Prof. Inga Prokopenko

Prof. Inga Prokopenko

PhD, e-One Health professor and director of the Multi-Omics Statistics Department at the University of Surrey, associate researcher at the University of Lille, and in charge of the “studies of links between diabetes and cancer risks” project within the Center National PreciDIAB

Today, Professor Inga Prokopenko tells us about his research projects within PreciDIAB and Mendelian randomization that could change the world of human genetics research.

Before focusing on your research on causal links between diabetes and cancer by Mendelian randomization, could you explain to us what led you to this particular area of research?

My decision to work in the field of genetics dates to the secondary school when I first learned about the mathematical beauty of Mendel’s laws. I was fascinated by the cross-disciplinary knowledge required to undertake research in human genetics.

After a BSc/MSc in Biology and Chemistry with specialisation in Human Genetics, my first junior-lead graduate research experience came from large-scale study on causes of birth defects and infertile marriages in Ukraine. Later, during an international genetic epidemiology and molecular genetics MSc and PhD in Pavia, Italy, I contributed to genetic studies for common diseases in population isolates, such as Sardinia, Italy, through extension of multi-allelic TDT test to incomplete trios using the microsatellite data from large families with Multiple Sclerosis. Subsequent employment at the GlaxoSmithKline R&D, the Psychiatry Translational Medicine & Genetics brought excitement of the first large-scale (several thousand individuals with/without depression) fine-mapping study, analyses of first Perlegen whole genome genotyping arrays, and one of the first multi-metabolite investigations on depression, relating it to blood insulin levels. As an applied statistical geneticist, on the tips of my fingers, I felt that soon all my expertise would become invaluable for novel genetic discoveries.

I then moved to a PostDoctoral position at the WTCHG/OCDEM, University of Oxford, UK to focus on large-scale studies investigating the genetic susceptibility to type 2 diabetes (T2D). During that time, within the MAGIC (Meta-Analysis of Glucose and Insulin related traits Consortium), we launched an innovative approach of performing genome-wide association studies for quantitative traits used as endophenotypes of clinical endpoints, such as T2D. My first-author Nature Genetics publication described a genome-wide significant association between a variant within the MTNR1B gene and variability of fasting glucose (FG) levels in non-diabetic individuals and related them to the pathophysiology of T2D. For the first time, we demonstrated a genetic link between circadian rhythms and T2D, underlying a range of metabolic dysfunctions. This discovery led to numerous follow-up studies.

Later, I became a Senior Lecturer followed by Reader Position in Human Genomics at the Department of Genomics of Common Diseases, Imperial College London, UK. Since 2019, I am a Professor in eOne Health at the University of Surrey, Vice-Chancellor’s Distinguished Chair and Head of Section of Statistical Multi-Omics as well as Visiting professor within PreciDIAB at the University of Lille, France. Since August 2021, I am also a Co-Director of the People-Centered AI Institute at the University of Surrey pushing forward the AI implementation for people’s health and wellbeing.

Could you explain to us what is Mendelian randomization and how this method is a real breakthrough for research in human genetics?

My research focuses on method development for the high-dimensional multi-omics data analyses and their application to the large-scale datasets. The innovative approaches use machine learning and AI as well as tackle dissection of the longitudinal multi-omics effects. From the applied perspective, my research focus is on the improved profiling, prevention and progression tracking, evaluation of trajectories in pathogenesis of human diseases. My major interest is in metabolic and early growth phenotypes, diabetes, its major comorbidities, including various cancer types and psychiatric traits, and dissection of multimorbidity through the life span. I am an active leader of major international efforts within DIAGRAMMAGIC, and EGG GWAS consortia. I am also a member of the European Human Exposome network funded through the EU H2020 programme and lead the University of Surrey efforts within the LONGITOOLS project tackling exposome of cardiometabolic traits and disease.

The causality between the exposures (obesity or adiposity for BMI) and their effects on complex traits and phenotypes (outcomes, such as breast cancer, BrC) are now very frequently studied using Mendelian randomisation (MR) approaches. MR (Figure) is a framework to evaluate the causal relationships from observational studies using genetic variants as instrumental variables (IVs) in the analysis. MR assumes that genetic variants are distributed in the populations at random, given the random nature of inheritance patterns and fixation of alleles at the point of conception. This principle of MR could be used in the same way as randomisation in clinical trials. In real life, it would be unethical to set a clinical trial to study obesity (adiposity) and risk of breast cancer relationship, while MR naturally overcomes ethical concerns. When the DNA variants are associated with the exposure, they can constitute the “genetic instrument”. Such instruments should influence the outcome, for example, breast cancer, only via their effect on the exposure (obesity) and should not correlate with potential confounders. Genetic instruments should be strong enough to draw valid inferences about causality, for example, we have used the FTO locus variants as instrument to demonstrate the causal effect of adiposity on insulin resistance in non-diabetic individuals (Fall et al., PLOS Medicine, 2013)1.
Recently, polygenic risk scores, PRSs have proven useful in constituting strong instruments2,3 to address a variety of aetiological questions and e.g. to suggest that adiposity has causal effect on a range of cardiovascular phenotypes1.

This method therefore opens up new perspectives in understanding the causal links between genetics and multiple factors. How are you going to apply this in PreciDIAB and what are the expected benefits of your projects?

Within PreciDIAB, I co-lead, together with Amélie Bonnefond and Philippe Froguel, the work package 2.2 on Markers of diabetes treatment failure and of cancer risk. Within this work package, two PhD students Jared Gichohi Maina and Vincent Pascat, are working together with us on advancement of analyses dissecting the shared genetics between metabolic phenotypes and several cancers, including pancreatic, colorectal, breast and prostate.

Growing observational and experimental evidence suggests that diabetes, obesity and cancers may share biological pathways in their pathogenesis. To disentangle the shared genetic component between diabetes and various types of cancer, we combine polygenic risk scores with hierarchical clustering of effects of these disease-associated DNA variants to disentangle the disease-specific genetic risk into its pathway-related components. We then test the effect of such pathway-related cumulative genetic risks on each disease and define the shared component. We also use Mendelian Randomisation (MR) to infer causal relationships between diabetes obesity, and cancers. Moreover, we implement multi-phenotype genome-wide association study (MPGWAS) approaches to yield a large improvement in the study power and help detecting shared genetic loci between co-morbid outcomes. Similarly, we study the relationships between the blood metabolite levels, diabetes and cancers. Additionally, we dissect the effects of each associated locus on all studied phenotypes to understand, whether the effect on multiple phenotypes, like that of TCF7L2 on diabetes and multiple cancers, is truly multi-phenotype or is mediated through a primary effect on one of the phenotypes. This is an exciting set of analytical efforts, which addresses a range important public health issues of disease comorbidity for better clinical care and prediction of the risks for comorbid conditions.
Within PreciDIAB, my research has already led to the publication of 4 papers, notably one in Nature Communication 4,5,6,7


  1. Fall, T. et al. The role of adiposity in cardiometabolic traits: a Mendelian randomization analysis. PLoS Med 10, e1001474 (2013).
  2. Fall, T. et al. Age- and sex-specific causal effects of adiposity on cardiovascular risk factors. Diabetes 64, 1841-52 (2015).
  3. Hagg, S. et al. Adiposity as a cause of cardiovascular disease: a Mendelian randomization study. Int J Epidemiol 44, 578-86 (2015).
  4. Lagou V, […], et al. Sex-dimorphic genetic effects and novel loci for fasting glucose and insulin variability. Nat Commun. 12(1):24 (2021).
    doi: 10.1038/s41467-020-19366-9.
  5. Balkhiyarova Z, et al. Relationship between glucose homeostasis and obesity in early life-A study of Italian children and adolescents. Hum Mol Genet. ddab287 (2021)
  6. Nouwen A, et al. Measurement invariance testing of the patient health questionnaire-9 (PHQ-9) across people with and without diabetes mellitus from the NHANES, EMHS and UK Biobank datasets. J Affect Disord. 292:311-318 (2021)
  7. Rayevsky A, et al. Functional Effects In Silico Prediction for Androgen Receptor Ligand-Binding Domain Novel I836S Mutation. Life. 11:659 (2021)

Discover the interviews...