Utility of Polygenic Scores for Differentiating Diabetes Diagnosis Among Patients With Atypical Phenotypes of Diabetes

Type 1 Diabetes Genetic Risk Scores: History, Application and Future Directions

PURPOSE OF REVIEW

To review the genetics of type 1 diabetes (T1D) and T1D genetic risk scores, focusing on their development, research and clinical applications, and future directions.

RECENT FINDINGS

More than 90 genetic loci have been linked to T1D risk, with approximately half of the genetic risk attributable to the human leukocyte antigen (HLA) locus, along with non-HLA loci that have smaller effects to disease risk. The practical use of T1D genetic risk scores simplifies the complex genetic information, within the HLA and non-HLA regions, by combining the additive effect and interactions of single nucleotide polymorphisms (SNPs) associated with risk. Genetic risk scores have proven to be useful in various aspects, including classifying diabetes (e.g., distinguishing between T1D vs. neonatal, type 2 or other diabetes types), predicting the risk of developing T1D, assessing the prognosis of the clinical course (e.g., determining the risk of developing insulin dependence and glycemic control), and research into the heterogeneity of diabetes (e.g., atypical diabetes). However, there are gaps in our current knowledge including the specific sets of genes that regulate transition between preclinical stages of T1D, response to disease modifying therapies, and other outcomes of interest such as persistence of beta cell function. Several T1D genetic risk scores have been developed and shown to be valuable in various contexts, from classification of diabetes to providing insights into its etiology and predicting T1D risk across different stages of T1D. Further research is needed to develop and validate T1D genetic risk scores that are effective across all populations and ancestries. Finally, barriers such as cost, and training of medical professionals have to be addressed before the use of genetic risk scores can be incorporated into routine clinical practice.

Validation of GenProb-T1D and its clinical utility for differentiating types of diabetes in a biobank from a US healthcare system

Atypical diabetes with overlapping clinical features of type 1 (T1D) and type 2 (T2D) is common and challenging diagnostically and for implementing effective treatment. Here, we validate a recently reported genetic probability of type 1 diabetes (GenProb-T1D) from the UK Biobank (UKB) for differentiating type 1 diabetes and type 2 diabetes in a diabetes patient cohort from a healthcare system-based biobank in the USA. Among 3,363 diabetes patients, we confirmed the performance of GenProb-T1D in differentiating typical type 1 diabetes vs type 2 diabetes. Furthermore, for 359 atypical diabetes patients, those with GenProb-T1D higher than the pre-defined cutoff derived from the UKB had clinical presentations more consistent with that of typical type 1 diabetes. Similar findings were found in participants of European and non-European ancestries. This study provides necessary validation to translate GenProb-T1D into genetic testing in a multi-ancestry cohort. Measuring underlying genetic susceptibility of type 1 diabetes and type 2 diabetes can supplement current clinical tools for earlier and more accurate diagnoses of diabetes.

Estimating Disorder Probability Based on Polygenic Prediction Using the BPC Approach

Polygenic Scores (PGSs) summarize an individual's genetic propensity for a given trait in a single value, based on SNP effect sizes derived from Genome-Wide Association Study (GWAS) results. Methods have been developed that apply Bayesian approaches to improve the prediction accuracy of PGSs through optimization of estimated effect sizes. While these methods are generally well-calibrated for continuous traits (implying the predicted values are on average equal to the true trait values), they are not well-calibrated for binary disorder traits in ascertained samples. This is a problem because well-calibrated PGSs are needed to reliably compute the absolute disorder probability for an individual to facilitate future clinical implementation. Here we introduce the Bayesian polygenic score Probability Conversion (BPC) approach, which computes an individual's predicted disorder probability using GWAS summary statistics, an existing Bayesian PGS method (e.g. PRScs, SBayesR), the individual's genotype data, and a prior disorder probability. The BPC approach transforms the PGS to its underlying liability scale, computes the variances of the PGS in cases and controls, and applies Bayes' Theorem to compute the absolute disorder probability; it is practical in its application as it does not require a tuning dataset with both genotype and phenotype data. We applied the BPC approach to extensive simulated data and empirical data of nine disorders. The BPC approach yielded well-calibrated results that were consistently better than the results of another recently published approach.

Personalized medicine in a community health system: the NorthShore experience

Genomic and personalized medicine implementation efforts have largely centered on specialty care in tertiary health systems. There are few examples of fully integrated care systems that span the healthcare continuum. In 2014, NorthShore University HealthSystem launched the Center for Personalized Medicine to catalyze the delivery of personalized medicine. Successful implementation required the development of a scalable family history collection tool, the Genetic and Wellness Assessment (GWA) and Breast Health Assessment (BHA) tools; integrated pharmacogenomics programming; educational programming; electronic medical record integration; and robust clinical decision support tools. To date, more than 225,000 patients have been screened for increased hereditary conditions, such as cancer risk, through these tools in primary care. More than 35,000 patients completed clinical genetic testing following GWA or BHA completion. An innovative program trained more than 100 primary care providers in genomic medicine, activated with clinical decision support and access to patient genetic counseling services and digital healthcare tools. The development of a novel bioinformatics platform (FLYPE) enabled the incorporation of genomics data into electronic medical records. To date, over 4,000 patients have been identified to have a pathogenic or likely pathogenic variant in a gene with medical management implications. Over 33,000 patients have clinical pharmacogenomics data incorporated into the electronic health record supported by clinical decision support tools. This manuscript describes the evolution, strategy, and successful multispecialty partnerships aligned with health system leadership that enabled the implementation of a comprehensive personalized medicine program with measurable patient outcomes through a genomics-enabled learning health system model that utilizes implementation science frameworks.