It remains largely unknown as to why some individuals experience substantial weight loss with obesity interventions, while others receiving these same interventions do not. Person-specific characteristics likely play a significant role in this heterogeneity in treatment response. The practice of precision medicine accounts for an individual’s genes, environment, and lifestyle when deciding upon treatment type and intensity in order to optimize benefit and minimize risk. In this review, we first discuss biopsychosocial determinants of obesity, as understanding the complexity of this disease is necessary for appreciating how difficult it is to develop individualized treatment plans. Next, we present literature on person-specific characteristics associated with, and predictive of, weight loss response to various obesity treatments including lifestyle modification, pharmacotherapy, metabolic and bariatric surgery, and medical devices. Finally, we discuss important gaps in our understanding of the causes of obesity in relation to the suboptimal treatment outcomes in certain patients, and offer solutions that may lead to the development of more effective and targeted obesity therapies.
|Original language||English (US)|
|Journal||Therapeutic Advances in Endocrinology and Metabolism|
|State||Published - Jan 1 2019|
Bibliographical noteFunding Information:
It was 30 years ago that the American Diabetes Association proposed two classes of diabetes mellitus (DM): insulin-dependent (type 1) and insulin-independent (type 2). 128 Over time, new subgroups were discovered, including latent autoimmune diabetes in adults and mature onset diabetes in the young (MODY). A recent cluster analysis suggested five DM subtypes in adults, each with different patient characteristics and risks for complications. 129 Moreover, it has become clear that the treatments for DM, including therapy type (e.g. sulfonylureas for HNF1A- or HNF4A-MODY) and efficacy, differ depending on the underlying cause. Our understanding of the etiologies underlying obesity may not be far ahead of where our understanding of the etiologies underlying DM were not long ago. Similar to DM, the substantial degree of heterogeneity seen in individual response to weight loss interventions is likely due to an equally large degree of heterogeneity in the cause. Without a clearer understanding of the specific etiology or distinct phenotypes, which may be complex and are unlikely based upon single features, the development of directed treatments will be challenging. While targeted treatments for several forms of monogenic obesity have emerged, most cases of obesity are polygenic in origin. In polygenic obesity, groups of alleles at different gene loci have variants each contributing a small additional effect towards body weight regulation. It may be that every individual with obesity carries his or her own specific polygenic variants. 130 , 131 While precision medicine, as an approach, may be presently better suited for the treatment of monogenic obesity, continued advancements in genetics, pharmacogenetics, and epigenetics may eventually elucidate pharmacotherapeutic options for polygenic forms. The rise of electronic health records (EHRs) and the subsequent creation of EHR-enabled clinical discovery cohorts may provide a valuable tool for examining person-specific characteristics associated with weight loss to interventions in the real-world setting. EHRs can be combined across multiple institutions to increase sample size and statistical power. 132 This is especially helpful for exploring outcomes to interventions in smaller groups of individuals, or for evaluating rare medication side effects. Integrating ‘-omic’ data (e.g. genomic, metabolomic) into the EHR will improve the capacity for identifying additional sources of variability in drug–response relationships that are too challenging to identify from smaller-scale studies. Further, large-scale observational studies combining EHR data with machine learning statistical techniques may allow us to better determine phenotypic characteristics associated with weight loss response to obesity interventions. That said, the heterogeneity in the approach to medical weight management and the inconsistent timing of patient evaluations leads to missing or flawed data, thereby limiting the amount of aggregated data that can be collected from EHR studies. Further, while correlation can be determined from such observational studies, causation cannot be, and compliance often cannot be readily assessed. Similar to the way that combining meta-analyses has increased our identification of the loci and SNPs contributing to the development of obesity and the metabolic syndrome, 10 , 11 combining data from obesity interventional trials may help us better identify subgroups of responders to various treatments. This is especially pertinent in pediatric obesity, where most studies are small and subgroup analyses are therefore limited. Fortunately, attempts are underway to standardize these processes, at least in the adult realm. The Accumulating Data to Optimally Predict obesity Treatment (ADOPT) Core Measures project was designed to provide investigators with tools to generate evidence through the use of common measures following four domains: behavioral, biological, environmental, and psychosocial. 133 Accumulating data on these factors will help inform the design and delivery of effective, tailored obesity treatments. 134 – As mentioned previously, several phenotypic predictors of weight loss response have already been elucidated. The most consistently identified predictors of later response to an intervention are early response and higher adherence, 36 which should be reported in clinical trials. Increased baseline appetite and decreased satiety predict better response, while the presence of disordered eating and psychopathology predict worse response to several interventions. 47 49 , 51 , 85 , 95 , 105 , 123 While these identified ‘primordial’ predictors represent the beginning of our understanding into person-specific characteristics predictive of weight loss response, many are not specific enough to help us tailor therapy. For example, given that increased hunger predicts greater weight loss response to exenatide, 85 topiramate, 89 and phentermine, 94 adding this variable to a pharmacotherapy selection algorithm may not help in the decision-making between these three options. In order to differentiate between which therapies to consider for each patient, we need to uncover personalized predictors that are specific to each intervention. Incorporating neuroimaging (e.g. fMRI), biobanks, and data repositories into studies evaluating characteristics associated with weight loss will help us discover predictors that are more precise. – Finally, future studies should also examine predictors of weight loss response to mobile health technologies, such as smartphone applications. Presently, evidence showing that these tools improve weight loss is mixed; 135 138 however, as with other interventions improved adherence appears to predict greater weight loss response. 139 , 140 Studies should also examine the optimal timing for treatment interventions. Such investigations should focus on determining the window of opportunity for when an intervention should be initiated in order to achieve the best possible response. Given that, among adolescents who develop obesity the most rapid weight gain appears to occur between the ages of 2 and 6 years, earlier interventions are likely needed. 141 The time course for beginning, discontinuing, or intensifying treatment in any population remains elusive and will require further investigation. Funding The authors received no financial support for the research, authorship, and/or publication of this article. Conflict of interest statement J.R.R. received research support in the form of drug/placebo from Boehringer Ingelheim. C.K.F. received research support from Novo Nordisk. S.D.S. received grant funding from Astra Zeneca Pharmaceuticals. A.S.K. received research support (drug/placebo) from Astra Zeneca Pharmaceuticals and served as a consultant for Novo Nordisk, WW, and Vivus Pharmaceuticals but did not accept personal or professional income for these activities. The other authors have no disclosures. ORCID iD Eric M. Bomberg https://orcid.org/0000-0002-8037-4314
© The Author(s), 2019.
- anti-obesity agents
- bariatric surgery
- obesity etiology
- precision medicine
- weight loss