Skeletal Muscle and Peripheral Nerve Histopathology in COVID-19

Discussion

The pathophysiology of SARS-CoV-2–associated myopathy is poorly understood. The possibility of skeletal muscle infection by the virus has been considered because muscle expresses angiotensin-converting enzyme 2, which is a cell surface receptor used by SARS-CoV-1 and SARS-CoV-2 for host cell entry.^22,23 Negative SARS-CoV-2 IHC in muscle in our study argues against this hypothesis. However, in 1 study that examined the diaphragm muscle obtained from 26 consecutive autopsies of critically ill COVID-19–infected patients who died, SARS-CoV-2 RNA was found in the muscle in 4 cases (15.4%).²⁴ In situ hybridization localized the RNA to inside the sarcolemma. This discrepancy with our study findings may be explained by differences in methods to detect the virus or examination of different muscles.

Twenty-four of 35 patients in our study had evidence of an inflammatory or immune-mediated myopathy with necrotic fibers, inflammatory cell infiltrates, or MHC-1 immunostaining of nonnecrotic/nonregenerating muscle fibers. Our observations suggest that muscle damage occurs secondary to an inflammatory response, including damage from cytokines.

To date, literature on skeletal muscle histopathology in COVID-19 is sparse. One study reported muscle biopsy findings in 3 patients infected with SARS-CoV-2 who were clinically suspected of having critical illness myopathy.²⁵ Biopsies revealed scattered necrotic and regenerating fibers in 1 patient and rare atrophic and regenerative fibers in 2 others. No biopsies stained positive for MHC-1 or membrane attack complex (C5b9). In these patients, the histopathologic findings likely reflected the clinically suspected critical illness myopathy rather than COVID-19–associated myopathy.

Myositis was reported in a 58-year-old patient with SARS-CoV-2 infection with facial weakness, nasal dysarthria, and dysphagia.¹³ Muscle biopsy showed perivascular and endomysial inflammation and MHC-1 expression. The patient had a dermatomyositis-specific autoantibody detected in the serum. Viral invasion of muscle was not seen on electron microscopy. Another autopsy series reported myositis in 2 of 10 autopsies, but it is unclear how myositis was defined.¹⁴ A figure showed necrotic fibers undergoing myophagocytosis in 1 patient. We recently reported a patient with marked weakness and elevated CK (up to 30,000 U/L) who had overexpression of MHC-1 and MxA on perifascicular muscle fibers and capillaries, suggestive of a type-1 interferonopathy.¹⁵ One patient in our autopsy series (patient 35) also had MHC-1 expression on perifascicular muscle fibers but without MxA expression.

How do our findings compare to myopathy associated with other coronavirus infections? In a small post-mortem series of patients who died of SARS-CoV-1 infection, myofiber atrophy and necrosis were also the most common histopathologic findings in skeletal muscle.²⁶ MHC-1 and MxA staining was not performed. Findings were thought to reflect critical illness myopathy or specific changes of SARS-CoV-1–associated myopathy. Another series found vasculitis in muscle.²⁷ The virus was not detected in muscle with methods of viral culture, electron microscopy, IHC, or in situ hybridization.^26,28 One postmortem case report on a patient with cutaneous T-cell lymphoma and Middle East respiratory syndrome coronavirus showed necrotic fibers and an inflammatory infiltrate made up of CD68-positive histiocytes and mixed CD4-positive and CD8-positive T cells.²⁹ Electron microscopy identified virus-like particles in macrophages infiltrating muscle but not in muscle fibers.

In nerve biopsies, we found perivascular and endoneurial inflammatory cell infiltrates (neuritis) in 9 patients. History of diabetes was present in 4 patients, connective tissue disease in 2, and immune checkpoint inhibitor use in 1, conditions in which neuritis can be seen.^30,31 None had signs or symptoms of Guillain-Barre syndrome. We cannot exclude these conditions as potential etiologies of the observed neuritis (e.g., diabetic lumbosacral radiculoplexus neuropathy or diabetic amyotrophy), although we think these are unlikely to be coincidental occurrence in the patients with inflammatory cell infiltrates in their nerves and MxA expression on capillaries, which would not be seen in these disorders. SARS-CoV-2 was not found in our nerve biopsies by IHC, suggesting that the virus does not infect peripheral nerve.

In contrast to our findings, a CNS-focused postmortem series reported SARS-CoV-2 immunostaining in cranial nerves (glossopharyngeal and vagal nerves), albeit in only 2 of 40 patients, raising the possibility that viral infection of peripheral nerve may occur.³² In that study, SARS-CoV-2 immunostaining was found in undefined cells within the medulla from which these cranial nerves originate. Contiguous spread of infection from the medulla to these cranial nerves is conceivable.

It is possible that viral invasion of muscle and nerve occurred at an earlier stage in the illness and that active viral infection resolved by the time of death, although 22 patients (63%) had detectable SARS-CoV-2 in the lower respiratory tract by IHC, suggesting ongoing infection in other tissues. Viral RNA may be cleared from muscle and nerve tissue due to efficient type 1 interferon response (e.g., including MxA expression) or other mechanisms but not be cleared from more highly burdened organs such as the lungs.

Notably, MxA expression was observed in endothelial cells in 9 of 35 muscle and 7 of 35 nerve biopsies in our autopsy series, which is likely the result of the host response to SARS-CoV-2 infection. MxA is a type 1 interferon–inducible protein that is normally expressed in response to viral infections and prevents viral replication in the host. However, overexpression of type 1 interferons can be toxic, and abnormal expression of MxA in various tissues is seen in type 1 interferonopathies, including dermatomyositis, systemic lupus erythematosus, and idiopathic pernio (chilblains). As mentioned, we previously reported a patient with COVID-19–associated myopathy who had overexpression of MxA on perifascicular muscle fibers and capillaries, as typical of dermatomyositis and suggestive of a type 1 interferonopathy.¹⁵ Perniosis with MxA expression in endothelial cells and surrounding dermal and epidermal tissues has been reported in children and young adults late in the course of mild confirmed or presumed COVID-19 infection.^33,34 It is speculated that efficient induction of type 1 interferons and activation of the innate immune system quickly eradicate the virus and result in a mild infection but may cause collateral tissue damage. In critically ill patients, similar acral manifestations can occur due to severe thrombotic retiform purpura, in which abnormal MxA expression is not seen.³⁴ Such cases could represent an insufficient type 1 interferon response to the virus. These studies suggest that the type 1 interferon response is protective in viral infection, but overexpression may be toxic to certain tissues. Peripheral nerve and skeletal muscle appear to be bystander victims of the host response and cytokine dysregulation. We suspect that an exaggerated type 1 interferon response might be involved in some cases of COVID-19–associated myopathy and neuropathy. However, the lack of consistent expression in all cases of necrotizing myopathy, myositis, and neuritis and expression in some cases even without these features indicate that other cytokines may be involved in muscle and nerve damage. Abnormal serum levels of several cytokines have been detected in patients with SARS-CoV-2 such as type 1 and gamma interferons, interleukin-1 and -6, and tumor necrosis factor-α, among others.^2,15,35,36 We do not think the histopathologic findings in this study simply reflect nonspecific changes in muscle and nerve of patients with severe illness because muscle and nerve biopsies obtained from autopsies of 10 control patients revealed necrotizing myopathy in only 1 patient and no cases with myositis or neuritis.

With regard to laboratory data, we noted that the proportion of SARS-CoV-2–positive patients in our study with elevated CK was higher (74%) than has been reported in other studies (9%–33%).^1,-,7 This is likely explained by the fact that we used peak CK rather than admission CK levels, and our cohort comprised patients with more severe COVID-19. Higher CK levels are associated with poorer outcomes.³⁷ It is possible that cardiac injury contributed to the elevated CK because 16 of the 20 patients with elevated peak CK levels had elevated peak high-sensitivity troponin T levels. However, only 3 of these patients had evidence of acute cardiac injury on pathologic examination. Troponin T may not be specific for cardiac damage and can be elevated in patients with myopathy without cardiac injury,³⁸ concordant with our clinical experience.

There are limitations to this study. First, we did not perform targeted histopathologic examinations of clinically symptomatic muscle and nerve. We do not know whether psoas muscle and femoral nerve were clinically affected. Clinical information was obtained retrospectively, and documentation of neuromuscular symptoms and examinations was limited. These were extremely ill patients who ended up sedated and on ventilators. Three died in the emergency room, and 5 more died within 2 days of admission. The focus on the evaluations of these patients before intubation was stabilization. Second, due to laboratory biosafety concerns, specimens were entirely fixed in formalin for paraffin-embedded sections, and frozen tissue, which is routinely used to assess muscle histopathology, was not available, nor were plastic sections and electron microscopy for muscle and nerve. In addition, because this is a postmortem case series of patients who ultimately died of the virus, our results may not reflect the full spectrum of histopathologic findings in patients with various degrees of illness severity. Our findings may be skewed to those patients with the most severe infections. Last, as mentioned, viral RNA may have been cleared from muscle and nerve tissue before death, possibly due to a robust type 1 interferon response.

Our observations suggest that SARS-CoV-2 is frequently associated with inflammatory cell infiltrates and MxA expression in endothelial cells in both muscle and nerve, as well as necrosis of muscle fibers and abnormal MHC-1 expression in muscle. Although we did not measure cytokine levels in blood, the histopathologic abnormalities seen in our patients suggest that these findings may be secondary to the storm of cytokine release rather than direct viral infection of these tissues. Further studies are needed to better understand the pathogenic mechanisms of myopathy and neuropathy associated with SARS-CoV-2.