Brain aging in bipolar disorder using a neuroimaging and machine learning-derived metric: Findings from the ENIGMA BD Working Group.

Bipolar disorder was associated with higher brain-predicted age difference (brain-PAD) compared to healthy comparison participants, primarily among older patients.

• The study included 1342 BD and 1577 HC adult participants aged 18-75 years old (mean = 37.2; SD = 12.3) from the ENIGMA-BD working group.
• A significant age by diagnosis interaction was found (+0.05 [SE: 0.02] years).
• This represents the largest study of brain age in BD to date with 2919 total participants.
• Brain-PAD was estimated using 77 MRI measures of regional subcortical and lateral ventricle volumes, cortical thickness, and surface area.
• Linear mixed models adjusted for sex and age as fixed effects and site as a random effect.

Individuals with BD on antiepileptic medications (AED) only or on both AED and second-generation antipsychotics (SGA) demonstrated greater brain-PAD compared to individuals not on any of the examined medications.

• BD individuals on AED only showed higher brain-PAD of +3.20 [SE: 0.78] years compared to those not on any examined medications.
• BD individuals on both AED and SGA showed higher brain-PAD of +3.74 [SE: 0.89] years compared to those not on any examined medications.
• AED use was 'generally related to more advanced brain age' according to the authors.

Individuals with BD taking lithium, whether alone or in combination with AED and SGA, showed no difference in brain-PAD compared to individuals not taking any of the examined medications.

• Lithium users showed lower brain-PAD compared to those on AED alone by -4.48 [SE: 0.84] years.
• Lithium users showed lower brain-PAD compared to those on AED and SGA combined by -5.01 [SE: 0.92] years.
• Authors describe this as consistent with 'a possible neuroprotective effect of lithium.'
• This finding was consistent with prior findings from ENIGMA-BD.

Individuals with a BD Type I subtype diagnosis had a higher brain-PAD compared to those with BD Type II or subtypes not otherwise specified (NOS).

• BD I subtype was associated with brain-PAD of +1.50 [SE: 0.55] years compared to BD II or NOS subtypes.
• This suggests differential brain aging effects across bipolar disorder subtypes.

Brain age was predicted using a machine learning model previously trained and validated using independent samples from the ENIGMA working group.

• The model used 77 MRI measures including regional subcortical and lateral ventricle volumes, cortical thickness, and surface area.
• Chronological age was subtracted from predicted age to produce an individual-level estimate known as brain-PAD.
• The model was leveraged from a previously established ENIGMA machine learning approach rather than trained de novo in this sample.
• Participants were drawn from the curated ENIGMA Bipolar Disorder working group dataset.