Unspecific post-mortem findings despite multiorgan viral spread in COVID-19 patients

Coronaviruses, including severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV), cause severe acute respiratory failure, which is associated with high mortality rates [1]. The novel SARS-CoV-2 strain exhibits phylogenetic similarities to SARS-CoV and causes coronavirus disease 2019 (COVID-19), which has resulted in more than 540,000 deaths worldwide so far. As the pandemic has progressed, the pathophysiology of this viral infection has become clearer; in particular, it has been shown that SARS-CoV-2 can directly alter cell function by a link to the angiotensin converting enzyme 2 (ACE2) receptor, which is almost ubiquitous in the human body [2].

Nevertheless, the mechanisms behind the high mortality and severe organ dysfunction associated with COVID-19 remain poorly understood. Controversies exist regarding the occurrence of fatal complications, such as pulmonary embolism or diffuse endothelial injury [3, 4], as well as on the roles of direct viral cellular injury or concomitant comorbidities in the fatality of this disease [5].

In this setting, autopsy is of great importance to help physicians understand the biological characteristics and the pathogenesis of COVID-19. Most of the previously reported post-mortem findings focused on lung morphology and few data are available on complete post-mortem analyses of other organs [6, 7]. The aim of this study was therefore to investigate the presence of specific features of viral injury as well as the distribution of the virus in different organs of patients who died from COVID-19.

This post-mortem study showed several histopathological abnormalities in COVID-19 non-survivors; however, none of the findings was specific for direct viral injury, even though SARS-CoV-2 was detected in all examined organs using RT-PCR. We decided to perform complete autopsies rather than other techniques such as post-mortem core biopsies, so as to obtain a better overview of all organs, especially the lungs (we collected 6 samples from each lobe). This approach enabled us to document the considerable heterogeneity of histological lesions and of SARS-CoV-2 spread through the body.

The diagnosis of SARS-CoV-2-related organ injury is challenging; post-mortem histological findings were heterogeneous and often associated with chronic underlying diseases. In a previous autopsy study in COVID-19 patients [3], the authors reported that DAD associated with viral pneumonia was almost impossible to distinguish from that caused by bacterial pneumonia. No obvious intranuclear or intracytoplasmic viral inclusions were identified in another report [6]. Desquamation of pneumocytes and hyaline membrane formation are frequently described in ARDS of many different causes, especially in early-phase ARDS [13]. The presence of multinucleated cells with nuclear atypia is used to diagnose herpes virus infection in daily practice. As in previous reports [6, 14], we also observed the presence of multinucleated cells within lung alveoli in three patients; however, the significance of multinucleated cells is unclear and may not be specific of SARS-CoV-2 infection [15]. Finally, some of the microscopic features of these patients are compatible with organ changes secondary to shock or systemic inflammation and no histological finding could be specifically ascribed to SARS-CoV-2.

In the absence of typical post-mortem viral features, our results show that RT-PCR is feasible on FFPE blocks and could be used in post-mortem analyses to identify the presence of SARS-CoV-2 in multiple organs and to understand the spread of the virus within the human body. The discordant RT-PCR and IHC results for detection of SARS-CoV2 in the lungs may be explained by the different sensitivity of these assays, which was higher for the RT-PCR, whereas low-level viral replication might not be detected by IHC. Moreover, IHC was based on the only available antibodies, which are targeted against SARS-CoV. New antibodies against SARS-CoV-2 need to be developed to improve the accuracy of IHC in the analysis of tissue samples from suspected or confirmed COVID-19 patients.

Most of the previous post-mortem studies in COVID-19 patients were conducted using needle biopsies and were therefore rather limited in terms of sampling; our complete autopsy analysis identified considerable heterogeneity of SARS-CoV-2 spread through the human body and provides a more accurate description of macroscopic and microscopic organ alterations. As for previous coronavirus diseases [16, 17], the lungs are the most affected organs in COVID-19. However, DAD findings were highly heterogeneous, including both early-onset and additional late lesions. This finding could be explained by the heterogeneity of the pulmonary injury, including compliant lungs in the early phase and a more dense and non-recruitable lung in the late phase [18]. As some patients died outside the ICU without receiving mechanical ventilation, we could not estimate lung compliance before death. The heterogeneity could also reflect different treatments (e.g., fluid administration or corticosteroids) or different complications; as an example, half of the patients had concomitant acute pneumonia and it is difficult to conclude whether the DAD reflected the natural time-course of the viral disease or was secondary to superimposed complications, such as nosocomial infections. In a recent report, needle post-mortem biopsies suggested that COVID-19 is not associated with DAD but rather with an acute fibrinous and organizing pneumonia (AFOP), consequently requiring corticoid treatment [19]. A diagnosis of AFOP is based on the absence of hyaline membranes and the presence of alveolar fibrin balls; however, hyaline membranes are heterogeneously distributed in the lung parenchyma with DAD and complete lung analysis, not just biopsies, are necessary to exclude their presence. Moreover, AFOP may be a fibrinous variant of DAD [20]. The limitation of lung biopsy was also shown in another study, in which only 50% of lung samples were positive for SARS-CoV-2 using RT-PCR [21], when compared to almost 100% in our series. In addition, we did not find specific “endothelitis” as previously reported in a small case series [4]. Considering the heterogeneity of post-mortem COVID-19 associated lesions, molecular and IHC assessments are mandatory in the histological analysis of COVID-19 tissue samples.

Patients with COVID-19 often have altered coagulation and a prothrombotic status, with the possible development of acute pulmonary embolism (PE) [22]. In our study, three patients had PE, already diagnosed before death. Four patients had pulmonary infarction. In a previous study, acute PE was considered as the main cause of death in four patients [3]; however, the inclusion of patients who died before hospital admission and the lack of specific thromboprophylaxis during the hospital stay may account for the differences in the severity of PE when compared to our study. Although we frequently observed the presence of microthrombi in the lung parenchyma, this feature is also reported in other forms of ARDS, regardless of etiology [13, 23]. As such, whether diffuse pulmonary thrombosis is a main contributor of the fatal course of severe hypoxemia in COVID-19 patients remains to be further studied. In a systematic review of pathological findings in COVID-19, Polak et al. [24] identified a timeline in the histopathological findings in the lung, with epithelial (DAD, denudation and reactive pneumocytes atypia) and vascular (microvascular damage, thrombi, intra-alveolar fibrin deposits) changes present at all stages of the disease, but fibrotic changes (interstitial fibrous changes) only appearing about 3 weeks after the onset of symptoms. Few patients had fibrosis at early stages, and in these cases, it was likely because of pre-existing lung disease. Our results are consistent with those of Polak et al. [24] except for the lack of late fibrotic changes, which may be related to the use of anti-inflammatory drugs at high doses for nearly all our patients (16/17).

We did not observe specific viral organ injury, such as myocarditis, hepatitis, or encephalitis. The cases of “acute cardiac injury” reported in COVID-19 clinical studies [25] do not necessarily translate into myocarditis or acute myocardial ischemia (only two had acute myocardial ischemia), similar to data reported in septic patients (i.e., elevated troponin without overt cardiac ischemia) [26]. However, using RT-PCR, we found the virus in almost all the examined organs; this suggests that the virus can bind to most cells, probably via the ACE2 receptor, which is ubiquitous, but may not directly cause organ injury. As extra-pulmonary direct viral injury (e.g., encephalitis, hepatitis, or myocarditis) has only been reported in very few cases, we suggest that SARS-CoV-2 infection may be just the trigger for an overwhelming host response, which could secondarily result in COVID-19-associated organ dysfunction. As RT-PCR might just detect residual viral genome, it remains unclear whether this represents active viral replication into the tissues or previous cellular infection, without clinically relevant significance [27].

This study has several limitations: (i) we only included patients who had had a positive RT-PCR on nasopharyngeal swab and/or broncho-alveolar lavage. To ensure that only true positive cases were enrolled, we decided not to include three patients who had had thoracic CT-scan findings suggestive of COVID-19 but had negative RT-PCR results. This limitation in our study reflects the difficulty of diagnosing COVID-19 on a clinical basis; (ii) the sample size was relatively small, and autopsies were only carried out from 72 to 96 h after death. This delay did not allow us to properly analyze the gastrointestinal tract and kidneys, which showed signs of autolysis; in particular, acute tubular injury in the proximal tubules was indistinguishable from autolysis; (iii) we could not determine the time-course and/or sequence of organ spread of the virus and no specific hypothesis regarding how SARS-CoV-2 spreads (e.g., hematogenously) could be identified; and (iv) the time to death differed from patient to patient as did the course of the disease and treatments received, which limits a precise clinical-pathological correlation of histological findings related to COVID-19. Finally, we did not evaluate specific mechanisms involved in the pathogenesis of organ injury.

These results underline the heterogeneity of organ injuries during COVID-19 disease and the absence of specific SARS-CoV-2 lesions. Using RT-PCR, SARS-CoV-2 could be detected in all organs, even those without evident microscopic lesions.