Dear Sirs,

Coronavirus disease 2019 (COVID-19) patients on intensive care units (ICU) frequently present with acute encephalopathy that appears to be distinct from other ICU-related encephalopathies regarding higher incidence, longer duration, and increased severity including autonomous dysregulation and non-satisfactory response to neuroleptic drugs [1]. According to the updated nomenclature of delirium and acute encephalopathy, we use the term acute encephalopathy to describe a rapidly developing pathobiological brain process, expressed clinically as delirium, subsyndromal delirium or coma with partly additional neurological findings such as extrapyramidal signs or seizures [2]. This type of acute encephalopathy is associated with delayed recovery, weaning failure, prolonged ICU or hospital stay, or even impaired clinical outcome.

First reports found positive responses of COVID-19-associated encephalopathy to immunotherapy [3,4,5] that supported the hypothesis of a possible inflammatory pathomechanism [6]. However, this encephalopathy might still be misjudged by intensivists as poorly treatable with limited prognosis, risking premature withdrawal of ICU therapy or even end-of-life decisions in affected patients. Therefore, our case series demonstrates a promising and rapid effect of intravenous immunoglobulins (IVIg) on otherwise treatment-refractory acute encephalopathy in COVID-19 patients on ICU.

This retrospective, single-center case series included 12 patients with critical courses of COVID-19 requiring treatment at ICU, who developed a severe encephalopathy (leading to clinical presentation of hyper- and/or hypoactive delirium [2]) of at least 1 week without satisfactory response to neuroleptic drugs and/or even sedatives. These patients were treated with 2 g/kg IVIg over 3–5 days as off-label, individual medical treatment. The dosage of IVIg was chosen pragmatically according to established therapeutic regimens in other neurological autoimmune-mediated diseases.

Other causes of encephalopathy such as increased blood levels of sedative medication, intoxication, metabolic changes (abnormal electrolyte concentrations, hyperuricemia, hepatic encephalopathy, hypo- or hyperglycemia), sepsis with fever, hypothermia, shock or hypoxia were excluded. All patients received treatment with 6 mg dexamethasone for at least 10 days according to the RECOVERY study protocol [7].

After occurrence of acute encephalopathy and prior to IVIg therapy, all patients received imaging with either cerebral CT or MRI and one-time cerebrospinal fluid (CSF) examinations (except patient #8, whose critical clinical status did not permit lumbar puncture) with measurement of standard laboratory parameters. Antineuronal autoantibodies (namely: IgG-antibodies against amphiphysin, PNMA2 (Ma2/Ta), Ri, Yo, Hu, CV2 (CRMP5), Tr (DNER), NMDA receptor, GABA-b receptor, AMPA receptor1/2 (GluA1/GluA2), mGluR5, Glycin receptor, Dopamin2 receptor, DPPX, LGI1, CASPR2, Aquaporin-4, Myelin, GAD65) were determined by cell-based indirect immunofluorescence assays at Labor Berlin or Euroimmun, Germany. A PCR-screening for common neurotropic pathogens (namely: SARS-CoV-2, herpes simplex virus 1/2, varicella zoster virus, human herpes virus 6, Epstein–Barr virus, cytomegalovirus, enterovirus, parechovirus, Streptococcus pneumoniae, Neisseria meningitidis, Haemophilus influenzae, Listeria monocytogenes, group B streptococcus, Escherichia coli (K1), and cryptococcus) excluded other common CNS infections. Additionally, indirect immunofluorescence technique on unfixed murine brain sections was applied with CSF and serum according to previously published protocols [8]. The inflammatory marker IL-6 was measured routinely every other day with Elecsys® IL-6 immunoassay (Roche) and Immulite IL-6 (Siemens). Neurofilament light chain was determined in serum with Simoa NF-light™ assay (Quanterix) and in CSF with NF-light™ ELISA (UmanDiagnostics).

Clinical neurological outcome was continuously evaluated by a neurologist and reported immediately before initiation and after termination of IVIg therapy and at discharge. Outcomes were assessed by Confusion Assessment Method for ICU (CAM-ICU), Richmond Agitation–Sedation Scale (RASS), Glasgow Coma Scale (GCS), Glasgow Outcome Scale (GOS) and modified Rankin Scale (mRS). Standardized scores were primarily ascertained by the treating neurologist. In case of missing data, scores were obtained from the daily routine assessment by the ICU nurses.

The patient cohort (n = 12) had a median age of 67 years (range 43–77, two females), required a median of 61 days (range 33–160) of ICU treatment and presented high rates of multi-organ failure with invasive ventilation in 11 (92%, n = 9 prone positioning, n = 2 veno-venous extracorporeal membrane oxygenation), pulmonary superinfections in 11 (92%), hepatic failure in five (42%), renal replacement therapy in nine (75%) and catecholamine therapy in all patients (Table 1). After a median of 23 days (range 0–37) after hospital admission due to COVID-19, all patients developed encephalopathy, requiring continuous intravenous sedation in 10 patients (83%, Table 2).

Table 1 Clinical presentation and duration of the disease
Full size table
Table 2 Detailed clinical and time course: changes of CNS medication
Full size table

Cerebral CT or MRI showed none or only unspecific findings, but no abnormalities explaining the observed encephalopathy (Table 3). CSF was taken in 11 patients. In accordance with a recently published study about CSF findings in COVID-19 [9], eight of our patients showed blood brain barrier (BBB) dysfunction and all yielded normal cell counts and negative PCR for common pathogens including SARS-CoV-2. Furthermore, nine patients displayed either low-titer antineuronal antibodies in serum (Myelin, CASPR2, NMDAR, Yo), or IgG binding of unknown specificity in indirect immunofluorescence of blood and CSF on unfixed murine brain sections, or both (Table 3). Increased neurofilament light-chain values in serum and CSF of all investigated patients (measured in n = 11 of 12) reflected relevant neuronal degeneration or damage.

Table 3 Diagnostic findings
Full size table

All patients received IVIg treatment at a median of 25 days (range 10–37) after acute encephalopathy onset and a median of 48 days (range 32–54) after hospital admission as ultima ratio therapy. Nine of 12 patients (#1–9, 75%, responders) presented milder symptoms at a median of 4 days (range 1–6) after IVIg initiation (Table 2). Three of 12 patients did not improve (#10–12, 25%, non-responders) and died of sepsis. 67% of responders had a negative CAM-ICU at discharge, while RASS improved from − 3 before therapy to 0 at discharge. Median GCS improved from 6 (range 3–12) to 14.5 (range 11–15), median GOS from 2 (range 2–3) to 3 (range 3–4) and median mRS from 5 (range 4–5) to 4 (range 3–5, Table 4). Before IVIg therapy, all patients required sedative (benzodiazepines, dexmedetomidine, clonidine, phenobarbital, propofol, esketamine, opiates) and/or neuroleptic or other CNS medication (risperidone, quetiapine, citalopram, amantadine) due to continuous encephalopathy, ongoing ventilation and ICU treatment. After IVIg administration, sedative or neuroleptic medication was successfully reduced in seven of 12 cases (Table 2). Patient #6 improved initially but died later due to sepsis. No adverse events attributable to IVIg therapy were observed.

Table 4 Results, primary clinical outcome
Full size table

Mean values of NfL in serum (responders 392 pg/ml, range 31–914 pg/ml), non-responders 2751 pg/ml, range 1348–4153 pg/ml) and CSF (responders 2406 pg/ml, range 981–6359 pg/ml, non-responders 14271 pg/ml, range 2215–28099 pg/ml) were higher in non-responders than in responders (Table 3). Similarly, mean values of IL-6 before IVIg treatment were lower in responders (98 ng/l, range 19–252 ng/l) compared to non-responders (438 ng/l, range 106–1087 ng/l) and decreased over the course of the treatment in responders (at discharge: 39 ng/l, range 7–99 ng/l) while increasing in non-responders (before death: 2983 ng/l, range 61–5905 ng/l, Table 3).

This case series demonstrates a therapeutic effect of IVIg in treatment-refractory COVID-19-associated acute encephalopathy. The pathophysiology of this entity remains speculative, but IVIg response points towards an immune-mediated mechanism, possibly succeeding COVID-19-induced cytokine release syndrome with BBB dysfunction and macrophage immigration or microglia activation, well in line with a neuroinflammatory hypothesis of COVID-19-associated acute encephalopathy [10]. The high frequency of intrathecal antineuronal antibodies detected by indirect immunofluorescence is in line with previous publications [8, 11] and might be a surrogate marker for autoimmune-mediated mechanisms, but whether they reflect specific or unspecific binding against CNS target epitopes causing acute encephalopathy currently remains open; especially since similar findings can be detected in asymptomatic persons. Immune-regulatory effects of IVIg are pleiotropic and involve Fc-receptor binding on macrophages and microglia, inflammation suppression including cytokines and chemokines, and transformation of activated microglia into a protective phenotype leading to reduction of neuronal cell death [12].

To date, one report described positive effects of IVIg on COVID-19-associated acute encephalopathy in five patients, but the therapeutic regimen overlapped with Tocilizumab and standardized methodology was lacking [13].

In our well-characterized cohort using standardized scores documenting responders and non-responders, natural remission coinciding the suspected IVIg effects cannot be excluded, but the long preceding interval without progress and the rapid improvement after IVIg initiation suggest a causal relationship. Due to the retrospective nature of our report and the individual treatment attempt as ultima ratio therapy, a control group without IVIg treatment is lacking. Unresponsiveness in three cases remains unexplained and might indicate variability in pathophysiological mechanisms of encephalopathy. Interestingly, IVIg non-responders presented very high NfL levels in serum and CSF that might also reflect a more intense neuronal damage leading to a more severe encephalopathy than in responders. Moreover, considering the fatal sepsis development in non-responders, the neurological improvement after IVIg therapy might have been masked by general disease severity with multi-organ failure. Of note, patient #1 started to improve within 1 day after beginning of IVIg treatment, which could be a coincidence. However, a more impressive and, thus, clearly definable improvement was seen 3–4 days after treatment of IVIg, which is why we tend to attribute the clinical change to our therapy.

In conclusion, IVIg are a promising immunotherapy for severe treatment-refractory COVID-19-associated acute encephalopathy. Further prospective controlled studies are required to validate safety and efficacy, monitor long-term outcome, and explore the mechanisms of the therapeutic IVIg effect.