Skip to main content

Part 8. Expected consequences on R&D and the future of MIDD

TL;DR​

  1. MIDD today still centers on in vivo clinical trials, which are the ultimate proof. In the future, in vivo trials plausibly become an instrument for validating models, while the evidence supporting the decisions of scientists, regulators, and doctors comes from simulations run with validated models.

  2. Working in silico reduces R&D time and cost, and fewer patients need to be recruited for RCTs.

  3. Faster adoption by the scientific community depends on a handful of challenges. Beyond the technical ones, which are more or less solved, regulators need to formalize their position on in silico clinical trials, and the assumption that limited knowledge blocks the use of models needs addressing.

Reduced time and cost​

The in silico clinical trial paradigm, grounded in the EM methodology, is a viable answer to the declining efficiency of therapeutic R&D. Multi-scale mechanistic models of selected diseases, applied to the VPs of interest, combine knowledge and data to account for the complexity of biology and illness.

Running in silico clinical trials is a time- and cost-effective way to explore the full space of potential targets and target combinations, so that the candidates taken into clinical development are more likely to show efficacy on clearly characterized responder subgroups. Back-of-the-envelope numbers make the case. Bloomberg estimated 1.3bn of spending on 78 clinical development programs by the four largest drugmakers in immuno-oncology (Bristol-Myers Squibb, Merck & Co, Roche, AstraZeneca), against more than five thousand possible combination pairs. Exploring all of them would cost roughly 100bn, about 0.5% of the US national debt. The same systematic exploration in silico would cost about $20m, covering the development of the necessary cancer models. That comparison understates the in silico figure, since real trials would still be needed to confirm the findings, but the savings remain substantial and the capital goes to the most beneficial candidates. If a cost-effective way exists to explore the whole space of immunotherapeutic combinations, we owe it to patients to use it. The same holds for Covid-19, with vaccines added to therapies.

Virtual patients also ease the problem of patient recruitment, in larger trials, in rare diseases, and in ethically difficult trial designs. Careful characterization of responders and of effect size reduces the number of subjects needed for the same statistical significance.

The cost of adopting this in silico paradigm is low next to the cost of a single late-stage failure in phase 2 or 3[1] [2]. Scannell and Bosley showed by simulation that a mere 0.1 absolute increase in the correlation between in vitro or in vivo test output and clinical outcomes in humans would improve R&D efficiency at least tenfold [3]. The cost/benefit ratio alone is a compelling reason to adopt it widely.

Perspectives​

The approach described here establishes in silico clinical trials as an attractive and viable way to advance clinical research in therapeutics, including the growing need to combine therapies. Proofs of concept span several therapeutic areas and several stages of the typical R&D process, and early success points toward wide acceptance. Faster adoption by the scientific community depends on a handful of challenges, the first of which is that regulators need to formalize their position on in silico clinical trials.

Stakeholder engagement​

Mathematical modeling has been advocated for pharmaceutical R&D since the early 2000s, and industry has been slow to adopt it. Hood and Perlmutter concluded that these approaches would revolutionize how new medicines are developed, in a more personalized framework, but asked who would lead the change and how industry would explore it[4]. Two examples showed computer modeling affecting the drug R&D pipeline: the discovery of the mode of action of EGFR inhibitors in cancer, and the use of predictive biosimulation in planning clinical studies[5] [6]. Building on that evidence, one of a series of PricewaterhouseCoopers reports analyzed the drug R&D process and argued for making research more predictive, with "virtualisation" through modeling and simulation supporting the "seismic shift" it considered necessary[7]. More recently, modeling and simulation was used to study efficacy, safety, and dosing schedules for rivaroxaban, a novel anticoagulant[8], and to find the best regimen for novel classes of anticancer drugs [9].

Interest in model-based drug development and quantitative systems pharmacology is growing, and both go beyond the classical PKPD modeling well established in the industry by adopting more mechanistic elements. Regulatory agencies have taken notice[10] [11] [12]. The US Food and Drug Administration (FDA) is moving toward accepting simulation-based evidence [13], and the European Medicines Agency (EMA) is expected to follow.

Pan-European initiatives are increasingly visible. After the European Commission-funded project "Avicenna: A Strategy for In Silico Clinical Trials" completed and published its Roadmap, discussions with the Commission led to the Avicenna Alliance for Predictive Medicine[14]. It brings together academia, through the Virtual Physiological Human Institute (VPHi[15]), and a wide representation of industry, to build on the project's outcome. At Commission level, interest shows in a specific call for projects (Horizon 2020 Health PM16, "In silico trials for developing and assessing biomedical products"). There is also a recognized need for expert views to inform policy frameworks where none currently exist, as these technologies reach the market. That is a key function of the Avicenna Alliance, working closely with the European Commission and connecting with similar activities in the USA and the Asia Pacific region to harmonize policy across territories ab initio as far as possible. These initiatives should eventually produce formal guidelines from the FDA, the European Commission, and the EMA, which will build confidence and accelerate adoption.

On the payer side, the European Network for Health Technology Assessment (EUnetHTA) has endorsed the EM as the only reliable methodology for exploring the relation between baseline risk and treatment effect, selecting patients to treat, and translating clinical efficacy data into real-world outcomes[16].

Where we are, and what comes next​

This document proposes M&S as a viable answer to the current challenges of drug R&D. Proofs of concept show it making a difference across therapeutic areas and across stages of the typical R&D process[17]. What remains to be demonstrated is that full MIDD beats the traditional approach: fewer failures, less money spent, fewer ethical losses. Faster adoption depends on a handful of challenges. The technical ones (see Table 6) are more or less settled, in that solutions exist for everything listed there, though they can be improved, and the details are beyond the scope of this document. The intellectual barriers matter more. First, the assumption that limited knowledge blocks the use of models. Second, regulators need to formalize their position on in silico clinical trials.

MIDD today still centers on in vivo clinical trials, which are the ultimate proof. In the future, in vivo trials plausibly become an instrument for validating models, while the evidence supporting the decisions of scientists, regulators, and doctors comes from simulations run with validated models.

Table 6: Technical hurdles

  • Adjusting the scope of the knowledge captured in the model so it best addresses the problem of interest
  • Choosing the appropriate level of granularity
  • Identifying and gathering the relevant documentation for validation, both knowledge and data
  • Mixing data of various origins and quality to build the VP
  • Constructing the joint distribution of the VP
  • Linking VP descriptors derived from model parameters to real recorded patient characteristics
  • Finding a compromise between the level of detail in the Computational Model and the available mathematical methods and computational facilities, while keeping the research objective reachable
  • Solving the sloppiness issue without impairing the accuracy of the prediction
  • Establishing validation scales and scores that support reliable conclusions given the modeling objectives
  • Accounting for the variability of the knowledge pieces integrated in the model, as captured in the VP, when marking the uncertainty of a prediction
  • Accounting for the strength of evidence of those knowledge pieces when marking the uncertainty of a prediction
  • Selecting simulation outputs that help interpret the predictions
  • Ensuring transparency throughout, from prediction back to knowledge
  • Choosing the time scaling for submodel integration
  • Choosing the time step for solving the equations

1. Scannell, J. W., Blanckley, A., Boldon, H. & Warrington, B. Diagnosing the decline in pharmaceutical R&D efficiency. Nat. Rev. Drug Discov. 11, 191–200 (2012) ↩︎

2. Pauker, S. G. & Kassirer, J. P. The threshold approach to clinical decision making. N Engl J Med 302, 1109–1117 (1980) ↩︎

3. Boissel, J. Individualizing aspirin therapy for prevention of cardiovascular events. Jama 280, 1949–1950 (1998) ↩︎

4. Eisenstein, M. The power of petabytes. Nature 527, 3–4 (2015) ↩︎

5. Hendriks, B. S. et al. Decreased internalisation of erbB1 mutants in lung cancer is linked with a mechanism conferring sensitivity to gefitinib. Syst. Biol. (Stevenage). 153, 457–466 (2006) ↩︎

6. Kansal, A. R. & Trimmer, J. Application of predictive biosimulation within pharmaceutical clinical development: examples of significance for translational medicine and clinical trial design. Syst. Biol. (Stevenage). 152, 214–220 (2005) ↩︎

7. Price Waterhouse Coopers. Pharma 2020: Virtual R&D. Which path will you take? (2008) ↩︎

8. Burghaus, R. et al. Evaluation of the efficacy and safety of rivaroxaban using a computer model for blood coagulation. PLoS One 6, (2011) ↩︎

9. Doudican, N. A. et al. Predictive simulation approach for designing cancer therapeutic regimens with novel biological mechanisms. J. Cancer 5, 406–416 (2014) ↩︎

10. Boissel, J.-P., Auffray, C., Noble, D., Hood, L. & Boissel, F.-H. Bridging Systems Medicine and Patient Needs. CPT Pharmacometrics & Syst. Pharmacol. 4, 135–145 (2015) ↩︎

11. Leil, T. A. & Bertz, R. Quantitative systems pharmacology can reduce attrition and improve productivity in pharmaceutical research and development. Front. Pharmacol. 5, (2014) ↩︎

12. Kovatchev, B. P., Breton, M., Man, C. D. & Cobelli, C. In silico preclinical trials: a proof of concept in closed-loop control of type 1 diabetes. J. diabetes Sci. Technol. 3, 44–55 (2009) ↩︎

13. (2018, December 13). (PBPK) modeling and simulation - European Medicines Agency |. Retrieved June 29, 2021, from https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-reporting-physiologically-based-pharmacokinetic-pbpk-modelling-simulation_en.pdf ↩︎

14. Clyde, R. G., Bown, J. L., Hupp, T. R., Zhelev, N. & Crawford, J. W. The role of modeling in identifying drug targets for diseases of the cell cycle. J. R. Soc. Interface 3, 617–627 (2006) ↩︎

15. VPH Institute - Building the Virtual Physiological Human. (n.d.). Retrieved May 29, 2021, from https://.www.vph-institute.org ↩︎

16. EUnetHTA. Levels of Evidence - Applicability of evidence for the context of a relative effectiveness assessment guideline. (2015). at http://eunethta.eu/outputs/levels-evidence-applicability-evidence-context-relative-effectiveness-assessment-amended-ja1 (opens new window)↩︎

17. Kovatchev, B. P., Breton, M., Man, C. D. & Cobelli, C. In silico preclinical trials: a proof of concept in closed-loop control of type 1 diabetes. J. diabetes Sci. Technol. 3, 44–55 (2009) ↩︎