Since the original description of the syndrome by Ashbaugh and Petty, and the meticulous work by Katzenstein a decade later, diffuse alveolar damage (DAD) and the presence of hyaline membranes have been considered the pathologic hallmark of ARDS [1, 2]. Animal models to guide discovery of new therapies for ARDS seek to recapitulate both the physiologic alterations seen in patients with ARDS and many of the histologic findings of DAD [3]. Attempts to identify lung-specific biomarkers of epithelial and endothelial injury and studies designed to identify genetic susceptibility traits for the development of ARDS implicitly assume that patients with ARDS have DAD [4]. However, this construct suffers from a number of inconvenient truths.

Patients succumbing to ARDS, often from multiple organ failure, have DAD at postmortem in only 40–60 % of cases, though most of the rest have bacterial pneumonia or edema without DAD [5, 6]. Patients with persistent non-resolving ARDS have similarly low percentages of DAD on lung biopsies [7, 8]. Pneumonia, emphysema, diffuse alveolar hemorrhage, metastatic malignancies, fibrosis, drug reactions, and crytopgenic organizing pneumonia make up the histologic diagnoses in the remainder [7, 8]. Perhaps even more disconcerting, 14 % patients meeting the Berlin definition of ARDS have no pulmonary lesions [5]. This explains, in part, the failure of even the best biomarkers to achieve an area under the receiver operator characteristic curve (AUROC) much over 0.75 [4, 5]. Furthermore, it appears that DAD is becoming less common in the modern era. Thille et al. observed that in patients with fatal ARDS, DAD at autopsy was less frequent between 2001 and 2010 (29 % of autopsies) versus the prior decade (44 %) [5]. This suggests that ventilator-induced lung injury contributed to the prevalence of DAD.

These findings have led some to question the link between DAD and ARDS. For example, the framers of the Berlin definition disagreed on this point and some argued that pneumonia and non-cardiogenic edema, the major findings when DAD is not present, are compatible with ARDS when clinical criteria are satisfied [9]. Supporting this view is the observation that low tidal volume ventilation reduced mortality in all clinical risk factors leading to ARDS, including patients who did not likely have DAD [10].

However, before abandoning the historical link between DAD and ARDS, it is important to review recent evidence suggesting that DAD matters (Table 1). First, the clinical course of ARDS in patients with DAD differs from the course of ARDS without DAD. Lorente et al. studied the clinical features of 149 patients with ARDS prior to death and observed that patients with DAD, in contrast to those without DAD, had a lower PaO2/FIO2 ratio and respiratory system compliance, a higher SOFA score and international normalized ratio, and were more likely to die of hypoxemia and less likely to die of shock [11]. A multivariate model, validated in a separate cohort of 57 patients, yielded an AUROC for the presence of DAD of 0.74 and 0.73 respectively, compared to 0.64 and 0.67 for the Berlin definition. In nonresolving ARDS, PaO2/FIO2 was lower and plateau pressure of the respiratory system higher in those patients with DAD on open lung biopsies than in those without DAD [7]. DAD is also more likely to be present in severe ARDS at postmortem [5]. Recently, a systematic review and meta-analysis of eight studies of open lung biopsies for 344 patients with ARDS demonstrated that the presence of DAD was associated with a higher mortality [12]. The pooled odds ratio for mortality in ARDS with DAD compared with ARDS without DAD was 1.81 (95 % confidence interval 1.14–2.80). These data suggest that current clinical definitions could be improved to enrich for the presence of DAD. Such enrichment would identify a subset with a higher mortality, a greater chance of dying from ARDS per se, and perhaps greater benefit from lung-targeted therapies.

Table 1 Clinical features of ARDS with DAD versus ARDS without DAD
Full size table

Recent clinical trials provide indirect evidence in support of this possibility. For example, the PROSEVA investigators enrolled patients with severe ARDS and confirmed the findings of a prior patient-level meta-analysis showing benefit of prone ventilation limited to patients with severe ARDS [13]. The control group mortality was expectedly high. Cisatracurium, which likely reduces ventilator-induced lung injury, also appears to benefit patients with more severe ARDS, with the entire benefit seen in the lower tertiles of PaO2/FIO2 (<120) [14]. While these observations suggest that focusing on severe ARDS has been successful in part owing to enrichment for the presence of DAD, it is very likely that patients without DAD, principally bacterial pneumonia, were included in these trials and may have benefited from these lung-targeted approaches, as for lower tidal volume [10]. We will never know just how important enriching for the presence of DAD has been for these trials as tissue or biomarkers for DAD are not available.

Obviously, lung biopsy is not a practical prospective method for further enrichment in clinical trials nor is it a practical strategy to reduce the noise and improve the signal for detection of biomarkers specific for DAD. However, other strategies hold promise for detecting higher-risk and treatment-responsive subsets. Calfee et al. used latent class analysis to identify two ARDS endotypes using clinical features, routine laboratory tests, and plasma biomarkers [15]. The so-called hyperinflammatory endotype 2 had a higher mortality that was reduced with randomly assigned higher PEEP while endotype 1 had a lower mortality rate that worsened with higher PEEP. It is unclear if either endotype is more or less likely to have DAD. However, even without this knowledge this approach holds promise for identifying high-risk and treatment-responsive subsets.

In summary, recent evidence suggests DAD, when present in a patient with ARDS, is associated with a different clinical course, greater chance of death from hypoxemia, and perhaps a different response to lung-targeted treatments. While these findings support the historical link between ARDS and DAD, we do not believe that these observations support renaming ARDS as DAD. Neither are diagnoses, both require a search for the underlying cause, and DAD requires a biopsy which is simply not feasible for most research purposes. Efforts to better identify DAD, or an endotype like it, with improved clinical definitions, molecular imaging, and evaluation of the transcriptome/proteome/microbiome are needed to identify responsive subsets and begin the era of individualized treatments for patients with ARDS.