Catch me if you can: SARS-CoV-2 detection in brains of deceased patients with COVID-19.

In the American biographical crime film Catch Me If You Can, FBI agent Carl Hanratty goes all out to catch the notorious impostor and cheque counterfeiter Frank Abagnale Jr. Hanratty's dogged pursuit of the culprit bears striking resemblance to current COVID-19 research efforts to find evidence of changes that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection might leave in the brain. In The Lancet Neurology, Jakob Matschke and colleagues give a detailed account of the histological alterations related to COVID-19 in the CNS. Through meticulous detective work, they mapped the brain's immunoinflammatory response to viral infection and detected SARS-CoV-2 protein expression in a substantial percentage of autoptic brains of patients with COVID-19.

Matschke and colleagues analysed 43 brains from a large cohort of patients who died with COVID-19, focusing on inflammatory changes and detection of SARS-CoV-2. To find out which CNS cell types are prone to SARS-CoV-2 infection, the authors screened gene expression datasets for signatures related to viral entry and persistence. Their in-silico analysis showed high expression of angiotensin-converting enzyme 2 (ACE2) in oligodendrocytes and of transmembrane serine proteases 2 and 4 (TMPRSS2 and TMPRSS4) in neurons—genes that code for proteins crucially implicated in SARS-CoV-2 host-cell entry (ACE2) and proteolytic priming of the virus-decorating spikes (TMPRSS2). Interestingly, a study using human brain organoids, published in September, 2020, showed that SARS-CoV-2 can readily infect and kill neurons. The neuronal cell death upon viral infection was preceded by aberrant intraneuronal localisation and hyperphosphorylation of Tau protein, similar to the pathogenesis of Alzheimer's disease and other neurodegenerative diseases.

Matschke and colleagues used quantitative RT-PCR (qRT-PCR) and immunohistochemistry with antibodies against nucleocapsid and spike proteins to detect SARS-CoV-2. Whereas viral RNA was detected in 48% of cases, viral protein detection was positive in 40%. Overall, the authors found SARS-CoV-2, either viral RNA or viral protein (or both), in 51% of the brains investigated. Remarkably, SARS-CoV-2 presence did not correlate with the severity of neuropathological alterations. While the replicative and infective potential of the viral RNA remains unclear, the in-situ detection of SARS-CoV-2 proteins is an important finding, as it confirms the presence of the virus in the brain. In this context, concerns5, 6 that the comparably low viral genome levels detectable by qRT-PCR in brain tissue might be blood-derived deserve mention.

Of note, the authors found virus protein expression to be confined to the medulla oblongata and to cranial nerves originating from the lower brainstem (most likely glossopharyngeal or vagal nerve). Considering the capability of SARS-CoV-2 to infect human gut enterocytes as well as pneumocytes,7, 8 this finding is of particular interest, warranting future investigations of vagal nerve tissue as a potential viral CNS access route in COVID-19.

The study also identified pronounced, brainstem-accentuated microglia activation, confirming previous work. As these brain-resident macrophage-like innate immune cells are highly heterogeneous with regard to gene expression, regional abundance, and perhaps functions, it seems worth testing whether microglia activated in a COVID-19 context correspond to a specific subtype, expressing sets of genes reflective of particular functional states.

Given the complex pathophysiology of COVID-19, any autopsy study is bound to have limitations (varying post-mortem intervals, incomplete or lacking clinical data, etc) and the present study is no exception in that regard. Confounding factors, such as the multiple comorbidities present among older patients with COVID-19 and, equally important, common COVID-19 treatment modalities, such as invasive ventilation (which might promote cerebral microbleeds) or dexamethasone medication (known to modulate immune responses), have to be considered when interpreting neuropathological findings. In the context of dexamethasone, it is unfortunate that no data on steroid medication were used to investigate correlations between some of the findings. Likewise, in the absence of appropriate control cohorts, it remains unclear to what extent microglia activation and brain infiltration by cytotoxic T-lymphocytes represent COVID-19-specific findings. Both sparse lymphocytic infiltrates and microglia activation were recently documented in the brains of individuals without COVID-19, and they appeared to be particularly pronounced in septic cases.

At a time when a potential second wave of infections is increasingly becoming of global concern, the question of whether the neuropathological alterations in COVID-19 directly result from SARS-CoV-2 brain infection as opposed to reflecting sequelae of an overstimulated systemic immune response is of high clinical importance. Whereas the first scenario would support the use of remdesevir or other antivirals, anti-inflammatory modalities appear to be the treatment of choice once damaging immunoinflammatory mechanisms take over. Teasing apart these fundamentally different scenarios is an ongoing task for neuropathology experts. The work by Matschke and colleagues represents an important step towards navigating the complex pathophysiology of COVID-19 in the brain. Just like agent Hanratty, the authors have done a superb job closing in on the culprit.