Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Obesity, Fat Intake and Chronic Disease

George Bray


The life insurance industry has repeatedly pointed out the increased risks of mortality associated with being overweight. Their statistics have shown that the overweight individual has a greater risk of developing diabetes mellitus, heart disease, high blood pressure, and gall bladder disease. With this impetus, a large body of research on the mechanisms underlying the development of obesity and its risks has been assembled. This chapter will provide a perspective on several of the recent developments. First, the newer techniques for measuring fatness will be described and the way they can be used to quantitate total and regional body fat. Second, current ideas about prevalence will be presented. Third, the role of diet composition in regulation of fat stores will be discussed using a feedback model. Finally, the relative impact of total fatness versus regional fatness will be discussed.


Obesity is defined as an abnormally high percentage of body fat. Overweight is defined as an increase in body weight above a standard related to height, but when sufficiently high is synonymous with obesity. Overweight and fat distribution are useful predictors of excess mortality and the risks of heart disease, hypertension, diabetes mellitus, gallbladder disease and some types of cancer among others. In order to determine whether an individual is obese or simply overweight due to increased muscle mass, one needs techniques for measuring body fat and standards against which to compare these numbers. Table 1 shows a list of both the classic and newer methods that can be used for assessing body fat and its distribution. It also provides an assessment of the relative ease, reliability, accuracy and expense of these various methods.

A. Techniques for Measuring Body Weigh and Body Fat

1. Anthropometric Methods

Measurements of height and weight, measurement of circumferences of the chest, waist, hips, or extremities and skinfolds in the triceps, biceps, subscapular, abdominal, thigh, calf, and sometimes other regions are relatively inexpensive to perform, and have been widely used in epidemiologic studies to assess total body fat, fat distribution, or the degree of overweight. Of these techniques, height and weight can be measured with the greatest accuracy (coefficient of variation <1%). Circumferences can also be measured accurately (cv ~ 2%), but skinfolds have more variability in their measurement (cv ~ 10%). Interobserver variation and variability in precise site location for skinfolds limit their use to studies where trained investigators are involved and in large samples. Because of the variability in skinfolds, they are of limited value in determining changes in body fat of individuals.

Anthropometric measurements provide several kinds of information. First, they provide height and weight from which the degree of overweight can be calculated. Second, the waist circumference alone and the ratio of the circumference of the waist divided by the circumference of the hips have provided useful epidemiologic tools for estimating the health risk associated with central fat distribution. Finally, skinfold measurements with all their attendant limitations can be used to quantitate subcutaneous fat distribution and for determining the relative amount of truncal and peripheral fat (57).

2. Quantitative Methods to Measure Body Compartments

a. Two Compartment Models

The data obtained from quantitative measurements of body composition can be used to partition the body into several compartments (96), some of which are summarized graphically in Figure 1. There are several methods that provide two compartment models where the first compartment is body fat and the second can be fat free mass, lean body mass, body potassium, or body water.

Body density has generally been considered the gold standard for measuring body fat. It assumes a density for fat equal to 0.900 and a density for fat free body mass equal to 1.100. These assumptions are satisfactory for normal individuals, but are frequently in error for children and adolescents, for athletes, for the markedly obese, and for the elderly. The simple technique of measuring body density provides data for a two compartment model of body fat and fat free mass. Whole body plethysmographs (Bod Pod) are an alternative to body density, but are more expensive. They can provide information for a two compartment model and are particularly useful in individuals who will not submerge in water.

Determination of total body water is a second method for calculating body fat. Total body water may be measured using tritiated water, deuterated water, 18O labeled water or appropriate chemicals such as antipyrine that dissolve in body water. The underlying assumption for this method is that water has a constant relation to fat free mass. This is assumed to be 73.2% body weight. Body fat is obtained by subtracting lean body mass [(Body Weight))0.732] from total body weight.

A third two compartment model is obtained by estimating total body potassium. The naturally occurring isotope of potassium, 40K, can be estimated from its gamma emission using a whole body scintillation counter. Assuming that lean body mass has 60 mEq/Kg of K in females and 66 mEq/Kg of K in males, one can calculate lean body mass, by subtracting lean mass from total body weight to obtain fat mass (31). In a study by Heymsfield and his colleagues (40), in a large group of males and females aged 52-58, the assumptions used in estimating total fat from total body water were shown to be highly reliable. Similarly, the density of fat free mass was also shown to be very close to the assumed values. However, the potassium concentrations were lower for both males and females than the values used in most equations, indicating that calculating body fat from potassium 40 could produce considerably larger errors.

Two techniques providing two compartment models of body fat use electrical conductivity. The first measures total body electrical conductivity using an expensive instrument called TOBEC (Total Body Electrical Conductivity). In contrast, determination of body electrical impedance using either a fixed or variable frequency electrical input at the wrist and ankle, can provide a relatively inexpensive method for calculating body fat providing that the degree of hydration is satisfactory. Using appropriate equations that include height, impedance, age, and sex can provide good estimates for body fat in all but the most obese individuals (79). Correlations between laboratories can provide coefficients of between r=0.95 and r=0.99. Since this method measures water, variation in hydration may bias the interpretation (30).

b. Multi-Compartment Models

The dual photon absorptiometer (DPA) and the dual energy x-ray absorptiometer (DEXA or DXA) were originally developed for determination of bone mass for studies of osteoporosis. This technique provides body mineral, body fat, and fat free mass (see Figure 1). Comparison of the estimated three compartment model obtained with these instruments to that obtained with neutron activation 40K and body water determinations show a very high correlations (39). Thus, the accuracy for this method of fat mass estimated with DPA or DEXA is high and it is probably the most appropriate "gold standard" for comparison of individuals who can be measured with this technique. Its current limitation is the mass of the individual who can be fitted on the table of these instruments. This is approximately 150 Kg.

The most expensive procedure for determination of total body mass and its components is neutron activation. At present, only the Brookhaven National Laboratory in New York is carrying out these highly expensive procedures (40). Along with whole body counting for potassium (40K) and tritiated water dilution for body water, the neutron activation instruments can provide measurements of potassium, sodium, chloride, calcium, nitrogen, and carbon as well as a number of minor elements. The limitation of the method is its radiation dose, which is of the order of 500 mRem. With knowledge about the proportion of calcium in the hydroxyapatite of bone crystal and the fact that nitrogen represents 16% of protein mass, one can calculate a four or six compartment model based on neutron activation (Figure 1).

3. Fat Distribution

Subcutaneous fat distribution can be determined by several methods. The first is the ratio of skinfolds measured on the trunk and on limbs (57). A second technique is the use of ultrasound, that measures subcutaneous fat thickness at defined trunk and limb sites. The use of magnetic resonance imaging and computed tomography to assess the proportion of fat to muscle and bone in limbs or trunk provides the most quantitative data (48, 80). The limitation of multiple CT scans is again radiation exposure, a problem that does not occur with magnetic resonance imaging.

Visceral fat is one of the most important fat depots, but is one of the most difficult depots to measure. The only reliable methods for measuring visceral fat at present are with computed tomography or magnetic resonance imaging. With either technique, a cut through the L-4, L-5 region provides a good estimate of the visceral to subcutaneous fat ratio.

4. Summary

a. DEXA appears to have replaced densitometry, total body water, and 40K as the gold standard for measuring body composition (39). For regional fat distribution, the ratio of trunk to peripheral skinfolds or the waist circumference alone or the waist circumference divided by hip circumference may be used. Visceral fat can only be adequately estimated at present by MRI or CT scans.

b. Once measured, body fat can be divided into 3 main components. The first component is the total amount of body fat expressed as a percent of body weight. The second component is the regional distribution of subcutaneous fat into central fat, also called male, android or upper body fatness versus female, gynoid or lower body fatness. This can best be done with CT scans, but subscapular fat fold, the ratio of truncal skinfolds to limb skinfolds or with sex-specific measurements of waist circumference or waist circumference divided by hip circumference (waist to hip ratio) can also be used. The third component of body fat is the amount of visceral fat located in the abdomen. This appears to be controlled differently than total fat or the regional distribution of subcutaneous fat and can only be accurately estimated with CT or MRI scans.

c. Body Weight For Epidemiologic Studies

For epidemiologic studies, two methods of relating body weight to height have been most widely used. These include relative weight and body mass index. Relative weight compares actual weight to the appropriate weight table for a given height obtained from life insurance or other tables. One problem with this approach is that many tables divide weight into "frame" sizes for which standards may be of dubious value.

The preferred method of relating height and weight is the body mass index that was developed by Quetelet more than 100 years ago. This relationship is expressed below and is usually called the body mass index (BMI), but might be more appropriately called the Quetelet Index or QI.

A nomogram for determining the body mass index is presented in Figure 2.

d. Fat Distribution For Epidemiologic Studies

Measurement of either the subscapular skinfold or the ratio of the circumference of the abdomen at the minimal point between the ribs and iliac crest and the circumference of the gluteal region at the maximal gluteal protuberance expressed as the WHR (waist to hip circumference ratio) is most widely used. A nomogram for WHR is shown in Figure 3. The ratio of trunk and peripheral skinfolds including biceps, triceps, subscapular abdominal, supra iliac, lateral calf, and lateral thigh can also be used. No satisfactory way other than CT or MRI scans exists to estimate visceral fat for epidemiologic studies.

B. Criteria for Determining Overweight and Obesity

Setting weight standards can be done in two ways. First, the normal distribution of body weight in relation to height can be measured in a large sample of the population and then arbitrarily divided into overweight and severely overweight categories. This approach has been used by the National Center for Health Statistics (47). This approach has given way to the second one where International Standards have been adopted.

The preferred way of arriving at healthy weights is to use body weights associated with the lowest overall risk to health. The minimal death rate in several prospective studies is associated with a body mass index between 22 and 25 Kg/m2. Analyses of the data from the Build study for 1979 (84), the population study from Norway (95), and the American Cancer Society study (54), showed that the body mass index associated with the lowest mortality for women increased with each decade of life. For men, on the other hand, the weight with the lowest mortality did not change with age in 2 of these 3 studies. Evaluation of the data has now led to a uniform set of guidelines for levels of body mass index which have been adopted by the World Health Organization (105) and the National Heart Lung and Blood Institute (64). These are shown in Table 2. The normal weight range is encompassed by a BMI between 18.5 and 25 Kg/m2.

Overweight is defined by the BMI range 25through 29 Kg/m2. Obesity is now defined as a BMI above 30 Kg/m2. Within the obesity categories dividing lines are drawn at 35 and 40 Kg/m2 to reflect potential therapeutic options within the group of obese patients.

Since body fat increases with age and is substantially higher for any given height/weight relationship in women than in men, assigning standards in terms of percent body fat is more difficult and on less solid ground. Guidelines for defining obesity in terms of body fat for each sex are provided in Table 3.

C. Criteria for Regional Fat Distribution

Several approaches have been used to estimate body fat distribution. The simplest is the circumference of the waist or the circumference of the waist divided by the circumference of the hips (WHR). A nomogram for determining this is presented in Figure 3. The ratio of skin folds on the trunk to the skin folds on the extremities (T/E ratio) provides a second index of fat distribution. A third approach is the ratio of skinfold to circumference measurements on the upper arm and upper thigh as originally proposed by Vague (93). The data from the Canadian National Survey for waist and hip circumference were used to plot the data in Figure 4 for both sexes.

D. Visceral Fat

Visceral fat can be estimated reliably with CT or MRI scans taken at the 4th to 5th lumbar vertebrae. Other estimates of visceral fat remain to be validated. Visceral fat increases with age and changes pari pasu with changes in body fat or fat distribution. At this writing no standards are available.

E. Stability of Body Weight in Childhood, Adolescence and Adult Life

Several epidemiological studies have examined the relationship between weight at two or more ages in the same population. The mean upward shift in body weight during adult life may well camouflage considerable individual and year to year fluctuations. In 1,302 women from Gothenburg, Sweden, the mean weight gain over a six year period was 1.4 " 5.1 Kg (SD); 28 of the women lost more than 10 kilograms and 59 gained more than 10 kilograms (67). In the normative aging study from Boston, 168 out of 1,396 men increased their weight more than 10%, whereas 75 lost more than 10%. The baseline weight for those gaining weight was higher (84.1 Kg) than for those losing weight (77.3 Kg). Williamson and colleagues (102) at the National Center for Disease Control have also reported on ten year changes in body weight. As expected, younger people gained more weight than older ones. In men the gain was less than in women before age 55. After 55, men tended to lose less than women. Between 14% and 16% became overweight between age 35 and 44. The 10 year incidence of gaining more than 10 Kg (> 5Kg/m2) peaks between age 25-34, and is 3.9% in men and 8.4% in women. Thus, women are more than twice as likely as men to gain significant amounts of weight in middle life.

A number of studies have tracked childhood weight into adult life but only a few of them have calculated the relative risk of being in the top weight category as an adult based on the weight status in childhood or adolescence. According to these reports, it is between 1.6 and 2.5 times more likely for the heaviest youngster to be overweight than the lightest youngster. In a 50 year follow-up of the Harvard Longitudinal Study of Child Health and Development, Casey et al (19) found that the BMI of females in childhood had essentially no correlation with their BMI as adults. Adolescent BMI of females in adolescence showed a better, but still low, correlation, with BMI at age 50 (r = .25 to .35). The low correlation of adolescent weight with later adult weights in women may account for the failure to show an effect of overweight in adolescent girls on mortality of adult women. In men, on the other hand, the correlation of BMI in childhood or adolescence with BMI at age 50 was better (r +.44 to +.55). Because the tracking of childhood and adolescent weight into adult life is of a low order, one must be careful in making public health recommendations to adolescents based on their adolescent weight status. A prospective follow-up over 36 years points to the variability of body weight with age (8). At age 36, 3,322 people born in 1946 were divided into weight categories according to BMI. In this cohort, 5.3% of the men and 8.4% of the women had a BMI greater than 30 Kg/m2 and 38% of the men and 24.2% of the women were overweight with a BMI between 25 and 29.9 Kg/m2. The correlation between BMI at age 26 and 36 was r=.64 for men and r=.66 for women. The authors draw the following important conclusions about weight stability. First, that 25% of the obese cohort, men as well as women, were obese both as children and adults. Second, the remaining 75% of this cohort became obese as adults, an event that could not be predicted from weights before age 20. Those who became obese between ages 11 and 36 were often not the heaviest during childhood. Only 50 to 60% of the men and women in the top decile at age 36 could be correctly predicted at age 26, using all socioeconomic, demographic, and other available weight data.

Family weight plays a role in the risk of childhood obesity becoming adult obesity. Whitaker et al (100) showed that overweight parents significantly enhance the risk of overweight children becoming overweight young adults. However, in overweight children under 3 years of age where neither parent is overweight, the risk of this child remaining overweight as an adult is very small. Thus, therapeutic advice should be significantly different in these two settings.


Prevalence data on obesity have been compiled from a number of countries around the world (105). The following discussion will focus first on factors affecting prevalence estimates within countries and then examine the cross-country prevalence data.

A. Factors Affecting Weight Within Countries

Within any country there are a number of clear-cut differences in the frequency of obesity. These differences are related to age, sex, ethnic factors and socio-economic status. The ensuing discussion will illustrate these risk factors with data from the National Center for Health Statistics (28), the World Health Organization (105) and the National Heart Lung and Blood Institute Report (65).

1. Age

The prevalence of overweight (BMI 25-30 Kg/m2) as well as obesity (BMI > 30 Kg/m2) increases with age. The peak prevalence of overweight males occurs at age 50 - 59, in females at age 70 - 74. For obesity, the peak is at age 40 - 49 for men and 50 - 59 for women.

2. Gender

In almost all populations, more women are overweight and obese than males.

3. Ethnic Factors

Among females in the United States there is a lower percentage of overweight white women than either African-American or Hispanic women (Figure 5). The nearly linear increase in prevalence with age is evident in all three groups, with peak values for overweight occurring between age 40 and 70. For males, the percentage of overweight is lower for African-Americans at all ages except the decade 40 - 49. Among males, the percentage of overweight becomes stable between the fourth and sixth decade of life.

White females are less obese at all ages than African-American or Hispanic females. The prevalence of obesity in the African-American and Hispanic groups rises to between 30% and 40% between the age of 40 to 70. This contrasts with a figure of 20% for white females. The differences in obesity among males between ethnic groups is much smaller than in women. White males have a smaller prevalence of obesity at all ages than either African-American or Hispanic males, but these differences are small with the prevalence among white males approximating 14% through most of their adult life whereas the prevalence rates among African-American and Hispanic males runs between 15 and 22%.

4. Socio-economic Status

Socio-economic status plays an important role in the prevalence of obesity. With the exception of the 20 - 29 year old women, those defined as below the poverty level by the United States government have a higher prevalence of obesity than those above the poverty level. This fact is clear for both white females and African-American females, but less so for Hispanic or Mexican/American females. These striking differences in the prevalence of obesity for white females are present but of much smaller values among the males.

5. Time Trends in Weight Status

Several studies suggest that there has been a progressive increase in weight for height in the United States throughout the entire 20th Century. Data on inductees into the military service show that for men 5' 8" tall weights rose from 147 lbs in 1863 to 168 lbs in 1962. A similar increase has been observed in the Framingham cohort, with males showing a steady increase throughout the early part of this century. By contrast, females showed a slight downward average weight in the Framingham Study. Life insurance data and data from the National Center for Health Statistics also report a small increase in average weight for height. NCHS data from surveys conducted in 1960-62, 1970-74 and 1980-81 (104) showed only a small change with time. However, recent data from 1988-94 shows a marked increase in both men and women (28) (Figure 6). Data from Denmark and The Netherlands also show a similar more recent increase. Particularly striking is the Danish data in inductees into the military service showing a sudden sharp increase in the prevalence rates for overweight and obesity in the late 1970's. The prevalence of obesity in children is also increasing (90).

B. Factors Affecting Weight Between Countries

1. Developmental Status

Figure 7 shows the time trends for weight change between countries. Two large Asian countries, China and Japan, have much lower prevalence rates than occidental countries. Among the highest prevalence rates are seen in Polynesian islanders. As the economic income of nations increase, fat consumption rises and the prevalence of obesity increases.

2. Diet

Dietary fat intake has a high correlation with the prevalence rates of obesity and overweight across countries (11).


A nutrient balance model can be used to examine the role of dietary fat in the development of obesity. In the normal weight adult, nearly 150,000 kcal of energy are available from the triacylglycerols stored in adipose tissue, 24,000 kcal in the peptide bonds and amino acids from protein, but barely 1,000 kcal in glucose and glycogen. Obesity is a failure of nutrient balance that occurs when the intake of nutrients exceeds the daily need for nutrients to stoke the metabolic furnace.

A. The Nutrient Balance Model

A feedback model for nutrient balance consists of 4 components. The first is the controlled system comprising dietary intake, digestion, absorption, storage, and metabolism of the nutrients in food. The second component is the feedback signals that tell the controller in the brain about the state of bodily systems. The third component is a controller located in the brain. And finally, there are the efferent control mechanisms that modulate nutrient intake and energy expenditure.

B. The Controlled Body System

1. Energy Intake.

The 150,000 kcal of energy contained in body fat of the normal adult human being are some six times larger than what is stored as protein. By comparison, the quantity of carbohydrate is minute. An individual eating 2000 kcal, of which 40% is carbohydrate, will take in an amount of carbohydrate each day between 50 and 100% of their total body stores of carbohydrate. In contrast, average daily protein intake is a little over 1% of total stores whereas fat intake considerably less than 1%. This is depicted in Figure 8. It should not be surprising that in studies of nutrient balance in experimental animals, changes in carbohydrate balance from day to day reciprocally affected carbohydrate intake on the subsequent day (29). That is, if carbohydrate balance is positive, i.e., the animal ate more carbohydrate than it oxidized, the animal would eat less carbohydrate the next day. Conversely, when carbohydrate balance is negative, the animal will eat more carbohydrate the next day. Although fat oxidation is related to body fat, (78) the day to day relationship between fat balance and fat intake is very weak (29). Addition of fat to a meal does not acutely change energy expenditure (29). Access to a diet with more than 30% fat consistently produces obesity in most animals. Thus, it should not be surprising that there is a positive but weak correlation between fat intake and body weight in humans (75, 106). As might be expected from this model, a low fat intake reduced body weight by a modest amount (55, 81).

2. Energy Expenditure.

Energy expenditure can be divided into several components. Resting metabolic rate is the largest component of energy expenditure in humans and is related to both fat free mass and surface area. Since obese individuals are heavier and usually have a higher fat free body mass they generally have higher levels of energy expenditure. Total energy expenditure can be estimated using the differential rate of disposal of stable isotopes of hydrogen and oxygen provided in doubly labeled water (DLW) to estimate carbon dioxide production and thus indirectly total energy expenditure. When energy intake from dietary records is compared with total energy expenditure using doubly labeled water, both normal weight and overweight individuals underreport their energy intake. Normal weight individuals underreport by an average of 20% and overweight individuals by up to 50%. Thus dietary histories are an unreliable way of estimating the energy intake of obese patients.

The thermogenic effect of food is a second component of energy expenditure. When a meal is ingested approximately 10% of the energy appears as heat and increased oxygen consumption over the next 6 hours. The thermic effect of food is highest with protein and lowest with fat. Overweight individuals have a lower thermogenic response to food which is related to insulin resistance. The sympathetic nervous system plays a role in thermogenesis in animals and human beings. In experimental animals much of this effect is mediated through activation of uncoupling protein (UCP) in brown adipose tissue. This thermogenic protein is a component of the mitochondrial membrane that, when activated by purine nucleotides, enhances the leakage of protons through the mitochondrial membrane rather than coupling to the generation of ATP. Using the knowledge of genetic structure of the UCP in BAT, at least two additional uncoupling proteins (UCP-2 and UCP-3) have been identified. UCP-2 is widely distributed in fat and muscle whereas UCP-3 appears to be located primarily in muscle. Interest in these two uncoupling proteins as potential targets for drug treatment of obesity has stimulated a great deal of research on their function and control.

Energy expended in physical activity is the final part of energy expenditure. The energy expended through physical activity may represent 30% of total daily energy expenditure and is both the most variable and under conscious control. As a primary strategy for weight loss, increasing physical activity is not very effective, but as a way of improving long term weight loss it is one of the most important things that an individual can do.

An analysis of this regulatory controlled system suggests the following concepts:

1) Each major nutrient may be regulated separately.

2) The time required to achieve balance for each nutrient varies as a function of the amount ingested each day in relation to the total body stores of that macronutrient. Thus, becoming obese by eating a high carbohydrate diet would appear to be more difficult than when eating a high fat diet because the body storage system for carbohydrate as glycogen is limited. Although excess carbohydrate can be converted to fatty acids, this is an energetically expensive transformation (25). Body fat stores, on the other hand, are many times larger than daily fat intake, implying a much greater capacity for fat storage and a much longer time constant to achieve balance.

3) Achievement of nutrient balance requires that the net oxidation of each macronutrient equals the average composition of the macronutrients in the diet (29, 78). That is, ingestion of a high fat diet requires greater oxidation of fat than when equilibrium is achieved eating a low fat diet.

Stability of energy stores (E = O)

Requires that Fat Intake = Fat Oxidation

Carbohydrate Intake = Carbohydrate Oxidation

Protein Intake = Protein Oxidation

4) There are major differences between individuals in the capacity to decrease rapidly the oxidation of carbohydrate after beginning a low carbohydrate (high fat) diet (110). Much of this difference may be genetic (7).

5) Physical training can increase oxidation of fatty acids by muscle, and thus regular aerobic exercise might reduce the tendency to become obese or help maintain lower body weight after losing weight.

6) The regulation of nutrient stores is subject to positive and negative feedback signals that operate through the central controller.

Nutrient composition of the diet plays a variable role in the development of obesity in man and animals. At one extreme are the types of obesity due simply to excess food intake regardless of composition. In these cases, obesity may develop when the diet is composed primarily of carbohydrates (vegetables, fruits, and meat or vegetable protein) or fats. In these instances, genetic factors probably play an important role because experimental animals with a recessively inherited tendency develop obesity regardless of the composition of the diet. At the other extreme are those types of obesity where dietary composition is central to the development of obesity. These include a high fat diet, access to beverages or solutions containing sucrose, or other soluble carbohydrates and diets with an abundance of highly palatable foods. Any of these types of dietary obesity can be controlled by changing diets or by restraint in the intake of food. In clinical studies, a low rate of fat oxidation (i.e., higher carbohydrate oxidation) in the basal state predicts an increase in body weight (110). The rate of fat oxidation is directly related to the degree of body fat (78). As weight is lost, fat oxidation declines. To keep from regaining weight that has been lost, the intake of fat must be reduced by approximately 20 g/d for each 10 kg of fat lost (78).

C. Afferent Feedback Signals

The brain receives information about the status of nutrient balance from several sources (9). Afferent signals can be transmitted over the somatic sensory nervous system via the autonomic nervous system or through blood borne signals.

1. Sensory Signals

Sight, smell and taste of food provide important signals for identifying potential environmental sources of food and for initiating food intake. Along with the texture and taste of food in the mouth, the olfactory and sensory cues about the quality of the food can serve as positive feedback signals for initiating and sustaining food ingestion. Negative feedback signals that eventually slow down, terminate, or abort an eating incident also arise from olfactory or gustatory senses. The ability of most animals to avoid foods that have previously made them sick, a phenomenon known as bait-shyness, is an example of these afferent sensory signals integrated with a central learning system. In addition to the classic tastes for salt, sweet, sour, bitter and savory, recent data suggests that the tongue has receptors that respond to individual fatty acids. The polyunsaturated fatty acids including linoleic and linolenic acid are potent inhibitors for a potassium channel in the taste bud (35).

2. Gastrointestinal Signals

Information about the presence of food or nutrients in the gastrointestinal tract can be initiated by several mechanisms. The first is gastrointestinal distension. The second is by the action of nutrients on the GI tract mucosa. For example, oleate, a long chain fatty acid, decreases food intake when infused into the duodenum. The third is release of gastrointestinal hormones when nutrients act directly on the gastrointestinal tract. The fourth is through the effects of absorbed nutrients. Apo-AIV, an apo-protein component of the chylomicrons that are synthesized in the GI tract, decreases food intake (33).

Several gastrointestinal hormones, as well as other peptides, have been implicated in the inhibition of feeding. The most prominent GI peptide is cholecystokinin (6). Intraperitoneal injections of CCK decrease food intake in hungry rats, sheep and humans, as well as inhabit sham feeding of rats and monkeys. The sequence of events associated with a response to CCK is similar to that of spontaneous postprandial satiety. CCK may terminate eating by acting on antral CCK-A receptors in the pylorus that constrict the pylorus and enhance gastric distension. This peripheral information generated in this way by CCK may be important in producing satiety because vagotomy and lesions to the central vagal connections of the vagus in the nucleus of the tractus solitarius will block the satiety effect of CCK.

Glucagon (34) and glucagon-like peptide-1 (GLP-1) are released from the gut and pancreas. Glucagon reduces food intake in animals and man (34). GLP-1 given intravenously to human subjects also reduces food intake (30).

Enterostatin is a third peptide of interest (26). It is a pancreatic co-enzyme produced in the intestine by cleavage of procolipase. Enterostatin is a pentapeptide that is conserved across most species. When enterostatin is injected either peripherally or centrally into experimental animals fat intake is specifically reduced. The concept of a nutrient-specific modulation of intake is suggested by these findings and merits additional research in a society where the intake of high fat foods is so prevalent.

3. Leptin

Leptin was discovered in 1994 by Friedman and his colleagues (109), and is produced primarily in adipose tissue and in placenta. Leptin is a member of the cytokine family and acts through the leptin receptor to modulate food intake, the sympathetic nervous system, and the hematopoietic system. Human beings who lack either leptin (61) or the leptin receptor (20) are obese, as is the obese (ob/ob) mouse which is leptin deficient (9, 52). When leptin is administered to either leptin deficient humans or mice, food intake is reduced and the other defects of this deficiency are corrected. In mice that overexpress the leptin gene, essentially all body fat can be eliminated indicating the potential value of this peptide as a treatment of obesity.

4. Nutrient Signals

Nutrient signals may also act on the liver or brain to induce satiety. 2-deoxy-D-glucose, a drug that inhibits cellular metabolism of glucose, increases food intake. Similarly, a drop in glucose precedes many meals in man and animals (18, 51). Manipulation of fatty acid oxidation also affects food intake. Blockade of fatty acid oxidation with either mercaptoacetate or methylpalmoxirate (etomoxir) increases food intake (17, 32). Fatty acid metabolites, such as 3-hydroxybutyrate or acetoacetate, also reduce food intake (77). These observations point to a potential for nutrients as afferent feedback signals.

D. The Controller

1. Anatomy

Several anatomic regions of the mid- and hind-brain play an important role in the control of nutrient balance. Destruction of the ventromedial or paraventricular hypothalamus is associated with hyperphagia and obesity in most homeothermic species that have been studied (9). Lesions in

the central nucleus of the amygdala also produce obesity. On the other hand, destruction of the lateral hypothalamus is associated with a decrease in food intake and a reduction in body fat.

2. Neurotransmitters

The neurotransmitters involved in regulation of nutrient intake can be divided into 3 groups:

1) The fast acting amino acids, which modulate ion channels;

2) The monoamines, which act more slowly through second messengers; and

3) The peptides, which may modulate monoamines and affect intake of specific nutrients (62).

Gamma aminobutyric acid (GABA) is one of the fast acting neurotransmitters that can increase or decrease food intake depending on where it is injected (70). Glutamate injected into the medial hypothalamus increases food intake (86).

In addition to GABA, there are a number of slow-acting neurotransmitters that are involved in modulating feeding, including norepinephrine, serotonin, dopamine, and histamine. Serotonin (5-HT) is derived from the dietary amino acid tryptophan. Increasing extra-neuronal concentrations of serotonin by any one of several methods decreases food intake (5). Norepinephrine can either decrease food intake by activating beta receptors in the periformical area or increase food intake by acting on alpha-2 adrenergic receptors in the paraventricular or ventromedial nucleus (42, 53). Dopamine D-1 receptors may be involved in decreasing food intake and in the hedonic effects of feeding. Activation of H-1 receptors in the ventromedial hypothalamus also reduces food intake.

A number of peptides modulate global food intake or the intake of a single nutrient (10, 62) (Table 4). Agouti-related peptide, beta-endorphin, dynorphin, galanin, growth-hormone-releasing hormone, melanin concentrating hormone, neuropeptide Y, and orexin A all stimulate food intake when injected into the third ventricle or the medial hypothalamus. A variety of other peptides, including a-melanocyte stimulating hormone, bombesin, CART (cocaine amphetamine regulated transcript), calcitonin (calcitonin gene-related peptide), cholecystokinin, corticotropin-releasing hormone, cyclo-his-pro, enterostatin, neurotensin, oxytocin, urocortin, and vasopressin inhibit feeding when injected into the medial hypothalamus or into the third ventricle (10).

One role for some of the neuropeptides in modulation of food intake is through their effects on specific types of eating (10). Neuropeptide Y injected into the paraventricular nucleus preferentially increases carbohydrate intake. Galanin injected into the same area increases fat intake in fat preferring animals and carbohydrate in carbohydrate preferring animals. Enterostatin, an activation pentapeptide from pancreatic procolipase, specifically reduces fat intake, whether the peptide is injected peripherally or into the central nervous system (69). These examples suggest that the way in which some peptides may act is to modulate specific components of the homeostatic system dealing with individual nutrients and their "appetites". This hypothesis can be called the nutrient-specific effect of peptides. The peptides that have been identified as affecting specific measurements are listed in Table 5.

Sensory specific satiety is a term that describes the fact that when given a choice of preferred foods including a food that had just been eaten, subjects will chose a new food that they have not eaten (74). The nutrient specific effect of peptides could provide the molecular basis for this phenomenon of sensory-specific satiety.

E. Efferent Controls

The efferent controls include the motor activities involved in identifying, obtaining and ingesting food, as well as the efferent effects produced by the autonomic nervous system and several circulating hormones. The complex sequence of motor activities that leads to the initiation of food seeking, the identification of food and the killing and ingestion of this food is integrated in the lateral hypothalamus, because electrical stimulation in this area will lead to food seeking and ingestive behavior.

1. Autonomic Nervous System

Both the sympathetic and parasympathetic nervous systems are involved in nutrient balance. In animal species that develop obesity following hypothalamic lesions, there is evidence for increased activity of the efferent parasympathetic nervous system (vagus nerve)(9). This may provide part of the explanation of the increase in insulin secretion that characterizes hypothalamic obesity.

Reduction in the thermogenic part of the sympathetic nervous system is also characteristic of obesity and may participate in the enhanced insulin secretion (9). In experimental animals there is an inverse relationship between the activity of the sympathetic nervous system and food intake (9). In spontaneously feeding rats there is a negative correlation between basal activity of the sympathetic nervous system that supplies brown adipose tissue and spontaneous food intake throughout the 24h (9). In addition almost all of the experimental maneuvers that increase food intake, such as lesions in the ventromedial hypothalamus or genetic obesity, decrease the activity of the sympathetic nervous system. Conversely, those maneuvers that decrease food intake, such as lateral hypothalamic lesions or injections of fenfluramine, an appetite suppressant drug, increase sympathetic activity (9).

Food intake is often initiated by a transient drop in circulating glucose levels (18, 51). In anticipation of food intake, efferent vagal activity increases, producing an early phase of insulin release from the pancreas. As food enters the stomach and intestine, its digestion triggers vagal afferents signaling a further rise in insulin secretion and an increase in peripheral efferent sympathetic activity that activates beta-3 adrenergic receptors and their thermogenic responses that may participate in mediating satiety. A rise in CCK, that slows gastric emptying and the release of other intestinal hormones, may also participate in the satiety sequence. In both animals and man, ingestion of a meal enhances sympathetic efferent activity that may serve as one of the inhibitory factors in feeding and act as part of the satiety system.

2. Efferent Hormonal Mechanisms

a. Insulin

Increased levels of insulin are characteristic of obesity. Peripheral injections of insulin to treat diabetes can increase food intake and body weight, probably by lowering glucose concentrations and triggering hunger (24). The increased food intake that follows insulin injections also produces mild degrees of obesity. Chronic infusions of insulin into the central nervous system, on the other hand, reduce food intake and body weight (94), and insulin has thus been proposed as a signal to the brain about the quantity of peripheral fat stores (71). One problem with this hypothesis is that most types of obesity are characterized by hyperphagia in the face of high concentrations of insulin. There are at least two other possible interpretations of the hyperinsulinemia of obesity. First, the rise in insulin may be a reflection of high levels of nutrient intake. Since insulin is essential for nutrient storage, increased flux of nutrients would be expected to increase insulin. Second, hyperinsulinemia may reflect actual or apparent hypothalamic resistance to the action of insulin. In this case, increased insulin secretion would be modulated by changes in the function of the autonomic nervous system resulting from resistance to the action of insulin in the central nervous system.

b. Adrenal Steroids

The development or progression of experimental obesity is either reversed or attenuated by adrenalectomy (9). In clinical medicine, Addison's disease with adrenal insufficiency is associated with loss of body fat, whereas Cushing's syndrome, with high levels of adrenal steroid secretion, is associated with obesity. The fact that almost all defects in genetically obese animals are reversed by adrenalectomy and that clinical changes in adrenal status can produce leanness or obesity suggests that glucocorticoids play a key role in the development and maintenance of the obese state. In addition, steroids modulate intake of specific nutrients. In adrenalectomized animals, injections of aldosterone increases fat intake through occupancy of Type I receptors in the brain. At normal levels of corticosterone, Type I receptors in brain are completely occupied and glucocorticoid effects are mediated through Type II receptors. Glucocorticoids such as corticosterone are essential for stimulation of carbohydrate intake following the injection of norepinephrine (88).


It is a cliche' to say that "overweight is risking fate". However, the data presented below will argue that not only is an excess quantity of fat risky, but increased abdominal fat distribution may be an even more important external guide to health risks (49, 50).

Epidemiologic data on the relationship between body mass index and a given risk, such as overall mortality, heart disease, diabetes mellitus, or gall bladder disease, are curvilinear and often described as J- or U- shaped. This effect on overall mortality is shown in Figure 9. Mortality or morbidity increases as body mass index increases. There may also be an increase in excess mortality at weights below a lower limit of body mass index of 18-19 Kg/m2, but this may also be related largely to smoking.

A. Effects of Obesity and Fat Distribution on Overall Mortality

1. Retrospective Studies

Both retrospective and prospective studies have contributed to our understanding of the relationship between overweight, fat distribution, and mortality. The primary retrospective studies looking at the relationship of weight and obesity have come from the life insurance industry. Life insurance statistics published most recently in 1979 (84), as well as those assembled at intervals throughout this century, have shown that excess weight is associated with higher mortality rates. The minimal mortality for both men and women occurs among individuals 10% below average weight. Deviations in body weight above or below this figure are associated with an increase in mortality. Based on the 1979 data, a body weight that is 10% above average weight is accompanied by an 11% increase in excess mortality for men and a 7% increase for women. If body weight is 20% above average weight, the excess mortality rises to 20% for men and 10% for women.

A second major retrospective study has examined the relationship between body weight and mortality in Norwegian men and women (95). The same curvilinear relationship is observed in this study with the minimum body mass index observed between 23 and 27 Kg/m2. Individuals with lower body weights showed an increase, giving the J-shaped relationship described above.

Overweight generally increases the risk of death, especially sudden death, although in many studies it may not be an independent variable. Three major problems plague the interpretation of studies in this area (59). First, many studies fail to separate smokers from non-smokers. Since smokers tend to have lower body weights and higher mortality, this influences the death rates and compounds the difficulty of assigning effects to body weight per se. Second, early mortality may bias the interpretation of weight status on life expectancy. Individuals who are losing weight at the time of a survey may die early and thus overemphasize the effect of low body weights as a cause of higher mortality. The failure to identify obesity as an independent risk factor, therefore, has led many people to suggest that it is unimportant. This denies the important relationship that obesity has to diabetes, hypertension and hyperlipidemia and through whose effects the increase in body weight is likely to cause ill health. Since obesity must modify some intermediate mechanism, such as cardiac function or the metabolism of lipids or glucose to produce death or disease, overweight may serve as a useful identifier of risk factors.

2. Prospective Studies of Obesity and Mortality

A large number of prospective studies have now been published relating obesity and mortality. Interpretations of these data have varied because some studies found no relationship between weight and excess mortality, whereas others did. In a careful review of this data, Sjostrom plotted the relationship between the numbers of subjects studied, the duration of follow-up and the mortality experience. As shown in Figure 10, this clearly demonstrates that smaller studies of long duration or shorter studies with large numbers of subjects lead to similar conclusions, i.e., that there is a relationship between initial body weight and subsequent excess risk of mortality.

3. Effects of Change in Body Weight on Risk Factors and Mortality

Weight gain in adults and children, has been associated with an increase blood pressure and blood lipids as well as glucose and uric acid and risk of heart disease. From the data obtained in Framingham, MA it was concluded that a 10% reduction in relative weight for men was associated with a fall in serum glucose of 2.5 mg/dl, a fall in serum cholesterol of 11.3 mg/dl, a fall in systolic blood pressure of 6.6 mm Hg, and a fall in serum uric acid of 0.33 mm/dl (4). For each 10% reduction in body weight of men, these data predicted there would be an anticipated 20% decrease in the incidence of coronary artery disease. If everyone were at optimal weight, there would be 25% less coronary heart disease, and 35% reduction of congestive heart failure and stroke, based on the data collected in Framingham (43). A 10 Kg increase in body weight in the Nurses Health Trial was associated with an increased frequency of cardiovascular events. Life insurance data also suggest that changes in body weight were associated with corresponding changes in relative risk of cardiovascular disease.

Losing and regaining weight, so-called weight cycling (82, 103), may also be hazardous. Data from the Chicago Gas and Electric Company Study (38) showed that those who gained and lost weight had a significantly higher risk of death from cardiovascular disease than the group of individuals with no change in weight. More recently, Lissner et al, using data from the Framingham study (56), showed that significant changes in weight, whether in the obese or nonobese, were associated with higher likelihood of mortality. A review of this controversial area by the Obesity Task Force (65) concluded that the risks of weight loss and regain had been overemphasized and were relatively unimportant relative to overweight. The weight loss resulting from surgical treatment of obesity lead to significant improvement in most risk factors (82). Similarly, voluntary weight loss may improve overall risks of mortality (103).

B. Effects of Regional Fat Distribution

Many studies have shown that central adiposity is positively correlated with increased mortality, as well as the risk for developing cardiovascular disease, diabetes mellitus, and stroke. Most of these studies have provided information about men. Among 14,462 women between age 38 and 60 years of age, the 12 year age specific incidence rates for myocardial infarction, stroke, and overall death rate were related to central adiposity. Among those in the highest quintile for central body fat, the relative risk of myocardial infarction was 8.2 times higher than that for the lowest quintile. For stroke and overall death rate, the relative risk was increased 3.8 and 2.8 times higher for those in the highest quintile compared to those in the lowest quintile. When women in the top five percent for central adiposity, measured as the ratio of the circumference of the waist divided by the circumference of the hips (WHR), were compared to women in the lowest quintile, the risk for myocardial infarction was increased 14.8 times, the risk of having a stroke was increased 11.0 times and the risk of death from all causes was increased 4.8 times. Data collected in Sweden allow a comparison of men and women in the same town. Larsson et al. (50) have suggested that differences in fat distribution between men and women may account for most of the sex-differences in rates of myocardial infarction.

C. Morbidity Related to Individual Organ Systems

1. Cardiovascular Morbidity

Increased weight and central adiposity both produce a number of important changes in cardiovascular function. Heart mass increases, both on postmortem examination (2) and as assessed by echocardiographic measurements of posterior wall and interventricular septal thickness (107). The increased cardiac mass is associated with increased blood volume and an increase in both intra and extracellular fluid volumes. Both cardiac output and cardiac stroke volume are elevated and positively correlated with body weight and with the degree of excess weight. Left and right ventricular end diastolic pressures are also higher, as are the pulmonary artery and pulmonary capillary wedge pressures. Studies using cardiac catheterization and echocardiography with pulsed Doppler techniques have revealed the presence of impaired left ventricular function in some obese patients (107), and a cardiomyopathy of obesity has been clearly identified (2). Abnormalities in both atrial and ventricular filling have been identified in 50% of patients with a BMI >40 kg/m2. Heart rate, however, does not increase in obesity. Thus, the increased cardiac output occurs by increased stroke volume from an enlarged heart. Electrocardiographic alterations show a leftward shift in the mean QRS complex with increased fatness for both men and women. The PR interval, QRS duration and QTc interval in voltage increase with increasing adiposity. A prolonged QTc interval was present in 28.3% of those tested.

A number of lipoprotein abnormalities are associated with obesity (36). First, high density lipoprotein cholesterol (HDL2) decreases in obese males and females. Second, serum total cholesterol is usually normal or only slightly elevated, although the transport of low density lipoprotein (LDL) cholesterol through the plasma compartment increases. This increased transport is consistent with the correlation between cholesterol production and obesity (66). As body fat accumulates, approximately 20mg of additional cholesterol is synthesized for each extra Kg of body fat. Third, the production of very low density lipoprotein triglyceride (VLDL) and the corresponding apoprotein B100 by liver, tend to increase in relation to the degree of obesity in Caucasians and Pima Indians. The increased hepatic VLDL production in obesity is probably a reflection of the associated hyperinsulinemia. Fourth, the high rate of apoprotein-B synthesis is probably related to the high rate of synthesis apoprotein-B for incorporation into VLDL. Fifth, lipoprotein lipase, the enzyme that hydrolyzes triacylglycerols in VLDL and chylomicrons, increases in adipose tissue. In contrast to most abnormalities in obesity, LPL frequently rises with significant weight loss (44), whereas most other abnormalities return towards normal. Finally, free fatty acid concentrations frequently increase in obesity, reflecting their higher rate of turnover.

2. Hypertension

Increased blood pressure, like increased levels of insulin, is characteristic of obesity. Indeed, these two events may be related through the mechanism of insulin resistance (23, 72). The use of indirect sphygmomanometric methods for determination of blood pressure requires use of an appropriately sized blood pressure cuff. When the blood pressure cuff is too short, greater differences are observed between systolic and diastolic pressures measured by direct intra-arterial methods than those obtained by indirect methods.

Hypertension has a striking correlation, not only with body weight but also with lateral body build, that is proportional to changes in both systolic and diastolic blood pressure. The cardiac response to hypertension can include both concentric hypertrophy and dilation (60). Both central body fat distribution, as well as an increase in total body fat, appear to be related to the appearance of hypertension. During periods of severe caloric deprivation, such as occurred in Europe during World War I and World War II, hypertension was almost nonexistent. In clinical studies correlating changes in blood pressure with weight reduction, approximately 50 to 70% of those who lose weight have a fall in blood pressure (58). One explanation might be reduced intake of salt, but careful studies have shown that blood pressure falls even if a fall in sodium intake is prevented by giving salt supplements (91). Weight reduction is more effective in lowering systolic blood pressure than diastolic blood pressure.

3. Pulmonary Function

A number of abnormalities in pulmonary function have been observed in obese subjects. At one extreme are patients with the Pickwickian syndrome, named after Joe the fat boy in Dickens' Pickwick Papers. This syndrome (Obesity-Hypoventilation Syndrome = OHS) is characterized by somnolence, obesity and alveolar hypoventilation. It is usually associated with obstructive sleep apnea, which may in turn reflect increased pharyngeal deposits of fat. Weight loss will markedly reduce the detrimental effects of this syndrome, as will oxygenation of the patients by using nocturnal continuous positive airway pressure (CPAP).

At the other extreme are the impairments in work capacity and pulmonary function due to obesity per se. There is a fairly uniform decrease in expiratory reserve function with obesity. Extensive alterations in pulmonary function are observed primarily in obese individuals with a BMI > 40Kg/m2 or in obese individuals with some other underlying respiratory or cardiovascular problem. Thus, vital capacity, inspiratory capacity, residual volume, and diffusing capacity remain fairly constant over a wide range of body weights, except in subjects who are massively overweight, ie., those with a weight to height ratio greater than 1cm/Kg.

4. Diabetes Mellitus and Obesity

The U.S. Diabetes Commission reported that the chance of becoming diabetic more than doubles for every 20% of excess body weight. A curvilinear relationship between diabetes and obesity clearly exists. This has been demonstrated in the Nurses Health Trial (21), in the Pima Indians (46), the Physicians Health Study (21), and in members of a weight loss club (73). The curvilinear relationship of diabetes with mortality is also seen in the data from the American Cancer Society Study (54). The excess mortality for diabetic individuals with a body mass index of 35 Kg/m2 increased by nearly 8 fold, compared to those of normal weight. All of the data cited above suggest a threshold effect for overweight and the development of Type II diabetes. When the body mass index nears 20 Kg/m2, there is essentially no risk for developing Type II diabetes.

Central fat deposition also increases the risk of diabetes. This was first suggested by Vague (93) and has been demonstrated repeatedly since that time (45, 68). There was a greater risk for developing diabetes as waist hip ratio (WHR) increased (ie., abdominal fat increased). For those in the lowest tertile for central fat distribution, increasing total fat had no significant effect. Haffner and his colleagues (37) in the San Antonio Heart Study have demonstrated that the presence of central adiposity in Mexican Americans was associated with high rates of Type II diabetes. In the women in this population, the BMI, a high waist/hip ratio, and the ratio of subscapular to triceps skinfold measurements all made independent contributions to the risk of developing NIDDM.

5. Gallbladder Disease and Obesity

The association of obesity with gallbladder disease has been documented in several studies. In one study, 88% of the variation in frequency of gallbladder disease could be accounted for by weight, age, and parity, with weight being the most important variable. Within each age group, however, the frequency of gallbladder disease increases at higher body weights. Women with a BMI >30 Kg/m2 had a yearly incidence of gallstones of 1%; those with a BMI >45 Kg/m2 had an annual rate of approximately 2% (85).

One mechanism for the increased risk of gallbladder disease is increased cholesterol production and secretion. As noted above, each Kg of excess body fat increases cholesterol production by approximately 20mg/d (66). With weight loss, bile becomes more highly saturated with cholesterol and, if nidation occurs, the risk of gallstone formation may increase sharply. Several recent studies have examined the formation of gall stones during the period of rapid weight loss. The incidence rates in these studies can be 15 to 25 fold higher than in the population of general obese subjects (98). The gall stones that form appear to produce symptoms in approximately 1/3 of the subjects, and a significant fraction may require surgery.

6. Cancer and Obesity

The American Cancer Society cohort study reported positive associations between excess weight and cancers of the gallbladder, biliary duct, endometrium, ovary, breast and cervix among women, and for cancers of the colon and prostate among men (54). The finding of an increase in the risk for endometrial cancer with increasing weight has been a consistent finding in the majority of case control studies. The relation of obesity to breast cancer has been primarily observed in post-menopausal women in studies in the Netherlands, Northern Italy, and Israel. Pre-menopausal women less frequently show an association of breast cancer and obesity. Indeed, Willett and his colleagues (101), using data from the Nurses Health Trial, have found an inverse relationship between BMI and age adjusted relative risk for breast cancer in pre-menopausal women. This discrepancy with other data might be explained by the recent finding that breast cancer risk is higher among women with upper body obesity (76).

7. Obesity and Joint Disease

Increasing body weight might be expected to add additional trauma to the weight bearing joints and, thus, accelerate the development of osteoarthritis. The National Health and Nutrition Examination survey (NHANES I) examined the prevalence of osteoarthritis in the hands and ankles in relation to weight, race and physical demands. Within each age group, there was a clear increase in the prevalence of osteoarthritis in relation to body weight for all groups. The slope of increase with weight was sharpest below 90 Kg, suggesting that body weight is only one factor. Weight loss was associated with a significant reduction in risk for osteoarthritis of the knee (27). In contrast with osteoarthritis, the risk of osteoporosis is reduced in the obese, possibly because of increased bone mass accrued during the early years of bone formation. Obesity is also associated with an increased risk of gout. In individuals whose weights were 15% above desirable, the frequency of gout was 3.0 times that of individuals who were less than 10% above desirable weight. There is also a significant correlation between uric acid levels and body weight (43).

8. Obstetrics and the Overweight Patient

Body weight before pregnancy and weight gain during pregnancy, influence the course of labor and its outcome. Infants born to heavy women weigh more than those born to light women. There is also a direct relationship between placental weight and pre-pregnancy body weight. When these infants were compared with weight changes at age 7, 50% of the incremental weight gain could be accounted for by the differences in placental weight at birth. The remaining 50% of the difference was accounted for by the post-natal environment. Naeye (63) found the fewest fetal and post-natal deaths occurred with mothers who were overweight at the beginning of pregnancy and who gained an average of 7.3 Kg or less. The optimal weight gain during pregnancy was 9.1 Kg for normal weight women and 13.6 Kg for those who were underweight.

9. Endocrine Function

Obesity produces a number of changes in endocrine function, but in almost all instances these appear to be secondary to the obesity rather than etiologic (108).


A. Guidelines

As the approaches to treatment for overweight and obesity has become more complex, the need for guidance in selecting the most appropriate methods has increased. This need has prompted several groups to develop and publish guidelines (1, 64, 89, 97, 105). Each of these documents has something to recommend it, but one feature is common to all but one. They have, with the exception of Weighing the Options accepted the weight standards presented in Table 2 where overweight is defined as a BMI between 25 and <30 Kg/m2 and obesity a BMI > 30 Kg/m2 with grades of obesity divided into 5 Kg/m2 BMI units.

B. Behavior Therapy, Diet and Exercise.

Behavior therapy was introduced into the treatment of obesity in 1967 (87) and rapidly became one of the cornerstones of all weight loss programs. When coupled with dietary advice, nutritional counseling and exercise, weight losses can average between 9 and 10 kg. Individuals who continue to monitor their eating behavior, who reduce their intake of dietary fat and who adopt strategies for increasing physical activity have a better likelihood of long term weight loss than individuals who do not (16).

C. Drug Treatment of Obesity.

Maintaining weight loss following a successful weight loss program has proven difficult because current treatments do not cure obesity. With behavior therapy programs approximately 1/3 of the weight is regained in the first year and the remainder within 3 years (97). For individuals with comorbidities associated with obesity, this is an unsatisfactory situation, and has led to the search for medications that can be used safely for an extended period of time when needed. All of the drugs available for the use in obesity prior to 1996 were approved for use for only a "few weeks" (13). In 1996 dexfenfluramine was approved by the FDA for long term use. Before its use could become very widespread, however, valvular abnormalities were reported in patients treated with the combination of fenfluramine and phentermine (22). The combination of fenfluramine and phentermine had been pioneered by Weintraub et al (99) who reasoned that the use of a drug acting on the noradrenergic system (phentermine) along with one acting on the serotonergic system (fenfluramine) might produce more weight loss than either one alone. With this combination of medications, weight losses were indeed better than in any of the programs reporting monotherapy alone. One lesson from the subsequent appearance of the valvulopathy associated with this combination of medications is that new combinations should be tested before being used, but also that combinations may indeed better than single drugs alone.

Two new drugs have either been approved or are likely to be approved by the FDA soon. These are sibutramine and orlistat.

1. Sibutramine

Sibutramine is a b-phenethylamine which blocks the uptake of norepinephrine, serotonin and dopamine. In experimental studies it reduces food intake, in animals and is also thermogenic (13). Clinical data also suggest that sibutramine reduces food intake, but the information on its thermogenic properties is conflicting. Three pivotal clinical studies are included in the package insert for sibutramine. In a multi-center double-blind dose-ranging study there was a dose-response in weight loss (15, 16). At all doses there was a significantly greater percentage of drug treated than placebo treated patients losing 5% or 10% of their initial body weight. In a year long trial comparing 10 and 15 mg/d of sibutramine against placebo both drugs were better than placebo, but there was no difference between them. In the third trial sibutramine was added after the initial period of weight loss and produced additional weight loss compared with weight regain in the placebo-treated patients. Sibutramine produces a small mean increase in blood pressure and heart rate which is not clinically important for most patients. However, the drug should be monitored carefully after its initial dosing to prevent treating patients who may have significant increases in blood pressure or heart rate. Current recommendations are to start the drug at 10 mg/d and increase to a maximum of 20 mg/d depending on response.

2. Orlistat

Orlistat has been approved in many countries and has received an "approvable" letter from the FDA. This drug acts in the intestine to partially inhibit pancreatic lipase, thus increasing the fraction of undigested triglyceride reaching the colon. Since the drug modulates fat digestion and has minuscule absorption, it is important to advise patients to adhere to a 30% fat diet. With low fat diets the drug has little effect and, with higher intakes of fat, it may produce steatorrhea.

Orlistat produces a dose-related reduction in body weight. Clinical trials using the drug in overweight individuals and in diabetic patients have been published (41, 83). In these trials the dose was 120 mg three times a day before meals. In two-year clinical trials in overweight patients the drug produced a nearly 10% weight loss in the treated patients compared to 5% to 6% in the placebo-treated patients at one year. During the second year, patients on the drug were more successful in preventing the weight regain. Effective dietary counseling is obviously important in the use of this drug to minimize the gastrointestinal side-effects.


This chapter has focused on some of the newer ideas about the definition and development of obesity, as well as the health risks associated with obesity and body fat distribution. Of these two, central or visceral fat has the highest relative risk for overall mortality, heart disease, stroke, hypertension, diabetes mellitus, and possibly cancer. Effective long-term treatment for the problem of obesity and excessive visceral fat, in particular, holds the promise of reducing national health budgets for a variety of chronic diseases.

published 2000