Clare Turnbull, PhD FRCP - PowerPoint Presentation

1. Clare Turnbull, PhD FRCP FRCPath MFPH MSc (Epidemiology)Professor in Translational Cancer Genetics, Institute of Cancer ResearchNHS Consultant in Clinical Genetics (Honorary), Royal Marsden NHS Foundation TrustConsultant in Public Health Medicine (Honorary), NDRS (NHS Digital/NHSE)Omics and Precision Public HealthPopulation Genomic ScreeningFPH MeetingSept 2023

2. Financial DisclosuresScientific Advisory Boards/ Educational Activities:Astra Zeneca, Guardant, RocheAll monies donated to charity

3. OVERVIEWCurrent applications of germline genomics in healthFuture (proposed) applications of germline genomics in healthPrinciples of population screeningNewborn Whole Genome Sequencing Population polygenic risk profiling (PRS)Conclusions

4. Current applications of germline genomics in healthDiagnosis in patients with features of rare disease; management of immediate familySmall gene panels (hereditary breast-ovarian cancer)Large gene panels (retinitis pigmentosa)Whole exomes/genomesAgnostic approachRapid service for acutely unwell neonatesSingle gene tests (NF1, CDH1, TP53)Very specific phenotype (NF1)Test with caution, ie when problematic management implications (CDH1, TP53)Though in USA/commercial gene panels widely included

5. Four horsemen of Genotype-Phenotype Prediction 1) Variable expressivity (burden of modifiers)4) Variant Interpretation 2) Biased Ascertainment3) Variant-specific risks

6. Variable Expressivity/Ascertainment biasMirshahi et al AHJG 2022 . 

7. Four horsemen of Genotype-Phenotype Prediction 1) Variable expressivity (burden of modifiers)4) Variant Interpretation 2) Biased Ascertainment3) Variant-specific risks

8. OVERVIEWCurrent applications of germline genomics in healthFuture (proposed) applications of germline genomics in healthPrinciples of population screeningNewborn Whole Genome Sequencing Population polygenic risk profiling (PRS)Conclusions

9. Future (proposed) applications of germline genomics in healthPopulation testing using whole genome sequencing?at birth: ascertainment of neonatal/paediatric rare diseaseNGP: 100,000 newborns from NHSGenomic prediction of common diseasesPolygenic risk scoresPrioritise/deprioritise individuals for disease screening/preventative interventionsOFH: 5 million NHS users

10. OVERVIEWCurrent applications of germline genomics in healthFuture (proposed) applications of germline genomics in healthPrinciples of population screeningNewborn Whole Genome Sequencing Population polygenic risk profiling (PRS)Conclusions

11. Epidemiology and Natural HistoryLethality/frequency: important public health problemLatent period well understood:Natural History/outcomes: well characterisedReliable association/robust marker for serious diseaseInterventionPresymtomatic intervention improves long-term outcomes (survival, QoL) compared to intervention when symptomaticAvailable to all (fully-funded)Screening TestPrecise (sensitivity/specificity), validatedKnown distribution of test values in the target populationScreening Prog ImplementationHigh quality randomised controlled trials show  mortality/ morbidityHealth economic caseClinically, socially and ethically acceptable: health professionals +public1968: Wilson and Jungner (WHO): Criteria for population screeningDetailed disease-centric evaluation

12. 1996: UK National Screening CommitteeEstablished to ensure robust review of new screening approaches3) Detailed health economic evaluation (using observed metrics)1) Detailed disease centred review of epidemiology, natural history studies and existing trial data (Wilson Jungner)2) Regional piloting with comparison of outcomes (akin to cluster randomised trial)

13. OVERVIEWCurrent applications of germline genomics in healthFuture (proposed) applications of germline genomics in healthPrinciples of population screeningNewborn Whole Genome Sequencing Population polygenic risk profiling PRS)Conclusions

14. Newborn WES/WGSProtein test(MS/MS)Genomic Test(WES)Sensitivity99.0%88%Specificity99.8%98.4%NBSeq project4.5 million newborns: bloodspots w FU dataCalifornia 2005-201348 Diseases (Inborn Errors of Metabolism)What does this mean in practice?Many cases of metabolic disease will be present by time WGS is returnedConfirmatory (diagnostic tests)Some cases will already have developed lethal/irreversible diseaseWhat to do with positive genomic result if diagnostic tests normalIs it a false positive or will the disease present laterNo immediate superior confirmatory “diagnostic” test can rule out future disease“Screen-positive, disease indeterminate (SPID)” CFFollow-up of SPIDs How: what tests/tools?How often?For how long?For dominant diseasesPenetrance lower; onset laterTechnology-centric approach: >200 diseasesAdhikari AN et al. Nature medicine 2020; 26(9): 1392-7.UK population700,000 newborns: 11,200 FPs/year

15. Epidemiology and Natural HistoryLethality/frequency: important public health problemLatent period well understood:Natural History/outcomes: well characterisedReliable association/robust marker for serious diseaseInterventionPresymtomatic intervention improves long-term outcomes (survival, QoL) compared to intervention when symptomaticAvailable to all (fully-funded)Screening TestPrecise (sensitivity/specificity), validatedKnown distribution of test values in the target populationScreening Prog ImplementationHigh quality randomised controlled trials show  mortality/ morbidityHealth economic caseClinically, socially and ethically acceptable: health professionals +public1968: Wilson and Jungner (WHO): Criteria for population screeningDetailed disease-centric evaluation// Technology-centric approach:

16. Epidemiology and Natural HistoryLethality/frequency: important public health problemLatent period well understood:Natural History/outcomes: well characterisedReliable association/robust marker for serious diseaseInterventionPresymtomatic intervention improves long-term outcomes (survival, QoL) compared to intervention when symptomaticAvailable to all (fully-funded)Screening TestPrecise (sensitivity/specificity), validatedKnown distribution of test values in the target populationScreening Prog ImplementationHigh quality randomised controlled trials show  mortality/ morbidityHealth economic caseClinically, socially and ethically acceptable: health professionals +public1968: Wilson and Jungner (WHO): Criteria for population screeningDetailed disease-centric evaluation// Technology-centric approach:

17. OVERVIEWCurrent applications of germline genomics in healthFuture (proposed) applications of germline genomics in healthPrinciples of population screeningNewborn Whole Genome Sequencing Population polygenic risk profiling (PRS)Conclusions

18. rarecommongeneticcenvironmental1) “Complex” diseases are complexMucci LA et al.. Jama 2016; 315(1): 68-76.Zdravkovic S,et al. J Intern Med 2002, 252:247-54.15%Breast CancerColorectal Cancer37%Myocardial Infarction (male, fatal) 31%

19. rarecommongeneticcenvironmental1) “Complex” diseases are complexMucci LA et al.. Jama 2016; 315(1): 68-76.Zdravkovic S,et al. J Intern Med 2002, 252:247-54.15%Breast CancerColorectal Cancer37%Myocardial Infarction (male, fatal) 31%13%

20. rarecommongeneticcenvironmental1) “Complex” diseases are complexMucci LA et al.. Jama 2016; 315(1): 68-76.Zdravkovic S,et al. J Intern Med 2002, 252:247-54.15%Breast CancerColorectal Cancer37%Myocardial Infarction (male, fatal) 31%13%7%

21. 21Top 20% of PRSBottom 80% of PRSTrue positives (37%)False positives (19-20%)PRS for Breast Cancer: AUC = 0.6463% of cancers

22. 2) PRS inherently limited in predictiveness    Female PopulationCurrent PRSFuture PRSOptimised PRS  AUC=0.64AUC=0.69AUC=0.71 Female PopulationTop 50% Top 20% Top 50% Top 20% Top 50% Top 20% % BC cases100%70%37%76%44%78%48%Lifetime Risk14.3%19.5%25.3%20.9%29.3%21.6%31.3%Risk age 40-491.7%2.4%3.2%2.6%3.8%2.7%4.1%Fritsche LG, et al. American journal of human genetics. 2020;107(5):815-36.Zhang YD, et al. Nature communications. 2020;11(1):3353.Hingorani A, Gratton J, et al. medRxiv. 2022.Huntley C, et al Lancet Oncology. 2023 (in press).

23. 2) PRS inherently limited in predictiveness    Female PopulationCurrent PRSFuture PRSOptimised PRS  AUC=0.64AUC=0.69AUC=0.71 Female PopulationTop 50% Top 20% Top 50% Top 20% Top 50% Top 20% % BC cases100%70%37%76%44%78%48%Lifetime Risk14.3%19.5%25.3%20.9%29.3%21.6%31.3%Risk age 40-491.7%2.4%3.2%2.6%3.8%2.7%4.1%Fritsche LG, et al. American journal of human genetics. 2020;107(5):815-36.Zhang YD, et al. Nature communications. 2020;11(1):3353.Hingorani A, Gratton J, et al. medRxiv. 2022.Huntley C, et al Lancet Oncology. 2023 (in press).

24. 2) PRS inherently limited in predictiveness    Female PopulationCurrent PRSFuture PRSOptimised PRS  AUC=0.64AUC=0.69AUC=0.71 Female PopulationTop 50% Top 20% Top 50% Top 20% Top 50% Top 20% % BC cases100%70%37%76%44%78%48%Lifetime Risk14.3%19.5%25.3%20.9%29.3%21.6%31.3%Risk age 40-491.7%2.4%3.2%2.6%3.8%2.7%4.1%Fritsche LG, et al. American journal of human genetics. 2020;107(5):815-36.Zhang YD, et al. Nature communications. 2020;11(1):3353.Hingorani A, Gratton J, et al. medRxiv. 2022.Huntley C, et al Lancet Oncology. 2023 (in press).

25. ABABRisk-StratificationPGSDetection of DiseaseScreening ToolALimited Predictiveness“High-Risk”: modest risk“Low-risk” group: majority of diseaseDo not predict lethal diseaseDifferential performance with ancestry Poor sensitivity for specificityOver-diagnosis of indolent diseaseLimited impact on survival“Enrich” for disease incidence in “high-risk” quantile3) Two-stage Screening Approach

26. 4) Population testing poorly studied in the actual populationMany unknowns around PRS risk profiling in real-world populationStudies to date (including OFH) based on atypical individuals who volunteer for screening studiesUptake of PRS as condition to access screeningWiden health disparities for already underserved populationsPRS less predictive in non-EuropeansUnintended Behavioural Implications“High” Risk individuals: anxiety, health-seeking“Low” Risk individuals: neglect symptoms, disregard health prevention adviceAcceptability of withdrawal of screening in “low-risk”who still have appreciable frequency of disease of equivalent lethalityjust modestly more expensive to screen for

27. OVERVIEWCurrent applications of germline genomics in healthFuture (proposed) applications of germline genomics in healthPrinciples of population screeningNewborn Whole Genome Sequencing Population polygenic risk profiling (PRS)Conclusions

28. Conclusions

29. Population Genomic Screeningvs Typical models of population screening1) Limitations to PredictivenessUncertainty in predictionRare pathogenic variants in Rare DiseaseModest enrichmentCommon Variants (PGS) in Common Disease2) Longterm (Lifelong) ImplicationsResource + cost: Monitoring/disease screeningPsychological and behavioural sequelae

30. Epidemiology and Natural HistoryLethality/frequency: important public health problemLatent period well understood:Natural History/outcomes: well characterisedReliable association/robust marker for serious diseaseInterventionPresymtomatic intervention improves long-term outcomes (survival, QoL) compared to intervention when symptomaticAvailable to all (fully-funded)Screening TestPrecise (sensitivity/specificity), validatedKnown distribution of test values in the target populationScreening Prog ImplementationHigh quality randomised controlled trials show  mortality/ morbidityHealth economic caseClinically, socially and ethically acceptable: health professionals +public1968: Wilson and Jungner (WHO): Criteria for population screeningDetailed disease-centric evaluation// Technology-centric approach:

31. Epidemiology and Natural HistoryLethality/frequency: important public health problemLatent period well understood:Natural History/outcomes: well characterisedReliable association/robust marker for serious diseaseInterventionPresymtomatic intervention improves long-term outcomes (survival, QoL) compared to intervention when symptomaticAvailable to all (fully-funded)Screening TestPrecise (sensitivity/specificity), validatedKnown distribution of test values in the target populationScreening Prog ImplementationHigh quality randomised controlled trials show  mortality/ morbidityHealth economic caseClinically, socially and ethically acceptable: health professionals +public1968: Wilson and Jungner (WHO): Criteria for population screeningDetailed disease-centric evaluation// Technology-centric approach:

32. Population Genomic screeningAll screening programs do harm; some do good as well, and, of these, some do more good than harm at reasonable cost”.Genomics is not an amateur sportMuir Gray, 2008National Screening CommitteeProf Maria Bitner Glindzicz, 2015Great Ormond St HospitalPatient GroupsGovernmentDiagnosticssectorPharmaEarly detection Targets