Abstract Rationale Acute respiratory failure (ARF) requiring mechanical ventilation affects over 1.5 million US adults annually. While ARDS is well characterized, most ventilated patients fall into non-ARDS subtypes with poorly defined clinical and biologic features. Gaps remain in understanding long term survival in non-ARDS ARF subtypes, and associations with biomarker-driven biologic subphenotypes. Methods We prospectively enrolled 846 adults with ARF requiring invasive mechanical ventilation (2011-2024, single academic center). Patients were classified by consensus into seven ARF subtypes—ARDS, at-risk for ARDS, airway controls, congestive heart failure (CHF), acute exacerbation of interstitial lung disease (AEILD), acute-on-chronic hypercapnic respiratory failure, and other/multifactorial. We compared physiologic indices, radiographic severity, applied organ support, plasma host-response biomarkers, and inflammatory subphenotypes at baseline and follow-up. Survival through three years and discharge destination were ascertained via medical records and vital statistics. Results Classification yielded ARDS (36.9%), at-risk (33.7%), airway controls (10.6%), other/multifactorial (7.2%), CHF (6.1%), acute-on-chronic hypercapnic failure (3.2%), and AEILD (2.2%). Subtypes differed in hypoxemia, ventilatory ratio, normalized elastance, and radiographic severity (all p 0.01). ARDS and at-risk displayed the highest plasma inflammatory biomarkers and hyperinflammatory subphenotype prevalence (up to 31%), while AEILD had 0%. CHF patients showed radiographic overlap and elevated biomarkers like ARDS, but lower ejection fraction on echocardiography and improved outcomes. Ninety-day mortality ranged from 15.6% (airway controls) to 78.9% (AEILD, p 0.001), with divergence persisting through 3 years. The other/multifactorial group showed increased mortality through 3 years. Airway controls had the lowest 3-year mortality risk (HR 0.64, 95% CI 0.45-0.92) and AEILD the highest (HR 2.70, 1.63-4.47) compared to ARDS (Figure). Discharge to independent settings was achieved by 79% of airway controls versus 39% of ARDS survivors (Figure). Baseline and evolving hyperinflammatory subphenotypes predicted worse survival within and across subtypes. Conclusions Mechanically ventilated patients with ARF demonstrate marked heterogeneity in physiology, biology, radiography, and outcomes. Systematic clinical classification combined with biologic phenotyping enables complementary prognostic enrichment, delineating subgroups with distinct trajectories and recovery profiles. To our knowledge, this is the first study to characterize 3-year survival by ARF subtype and by inflammatory subphenotype. These findings support the integration of clinical and biomarker data to advance precision prognostication, individualized management, and targeted trial design in ARF. This abstract is funded by: NIH (R01HL176668, R35GM160146)
Kitsios et al. (2026) studied this question.