Forty Postmortem Examinations in COVID-19 Patients

Two Distinct Pathologic Phenotypes and Correlation With Clinical and Radiologic Findings

Simona De Michele, MD; Yu Sun, MD, PhD; Mine M. Yilmaz, MD; Igor Katsyv, MD, PhD; Mary Salvatore, MD; Amy L. Dzierba, PharmD; Charles C. Marboe, MD; Daniel Brodie, MD; Nina M. Patel, MD; Christine K. Garcia, MD, PhD; Anjali Saqi, MD, MBA


Am J Clin Pathol. 2020;154(6):748-760. 

In This Article


Forty sequential lung specimens from postmortem examinations of patients infected with SARS-CoV-2 were evaluated. All patients had nasopharyngeal swabs, either premortem (n = 32) or postmortem (n = 8), that confirmed SARS-CoV-2 infection. Complete postmortem examinations, with the exclusion of the head in a few cases, were performed.

Clinical Characteristics and Imaging

Patients ranged in age from 38 to 97 years (median age, 71.5 years) Table 1. The majority were men (70%) and Hispanic or Latino (57.5%). Hypertension was present in 85%, diabetes mellitus in 50%, preexisting heart disease in 38%, and preexisting lung disease in 25%; one or more of these were present in 37 (92.5%) of 40 patients. Twenty-six (65%) patients received supplemental oxygen, and 23 (57.5%) were intubated or received invasive mechanical ventilation. The mean time from admission to death was 10.4 days (range, 0–45 days).

Laboratory data (including platelet count for 33, international normalized ratio on admission for 32, D-dimer for 26, troponin-T for 32, N-terminal prohormone of brain natriuretic peptide for 20, interleukin 6 [IL-6] for 24) were available for a subset of 40 patients. Chest radiographs were available for 32 of 40 patients, 26 had multiple chest radiographs, and three had chest CTs.

Pulmonary Pathology: Major and Minor Microscopic Patterns

Three major pulmonary findings included ALI in 29 (73%) of 40, IFPAs in 36 (90%) of 40, and vascular congestion and hemangiomatosis-like change (VCHL) in 20 (50%) of 40 Figure 1, Image 1, and Image 2. ALI comprised the histologic spectrum of DAD with an acute (exudative) phase demonstrating hyaline membranes with or without an organizing (proliferative) phase exhibiting interstitial fibroblastic proliferation as well as a single case with predominantly fibrin, compatible with acute fibrinous and organizing pneumonia (AFOP). When present, ALI was evident in 33.3% to 100% of H&E slides for each case (mean, 78.4%; median, 85.7%).

Figure 1.

The spectrum of lung pathology in coronavirus disease 2019 was categorized as "major" (A) and "minor" (B) pathologic patterns. ALI, acute lung injury; IFPA, intravascular fibrin or platelet aggregate; VCHL, vascular congestion and hemangiomatosis-like change. a Candida species, herpes simplex virus, and Aspergillus species. bCongestion, fibrosis, or chronic inflammation.

Image 1.

Lung with and without acute lung injury (ALI). A, B, Lung with ALI phenotype. A, Eosinophilic hyaline membranes, the hallmark of diffuse alveolar damage, line alveolar septa (arrows) (H&E, ×8.3). B, A pulmonary artery has an organizing thrombus (H&E, ×1.2). C, D, A non-ALI phenotype. C, An intravascular fibrin-platelet aggregate (arrow) in a small-caliber vessel in a background of otherwise unremarkable lung parenchyma (H&E, ×40). D, CD61 immunohistochemical stain highlights the platelets within small vessels and capillaries of alveolar septa (×20).

Image 2.

A, B, Lung from a region of vascular congestion and hemangiomatosis-like change (VCHL). The alveolar walls have patchy thickening of alveolar septa with a complex mesh-like framework (H&E, ×15). B, Thickened alveolar septa of VCHL and interspersed unremarkable septa (reticulin, ×40). C, Control case has vascular congestion but lacks the mesh-like framework and corresponding thickening of alveolar walls (H&E, ×15).

VCHL was present in 50% of the cases and characterized by a widening of alveolar septa by a proliferation of dilated and engorged capillary channels arranged perpendicular to each other, forming a complex mesh-like framework. IFPAs, the most common of the major pulmonary findings, were large and present in either pulmonary arteries (19/40; 47.5%) or small vessels (17/40; 42.5%) and were often focal (Table 1).

Multiple minor microscopic patterns were identified, including infarcts, bronchopneumonia, aspiration, microorganisms, underlying chronic lung disease (eg, interstitial lung disease), and minimal chronic inflammation unassociated with ALI (Figure 1).

Novel and Discordant Findings

ALI, one of the three major patterns and the most commonly reported in lungs of COVID-19 decedents, was present in 29 (72.5%) of 40 cases. Eleven (27.5%) of 40 cases did not show findings of ALI—designated "non-ALI"—and were categorized as "novel." This suggested two overarching phenotypic patterns of pulmonary pathology—ALI and non-ALI—which framed subsequent analyses.

In general, there was concordance between ALI and non-ALI and the chest radiograph findings (Table 1). For the ALI group, 96% (25/26) of the patients had evidence of alveolar infiltrates, and there was evidence of disease progression for 59% (13/22), stability for 36% (8/22), and regression in 4% (1/22). For the non-ALI group, alveolar infiltrates were absent (83%; 5/6) or minimal and inconsistent with ARDS (1/6; 17%), and none (0/4) had evidence of disease progression on chest radiographs (Supplemental Table 2).

ALI vs Non-ALI: Comparison of Pathologic, Clinical and Laboratory Data, Imaging, and Cause of Death

Clinical and laboratory data and imaging findings from these two groups—ALI and non-ALI—were compared. Greater imaging and laboratory data were available for the ALI group than the non-ALI group due to the shortened interval of time from admission to death in the non-ALI group.

Patients with ALI were hospitalized longer (P = .02), were treated with supplemental oxygen longer (P = .003), and had evidence of more alveolar infiltrates on the last chest radiograph (P = .01) than the non-ALI patients (Figure 2B and Supplemental Figure 1). Higher radiologic grade on the last available chest radiograph significantly correlated with presence of microscopic ALI (r = 0.79, adjusted 95% confidence interval [CI], 0.29–0.95; P = .0004; Figure 2A). Although there was a positive trend between radiologic grade on initial chest radiography and presence of ALI, it was not statistically significant (P = .06). There was no significant association between ALI and D-dimer, troponin-T, pro-BNP, and IL-6 levels; the non-ALI decedents had limited laboratory data for similar analysis. There was no statistically significant difference in O2 saturation at presentation, dyspnea, and Tmax between the ALI and non-ALI patients.

Figure 2.

Associations between clinical, laboratory, radiologic, and microscopic findings. A, Correlation matrix of clinical, laboratory, major microscopic, and radiologic data. Circle size and color intensity are proportional to correlation coefficient, with red representing positive correlations and blue representing negative correlations. *Statistical significance (Holm-adjusted P < .05). B, Violin plot and dot plots for nonbinary and binary data, respectively, of clinical, laboratory, microscopic, and radiographic data distributions by ALI status. For any given clinical feature, the violin and dot color are proportional to the –log10 (Holm-adjusted P value) of a Wilcoxon rank-sum or Fisher exact test for nonbinary and binary data, respectively, comparing the distribution of that feature in ALI vs non-ALI. Each circle corresponds to a single case. For binary data, 1 = presence, 0 = absence. ALI, acute lung injury; CPR, cardiopulmonary resuscitation; CXR, chest x-ray; IFPA, intravascular fibrin or platelet-rich aggregate; INR, international normalized ratio; pro-BNP, N-terminal prohormone of brain natriuretic peptide; Tmax, maximum temperature; VCHL, vascular congestion and hemangiomatosis-like change.

The two other major patterns—IFPAs and VCHL—were evaluated in the context of ALI and "novel" non-ALI phenotypes as well as independently to assess for correlation. There was a significant negative correlation between the presence of ALI and pulmonary VCHL (r = –0.62; adjusted 95% CI, −0.87 to −0.11; P = .005; Figure 2A). No statistically significant correlation was identified between the presence of ALI and IFPAs. All non-ALI cases (11/11) had concurrent VCHL and IFPAs (Image 1B). Meanwhile, 27 of 29 ALI cases had IFPAs (18), VCHL (2), or both (7); two had neither VCHL nor IFPAs.

A pathologic microscopic or macroscopic cause of death was identified in only two of 11 non-ALI decedents and attributable to pulmonary thromboembolism and intracranial hemorrhage. The cause of death for the remaining nine of 11 non-ALI cases was based on adjudication of clinical and pathologic findings Table 2 and included cardiac arrest or respiratory failure in eight and septic shock in one; all had mild to moderate myocyte hypertrophy and interstitial fibrosis; seven of nine had coronary artery disease.


Review of pulmonary H&E slides from seven ALI "controls" showed large IFPAs in two cases and small, focal IFPAs in five cases. The non-ALI controls, which were from decedents who had similar median ages, sex distribution, abbreviated hospital course, comorbidities, and cause of death as the non-ALI cohort, lacked IFPAs and had focal congestion but not the accompanying hemagiomatosis-like meshwork, which was present diffusely in nine of 11 and focally in two of 11 non-ALI cases (Image 2).