Working memory training in survivors of pediatric cancer: A randomized pilot study

Publication: Psycho-Oncology

Institution: Children’s National Medical Center, George Washington School of Medicine, Duke University

Investigator(s): Kristina K. Hardy, Victoria W. Willard, Taryn M. Allen, and Melanie J. Bonner

Program: Cogmed RM

Background & Aim: Brain tumor and acute lymphoblastic leukemia (ALL) survivors are at increased risk for developing neurocognitive late effects such as difficulties with attention, working memory (WM), and processing speed. Impairments in these primary functions are associated with declines in cognitive, academic, social, and vocational outcomes. Although psycho-stimulant medication has reduced inattentive symptoms in survivors, susceptibility to side effects in this group are greater than for ADHD children, long term effects are unknown, and there is reluctance to using such medications. Cognitive remediation programs requiring intensive, therapist directed, in person intervention have also been employed for survivors yet, compliance is frequently low and effect sizes small. Given the limitations to existing interventions, there is thus a need to explore the efficacy of interventions targeting late effect symptoms that are also feasible for cancer survivors.

The aim of the current study was to describe the feasibility and preliminary efficacy of Cogmed RM in a small, randomized clinical trial of pediatric brain tumor and ALL survivors. Both participants and caregivers completed an in-house feasibility and acceptability survey. The primary outcome measure was performance on a visuo-spatial working memory task (Symbolic Working Memory; Wide Range Assessment of Memory and Learning, 2nd Edition (WRAML-2)). Secondary outcomes included verbal working memory (Verbal Working Memory; WRAML-2), auditory and visual short term memory (Letter Number and Finger Windows; WRAML-2) and caregiver attention and learning problem ratings (Inattention and Learning Problems; Conners’ 3 Rating Scale). Participants also completed the Side Effects Rating Scale so to systematically record any adverse events.

Population & Sample Size: N = 20 pediatric cancer survivors (ALL or BT), ages 8 -16 years with attention and working memory difficulties, off treatment and medically stable for at least 1 year

• n = 11 survivors in adaptive Cogmed training group
• n = 6 survivors in non-adaptive (placebo) Cogmed training group
• n = 3 survivors excluded due poor compliance, 1 due to lack of post-testing data

Design: Randomized, placebo controlled, test-retest, 3 month follow up

T1 = baseline, T2 = post-test, T3 = 3 month follow up


I. Survivors in adaptive training improved significantly over the placebo control group on a non-trained measure of visuo-spatial working memory (Symbolic Working Memory; WRAML-2) at T2 with gains maintained at T3.

II. No significant group differences at T2 and T3 on:

1) Verbal working memory (WRAML-2)
2) Auditory short term memory (Letter Number; WRAML-2)
3) Visual short term memory (Finger Windows; WRAML-2)
4) Caregiver ratings of attention (Inattention; Conners’ 3 Rating Scale)

III. Survivors in adaptive training improved significantly over the placebo control group on caregiver ratings of learning problems (Conners’ 3 Rating Scale) at T2. No significant difference at T3.

Summary and Implications: This research was the first randomized, placebo controlled trial using Cogmed RM with pediatric cancer survivors. Participants in both groups reported no adverse events on the Side Effects Rating Scale and high levels of satisfaction with training. Compliance was also high, with 85% of participants completing at least 20 out of the 25 required training sessions and 75% completing all of the sessions.

Participants in adaptive Cogmed RM training improved significantly at post-training on a non-trained task of visuo-spatial WM and gains were maintained at 3 month follow up, with effect sizes of 0.94 and 1.22 respectively. Caregiver rating of learning problems also significantly improved for the adaptive training group over the placebo control at post-test (d = 0.80), although the difference was no longer significant at follow up. The researchers also calculated the Reliable Change Index (RCI) to determine whether any participants exhibited clinically meaningful improvement in performance-based measures or ratings of functional impairment. Thirty six percent and 45% of participants exceeded the Reliable Change Index (RCI) threshold for clinically meaningful improvements in visuo-spatial WM and learning problems respectively. Although significant group differences were not observed for the remaining outcomes, a medium sized effect of 0.67 was observed on the visual attention task (Finger Windows; WRAML-2) for the adaptive Cogmed RM training group at post-test.

Limitations of this study included a small sample size, limited battery of attention and WM outcome measures, raters not blinded to group assignment, and lack of teacher behavioral ratings. Despite these limitations, this research provided preliminary support for the feasibility of Cogmed RM as an intervention for ameliorating cognitive late effects in pediatric cancer survivors. Consistent with previous investigations of Cogmed RM, gains post-training were robust for visuo-spatial WM. However, the lack of significant improvement on verbal WM and attention tasks represented a departure from findings of previous research. The researchers posited that one possible reason for this lack of significance was that the verbal WM task from the WRAML-2 was not similar to the trained verbal tasks or to the tasks used to measure verbal WM gains (e.g., digit span task) in previous research. Another possibility was that the population required more intensive verbal WM training.

Future studies should evaluate whether training “dose” will lead to greater transfer and maintenance of training gains. The researchers suggested that cancer survivors should not be expected to demonstrate the same level of benefit as children with other neurodevelopmental conditions and that the mechanisms underlying training-related improvements may be different for cancer survivors. The feasibility of Cogmed RM as a “preventative” tool to offset the development of cognitive late effects is also another avenue of potential research.

Funding: This study was supported by grant RO3 CA 132570 from the National Cancer Institute to Kristina K. Hardy.