Why has the genetics of obesity been difficult to study?

Defining the phenotype
One of the first problems faced in attempting to define a genetic basis for obesity is deciding what kind of effect (phenotype) we seek to examine. Obesity is a heterogeneous clinical disorder. Whilst it can be conveniently defined and clinically measured in terms of elevated body mass index (BMI), this is a definition chosen to define people or populations thought to be most at risk from its complications.

It is a composite measure of body mass in relation to height, and is dependant on fat mass, lean tissue, bone and fluid mass, all of which may be subject to independent genetic or environmental influence. As BMI is not based on any specific pathophysiological process,it is an inherently unsatisfactory endpoint when looking for the effects of single genes or gene clusters.

For this reason, some studies have investigated genetic influence on more specific measures of body composition such as percentage body fat, total fat mass, visceral fat mass, subcutaneous fat mass or waist – hip ratio. These variables can be measured by bioelectrical impedance, computed tomography, magnetic resonance imaging, dual-energy X-ray absorbimetry scanning or underwater weighing.

Whilst there is often a reasonable correlation with BMI (e.g. 0.83 between BMI and fat mass in one recent study; Deng et al.,2002), the strength of such relationships varies unpredictably according to sex and age. Further- more, these variables are not easy to measure in the large populations required for genetic study and even then may result from complex individual interrelationships between factors such as energy intake, age, sex, resting, voluntary and diet-induced energy expenditure and environmental (e.g. dietary, psychological and sociopolitical) influences.

Some researchers have attempted to examine the effects of putative genes on basic 'intermediary' biological measures such as resting metabolic rate, which may be affected by 'candidate' genes such as those encoding mitochondrial uncoupling proteins (UCPs) or elements of the sympathetic nervous system such as the β 3-adrenoceptor. This approach also has drawbacks; subjects with established obesity tend to have high rather than low resting metabolic rates, for example. Using more complex constructs such as the difference between predicted and measured metabolic rates possible but moving away from the clinical phenotype may mean that people displaying abnormalities are not obese.

Warden CH and Fisler JS
Katsanis N, Beales PL, Woods MO