COVID‐19 and the peripheral nervous system. A 2‐year review from the pandemic to the vaccine era

Abstract Increasing literature has linked COVID‐19 to peripheral nervous system (PNS) diseases. In addition, as we move from the pandemic to the vaccination era, literature interest is shifting towards the potential association between COVID‐19 vaccines and PNS manifestations. We reviewed published literature on COVID‐19, COVID‐19 vaccines and PNS manifestations between 1 January 2020 and 1 December 2021. For Guillain‐Barré syndrome (GBS), isolated cranial neuropathy (ICN) and myositis associated with COVID‐19, the demographic, clinical, laboratory, electrophysiological and imaging features were included in a narrative synthesis. We identified 169 studies on COVID‐19‐associated complications, including 63 papers (92 patients) on GBS, 29 papers (37 patients) on ICN and 11 papers (18 patients) on myositis. Additional clinical phenotypes included chronic inflammatory demyelinating polyneuropathy, vasculitic neuropathies, neuralgic amyotrophy, critical care‐related complications, and myasthenia gravis. PNS complications secondary to COVID‐19 vaccines have been reported during randomized clinical trials, in real‐world case reports, and during large‐scale surveillance programs. These mainly include cases of GBS, Bell's palsy, and cases of neuralgic amyotrophy. Based on our extensive review of the literature, any conclusion about a pathophysiological correlation between COVID‐19 and PNS disorders remains premature, and solely supported by their temporal association, while epidemiological and pathological data are insufficient. The occurrence of PNS complications after COVID‐19 vaccines seems limited to a possible higher risk of facial nerve palsy and GBS, to a degree that widespread access to the ongoing vaccination campaign should not be discouraged, while awaiting for more definitive data from large‐scale surveillance studies.


| INTRODUCTION
An increasing body of literature, including cohort studies, [1][2][3][4][5][6][7][8][9] has linked COVID-19 to the development peripheral nervous system (PNS) diseases. However, findings are divergent due to methodological differences and largely variable sample sizes. Few studies used a prospective design 8,10 and focused on defined diagnoses rather than symptoms alone. 5 Some studies relied on self-administered questionnaires and others on hospital records, and in some cases, the diagnosis was not confirmed by neurologists. 8 Individual diagnoses were not always supported by laboratory, electrodiagnostic (EDX), and pathology findings. As a result, PNS involvement was quite variable, ranging from 1.3% to 9.5% of cases 4,8 if individual diagnoses were considered (eg. neuropathy, myopathy, etc.), and up to 70.2% if individual symptoms were included (eg. myalgia, paresthesia, etc.). 3 In a very large retrospective cohort including 1760 COVID-19 patients from a single epidemic hotspot (Bergamo, Italy), 6  Here, we present a comprehensive narrative on 169 studies published between 1 January 2020 and 1 December 2021 on PNS involvement. Our main aims were investigating the association between COVID-19 and PNS diseases and understanding whether COVID-19 had any clinically meaningful impact on clinical presentation, diagnosis, and therapeutic approaches. Furthermore, as we are moving from the pandemic to the vaccination era, we provide an overview of the potential association between COVID-19 vaccines and PNS diseases discussing the findings reported so far.

| METHODS
Given the extent and heterogeneity of the topics reviewed in this paper, we aimed to provide a synthetic albeit comprehensive narrative on the published literature. However, our approach was not meant to be systematic, as commonly defined by Cochrane and PRISMA statements. A systematic approach has been attempted in the past, during the early and later stages of the pandemic, but on GBS cases only. 11,12 The data we extracted therefore were not incorporated in a metaanalysis, but were instead the basis for our expert opinion commentaries.
We restricted our search to peer-reviewed studies, published in English, and importantly, to papers published between 1 January 2020 and 1 December 2021.

| Pathophysiological insights into PNS involvement
The causal association between COVID-19 and nervous system manifestations has been solely inferred from their temporal co-occurrence. Two patterns have been described: (a) neurological complications occurring together with COVID-19 symptoms and suggesting a direct viral mechanism ("para-infectious" hypothesis), such as neuroinvasion; (b) neurological complications developing after the initial infectious symptoms and supporting indirect mechanisms ("post-infectious" hypothesis), likely immune-mediated.
The "post-infectious" immune-mediated hypothesis is supported by evidence that COVID-19 causes a proinflammatory state due to the release of multiple cytokines, such as IL1, IL6, and TNF, as well as immune-cell hyperactivation. 14 The umbrella term "cytokine storm" has been used to describe this phenomenon, although its appropriateness for COVID-19 is still debated. 25 In the lung, this has been linked to the progression towards acute respiratory distress syndrome (ARDS). 25 This mechanism has been proposed for other systemic complications of the disease, such as skin vasculitis, Kawasaki-like syndrome, myocarditis, and hemophagocytic lymphohistiocytosis. 25 Similar to other systems, vasculitis may affect cerebral small vessel causing stroke. 26 In the PNS, individual reports of multiplex mononeuropathy suggested possible vasculitis that, however, could not be confirmed due to the lack of neuropathological data. [27][28][29][30] The isolation of pathogenic anti-neuronal antibodies (eg, anti-contactin-associated protein 2), 31 has been invoked as a proof of the immune-mediated mechanism underlying post-COVID-19 myelitis and encephalitis. 14 Although uncommon in our review (Table 1), the presence of disease-specific antibodies, such as anti-ganglioside antibodies has been observed in post-COVID neuropathies. The shared pathogenetic hypothesis is that the molecular mimicry between SARS-CoV-2 surface proteins and self-antigens may lead to the production of autoantibodies targeting neuronal antigens or nodal/ paranodal proteins in the CNS and PNS, respectively. Although there is clinical evidence that some of the high-affinity SARS-CoV-2-neutralizing antibodies cross-react with human self-antigens, including self-antigens found in the CNS, their ability to cross the brain-or nerve-blood barrier has not been demonstrated. 32 By using in silico analysis, Keddie et al demonstrated that there is no linear homology between SARS-CoV-2 proteins and any axonal or myelin surface proteins, thus, making the molecular mimicry hypothesis unlikely. 33 An alternative immune-mediated hypothesis has been proposed by Suh et al. 23 The authors conducted a post-mortem histopathological study on the psoas muscle and femoral nerve of 35 patients who died of severe COVID-19 compared to 10 critically-ill patients who were negative for SARS-CoV-2 but died during the COVID-19 pandemic. 23 They observed overexpression of the major histocompatibil- An additional effect of uncontrolled systemic inflammation is the occurrence of coagulopathy resulting mainly, although not exclusively, in venous thromboembolic events. 35 In the CNS, this has been linked to an increased incidence of stroke in specific epidemiological scenarios, while its significance for PNS complications remains unclear.
Some of the neuropathies secondary to COVID-19 in our review could be secondary to thrombotic mechanisms, but much needed pathological data remain unavailable.

| Guillain-Barré syndrome
We identified 63 publications and 92 patients, 36 100 and UK (25 cases, including 13 "definite" and 12 "probable COVID-19"). 33 We did not include these additional 66 cases in Table 1 104 In the UK and Italian COVID-19 cohorts, a higher than expected frequency of AIDP was noted, 33,99 although it was statistically significant only in one study. 99 This likely reflected a higher than expected prevalence of axonal forms in their GBS control cases (up to 41.2%; seven out of 17 cases). 99 In a separate study on the same Italian COVID-19 cohort, the electrophysiological features of AIDP were compared to non-COVID-19 AIDP. 107 Distinctive features among COVID-19 patients were higher percentage of cases with absent F waves, which was attributed to motor neuron hypoexcitability, and increased duration of distal compound motor action potential (CMAP) without changes in distal latencies, which was interpreted as conduction slowing within muscle fibers. A major confounder in this study was that a large proportion of COVID-19 AIDP patients had a critical illness, which in some of the cases was due to the underlying lung infection. 107 Laboratory testing disclosed albumin-cytological dissociation in 75.3% of cases, comparable to the 64% seen in the overall GBS population, 104  Unexpectedly, among the 11 MFS cases only one tested positive for anti-GD1b 93 and none for anti-GQ1b, as compared to 90% positivity in non-COVID-19 MFS patients. 104 The same finding was reported in the UK cohort. 33 Brain and spine MRI were performed in 28 and 31 cases. Common findings, although present in less than 30% of cases, included enhancement of cranial nerves, spinal nerve roots, and cauda equina.
Far from being specific to COVID-19, these findings added diagnostic certainty to GBS diagnosis.
The initial concern that intravenous immunoglobulin (IVIG) might impair the humoral immunity towards SARS-CoV-2 prompted some clinicians to prefer plasmapheresis (PEX) as first-line therapy. However, there was no report of clinical deterioration after IVIG, which was the preferred treatment in 81.3% of cases (74 patients). This was the same approach reported in the GBS/COVID-19 cohorts. 33,99,100 The data on clinical outcomes were scarce among published cases.  33,99 However, these percentages are higher than the 20% reported in the literature for the general GBS population. 104,108 The mortality rate among COVID-19/ GBS cases was 6.5% (Table 1), similar to reports in the general population. 105 No deaths related to GBS and neuromuscular weakness were reported in the three cohort studies. 33,99,100 Overall, based on our extensive review of published literature, GBS phenotype among COVID-19 patients did not show distinctive features. This conclusion was also reached by the UK cohort study. 33 Few clinical findings, such as frequent need for ICU stay 99,100 and invasive ventilation, 33 and possibly more severe disability outcomes ( Larger and longer case-control studies and surveillance data from multiple geographic regions will ultimately be able to prove any epidemiological association between COVID-19 and GBS. In this regard, a recent international prospective cohort study by the International GBS Outcome Study consortium enrolled incident GBS cases between 30 January 2020 and 30 May 2020 and found no increase in patient recruitment during the pandemic. 110 A higher prevalence of COVID-19 was noted among GBS cases when compared to the general population, but this could have been secondary to substantial recruitment bias. 110

| Cranial neuropathies
Isolated or multiple cranial neuropathies not associated with polyneuropathy or other neurological disorders have been reported (Table 2). 93 cytic infiltrate and axonal damage on brain autopsy. 112 Evidence that these symptoms may be transitory in the majority of patients has suggested a competitive mechanism on the olfactory receptors rather than a permanent cell damage. 137 For yet unknown reasons, olfactory symptoms are strongly associated with gustatory dysfunction (ie, ageusia or dysgeusia). 136 We identified two literature reports of isolated optic neuropathy. 113 The facial nerve has attracted much interest given the frequent occurrence of gustatory symptoms among COVID-19 patients.
Reports of severe unilateral dysgeusia leading to the diagnosis of ipsilateral facial palsy however remain isolated. 68 Three case of hypoglossal nerve neuropathy have been reported. [132][133][134] All occurred unilaterally and were temporally related with either endotracheal intubation or prone positioning, suggesting a iatrogenic etiology.
Based on available evidence, any direct etiological association between COVID-19 and cranial neuropathies seems inconclusive.
Larger and longer case-control studies will be needed to address any causal link.

| Chronic inflammatory demyelinating polyneuropathy and other neuropathies
Because GBS patients reported in Table 1 were not prospectively followed up, it is unclear whether some of them were in fact acuteonset chronic inflammatory demyelinating polyneuropathy (CIDP), which is reported in up to 5% to 16% of patients in pre-COVID-19 studies. 147 In our literature review, we did not find any report linking COVID-19 to a new diagnosis of CIDP.
The potential of COVID-19 to precipitate CIDP has been pro- with a systemic immune-mediated process similar to multisystem inflammatory syndrome. 28 The potential of COVID-19 to cause small fiber neuropathy (SFN) has been postulated based on the occurrence of autonomic dysfunction among COVID-19 patients presenting with GBS (Table 1). In one of such cases, autonomic dysfunction in the form of profuse sweating, constipation and erectile dysfunction preceded motor symptoms. 97 In the ALBACOVID registry, 2.5% of patients were diagnosed with autonomic dysfunction. 2 Although growing literature interest is directed toward SFN in "long COVID" (see below), in our review we did not find any additional literature report on isolated SFN being diagnosed in the acute setting.

| Muscle and neuromuscular junction
Myalgia and asymptomatic CK elevation are common findings among Growing literature has reported the occurrence of myositis in the setting of COVID-19 (Table 4). 10,157-164 These include patients with classic proximal myopathy, 10,157-159,164 cases with marked bulbar involvement, 160 presentations consistent with dermatomyositis, 161 including cases with amyopathic dermatomyositis and interstitial lung disease. 165,166 In the majority of cases, the diagnosis of myositis was determined based on clinical presentation supported by laboratory findings (ie, CK elevation, and, when available, myositis-specific autoantibodies) in the setting of a molecular diagnosis of viral (ie, SARS-CoV-2) infection. In few patients, the diagnosis was confirmed by muscle biopsy 10,160 and/or muscle MRI. 158,160,162 In one case, electron microscopy ruled out direct viral invasions as pathophysiological mechanism of muscle damage. 160 A dramatic case of limb ischemia complicated by severe muscle injury, inflammation and compartment syndrome was attributed to COVID-19-induced hypercoagulability. 163 In a case series of COVID-19-associated paraspinal myositis, 162 seven out of nine patients who underwent spine MRI for back pain, lower extremity weakness, or lower extremity paresthesia were found to have edema and enhancement of the paraspinal muscles (ie, erector spinae and multifidus paraspinal muscles) at the lumbar level.
Although the clinical relevance of these finding was unclear, the authors hypothesized that myositis could be relatively common in COVID-19 patients. It cannot be ruled out that the paraspinal involvement could have been secondary to a protracted immobilization among severe COVID-19 patients.   184 and on similar experiences with influenza A and B infection, early administration of IVIG has been proposed as a potential preventative intervention for CIM/CI, although evidence is still limited to individual case reports. 185 Prone positioning has been found beneficial in ARDS and successfully translated to the COVID-19 care in the ICU, but it has also posed unexplored challenges. 186 In the setting of PNS disease, entrapment neuropathies have been the most common complication. [187][188][189][190][191] Among 83 patients admitted for COVID-19-related ARDS and requiring prone ventilation, 12 (14.5%) developed this complication. 187 The most frequent sites of injury were ulnar nerve (28.6%), radial nerve (14.3%), sciatic nerve (14.3%), brachial plexus (9.5%), and median nerve (9.5%). A similar study in Italy involved 135 COVID-19 patients requiring prone ventilation of whom 7 (5.2%) developed entrapment neuropathies, with again the ulnar nerve (five out of seven) and the brachial plexus (two out of seven) being the most frequently affected.
In the majority of cases, axonotmesis was evident on neurophysiological exams. 188 These percentages are higher than expected based on the clinical trials that have validated the use of prone ventilation in the pre-COVID era. 186 One hypothesis is that patients with COVID-19 ARDS may be more vulnerable to peripheral nerve injury, but no control patients were included to address this question in both studies. 187,188 Long and repeated prone positioning, and the comorbidities associated with severe COVID-19 (eg, diabetes, obesity, old age) rather than direct mechanisms could explain a predisposition to more frequent nerve injury among COVID-19 patients.
Compression of the lateral femoral cutaneous nerve at the level of the anterior-superior iliac spine or inguinal ligament may be a relatively uncommon but specific complication of prone positioning. 192,193 Additional complications have been linked to nerve injury during endotracheal tube insertion or as a result of its displacement during prone positioning. A case of Tapia syndrome (ie, concomitant paralysis of hypoglossal and vagus nerves) has been described in one COVID-19 patient. 132 At least one of the two cases of hypoglossal paralysis reported in Table 2 was likely due to orotracheal intubation and prone ventilation rather than to multineuritis as hypothesized. 133  has been reported in up to 38% and 18% of patients, respectively. 196,197 These findings are supported by impaired performance on neuropsychological testing and by evidence of frontal and parietal hypometabolism on FDG-PET. 198,199 Proposed mechanisms include long lasting neuronal damage caused by hypoxia, neuroinflammation, or virus permanence. 200 More recently, an increasing body of literature has suggested an involvement of the PNS during the later stages of COVID-19. As pointed above, anosmia and dysgeusia besides being common early COVID-19 symptoms, seem to persist in up to 27% of patients after the acute phase, possibly suggesting irreversible damage to the nerve terminals or the sensory receptor cells. 197 Pain is one of the most common long-term PNS symptoms after COVID-19, reported by up to 30% of patients depending on the cohorts. 200 Localized pain, such as chest pain, joint pain, and headache is the most frequent complaint, but there is an increasing number of reports on a more diffuse and ill-defined pain among long-haulers, frequently associated with descriptors such as fatigue, myalgia, and paresthesia. 196,197,200 One hypothesis is that the release of proinflammatory cytokines during the acute infection may cause hypersensitization of peripheral nociceptors followed by plastic changes and central sensitization during the chronic stage.

| Long COVID and PNS involvement
Muscle atrophy seems to be an early feature of severe COVID-19, in possible relation to the release of proinflammatory cytokines (TNF-alpha, IL-1 and IL-6), a mechanism that has been well established in other diseases, such as AIDS and cancer, where muscle loss is a prominent symptom. 201 Additional mechanisms, specific to COVID-19, could be prolonged immobilization with type 2 muscle atrophy, use of high-dose steroids and neuromuscular blockade, and nutritional deficiencies related to prolonged feeding assistance. Whether these manifestations are reversible and their long-term impact on COVID-19 patients remains to be determined, and prospective studies are still lacking.
Fatigue has been described in as high as 53% of patients at The spectrum of symptoms associated with long COVID has prompted comparisons with myalgic encephalomyelitis or chronic fatigue syndrome (ME/CFS). 208 This is not surprising as ME/CFS could be secondary to viral infections such as EBV, rotavirus, or HHV-6, among others. 209 One may hypothesize that at least some of the symptoms observed with long COVID could be a non-specific response to an infectious (viral) illness in predisposed individuals, as it has been proposed for ME/CFS. 208 Despite the mounting pressure from the public opinion, which parallels the increasing frequency of referrals to neurology for "long-COVID" symptoms, the quality and quantity of literature on this topic is still limited. Many questions remain unanswered, including the temporal criteria for defining "long-COVID" itself, whether this is a single entity or an umbrella category for multiple and unrelated presentations, and, more importantly, whether it is secondary to a non-neurological pathological process, such as a psychiatric disorder (such as post-traumatic stress disorder or depression), or expression of the pulmonary and/or cardiac involvement in the early stages of the disease. The potential of COVID-19 vaccines to cause GBS, particularly the adenovirus vector-based, vaccines has been initially suggested by cases that occurred during the phase III clinical trials. The association between GBS and vaccination has been long debated since initial reports of increased incidence of GBS after the swine influenza vaccine during the USA/New Jersey 1976 vaccination campaign. 104 Thereafter, similar concerns have been raised for multiple vaccines, including oral polio, DPT, rabies, hepatitis B, and quadrivalent conjugated meningococcal vaccines. 220 However, large case-control studies have failed to show causal association. 220  where it was available), whereas in a single case it was AMSAN.

| COVID-19 vaccines: An overview
Patients were treated with conventional therapies (ie, IVIG) and the outcome was positive, with GBS disability score ≤ 2 in 10 out of 14 cases where this information was available. The occurrence of GBS after mRNA-based vaccines has been also reported, although less commonly (n = 5 cases). 223,227,228,233,234 237 Patients presenting with GBS after AstraZeneca vaccine had more frequent facial and bulbar involvement than the historical cases, and more commonly they had the bifacial weakness and distal paresthesia GBS variant, similar to the above-mentioned reports. [237][238][239] Large surveillance programs to identify any excess of GBS cases In contrast, during surveillance studies for the mRNA-based vaccines, no vaccine-outcome association, including the occurrence of GBS, met the pre-specified requirement for a signal. 244 This may suggest that, similar to thrombotic complications (see below), antigens mimicking neural components may be related to the structure of the adenovirus vectors, and this would explain the relative safety of mRNA vaccines. 245 Overall, these data are preliminary and should be taken cautiously, without leaping to costly conclusions. 246

| COVID-19 vaccines, CIDP, and other neuropathies
In our review, we found isolated reports of "acute onset" CIDP following Moderna 259 and AstraZeneca vaccines, 235,260 in all cases 3 weeks after the first dose. Of note, one of the cases had a similar presentation years prior after the influenza vaccine, but he was asymptomatic since then. 259 All cases had a good recovery after standard treatment. In the UK cohort on GBS within 4 weeks after COVID-19 vaccines, four out of 16 patients were diagnosed with acute-onset CIDP. 237 No clinical or neurophysiological exams are available for these patients, besides the fact that two were subsequently re-treated with IVIG, one patient with corticosteroids, and one with plasma exchanges with good outcome. 237 In the pre-COVID era, the Italian CIDP database identified vaccination as the anteceding event in 1.5% of 411 patients 1 to 42 days before the diagnosis of CIDP. 261 Given the chronic nature of CIDP and the fact that its diagnosis requires a progression over 8 weeks, attributing its onset to a single event (ie, vaccination) is challenging.
As an example, one of the post-COVID-19 vaccine CIDP patients had also received the influenza vaccine 6 weeks prior to the onset of symptoms, while the COVID-19 vaccine had been administered 3 weeks prior. 235 When reviewing the potential of COVID-19 vaccines to exacerbate or worsen CIDP in patients with an established diagnosis, we did not find any report so far. Our current knowledge on the use of vaccines of any kind in patients with a prior CIDP diag-  272 Interestingly, one patient complained of onset of pain around the injection site that spread to the shoulder and the arm. 270 Overall, the incidence of post-vaccination Parsonage-  strength, consistency, specificity, temporality, biologic gradient, plausibility, coherence, experiment, and analogy. 282 So far temporality and, possibly, plausibility seem to be the only criteria met by the conditions reviewed in this paper. Therefore, based on available data, any conclusion about a pathophysiological correlation between COVID-19, vaccines and PNS disorders remains premature, while epidemiological, clinical and pathological data are insufficient. 283 The occurrence of PNS complication after COVID-19 vaccines seems very rare and limited to a possible higher risk of facial nerve palsy and possibly GBS, however, in a range that should not raise any concern on the need to pursue the vaccination campaign. Based on experiences with other vaccination campaigns and data coming from adverse monitoring systems, there is widespread consensus that the benefits of vaccination outweigh the risks related to adverse events.

| COVID-19 vaccines and muscle involvement
Although large cohort studies are still lacking, there is early evidence that the administration of COVID-19 vaccines, specifically the Pfizer vaccine, among patients with known history of GBS is not associated with a significant risk of relapse. 284