Arevir Meeting, Bonn, April 23, 2009 M. Zazzi on behalf of the EuResist Network (www.eurestist.org)
EuResist status � Funded by the EU JAN-06 to JUN-08, then set as a European Network (legal entity) � Data collected from ~30,000 patients (Italy, Germany, Sweden, Luxembourg, Belgium, Spain) � Data modeling by IBM Israel, Max Planck Institute for Informatics, Informa & Rome TRE University � Several methods investigated, much effort on feature selection and derivation � DB still expanding, models being updated and refined
EuResist – TCE definition CD4 Viral load Treatment switch Genotype Viral load time 0 to 12 weeks Short-term model: 4-12 weeks Pre-therapy HIV RNA Patient demographics (age, gender, race, route of infection) Past AIDS diagnosis Past treatments Past genotypes
EuResist – labeling therapies Baseline data HIV genotype at 0 to 12 weeks before treatment VL at 0 to 12 weeks before treatment Additional variables when available Treatment switch VL at 4 to 12 weeks (8-week outcome) SUCCESS FAILURE Undetectable or >2 log Detectable and not >2 decrease VL log decrease VL
The data funnel…
EuResist – engines � Three prediction engines developed independently � Generative-Discriminative (by IBM) � Evolutionary (by Max-Planck Institute) � Mixed effects (by Rome TRE & Informa) � Then, combined into a unique engine and made freely available on the web
Data-driven systems –Eu Resist Generative/Discriminative engine Model response to treatment in the absence of genotype with a Bayesian network For any defined regimen, compute a probability of success (Generative step) Use the probability as an additional feature for logistic regression together with genotype and other covariates (Discriminative step)
Data-driven systems –Eu Resist Evolutionary engine Model HIV evolution under therapy from longitudinal and cross-sectional sequence data For any defined genotype, neighbor mutants can be computed in silico and the contribution of the expected mutants to resistance can be calculated Functions weight for probability and expected time for mutants to occur Probability to remain susceptible to a drug (below a defined phenotypic threshold) Altmann et al, AVT 2007 GENETIC BARRIER
Data-driven systems –Eu Resist Mixed-Effects engine Focuses on interactions among variables � • drug x drug • drug x drug x drug • drug x mutation • drug x previous drug class exposure • drug x previous drug exposure • mutation x mutation
Data-driven systems –Eu Resist Combined engine � The combination (mean) of the engines performs equal to or better than the individual engines � The combined engine learns faster, i. e. it is more accurate when trained on limited data sets Altmann et al, PLoS ONE 2008
The EuResist combined engine 3143 therapies, Short-term outcome (8 weeks)
EuResist vs. Expert interpretation (EVE study) Form the invitation letter: The requested response include a categorical (C) answer and a quantitative (Q) estimate: C) Given this HIV genotype and patient information, will the indicated therapy be successful (i. e. will it make HIV RNA decrease by at least 2 logs or to undetectable levels in 8 weeks) ? Q) Given this HIV genotype and patient information, what probability of success would you estimate for the indicated therapy?
EuResist vs. Expert interpretation (EVE study)
Distribution of the 25 therapies by year and type 5 4 No. of 3 cases 2 1 1998 1999 2000 2001 2002 2003 2004 2005 2006 Year of therapy NRTI-only therapy NNRTI-based therapy *PI-based therapy *17 cases with boosted PI, 2 ATV, 1 NFV
Patient characteristics FEATURE MEDIAN (IQR) Baseline log viral load 4.67 (4.38-4.99) Baseline CD4 counts 298 (134-412) Number of previous treatment lines 5 (3-6) Number of NRTI mutations at baseline 3 (3-4) Number of NNRTI mutations at baseline 1 (0-2) Number of PI mutations at baseline 2 (0-3) Number of available previous viral load data 15 (8-25) Number of available previous CD4 counts 14 (10-30) Number of available previous genotypes 1 (0-3)
EuResist vs. Expert interpretation (EVE study) 25 HAART cases randomly selected form the EuResist db: • Obsolete therapies excluded • Wild type genotype excluded • All clinical and virological information available 12 experts enrolled, response obtained from 10: • On ‐ line anonymous rating • Only European (E) vs. non ‐ European (N) setting traceable • Use of any interpretation system allowed (and declared)
AUC 95% CI Best_expert 0.853 0.655 - 0.961 euresist 0.787 0.578 - 0.923 P = 0.011 mean_expert 0.777 0.567 - 0.917 Worst_expert 0.653 0.438 - 0.830
EuResist vs. Expert interpretation (EVE study) Correlation between EuResist and mean(expert) probability of success
EuResist vs. Expert interpretation (EVE study) 50 +1.96 SD 40 EuResist - mean(expert) 38.9 30 20 Mean 10 7.0 0 -10 -20 -1.96 SD -25.0 -30 0 20 40 60 80 100 AVERAGE of EuResist and mean(expert) xxx
No. of errors vs. no. of interpretation systems used
How confident are you in your prediction? Average absolute difference between the predicted probability of success and the cut-off value (50%)
EuResist vs. Expert interpretation (EVE study)
EuResist vs. Expert interpretation (EVE study) EuResist & most experts incorrect EXPERT 1 EXPERT 3 EXPERT 4 EXPERT 5 EXPERT 6 EXPERT 7 EXPERT 8 EXPERT 9 EXPERT 10 EXPERT 11 EuResist ACTUAL F S F F S F F F F F F S S S S S S S S S F F S F F F F S S F F F F S F OUTCOME Incorrect prediction Treatment failure F S Treatment success
EuResist vs. Expert interpretation (EVE study) Unexpected drug efficacy EXPERT 1 EXPERT 3 EXPERT 4 EXPERT 5 EXPERT 6 EXPERT 7 EXPERT 8 EXPERT 9 EXPERT 10 EXPERT 11 EuResist ACTUAL F S F F S F F F F F F S S S S S S S S S F F S F F F F S S F F F F S F OUTCOME Incorrect prediction Treatment failure F S Treatment success
Case #12843 (patient 17363) 1 (27/07/2001: ABC, D4T, LPV/r ) a � Past treatment lines � AZT � DDI � AZT DDC � DDC SQV 3TC � D4T IDV D4T � EFV RTV � Nadir CD4: unknown � Zenith VL: 72,300
Case #12843 (patient 17363) 1 (27/07/2001: ABC, D4T, LPV/r ) b � Genotype � L10I M36I G48V I54V L63P A71V T74S V77I V82A L90M I93L � D67N T69D K70R K103N V118I G190A T215C K219Q G333E � Past drug resistance mutations � unknown � Baseline VL: 72,300 � Follow-up VL: 314 (-2.36 log)
Case #12843 (patient 17363) 1 (27/07/2001: ABC, D4T, LPV/r ) c � Treatment more effective than expected � T215C revertant? � Transient success? � Patient lost to follow-up � Definition of success
Case #14503 (patient 19816) 2 (05/10/2001: D4T, EFV, LPV/r ) a � Past treatment lines � AZT DDC � 3TC AZT � 3TC D4T IDV � DDI NVP SQV/rtv � D4T DDI LPV/rtv � Nadir CD4: 8 � Zenith VL: 794,328
Case #14503 (patient 19816) 2 (05/10/2001: D4T, EFV, LPV/r ) b � Genotype � L10I G48V I54V Q58E L63P A71V V77I V82C I84V � M41L D67N L74V K101N V118I Y181C L210W T215C K219E � Past drug resistance mutations � unknown � Baseline VL: 794,328 � Follow-up VL: 1,000 (-2.90 log)
Case #14503 (patient 19816) 2 (05/10/2001: D4T, EFV, LPV/r ) c � Treatment more effective than expected � T215C revertant? � V82C not resistant to LPV? � Transient success? � EFV with Y181C? � Later VL rebound to 15,900 � Definition of success
EuResist vs. Expert interpretation (EVE study) Adherence issues? EXPERT 1 EXPERT 3 EXPERT 4 EXPERT 5 EXPERT 6 EXPERT 7 EXPERT 8 EXPERT 9 EXPERT 10 EXPERT 11 EuResist ACTUAL F S F F S F F F F F F S S S S S S S S S F F S F F F F S S F F F F S F OUTCOME Incorrect prediction Treatment failure F S Treatment success
Case #25745 (patient 9492) 3 (14/04/2005: 3TC, TDF, ATV/r ) a � Past treatment lines � AZT � AZT DDC � 3TC AZT IDV � 3TC AZT NVP � Nadir CD4: 289 � Zenith VL: 18,000
Case #25745 (patient 9492) 3 (14/04/2005: 3TC, TDF, ATV/r ) b � Genotype � L63P I93L � M41L E44A D67G L74I V118I V179I Y181I M184V L210W T215Y K219D � Past drug resistance mutations � L10I A71V I84V � Y181C � Baseline VL: 18,000 � Follow-up VL: 21,000
Case #25745 (patient 9492) 3 (14/04/2005: 3TC, TDF, ATV/r ) c � Poor adherence? � But 3 years with undetectable VL in the past � Underestimated resistance? � Impact of past I84V on ATV � L74V as a proxy of hidden K65R impacting TDF
Case #43708 (patient 8477) 4 (22/04/2005: AZT, EFV, ATV/r ) a � Past treatment lines � ABC DDI EFV � 3TC TDF EFV � LPV/r � TDF EFV NFV � AZT EFV NFV � 3TC TDF EFV � Nadir CD4: 11 � Zenith VL: 500,000
Case #43708 (patient 8477) 4 (22/04/2005: AZT, EFV, ATV/r ) b � Genotype � M36I T74S I93L � K65R L74V V90I Y115F M184V G190Q K219N � Past drug resistance mutations � Same as at last time point � Baseline VL: 114,370 � Follow-up VL: 3,816 (-1.48 log) � But earlier 97 (-3.07 log)
Case #43708 (patient 8477) 4 (22/04/2005: AZT, EFV, ATV/r ) c � Limited adherence? � Three AZT hypersusceptibility mutations (K65R L74V M184V) � Transient response � Underestimated resistance? � G190Q impact on EFV?
Recommend
More recommend
