Skip to content

COPD biomarkers and imaging phenotypes

TL;DR — No blood or imaging biomarker replaces clinical assessment plus spirometry for COPD diagnosis. Blood eosinophils are the most actionable predictive marker: combined with exacerbation history they estimate benefit from ICS and, at higher thresholds in selected chronic-bronchitis COPD, dupilumab (David 2021, PMID 33122447; Bhatt 2024, PMID 38767614). CT can quantify emphysema, airway walls, gas trapping, mucus plugs and pulmonary vessels; MRI adds ventilation/perfusion information without ionizing radiation (Elbehairy 2024, PMID 38548292). Imaging is clinically established for differential diagnosis and procedure selection, but most quantitative features remain prognostic rather than treatment-predictive. Biomarker claims require analytical validity, external validation and prospective treatment interaction.

Biomarker roles

Role Question COPD example
Diagnostic Does the patient have COPD? Post-BD spirometry remains anchor
Prognostic What will happen? CT emphysema, fibrinogen, event history
Predictive Which treatment helps? Eosinophils × ICS/dupilumab effect
Monitoring Has biology changed? Repeated physiology/symptoms
Pharmacodynamic Did a drug hit target? Trial-specific pathway marker

Blood biomarkers

Marker Use Limitation
Eosinophils ICS/biologic treatment selection Variable, non-specific, threshold context
Fibrinogen Prognostic/enrichment Systemic and non-specific
CRP Inflammation/infection context Obesity and comorbidity confounding
Alpha-1 antitrypsin Detect deficiency Acute-phase elevation; needs genotype/phenotype
CBC/hemoglobin Anemia/polycythemia/eosinophils Not COPD-specific
BNP/troponin Cardiac mimic/risk Decision-specific, not COPD marker

Blood eosinophil thresholds relate continuously to exacerbation patterns; fixed cut points are decision aids rather than biological borders (Yun 2018, PMID 29709670).

CT phenotyping

CT feature Quantification Potential use
Emphysema Low-attenuation area, density percentile Prognosis and volume-reduction target
Air trapping Expiratory attenuation/PRM Small-airway disease
Airway wall Pi10 and segmental measures Airway-predominant phenotype
Mucus plugs Segment count/occlusion Exacerbation and type-2 research
Fissure integrity Quantitative completeness Valve eligibility
Vascular pruning Vessel volume/morphology Pulmonary vascular phenotype
Incidental nodule Standard nodule pathway Cancer detection

CT density depends on inspiration, scanner, reconstruction and calibration. Longitudinal change is not interchangeable across protocols (Elbehairy 2024, PMID 38548292).

Unsupervised machine learning can identify emphysema subtypes, but transportability across scanners and cohorts is necessary before clinical assignment (Angelini 2023, PMID 37268414).

MRI and functional imaging

Hyperpolarized-gas MRI visualizes regional ventilation; proton MRI can assess perfusion and structure. MRI avoids radiation and offers repeated functional measurement, but availability, standardization and multicenter validation constrain use (Elbehairy 2024, PMID 38548292).

Nuclear ventilation/perfusion imaging and PET answer narrower physiological or inflammatory questions. Imaging signal must be connected to an actionable decision to justify complexity.

Physiology as biomarker

Measure Biological information Limitation
FEV1 Integrated expiratory flow Weak symptom correspondence
Lung volumes Hyperinflation/gas trapping Equipment and reference dependence
DLCO Alveolar–capillary transfer Anemia, vascular disease confound
Exercise test Integrated limitation Requires mechanism attribution
Oscillometry Peripheral mechanics Clinical thresholds unsettled

From phenotype to treatment prediction

A phenotype becomes clinically useful when it changes an outcome under one treatment versus another. Fissure integrity for valves and eosinophils for anti-inflammatory therapy approach this standard (Criner 2018, PMID 29787288; Bhatt 2024, PMID 38767614).

Radiomics and machine-learning classifiers risk leakage, site effects and optimistic internal validation. External calibration and prospective decision-impact trials are mandatory.

Composite treatable traits

Trait Measurement bundle Candidate intervention
Type-2 exacerbator Events + repeated eosinophils + chronic bronchitis ICS/dupilumab
Hyperinflated emphysema CT + RV/TLC + symptoms + fissures Valve/LVRS
Chronic hypercapnia Stable PaCO2 + sleep/obesity assessment Home NIV
Frequent infective events Sputum/CT/cultures Infection/bronchiectasis pathway
AATD Concentration + genotype/phenotype Family testing/augmentation selection

Biomarker hierarchy: association is not utility

Level Required question COPD example
Analytical validity Is the measurement reproducible? Blood eosinophils vary over time; CT density depends on inspiration and reconstruction (David 2021, PMID 33122447; Elbehairy 2024, PMID 38548292).
Prognostic validity Does it predict outcome beyond standard variables? Mucus-plug burden predicted mortality after adjustment (Diaz 2023, PMID 37210745).
Predictive validity Does it identify differential treatment effect? Eosinophil/chronic-bronchitis enrichment predicted dupilumab benefit (Bhatt 2023, PMID 37272521).
Clinical utility Does biomarker-guided care improve net outcomes? Eosinophil-guided acute steroids reduced exposure in CORTICO-COP (Sivapalan 2019, PMID 31122894).
Implementation utility Can systems deliver it equitably and act on it? Quantitative CT and MRI need acquisition/analysis standardization.

Quantified prognostic signals

Marker Result Limitation
CT mucus plugs Mortality 34.0%, 46.7% and 54.1% for 0, 1–2 and ≥3 plugged segments over median 9.5 years (Diaz 2023, PMID 37210745). Observational; treatment response unproven.
BODE Mortality HR 1.34 (95% CI 1.26–1.42) per point; C statistic 0.74 versus 0.65 for FEV1 (Celli 2004, PMID 14999112). Prognostic index, not a molecular endotype.
Persistent systemic inflammation Mortality 13% versus 2%, exacerbations 1.5 versus 0.9/year (Agustí 2012, PMID 22624038). Limited biomarker panel and no causal proof.
Frailty Mortality HR 1.68 (95% CI 1.37–2.05), 6MWD −90.23 m (Wang 2023, PMID 37173728). Definition-dependent prevalence.
Lung-function trajectory Normal-attainment/rapid-decline COPD versus low-attainment COPD all-cause mortality HR 1.93 (95% CI 1.14–3.26) (Marott 2020, PMID 32289231). Requires longitudinal data unavailable in routine care.

Imaging as a predictive biomarker

Terminal-airway loss can precede emphysema (McDonough 2011, PMID 22029978), while machine-derived CT subtypes capture spatial heterogeneity (Angelini 2023, PMID 37268414). The strongest demonstrated imaging utility is procedural: absent collateral ventilation and target-lobe anatomy enriched large valve effects, with TRANSFORM showing ≥12% FEV1 response in 55.4% versus 6.5% but 29.2% pneumothorax (Kemp 2017, PMID 28885054). That is a treatment-specific predictive biomarker, not validation of CT clusters for every therapy.

Eosinophil controversy after multiple biologic programs

Benralizumab was neutral in primary phase 3 comparisons despite eosinophil enrichment (Criner 2019, PMID 31112385). Dupilumab produced replicated benefit at ≥300 cells/µL with chronic bronchitis (Bhatt 2023, PMID 37272521; Bhatt 2024, PMID 38767614), and MATINEE later reduced exacerbations with mepolizumab (rate ratio 0.79, 95% CI 0.66–0.94) without significant symptom/QoL benefit (Sciurba 2025, PMID 40305712). Blood eosinophils therefore mark a probability distribution interacting with pathway, population and endpoint—not a stand-alone diagnosis.

Fibrinogen illustrates association, enrichment and surrogate failure

Pooled data from 6,376 people with spirometric COPD found fibrinogen at least 350 mg/dL in 44.7%; this threshold predicted hospitalized exacerbation within 12 months (HR 1.64, 95% CI 1.39–1.93) and death within 36 months (HR 1.94, 1.62–2.31) (Mannino 2015, PMID 25685850). This led to FDA qualification of plasma fibrinogen in 2015 as a prognostic/enrichment drug-development biomarker—the first such qualified tool in COPD—not as a treatment-response surrogate or clinical diagnostic test (Miller 2016, PMID 26745765).

The broader biomarker literature reinforces that distinction. Across 61 studies, mortality HRs were 0.80 (95% CI 0.73–0.89) per 50 m longer six-minute walk distance, 1.10 (1.02–1.18) per 10-bpm higher heart rate, 3.13 (2.14–4.57) per doubling of fibrinogen, 1.17 (1.06–1.28) per doubling of CRP and 2.07 (1.29–3.31) per doubling of leukocyte count (Fermont 2019, PMID 30617161). Strong prognosis does not prove that lowering the marker improves outcome; validation must specify the use case before judging performance.

No single-marker history has yet escaped the use-case problem: blood CRP/fibrinogen, sputum neutrophils, CCL18, CC16, integrative indices and CT phenotypes have all shown association, but none is a universal COPD diagnostic or treatment-response marker (Rosenberg 2012, PMID 22424427). In 814 tobacco-exposed SPIROMICS participants with preserved spirometry, high RV/TLC identified more functional small-airway disease, symptoms and later progression than high TLC, despite no significant difference in exacerbation, hospitalization or mortality incidence over follow-up (Arjomandi 2025, PMID 39586032). Even composite physiology therefore requires an explicitly defined prognostic or predictive decision.

Incremental prediction is stricter than association

In 640 stable and 262 hospitalized COPD participants, doubling fibrinogen and MR-proADM was associated with 3-year mortality HRs of 2.2 (95% CI 1.3–3.7) and 2.1 (95% CI 1.5–3.0), respectively. Adding MR-proADM to fibrinogen increased one-year mortality AUC from 0.78 to 0.83 (p=0.02), but adding fibrinogen to MR-proADM did not improve AUC (0.83 versus 0.82; p=0.34) (Zuur-Telgen 2021, PMID 34886719). Qualification for enrichment and independent prognostic association therefore do not guarantee incremental bedside utility.

In 1,830 MESA participants without baseline chronic lower-respiratory disease, each SD greater CT Pi10 was associated with 9% faster FEV1 decline (95% CI 2%–15%), incident COPD OR 2.22 (95% CI 1.43–3.45), and 57% higher risk of first respiratory hospitalization or death over 78,147 person-years (Oelsner 2018, PMID 29529382). Pi10 integrates wall thickness and lumen caliber and is scanner/segmentation dependent; outcome association alone does not specify a treatment.

Open questions

  • Can quantitative CT prospectively assign inhaled or biologic therapy? (Elbehairy 2024, PMID 38548292)
  • What repeated eosinophil measure best predicts net ICS benefit? (David 2021, PMID 33122447)
  • Which mucus-plug metrics are causal and reversible? (Lee 2024, PMID 39624959)
  • Can MRI provide a multicenter surrogate endpoint responsive to treatment? (Elbehairy 2024, PMID 38548292)
  • A PubMed search repeated on 2026-09-02 for COPD biomarker-algorithm recalibration across ancestry and scanner platforms again retrieved no directly responsive study; external validation across both sources of measurement shift remains an evidence gap.

References

  1. Elbehairy AF, et al. Advances in COPD imaging using CT and MRI. Eur Respir J. 2024. PMID 38548292
  2. David B, et al. Eosinophilic inflammation in COPD: marker to treatable trait. Thorax. 2021. PMID 33122447
  3. Bhatt SP, et al. Dupilumab for COPD with blood-eosinophil evidence of type-2 inflammation. N Engl J Med. 2024. PMID 38767614
  4. Yun JH, et al. Blood eosinophil count thresholds and exacerbations. J Allergy Clin Immunol. 2018. PMID 29709670
  5. Angelini ED, et al. Emphysema subtypes by unsupervised CT machine learning. Thorax. 2023. PMID 37268414
  6. Criner GJ, et al. LIBERATE endobronchial-valve trial. Am J Respir Crit Care Med. 2018. PMID 29787288
  7. Lee YL, et al. Eosinophilic inflammation in COPD pathogenesis and progression. Ann Med. 2024. PMID 39624959
  8. Diaz AA, et al. Airway-occluding mucus plugs and mortality in COPD. JAMA. 2023. PMID 37210745
  9. Celli BR, et al. The BODE index in COPD. N Engl J Med. 2004. PMID 14999112
  10. Agustí A, et al. Persistent systemic inflammation and poor COPD outcomes. PLoS One. 2012. PMID 22624038
  11. Wang L, et al. Frailty prevalence and clinical impact in COPD. BMC Pulm Med. 2023. PMID 37173728
  12. Marott JL, et al. Lung-function trajectories as predictors of mortality. Am J Respir Crit Care Med. 2020. PMID 32289231
  13. McDonough JE, et al. Small-airway obstruction and emphysema in COPD. N Engl J Med. 2011. PMID 22029978
  14. Kemp SV, et al. TRANSFORM Zephyr-valve trial. Am J Respir Crit Care Med. 2017. PMID 28885054
  15. Sivapalan P, et al. Eosinophil-guided corticosteroids in hospitalized exacerbations. Lancet Respir Med. 2019. PMID 31122894
  16. Criner GJ, et al. Benralizumab for COPD exacerbation prevention. N Engl J Med. 2019. PMID 31112385
  17. Bhatt SP, et al. Dupilumab for COPD with type-2 inflammation. N Engl J Med. 2023. PMID 37272521
  18. Sciurba FC, et al. Mepolizumab in eosinophilic COPD. N Engl J Med. 2025. PMID 40305712
  19. Mannino DM, et al. Plasma fibrinogen as a biomarker for mortality and hospitalized exacerbations in COPD. Chronic Obstr Pulm Dis. 2015. PMID 25685850
  20. Miller BE, et al. Plasma fibrinogen qualification as a drug-development tool in COPD. Am J Respir Crit Care Med. 2016. PMID 26745765
  21. Fermont JM, et al. Biomarkers and clinical outcomes in COPD: systematic review and meta-analysis. Thorax. 2019. PMID 30617161
  22. Rosenberg SR, et al. Biomarkers in COPD. Transl Res. 2012. PMID 22424427
  23. Arjomandi M, et al. Phenotypes and trajectories of tobacco-exposed persons with preserved spirometry. Ann Am Thorac Soc. 2025. PMID 39586032
  24. Zuur-Telgen MC, et al. Predicting mortality in COPD with fibrinogen and mid-range proadrenomedullin. COPD. 2021. PMID 34886719
  25. Oelsner EC, et al. Prognostic significance of large-airway dimensions on CT in the general population: MESA Lung. Ann Am Thorac Soc. 2018. PMID 29529382