PhenX: a toolkit for interdisciplinary genetics research

Abstract

Purpose of review

To highlight standard PhenX (consensus measures for Phenotypes and eXposures) measures for nutrition, dietary supplements, and cardiovascular disease research and to demonstrate how these and other PhenX measures can be used to further interdisciplinary genetics research.

Recent findings

PhenX addresses the need for standard measures in large-scale genomic research studies by providing investigators with high-priority, well established, low-burden measurement protocols in a web-based toolkit (https://www.phenxtoolkit.org). Cardiovascular and Nutrition and Dietary Supplements are just 2 of 21 research domains and accompanying measures included in the PhenX Toolkit.

Summary

Genome-wide association studies (GWAS) provide promise for the identification of genomic markers associated with different disease phenotypes, but require replication to validate results. Cross-study comparisons typically increase statistical power and are required to understand the roles of comorbid conditions and environmental factors in the progression of disease. However, the lack of comparable phenotypic, environmental, and risk factor data forces investigators to infer and to compare metadata rather than directly combining data from different studies. PhenX measures provide a common currency for collecting data, thereby greatly facilitating cross-study analysis and increasing statistical power for identification of associations between genotypes, phenotypes, and exposures.

Introduction

Gene–environment interactions underpin virtually all human diseases, including variations in their severity and clinical presentation. Fundamental understanding of cause of complex disease offers the opportunity to identify and classify genetic risk alleles and to prescribe genetically informed pharmaceutical therapies, as well as to alter environmental exposures to prevent and manage disease initiation and progression. An individual’s genetics influences the capacity to adapt to changing environments and elicits physiological responses in both health and disease. In turn, environmental agents can affect chromatin structure and stability, and gene expression levels. The introduction of high-throughput genotyping to population-based studies has led to the emergence of genome-wide association studies (GWAS) and a revolution in the way that scientists think about the interplay of genetics and common, complex diseases [1].

GWAS provide the potential to inspire new hypotheses and lead to the elucidation of the biochemical and physiological basis of pathogenesis [2]. However, the lack of inclusion of standard measures in GWAS has made it difficult to compare and/or combine these studies despite the fact that many risk factors and phenotypes are common across multiple conditions (e.g. obesity, diet, and smoking) and is a recognized limitation.

PhenX (consensus measures for Phenotypes and eXposures) addresses the need for standard measures in large-scale genomic research studies by providing investigators with high-priority, well established, low-burden measures [3^•]. The measures and protocols in the PhenX Toolkit are chosen by content experts and are reviewed by the PhenX Steering Committee. Table 1 shows the 21 high-priority research domains that were selected by the PhenX Steering Committee for inclusion in the toolkit.

Table 1.

PhenX research domains

Alcohol, tobacco, and other substances	Ocular
Anthropometrics	Oral health
Cancer	Physical activity and physical fitness
Cardiovascular	Psychiatric
Demographics	Psychosocial
Diabetes	Reproductive health
Environmental exposures	Respiratory
Gastrointestinal	Skin, bone, and muscle and joint
Infectious diseases and immunity	Social environment
Neurology	Speech and hearing
Nutrition and dietary supplements

PhenX domains

Each PhenX domain is defined as a field of research with a unifying theme that includes easily enumerated quantitative and qualitative measures. A Working Group of expert scientists is assembled for each domain. Through a consensus process and with input from the broader scientific community, the Working Group selects up to 15 measures to be included in the PhenX Toolkit.

The article focuses on measures from two PhenX domains, Nutrition and Dietary Supplements and Cardiovascular, and the potential application of measures within those domains to assess metabolic syndrome.

Nutrition and dietary supplements

Gene–nutrient interactions influence risk for the initiation and/or progression of common chronic diseases, including diabetes (type 2), metabolic syndrome, cardiovascular and neurological disease, osteoporosis, and cancers [4^••,5,6,7^••]. Whereas it is established that human genetic variation can confer food and even individual nutrient tolerances/intolerances (e.g. lactose, iron) [8], there is increasing evidence that genetic variation influences individual nutrient requirements required to maintain health [9^••] as well as nutrient utilization and metabolism in chronic disease [7^••]. GWAS permit the comprehensive identification of genetic variation that interacts with dietary components to confer risk and/or protective phenotypes, and thereby enable the derivation of genetically informed dietary approaches to promote health and prevent and/or manage chronic disease in genetically diverse populations [10].

The scope of the PhenX Nutrition and Dietary Supplements domain is complex because of the number of essential dietary components. There are more than 40 nutrients (minerals, vitamins, fatty acids, amino acids) that are essential for human health. The Working Group was challenged by the structure of the PhenX framework and initially struggled to balance the competing priorities of burden, utility, scope of the domain, and scientific rigor. The Working Group first considered measures and methodologies related to diet, which are best described as dietary exposures or indicators of nutritional status. Dietary exposure levels can be obtained from questionnaires and nutrient composition databases. For some, but not all nutrients, dietary exposure predicts nutritional status, but this relationship is modifiable by genetics and environment. For selenium and vitamin D, diet does not predict nutritional status and therefore biomarker assays are required to quantify nutritional status. This includes physiological measurements of dietary components in human blood or tissue. These nutritional status measurements require analytical chemistry/biochemistry-based methodologies that quantify biomarkers. The Working Group was able to identify both dietary exposure and nutrient status measures that encompassed all essential nutrients to be included in the PhenX Toolkit. The rationale of the Working Group was that all essential nutrients are required for health and therefore should not be ranked or prioritized for GWAS unless the study is driven by a specific and narrow hypothesis.

The Nutrition and Dietary Supplements Working Group chose 13 measures (Table 2) that, taken together, assess all essential nutrients as dietary exposures and/or nutrient status measurements. The Working Group identified ‘gold standard’ methodologies that will enable data to be compared across ethnically diverse populations with acceptable burden to investigators and study participants. The Working Group recommended the multiple pass 24-h recall protocol for measuring total dietary intake [11]. This protocol enables virtually all aspects of diet to be quantified, including dietary intake of most essential nutrients. Because vitamin D and selenium cannot be quantified from a questionnaire or the 24-h recall, the Working Group deemed it important to include bioassay protocols to address these nutrients. In addition, the Working Group identified a limited number of low-burden, validated screeners that are used to quantify dietary exposures. The selected measures will permit the research community to standardize data collection in nutrition, thus facilitating identification of gene–nutrient interactions in health promotion, disease prevention, and disease progression.

Table 2.

PhenX measures (Nutrition and dietary supplements domain)

Breast-feeding

Calcium intake

(a) Calcium intake by adults (daily)

(b) Calcium intake by children

Caffeine intake

Dairy food intake (daily servings)

Dietary supplements use

Fiber intake

Fruits and vegetables intake

Percentage energy from fat

Selenium

Sugar intake (added)

Vitamin D

Total dietary intake

Cardiovascular

Cardiovascular diseases (CVDs) have complex cause and are a prevalent cause of mortality and morbidity. The pathogenesis of CVD involves nearly all organ systems and metabolic pathways. Moreover, the evidence for multiple risk factors, both genetic and environmental, has been established, whereas other putative risk biomarkers are being investigated. Through observational and interventional epidemiologic studies the genetics of infrequent but high-impact effects of single genes, such as those for familial hypercholesterolemia [12], have been documented. However, much of the genetic risk for CVD is dependent on the relatively small effects of multiple genes and the interaction of these genes with environmental factors [13]. CVD provides a good example of the need to integrate a broad range of phenotypic measurements, both risk biomarkers and clinical outcomes, into GWAS.

The Cardiovascular Working Group developed a parsimonious list of measures that captures important components of both risk and disease expression, but with minimal personal and technical burden to both study subjects and investigators. The Working Group recognized that these considerations are critical to gaining widespread incorporation of the measures into studies that are not focused on CVD. This approach required compromises between technically sophisticated and expensive methods that could provide specific assessments and less burdensome, but less precise assessments drawn from numerous epidemiologic studies of cardiovascular conditions.

The PhenX Toolkit provides protocols for making robust measures of CVD risk markers, including environmental components (such as diet, obesity), and of clinical conditions and pharmacologic interventions. In addition, the Cardiovascular Working Group selected 14 measures (Table 3) that can assess clinical conditions and that are widely accepted and standard measures. For example, a measure for a lipid panel includes protocols for total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides. The lipid protocols are measured as continuous variables without classification (e.g. normal, borderline, or abnormal), facilitating statistical analysis for GWAS and avoiding the potential issues that can arise as categorical variables are reinterpreted or revised by expert committees seeking to refine clinical treatment goals.

Table 3.

PhenX measures (Cardiovascular domain)

Family history of heart attack

Lipid profile

Blood pressure (adult/primary)

High blood pressure during pregnancy

Heart valve function

Angina

Sudden cardiac arrest

Myocardial infarction

Peripheral arterial disease

Abdominal aortic aneurysm

Arrhythmia (atrial and ventricular)

Deep venous thrombosis

Pulmonary embolism

Rheumatic fever/rheumatic heart disease

The field of genetics has played an important role in the understanding of CVD. For example, the correlation between Lp(a) levels and coronary disease had been observed frequently but a causal relationship between the two was not supported. However, this causal relationship was recently elucidated by demonstrating a relationship between a genotype for increased Lp(a) and an increased risk of coronary disease [14^•]. The association of polymorphisms at the LPA locus on 6q26–27 for Lp(a) explained the majority of variability for coronary disease and established this link to polymorphisms predicting coronary heart disease [14^•].

Using PhenX measures: metabolic syndrome

Metabolic syndrome is a cluster of inter-related risk biomarkers for CVD and type 2 diabetes. This clustering reflects both environmental and genetic factors and their interaction. As such, metabolic syndrome provides a salient example of the utility of the PhenX Toolkit to investigators to test associations across different clusters of phenotypes and exposures to define complex conditions or syndromes. A harmonized definition of metabolic syndrome includes elevated blood pressure, dyslipidemia (elevated triglycerides and lower high-density lipoprotein cholesterol), raised fasting glucose, and central obesity [15^••]. These factors for metabolic syndrome are strongly influenced by genetic factors and the molecular and physiological relationships require accurate phenotypic and genetic assessment to determine potential explanatory mechanisms [16^••].

Figure 1 illustrates the interaction of genome and environment and the potential effect on risk phenotypes, highlighting the importance of multiple cross-cutting variables chosen from varying disciplines. In addition to the measures identified by the Cardiovascular and Nutrition Working Groups, the obesity risk phenotype may include waist circumference and body composition, measures identified by the Anthropometrics Working Group. The risk phenotype metabolic syndrome might further expand the list of variables of interest by including age (Demographics Working Group) and smoking (Alcohol, Tobacco, and Other Substances Working Group), in addition to those already included in obesity as well as those highlighted by both the Nutrition and Cardiovascular Working Groups.

Figure 1. Gene–environment interactions in the ontogeny of disease.

The diagram illustrates how an individual’s genome and environment interact and contribute to risk phenotypes which are associated with the development of disease.

Thus, PhenX provides a common platform on which to build an integrated approach for GWAS to define the interacting risk factors and exposures that underpin CVD, diabetes, and their common complications. More specifically, FTO was identified as a risk allele for metabolic syndrome through GWAS and has been implicated in hunger/satiation control, energy intake, and energy metabolism [17^•]. FTO variants are associated with body mass index, hip circumference, and body weight [18] and have recently been shown to interact with the macronutrient composition of diet [19]. The use of PhenX measures will lead to the ability to analyze and reanalyze data from numerous studies and to systematically establish risk factors for obesity-related morbidity. GWAS of coronary artery disease risk have also identified unexpected associations with metabolic pathways whose function is dependent on essential nutrients, and thereby reveal new gene–nutrient interactions and the potential for novel nutritional interventions, including genes involved in mitochondrial folate-dependent one-carbon metabolism [13] and uric acid metabolism [20]. Likewise, investigations involving the influences of maternal diet on fetal genome programming leading to risk of metabolic syndrome in offspring are also amenable to association studies [21].

The PhenX Toolkit

The PhenX Toolkit presents the measures and protocols free of charge to the scientific community in a web-based resource (https://www.phenxtoolkit.org/). Within the toolkit, a brief description of each measure, its purpose, the rationale for its inclusion, the standard protocols for collecting the data, and references are provided. The requirements (e.g. training, personnel, and equipment needed to collect the data) are described. Toolkit users can search or browse for measures and protocols. Users select measures by adding them to a Cart (i.e. a specific collection of measures). Registered users can also save multiple collections of measures (Carts) and share their Carts with other registered users via a Toolkit Network. This enables investigators who are planning studies or expanding an existing study to work together to include a common set of PhenX measures for future analyses. Users can download a report that details their selections and can easily ‘cut and paste’ protocols to incorporate them into their study design.

Conclusion

The PhenX Toolkit is designed to assist investigators who are planning genomic studies and to promote the use of standard measures. The PhenX Toolkit makes it easier for researchers to expand their studies by incorporating measures of scientific interest that are outside of their primary research focus. As noted earlier, metabolic syndrome is just one example of an area of research that can benefit from the use of broadly accepted measures in two differing fields of study such as those included in the Cardiovascular and Nutrition and Dietary Supplements domains. As scientists seek to broaden their understanding of underlying determinants of complex diseases and effects of gene–environment interactions, the use of high-priority, standardized measures in fields which may not be their area of expertise will promote cross-study analysis, effectively increasing sample size and statistical power for validating prior study results and detecting relatively modest genetic associations. Increased statistical power will facilitate the identification of gene–gene and gene–environment interactions and lead to a better understanding of risk phenotypes. This knowledge should promote development of multi-faceted strategies for the prevention, management, and treatment of disease.