Pulmonary pathology of COVID-19: a review of autopsy studies

Abstract

Purpose of review 

COVID-19 lung injury is a common manifestation of severe illness. Lung tissue examination has been largely derived from autopsy – a combination of case reports, small and moderately sized series with international scope. Common and uncommon histopathology provides insight into the progression of severe, fatal disease.

Recent findings 

COVID-19 lung histology is most commonly diffuse alveolar damage as part of acute respiratory distress syndrome. Lung injury can be temporally heterogeneous, with patterns of healing alongside new injury. Viral studies, including immunohistochemistry, RNA in-situ hybridization, and tissue-based Polymerase chain reaction (PCR) assist in discerning complications of therapy (e.g. ventilator-associated pneumonia) from primary viral-induced injury. Response to viral infection produces systemic effects, and one major manifestation is thrombosis of micro-circulation and larger vessels. Less common patterns include neutrophil-rich inflammation, raising speculation that neutrophil extra-cellular traps may play a role in both viral control and exaggerated immune response.

Summary 

The heterogeneity of fatal cases- persistence of viral infection in lung, clearance of virus but severe lung injury, thrombosis, and exaggerated immune response – suggest that antiviral, antithrombotic, anti-inflammatory, and supportive therapy play a role in treatment, but that the patient-specific cause and timing of the lung injury is important in choosing intervention.

INTRODUCTION

The understanding of severe COVID-19 pathology includes the analysis of post-mortem examinations. The disease, while potentially affecting many organs with a systemic effect, is frequently associated with lung injury. This COVID-19 lung injury is a significant contributor to the cause of death in these patients, and as a result, its histology has been the focus of several autopsy series, This has led to insights into a heterogeneous lung disease that encompasses simultaneous patterns of exudative and proliferative organizing phases of diffuse alveolar damage (DAD), but in addition includes other patterns of acute, subacute and now chronic injury with prolonged illness leading to healing phases, including those with pulmonary fibrosis. The interplay of persistent viral-induced pulmonary injury and potential superimposed effects of secondary bacterial and fungal infections has made it difficult to emerge with a single pathway of viral-induced progression. Immune response and systemic effects, including the impact of large vessel and microscopic thrombosis on disease severity is significant, and may be more frequent in severe COVID-19 lung injury than other forms of acute lung injury. Modern techniques have played a role in teasing out these issues of pathogenesis, incorporating immunohistochemistry, RNA in-situ hybridization, fixed tissue polymerase chain reaction tests alongside traditional histomorphology and electron microscopy. Also, novel in this period is the incorporation of limited autopsy, in some cases guided by imaging techniques. 

Box 1: no caption available

REVIEW

A PubMed search of autopsy, lung, and COVID-19 yielded 556 unique cases, with 488 on which tissue sampling with histopathology were reported and 68 cases with clinical findings but lacking tissue pathology. This information was contained in 44 publications ^{[1▪,2▪,3–6,7▪,8–11,12▪,13–22,23▪▪,24–26,27▪,28–34,35▪▪–37▪▪,38–42,43▪,44]} from North America, South America, Europe, and Asia. Men exceeded women 2:1, with a mean age of 71.45 among studies with individual case data. The age range was wide, from 22 to 100, with 30 of 44 studies reporting at least one patient under the age of 60. Co-morbid conditions were common, with hypertension, cardiac disease and diabetes most frequently reported. Of note, chronic obstructive pulmonary disease was reported in 14% of cases and malignancy, both solid tumor and leukemia/lymphoma, in 14%. Cough was reported in 47% of cases, fever in 70% and dyspnea/shortness of breath in 68%. Of the 476 cases in which it was specifically mentioned, 50% (237) were mechanically ventilated during their care. The summary of clinical data from these publications is contained in Table 1.

Table 1 - Summary of patient characteristics among autopsy series

		N
Sex (M:F)	2.05:1	546
Age (mean)	71.45	475
Age range	22–100
Co-morbid conditions
Hypertension	52%	554
Cardiac disease	38%	554
Diabetes mellitus	32%	554
Chronic obstructive pulmonary disease	14%	554
Asthma	3%	554
Obesity	11%	504
Kidney disease	12%	537
Malignancy	14%	554
Presenting symptom
Fever	70%	259
Cough	47%	319
Dyspnea/shortness of breath	68%	273
Ventilatory support	50%	476

COPD, chronic obstructive pulmonary disease.

AUTOPSY HISTOPATHOLOGY OF PULMONARY COVID-19 INFECTION

Tracheobronchitis

Although descriptions of mucous plugging and airway obstruction were reported in the patient care setting ^[45], these were not prominently encountered in autopsy series, with one exception in which this findings was found with suppuration ^[23▪▪]. An inflammatory, often neutrophil-rich tracheitis and tracheobronchitis (Fig. 1a and b) is described in 47% of cases in one series; of those cases, the majority were never intubated, excluding airway trauma from intubation as a cause ^[2▪]. This finding was also described in two other series at a somewhat lower rate, but also as an acute tracheitis or tracheobronchitis ^{[13,37▪▪]}. Chronic airway inflammation was also reported ^{[2▪,23▪▪]}. These differences may be the result of the extent of large airway sampling.

FIGURE 1: Lung Pathology in COVID-19 – (a) Large airway demonstrating area of erythematous patches (black arrow) that appear as neutrophil rich exudates and mucosal ulceration in (b). (c) Diffuse alveolar damage, shown here with hyaline membranes of exudative DAD alongside type 2 cell hyperplasia and fibroblastic proliferation of organizing phase of DAD. (d) Some cases have predominant fibroblastic proliferation without new area of injury (a, Macroscopic image, b–d Hematoxylin and eosin, original magnification b, c ×100, d ×150).

Diffuse alveolar damage and lung weights

The main histologic pattern seen at autopsy in COVID-19 lung injury is DAD as part of the acute respiratory distress syndrome (ARDS). This type of ALI has well-defined stages that include early pulmonary edema followed by injury to the epithelial and vascular bed, resulting in epithelial cell necrosis and exudation of proteins from blood. This is morphologically described as hyaline membranes and begins about 48 h after an injury. This is followed by type 2 pneumocyte hyperplasia, followed by fibroblastic proliferation.

The phase of edema is often manifested by increased lung weights, and this was reported in nearly every series ^{[2▪,8,23▪▪,36▪▪]}. In one series, 90% of cases had combined lung weights over 1300 g. The exudative phase with hyaline membranes was frequently observed in COVID-19 lung injury, as was the proliferative phase with type 2 cell hyperplasia. Fibroblastic proliferation, and with time fibrosis (see later section), were reported. Although increasing days of disease tracked with the proliferative and fibroblastic phases, one of the features of COVID-19 disease is heterogeneity, with new hyaline membrane disease alongside some of the more subacute patterns ^{[2▪,3,5,10,13,16,17,20,46]}. This heterogeneity of injury indicates ALI persists even after other areas are already in a proliferative and healing phase (Fig. 1c). Although this can be explained by the many causes of DAD (e.g. sepsis, ventilator-induced injury, hypotension), results of tissue-based viral studies support the view that some of the new injury is caused by persistent viral infection. In other patients, virus in tissue becomes undetectable, and a healing phase characterized by interstitial fibroblastic proliferation ensues (Fig. 1d). Although the outcome of this phase is not certain in an individual patient, fibrotic lung is the endpoint in some. In addition, it is also clear that some patients have superimposed lung infections, septic shock, and thromboembolic disease as complications, rather than ongoing viral infection. The importance of these observations is that supportive care, antibiotics, and anticoagulation may have to accompany antiviral and anti-inflammatory strategies, and that this approach may vary from patient to patient depending on the pathway of their severe injury.

Other histologic features of DAD include the presence of multinucleation ^[6,16] that was not consistently associated with viral cytopathic change. In fact, characteristic viral inclusions have not been definitively identified, even with descriptions of eosinophilic bodies as well as basophilic inclusions (Fig. 2a and b). In fact, one series suggests that COVID-19 lung injury is indistinguishable from ARDS of other causes. The authors acknowledge that the rate of microthrombi was potentially increased ^[27▪].

FIGURE 2: Histopathologic features of COVID-19 lung injury (a) Multinucleation in likely type 2 pneumocyte. (b) Basophilc inclusion in type 2 cells are not clearly viral inclusions by other studies. (c) Resolving focus of organizing pneumonia, with intra-alveolar fibroblastic plug. (d) Squamous metaplasia in distal airway and alveolar space (Hematoxylin and eosin stains Original magnification, a, b ×150, c, d ×100).

Other histology

Organizing pneumonia (Fig. 2c) was seen in a subset of cases, characterized by intra-alveolar fibroblastic plugs. This was described in 50% of cases in one series ^[5], but less frequently in others ^[23▪▪]. In addition, squamous metaplasia of distal airway and in alveolar zones was seen (Fig. 2d), a feature generally associated with longer duration of disease ^{[2▪,6,16,23▪▪,30]}. In one case ^[28], features associated with severe acute and rapid progression, but with squamous metaplasia, raise the possibility that viral infection was present for some time before the decompensation and in support of this view, viral testing was positive one month prior to the severe decompensation. Several groups also describe pulmonary hemorrhage in a significant proportion of cases ^[23▪▪].

One series suggested the presence of acute fibrinous and organizing pneumonia ^[47,48], a pattern described in ARDS; this was observed in two other series ^[25,27▪]. Overall, this pattern was not commonly reported and may reflect a combination of organizing pneumonia, alveolar hemorrhage, and hyaline membranes.

Fibrosis

Although many of the patients had rapid progression and death within 14 days, a subset with longer disease course were also reported. These patients had organizing phase of DAD with progression to fibrosis ^{[5,6,10,21,35▪▪,39]}. In one case, serial testing for virus showed disappearance at 23 days, and autopsy at 45 days revealed lung fibrosis. Although fibrosis is a feature observed with longer duration of illness, it is unclear which patients will develop fibrosis.

Imaging

A few series attempt to correlate pathology with radiologic pattern. The combination of postmortem imaging and cryosamples showed ^[12▪] normal imaging corresponding to normal histology, but importantly, observed ground glass regions corresponding to early DAD, combinations of ground glass and consolidation with proliferative DAD, and as expected honeycomb pattern with late fibrosis. One study with postmortem imaging showed bilateral opacities and ’crazy paving’ as histologic DAD in different phases ^[15]; another study found ground glass opacity ^[49] with exudative DAD.

Thrombosis

Large artery, small artery, and capillary bed microthrombi appear commonly in patients with severe COVID-19 lung injury. Although such thrombi, including capillary bed microthrombi, can be seen in ALI patients of various causes, the frequency of this observation in lung as well as extrapulmonary sites for thrombosis of all types has been sufficiently consistent as to include this finding as typical of COVID-19 disease. Microthrombi are well described. Combining data from several multicase series, 72% of cases (163/227) had microthrombi, with platelet or fibrin and platelet thrombi ^{[5,10,13,16,17,19,20,23▪▪,46]}. In addition, several series describe small and large vessel thrombi, some with pulmonary infarcts ^{[2▪,16,17,36▪▪]}, but at a lower rate than microthrombi (Fig. 3a–c). Extrapulmonary thrombi, including heart, kidney, and prostatic bed are reported ^[4,24,50▪]. The degree and extent of thrombosis have contributed to the idea that COVID-19 infection leads to a systemic response with endothelial dysfunction.

FIGURE 3: Thrombosis in COViD19- lung injury (a) Fibrin and platelet thrombi in small vessels (b) Platelet rich thrombi in capillary bed. (c) Macroscopic thrombus, and associated wedge shaped hemorrhagic pulmonary infarct (a – Hematoxylin and eosin, original magnification ×50, b – CD61 immunohistochemistry, ×150, c – Macroscopic image).

Endotheliitis and endothelial sprouting

The presence of thrombi had led to speculation regarding vascular injury, inflammation, and vascular remodeling ^[5,43▪] possibly from COVID-19 infected endothelial cells. In one series ^[32], perivascular lymphocytes, neovascular growth, and microthrombosis was 9-fold more frequent than in influenza matched cases. Intussusceptive and sprouting angiogenesis were seen at a rate nearly 3-fold higher than in other ALI. This may match the morphology seen in non-ALI cases in a different series ^[3]. Although not an autopsy study, noninvasive examination of microvascular dysfunction demonstrates a reduction in capillary bed perfusion and was seen in severe, hospitalized COVID-19 patients. The reduction in capillary perfusion was speculated to be a correlate of microthrombosis ^[51]. In addition, examination of ICU and non-ICU patients suggests an endotheliopathy in severely ill patients ^[52].

THE ETIOLOGY OF ACUTE INFLAMMATION – A LINK BETWEEN PNEUMONIA AND THROMBOSIS?

Neutrophils and neutrophil extracellular trap formation

Studies have examined the inflammatory cell milieu in lung. The majority of cases described with DAD have a mononuclear cell infiltrate of lymphocytes and macrophages ^{[1▪,2▪,22,23▪▪]} some with interstitial lymphocytes and intra-alveolar macrophages. Also reported are a subset with neutrophils, some with distinct interstitial accumulation of neutrophils and thrombi ^[2▪,40,53] (Fig. 4a). Suppuration is described in many series, described as acute bronchopneumonia and assumed to be superinfection ^{[1▪,2▪,16,20,24,50▪]} and in a significant proportion, confirmed as such (Fig. 4b,c). One group performed sequencing for other pathogens and found 3 of 16 with bacterial infections. Cases remain in which a second pathogen is not identified. What this data underscores is the difficulty to resolve a neutrophilic infiltrate as primarily due to COVID-19 as opposed to superinfection, and introduces a pattern, likely a less common one, in which primary COVID-19 infection induces localized areas of neutrophilic inflammation.

FIGURE 4: Neutrophils in COVID-19 lung injury cases (a) Interstitial neutrophils are prominent and condensed on alveolar wall, with comparatively sparse intra-alveolar neutrophils (b). Macroscopic lung with abscess cavity (arrow) and patchy bronchopneumonia (arrowhead) that histologically in (c) represent acute suppurative pneumonia, in this case positive for Gram negative bacilli (not shown) (Hematoxylin and eosin stain, original magnification a ×100, c ×50, b, macroscopic image).

Thus, there remains some controversy as to whether the presence of neutrophils in the lung are a primary manifestation of innate immune response to COVID-19, or a superinfection with bacterial or fungal organisms. This problem of overlapping histology can be difficult to resolve ^[20]. However, the use of ancillary studies including special stains, postmortem culture, micro-organism sequencing and viral immunohistochemistry and ribonucleic acid in-situ hybridization can help sort out cases with and without super-infection. Review of the literature does suggest a pronounced neutrophil response without evidence of superinfection. For example, one series showed DAD in different stages, acute inflammation, and with virus in lung or large airway identified in five of these patients ^[17]. A case report from Japan noted interstitial and alveolar neutrophils without associated infection or mechanical ventilation ^[18]. Another series reported 9 of 14 cases with neutrophils, but only 3 with bronchopneumonia ^[13]. A distinct subset of reported cases have abundant interstitial neutrophils ^{[2▪,8,22,23▪▪,53]}. Nevertheless, a larger study employing multipathogen sequencing could help resolve this important issue.

This observation has led to studies examining the relationship of these neutrophils to the formation of neutrophil extracellular traps (NET). Demonstration of citrullinated histone H3 supports this hypothesis ^[54] and the combination of such staining with neutrophils and thrombi suggests a potential relationship between the neutrophils, NET formation and microthrombi ^[9]. The NET hypothesis was proposed from a clinical study with neutrophilia ^[55] as well additional tissue based studies with emphasis on microthrombi and neutrophil activation ^{[56▪,57,58,59▪]}. NET formation is thought to be involved in the killing of pathogens as a response to extracellular DNA; however, it is possible that excess NET formation and immune activation could cause damage to the lungs, inducing thrombosis.

Virus and lung

Several studies have incorporated tissue based viral testing in lung tissue as an adjunct to histopathology. In an early case report from Germany ^[26], DAD with hyaline membranes and microthrombi were reported and tissue for quantitative PCR tested from various sites, with only lung showing presence of viral RNA. In a series of 17 patients, 11 had immunohistochemistry evidence of virus in lung, overlapping with patterns of exudative DAD. One series showed a temporal relation of viral RNA limited to cases at <10 days with acute changes of DAD and microthrombi ^[46]. Other groups showed the presence of viral protein and RNA at later time points, albeit at lower rates with time but including reports of positives at nearly 4 weeks of illness ^[39]. In another study ^[9], median days of illness in lung virus positive cases was 14 days, with virus identified in one case over 30 days. However, this group also showed that virus negative lung tissue cases had median days of illness of 27 days, reinforcing a decreasing rate of lung virus positivity with time. Based on the literature review and cases studied at Weill Cornell, viral positive immunohistochemistry and ribonucleic acid in-situ hybridization was frequently associated with hyaline membranes, even if alongside proliferative DAD (Fig. 5). This indicates in some cases, ongoing lung injury is associated with evidence of virus, even at later time points of illness ^[2▪]. The molecular underpinnings of this observation are only beginning to be elucidated. One group ^[1▪] identified that high viral load patients had high interferon response, less pulmonary damage but early death, while in contrast a low interferon stimulated group had low viral load with later fatality at a point with more healing and re-modeling. Intra-alveolar hemorrhage was associated with CXCL9, 10 and 11 whereas the later duration of disease group had CD8 positive T-cells, macrophages and a higher rate of thromboemboli.

FIGURE 5: Ancillary studies in COVID-19 lung (a) An area of lung injury with hyaline membranes and type 2 pneumocyte hyperplasia (b) Immunohistochemistry for COVID-19 spike protein showing predominant hyaline membrane reactivity and foci of cellular staining (red) (c) RNA in situ hybridization for viral spike gene message showing numerous cells staining, many of which are morphologically type 2 pneumocytes (brown) (a, Hematoxylin and eosin Original magnification ×100, b, Alkaline phosphatase immunohistochemistry ×100, c, RNA in situ, RNAscope ACD probe for V-nCoV2019-S, ×100).

The relevance of these studies is that viral presence in lung tissue persists and is associated with exudative DAD, at a higher rate in the first 2 weeks of disease, but in some later than that timeframe. This co-localization of virus and persistent acute injury foci suggest viral clearance does not occur in a subset of patients.

Biopsy only autopsy

One of the recurring themes during the COVID-19 pandemic is restriction of autopsy or lack of proper autopsy facilities for infected cases. One solution was a biopsy based autopsy some with radiologic guidance ^[60]. Using ultrasound and 14-gauge biopsy cores, samples were obtained from a variety of organs, including lung. In one study, 5 autopsies of 60 expirations were obtained in this way, indicating that the low autopsy rate had other root causes such as consenting process, rather than solely autopsy extent. Virus was detected in lung tissue, with lower levels in liver, spleen and lowest levels in heart. In another series focusing on fibrosis, biopsies showed DAD in all phases in 93% of cases, with longer duration of disease, higher rate of ventilator support, longer hospitalization and younger age associated with fibrosis. One group performed post mortem cryosamples in 8 patients – with 63 attempts with 39 having successful acquisition of target tissue, indicating a need for training to achieve higher rate of success ^[12▪]. In a biopsy series from China, 9 of 10 had exudative DAD and suppuration, with 2 cases showing super-infection. It is likely that large airway disease and large vessel thrombi may be under-detected ^[44] in biopsy only autopsy. However, DAD, squamous metaplasia, organizing pneumonia suppuration, microthrombi and AFO – all patterns seen in full autopsies - were described ^[25,30,38]. Overall, biopsy only autopsies have a role if full autopsies are not possible.

CONCLUSION

COVID-19 lung injury at autopsy is predominantly DAD, with other injury patterns often co-existing. Patterns such as organizing pneumonia, squamous metaplasia and proliferative phases of DAD seen alongside exudative DAD indicate temporal heterogeneity of ongoing injury in some patients, whereas others progress to proliferative phase only, some with fibrosis. Thrombi, especially micro-thrombi are common. The interplay of persistent viral induced injury, immune response, systemic endothelial dysfunction, and thrombosis suggests that combination of disease states may vary and may impact therapeutic approaches.

Acknowledgements

The translational research laboratory in the department of pathology performed the immunohistochemistry and in-situ hybridization.

Financial support and sponsorship

None.

Conflicts of interest

There are no conflicts of interest.