Facilitating precision medicine through analysis of next-generation sequencing projects

Abstract

Precision medicine constitutes an emerging strategy that aims at the individualization of healthcare by considering the personal molecular features and environmental factors of the patient in question. Genetic biomarkers constitute one dimension of a patient’s molecular phenotype that can allow for treatment stratification. As such, incorporating genetic variability into clinical decision making has raised great interest with drug developers, regulators and in the wider medical community. Importantly however, most studies that evaluated associations of genetic variability with drug reponse or toxicity interrogated only selected, mostly common candidate variants and the prevalence and relevance of rare variants for pharmacogenetics remained largely unexplored. This thesis demonstrates how population-scale Next-Generation Sequencing (NGS) data can be leveraged to map the interindividual and ethnogeographic variability of genes with medical importance.

Papers I and II focused on ATP-binding casette (ABC) transporters, as an example of a pharmacogenetically relevant gene family, and show how their variability can have potential predictive value in breast cancer chemotherpy. The human ABC transporter family consists of 48 functionally important membrane proteins which mediate the active transport of a plethora of substrates, including a multitude of endogenous substrates as well as drugs, such as calcium channel blockers and various chemotherapeutics. Because of this physiological and clinical importance, Paper I systematically investigated the interindividual and ethnogeographic variability in the ABC transporter superfamily using NGS data of 138,632 unrelated individuals worldwide, and used an list of sophisticated computational algorithms to estimate their functional relevance. In total, 62,793 exonic variants were discovered, of which 98.5% were rare with minor allele frequencies (MAF) <1.5%. Based on these data, individuals were found to harbor between 9.3 and 13.9 deleterious ABC variants, only 0.3% of which were shared among all populations. As such, this work analyzed the landscape of ABC transporter variability on an unprecedented scale and revealed large interindividual and ethnogeographic variability with potential relevance for the treatment with ABC transporter substrates.

Paper II built on these findings by evaluating whether ABC transporter variability was associated with drug response. As drug resistance due to facilitated ABC transporter-mediated efflux of chemotherapeutics constitutes an important cause of morbidity and mortality, ABC transporter variability was evaluated whether it could predict treatment outcomes in breast invasive carcinoma (BRCA), clear cell renal carcinoma (ccRCC) and hepatocellular carcinoma (HCC). In contrast to previous studies, these analyses did not only consider common ABC polymorphisms but considered also rare genetic variants using mutational burden testing. Importantly, variant burden of ABCC1 was found to significantly assoiate with reduced survival in BRCA patients, specifically in those subgroups treated with the MRP1 (the transporter encoded by ABCC1) 2 substrates doxorubicin (p=0.0088) and cyclophosphamide (p=0.0011). In contrast, no association was discovered in tamoxifen-treated patients (p=0.13). Multiple variants enriched in the high mutational burden group affected residues in functionally important transporter domains providing additional mechanistic support. Combined, these results argue for a model in which multiple variants with individually small effect sizes shape drug resistance, thus incentivizing a shift in strategy away from the interrogation of candidate variants and towards the incorporation of germline data for precision cancer medicine.

Paper III indicated how publically available sequencing data from individuals can be used to provide accurate estimates of population-specific carrier rates and genetic complexity of 450 human autosomal recessive (AR) diseases. Specifically, population-scale NGS data of individuals free from clinically diagnosed congenital disorders was used to identify disease allele carrier frequencies for 450 AR disorders. Using 85 diseases with known epidemiology, the data showed that our prevalence estimates corresponded well to clinically reported incidences (p<0.001; R=0.68). Furthermore, these data allowed for the first time to evaluate the genetic complexity of the human AR diseasome and estimate population-specific founder effects. As such, these analyses reveal the molecular genetics of AR diseases with unprecedented resolution and provide important insights into epidemiology, complexity and population-specific founder effects, which can provide a powerful resource for clinical geneticists to inform population-adjusted genetic screening programs, particularly in otherwise understudied ethnogeographic groups.

In conclusion, by utilizing sophisticated computational methods for the analysis of publically available population-scale sequencing data of >130,000 individuals, this thesis uncovered the landscape of genetic variability in genes with importance for pharmacogenetics and congenital disease. The resulting findings aspire to improve pharmacogenetic interpretations and carrier screening programs and, hopefully, can contribute to the advancement of precision medicine.