any subjects who seem to have certain health characteristics at the outset of the study. Although most of these studies have shown that when analyses are so restricted, the effects of WL seem to be less deleterious and, in some studies, even moderately beneficial (9), this has not been the case in all studies. Some large, seemingly well-done studies have observed that apparently intentional WL is associated with increased mortality rate (10). However, as pointed out elsewhere (9), even among obese people who state that they are intending to lose weight, not all WL is necessarily intentional. Thus, even among this group, a large potential for confounding still exists. Two approaches are available to address this concern. In the first, one can try to statistically disentangle the effects of intentional and unintentional WL with more complex statistical models (C.S. Coffey, G.L. Gadbury, K.R. Fontaine, C. Wang, R. Weindruch, and D.B. Allison, unpublished data). In the second, one can try to use careful experimental design to minimize this potential confounding by random assignment to conditions that produce or do not produce clinically meaningful WL that is sustained throughout life. Unfortunately, doing this in a sample of humans sufficiently large to produce meaningful results is impractical given current technology; even if it could be done, it would be many decades before results were available. In such situations, an animal model, although potentially imperfect, represents a reasonable alternative. Such a model cannot be used to make unequivocal statements about effects in humans but can add substantially to the body of evidence. In fact, it is noteworthy that the NIH draft strategic plan for addressing obesity includes the following goal: “use animal models to determine whether sustained weight reduction after obesity has developed affects morbidity and mortality” (11). Given the vast literature documenting the life-prolonging effects of caloric restriction in multiple species, one might expect that a study demonstrating that WL in obese rodents prolongs life would have already been done. However, to the best of our knowledge, this is not the case. All studies of caloric restriction of which we are aware either began caloric restriction early in life or, even in those studies where caloric restriction was begun in adulthood, applied it to groups of animals that were not uniformly obese (e.g., 12) and reduced animals to caloric intakes that might be described as below normal rather than to simply nonobese levels. The purpose of this study was to offer a first experimental test of the hypothesis that sustained WL induced by 694


caloric restriction begun in adulthood in a dietary obese animal model causes reduction in mortality rate and prolongation of life span.

Research Methods and Procedures Animals and Initial Randomization Outbred Sprague-Dawley IGS rats were purchased from Charles River Breeders (Wilmington, MA) at age 7 weeks. For ease of handling, three cohorts of 104 rats each (52 males and 52 females) were received at weekly intervals, for a total of 312 rats purchased. Rats were housed individually in hanging stainless steel cages with wire mesh bottoms and were maintained at ⬃23 °C on a 12-hour light/dark cycle (lights on at 6 AM). The animals were habituated for 3 weeks during which they had ad libitum (AL) access to water and powdered laboratory food. During this period, several animals were removed from the study because of growth failure or development of malocclusions. At age 10 weeks, remaining rats were randomly assigned to one of three groups. Randomization was accomplished with computer-generated pseudorandom numbers and took place within the strata of cohort, sex, and baseline body weight. This ensured that sex and cohort were orthogonal to group assignment and that among-group variations in baseline body weight were minimized. Randomization was carried out for the three cohorts at 1-week intervals. The first group (n ⫽ 40; 20 males and 20 females) was assigned to a condition referred to as the ever-lean (EL)-low-fat (LF) group. This group was fed an LF diet (18.7% calories as fat, Table 1) and given food sufficient only for achieving a mean body weight that was ⬃90% of the mean body weight for this strain, sex, and age fed AL throughout life [reference data for Sprague-Dawley rats obtained from Charles River Breeder Hooks (13)]. To ensure that the mean body weight of the group was maintained at target level throughout the study, the body weight of the rats was measured twice weekly, and the daily caloric intake of the animals was adjusted accordingly. The second group (n ⫽ 41; 21 males and 20 females) is referred to as EL-high-fat (HF) group. This group was maintained identically to EL-LF except for the composition of the diet administered, which was a vitamin-, mineral-, and protein-fortified HF diet (45% calories as fat, Table 1). EL-LF and EL-HF were maintained as described above for the remainder of the experiment. The third group consisted of 210 rats assigned to an obesity induction phase. All of these animals had AL access to an HF diet (45% calories as fat, Table 1), similar to a 45% fat diet previously demonstrated to produce obesity in rats (14). This group was referred to as AL-HF, and a representative sample of animals from this group (n ⫽ 24) was weighed once weekly up to age 46 weeks.


Table 1. Composition of AIN-93G modified diets*

Gross energy (kcal/g) Ingredients (g/kg) Casein, high nitrogen DL-methionine L-cysteine Sucrose Cornstarch Hydrogenated soybean oil Soybean oil (stabilized with tertiary butylhydroquinone) Cellulose Mineral mix* Potassium phosphate, mono Potassium citrate, H2O Vitamin mix* Vitamin B12 and K1 supplement Choline bitartrate

Low fat

High fat

3.85

4.65

290.5 2.2 4.2 242.8 263.5

High fat fortified 4.61

290.5 2.2 4.2 167.5 186.8

357 2.7 5.4 128.5 142.2

40

116

116

40.0 50.6 35

116.0 50.6 35

116.0 50.6 43.2

3.6 5.1 10

3.6 5.1 10

4.4 6.3 12.3

10 2.5

10 2.5

12.3 3.1

* Based on Reeves et al. (15). Low-fat diet, percentage of total calories: 50.0%, carbohydrate; 18.7%, fat; 31.3%, protein; high-fat diet, percentage of total calories: 30.0%, carbohydrate; 45.0%, fat; 25.0%, protein; high-fat fortified diet, percentage of total calories: 26.0%, carbohydrate; 45.3%, fat; 28.7%, protein.

Final Randomization of the AL-HF Group At age 46 weeks, the lightest one-third of AL-HF was discarded to ensure that obesity was present in the remaining animals. Remaining animals were then randomly assigned to one of three conditions. One group (n ⫽ 49; 25 males and 24 females), referred to as the ever-obese (EO) group (EO-HF), was allowed to continue eating the HF diet AL for the remainder of its life. A second group (n ⫽ 50; 25 males and 25 females), referred to as the WL-HF group, was switched to the fortified HF diet and had its food restricted such that the achieved body weights closely matched those of the EL groups (⬃80% of their week 46 AL body weight). Reduction of body weight was achieved over 6 weeks by caloric restriction starting at 90% of week 46 AL caloric intake and increasing to 60% of week 46 AL caloric intake by week 52. Thereafter, caloric intake was individually adjusted twice weekly to maintain a 20% body weight reduction. The third group (n ⫽ 47; 23 males and 24

females), referred to as the WL-LF group, was assigned to the same LF diet as EL-LF and had its caloric intake restricted similarly as described above over 6 weeks to achieve body weights ⬃80% of their week 46 AL body weight. All animals were subsequently followed until time of death. A blood sample (0.5 mL) was obtained from the tail of a representative sample of rats (n ⫽ 6 to 8) from each group at week 84 and analyzed for leptin content using an RIA specific for rat leptin (Linco, Inc., St. Louis, MO). The complete experimental design is depicted in Figure 1. Diets The composition of the three purified diets used in the study is shown in Table 1. The diets were modified versions of the AIN 93G diet (15) and were prepared in powdered form. The use of choline bitartrate in the diets was discontinued during the maintenance phase of this study because of possible contamination of this substance during manufacture by the supplier. Choline chloride was then substituted for this additive (16). The diets were sterilized by irradiation before receipt at the animal facility. All diets were prepared by Dyets, Inc. (Easton, PA). Animal Husbandry All animals were housed within a high efficiency particulate air-filtered specific pathogen-free barrier (AR Positive-Pressure Clean Room; BioBubble, Inc., Fort Collins, CO). The St. Luke’s-Roosevelt Hospital Animal Care Facility is U.S. Department of Agriculture-licensed and Association for Assessment and Accreditation of Laboratory Animal Care-approved, and is regularly attended by a consulting veterinarian. All rats were weighed twice weekly. Powered diet was provided in spill-proof ceramic jars. For rats maintained on caloric restriction, fresh jars containing the diet ration were placed in the cages daily. The condition of all animals was monitored daily by study technicians. As the study progressed, obese rats that developed foot ulcers or skin lesions were switched from hanging cages to polypropylene living cages lined with wood chip bedding. During the study, individual rats were euthanized at the direction of the consulting veterinarian for tumors that impaired movement, bleeding abdominal or foot ulcers, respiratory infections, or achievement of a moribund state (inability to eat or drink or labored breathing). A significant number of animals developed kidney, urinary tract, or bladder stones as the study progressed because of a contaminated source of choline bitartrate in the diet (16). These animals were initially diagnosed as having urinary tract infections, but this proved not to be the case. Accordingly, animals demonstrating an impaired ability to urinate were euthanized at the request of the consulting veterinarian, and a necropsy was conducted to confirm the presence of kidney, urinary tract, or bladder stones. In all cases where euthanasia was requested, the date of euthanasia was treated as the date of death. OBESITY RESEARCH Vol. 13 No. 4 April 2005

695


Figure 1: Study protocol.

Statistical Analysis The primary statistical analysis consisted of Cox proportional hazards regression (17). A two-tailed ␣-level of 0.05 was used. All models controlled for sex, cohort, and baseline body weight, where baseline was defined as weight at age 10 weeks (before the first randomization). As a form of sensitivity analysis, additional analyses were conducted using two rank tests (Savage and Wilcoxon test) and a likelihood ratio test for testing the homogeneity of survival functions across groups (17). However, results were virtually identical regardless of which analysis was used. There696


fore, herein we report the results of the Cox proportional hazards regression only. All analyses were implemented using SAS 9.0 (PROC PHREG and PROC LIFETEST). Finally, because of concerns about a possible contaminant in the food leading to kidney, urinary tract, and/or bladder stones (16), all analyses were conducted three times. One analysis included all animals and made no use of information with respect to kidney, urinary tract, or bladder stones. This was considered the primary analysis and was the original analysis planned. A second supplementary analysis excluded any animal that had any signs of kidney, urinary

60.7 13.6 84.7 79.7 NA 629.1 363.0 652.2 395.9 755.3

690.3 397.6 679.9

6 3 10 0 7 3 7 4 1 0 39.2 45.1 31.0 17.5 44.4 42.2 69.3 67.2 208.0 187.4 708.8 441.1 723.2 453.3 664.3 338.3 634.0 374.6 1094.2 703.6 11 10 16 13 16 6 13 9 12 6 24.3 20.6 38.8 32.7 57.9 59.5 75.8 73.0 127.9 172.3 691.6 392.8 691.7 391.1 672.6 372.8 671.2 409.0 887.6 576.7 20 17 21 18 22 18 24 20 24 16 16.5 7.8 27.1 7.9 75.3 82.0 90.0 112.5 82.3 100.1 614.7 330.8 620.0 342.1 849.4 489.8 842.9 488.6 820.9 508.0 20 20 21 20 23 (1 missing) 24 25 25 25 24 37.3 20.6 22.2 10.8 23.9 16.7 39.7 18.2 25.6 17.7 316.0 212.9 314.0 211.4 325.1 214.8 320.9 222.8 318.8 219.3 EO-HF

WL-HF

WL-LF

EL-HF

Male Female Male Female Male Female Male Female Male Female EL-LF


NA, not applicable.

SD Mean (g) Sex Group

N

Mean (g)

SD

N

Mean (g)

SD

N

Mean (g)

SD

N

Week 94 Week 70 Baseline (week 46) Before first randomization

Survival Analysis Figure 4 depicts the survival data for the five groups. Data for males and females were combined because there were no significant interactions between sex and group assignment. Cox proportional hazards regression using all animals indicated that the omnibus test for the effect of group assignment was significant (for the likelihood ratio test, ␹2 ⫽ 24.7, df ⫽ 7, and p ⬍ 0.001). We subsequently looked at specific effects for the EL condition, the WL condition, and the EO condition, the main effect of diet (LF versus HF), and their interactions. Results are given in Table 4. The groups living longest were the two EL groups. Although the mortality rate was numerically higher in the two WL groups, this difference was not statistically significant (p ⫽ 0.18). Importantly, WL was associated with a significant reduction in mortality rate relative to being in the EO group (p ⫽ 0.002). Thus, the primary hypothesis of our experiment was supported. Although there was a statistically significant main effect of sex, there was no sex-bygroup interaction, suggesting that WL induced by caloric restriction was equally beneficial in male and female rats. Finally, diet had no statistically significant effect (p ⫽

Table 2. Body weight of animals before the first randomization and after week 46

Descriptive Statistics Table 2 contains descriptive statistics on the body weight of groups at multiple points throughout life. Table 3 lists the minimum value, the 10th percentile, the mean, the median, the 90th percentile, and the maximum of life span for each group. Figure 2 displays the mean body weights of all male (Figure 2A) and female (Figure 2B) groups throughout life. We achieved the body weight reduction goals after week 46 for the two WL groups. Also, although the AL-HF males and females were obese by week 46 (before final randomization), EO-HF became even more obese as the study progressed. Figure 3 depicts leptin levels of the groups measured at age 84 weeks, after achievement of the WL goals for WL-LF and WL-HF. EO-HF males and females had significantly elevated leptin levels compared with all other groups (p ⬍ 0.01). However, there were no significant differences of leptin levels between the EL and WL male groups (Figure 3A); a similar result was seen for the female groups (Figure 3B) except that WL-HF had significantly decreased leptin levels compared with the EL groups. Because leptin levels reflect the level of body fat in rodents (18,19), we can assume that not only body weight but total body fat of the respective EL and WL groups were quite comparable.

N

Results

20 20 21 20 24 24 25 24 25 24

Mean (g)

Week 118

SD

tract, or bladder stones or urinary tract infections. A third analysis included all animals but included terms for an interaction effect between experimental condition and development of kidney, urinary tract, or bladder stones.

6.9 46.7 23.9


697


Table 3. Distribution of life span in weeks for all groups Group

Sex

N

Mean (week)

EL-LF

Male (sex ⫽ 1) Female (sex ⫽ 0) All Male Female All Male Female All Male Female All Male Female All

19 20 39 22 19 41 24 24 48 24 25 49 25 24 49

101.7 91.8 96.6 108.6 95.4 102.5 100.9 82.3 91.6 98.5 90.5 94.4 92.3 79.2 85.9

EL-HF

WL-LF

WL-HF

EO-HF

Median

10th Percentile

90th Percentile

Minimum

Maximum

94.6 93.3 94.6 110.5 100.3 105.0 102.9 76.4 91.4 94.1 87.0 88.7 93.7 80.0 86.6

76.1 63.9 72.9 86.1 69.7 72.0 74.3 55.9 59.7 74.0 61.9 65.0 71.6 49.7 59.9

133.7 118.3 129.1 130.4 116.9 125.0 124.6 118.7 124.6 124.1 130.0 127.9 110.0 108.4 108.7

74.9 57.6 57.6 70.9 61.4 61.4 63.6 50.7 50.7 65.0 55.7 55.7 60.7 48.6 48.6

133.7 123.6 133.7 132.6 117.1 132.6 132.7 129.1 132.7 128.9 132.7 132.7 124.3 116.9 124.3

0.518) nor was there any diet-by-caloric intake condition interaction (p ⫽ 0.595). This indicates that WL among dietary obese rats was successful in reducing mortality rate regardless of sex and which of the two diets was consumed. Supplementary Analyses As stated above, we experienced an outbreak of what we initially believed were urinary tract infections in the animals but that subsequently proved to be kidney, urinary tract, or bladder stones. Through discussions with colleagues, we learned that other laboratories around the country were observing similar effects in certain long-term feeding studies with rats. Eventually, it became clear that all of the studies experiencing this difficulty had received batches of food at approximately the same time from one of two companies, including Dyets, Inc. Further investigation suggested that these batches of diet had been contaminated with an adulterated batch of choline bitartrate (16). Immediately after this was discovered, fresh batches of diet without choline bitartrate from the suspect chemical supplier were prepared, shipped to us, and used for the remainder of the study. Table 5 identifies the condition of the animals with respect to the occurrence of urinary tract, bladder, or kidney stones or what was initially identified as urinary tract infection. These events occurred with roughly equal frequency across all groups. We reanalyzed the data excluding any animal that had any indication of urinary tract stones or infection. Although the degree of statistical significance was generally reduced, presumably because of the smaller 698


Figure 2: Body weight of the groups after 10 weeks of age.


Figure 4: Survival curves for different groups. Survival time of the EO-HF groups was significantly decreased in comparison to all other groups.

Figure 3: Plasma leptin levels of the groups at 84 weeks of age. (*) Different from EL groups, p ⬍ 0.05. (**) Different from EL and WL groups, p ⬍ 0.01.

sample size, the pattern of results and the significance of effects were largely unchanged. Finally, we repeated the analyses with all animals but incorporated an interaction term between group assignment and whether the animal had any indication of stones or infection. This interaction was not significant (p ⫽ 0.170 to approximately 0.992). Thus, our conclusions seem to be unaffected by this contaminantinduced problem.

Discussion This is the first study, to our knowledge, to show that among organisms randomly assigned to lose or not lose

weight to achieve normal body weights after the onset of obesity, caloric restriction resulting in WL prolongs life. The effect was strong, reducing the hazard rate by 45%. WL also increased the mean, median, and the 90th percentile of life span, respectively, by 6.67%, 5.53%, and 14.6% for WL-LF and by 9.96%, 2.47%, and 17.6% for WL-HF. Whether this result can be generalized to humans is open to speculation, but, if it can be, the clinical and public health implications are profound. This study has many strengths: its randomized design, the distinction between the amount of diet consumed and resulting body weight, the distinction between the amount of diet consumed and the composition of the diet, the demonstration that results are consistent across both sexes, the use of an outbred strain made obese through dietary means, the reduction only to normal body weights rather than unusually low body weights, and the fact that animals were followed for their entire life span. One limitation of this study is the fact that only two diets were studied. We cannot say that these results would be obtained with WL induced on any diet, but it does seem reasonable to assert that the life-prolonging effect of WL among obese animals was not dependent on the level of fat in the diet. Another interesting phenomenon of this study is the abrupt weight gain of animals in the EO-HF Group at Week 80, which is largely the result of unusually low-weight animals dying and, therefore, no longer contributing to the mean. The loss of body weight before death is a well-known phenomenon, and the subsequent reversal of this jump in group mean results from the remaining animals beginning to lose weight with age and having progressively poorer health. OBESITY RESEARCH Vol. 13 No. 4 April 2005

699


Table 4. Group and diet effects on mortality Model 1

2

3*

4

Variables in the model

df

Estimate

SE

EL WL HF Sex Baseline Cohort EL WL HF Sex Baseline Cohort ELbyHF EL Sex Baseline Cohort HF Sex Baseline Cohort

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

⫺0.81 ⫺0.60 ⫺0.10 ⫺0.80 0.00 0.02 ⫺0.93 ⫺0.64 ⫺0.17 ⫺0.80 0.00 0.01 0.16 ⫺0.21 ⫺0.56 0.00 ⫺0.03 0.09 ⫺0.70 0.00 0.02

0.20 0.20 0.15 0.39 0.00 0.09 0.31 0.21 0.21 0.39 0.00 0.09 0.31 0.15 0.43 0.00 0.09 0.14 0.38 0.00 0.09

␹

2

15.91 9.28 0.43 4.26 0.49 0.03 9.20 9.17 0.71 4.33 0.48 0.02 0.28 1.78 1.70 0.00 0.08 0.42 3.47 0.24 0.05

p

HR

⬍0.01 ⬍0.01 0.51 0.04 0.48 0.86 ⬍0.01 ⬍0.01 0.40 0.04 0.49 0.89 0.60 0.18 0.19 0.95 0.78 0.52 0.06 0.62 0.82

0.45 0.55 0.90 0.45 1.00 1.02 0.39 0.53 0.84 0.45 1.00 1.01 1.18 0.82 0.57 1.00 0.97 1.10 0.50 1.00 1.02

Confidence interval 0.30 0.37 0.67 0.21 1.00 0.86 0.22 0.35 0.56 0.21 1.00 0.85 0.64 0.60 0.24 0.99 0.81 0.83 0.24 1.00 0.86

0.66 0.81 1.22 0.96 1.01 1.20 0.72 0.80 1.26 0.96 1.01 1.20 2.17 1.10 1.33 1.01 1.17 1.44 1.04 1.01 1.21

HR, hazards ratio. * Comparison of the WL and EL groups. The EO group was excluded.

In future research, it might be interesting to compare the effects of various diets on which humans sometimes lose weight with respect to their effects on longevity in obese animals. Perhaps more interestingly, future studies might build on these observations by comparing different methods of producing WL beyond caloric restriction. Specifically, one could evaluate whether WL among obese animals induced by exercise has similar effects. On the one hand, one might expect the effects to be even better given the common belief that, above and beyond body weight, being more active results in better health and greater longevity (20,21). On the other hand, studies comparing the effects of exercise with those of caloric restriction in nonobese animals have generally not found exercise to be superior. In fact, these studies may show that caloric restriction is substantially superior to exercise in terms of prolonging mean life span and, even more so, in lengthening lifespan in the upper quartiles of the life span distribution (22). Another potential direction for future research would be to evaluate whether WL induced via pharmaceutical agents produces an equivalent prolongation in life span. Surpris700


ingly, despite the fact that millions of people take prescription drugs for WL (23), to the best of our knowledge, there has never been a single published report indicating that any antiobesity prescription drug prolongs life. However, it is quite plausible that WL drugs do help to prolong life; indeed, this is part of the justification for their use (24). The merit of this speculation is supported by the fact that ephedrine, until recently a compound used frequently for WL, although not as part of a prescription drug, caused lower body weight in male and female rats and mice and, at least among female Fisher 344 rats, prolonged life (25). A recent report from investigators at the National Institutes of Aging indicated that when rats were given metformin, they weighed less and lived longer (26). Although these studies are interesting, neither evaluated animals that were already obese. It is possible, of course, that different drugs may have better or worse effects in animals after the onset of obesity. Moreover, it is not reasonable to assume that because the effects of obesity and weight gain are deleterious, WL will necessarily reverse all those effects. Some of the effects of weight gain may be irreversible. An example of this may be


Table 5. Distribution of stones and apparent urinary tract infections among groups Group EL-LF

Disease Stones Urinary tract infection

EL-HF

Stones Urinary tract infection

WL-LF


WL-HF


EO-HF


some aspects of hypertension. Evidence from a Swedish study of obese subjects indicated that surgically induced WL in humans, even when maintained for many years, produced an initial reduction in the prevalence of hypertension but that after 8 years of follow-up, blood pressure levels and hypertension rates returned to those of a control group (27). This suggests the possibility that obesity imposes some irreversible change in anatomy and physiology such that WL may be incapable of fully reversing the hypertensive effects of obesity. In summary, the effects of intentional WL on mortality rate and longevity in humans remain open to question, and it is unlikely that we will ever have a definitive study to answer this question in humans. Randomly assigning a sufficiently large number of humans to a successful WL condition for a sufficient length of time is highly impractical. Even were we able to do this, questions would remain because we can ultimately randomly assign subjects only to conditions that do or do not produce WL and not to degrees of WL itself. Thus, the primary independent variable cannot be independently manipulated, losing the true power of randomization. That same criticism can be offered of this and any other animal study as well. However, at present, these types of animal studies offer the most rigorous level of experimental control we can achieve. Judgments about the effects of intentional WL on mortality rate and longevity in humans will have to be made based on consideration of the

Status

Frequency

Percentage

No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes

19 20 38 1 23 18 40 1 37 11 46 2 31 18 47 2 29 20 48 1

48.7 51.3 97.4 2.6 56.1 43.9 97.6 2.4 77.1 22.9 95.8 4.2 63.3 36.7 95.9 4.1 59.2 40.8 98.0 2.0

entire body of evidence. We believe that this study, the first, to our knowledge, to demonstrate that caloric restriction resulting in WL initiated after the onset of obesity in a uniformly obese sample prolongs life in animals, is an important part of that body of evidence and tips the scales of evidence toward the belief that intentional WL will prolong life among obese humans.

Acknowledgments This work was supported, in part, by NIH Grants R01DK054298, P30DK056336, and P30DK026687. We thank Eleen Daly, Kesha Robinson, and Neville Cummins for their dedicated assistance in performing this study. References 1. Allison DB, Fontaine KR, Manson JE, Stevens J, VanItallie TB. Annual deaths attributable to obesity in the United States. JAMA. 1999;282:1530 – 8. 2. Fontaine KR, Redden DT, Wang C, Westfall AO, Allison DB. Years of life lost due to obesity. JAMA. 2003;289:187– 93. 3. Weindruch R, Sohal RS. Seminars in medicine of the Beth Israel Deaconess Medical Center: caloric intake and aging. N Engl J Med. 1997;337:986 –94. 4. Kealy RD, Lawler DF, Ballam JM, et al. Effects of diet restriction on life span and age-related changes in dogs. J Am Vet Med Assoc. 2002;220:1315–20. 5. Allison DB, Pi-Sunyer FX. Obesity treatment: examining the premises. Endocr Pract. 1995;1:353– 64. OBESITY RESEARCH Vol. 13 No. 4 April 2005

701


6. Pi-Sunyer FX. Short-term medical benefits and adverse effects of weight loss. Ann Intern Med. 1993;119:722– 6. 7. Knowler WC, Barrett-Connor E, Fowler SE, Lachin JM, Walker EA, Nathan DM; Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346:393– 403. 8. Fontaine KR. Weight loss and health-related quality of life. Am J Manag Care. 2001;7:926 –7. 9. Yang D, Fontaine KR, Wang C, Allison DB. Weight loss causes increased mortality: cons. Obes Rev. 2003;4:9 –16. 10. Sorensen TI. Weight loss causes increased mortality: pros. Obes Rev. 2003;4:3–7. 11. National Institutes of Health. Strategic Plan for NIH Obesity Research. http://obesityresearch.nih.gov/about/strategic-plan.htm (accessed September 11, 2004). 12. Weindruch R, Walford RL. Dietary restriction in mice beginning at 1 year of age: effect on life-span and spontaneous cancer incidence. Science. 1982;215:1415– 8. 13. Charles River Laboratories. Research Models: Products and Services 2004. http://www.criver.com/04cat/catalog_ 2004W.pdf (accessed March 10, 2005). 14. Boozer CN, Brasseur A, Atkinson RL. Dietary fat affects weight loss and adiposity during energy restriction in rats. Am J Clin Nutr. 1993;58:846 –52. 15. Reeves PG, Rossow KL, Lindlauf J. Development and testing of the AIN 93 purified diets for rodents: results on growth, kidney calcification and bone mineralization in rats and mice. J Nutr. 1993;123:1923–31. 16. Klurfeld DM. Kidney and bladder stones in rodents fed purified diets. J Nutr. 2002;132:3784.

702


17. Cox DR, Oakes D. Analysis of Survival Data. London, United Kingdom: Chapman & Hall; 1984. 18. Du D, Higginbotham DA, White BD. Food intake, energy balance and serum leptin concentrations in rats fed low-protein diets. J Nutr. 2000;130:514 –21. 19. Frederich RC, Hamann A, Anderson S, Lollmann B, Lowell BB, Flier JS. Leptin levels reflect body lipid content in mice: evidence for diet-induced resistance to leptin action. Nat Med. 1995;1:1311– 4. 20. Blair SN, Cheng Y, Holder JS. Is physical activity or physical fitness more important in defining health benefits? Med Sci Sports Exerc. 2001;33(Suppl 6):S379 –99. 21. Katzmarzyk PT, Janssen I. Ardern confidence interval: physical inactivity, excess adiposity and premature mortality. Obes Rev. 2003;4:257–90. 22. Holloszy JO. Mortality rate and longevity of food-restricted exercising male rats: a reevaluation. J Appl Physiol. 1997;82: 399 – 403. 23. Stafford RS, Radley DC. National trends in antiobesity medication use. Arch Intern Med. 2003;163:1046 –50. 24. Manson JE, Faich GA. Pharmacotherapy for obesity: do the benefits outweigh the risks? N Engl J Med. 1996;335:659 – 60. 25. National Toxicology Program. NTP toxicology and carcinogenesis studies of ephedrine sulfate (CAS no. 134-72-5) in F344/N rats and B6C3F1 mice (feed studies). Natl Toxicol Program Tech Rep Ser. 1986;307:1–186. 26. Ingram DK, Anson RM, DeCabo R, et al. Development of calorie restriction mimetics as a prolongevity strategy. Ann N Y Acad Sci. 2004;1019:412–23 . 27. Sjo¨stro¨m CD, Peltonen M, Wedel H, Sjo¨stro¨m L. Differentiated long-term effects of intentional weight loss on diabetes and hypertension. Hypertension. 2000;36:20 –5.