To describe the phenotypic spectrum, serum-predominant seropositivity and treatment response in three patients with anti-IgLON5 disease.

IgLON5 is a neuronal cell-adhesion protein; anti-IgLON5 disease bridges autoimmunity and neurodegeneration. The hallmark is a distinctive sleep disorder, and IgLON5-IgG is serum-predominant. IgLON5 antibody can cause encephalitis that can be treated if diagnosed early.

Retrospective review of three consecutive patients; IgLON5-IgG by cell-based assay (serum and/or CSF). Patient consent was taken and ethical approval sought.

Three men (88, 71 and 53 years) presented with a prominent sleep disorder, with behavioural change, cognitive decline, agitation, hallucinations and gait/bulbar features.

Case 1: sleep disturbance, cognitive decline, agitation, minimal hallucinations; IVIG and rituximab; back to baseline cognition by week 4.

Case 2: sleep disturbance, agitation, visual hallucinations; attributed to alcohol cirrhosis (not withdrawal); steroids, IVIG and rituximab; back to baseline by week 6.

Case 3: sleep disturbance, unsteadiness/falls, nocturnal hallucinations; managed as alcohol withdrawal, progressing to respiratory failure requiring intubation; PET-CT showed stable pulmonary nodules with treated testicular malignancy and previous TB; IVIG then plasma exchange and IV methylprednisolone; very good response, rehabilitation then home at baseline cognition.

MRI was non-specific in all three and CSF protein, glucose and cells were within normal limits. Serum IgLON5-IgG was positive in 3/3, CSF in only 1/3. Alcohol comorbidity delayed diagnosis in 2/3. No active malignancy was found. All three improved, recovering from being dependent in all care to discharge home, functioning without any support, and one returned to work; best responses followed earlier or escalated treatment.

In this small series, a prominent sleep disorder was the consistent clue. Serum testing appeared more sensitive than CSF; a CSF-only strategy may under-diagnose, similar to literature. MRI was non-specific and did not aid diagnosis. Earlier immunotherapy appeared to accompany better outcomes, but n = 3 precludes causal inference; findings are hypothesis-generating.