multifactorial breast cancer risk assessment
TRANSCRIPT
Multifactorial Breast Cancer Risk Assessment
Paul JamesParkville Familial Cancer Centre
Breast Cancer risk is a complex trait with contributions from multiple interrelated domains
• High Risk Genes – well characterized average risks• Moderate / Low Risk genes – improving risk estimates• Polygenic Risk Score
• Family History
• Epidemiological Risks: intrinsic: age, height, menstrual hx
• modifiable: weight, hormone use, alcohol
• Mammographic Density
All forms of risk contribute with evidence of minimal interaction / epistasisYang J Clin Oncol 2020, Gallagher JAMA Netw Open 2020, Borde JNCI 2020, Kapoor JNCI 2021
For Mod Risk genes modification can substantially alter clinical interpretation
High Risk
Moderate Risk
CHEK2: PRS Modification
Average
High PRS Quintile
Low PRS QuintilePopulation Risk
Increasing Data on the combined effects of multiple risk factors
How does Multifactorial Risk Assessment work at an individual level?
High Risk Cohort
3092 women affected by BCAssessed by a Familial Cancer Clinic:
• Young; mean age dx 44y• Familial: 66% 1st/2nd Deg with BC
37% multiple relatives BC• Enriched: ↑grade, ER-/TNBC,
multiple primary BC, OC, Male BC
Previous clinical genetic testing excluded BRCA1, BRCA2
Low Risk Cohort
4823 women, unaffected by BC
Older (mean age 64)Ongoing screening – with continued cancer status updates
Family Hx: 32% 1st/2nd Deg BC9% multiple relatives BC
• Panel Gene Testing – 10 established BC predisposition genesPALB2, CHEK2, ATM, BARD1, RAD51C, RAD51D, TP53, CDH1, PTEN STK11
• SNP Genotyping (Mavaddat et al., 2015 PRS-77) – PRS Score (70 SNPs)• Family History (all cancers to 3rd degree relatives)
RESULTS
• The strength of each risk factor assessed by regression modelling.
Pathogenic Variants in Panel Genes:5% Clinic Cases1.4% Healthy Screened Controls
PRS-70: HR per SD 1.64 (1.55-1.71)
Effect of polygenic risk greater for:
Early onset: Before 50 years: HR 1.64 (1.56-1.72)After 50 years: 1.43 (1.36-1.50)
ER Positive disease: ER+ : HR 1.76 (1.65-1.86)ER - : 1.41 (1.31-1.52) Cumulative Lifetime Risk of BC (%)
Unaffected
Affected
AUC 0.64 (0.63-0.65)
RESULTS• The strength of each risk factor assessed by regression modelling.
• Weighted adjustment of all BC/OC family history to account for ascertainment in cases
Family History:OR 95% CI
Breast cancer 1st/2nd Deg Relative 1.33 1.28-1.39
1st deg relative diagnosed <40 y 1.72 1.38-2.14
Multiple BC (≥3) 2.55 2.26-2.87
RESULTS• The strength of individual risk factors assessed by regression modelling.
PRS, Single Genes, and FHx• All contribute risk• Are essentially independent
PRS HR in PV carriers: 1.75 (1.26‐2.45) p=0.001
Attenuation 2‐10%
Individually Combined
Logistic Regreesion Multiple Regresion
OR 95% CI OR 95% CIPRS-70Per SD 1.64 1.56-1.71 1.61 1.54-1.69
Pathogenic Variants10 genes
3.80 2.79-5.16 3.51 2.55-4.83
BC Family historyPer 1st/2nd Degree Relative
1.33 1.28-1.39 1.29 1.24-1.35
Family History PRS
Panel Test 0.02 0.56
Famliy History - 0.65
Interaction Termsp-values
Panel Gene TestingBest estimate of single gene risks
Polygenic RiskAttenuated HR per SD
BC Family HistoryAttenuated OR for FDR/SDR +
age of diagnosis
LTRi
LTRi – individual’s lifetime riskLTR0 – population average LTR
Composite Lifetime Risk of Breast cancer
75% 37%
15% 37%
10% 26%
BC Unaffected
CHEK2 pathogenic variantsn = 75
High
Mod
Pop
OR 2.5
Composite Risk
Range of Observed Composite Risk in carriers of Pathogenic Variants
PALB2 CHEK2 ATM RAD51C RAD51D BARD1 TP53, PTEN, CDH1
LifetimeRisk
Median – } Standard deviation. Lines = 90% range, Outliers: = BC cases, = Unaffected
79% 46%
17% 36%
4% 17%
40% 29%
60% 71%
0% 0%
High
Mod
Pop
Genetic Testing Composite Risk
Cases Controls
LoF Variants in 10 BC Panel Genesn=217: PALB2, ATM, CHEK2, RAD51C/D, BARD1, TP53, CDH1, PTEN, STK11
Cases Controls
Genetic Testing
Genes + FHx
Genes + FHx + PRS
Affected
Unaffected
50% Actionable
Increased Risk
50% BelowAverage Risk
X2 =1264 p < 0.0001
Integrated risk assessment improves the accuracy in the Familial Cancer Clinic…
…and in general population screening
n=3092
n=4823
3.4% 49% 29% 63%Net Reclassification Improvement:
CONCLUSIONS
• Polygenic risk and FHx combine to explain an important component of ‘penetrance’ in individuals with rare pathogenic variants in BC predisposition genes…
• And all three elements together provide the best assessmentof breast cancer risk available…with the addition of personalrisk factors
• ?? any reason not to include PRS with genetic testing currently offered to women referred to clinical services
Unaffected
Affected
Cumulative Lifetime Risk of BC (%)
AUROC 0.74 (.73‐.75)
USING POLYGENIC RISK MODIFICATION TO IMPROVE BREAST CANCER PREVENTION
An RCT of personalised risk assessment and risk management for unaffected women undergoing assessment for HBOC
2400 Women undergoing PT randomised to receive composite risk assessment or single gene test result
5yr Study of PRS in Practice
Inclusion criteria• Female, no phx DCIS BrCa OvCa• Aged >18 years• Undergoing BRCA1 BRCA2 PALB2
CHEK2 ATM RAD51C RAD51Dpredictive test
• Have access to, and basic familiarity with, digital platforms
Exclusion criteria• Unable to read/understand
participant materials• Undergoing cancer treatment• Previous genomic testing
ELIGIBILITY
RECRUITMENT
Dedicated clinician education program
2000 participants, enrolled before first appointment
Randomised to receive:
1. Integrated risk assessment = FHx / PRS / genetic test result+ personal risk factors
Or
2. Standard care (single gene result)
Followed for 3+ years
All data collection online
Genotyping with the Illumina ‘Confluence’ custom array
PROTOCOLProject ManagerResearch Coordinator
Short term:i. Compare distribution of
assessed riskii. RM recommendations and
intentions
Medium to long term:iii. RM uptake / adherenceiv. RA accuracy and calibration
AIMS & OBJECTIVES
CLINICAL
ANALYSE PERFORMANCE OF BREAST CANCER RISK MANAGEMENT THAT
INCORPORATES PRS
HEALTH SERVICES
ESTABLISH HEALTH SERVICES IMPLICATIONS OF IMPLEMENTING PRS
PATIENT EXPERIENCE
EVALUATE PATIENT EXPERIENCE OF UNDERGOING PERSONALISED
GENOMIC TESTING
i. Evaluate cancer‐related anxiety, adaptation, impact, value, patient‐assessed acceptability
ii. Qualitative interviews in a subset of participants
i. Simulation modelling long‐term implications for cancer outcomes and healthcare resources
ii. Economic evaluation to assess the incremental cost‐effectiveness
Acknowledgements
Supported by: Thanks - : Participating patients and their families
Staff from Victorian, NSW and Tasmanian Familial Cancer Centres
Simone McInernyLyon MascarenhasJo McKinleyMaryAnne YoungMelissa SoutheyGeorgia Chevenix‐TrenchTatiane YanesFamilial Cancer Research Grp
Cancer Genomics Lab (PMCC)Ian CampbellNa LiKylie GorringeBelle Lim
Familial Cancer CentresMarion HarrisAlison Trainer
Geoffrey LindemanIngrid Winship
Yoland AntilAinsley Campbell
Pathology North NSWRodney Scott
LifePool StudyLisa Devereux