Humain Lens: The Disruptive Impact of AI and Precision Lung Data on Respiratory Trials

By Susan Wood

We are experiencing a worldwide lung crisis. Lung disease mortality rates are much higher than for heart disease, stroke or cancer, and global economic burden across all respiratory diseases is more than $1.3 billion annually.¹ Lung disease certainly deserves more attention from researchers, clinicians and biotech.

Historically, reliance on imprecise and subjective endpoints has stalled progress and sidelined R&D investment for lung and respiratory disease therapies. A six-minute walk test (6MWT) or spirometry, common measures used in lung and respiratory clinical trials, appear especially subjective when compared to the readout from a cancer gene panel, for example.

Certain lung and respiratory diseases are complex to diagnose, categorize, monitor, and manage. Until now, available tools, such as the 6MWT and spirometry, have been proven inadequate for assessing disease burden, progression, and response to therapy. Fortunately, for the billons who suffer from lung and respiratory disease, artificial intelligence (AI) is more than adequate, it’s essential and disruptive.

Where respiratory disease, AI and precision medicine intersect chronic obstructive pulmonary disease (COPD), the third-leading cause of death worldwide, is an ideal use case for exploring the potential of AI. Investments made to more precisely diagnose and treat COPD would certainly be offset by long-term healthcare savings. The economic impact worldwide would be transformative.

For new precision medicine approaches, establishing clinical utility is a prerequisite for approval and widespread use. In the case of AI and COPD, the challenge has been how to aggregate and apply sufficient data to identify, quantify and validate clinically relevant biomarkers. These AI-derived biomarkers now exist, however, and the menu is growing, so efforts can now shift to investment in new diagnostic approaches and lung and respiratory therapy pipelines. Clinical trial sites can be a particularly rich source of data for training AI and discovering new biomarkers. And this will only improve as use proliferates. Eventually we’ll have a virtuous cycle, fueled by rapidly accumulating patient data, that simultaneously accelerates therapy pipelines while improving diagnostic accuracy at the point of care.

Addressing the crisis of the lung starts with assurances that diagnosis and treatment can indeed be more objective and evidenced-based. As we’ve seen with other precision medicine approaches, rigid recruiting criteria and quantitative endpoints raise the bar for trial matching, but they also reduce variability and improve reliability. This is where AI, with similar clinical utility as genetic or other biomarker data, can deliver the most immediate and meaningful disruption.

AI-powered lung intelligence

AI-powered lung intelligence, manifested as imaging biomarkers, provide objective insights into mechanistic disease processes. Consider mucus plugs, which affect 25-57% of patients with COPD^2-4 and in approximately 58% of patients with asthma.⁵ Mucus plugs are now regarded as clinically relevant to patients’ outcomes, including mortality.⁶

For biopharma developers, the ability to quantitatively measure mucus plugs offers a more reliable measure for therapy response. Until recently, measuring mucus plugs was a costly and time-consuming manual process, making it unsuitable as a primary endpoint for most trials.

Today, through semi-automated, AI-driven mucus scoring workflows, radiologists can assess each lung segment, arming them with visualizations and quantitative results. The workflow is efficient, and the measurement is objective, providing a precise biomarker for assessing response to therapy.

From object endpoints to precision therapies

A mucus plug burden biomarker is just one of more than 50 that therapy developers can use for clinical trial matching and endpoints. It’s impossible to overstate how important – and disruptive – these biomarkers are to the future of lung and respiratory trials and therapy approvals. In fact, according to a 2021 study, the probability of success for a drug to move from Phase I to approval doubles when preselection biomarkers are used.⁷

For one, the use of AI-powered biomarkers improves and accelerates trial enrollment. Today, recruitment costs an estimated $3.2 billion, and even with this high cost 37% of studies fail to reach target study size.^8,9 Developers who discard conventional, subjective measures in favor of these new biomarkers can significantly reduce unnecessary costs of trial delay, which some estimates put at between $600,000 to $8 million per day per drug.¹⁰

AI-powered biomarkers can also enable trial sponsors to reduce trial size altogether. A more representative cohort can be smaller without compromising study design and performance. Months can be saved by identifying the right cohort and having quantitative endpoints that further accelerate validation and regulatory review.

Conclusion

The high economic burden of lung disease is crippling health systems, and the pandemic continues to stretch scarce resources. New approaches to treatment and diagnosis are desperately needed. This will require more investment, preceded by assurances that variability and subjectivity will no longer impede progress in therapy development. AI-powered biomarkers offer that assurance.

AI-powered lung intelligence is truly disruptive, providing clear and objective evidence of therapeutic efficacy. As the use of AI and imaging biomarkers becomes standard of care across clinical trial sites globally, it will create a virtuous cycle of data and insights. This will embolden more sponsors to invest in bringing precision therapies to patients and meet the global crisis of the lung head on.

References

1. 1. ERS: “The total cost of respiratory disease in the 28 countries of the EU alone amounts to more than €380 billion annually. Available at: www.erswhitebook.org/chapters/the-economic-burden-of-lung-disease
   2. 5 – Okajima Y, Come CE, Nardelli P, et al. Luminal Plugging on Chest CT Scan: Association with Lung Function, Quality of Life, and COPD Clinical Phenotypes. Chest. 2020;158(1):121-130. doi:10.1016/j.chest.2019.12.046
   3. Kim V, Dolliver WR, Nath HP, et al. Mucus plugging on computed tomography and chronic bronchitis in chronic obstructive pulmonary disease. Respiratory Research. 2021;22(1):110. doi:10.1186/s12931-021-01712-0
   4. Dunican EM, Elicker BM, Henry T, et al. Mucus Plugs and Emphysema in the Pathophysiology of Airflow Obstruction and Hypoxemia i n Smokers. Am J Respir Crit Care Med. 2021;203(8):957-968. doi:10.1164/rccm.202006-2248OC
   5. Dunican EM, Elicker BM, Gierada DS, et al. Mucus plugs in patients with asthma linked to eosinophilia and airflow obstruction. J Clin Invest. 2018. 128(3):997-1009. doi:10.1172/JCI95693
   6. Okajima Y, Come C, Nardelli P, et al. Mucus plugging on CT and mortality in smokers. European Respiratory Journal. 2019;54(suppl 63). doi:10.1183/13993003.congress-2019. OA1917
   7. GBD Chronic Respiratory Disease Collaborators. Prevalence and attributable health burden of chronic respiratory diseases, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet Respir Med. 2020;8(6):585-596.
   8. Biotechnology Innovation Organization (BIO), Pharma Intelligence, Quantitative Life Sciences. Clinical Development Success Rates, 2011-2020. Available at: pharmaintelligence.informa.com/~/media/informa-shop-window/pharma/2021/files/reports/2021-clinical-development-success-rates-2011-2020-v17.pdf. Accessed April 15. 2022.
   9. Office of Disease Prevention and Health Promotion (ODPHDP). Respiratory Diseases.    Available at: www.healthypeople.gov/2020/topics-objectives/topic/respiratory-diseases.
   10. Munda, J., PharmaIT, First Analysis Quarterly Insights, April 2021

Dr. Susan Wood is president & CEO of VIDA Diagnostics. She has 25+ years of experience championing clinical intelligence solutions into routine clinical use. Dr. Wood received her PhD from the Johns Hopkins Medical Institutions, School of Hygiene and Public Health. Her PhD work combined quantifying three-dimensional lung structure with changes in lung function using high-resolution CT imaging. She also holds a Master of Science degree in Biomedical Engineering from Duke University, and a Bachelor of Science in Engineering from the University of Maryland, College Park.