Neuronophagia and microglial nodules in a SARS-CoV-2 patient with cerebellar hemorrhage

Symptomatic SARS-CoV-2 infection presents as a respiratory syndrome with upper and lower respiratory systems manifestations, characterized by cough, dyspnea, fever, chills, hyposmia, and ageusia [5, 20]. Male patients [9] as well as patients with comorbidities such as diabetes, obesity, hypertension, cardiac disease, pulmonary disease, and other chronic diseases [32], are prone to more severe manifestations. While the majority of infections are mild, severe infections can lead to serious end-organ damage due to respiratory failure, acute kidney injury, disseminated intravascular coagulation-like systemic coagulopathy [12], and thrombosis [16]. Neurologic sequelae have been reported in a number of SARS-CoV-2 patients including ischemic stroke, seizures, Guillain–Barre Syndrome, and acute necrotizing hemorrhagic leukoencephalopathy [3, 8, 14, 15, 19, 25, 35]. It is yet to be determined whether CNS manifestations are directly attributed to CNS infection by SARS-CoV-2, secondary to systemic effects in the context of hypoxia and multiorgan damage, or both.

Pathologic findings in the CNS from autopsy studies of SARS-CoV-2 patients are scarce. Most studies thus far lacked neuropathologic characterization altogether [1, 2, 30, 31, 34]. A case series describing 10 autopsies of SARS-CoV-2 decedents found no signs of encephalitis or vasculitis [22]. Another study of autopsy findings of four brains showed mild hypoxemic changes in three out of the four brains examined, but no encephalitis [21]. A recent case report of neuropathologic findings in one case of SARS-CoV-2 documented a picture of multifocal white matter hemorrhages, axonal damage in a variable perivenular distribution, and no significant perivascular inflammation. While a cortical organizing infarct was noted, neuronal necrosis, neuronophagia, microglial nodules, and vascular necrosis were not identified [24]. Most recently, a case series of 18 SARS-CoV-2 autopsies documented the presence of hypoxemic changes, perivascular inflammation in a subset of cases, and a single microglial nodule, however, the authors concluded there was no evidence of encephalitis or vasculitis. Real-time PCR identified viral transcripts in the frontal lobe, medulla and olfactory nerves in the majority of cases [29].

In this report, we describe the neuropathologic findings from a confirmed SARS-CoV-2 autopsy, where we describe a case of marked neuronophagia involving the inferior olives and dentate nuclei.

A 73-year-old man with hypertension and type II diabetes mellitus presented with sudden onset of headache, followed by nausea, vomiting, and loss of consciousness. He was fully functional at baseline and had no recent infectious symptoms prior to presentation. On the day of admission, he complained to his caretaker of a sudden severe headache while eating lunch. When a second caregiver arrived, he had become unarousable. Emergency medical services were called, and his initial vital signs revealed an oxygen saturation of 30%. He was intubated and oxygen saturation of 80% was achieved with bag and valve ventilation. During transportation to the emergency department, he went into cardiac arrest with pulseless electrical activity. During the resuscitation, out of suspicion for esophageal placement of the endotracheal tube, he was re-intubated. Return of spontaneous circulation was achieved after at least 17 min. His initial vital signs were notable for temperature 36.2 °C, heart rate 113 beats per minute, blood pressure 113/72 mmHg, and oxygen saturation 100% on mechanical ventilation. Chest radiograph showed hazy bilateral opacities, and nasopharyngeal swab for SARS-CoV-2 RNA was positive. Laboratory studies were notable for WBC 13.3 × 10⁹/L, hemoglobin 14.6 g/dL, platelets 346 × 10⁹/L, serum creatinine 1.16 mg/dL, and serum lactate 12.0. C-reactive protein (CRP) was elevated 8.7 mg/L (normal is less than 3 mg/L), and erythrocyte sedimentation rate (ESR) was 27 mm (normal is 0–10 mm). No other inflammatory studies were performed. Liver chemistries and coagulation studies were within normal limits. Computed tomography (CT) of the head showed a 6 × 4 cm right cerebellar intra-parenchymal hemorrhage (Fig. 1a), edema and compression of the medulla (Fig. 1b), and tonsillar herniation, The fourth ventricle was compressed and there was secondary obstructive hydrocephalus as well as intraventricular hemorrhage. CT angiography of the head did not reveal an underlying vascular lesion. With no improvement after 18 h, his care was transitioned to comfort measures and he died within minutes of palliative extubation.

The unique findings we describe in this report revealed marked neuronophagia and microglial nodules in the inferior olives and to a lesser extent the dentate nuclei, and only mild perivascular lymphocytic infiltrates. Parenchymal inflammation in the inferior olives and dentate nuclei has been described in neuroinvasive enterovirus 71 infection [33]. Radiologic evidence of encephalitis was described in a recent report of SARS-CoV-2 [26], however, the lack of prominent inflammation in our case makes encephalitis less likely. Moreover, microglial nodules with white matter perivascular hemorrhages were described by Hart and Earle [13] and Russell [27] and designated as “perivenous encephalitis”, which is associated with childhood mumps, measles, chickenpox, and less commonly vaccination. The inflammation in perivenous encephalitis is more pronounced than what we observed in this case, and prominent perivascular hemorrhages are lacking. The absence of prominent perivascular hemorrhage and demyelination rules out Acute Hemorrhagic Leuko-Encephalitis, which was recently inferred radiologically in a SARS-CoV-2 case [23].

We hypothesize that the marked neuronophagia in our case results from the cytokine-induced microglial activation, that primes microglia to phagocytose hypoxic neurons. SARS-CoV-2 infection is associated with increased systemic cytokine levels, including Interleukin-6 and Interferon-gamma [6, 17], which are known to activate microglia, and can up-regulate the expression of microglial receptors involved in phagocytosis [18]. Concomitantly, hypoxia can induce the expression of “eat me” signals on the surface of neurons, making hypoxic neurons more vulnerable to phagocytosis. Phagocytosis of viable or dying neurons has been described in hypoxic and inflammatory conditions [4]. In this way, we propose that the hypoxia and systemic inflammation act synergistically to induce neuronophagia. The main alternative hypothesis is the direct viral infection of neurons leading to neuronal death and neuronophagia. We find this alternative hypothesis unlikely because we did not identify viral proteins by immunohistochemistry, which detected viral proteins in the lung. Furthermore, viral RNA was extremely low in the brain and is likely explained by the presence of viral RNA in the blood within the specimens. This possibility is supported by our detection of viral RNA in the cerebellar blood clot. Moreover, viral infections such as Herpes Simplex and Cytomegalovirus are associated with a more prominent inflammatory response [28]. Therefore, a direct infection of neurons in our case is unlikely, especially that the medulla—where neuronophagia was most abundant—was devoid of viral transcripts. Another alternative hypothesis is that neuronophagia may be secondary to disruption of the olivary-cerebellar circuitry resulting in trans-synaptic degeneration of the olivary neurons. We do not favor this possibility because disruption of the Guillain–Mollaret triangle is generally associated with pseudo-hypertrophy of the inferior olivary nucleus [10]. This process does not involve neuronophagia, rather enlargement and cytoplasmic vacuolation of olivary neurons beginning three weeks after injury, followed later by astrocytic hypertrophy (gemistocytic astrocytes), and neuronal loss months later. In fact, in the acute phase (< 7 days), no histologic abnormalities except for pallor of the amiculum were described [11]. In addition, neuronophagia involving both inferior olives is not predicted from interruption of the circuitry on one side. Therefore, the clinico-pathologic findings in the case reported herein are incompatible with olivary pseudo-hypertrophy.

The cause of the cerebellar hemorrhage is most likely hypertensive vasculopathy. Perivascular hemosiderin-laden macrophages were identified in the cerebellar white matter adjacent to the bleed. It is interesting that the severity of these hypertensive changes was only mild in the white matter, thalamus, corpus striatum, and the kidney. We also cannot fully exclude SARS-CoV-2 vasculopathy as a potential contributor to the cerebellar hemorrhage. However, the vascular changes in this case are likely secondary to the herniation and subsequent acute infarcts. We did not identify definitive evidence of vasculitis or amyloid angiopathy.

Overall, the constellation of findings we describe are unique. To our knowledge, this report is the first to document extensive neuronophagia and microglial nodules in the inferior olives and dentate nuclei without pronounced inflammatory infiltrates in a confirmed SARS-CoV-2 autopsy.