Interstitial Lung Abnormalities and Aging Biomarkers: A Mediation

Benjamin Franklin famously wrote that “life’s tragedy is that we get old too soon and wise too late.” Although one may be faulted for getting wise too late, getting old too soon may be out of his or her control. Molecular risk factors that predispose to accelerated aging have been identified across the genetic, epigenetic, transcriptomic, and proteomic landscape (1). Those with evidence of accelerated biologic aging are more likely to develop common age-related conditions such as heart disease and malignancy but also appear to be at increased risk of lung disease. Short telomere length and DNA methylation, both potential markers of accelerated biologic aging (1), predispose to the development of fibrosing interstitial lung disease (ILD), resulting in progressive lung function decline and high mortality (2–4). A number of biomarkers that reflect biologic aging processes are also key mediators of fibrogenesis (5–8), suggesting a biologic link between fibrotic ILD and accelerated aging.

In this light, there exists high potential for the biologic processes of normal aging and pathologic accelerated aging to shed light on those contributing to the development of fibrosing ILD. Because the presence of pulmonary fibrosis suggest that the horse has already left the proverbial barn, studies aimed at elucidating the pathogenesis of fibrosing ILD could be particularly informative when performed in those with the earliest features of ILD, termed “interstitial lung abnormalities” (ILAs) (9). Most ILAs progress over time, increase in prevalence with age and smoking history, and likely have a negative impact on survival (10–13), supporting their classification as early ILD. Moreover, these associations appear to strengthen in those with fibrotic ILAs (11), suggesting that ILAs capture the entire continuum of ILD from early morphologic changes to clinically relevant disease.

In this issue of the Journal, Sanders and colleagues (pp. 1149–1157) seek to address the link between ILAs detected by chest computed tomography (CT) in participants from the Framingham Heart Study and plasma biomarkers of accelerated aging (14). These authors assessed 10 previously reported biomarkers of aging (8) and found that increasing plasma concentration of GDF15 (growth differentiation factor 15), TNFR (tumor necrosis factor α receptor II), IL-6, and CRP (C-reactive protein) was associated with increased odds of ILA presence. GDF15 and TNFR in particular were strongly associated ILA presence, with each unit increase in log transformed value increasing the odds of ILA presence by greater than threefold. Importantly, these results were robust to adjustment for other age-related conditions such as cancer, hypertension, and coronary artery disease, which was measured by coronary artery calcium score. To validate their GDF15 findings, Sanders and colleagues tested the association between this biomarker and ILAs in an independent cohort drawn from COPDgene study (NCT00608764). GDF15 was again associated with ILA presence, with each unit increase in log transformed value increasing the odds of ILA presence by greater than eightfold.

Sanders and colleagues then assessed the association between these aging biomarkers and mortality risk. Although no association reached statistical significance after adjustment for multiple testing, increased concentrations of GDF15, TNFR, and an insulin growth factor were associated with an increase in mortality risk. Among COPDgene participants, GDF15 was also associated with survival, with each unit increase in log transformed value increasing the hazard of death by 60%, supporting the findings in the Framingham cohort. Finally, given the shared associations between these biomarkers, aging, and the presence of ILAs, these authors conducted causal mediation analysis to determine the extent to which each biomarker mediated the association between ILAs and age. They found that TNFR and IL-6 mediated just under 10% of the association between ILAs and age, whereas GDF15 mediated 22% but failed to cross the statistical significance threshold after adjustment for multiple testing. GDF15 was again tested in COPDgene participants and was found to mediate 58% of the association between ILAs and age.

Taken together, these findings provide new and important insight into the shared pathobiology of ILA development and aging and support the work of others showing GDF15 to be an important mediator of fibrosing ILD (7). Validation of the observed associations between ILA presence and IL-6, TNFR, and CRP concentrations would strengthen these findings even further, as these biomarkers were not available in the COPDgene cohort. Another limitation worth noting stems from the time interval between biomarker determination and chest CT acquisition. At least 1 year elapsed for all 10 biomarkers assessed, and more than 5 years elapsed for several of the key biomarkers identified in this study, including GDF15 (7 yr), IL-6 (5.1 yr), and TNFR (5.1 yr). These time intervals make the Framingham cohort results difficult to interpret in isolation, but GDF15 validation in COPDgene participants suggests true association and greatly strengthens the findings overall.

The findings presented here provide not only important pathobiologic information but also an exciting glimpse into the future of ILD screening. Like wisdom, fibrosing ILD is an evolution rather than discrete state of being and has early and late phases that bookend an intermediary phase of variable length. Detection of early-phase ILD remains a major challenge. Despite prior studies showing ILAs to be present in up to 10% of screened individuals (10–13), the proportion of these individuals who develop clinically significant ILD and the timeframe over which that happens remains unclear. Until these questions are answered, systematic ILD screening using chest CT is likely to remain unrealistic and cost ineffective. Consequently, the findings presented here may serve as a stepping-stone toward blood-based ILD screening. This will undoubtedly require the identification and validation of additional biomarkers beyond those presented here, but this study represents an exciting first step in the right direction.