Back in the 1990s, the idea of running computer-simulated clinical trials would have seemed far-fetched. The internet had only just emerged, personal computers were the size of suitcases, and...
Faster drug approvals based on less evidence mean new tools are needed to track cost-effectiveness. As regulators approve more drugs faster , health technology assessors are under growing pressure to rapidly quantify the added benefit of these therapies. Regulators and health technology assessors are challenged by limited evidence of clinical efficacy on short deadlines. Health technology assessment (HTA) organizations also face squeezed budgets, making their task even tougher.
Hence the scramble for new tools to help bridge the gap between desired and available evidence, also covered here . Real-world evidence is increasingly used, featuring within a growing list of non-traditional evidence sources that the UK’s National Institute for Health and Care Excellence (NICE) is seeking to integrate into its assessments as part of its strategy and methods review .
Evidence of benefits shown in a clinical trial rarely translate perfectly into a real-world setting. Adjustments are required. Identifying and quantifying those adjustments can have a big impact on price and value.
Take the US’s Institute for Clinical and Economic Review (ICER)’s recent reassessment of a trio of medicines used to help prevent attacks of hereditary angioedema, or severe swelling.The original health economic assessment from 2018 based on Phase III trial data over-estimated the number of attacks patients were having before treatment began. Adjusting that baseline figure using health claims data generated a significantly lower value-based price recommendation for the drugs.
Things could, in theory, have gone the other way: a drug might have shown even more benefit for some patients in the real-world than in a clinical trial. Regardless, patients at large often behave differently to those in a controlled trial. And patients change over time – along with disease prevalence, severity and treatment choices.
Regularly collecting and crunching various forms of real-world data, such as claims data, is one option. But it’s cumbersome and expensive, particularly when carried out repeatedly for multiple drugs. Such evidence, too, comes with its own biases and flaws.
In silico disease models and simulated trials among “virtual” populations offer a valuable additional evidence source during R&D, as we’ve outlined here and here . They can also add significant value post-approval, either by directly helping demonstrate drugs’ cost-effectiveness, and/or by guiding more informed, efficient real-world data collection.
In silico disease and treatment models are increasingly comprehensive, detailed and accurate, thanks to extraordinary growth in scientific knowledge and computer processing power. They allow multiple hypotheses to be tested, quickly and repeatedly, with no burden on real patients or health workers, and at a far lower cost than real-life trials.
For example, by defining “virtual” patients with different disease characteristics, in silico models can help identify and characterize sub-groups that might benefit more – or less – than others from a given treatment. Lifestyle or behavior characteristics could be modeled, too. There’s no limit on the number of virtual patients – in silico trials can be bigger and can simulate a longer time-span than would be economically or ethically feasible in the physical world.
As more drugs receive expedited review and are approved on condition of further data collection, the speed and cost advantages of in silico-generated evidence are gaining prominence.
Even controlled clinical trials carried out post-launch may not generate clear-cut results. In early 2021, a handful of checkpoint inhibitors approved under FDA’s Accelerated Approval pathway failed to generate evidence of benefit in some indications. FDA committee experts were divided on whether such therapies should be withdrawn. Some argued that a subset of patients were benefiting, but they hadn’t been clearly characterized while others wanted to uphold statistical principles at all costs.
In silico modeling could help bridge the divide. Knowledge-based models used with simulated populations offer the option to test multiple “virtual” patient subgroups (for instance, those with specific tumor characteristics or immunological profiles), guiding more targeted, and thus faster, Phase IV/post-approval data collection.
In silico modeling won’t replace in-person clinical trials or real-world studies, but it can ensure that trials are designed to be as efficient and powerful as possible.
Trial efficiency is now more important than ever. As healthcare costs escalate, regulators’ attempts to balance rapid access with the need for rigorous evidence of efficacy are coming under increased scrutiny , particularly around the use of surrogate endpoints. Meanwhile, HTA bodies like NICE aspire to a more flexible, iterative approach with regular drug re-assessments replacing isolated yes/no reimbursement decisions
In silico is already gaining traction with regulators as they seek new sources of information to help understand increasingly complex therapies. Knowledge-based models of disease and treatment could provide health technology assessors with additional insights, alongside those from economic models, into the impact of these therapies on patients, and health budgets, in the real world.