The ultimate goal of obesity genetics is to identify a gene defect found exclusively in obese patients producing a functional variant (for example with altered or absent protein function,  the so-called ‘smoking gun’).  The approaches that may be used in the attempt to identify such mutations depend to a large extent on what is known a priori about the function of the normal protein product.

The candidate gene approach
If there is knowledge that an abnormal   protein product   is capable of causing obesity (or counteracting   it),  evidence for the presence   of mutations in the responsible ‘candidate gene’ may be sought in the population at large and related to measures of adiposity.  In general,  a gene may be considered a candidate gene for obesity based either on knowledge of its physiological role or because it becomes implicated in one or more forms of experimental or naturally occurring animal or human obesity.

Candidate genes based on physiological function
The candidate gene approach has been applied directly to examine the role of a number of polymorphisms in genes known to encode proteins with a role in energy homeostasis including the Trp64Arg mutation of the β 3-adrenoceptor, mutations of the mitochondrial UCPs1-3 and lipoprotein lipase. Unfortunately, the results of this direct approach have often been confusing. Two meta-analyses published in the same year of the Trp64Arg β3-adrenoceptor mutation,  with data from over 40 studies of 7000 subjects, have concluded with rather different assessments of its significance (Allison et al., 1998; Fujisawa et al., 1998).

A common reason for discordant results is that differences in polymorphism frequency   in different populations   may   give rise to population   stratification effects. Furthermore, the presence of a polymorphism may not necessarily lead to alterations in protein structure or function and even if the protein product is altered,  this may not always lead to obesity (an effect seen with certain null mutations of the MC-4 receptor; Hirschhorn and Altshuler, 2002). In addition, publication bias towards positive results may overestimate the strength of an association. Conversely, true candidate gene effects may be missed if they are modulated by the presence of gene–environment interactions (such as those reported to occur between exercise and polymorphisms in the β2-adrenoceptor gene;  Macho-Azcarate et al.,  2002)  or have effects specific to certain ethnic groups or sexes.

At the time of writing, there have been positive reports in human populations of significant association between some 58 candidate genes and various measures of body weight/body composition/energy expenditure/serum leptin and temporal changes in body weight,  although the evidence is somewhat conflicting for a number of these associations (e.g. leptin receptor,  β3-adrenoceptor,  lipoprotein lipase,  UCP-1–3,    pro-opiomelanocortin     (POMC)  tumour necrosis factor and peroxisome proliferator-activated receptor-Y ; Rankinen et al., 2002).

The search for candidate genes amongst animal and human forms of monogenic or syndromic obesity has often been more illuminating.

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