Susan Wood, PhD

The pandemic experience has been sobering on many levels: As we’ve pressure tested our health systems, we’ve exposed systemic weaknesses, especially in how we address declining lung health. Respiratory disease was already among the leading causes of death globally; now, with COVID, the sense of urgency is even more acute.

Today, patients with “long COVID” or related complications are straining many health systems.And according to researchers at Penn State College of Medicine, more than half of the 236 million patients diagnosed with COVID worldwide since December 2019 will experience post-COVID symptoms up to six months after recovering.

As investment in long COVID research grows, so too will focus on lung and respiratory disease overall, and this will lead to greater scrutiny of respiratory clinical trials and the high costs and failure rates. One area of focus will be the systemic implications of endpoint subjectivity. Existing endpoints used in respiratory trials, such as physical activity performance for COPD, asthma, and idiopathic pulmonary fibrosis, “represent a challenging domain to accurately measure,” and this underscores the need for biomarkers and more precise endpoints that better predict patient response.

There is no single reason respiratory clinical trials fail, but the inability to recruit and retain suitable cohorts is certainly among them. Recruiting challenges have led many to suggest that overly rigid inclusion criteria are to blame, but that’s a double-edged sword. A looser set of inclusion criteria can pull in more participants, but at the risk of creating an overly heterogeneous subject pool with the potential to increase variability and trial failure. Precision imaging phenotypes help study designers of respiratory trials strike the right balance in setting inclusion criteria that’s highly targeted, yet feasible for recruitment.

Instead of unnecessarily restricting the addressable trial population, objective, highly quantifiable measures provide an even stronger basis for expanding recruitment. Much of what’s needed to identify and qualify more suitable candidates—existing CT scans for example—is readily available. The biomarkers are not only there to guide study recruitment but also to establish multiple endpoints for “lung intelligence” that enables even deeper, more granular analysis of mechanisms of action.

Natural language processing (NLP) and artificial intelligence (AI) are also showing tremendous potential to improve clinical trial recruiting. This is true in the early stages of design, where NLP and AI can be used in literature review and to mine studies retrospectively, and throughout a study, and where more precise endpoints, such as AI-enabled biomarkers in an acute respiratory distress syndrome trial, can help investigators establish rigid but more quantifiable criteria that effectively expand access to a larger addressable trial population.

In the case of drug trials for lung and respiratory therapies, this increasing focus on AI and AI-enabled biomarkers is transformative. Now, where highly subjective results from a six-minute walk test were once a standard of care for patient inclusion and assessing mechanisms of action, we can now use more objective and precise measures such as air trapping, mucus plug scoring, total airway count (TAC), and tissue density metrics. As stated by pulmonologist Marc Miravitlles, MD, PhD: “A drug that is expected to interfere with the progression of emphysema should be evaluated by changes in lung density (HRCT) and not by its possible effect on spirometry, exacerbations, or St. George Respiratory Questionnaire (SGRQ) scores. Choosing the right primary outcome may have a strong impact on the future of treatments for emphysema in patients with COPD.”

A good example of precision imaging’s impact on therapy response is from a recent 180 subject alpha-1 COPD trial in which CT-based imaging biomarkers showed a 34% reductionin the rate of decline of lung density; meanwhile, conventional tests (spirometry, questionnaire, and exacerbation rates) showed no impact of the therapy.

Having more precise measures give trial sponsors options that didn’t previously exist, including how and where trials take place. One of the serendipitous outcomes of COVID has been wider-spread acceptance that decentralized trials can be viable and beneficial for sponsors and patients alike. During the height of the pandemic, for example, medical device companies began offering “at home” tools to avoid trial disruption, including home spirometry kits to remotely assess lung function.

A recent Morning Consult poll suggests that the public is more than ready. Roughly four in five adults say it’s important that trials are easy to get to and diverse. Moreover, adults were more likely to be interested in participating in a clinical trial if site visits were virtual or took place closer to them, even if closer was a retail clinic versus a hospital.

Policymakers are also focused on the potential of expanding clinical trial participation and optionality. Likely persuaded by the progress made during the pandemic, for both COVID- related and ongoing non-COVID trials, the 21st Century Cures Act 2.0, Section 310, directs the Secretary of Health and Human Services to “convene a multistakeholder meeting to explore innovative ways and incentives to foster the adoption of decentralized trials.”

According to a recent McKinsey report, pharmaceutical sponsors have also done a rapid pivot in the face of COVID. Pre pandemic (2019), an Industry Standard Research survey found that 38% of pharma and contract research organizations expected virtual trials to be a major component of their portfolios. When McKinsey asked the same questions a year later, 100% said yes.

A decentralized model for respiratory clinical trials is increasingly more viable, even in those that involve precision imaging. Having precise, objective criteria for recruiting means that multiple, geographically distributed sites can now do the work of a few larger sites, and more efficiently. With new technologies such as AI and assiduous data management we can reduce site-to-site variability in data collection, instrument calibration, training and other requirements and cast a wider net for viable trial candidates.

Trillions of terabytes of healthcare data are captured each day, yet so little of that data is made actionable. This missed opportunity is especially concerning when it comes to data from the height of the pandemic and, moving forward, long COVID. We have an opportunity in lung and respiratory disease to pool data on a massive scale, and the learnings from that collaboration would accrue to clinical trial performance across all disease areas.

Some, including contributors to this publication, argue for the importance of data and AI in helping streamline regulatory reviews. “AI can handle the quantum increase in data for regulatory domain by processing tedious tasks and increasing the speed and accuracy of certain regulatory functions such as regulatory information processing, generation of impact analysis reports, gathering insights on current trends, etc.”

 Accelerating regulatory review while improving accuracy would be a game changer.

With AI, we also have the luxury of retrospective analysis. Retrospective analyses, looking into past or similar trials, can guide decision-making about proceeding to a next trial phase and, equally transformative, can provide new and deeper insights about mechanisms of action. Retrospective analyses will have a significant impact on clinical trial effectiveness and cost moving forward.

The potential to mine provider databases, including electronic health records, is limitless. This opens the door for increasing trial access and diversity, objectives the public overwhelmingly supports. Having more precise measures for recruiting, management and oversight improves the overall value proposition for clinical trials, for all stakeholders. The destination is a true matching system that ensures equal access to trials, fewer trial failures and a much better ROI for trial sponsors and society overall.

In the area of respiratory disease, the crisis of the lung has intersected squarely with COVID, and we’re already seeing massive changes in clinical trial design, better recruiting and, soon, more successful outcomes. This progress in lung and respiratory trials is the realization of more than a decade of innovation with AI and AI-enable biomarkers. It mirrors progress with genomic data and reflects a greater willingness to discover and apply more objective measures that improve the speed, equity, and cost-effectiveness of clinical trials.

Equally important in the coming months will be decisive action in Washington and other capitals, such as the FDA’s Biomarker Qualification Program, under the Center for Drug Evaluation and Research, that encourage the validation and widespread application of quantitative biomarkers in clinical trials and as companion diagnostics.

Make no mistake, the issue is a global one. A recent 

 article, entitled “Respiratory research funding is inadequate, inequitable, and a missed opportunity,”

 calls particular attention to the crisis of the lung, but it advocates for scientific collaboration that would transcend a single disease. The authors see the World Health Organization’s “Coordinated Global Research Roadmap” as the model to emulate. This is yet another unintended, but positive outcome of the pandemic.

Five years from now, if not sooner, respiratory clinical trials will be far less costly, failure rates much lower and therapies more widely and equitably available. COVID may still be with us endemically, but its pandemic legacy will be as a catalyst for widespread and systemic changes in healthcare, including clinical trials. Greater reliance on AI, genomic data, and other more precise imaging measures in respiratory trials and at the point of care will enable massive change, and the hard-won lessons of COVID can and must guide us as we go.

Miravitlles M, Anzueto A. Use of CT Lung Densitometry as Outcome Measure for Emphysema Progression: The Case of Losartan. Am J Respir Crit Care Med. Published online June 2, 2022:rccm.202205-0927ED. doi:10.1164/rccm.202205-0927ED

Chapman KR, Burdon JGW, Piitulainen E, et al. Intravenous augmentation treatment and lung density in severe α1 antitrypsin deficiency (RAPID): a randomised, double-blind, placebo-controlled trial. Lancet. 2015;386(9991):360-368. doi:10.1016/S0140-6736(15)60860-1

Morning Consult National Tracking Poll #2112028, December 03-06, 2021, 

https://assets.morningconsult.com/wp-uploads/2021/12/20125132/2112028_crosstabs_MC_HEALTH_CLINICAL_TRIALS_Adults.pdf

Gaurav Agrawal, Rachel Moss, Ralf Raschke, Stephan Wurzer, and John Xue, McKinsey’s Pharmaceuticals & Medical Products Practice. No place like home? Stepping up the decentralization of clinical trials. 

https://www.mckinsey.com/industries/life-sciences/our-insights/no-place-like-home-stepping-up-the-decentralization-of-clinical-trials

Rohit Kadam, Saurabh Das, Ashutosh Pachisia, Niketan Panchal, Anita Ramachandran, Dr. Ashish Indani, Prashant Chaturvedi. The Future of Regulatory Intelligence with Conversational AI, 

https://www.appliedclinicaltrialsonline.com/view/the-future-of-regulatory-intelligence-with-conversational-ai

Williams S, Sheikh A, Campbell H, et al. Respiratory research funding is inadequate, inequitable, and a missed opportunity. Lancet Respir Med. 2020;8(8):e67-e68. doi:10.1016/S2213-2600(20)30329-5, 

https://pubmed.ncbi.nlm.nih.gov/32763207/

Articles

Pandemic has brought new sense of awareness to respiratory diseases and clinops.

How COVID is Forcing a Reckoning in Respiratory Clinical Trials

The lung is a complex organ, and the clinical trials processes that support new therapies for lung and respiratory disease are equally complex. Streamlining these processes with new approaches, including artificial intelligence (AI), could significantly improve the speed and quality of respiratory clinical trials while lowering costs. This article will discuss steps investigators and sponsors can take now to streamline trials and accelerate the time to market for lifesaving and life-changing therapies.

According to Eastern Research Group, it can take more than a decade to get a new drug to market, and respiratory trials are among the most expensive. And, according to the Biotechnology Innovation Organization (BIO), 58% of respiratory drugs fail in the last phase of the clinical trial process.

A first reason cited for failure is that pulmonary trials require a high number of subjects to achieve strong statistical significance. The conventional measures used to assess trial participants, such as the six-minute walk test, can introduce variability that is not directly related to lung health, and this affects the data. A more precise quantitative method for assessing lung structure and function in clinical trials would complement existing measures and potentially require fewer subjects.

Finding and recruiting appropriate subjects that meet inclusion criteria is a challenge for nearly all trials. Unfortunately, patient data is not readily accessible to investigators, and physicians have a difficult time staying informed about current trial opportunities for their patients.

Lastly, Roots Analysis data show that recruitment and retention issues make it difficult to find viable clinical trial sites. In fact, 37% of research sites do not meet subject accrual goals and more than 10% fail to enroll a single patient.

 Respiratory trials that involve imaging are especially challenging for sites given the complexities of subject breathing training, acquisition protocols, data security and more. These challenges create costly delays for new therapies and stop some promising trials altogether. These delays come at a time of continual increases in incidence, prevalence and mortality from lung disease and a need for technological and therapeutic advances that can be addressed with better approaches to imaging, recruiting and clinical site preparedness.

Over the past decade, better hardware and software have enabled trial sponsors to rely on more precise quantitative imaging-based data for respiratory clinical trials, providing much more detailed information about the condition of the lung.

Quantitative imaging has several advantages over conventional measures. First, imaging biomarkers offer reliable, objective measures of structural lung changes, such as the ability to measure airway wall thickness with millimeter precision using quantitative CT (QCT). A second advantage is repeatability. While a subject might have some variability in a spirometry effort, imaging data is likely more reliable and repeatable, especially when sites are adequately trained to collect imaging for clinical trials.

Imaging biomarkers can also be useful for subject selection by applying inclusion and exclusion criteria. A BIO study found that the probability of success for a given drug to graduate from Phase I to approval doubles when preselection biomarkers are used.

If highly precise imaging measures can be used in a trial, it’s likely that fewer subjects will be necessary to reach a statistically significant result. One study on this topic showed that a trial with 550 subjects using conventional measures could have been reduced to 130 patients if QCT measures were used.

 Reducing the number of subjects can speed trials and lower costs.

Focus on the right clinical trial subjects

Reducing the number of required trial subjects is a great start, but it can still be challenging to find the right patients for a respiratory trial; research firm Roots estimates $3.2 billion is spent on patient recruitment alone.

 Rich insights contained within imaging data can help address this challenge, however. Thanks to AI, sponsors can use imaging-based biomarkers to objectively and quantitatively determine which patients are ideal candidates for a clinical study.

Markers of emphysema, COPD, interstitial lung disease (ILD), asthma, and lung cancer are now used to identify viable trial participants. If, for example, a new emphysema drug trial is targeting individuals with upper-lobe predominant emphysema with a disease burden of 50% or more, an automated AI-powered analysis of multiple chest CT scans can surface candidates who match the specific criteria. The entire sourcing process is automated, which is an obvious benefit for trial sponsors as well as treating physicians whose patients may be included in promising trials based purely on their existing scan data.

Clinical trial site training and best practices

The challenges of managing a clinical trial site include proper training, consistency, equipment calibration, and maintaining high standards for efficiency and compliance. It’s not easy. Staff turnover is high, scanners get upgraded, and the need to retrain makes the trial process even longer and more costly. Adding imaging-based biomarkers to this process needn’t add complexity. We’ve successfully trained site personnel to ensure images are acquired using specific CT protocols and process controls that ensure scanner settings and breathing instructions are executed properly by a trained, certified technician on a calibrated CT scanner.

Technology, including AI, that supports earlier and more accurate diagnosis and management of lung and respiratory disease, is now more accepted across healthcare and biopharma. For clinical trials specifically, bringing quantification and objectivity to patient selection and trial management will improve the economics and patient impact of drug discovery and commercialization. It’s not hard to see how greater lung imaging precision, providing AI-enabled data on exact location and nature of damage, is invaluable in drug development, especially when this same technology is also becoming the standard of care for patient diagnosis and therapeutic guidance around the world.

, is the President & CEO of VIDA Diagnostics

Thomas DW, Burns J, Audette J, et al. Clinical development success rates 2006-2015. Biotechnology Innovation Organization, Washington DC. June 2016.

Dirksen and colleagues found that “using quantitative CT (QCT) measures, power analysis showed that the effect would be significant in a similar trial with 130 patients” vs. 550 subjects with conventional measures. Source: 1 - Dirksen et al, A Randomized Clinical Trial of a1-Antitrypsin Augmentation Therapy, Am J Respiratory Critical Care, Vol 160, 1468-1472, 1999

Biotechnology Innovation Organization (BIO), Pharma Intelligence, Quantitative Life Sciences; “Clinical Development Success Rates, 2011-2020,” February 2021, Page 18

Dirksen et al, A Randomized Clinical Trial of a1-Antitrypsin Augmentation Therapy, Am J Respiratory Critical Care, Vol 160, 1468-1472, 1999

5 Ways to Lower Clinical Trial Patient Recruitment Costs, Antidote, 2021

Steps that investigators and sponsors can take to streamline trials and accelerate the time to market for lifesaving therapies.

Susan Wood, PhD

Articles

How COVID is Forcing a Reckoning in Respiratory Clinical Trials

Streamlining the Complex Respiratory Clinical Trial Infrastructure