1 / 32

Opportunities in HCC Risk Assessment: Cohorts, Biomarkers, and Decision-Making

Explore the latest advancements in HCC risk assessment, including prospective cohorts, effective surveillance strategies, cost-effectiveness modeling, new biomarkers, and decision-making aids.

dlarry
Download Presentation

Opportunities in HCC Risk Assessment: Cohorts, Biomarkers, and Decision-Making

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Assessment of risk and decision analysis Chicago, ILCA 2019 Pierre Nahon Service d’Hépatologie Hôpital Jean Verdier Bondy – Université Paris 13 INSERM 1162 - Paris 5 Génomique fonctionnelle des tumeurs solides

  2. Financial Disclosures Honoraria or consultation fees: Abbvie, Astra-Zeneca, Bayer, Bristol-Myers Squibb, Gilead Sciences, IPSEN

  3. Opportunities in HCC Risk assessment • Large prospective multicentre cohorts and consortiums • Evidence for effective surveillance, intervention and prevention strategies in high risk individuals • Design for chemoprevention and/or improved surveillance trials • Modeling cost-effectiveness and burden of disease by stratifying the population by risk and intervention • New risk assessment methodologies and evaluation techniques • Promising new biomarkers • Progress in research for communicating risk, decision-making and decision aids

  4. Opportunities in HCC Risk assessment • Large prospective multicentrecohorts and consortiums • Evidence for effective HCCsurveillance, intervention and prevention strategies in high risk individuals • Design for chemoprevention and/or improved surveillance trials • Modeling cost-effectiveness and burden of disease by stratifying the population by risk and intervention • New risk assessment methodologies and evaluation techniques • Promising new biomarkers • Progress in research for communicating risk, decision-making and decision aids

  5. 2018: HCC surveillance becomes a collective responsability Extension of at-riskpoulation

  6. Level of evidence: multicentre cohortstudiestakingintoaccount lead-time bias CirVir CO12 Costentin et al, Gastroenterology 2018 Compliance Earlydetection

  7. Cooper, Gastroenterology 2018 Risk-basedstrategyincorporatingprecisionmedicine Refinement of periodicity/modality of surveillance Compliance for long-term monitoring Reducedcost-effectiveness in specificsubgroups Pitfalls of US screening • Sensitivity (15-20% of patients diagnosedoutside Milan) • Performance in obese individuals Lack of reliablebiomarkers for prediction/earlydetection

  8. Refinement of screening strategies Increasing surveillance rates Risk stratification Promotingeducation • Patients education • Practitionners training • Enlistment of primary care • providers Prediction biomakers Machine learning Low HCC risk Moderate/ High HCC risk Improvingcompliance Screening using Contrast-enhanced imaging Recommended semi-annual ultrasound • Systems-level interventions • Dedicatedclinicalpathways • Navigation programs • Mailedoutreach Earlydiagnosis biomakers Reviewed in Singal, Lampertico, Nahon. J Hepatology (in press)

  9. 407 patients withcirrhosis at « high HCC risk » (>5%/yr) Withboth US and MRI for surveillance Kim et al, Hepatology 2019 Cost-effectivenessstudy

  10. Cancer risk-basedmodels and surveillance: the example of lung cancer in the general population

  11. Personalisation of HCC screening Allocation of HCC risk classes Low HCC risk Intermediate HCC risk High HCC risk <1.5% 1.5 - 3% >3%

  12. Personalisation of HCC screening Allocation of HCC risk classes Low HCC risk Intermediate HCC risk High HCC risk <1.5% 1.5 - 3% >3% Decision • Reinforced US surveillance • Education programs • Mailedoutreach • Dedicatedclinicalpathway • Optimization of surveillance modality • Imaging (CT scan, MRI)? • Biomarkers for earlydetection? • shorter interval? Recommended US surveillance Or Dropping surveillance? COSTS

  13. Determinants and outcomes Insult Host factors Environmental factors Liverinjury Virus Metabolic syndrome Alcohol Histologicalfeatures Gender Age Ethnicity Genetics ? Extra-hepatic mortality Liver-related mortality HCC ?

  14. A « global » annual incidence rangingfrom 1.5% to 3% in cirrhosis in 2019* HBV (n=528) Alcohol(n=652) Papatheodiris, Hepatology 2017 2.9% Ganne-Carrié, J Hepatology 2018 HCV (n=1372) NASH (n=7068) 18.1% Nahon, Gastroenterology 2017 2.7% 6.7% Ioannou, J Hepatology 2019 *Based on European multicentre prospective cohorts of patients included in surveillance programs

  15. HCC-risk assessment models REACH-B score 17-point scoring system: - Male: 2 - Per 5-year increaseabove 35: 1 - ALAT >15: 1; ALT >45: 2 - HBeAg (+): 2 - DNA >4 log: 3; >5 log: 5 Cumulative risk score and associated 5-year risk of developing HCC in patients with CHB Training set: 3584 patients (REVEAL cohort) Validation set: 1505 patients (Hong Kong and Korea) Yang HI, et al. Lancet Oncol 2011;12:568–74. REACH-B, Risk Estimation for HCC in Chronic Hepatitis B; ALAT, alanine aminotransferase; ALT, alanine transaminase

  16. HBV-controlledcaucasianswithcirrhosis? PAGE-B Older men withcirrhosis ! Brichler et al, JVH 2018 Papatheodiris J Hepatol 2016

  17. HCC riskmodels in non-viral cirrhosis Ioannou, J Hepatology 2019

  18. CirVir CO12 HCV: Can we“predict” HCC risk at the individuallevel? Score ≤5 6 ≤ score ≤10 11 ≤score ≤14 Score >14 0.60 P<0.0001 100 0.50 90 80 0.40 1-year risk 3-year risk 5-year risk 70 Cumulative incidence of HCC (HCV) 0.30 60 HCC risk (%) 0.20 50 40 0.10 30 0.00 20 12 24 36 48 60 72 84 0 10 Time since inclusion (months) 0 • Age >50 years • Alcohol • GGT >N • Plat <100 103 • SVR Score ≤5: low Score 6–10: intermediate Score 11–14: high Score >14: maximal Riskmodelling Ganne-Carrié et al, Hepatology 2016

  19. Specific HCC risk factors in patients with SVR? Influence of metabolic syndrome according to SVR status 1000 SVR patients followed 5.7 yrs: 842 cirrhotics,158 bridgingfibrosis Van der Meer AJ, et al. J Hepatol 2016 Nahon P, et al. Gastroenterology 2017

  20. How can weimproverisk stratification? • Limits of conventional analytic approaches (Cox models) • Useful to quantify the relative importance of independent predictors (HRs) • Unfit to deal with highly heterogeneous populations and to detect specific relationships in specific subgroups • Decision-tree based approaches using Machine-learning • Effective for modeling complex relationships between correlated variables • Automatic detection of optimal thresholds • High illustrative value

  21. CirVir CO12 Identifying residual risk of HCC following HCV eradication in compensated cirrhosis:Machine learningapproaches (decisiontreeanalysis) N=836 JCO, in revision

  22. Forest plots (or variable importance plots): hierachisation of riskfactorstakingintoaccounttheir interactions and internal validation in an ensemble of 1000 trees (stability, robustness)

  23. A. HCC cumulative incidence: Cox proportional hazards model B. Calibration plot: Cox proportional hazards model C 1.00 C H f o e c n e d i c 0.75 n i e v i t a l u m u C 0.50 0.25 0.00 0 12 24 36 48 60 Time (months) High predicted risk Moderate predicted risk Low predicted risk C 1.00 C H f o e c n e d i c 0.75 n i e v i t a l u m u C 0.50 0.25 0.00 0 12 24 36 48 60 Time (months) High predicted risk Moderate predicted risk Low predicted risk External validation and calibration of models are essential* 100% Observed HCC status 75% 50% 25% Cox model 0% 0% 25% 50% 75% 100% Predicted HCC probability C-Index=0.66 C-Index=0.74 Machine learning • Random survival forest (RSF) combining 1000 decision trees E. HCC cumulative incidence: random survival forest F. Calibration plot: random survival forest 100% Observed HCC status *Validation in the CO22 Hepather cohort (Carrat F et al, Lancet 2019, n=668 patients with HCV-relatedcirrhosis) 75% 50% 25% 0% Predicted HCC probability 0% 25% 50% 75% 100%

  24. Non invasive biomarkers for HCC risk stratification (and earlydiagnosis) Identify subroupsat differentrisks Facilitate HCC early detection Improvement of staging and prediction of treatmentresponse Precancerous Focal lesion Liquidbiopsy Cirrhosis HCC

  25. HALT-C cohort Gastroenterology 2011 SNP + clinical data • N = 816 • Follow-up: 6,1 yrs • HCC=66 Hepatology 2005

  26. Integration of genetic data into HCC-riskassessmentmodels: whichincremental value? • ♀=95% • BMI=24 kg/m2 • Diabetes=13,5% • PNPLA3(GG)=5% Age + Gender + BMI + Diabetes PNPLA3 + Clinical factors 1.0 Patients sans CHC (%) 0.8 True positive fraction 0.6 PNPLA3 (rs738409 C>G) 0.4 • ♂=100% • BMI=31,6 kg/m2 • Diabetes=56,1% • PNPLA3(GG)=22,5% False positive fraction 0.2 Guyot et al, J Hepatol 2013 0.0 1.0 0.6 0.8 0.36 0.4 0.0 0.2

  27. Refinement of riskprediction by reclassification of individuals Manolio T, NEJM 2010

  28. Reviewed in Trépo, Romero, Zucman-Rossi, Nahon; J Hepatol 2016

  29. Towardsindividualized HCC riskassessment: « user-friendly » interface for decision-making process

  30. SNPs + clinical data

  31. Conclusions et perspectives (1) • HCC incidence tends to begloballysimilar in non-viral and viral cirrhosisfollowing HBV control/HCV eradication • HCC riskfactors in these patients includevariousfeaturesrelated to 1) hostcharacteristics, 2) environmentalfactors, 3) liver tests impairment • The incremental values of circulatingbiomarkers(genetic variants) to improve HCC riskassessmentremains to bedemonstrated • Combiningthese simple routine parametersusingclassicallogisticregressionis able to stratifycirrhotic patients into distinct HCC risk classes but onlyprovides information on average global effects

  32. Conclusions et perspectives (2) • Machine learningapproaches enable : • more effective combinations between HCC risk factors by better accounting for patient’s complexity • the identification of unexpected “extreme phenotypes” • High illustrative value of the long course of cirrhosis • HCC risk stratification willform the basis for future trialsexploring: • Refinement of surveillance modalities • The identification of new biomarkersuseful for HCC prediction/earlydiagnosis/classification • Preventionstrategies • Cost-effectivenessstrategies in HCC management

More Related