CHECKPOINT-INHIBITOR TOXICITY

Nervous system immune-related adverse events

Neurological irAEs are mostly non-specific grade 1-2 events, and high-grade ones run below 1% — but the phenotypes that do reach this atlas are among the most lethal in it. Myasthenia gravis carries roughly 20% fatality in the pharmacovigilance record, arrives earliest of the neurological events, and travels with myocarditis and myositis.

Symptoms overlap heavily here: weakness can be myasthenia gravis, myositis, or Guillain-Barré, and the first two also implicate the heart. SITC evaluates the triad with one shared work-up and refers every neurological irAE to a specialist regardless of severity — this atlas carries that rule rather than a grade threshold. Educational reference, not medical advice.

Anchored on the SITC chapter for this organ. Every source is named beside the position it supports, and labelled with what kind of source it is.

At a glance

ICI-associated myasthenia gravis
RareOnset weeks 1–6Critical
0.47% of ICI reports vs. 0.04% of the full database (ROR 16.5, 95% CI 14.5-18.9) — VigiBase spontaneous reports — a share of REPORTS and a disproportionality signal, not an incidence in treated patients
ICI-associated encephalitis
RareOnset weeks 6–12Critical
0.51% of ICI reports vs. 0.05% of the full database (ROR 10.4, 95% CI 9.2-11.8) — VigiBase spontaneous reports — a share of REPORTS with a disproportionality signal, not an incidence in treated patients
ICI-associated Guillain-Barré syndrome
Very rareOnset weeks 1–6Critical

ICI-associated myasthenia gravis

Also called: ICI myasthenia gravis · myasthenic syndrome · MG · checkpoint inhibitor myasthenia

The earliest and deadliest of the neurological irAEs — median 29 days, roughly 20% fatality in the pharmacovigilance record — and the one that travels with myocarditis and myositis. Seronegative disease does not exclude it, and the threat to life is respiratory, so the pulmonary assessment is part of the diagnosis rather than a complication check.

rarecriticalonset: weeks 1–6

Reported frequency

0.47% of ICI reports vs. 0.04% of the full database (ROR 16.5, 95% CI 14.5-18.9) — Share of ICI adverse-event reports naming myasthenia gravis
of 48,653 ICI reports · VigiBase spontaneous reports — a share of REPORTS and a disproportionality signal, not an incidence in treated patients
pharmacovigilance · 31118078
3.8% anti-CTLA-4, 6.1% anti-PD-1, 12.0% combination — Overall neurological adverse events across ICI trials (ALL neurological phenotypes, not this one)
of 59 clinical trials totalling 9208 patients · Systematic review to February 2016. This is the ORGAN-level denominator, dominated by non-specific grade 1-2 events such as headache — it is carried here as context for how small a fraction myasthenia gravis is, never as this entity's rate
narrative review · 28064139

Severity and class

below 1% — High-grade neurological adverse events (ALL neurological phenotypes)
of 59 clinical trials totalling 9208 patients · Systematic review, all ICI classes. Organ-level, not specific to myasthenia gravis — no source cited here reports a grade ≥3 rate for this entity alone
narrative review · 28064139
~ 20% — Fatality among reported myasthenia gravis cases
of Myasthenia gravis reports within 48,653 ICI adverse-event reports · VigiBase spontaneous reports. A fatality share among REPORTS, which over-represents severe outcomes — but note the direction of the contrast the study draws: the other neurological phenotypes sat at 6-12% in the same database
pharmacovigilance · 31118078
Associated with anti-PD-1 — ICI class associated with myasthenia gravis reports
of 48,653 ICI adverse-event reports against a full database of 18,518,994 · VigiBase disproportionality analysis to September 2018. A statement of association direction, not a ratio between classes — the study reports which class the phenotype tracked with, and this field carries that and nothing more
pharmacovigilance · 31118078

Onset

Median 29 days — the earliest of the neurological phenotypes in the pharmacovigilance record, against 61-80 days for the others. The organ-level median across trials was 6 weeks, which is a different and slower population. Stored in the units each source printed; no median is converted to weeks.

Presentation

· Fatigable or fluctuating muscle weakness, proximal (neck and shoulder) more than distal
· Ptosis, double vision from extraocular movement abnormality, facial weakness, difficulty swallowing — bulbar and ocular muscles are commonly affected
· Diaphragmatic weakness with respiratory compromise, which is what makes this entity lethal
· Concurrent myocarditis or myositis — SITC calls the combination potentially dangerous, and the pharmacovigilance record found it frequent
· Seronegative presentation is common enough to matter: anti-AChR was positive in 66.7% (30/45) and anti-MuSK in 5.3% (1/19) of one 47-patient series, and SITC states toxicity can occur independent of positive serology

Differential

· Myositis — the reason SITC offers electrodiagnostic studies specifically to distinguish the two, and the reason finding one is not a reason to stop looking
· Myocarditis, which shares the triad and is carried as its own card in this atlas; troponin belongs in the work-up of weakness here, not only of chest symptoms
· Thyroid dysfunction, which SITC includes in the same evaluation
· Guillain-Barré syndrome and other polyneuropathies — ascending weakness with areflexia points there instead, and the treatment paths diverge
· Progression of the malignancy, including leptomeningeal or CNS disease

Work-up

Neurology consultation · All patients
SITC attaches it to ANY grade of myasthenic symptoms, and refers every neurological irAE to a specialist regardless of severity. Not a grade threshold.
34172516
Diagnostic antibody testing (anti-AChR, anti-MuSK) · All patients
Confirms the diagnosis when positive. A negative panel does NOT exclude it — SITC states toxicity can occur independent of positive serology, and one series found anti-MuSK positive in 1 of 19.
34172516
Pulmonary function with negative inspiratory force and vital capacity · All patients
The measurement that detects diaphragmatic weakness before it becomes respiratory failure. SITC asks for it in the work-up and then for frequent repeat assessments during treatment.
34172516
Evaluation for concurrent myositis, myocarditis and thyroid dysfunction · All patients
The shared triad work-up. The pharmacovigilance record found concurrent myocarditis and myositis frequent in this phenotype specifically, and myocarditis is the member with 50% reported fatality.
34172516, 31118078
Electrodiagnostic studies · If atypical
Offered by SITC to distinguish myasthenia gravis from myositis, which present with overlapping weakness.
34172516

Grade ladder

Grade 1

Hold the ICIConsider admission

Grade 1 — ASCO's cross-organ default continues an ICPi at grade 1 and names neurologic toxicity as one of its exceptions. SITC fills that exception without grading it: any grade of myasthenic symptoms gets a neurology consultation, and a diagnosis of myasthenia gravis discontinues the ICI. The dose on this rung is SITC's OCULAR myasthenia figure and applies only to ocular disease of grade ≤2.

Corticosteroid 0.5–1 mg/kg/day (oral)

Taper within 6 weeks or longer, starting once: Not stated for this entity. SITC gives the ocular grade ≤2 dose without a taper window; the ceiling shown is ASCO's cross-organ 'at least 4-6 weeks', which is a FLOOR in its own text and is rendered here as an open-ended plan rather than a completion date.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy and pyridostigmine — A diagnosis of myasthenia gravis — not steroid failure. SITC names this the standard of care on diagnosis, alongside discontinuing the ICI.

· Neurology consultation at any grade of myasthenic symptoms
· Assess negative inspiratory force and vital capacity before treating this as mild
· Look for myocarditis and myositis — the triad is why an isolated-looking weakness is not isolated

Escalate when: Any bulbar symptom, any drop in negative inspiratory force or vital capacity, or any evidence of myocarditis or myositis. Weakness that is fatigable and progressing does not wait for a grade.

Grade 2

Hold — consider permanent discontinuationAdmit

Grade 2 — ASCO's general ladder may suspend the ICPi for most grade 2 toxicities. SITC's entity rule is stronger and ungraded: a diagnosis of myasthenia gravis discontinues ICI therapy and starts IVIG or PLEX with corticosteroids and pyridostigmine. The 0.5-1 mg/kg figure is carried here ONLY where the disease is ocular and grade ≤2; a generalised grade 2 myasthenia gravis is outside the scope of that sentence and no other dose is stated.

Corticosteroid 0.5–1 mg/kg/day (oral)

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length, rendered open-ended.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy and pyridostigmine — Diagnosis of myasthenia gravis, at any grade.

· Discontinue the ICI on diagnosis
· Frequent pulmonary assessments — SITC's word is frequent, and this is the organ that kills
· Evaluate concurrent myositis, myocarditis and thyroid dysfunction

Escalate when: Falling vital capacity or negative inspiratory force, dysphagia, or any cardiac finding. Respiratory compromise is the failure mode, not limb weakness.

Grade 3

Discontinue permanentlyAdmit

Grade 3 — ASCO suspends the ICPi and starts high-dose corticosteroids at grade 3 across organs. SITC's entity rule already discontinues the drug and starts IVIG or PLEX. The ocular grade ≤2 dose does NOT reach this rung, and SITC's high-dose pulse corticosteroid sentence for myasthenia gravis states no dose, so none is encoded.

Dose not carried

SITC's standard of care for autoimmune myasthenia gravis includes high-dose pulse corticosteroids, and that sentence names no dose, route or duration. The only myasthenia figure in the chapter is scoped to OCULAR disease at grade ≤2, which this rung is not. Importing the encephalitis or Guillain-Barré pulse regimen from two paragraphs away would be a cross-entity borrow of exactly the kind D-STEROID forbids, so the rung states that the dose is not carried rather than supplying one.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy and pyridostigmine — Diagnosis of myasthenia gravis. SITC names IVIG and PLEX as the standard of care, not as a rescue.

· Permanent discontinuation of the ICI
· Frequent pulmonary assessments, with a threshold for ventilatory support set by neurology and critical care
· Rule in or out the myocarditis that travels with this phenotype before attributing everything to weakness

Escalate when: Respiratory failure, myasthenic crisis, or concurrent myocarditis.

Grade 4

Discontinue permanentlyICU

Grade 4 — ASCO's cross-organ rule is permanent discontinuation at grade 4, with an endocrine exception that does not apply. SITC discontinues on diagnosis regardless of grade, so the two agree here.

Dose not carried

No dose for grade 4 myasthenia gravis is stated in any chapter held here. The ocular grade ≤2 figure does not extend, and no pulse dose is stated for this entity.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy and pyridostigmine — Diagnosis of myasthenia gravis.

· Critical care with airway and ventilatory management
· Permanent discontinuation of the ICI
· Continued evaluation for myocarditis and myositis — the triad does not resolve because the myasthenia is being treated

Escalate when: Already the top rung. Management of established myasthenic crisis is beyond what the chapters held here state.

Rechallenge

contraindicated

SITC discontinues ICI therapy on a diagnosis of myasthenia gravis, without a grade condition and without describing a route back. ASCO's grade 4 rule is permanent discontinuation. Against roughly 20% reported fatality, the earliest onset of any neurological phenotype, and a frequent association with myocarditis, no source cited here describes conditions under which the drug would be resumed. The stance is the conservative reading of an absence, and the empty recurrence field says the question has not been answered here.

· None are stated by any source cited here. A prerequisite list would imply a gated route that no chapter held here describes

Deliberately not carried

divergences — Empty for the same structural reason as the cardiac card: only ONE society chapter for this organ is held here. ESMO and ASCO neurology chapters were never read — ASCO appears via its cross-organ ladder quoted from the abstract, which is why 'neuro' is deliberately absent from ASCO's IRAE_GUIDELINE_ORGANS row. Where the two touch the same decision they agree, and ASCO says so itself by naming neurologic toxicity as an exception to its grade-1 default. Read the empty list as one voice, not as consensus.
incidence — Neither figure carried is an incidence in treated patients. The 0.47% is a share of spontaneous REPORTS with a disproportionality signal beside it; the 3.8/6.1/12.0% figures are the ORGAN-level trial rate across all neurological phenotypes, dominated by headache. Both framings say so in their population field. No source cited here reports how often ICI myasthenia gravis occurs per patient treated.
gradeThreePlus — The sub-1% high-grade figure carried is organ-level across all neurological phenotypes, and its population field says so. No source cited here reports a grade ≥3 rate for myasthenia gravis specifically. It is carried rather than left null because the organ-level ceiling is genuinely informative — high-grade neurological events are rare in aggregate — but it must not be read as this entity's severity rate.
onset.medianWeeks — The phenotype-specific median is 29 days and the organ-level median is 6 weeks; they come from different sources and different populations. Converting 29 days to 4.14 weeks would print a number no source stated, and averaging the two would be worse. Both are in onset.note in their published units.
ladder[].steroid dose, grades 3-4 — SITC states high-dose pulse corticosteroids for autoimmune myasthenia gravis and names no dose. The chapter's only myasthenia dose is scoped to OCULAR disease at grade ≤2 and is carried only on the rungs it covers. The 1000 mg IV methylprednisolone regimen in the same chapter is written for encephalitis and Guillain-Barré, and is not borrowed here — for a critical-tier entity this is the most consequential gap on the card and it is stated rather than filled.
ladder[].taper — No taper window, start trigger or step cadence is stated for myasthenia gravis anywhere in the chapter held here. The grade 1-2 rungs show ASCO's cross-organ 'at least 4-6 weeks' as an open-ended ceiling, sourced to ASCO and labelled as such in startTrigger, and all three cadence fields are null together.
serology test performance — The 66.7% and 5.3% positivity rates are stored as SITC states them, with their denominators (30/45 and 1/19), because the clinically load-bearing point is that a negative panel does not exclude the diagnosis. They are positivity rates in a 47-patient series, not sensitivity, and no negative predictive value is asserted.
guidelinePmids — Left empty although ASCO and SITC are rows in guidelines.ts: their stored keyRecommendation strings are renal-scoped, and linking them would surface kidney dosing on a neurology card. The chapter is carried in citations with its own quote.

Sources

Brahmer JR, et al. (2021) Society for Immunotherapy of Cancer (SITC) clinical practice guideline on immune checkpoint inhibitor-related adverse eventsCited for: The presenting pattern of ICI myasthenia gravis and its bulbar, ocular and diaphragmatic involvement; that toxicity occurs independent of positive serology; the anti-PD-(L)1 skew; the myocarditis and myositis association; the standard of care (IVIG or PLEX with corticosteroids and pyridostigmine); the ICI discontinuation rule; the ocular grade ≤2 prednisone dose; and the work-up, including the pulmonary assessments and the electrodiagnostic study that separates myasthenia gravis from myositis.Supporting text: the full text
Schneider BJ, et al. (2021) Management of Immune-Related Adverse Events in Patients Treated With Immune Checkpoint Inhibitor Therapy: ASCO Guideline UpdateCited for: That ASCO's continue-at-grade-1 default explicitly EXCLUDES some neurologic toxicities, and that permanent discontinuation is its grade 4 rule. Cited as the cross-organ ladder it is — no ASCO neurology chapter is held in this repository.Supporting text: the PubMed abstract (checkable at the link above)
Johnson DB, et al. (2019) Neurologic toxicity associated with immune checkpoint inhibitors: a pharmacovigilance study.Cited for: Per-phenotype reporting shares and reporting odds ratios for myasthenia gravis, encephalitis, peripheral neuropathy and meningitis; the class split (myasthenia gravis and encephalitis with anti-PD-1, the others with anti-CTLA-4); and the fatality-and-onset contrast that separates myasthenia gravis from the rest — roughly 20% fatality at a median of 29 days, against 6-12% at 61-80 days, with frequent concurrent myocarditis and myositis.Supporting text: the PubMed abstract (checkable at the link above)
Cuzzubbo S, et al. (2017) Neurological adverse events associated with immune checkpoint inhibitors: Review of the literature.Cited for: Overall neurological adverse-event incidence by ICI class across 59 trials, the observation that most events are grade 1-2 and non-specific, the sub-1% high-grade rate, the median 6-week onset, and that drug interruption plus steroids led to neurological recovery in most cases — including in Guillain-Barré syndrome, where steroids are not the usual recommendation.Supporting text: the PubMed abstract (checkable at the link above)

ICI-associated encephalitis

Also called: immune-related encephalitis · autoimmune encephalitis · encephalopathy · ICI encephalitis

Under 1% by SITC's estimate, and treated at full intensity from grade 1 — pulse-dose methylprednisolone plus IVIG or PLEX at any grade. Antibody panels are usually negative, so a normal autoimmune screen argues against nothing, and antivirals run empirically until viral encephalitis is excluded.

rarecriticalonset: weeks 6–12

Reported frequency

0.51% of ICI reports vs. 0.05% of the full database (ROR 10.4, 95% CI 9.2-11.8) — Share of ICI adverse-event reports naming encephalitis
of 48,653 ICI reports · VigiBase spontaneous reports — a share of REPORTS with a disproportionality signal, not an incidence in treated patients
pharmacovigilance · 31118078

Severity and class

below 1% — High-grade neurological adverse events (ALL neurological phenotypes)
of 59 clinical trials totalling 9208 patients · Systematic review, all ICI classes. Organ-level, not specific to encephalitis — no source cited here reports a grade ≥3 rate for this entity alone
narrative review · 28064139
6-12% — Fatality among reported neurological cases OTHER than myasthenia gravis
of Encephalitis, peripheral neuropathy and meningitis reports within 48,653 ICI adverse-event reports · VigiBase spontaneous reports. The study gives this as a RANGE across the non-myasthenic phenotypes rather than a figure for encephalitis alone, and the framing is stored at that resolution rather than narrowed to one point
pharmacovigilance · 31118078
Associated with anti-PD-1 — ICI class associated with encephalitis reports
of 48,653 ICI adverse-event reports against a full database of 18,518,994 · VigiBase disproportionality analysis to September 2018. SITC states the same direction in prose — higher risk with anti-PD-(L)1 monotherapy or combination than with anti-CTLA-4 — and quantifies it no further
pharmacovigilance · 31118078

Onset

The pharmacovigilance record groups encephalitis with the non-myasthenic neurological phenotypes at a median of 61-80 days — later than myasthenia gravis at 29 days. That is a range across several phenotypes, not a median for encephalitis alone. The organ-level median across trials was 6 weeks. No encephalitis-specific median is stated in any source cited here.

Presentation

· Altered behaviour, confusion, agitation
· Short-term memory impairment
· Speech abnormality
· Seizures
· Positive autoimmune encephalitis or paraneoplastic antibodies are RARE — SITC's word — so a negative panel does not argue against the diagnosis

Differential

· Viral encephalitis, which is why SITC runs empirical antiviral treatment until it is ruled out rather than after
· Aseptic meningitis — SITC distinguishes them by mental status, which is typically normal in meningitis and not in encephalitis
· Hypophysitis, which SITC asks to be excluded in a patient presenting with headache
· CNS metastases and leptomeningeal disease
· Metabolic encephalopathy, sepsis, and drug effects — the ordinary causes of confusion in a patient with cancer

Work-up

MRI of the brain · All patients
First item in SITC's encephalitis work-up, and what separates structural disease from an immune cause.
34172516
Lumbar puncture with PCR for infectious encephalitis · All patients
Infection is the diagnosis that must not be missed, and the reason antivirals start before results return rather than after.
34172516
Autoimmune encephalopathy and paraneoplastic panels of blood and CSF · All patients
Ordered by SITC, with the caveat SITC itself states: positive results are rare, so a negative panel does not exclude ICI encephalitis and must not delay treatment.
34172516
Electroencephalogram · All patients
Part of SITC's work-up; seizures are within the presenting range and may be non-convulsive.
34172516
CBC, CMP and thyroid panel · All patients
SITC's baseline set, covering the metabolic and endocrine causes of an altered mental state.
34172516
Specialist referral · All patients
SITC refers every neurological irAE to a specialist regardless of severity.
34172516

Grade ladder

Grade 1

Hold the ICIAdmit

Grade 1 — ASCO's continue-at-grade-1 default names neurologic toxicity as an exception, and SITC supplies an explicitly ungraded regimen: patients with ANY grade of encephalitis receive pulse-dose methylprednisolone and additionally IVIG or PLEX. There is no low-intensity rung for this entity in the chapter held here.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — SITC states this for ANY grade of encephalitis, with IVIG or PLEX given in addition rather than on failure.

Taper within 6 weeks or longer, starting once: Not stated for this entity. SITC gives the pulse regimen and no taper; the ceiling shown is ASCO's cross-organ 'at least 4-6 weeks', which is a FLOOR in its own text and is rendered open-ended rather than as a completion date.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX) — Given WITH the pulse steroid at any grade — SITC's word is additionally, not instead.
IVIG 2 g/kg in divided doses over 5 days, PLEX one session every other day for 5–7 cycles, or rituximab 375 mg/m² weekly for 4 weeks — ICI-related encephalitis that does not respond to pulse-dose corticosteroids.

· Empirical antiviral treatment until viral encephalitis can be ruled out
· Exclude hypophysitis in a patient whose presentation includes headache
· Specialist referral regardless of severity

Escalate when: No response to pulse-dose corticosteroids, seizures, or a falling level of consciousness.

Grade 2

Hold — consider permanent discontinuationAdmit

Grade 2 — ASCO may suspend the ICPi for most grade 2 toxicities. SITC's encephalitis regimen does not change by grade: the same pulse-dose methylprednisolone with IVIG or PLEX.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — Any grade, with IVIG or PLEX in addition.

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX) — Given with the pulse steroid at any grade.
IVIG 2 g/kg in divided doses over 5 days, PLEX one session every other day for 5–7 cycles, or rituximab 375 mg/m² weekly for 4 weeks — No response to pulse-dose corticosteroids.

· Continue empirical antivirals until viral encephalitis is excluded
· EEG if the mental state fluctuates — non-convulsive seizures are within range
· Do not let a negative autoimmune panel delay treatment

Escalate when: Refractory disease, status epilepticus, or airway compromise from a falling conscious level.

Grade 3

Discontinue permanentlyICU

Grade 3 — ASCO suspends the ICPi and starts high-dose corticosteroids at grade 3; SITC has already done both at grade 1 for this entity. What changes at this rung is the care setting and the threshold for moving to the named refractory regimens.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — Unchanged by grade in the chapter held here.

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG 2 g/kg in divided doses over 5 days, PLEX one session every other day for 5–7 cycles, or rituximab 375 mg/m² weekly for 4 weeks — No response to pulse-dose corticosteroids. These are the doses SITC states.

· Critical care with airway protection and seizure management
· Antivirals continue until infection is definitively excluded
· Neurology leads; the diagnosis is refined alongside treatment, not before it

Escalate when: Status epilepticus, cerebral oedema, or continued deterioration on second-line therapy.

Grade 4

Discontinue permanentlyICU

Grade 4 — ASCO's cross-organ rule is permanent discontinuation at grade 4, with an endocrine exception that does not apply here.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — Unchanged by grade in the chapter held here.

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG 2 g/kg in divided doses over 5 days, PLEX one session every other day for 5–7 cycles, or rituximab 375 mg/m² weekly for 4 weeks — No response to pulse-dose corticosteroids.

· Critical care; permanent discontinuation of the ICI
· Infection excluded definitively before immunosuppression is escalated further

Escalate when: Already the top rung. Management of refractory ICI encephalitis beyond the named regimens is outside what the chapters held here state.

Rechallenge

contraindicated

No source cited here describes resuming an ICI after encephalitis. SITC states no rechallenge position for this entity, and ASCO's grade 4 rule is permanent discontinuation. Given a phenotype treated at pulse-dose intensity from grade 1, the absence of a described route back is read conservatively rather than as an open question.

· None are stated by any source cited here

Deliberately not carried

divergences — Empty because only ONE society chapter for this organ is held here — the same structural limit as the other wave-2 cards, not a finding that the sources agree. ASCO appears via its cross-organ ladder quoted from the abstract and establishes no neurology chapter coverage.
incidence — SITC estimates encephalitis at <1% and states no denominator, population or study behind that estimate, so it is not entered as a framing — a bare '<1%' with no denominator is exactly the shape this schema exists to prevent. The figure carried instead is the pharmacovigilance reporting share, whose population field says what it is. The SITC estimate appears in the entry's oneLiner attributed to SITC as an estimate.
mortality — The 6-12% carried is the pharmacovigilance study's range across ALL non-myasthenic neurological phenotypes — encephalitis, peripheral neuropathy and meningitis together. No encephalitis-specific fatality figure is stated in any source cited here, so the framing is stored at the resolution the source published rather than narrowed to a single number that would look phenotype-specific.
onset — Same resolution problem: 61-80 days is a range across the non-myasthenic phenotypes, not an encephalitis median. medianWeeks is null and the note says which population the range describes.
ladder[].taper — SITC states the pulse regimen and the refractory regimens with their doses, and no taper window, start trigger or step cadence for encephalitis. Every rung shows ASCO's cross-organ 'at least 4-6 weeks' as an open-ended ceiling, labelled as ASCO's in startTrigger, with all three cadence fields null together.
ladder[].hospitalization — No chapter held here states a care setting for encephalitis at any grade. The values are the atlas's reading for an entity SITC treats with pulse-dose steroids and IVIG or PLEX from grade 1 and works up with lumbar puncture and EEG, and each gradeDefinition says the ladder's intensity is SITC's while the setting is not.
guidelinePmids — Left empty although ASCO and SITC are rows in guidelines.ts: their stored keyRecommendation strings are renal-scoped and would surface kidney dosing on a neurology card.

Sources

Brahmer JR, et al. (2021) Society for Immunotherapy of Cancer (SITC) clinical practice guideline on immune checkpoint inhibitor-related adverse eventsCited for: The estimated encephalitis incidence, its symptom range, the anti-PD-(L)1 and combination skew, the rarity of positive autoimmune or paraneoplastic antibodies, the full diagnostic work-up, the any-grade pulse-dose steroid with IVIG or PLEX, the named refractory regimens with their doses, and the empirical antiviral cover until viral encephalitis is excluded.Supporting text: the full text
Schneider BJ, et al. (2021) Management of Immune-Related Adverse Events in Patients Treated With Immune Checkpoint Inhibitor Therapy: ASCO Guideline UpdateCited for: That ASCO's continue-at-grade-1 default explicitly EXCLUDES some neurologic toxicities, and that permanent discontinuation is its grade 4 rule. Cited as the cross-organ ladder it is — no ASCO neurology chapter is held in this repository.Supporting text: the PubMed abstract (checkable at the link above)
Johnson DB, et al. (2019) Neurologic toxicity associated with immune checkpoint inhibitors: a pharmacovigilance study.Cited for: Per-phenotype reporting shares and reporting odds ratios for myasthenia gravis, encephalitis, peripheral neuropathy and meningitis; the class split (myasthenia gravis and encephalitis with anti-PD-1, the others with anti-CTLA-4); and the fatality-and-onset contrast that separates myasthenia gravis from the rest — roughly 20% fatality at a median of 29 days, against 6-12% at 61-80 days, with frequent concurrent myocarditis and myositis.Supporting text: the PubMed abstract (checkable at the link above)
Cuzzubbo S, et al. (2017) Neurological adverse events associated with immune checkpoint inhibitors: Review of the literature.Cited for: Overall neurological adverse-event incidence by ICI class across 59 trials, the observation that most events are grade 1-2 and non-specific, the sub-1% high-grade rate, the median 6-week onset, and that drug interruption plus steroids led to neurological recovery in most cases — including in Guillain-Barré syndrome, where steroids are not the usual recommendation.Supporting text: the PubMed abstract (checkable at the link above)

ICI-associated Guillain-Barré syndrome

Also called: GBS · ascending polyradiculoneuropathy · acute inflammatory demyelinating polyradiculoneuropathy · AIDP

The one entity in this atlas whose ICI recommendation carries the word always: SITC permanently discontinues the drug on diagnosis, at any grade. It arrives early — usually within the first 3 cycles — and unlike idiopathic Guillain-Barré, corticosteroids are given here alongside IVIG or PLEX.

very-rarecriticalonset: weeks 1–6

Severity and class

below 1% — High-grade neurological adverse events (ALL neurological phenotypes)
of 59 clinical trials totalling 9208 patients · Systematic review, all ICI classes. Organ-level — no source cited here reports a rate for Guillain-Barré syndrome specifically at any grade
narrative review · 28064139
More common with anti-CTLA-4 or anti-CTLA-4 plus anti-PD-1 than with anti-PD-(L)1 alone — ICI class associated with Guillain-Barré syndrome
of Not stated — SITC gives the direction of the comparison and no numerator or denominator · SITC's chapter statement. Carried because the direction is opposite to myasthenia gravis and encephalitis, which is decision-relevant, but it is a qualitative comparison and the value says so
guideline · 34172516

Onset

SITC states that Guillain-Barré syndrome can develop soon after ICI treatment is started, usually within the first 3 cycles. Cycles are not weeks — the interval depends on the agent and schedule — so no week bounds are entered and the bucket reflects 'early' without converting a cycle count into a duration. The organ-level median onset across trials was 6 weeks.

Presentation

· Early lower back or thigh pain, preceding the weakness
· Ascending weakness, sensory loss and areflexia — the main symptoms
· Facial weakness and impaired extraocular movement from cranial neuropathies
· Autonomic dysregulation
· Nerve root enhancement and thickening on imaging

Differential

· A compressive spinal lesion, which is why SITC's work-up opens with MRI of the spine rather than with nerve conduction studies
· Myasthenia gravis — fatigable weakness with bulbar and ocular signs rather than ascending weakness with areflexia, and a different treatment path
· Myositis, which shares limb weakness and is separated electrodiagnostically
· Non-Guillain-Barré polyneuropathies, which SITC manages differently and at a lower steroid dose — the distinction changes the regimen, so it is made before treating
· Malignant infiltration of nerve roots or the leptomeninges

Work-up

MRI of the spine · All patients
First in SITC's work-up, explicitly to rule out a compressive lesion. Ordered before the diagnosis is treated as immune.
34172516
Lumbar puncture · All patients
Part of SITC's diagnostic set for suspected Guillain-Barré syndrome.
34172516
EMG and nerve conduction studies · All patients
Confirms the polyradiculoneuropathy and separates it from myositis and myasthenia gravis.
34172516
Ganglioside antibody panel of blood and CSF · All patients
SITC's serological component of the work-up.
34172516
Frequent pulmonary assessments · All patients
SITC asks for these specifically in Guillain-Barré syndrome. Ascending weakness reaches the diaphragm, and respiratory failure is the way this entity kills.
34172516
Specialist referral · All patients
SITC refers every neurological irAE to a specialist regardless of severity.
34172516

Grade ladder

Grade 1

Discontinue permanentlyAdmit

Grade 1 — ASCO's continue-at-grade-1 default names neurologic toxicity as an exception, and SITC's rule here is the strongest in the atlas: patients diagnosed with Guillain-Barré syndrome should ALWAYS permanently discontinue ICI therapy. The steroid regimen is likewise ungraded — any grade receives pulse-dose methylprednisolone with IVIG or PLEX.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — SITC states this for ANY grade of Guillain-Barré syndrome, with IVIG or PLEX in addition. This departs from idiopathic Guillain-Barré, where steroids are not usually recommended — the review cited here notes recovery on drug interruption plus steroids even in that condition.

Taper within 6 weeks or longer, starting once: Not stated for this entity. SITC gives the pulse regimen and no taper; the ceiling shown is ASCO's cross-organ 'at least 4-6 weeks', a FLOOR in its own text, rendered open-ended.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy — Diagnosis of Guillain-Barré syndrome — given with the steroid, not after it fails.

· Permanent discontinuation of the ICI — SITC's word is always, with no grade condition
· Frequent pulmonary assessments from the outset
· Exclude a compressive spinal lesion before treating the weakness as immune

Escalate when: Ascending weakness reaching respiratory muscles, bulbar involvement, or autonomic instability.

Grade 2

Discontinue permanentlyAdmit

Grade 2 — nothing in SITC's Guillain-Barré instruction is grade-conditioned. Same permanent discontinuation, same pulse-dose regimen, same IVIG or PLEX.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — Any grade, with IVIG or PLEX in addition.

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy — Diagnosis of Guillain-Barré syndrome, at any grade.

· Frequent pulmonary assessments — vital capacity trends decide the care setting
· Neurology leads management
· Autonomic monitoring; dysautonomia is part of the syndrome, not a complication of it

Escalate when: Falling vital capacity, bulbar weakness, or autonomic instability.

Grade 3

Discontinue permanentlyICU

Grade 3 — ASCO suspends the ICPi and starts high-dose corticosteroids at grade 3 across organs; SITC has already permanently discontinued the drug at grade 1 for this entity. The regimen does not change, the care setting does.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — Unchanged by grade in the chapter held here.

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy — Diagnosis of Guillain-Barré syndrome.

· Critical care with ventilatory support available
· Continued autonomic monitoring
· Permanent discontinuation of the ICI

Escalate when: Respiratory failure or haemodynamic instability from dysautonomia.

Grade 4

Discontinue permanentlyICU

Grade 4 — ASCO's cross-organ rule is permanent discontinuation at grade 4, which SITC's unconditional rule has already applied at every rung below.

Corticosteroid 1–2 mg/kg/day (intravenous)

Pulse: Methylprednisolone 1000 mg IV daily for 3–5 days — Unchanged by grade in the chapter held here.

Taper within 6 weeks or longer, starting once: Not stated for this entity; the ceiling is ASCO's cross-organ minimum taper length.

No step size or interval is carried: the source guideline states a total window and a response marker, and does not specify a cadence.

Second line

IVIG or plasma exchange (PLEX), with corticosteroid therapy — Diagnosis of Guillain-Barré syndrome.

· Ventilatory support and critical care
· Permanent discontinuation of the ICI

Escalate when: Already the top rung. Management of established respiratory failure in Guillain-Barré syndrome is beyond what the chapters held here state.

Rechallenge

contraindicated

This is the clearest rechallenge position in the atlas and it is stated rather than inferred: SITC says patients diagnosed with Guillain-Barré syndrome should ALWAYS permanently discontinue ICI therapy, with no grade condition and no exception. ASCO's grade 4 rule points the same way. No recurrence figure exists here because the question does not arise.

· None — the source's instruction is unconditional, so a prerequisite list would misrepresent it as a gated pathway

Deliberately not carried

divergences — Empty because only ONE society chapter for this organ is held here. ASCO appears via its cross-organ ladder quoted from the abstract and establishes no neurology chapter coverage.
incidence — EMPTY. No source cited here reports a rate for ICI-associated Guillain-Barré syndrome. The pharmacovigilance study breaks out myasthenia gravis, encephalitis, peripheral neuropathy and meningitis, and does not report Guillain-Barré separately; the systematic review counts Guillain-Barré-like syndromes only among its 27 case reports, which is not a denominator. The list is empty rather than borrowing the peripheral-neuropathy share, which is a different phenotype with a different class association.
mortality — Null. The 6-12% figure in the pharmacovigilance record covers encephalitis, peripheral neuropathy and meningitis, and Guillain-Barré is not among the phenotypes it breaks out. Nothing is carried rather than attributing another phenotype's fatality to this one.
onset — SITC states 'usually within the first 3 cycles', which is not a duration — cycle length varies by agent and schedule, so converting it into weeks would manufacture a window. medianWeeks and both range bounds are null and the note carries the cycle statement verbatim in substance.
agentClassSkew — The class comparison is carried, but it is qualitative — SITC states the direction and gives no numerator, denominator or ratio, and its basis is 'guideline' rather than a measured study. It is retained because the direction is opposite to myasthenia gravis and encephalitis and that difference is decision-relevant, and the value and denominator fields say plainly that no number was published.
ladder[].taper — SITC gives the pulse regimen and no taper for this entity. Every rung shows ASCO's cross-organ 'at least 4-6 weeks' as an open-ended ceiling labelled as ASCO's, with all three cadence fields null together.
ladder[].hospitalization — No care setting is stated for Guillain-Barré syndrome in the chapter held here. The values are the atlas's reading for an entity SITC treats with pulse steroids and IVIG or PLEX from grade 1 and monitors with frequent pulmonary assessments; each gradeDefinition attributes the intensity to SITC and not the setting.
guidelinePmids — Left empty although ASCO and SITC are rows in guidelines.ts: their stored keyRecommendation strings are renal-scoped and would surface kidney dosing on a neurology card.

Sources

Brahmer JR, et al. (2021) Society for Immunotherapy of Cancer (SITC) clinical practice guideline on immune checkpoint inhibitor-related adverse eventsCited for: The anti-CTLA-4 and combination skew for Guillain-Barré syndrome, its timing within the first 3 cycles, the presenting sequence from lower back or thigh pain to ascending weakness and areflexia, the cranial and autonomic involvement, the imaging finding, the work-up including the compressive-lesion exclusion, the any-grade pulse-dose steroid with IVIG or PLEX, the unconditional permanent discontinuation, and the frequent pulmonary assessments.Supporting text: the full text
Schneider BJ, et al. (2021) Management of Immune-Related Adverse Events in Patients Treated With Immune Checkpoint Inhibitor Therapy: ASCO Guideline UpdateCited for: That ASCO's continue-at-grade-1 default explicitly EXCLUDES some neurologic toxicities, and that permanent discontinuation is its grade 4 rule. Cited as the cross-organ ladder it is — no ASCO neurology chapter is held in this repository.Supporting text: the PubMed abstract (checkable at the link above)
Cuzzubbo S, et al. (2017) Neurological adverse events associated with immune checkpoint inhibitors: Review of the literature.Cited for: Overall neurological adverse-event incidence by ICI class across 59 trials, the observation that most events are grade 1-2 and non-specific, the sub-1% high-grade rate, the median 6-week onset, and that drug interruption plus steroids led to neurological recovery in most cases — including in Guillain-Barré syndrome, where steroids are not the usual recommendation.Supporting text: the PubMed abstract (checkable at the link above)

Educational use only. Educational synthesis of the published literature — not a treatment protocol, dosing guide, or medical advice. Regimens and agents shown are illustrative of what the literature describes; verify against current guidelines (ASON / KDIGO / ASCO / NCCN) and individualize to the patient. Using this site creates no clinician–patient relationship.

Spotted an error? Tell us. Sourcing and method are described on Methods.