Using the animal model framework to explore translation of safety pharmacology cardiovascular data...

1
on the nervous and GI systems account for nearly 50% serious drug adverse reactions during clinical development. Consequently, there is a real pre-clinical need to accurately predict the potential for compounds to affect the cardiovascular system as well as other organ systems. Following the success of the animal model framework approach in defining the sensitivity, specificity and predictability of pre-clinical cardiovascular data, the companies involved have begun to apply this approach to the in vivo assessment of nervous, respiratory, GI and renal function. Because there is little or no quantitative Phase I data for other organ systems, it is hard to correctly align pre-clinical and clinical parameters. This presentation will outline how the framework has been modified in order to facilitate translation of such data sets through the creation of functional equivalence tables. The presentation will also address the challenges of sourcing and categorising both pre-clinical and clinical data and highlight gaps in our predictive capacity. Ultimately, the presentation will ask how the safety pharmacology discipline can improve utilisation of pre-clinical data in Phase I study design to reduce compound attrition due to serious drug adverse reactions or toleration issues. doi:10.1016/j.vascn.2010.11.097 Poster Number: 94 Board Number: 33 Impact and prevalence of safety pharmacology-related toxicities throughout the pharmaceutical life cycle Will S. Redfern a , Russ Bialecki b , Lorna Ewart a , Tim G. Hammond a , Lew Kinter b , Silvana Lindgren c , Chris E. Pollard a , Mike Rolf a , Jean-Pierre Valentin a a AstraZeneca R&D, Macclesfield, United Kingdom b AstraZeneca R&D, Wilmington, DE, United States c AstraZeneca R&D, Sdertlje, Sweden Prioritisation of safety pharmacology resources according to impact and prevalence of toxicities/adverse drug reactions (ADRs) is often based on opinion, anecdotal examples, recent issues with specific projects, and regulatory focus. There is a need for evidence- based decision-making, which requires reliable data on impact and prevalence of drug-induced toxicities across a range of organ functions. We have collated information from published reviews to provide a rough guide to this across the organ functions falling within the domain of safety pharmacology: cardiovascular (CVS), nervous system (NS), respiratory (RESP), gastrointestinal (GI) and renal (REN). Some datasets relate to frequency of candidate/marketed drugs associated with the toxicity; others contain data on prevalence of the toxicity in subjects/patients, as follows. 1. Preclinical attrition (based on 88 CDs stopped): CVS (27%)> NS (14%) > GI (3%) > REN/ RESP (2% each). 2. Phase I serious ADRs (based on 43 events in 1,015 subjects): NS (28%) > GI (23%) > CVS (9%) > REN (5%) > RESP (0%). 3. Marketed drugs -serious ADRs (based on 21,298 patients): NS (30%)>CVS (25%)>GI (14%)>RESP (8%) > REN (2%). 4. Marketed drugs -withdrawals (based on 47 drugs withdrawn 1975-2007): CVS (45%)>NS/GI/REN/RESP (2% each). These data require cautious interpretation, but are a first step to assessing frequency and impact of different drug-induced adverse effects throughout drug discovery, development and marketing. Sources: 1. ADD (2006) 1;53-65; 2. EJCP (1998) 54;13-20; 3. JAMA (2006) 296;1858-66; 4. DDT (2009) 14;162-67. doi:10.1016/j.vascn.2010.11.098 Poster Number: 95 Board Number: 34 Using the animal model framework to explore translation of safety pharmacology cardiovascular data to phase I clinical trial Rob Wallis a , Russell Bialecki b , Lorna Ewart c , Tim Hammond b , Philip Jarvis c , Derek Leishmann d , Silvana Lindgren c , Vicente Martinez e , Chris Pollard c , Will Redfern c , Jon Scatchard a , John Sherington a , Jean-Pierre Valentin c a Pfizer, Macclesfield, Cheshire, United Kingdom b AstraZeneca, Wilmington, DE, United States c AstraZeneca, Macclesfield, Cheshire, United Kingdom d Eli Lilly and Company, Indianapolis, IN, United States e Universidad Autonoma de Barcelona, Barceona, Spain Despite the wide use of animal models in the pharmaceutical industry, there are few examples that demonstrate how well these models predict outcome in man taking into account drug exposure across species. Understanding the translation of non-clinical models to clinical outcomes is critical to improving prediction in safety pharmacology and ultimately reducing late-stage attrition. A framework approach has been developed by AstraZeneca, Eli Lily and Pfizer to assess how well safety pharmacology models predict the outcome of Phase I clinical trials. The approach assesses the degree of confidence in the animal model and in the translation of data to man. The companies involved have generated data on a combined pool of 64 compounds that have been assessed in the conscious dog telemetry model for effects on haemodynamics and the ECG. Statistical measures of sensitivity (i.e. the proportion of compounds whose effects in man are correctly identified by model), specificity (i.e. the proportion of compounds without an effect in man, that is correctly identified by model) and predict- ability (i.e. a measure of how often a model will predict true positive or negative outcomes) have been calculated with the appropriate confidence intervals. This presentation will outline the animal model framework and will address data interpretation and its impact on decision making within the drug discovery and development process. doi:10.1016/j.vascn.2010.11.099 Poster Number: 96 Board Number: 35 Can retrospective analysis of preclinical cardiovascular safety data improve predictivity of future results? Pieter-Jan Guns a,c , Ard Teisman a , Karel Van Ammel a , Rob Towart a , Roel Straetemans b , Hidde Bult c , David Gallacher a a J&J EDC, Beerse, Belgium b J&J Biostatistics Department, Beerse, Belgium c University of Antwerp, Antwerp, Belgium Background: Drug-induced TdP is a major cause of failure of new drug candidates. To identify this liability, different surrogate biomarkers such as hERG inhibition, action potential prolongation and QT-prolongation are applied in different preclinical models. The current work represents the conclusions of a retrospective data warehouse-assisted comparison of different CVS models applied within J&J, on over 1000 compounds, including both internal J&J drug candidates and external reference drugs. Abstracts e29

Transcript of Using the animal model framework to explore translation of safety pharmacology cardiovascular data...

on the nervous and GI systems account for nearly 50% seriousdrug adverse reactions during clinical development. Consequently,there is a real pre-clinical need to accurately predict the potentialfor compounds to affect the cardiovascular system as well as otherorgan systems. Following the success of the animal model frameworkapproach in defining the sensitivity, specificity and predictabilityof pre-clinical cardiovascular data, the companies involved havebegun to apply this approach to the in vivo assessment of nervous,respiratory, GI and renal function. Because there is little or noquantitative Phase I data for other organ systems, it is hard tocorrectly align pre-clinical and clinical parameters. This presentationwill outline how the framework has been modified in order tofacilitate translation of such data sets through the creation offunctional equivalence tables. The presentation will also address thechallenges of sourcing and categorising both pre-clinical and clinicaldata and highlight gaps in our predictive capacity. Ultimately, thepresentation will ask how the safety pharmacology disciplinecan improve utilisation of pre-clinical data in Phase I study designto reduce compound attrition due to serious drug adverse reactionsor toleration issues.

doi:10.1016/j.vascn.2010.11.097

Poster Number: 94Board Number: 33

Impact and prevalence of safety pharmacology-related toxicitiesthroughout the pharmaceutical life cycleWill S. Redferna, Russ Bialeckib, Lorna Ewarta, Tim G. Hammonda,Lew Kinterb, Silvana Lindgrenc, Chris E. Pollarda, Mike Rolfa,Jean-Pierre Valentina

aAstraZeneca R&D, Macclesfield, United KingdombAstraZeneca R&D, Wilmington, DE, United StatescAstraZeneca R&D, Sdertlje, Sweden

Prioritisation of safety pharmacology resources according toimpact and prevalence of toxicities/adverse drug reactions (ADRs)is often based on opinion, anecdotal examples, recent issues withspecific projects, and regulatory focus. There is a need for evidence-based decision-making, which requires reliable data on impact andprevalence of drug-induced toxicities across a range of organfunctions. We have collated information from published reviewsto provide a rough guide to this across the organ functions fallingwithin the domain of safety pharmacology: cardiovascular (CVS),nervous system (NS), respiratory (RESP), gastrointestinal (GI) andrenal (REN). Some datasets relate to frequency of candidate/marketeddrugs associated with the toxicity; others contain data on prevalenceof the toxicity in subjects/patients, as follows. 1. Preclinical attrition(based on 88 CDs stopped): CVS (27%)>NS (14%)>GI (3%)>REN/RESP (2% each). 2. Phase I serious ADRs (based on 43 events in 1,015subjects): NS (28%)>GI (23%)>CVS (9%)>REN (5%) > RESP (0%).3. Marketed drugs -serious ADRs (based on 21,298 patients): NS(30%)>CVS (25%)>GI (14%)>RESP (8%) > REN (2%). 4. Marketeddrugs -withdrawals (based on 47 drugs withdrawn 1975-2007):CVS (45%)>NS/GI/REN/RESP (2% each). These data require cautiousinterpretation, but are a first step to assessing frequency and impactof different drug-induced adverse effects throughout drug discovery,development and marketing.

Sources: 1. ADD (2006) 1;53-65; 2. EJCP (1998) 54;13-20; 3. JAMA(2006) 296;1858-66; 4. DDT (2009) 14;162-67.

doi:10.1016/j.vascn.2010.11.098

Poster Number: 95Board Number: 34

Using the animalmodel framework to explore translation of safetypharmacology cardiovascular data to phase I clinical trialRob Wallisa, Russell Bialeckib, Lorna Ewartc, Tim Hammondb,Philip Jarvisc, Derek Leishmannd, Silvana Lindgrenc,Vicente Martineze, Chris Pollardc, Will Redfernc, Jon Scatcharda,John Sheringtona, Jean-Pierre Valentinc

aPfizer, Macclesfield, Cheshire, United KingdombAstraZeneca, Wilmington, DE, United StatescAstraZeneca, Macclesfield, Cheshire, United KingdomdEli Lilly and Company, Indianapolis, IN, United StateseUniversidad Autonoma de Barcelona, Barceona, Spain

Despite the wide use of animal models in the pharmaceuticalindustry, there are few examples that demonstrate how well thesemodels predict outcome in man taking into account drug exposureacross species. Understanding the translation of non-clinicalmodels to clinical outcomes is critical to improving prediction insafety pharmacology and ultimately reducing late-stage attrition.A framework approach has been developed by AstraZeneca, EliLily and Pfizer to assess how well safety pharmacology modelspredict the outcome of Phase I clinical trials. The approach assessesthe degree of confidence in the animal model and in the translationof data to man. The companies involved have generated data on acombined pool of 64 compounds that have been assessed in theconscious dog telemetry model for effects on haemodynamicsand the ECG. Statistical measures of sensitivity (i.e. the proportionof compounds whose effects in man are correctly identified bymodel), specificity (i.e. the proportion of compounds without aneffect in man, that is correctly identified by model) and predict-ability (i.e. a measure of how often a model will predict truepositive or negative outcomes) have been calculated with theappropriate confidence intervals. This presentation will outlinethe animal model framework and will address data interpretationand its impact on decision making within the drug discovery anddevelopment process.

doi:10.1016/j.vascn.2010.11.099

Poster Number: 96Board Number: 35

Can retrospective analysis of preclinical cardiovascular safety dataimprove predictivity of future results?Pieter-Jan Gunsa,c, Ard Teismana, Karel Van Ammela, Rob Towarta,Roel Straetemansb, Hidde Bultc, David Gallachera

aJ&J EDC, Beerse, BelgiumbJ&J Biostatistics Department, Beerse, BelgiumcUniversity of Antwerp, Antwerp, Belgium

Background: Drug-induced TdP is a major cause of failure of newdrug candidates. To identify this liability, different surrogatebiomarkers such as hERG inhibition, action potential prolongationand QT-prolongation are applied in different preclinical models.The current work represents the conclusions of a retrospective datawarehouse-assisted comparison of different CVS models appliedwithin J&J, on over 1000 compounds, including both internal J&J drugcandidates and external reference drugs.

Abstracts e29