Objectives & Agenda - psiweb.org

Objectives & Agenda - psiweb.org

From Molecule to Medicine: a Case Study in how a Statistician can Provide Strategic Input to Drug Development Nelson Kinnersley, PhD Director of Biostatistics & Site Head, Roche Products Ltd, UK PSI Conference, London, 14-17 May 2017 Objectives & Agenda Objective: Use Case Study to propose suggestions for how statisticians can be better equipped to contribute to the cross-functional teams developing new medicines Agenda Definitions - What is Strategic Drug Development? Case study (anonymised) Offer suggestions for statisticians to consider when working-up a Clinical Development Plan Q&A 2 Definitions What is Strategic Drug Development? Strategy: No single consensus some common themes Not just for senior leaders for everyone Frameworks can help guide you:

e.g. Behaviours Drug development Not just for medics for everyone Frameworks can help guide you: Classical/vanilla (Ph 1 Ph 2 Ph 3) Non-classical (every other variation!) Statistical approaches e.g. quantify uncertainty operating characteristics of trial design(s) Analyse data, make inferences, Oh, and I think we can agree on one thing 3 Objectives & Agenda Objective: Use Case Study to propose suggestions for how statisticians can be better equipped to contribute to the cross-functional teams developing new medicines Agenda Definitions - What is Strategic Drug Development? Case study (anonymised) Offer suggestions for statisticians to consider when working-up a Clinical Development Plan Q&A 4 Strategy Definitions (1) Schoemaker et al (2013), HBR 6 skills identified from work with 20,000 executives:

5 Strategy Definitions (2) Bowman (2016), HBR 4 key abilities: 1. Know: Observe and Seek Trends 2. Think: Ask the Tough Questions 3. Speak: Sound Strategic 4. Act: Make Time for Thinking and Embrace Conflict (General) examples of Tactical (T) vs Strategic (S): 1. Know: T: Hire statistician today with post-grad in Med Stats S: Identify data science skills needed in 2021 2. Think: Ask the Tough Questions T: What difference in survival is meaningful? S: Given likely approved drugs in 2023, what endpoints will be meaningful to Regulators, HTAs, patients? 3. Speak: Sound Strategic T: 3-page Stats Methods email attachment S: Concise bullet with how team can win (follow-up with tactical details) 4. Act: Make Time for Thinking and Embrace Conflict T: Attending every meeting to show presence

S: Block calendar time; develop/test/refine options 6 Strategy Definitions (3) Mindtools (2017) Amongst the simplest definitions (Ive encountered): "Determining how we will win in the period ahead." In business, different levels of strategy: Corporate strategy focuses on the organization as a whole Business unit strategy focuses on an individual business unit or market. Team strategy identifies how a team will help the organization meet its overall goals and objectives. 7 Summary What is Strategic Drug Development? No consensus Working definition: Determining how we will be successful in the period ahead Behaviours: Anticipate, Challenge, Interpret, Decide, Align, Learn 8 Objectives & Agenda Objective: Use Case Study to propose suggestions for how

statisticians can be better equipped to contribute to the cross-functional teams developing new medicines Agenda Definitions - What is Strategic Drug Development? Case study (anonymised) Offer suggestions for statisticians to consider when working-up a Clinical Development Plan Q&A 9 Case Study (Anonymised) For simplification, consider only USA & EU Scenario (2017): Oncology, Disease X Patients with symptoms present; Physician diagnosis (CT, MRI) First Line (1L) treatment options: D D DD Approved: Drugs D1, D2, D3 (USA & EU) 1 2 23 SoC: Drug D1 in USA; Drug D2 inDEU D 1 2 Primary endpoint: Progression Free Survival (PFS) Project Team Leader:

What Ph 3 sample size do we need to get (our) Drug E E 10 One approach What Ph 3 sample size do we need to get Drug E approved? Ph 3, two-arm RCT E D E vs D1 1 Use public domain (e.g. clinicaltrials.gov) Median PFS on D1=40mos (1212d) Assume Exponential dist Suppose target Hazard Ratio (HR)=0.75; alpha=0.05; 80% power* To see 380 PFS events requires N=500; Trial duration ~77 to 84 months** i.e. read-out 2024 File Regulatory Approval 2025, HTA 2026-27 Patient access ~9 years from now Project Team Leader, either: 1. Thank you; job done (& lets hope D1 becomes SoC in EU) 2. Faster please * For sake of generalisation; ** recruitment assumptions 11 One approachcontinued

What Ph 3 sample size do we need to get Drug E approved? E Ph 3, three-arm RCT E vs D1; E vs D2 D D 1 2 Extend previous approach Use public domain for D2 Type I error control; specify when each comparison performed (& study integrity maintained); etc etc Patient access ?? >9 years from now Project Team Leader, either: 1. Thank you; job done 2. Faster please 12 Schematic representation 2016 2017 2018 2019

2020 2021 2022 Approved 1L Weak chemo versus D2 E versus D1 2024 2025 Primary Analysis Reg & HTAs Weak chemo versus D1 2023 E versus D1 & D2 Can we do better? Probably!

13 Objectives & Agenda Objective: Use Case Study to propose suggestions for how statisticians can be better equipped to contribute to the cross-functional teams developing new medicines Agenda Definitions - What is Strategic Drug Development? Case study (anonymised) Offer suggestions for statisticians to consider when working-up a Clinical Development Plan Q&A 14 Returning to the Schematic representation 2016 2017 2018 2019 2020 2021 2022

Approved 2023 2025 Primary Analysis Weak chemo versus D2 E versus D1 Reg & HTAs Weak chemo versus D1 1L 2024 E versus D1 & D2 Can we do better? Probably! Trial optimisation

What is the estimand? Target delta - justification for HR=0.75 ? (see e.g. Cook et al (2015)) Interim analysis (alternative boundaries?); not too early (Type I and II error) ! Alternative comparator? D3? lower hurdlebut not SoC etc Many of these topics covered at this Conference great ! my focus is looking at wider scope 15 Example of trials Strategic Approach 2016 2017 2018 2020 2021 Interim Analysis 2022 2023 Primary PFS

2024 2025 OS Analysis Weak chemo versus D2 (n = 750) X1 vs D1 (Ph 2, n=130) X1 vs D1 (Ph 3, n=350) Reg & HTAs 1L Weak chemo versus D1 (n = 810) 2019 X2 vs D2 (Ph 3, n=450) X3 (Ph 1, n=150)

What might this tell us? If successful, X1 & X2 approved before (our) E Unclear which of X1, X2 approved first: What if Stunning Ph 2 data for X1? ? submission 4Q2017 (for Acc./Cond. Approval) ? Data package for HTA submission Why didnt X1 have IA? What does this tell us lhood Ph 2 success for X1? Why 2 Interims for X2 vs D2? Futility then Efficacy? How many PFS events? Sizeable Ph 1 trial for X3 Signal seeking? Expansion cohorts? What could lead to PRIME / BTD status request? Use of Historical data? 16 Alternative approach What Ph 3 sample size do we need to get Drug E approved? Consider what will meet Reg, HTA and patient needs at (planned) Filing: Will D1 and D2 still be SoC? What will be meaningful clinical benefit? What evidence do we have already? How could we use that evidence to manage (gate) risk in large Ph 3?

Understand/Interpret Estimand & Target Product Profile (TPP) see later As before: Use public domain for input data (summary or patient-level) Challenge/verify assumptions: simulations (recruitment, PFS events, conditional probabilities) Align & Decide at Milestones: Internal: present your work to Project Team, Review Committees External: HA/HTA meetings; Filing; Outcomes / benefits: 1. Integral to the team (vs service provider) 2. Improved understanding of uncertainty / risks (at all levels) 3. No guarantees shorter timelines, but arguably better decision-making along-theway 4. Stronger sense of ownership in the outcome 17 Possible extensions e.g. what do Later Line (2L+) trials tell us?

2016 2017 2018 Weak chemo versus D2 (n = 750) X1 vs D1 (Ph 2, n=130) X1 vs D1 (Ph 3, n=350) 2020 2021 Interim Analysis 2022 2023 Primary PFS 2024 2025

OS Analysis Reg & HTAs 1L Weak chemo versus D1 (n = 810) 2019 X2 vs D2 (2L, Ph 3, n=230) X3 vs D1 (3L, Ph 2, n=120) Reg & HTAs 2L+ X2 vs D2 (Ph 3, n=450) X3 (Ph 1, n=150) Patient characteristics

Potential Cure Fraction in 1L? Incorporate that when translating 2L+ to 1L Efficacy X2 +ve PFS in 2L, awaiting OS& they started Ph 3 in 1L. ??Increase lhood X2 becomes SoC in 1L too? Assuming PFS & OS shorter in 2L+, quantify relationship (if any) PFS & OS Prediction PFS, OS, recruitment HR=0.75 too? 18 Possible extensions e.g. what do Later Line (2L+) trials tell us? 2016 2017 2018 2019 2020 Weak chemo versus D2 (n = 750) X1 vs D1 (Ph 2, n=130) X1 vs D1 (Ph 3, n=350) Interim

Analysis 2022 2023 Primary PFS 2024 2025 OS Analysis Reg & HTAs 1L Weak chemo versus D1 (n = 810) 2021 X2 vs D2 (2L, Ph 3, n=230) X3 vs D1 (3L, Ph 2, n=120) Reg & HTAs

2L+ X2 vs D2 (Ph 3, n=450) X3 (Ph 1, n=150) Additional uncertainty Why (& how) such a quick 3L trial for X3? Looks like it started while 1L Ph 1 started early signal in 1L? Biomarker subgroup? How might subsequent 1L and 2L+ look for X3? 19 Possible extensions e.g. what do Later Line (2L+) trials tell us? 2016 2017 2018 2019 2020 2021

Primary PFS Weak chemo versus D2 (n = 750) X1 vs D1 (Ph 2, n=130) X1 vs D1 (Ph 3, n=350) X2 vs D2 (Ph 3, n=450) X3 (Ph 1, n=150) 2024 2025 OS Analysis Biomarker +ve, X3 vs D1 (1L, Ph 3, n=300) X2 vs D2 (2L, Ph 3, n=230) X3 vs D1 (3L, Ph 2, n=120)

Biomarker +ve, X3 vs D1 (3L, Ph 3, n=180) Hypothetical Hypothetical Reg & HTAs 2L+ Interim Analysis 2023 Reg & HTAs 1L Weak chemo versus D1 (n = 810) 2022 Patient characteristics Will success of X3 drive segmentation of treatment? Potential first Approvals & SoC 1L: X1 (Accel/Cond) or X2 (Full). ?X3 in biomarker +ve? 2L: X2 3L: X3 in biomarker +ve

20 Use of a Target Product Profile (TPP) See e.g. FDA (2007) draft Guidance Tyndall et al (2017) The TPP embodies the notion of beginning with the goal in mind. sponsor specifies the labeling concepts that are the goals of the drug development program, documents the specific studies intended to support the labeling concepts, and then uses the TPP to assist in a constructive dialogue with the FDA. The ideal version of what the sponsor would like to claim in labeling guides the design, conduct, and analysis of clinical trials to maximize the efficiency of the development program. Ideally, the final version of the TPP will be similar to the USPIapplication & annotated draft labeling submitted withExample a new drug or SmPC biologics license application. 21 Generic TPP (1L patient population) Indication Efficacy

Safety (initial) Dosing Optimistic Target Previously untreated patients with Disease X Treat until disease progression or intolerance Minimal 22 Generic TPP (1L patient population) Indication Efficacy Safety (initial) Dosing Optimistic

Safety and tolerability similar to D1 Target Minimal Previously untreated patients with Disease X Safety and tolerability similar to D1 Treat until disease progression or intolerance Side effects are manageable and reversible 23 Generic TPP (1L patient population) Indication Optimistic Efficacy Safety (initial)

Efficacy 75% better than D1 (HR 0.57) Primary: PFS 40 70 months Secondary: OS, PRO Safety and tolerability similar to D1 Efficacy 50% better than D1 (HR 0.67) Previously untreated Target patients with Disease X Primary: PFS 40 60 months Secondary: OS, PRO Safety and tolerability similar to D1 Minimal Efficacy 33% better than D1 (HR 0.75) Primary: PFS 40 53 months Secondary: OS, PRO Dosing Treat until disease

progression or intolerance Side effects are manageable and reversible Opportunity to use Estimands framework? 24 Recap Determining how we will be successful in the period ahead Used internal & external information Target Product Profile Clinicaltrials.gov, clinicalstudydatarequest.c om, publications, Assessment reports (EPAR, FDA, NICE,) Posed hypotheticals What if I were in competitors shoes? What does this tell me about HA and HTA perspective? Generated alternative scenarios Sample size calcs, simulations

25 Some potential next steps Quantify uncertainty Likely treatment effects Timing of readouts Probability (Technical) Success at readout (PTS) Prevalence of biomarker subgroups etc etc Convey & influence Who is decision-maker? What format will be most effective? Evolve Follow literature (eAlerts, HA/ HTA decisions,) Incorporate new readouts & share 26 Leading toPossible Alternative Comparator D1 (required by TPP) Also, increased confidence D1 will become SoC PFS HR=0.67 (TPP) smaller trial

Analyses Futility (manage risk) Efficacy Interim (upside) PFS & OS (support HA & HTAs) Timelines NO GUARANTEE QUICKER! but Scenario planning allows iteration to balance trade-offs X1 and X2 may be First in Class approval but E might be first if Interim Analysis successful Success? Prob Success quantifies each of these outcomes Can be updated as internal & external data arises 27 Objectives & Agenda Objective: Use Case Study to propose suggestions for how statisticians can be better equipped to contribute to the cross-functional teams developing new medicines Agenda Definitions - What is Strategic Drug Development? Case study (anonymised) Offer suggestions for statisticians to consider when working-up a Clinical Development Plan

Summary and Q&A 28 Summary (or, What might I do differently?) Consider taking survey in Schoemaker et al (2013): think about the work you have done over the past year related to developing new strategies, solving business challenges, and making complex decisions. Average your scores for each of the six leadership skills and then address your weakest area first, following the recommendations described in this article. What does success look like for range of stakeholders? Not just person who asked you the (study design) question! 29 Selected Literature Strategy definitions: Bowman, N (2016). 4 Ways to Improve Your Strategic Thinking Skills https://hbr.org/2016/12/4-ways-to-improve-your-strategic-thinking-skills Mindtools (2017) https://www.mindtools.com/pages/article/what-is-strategy.htm Schoemaker, P.J., Krupp, S. and Howland, S. (2013). Strategic leadership: The essential

skills. Harvard Business Review, 91(1), pp.131-134. Target Product Profile / Effect size: Cook, J.A., Hislop, J., Altman, D.G., Fayers, P., Briggs, A.H., Ramsay, C.R., Norrie, J.D., Harvey, I.M., Buckley, B., Fergusson, D., Ford, I. & Vale, L.D. (2015) Specifying the target difference in the primary outcome for a randomised controlled trial: guidance for researchers. Trials, 16 (1), p.12. FDA draft Guidance (2007) https://neuroscienceblueprint.nih.gov/resources/target-product-profile.htm Tyndall, A., Du, W. & Breder, C.D. (2017) Regulatory watch: The target product profile as a tool for regulatory communication: advantageous but underused. Nature Reviews Drug Discovery. 16 (3), 156156. Probability of Success: Frewer, P., Mitchell, P., Watkins, C. & Matcham, J. (2016) Decision-making in early clinical drug development. Pharmaceutical Statistics. 15 (3), 255263. Rufibach, K., Burger, H.U. & Abt, M. (2016) Bayesian predictive power: choice of prior and some recommendations for its use as probability of success in drug development. Pharmaceutical Statistics. 15 (5), 438446. Rufibach, K., Jordan, P. & Abt, M. (2016) Sequentially updating the likelihood of success of a Phase 3 pivotal time-to-event trial based on interim analyses or external information. Journal of Biopharmaceutical Statistics. 26 (2), 191201. 30 What does success look like for a range of

stakeholders? Not just person who asked you the (study design) question! [email protected] https://www.linkedin.com/in/nelson-kinnersley3b31435 31

Recently Viewed Presentations

  • Subcultures - Sociology at Girton

    Subcultures - Sociology at Girton

    A youth subculture is a youth-based subculture with distinct styles, behaviors, and interests. Youth subcultures offer participants an identity outside of that ascribed by social institutions such as family, work, home and school. Youth subcultures that show a systematic hostility...
  • Program Evaluation Webinar Series Part 1

    Program Evaluation Webinar Series Part 1

    A key point about our framework is that those early steps are an underlining guidance for all of the things that we do in the evaluation. And particularly this idea of stakeholders and focus end up influencing how we make...
  • Sedimentology (ESCI 332) Clastics: Dr. John Anderson and

    Sedimentology (ESCI 332) Clastics: Dr. John Anderson and

    Walther's Law (1894): •Johannes Walther pioneered the study of modern environments as a tool for interpreting sedimentary strata. The various deposits of the same facies area and similarly the sum of the rocks of different facies areas were formed beside...
  • Chapter 16 CHAPTER 16 LINKS TO OBJECTIVES  Create

    Chapter 16 CHAPTER 16 LINKS TO OBJECTIVES Create

    Times New Roman Arial Arial Narrow Georgia Comic Sans MS Tahoma Arial Black Default Design Custom Design Slide 1 Signature: Microsoft Word 2003 Performance Objectives Creating Footnotes and Endnotes Creating Footnotes and Endnotes…/2 Printing Footnotes and Endnotes Viewing and Editing...
  • Abraham And The Land Promised To His Descendants

    Abraham And The Land Promised To His Descendants

    Islam was established by Mohammed in Mecca. 638. Caliph Omar conquered Jerusalem. 691. Construction of the Dome of the Rock and Al-Aqsa mosque. Moslem presence established on the Temple Mount in Jerusalem. The Nation of Israel - Timeline. 1517.
  • Statute of the ICJ, Article 38

    Statute of the ICJ, Article 38

    Statute of the ICJ, Article 38. The Court, whose function is to decide in accordance with international law such disputes as are submitted to it, shall apply: a. international conventions, whether general or particular, establishing rules expressly recognized by the...
  • Symbols of Elements

    Symbols of Elements

    Atomic Mass. Listed on the periodic table. Gives the mass of "average" atom of each element . Average . atom based on all the isotopes and their abundance % Atomic mass is not a whole number. Na. 22.99
  • Mission

    Mission

    Extend Eclipse Plug-ins to support grid computing High Performance Computing and Monte Carlo Methods Host Institution - North Carolina A&T State University Dr. Yaohang Li, Primary Instructor Spring 2005, 10 students enrolled at NCAT from Departments of Computer Science, Mechanical...