Predictors of diffuse alveolar damage in patients with acute respiratory distress syndrome: a retrospective analysis of clinical autopsies

We analyzed the 356 patients with ARDS who died and had a clinical autopsy between 1 January 1991 and 31 December 2010 in the 18-bed intensive care unit (ICU) at the University Hospital of Getafe in Spain. We systematically requested informed consent from patients’ relatives for both clinical autopsy and potential use of tissue samples for subsequent data analysis in research or teaching purposes, and the study was approved by the institutional review board.

All medical files of patients having had an autopsy were reviewed by two intensivists blinded to the autopsy findings to determine if patients fulfilled the clinical criteria for ARDS at the time of death. The diagnosis of ARDS was established by consensus of the two intensivists and resolved by a third intensivist in case of discrepancies. According to the Berlin definition [16], patients were considered to have the diagnosis of ARDS if they had (1) acute respiratory failure not fully explained by cardiac failure or fluid overload; (2) bilateral opacities not fully explained by effusions, lobar/lung collapse, or nodules on the chest radiograph or the CT scan; and/or (3) onset within 1 week after a known clinical insult or new/worsening respiratory symptoms.

We collected demographic variables (age, sex, severity score at ICU admission), reasons for ICU admission and mechanical ventilation, duration of mechanical ventilation, and date of the onset of risk factor for ARDS. Patients were considered to have a pulmonary risk factor for ARDS if they had a diagnosis of pneumonia, aspiration, inhalation, or lung contusion, as well as an extrapulmonary risk factor for ARDS if they had a diagnosis of sepsis, shock, multiple trauma or transfusion, or pancreatitis. Severity was defined according to degree of hypoxemia, and ARDS was considered severe if the ratio of partial pressure of arterial oxygen to fraction of inspired oxygen (PaO₂/FiO₂) was ≤ 100 mmHg, moderate if PaO₂/FiO₂ was between 101 and 200 mmHg, and mild if PaO₂/FiO₂ was between 201 and 300 mmHg, in all cases using a PEEP level ≥ 5 cmH₂O.

We also distinguished the more severe patients according to the four ancillary variables proposed by the Berlin definition for severe ARDS: (1) radiographic severity with more extensive opacities involving three or four quadrants compared with patients fulfilling the minimal criterion of “bilateral opacities” (two quadrants), altered respiratory system compliance (C_RS ≤ 40 ml/cmH₂O), large anatomical dead space as indicated by high corrected expired volume per minute (≥10 L/minute), and positive end-expiratory pressure (PEEP) ≥ 10 cmH₂O. Because plateau pressure was not measured over the whole period of the study, we calculated dynamic driving pressure (expressed as centimeters of water) as the difference between peak inspiratory pressure minus PEEP, and dynamic compliance of the respiratory system, and not static C_RS, as tidal volume divided by peak inspiratory pressure minus PEEP and expressed in milliliters per centimeter of water. Corrected expired volume per minute was calculated as the measured minute ventilation multiplied by the partial pressure of arterial carbon dioxide (PaCO₂) divided by 40 mmHg.

In all patients with ARDS, we collected ventilatory settings, arterial blood gas results, and chest radiography findings within 48 h before death. Radiographic severity was assessed by two intensivists, and extent of opacities was classified according to the number of quadrants involved (two, three, or all four). In case of disagreement, discrepancies were resolved by a consensus of three physicians, including a third intensivist or a radiologist blinded to the autopsy findings. Patients with four quadrants involved were those with bilateral and diffuse infiltrates without pleural effusion or lung collapse.

All our autopsy examinations were performed within 12 h of death using the same methodology for over 20 years, as already detailed in our previous studies in this field of investigation [4, 7, 17‐19]. After removal from the thorax, the lungs were inflated at a pressure of 35 cmH₂O and were fixed in blocks with 10% formalin. We took samples for microscopic analysis from each pulmonary lobe and additional samples from areas with macroscopic injuries. All samples were stored from the time of autopsy and were subsequently independently analyzed by two pathologists who had no clinical information. A third pathologist, also blinded, resolved any discrepancies. Criteria for DAD were hyaline membranes plus at least one of the following: intraalveolar edema, cell necrosis of type I alveolar cells, proliferation of type II alveolar cells (cuboidal cells), interstitial proliferation of fibroblasts and myofibroblasts, or organizing interstitial fibrosis [2, 3, 17].

Patients with ARDS and DAD and those with ARDS without DAD were compared using the chi-square test or Fisher’s exact test for categorical variables and Student’s t test or the Mann-Whitney U test for continuous variables when appropriate. Continuous variables were expressed as mean ± SD or as median and IQR (25th–75th percentiles). Qualitative variables were expressed as frequency and percent.

In this large study of clinical autopsies performed in patients who fulfilled all the criteria for ARDS at the time of death, a typical histological pattern of DAD was observed in less than half of the patients and was significantly more frequent in patients with ARDS for > 3 days, in most patients with hypoxemia, in those with more increased dynamic driving pressure, and in patients with extensive opacities involving the four quadrants. Therefore, the degree of hypoxemia, the driving pressure, and the radiographic severity could be strong non-patient-dependent markers of DAD and could be easily assessed at the bedside.

In addition to the degree of hypoxemia, the current Berlin definition proposes the following four ancillary variables to identify severe ARDS [16]: altered respiratory system compliance ≤ 40 ml/cmH₂O, high minute ventilation corrected by PaCO₂ ≥ 10 L/minute (indicating a large anatomical dead space), high PEEP levels ≥ 10 cmH₂O, and radiologic severity with extensive opacities involving at least three quadrants. In a previous study, patients with DAD had a lower PaO₂/FiO₂ ratio and lower respiratory system compliance [20]. However, the extent of opacities on chest radiographs was not assessed therein. After multivariate analysis including radiographic severity, altered respiratory system compliance (<40 ml/cmH₂O) did not remain significantly associated with DAD, whereas increased dynamic driving pressure and extensive opacities involving the four quadrants (and not three quadrants) constituted two factors independently associated with DAD in addition to hypoxemia severity.

We previously reported that DAD was markedly more frequent in most patients with hypoxemia, particularly after ≥ 3 days of evolution of ARDS [7]. However, in this previous study, we did not assess factors independently associated with DAD as we did in the present study. In addition to degree of hypoxemia and duration of evolution, we have now identified two other predictors of DAD: altered dynamic driving pressure and radiographic severity with diffuse opacities. Our data confirm that, even after adjustment for other variables, duration of evolution was independently associated with DAD. The absence of hyaline membranes during the first 3 days may be due to short duration of the process because hyaline membrane formation may take 2–3 days [2]. Moreover, patients with DAD had heavier lungs than the others, meaning that they had more alveolar and interstitial edema, one of the criteria for DAD. Although tidal volumes did not differ between patients with DAD and those without DAD, we found that the reduction of tidal volumes in the age of protective ventilation was associated with a parallel reduction of DAD in patients with moderate or severe ARDS, suggesting that low tidal volumes may attenuate DAD.

It has been shown that patients with ARDS with diffuse opacities may differ from those with lobar opacities [21‐24]. In a previous study, patients with diffuse ARDS were found to be more likely to have pulmonary ARDS, and also had lower respiratory compliance and higher mortality, than those with lobar ARDS [21]. The same authors found that lung morphology markedly influenced the effects of PEEP [22]. Indeed, PEEP induced marked alveolar recruitment in patients with diffuse opacities, whereas in patients with lobar opacities, PEEP induced mild alveolar recruitment associated with overdistention of previously aerated lung areas [22]. The authors suggested that in cases of lobar ARDS, there was mechanical loss of aeration in the lower lobes, whereas in cases of diffuse ARDS, an inflammatory mechanism may have occasioned aeration loss. In another study, effects of the prone position were more pronounced, with better oxygenation improvement in patients with lobar ARDS than in those with diffuse ARDS [23]. In a recent study, plasma biomarkers such as the soluble form of the receptor for advanced glycation end products or plasminogen activator inhibitor-1 were higher in diffuse ARDS than in lobar ARDS [24]. The high levels of these markers resulted from DAD and were associated with higher mortality. Consequently, patients with diffuse ARDS with extensive opacities involving the four quadrants could represent a homogeneous subphenotype of ARDS with behavior different from the others. Recently, Calfee and colleagues described several subphenotypes of ARDS that may have different responses to therapeutic strategies [25, 26]. In a subanalysis of a randomized controlled trial, the authors found that high PEEP levels decreased mortality only in a subgroup of patients characterized by particularly severe inflammation [26]. These patients were more likely than others to experience shock, metabolic acidosis, and higher levels of inflammatory biomarkers, and their outcomes were noticeably worse. ARDS is a syndrome involving highly heterogeneous patients, and identification of homogeneous subphenotypes could have an impact on the effects of different therapeutic strategies.

Obviously, clinical autopsy allows for analysis only of patients who have died, and therefore previously they have been among the most severely ill. Nearly half of the patients had severe ARDS at the time of death, and histological findings may differ from those in patients with less severe ARDS. However, in only 20% of cases was refractory hypoxemia the main reason for death. Although a high proportion of patients indeed died with ARDS, the proximate cause of death was septic shock, cardiac arrest, or withdrawal of life support. Because the latter was not always directly related to ARDS, these patients did not present the most pronounced respiratory severity at the time of death. In fact, they were in different stages of respiratory severity at the time of death, including 14% of those with mild ARDS, which means that our result could remain generalizable to patients with less severe ARDS. By contrast, open lung biopsy series included a subgroup of patients who may not have been representative; in fact, they were limited to those with persistent ARDS beyond 5–7 days without an identified cause [8‐14], which represented only a small proportion of patients with ARDS.

Our study was conducted in a single center and over a lengthy period of time, at a time when ventilator strategies were markedly evolving, especially after 2000 and the widespread use of low tidal volumes [27‐29]. A major limitation is that we could not collect accurate tidal volumes in milligrams per kilogram of predicted body weight over the whole period of the study, and consequently, although reduction in tidal volumes was associated with a parallel reduction in the incidence of DAD, we cannot assert that high tidal volumes really promote DAD. Similarly, we could not collect records of plateau pressure in all patients, and we calculated dynamic instead of static compliance and static driving pressure.

Another limitation was due to interobserver variability in the analysis of chest radiographs [30, 31], which may have led to misinterpretation of bilateral opacities. Normal lungs were found in 8% of the patients at autopsy examination, meaning that bilateral opacities seen on chest radiographs were probably atelectasis recruited during inflation at high pressure before microscopic analysis. In a previous study in which researchers compared the accuracy of chest radiography with computed tomography for the diagnosis of ARDS, many patients with ARDS based on computed tomography were detected using chest radiography [32]. However, sensitivity was higher for diffuse ARDS than for focal ARDS, and consequently diffuse opacities involving the four quadrants seem likely easier to identify.

DAD proportion may depend on the definition of ARDS that is being applied. In our study, the proportion of DAD was only 45% among all patients fulfilling the criteria for ARDS according to the Berlin definition (i.e., using “bilateral opacities” as radiologic criteria involving two, three, or four quadrants). In a previous study, the proportion of DAD reached 66% among all patients fulfilling the criteria for ARDS according to the American-European Consensus Conference definition and using diffuse opacities involving the four quadrants as radiologic criteria [4]. These two findings confirm the higher proportion of DAD among the most hypoxemic patients and those with diffuse infiltrates.