Abstract
BACKGROUND AND OBJECTIVES: Working memory training (WMT) has been shown to improve WM in healthy older adults, patients with mild cognitive impairment, and individuals with HIV, including improvements in WM network efficiency. This randomized study explored near-transfer and far-transfer effects of WMT and examined whether LMX1A or APOE ɛ4 genotype variations influenced these effects in individuals with HIV and HIV-seronegative (SN) controls. METHODS: Participants were recruited from local communities. After screening and baseline evaluation, participants were randomized to either 25 sessions of adaptive or nonadaptive WMT (as an active control) over 5-8 weeks. Nontrained near-transfer WM tests, far-transfer tests, and self-reported executive functioning were performed at baseline, 1-month, and 6-month follow-ups. Genotyping included LMX1A-rs4657412 and APOEɛ-rs429358/rs7412. Primary outcomes included near-transfer and far-transfer training gains and genotype associations at 1 and 6 months. Secondary outcomes were genotypic effects on self-reported functioning. Linear mixed-effect models were used to compare the training gains across training types, HIV serostatus, and genotype variations. RESULTS: A total of 107 participants (60 with HIV, 47 SN) completed adaptive WMT, and 70 active controls (36 with HIV, 34 SN) completed nonadaptive WMT. Overall, 96 adaptive WMT participants and 68 active controls completed the 1-month follow-up while 77 adaptive WMT participants and 37 active controls completed the 6-month follow-up. Adaptive WMT led to higher improvement indices in SN participants than in the HIV group (training*HIV serostatus: p = 0.038, β = -0.22 [-0.33, -0.12]) and similar near-transfer and far-transfer effects (training type*visit: p < 0.001-0.028, β = 0.25-0.64 at 1 month, β = 0.29-0.77 at 6 months) in both groups, but greater improvements in far-transfer tasks in the HIV group at 6 months (training type*HIV serostatus*visit: p = 0.001-0.039, β = 0.60-1.11 at 6 months). Compared with LMX1A-G carriers, LMX1A-AA carriers showed higher improvement indices (training*LMX1A genotype: p < 0.001, β = 0.39 [0.29, 0.49]). While both genotype groups showed near-transfer and far-transfer gains, LMX1A-G carriers gained more after adaptive WMT than nonadaptive WMT in WM (training type*LMX1A genotype*visit: p = 0.015, β = 0.90 [0.31,1.48] at 6 months) and visuoperceptual (training type*LMX1A genotype*visit: p < 0.001, β = 1.18 [0.66, 1.69]) domains. Compared with non-ɛ4 carriers, APOE ɛ4 carriers showed greater improvement on processing speed (APOE ɛ4 genotype*training type*visit: p = 0.008, β = 0.68 [0.14, 1.22] at 1 month, and β = 0.90 [0.25, 1.55] at 6 months). DISCUSSION: Adaptive WMT improved near-transfer WM, far-transfer performance, and self-reported executive functioning in all participants, but more sustained effects among participants with HIV. These findings suggest that adaptive WMT can be an effective adjunctive therapy for cognitive deficits in PWH, especially in those with LMX1A-G/GA and/or APOE ɛ4 carriers. CLASSIFICATION OF EVIDENCE: This study provides Class IV evidence that adaptive WMT improves near-transfer WM, far-transfer performance, and self-reported executive functioning in patients with or without HIV, regardless of LMX1A or APOE ɛ4 genotype status. NCT02602418 (registered at ClinicalTrials.gov, Registered: May 11, 2015. First patient enrolled: September 15, 2012.