The Effect of Exercise on Cancer-Related Cognitive Impairment and Applications for Physical Therapy: Systematic Review of Randomized Controlled Trials

Abstract

Background: Cancer-related cognitive impairment (CRCI), often called "chemo-brain" or "chemo-fog," is a common side effect among adults with cancer, which can persist well after treatment completion. Accumulating evidence demonstrates exercise can improve cognitive function in healthy older adults and adults with cognitive impairments, suggesting exercise may play a role in managing CRCI.

Purpose: The purpose was to perform a systematic review of randomized controlled trials (RCTs) to understand the effect of exercise on CRCI.

Data sources: Relevant literature was retrieved from CINAHL, Medline (Ovid), and EMBASE.

Study selection: Eligible articles were RCTs that prescribed aerobic, resistance, combined aerobic/resistance, or mind-body (eg, yoga or Qigong) exercise during or following cancer treatment and included cognitive function outcome measures.

Data extraction: Descriptive information and Cohen d effect sizes were directly extracted or calculated for included trials.

Data synthesis: Twenty-nine trials were included in the final analysis. A statistically significant effect of exercise on self-reported cognitive function, both during and postadjuvant treatment, was reported in 12 trials (41%) (Cohen d range: 0.24-1.14), most commonly using the EORTC QLQ-C30. Ten trials (34%) performed neuropsychological testing to evaluate cognitive function; however, only 3 trials in women with breast cancer reported a significant effect of exercise (Cohen d range: 0.41-1.47).

Limitations: Few RCTs to date have evaluated the effect of exercise on CRCI as a primary outcome. Twenty-six trials (90%) in this review evaluated CRCI as secondary analyses.

Conclusions: Evidence supporting exercise as a strategy to address CRCI is limited. Future research evaluating CRCI as a primary outcome, including self-reported and objective measures, is needed to confirm the possible role of exercise in preventing and managing cognitive impairments in adults with cancer.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagram of the literature search used.

Figure 2

Authors’ risk of bias assessment using the Cochrane Risk of Bias tool.

Figure 3

Effect sizes for interventions using the cognitive function subscale of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire–Core 30 (EORTC QLQ-C30) as a self-reported outcome measure of cognitive function. Oechsle et al (2014) also used the EORTC QLQ-C30 but did not report means or SDs. As such, effect size could not be calculated. Risk of bias: A = random sequence generation; B = allocation concealment; C = masking of participants/personnel; D = masking of outcome assessors; E = incomplete outcome data; F = selective reporting; G = other bias. Red = high risk of bias; yellow = unclear risk of bias; green = low risk of bias. ^*Statistically significant effect at P < .05 as reported in original data. ^‡Different exercise intervention arm of the same trial. ~, data from different publication of the same trial. Black markers = group effect. Gray markers = group x time effect.

Figure 4

Effect sizes for interventions using other self-reported outcome measures of cognitive function. Oechsle et al (2014) noted a significant benefit of exercise for self-reported cognitive function as measured by the Modified Fatigue Impact Scale (MFIS)–Cognition subscale. Bryant et al (2018) used the Patient-Reported Outcomes Measurement Information System (PROMIS)–Applied Cognition and General Cognitive Concerns as a self-reported outcome measure of cognitive function but did not observe a significant effect. Neither trial reported means or SDs, so effect sizes could not be calculated. App Cog = Applied Cognition; BCPT = Breast Cancer Prevention Trial–Cognitive Problems; Cog Failures = Cognitive Failures Questionnaire; Comments = Comments of others; FACT-Cog = Functional Assessment of Cancer Therapy–Cognitive Function; FAQ = Fatigue Assessment Questionnaire–Cognitive Fatigue; Gen Cog = General Cognition Concerns; MDASI = MD Anderson Symptom Inventory–Memory Difficulty; MFI = Multidimensional Fatigue Index–Mental Fatigue; PCA = Perceived Cognitive Abilities; PCI = Perceived Cognitive Impairments; PFS = Piper Fatigue Scale–Cognitive/Mood Fatigue; POMS = Profile of Mood States–Confusion; QOL = Quality of Life; SCTS = Stem Cell Transplantation Symptoms–Cognitive Cluster; SOSI = Symptoms of Stress Inventory–Cognitive Disorganization. Risk of bias: A = random sequence generation; B = allocation concealment; C = masking of participants/personnel; D = masking of outcome assessors; E = incomplete outcome data; F = selective reporting; G = other bias. Red = high risk of bias; yellow = unclear risk of bias; green = low risk of bias. ^*Statistically significant effect at P < .05 as reported in original data. ^‡Different exercise intervention arm of the same trial. ~, data from different publication of the same trial. Black markers = group effect. Gray markers = group x time effect.

Figure 5

Effect sizes for interventions using objective outcome measures of cognitive function. Campbell et al (2018) used the Stroop test and Peterson et al (2018) used the Controlled Oral Word Association Test (COWA) as objective measures of cognitive function. Neither study noted a significant effect. Neither study reported means or SDs, so effect sizes could not be calculated. ACT = Auditory Consonant Trigram; Animals = animal naming; Attn to Rep = Attention to Response Task; BCOG = Brief Cognitive Status; D2 = D2 Test of Attention; Digit Span B = Digit Span Backwards; Digit Span F = Digit Span Forwards; HVLT = Hopkins Verbal Learning Test; LNC = Letter/Number Coding; LNS = Letter/Number Sequencing; Log Mem 1/2 = Logical Memory 1/2; RAVLT = Rey Auditory Verbal Learning Test; TMT A/B = Trail Making Test A/B. Risk of bias: A = random sequence generation; B = allocation concealment; C = masking of participants/personnel; D = masking of outcome assessors; E = incomplete outcome data; F = selective reporting; G = other bias. Red = high risk of bias; yellow = unclear risk of bias; green = low risk of bias. ^*Statistically significant effect at P < .05 as reported in original data. ^‡Different exercise intervention arm of the same trial. ~Data from different publication of the same trial. Black markers = group effect. Gray markers = group x time effect.