Cognitive functioning following brain irradiation as part of cancer treatment: Characterizing better cognitive performance

Abstract

Objective Although brain radiation therapy (RT) impacts cognitive function, little is known about the subset of survivors with minimal cognitive deficits. This study compares the characteristics of patients receiving brain irradiation as part of cancer treatment with minimal cognitive deficits to those with poorer cognitive functioning. Methods Adults at least 6 months postbrain RT (N = 198) completed cognitive measures of attention, memory, and executive functions. Cognitive functioning was categorized into better‐ and poorer‐performing groups, with better‐performing survivors scoring no worse than 1.5 standard deviations below the published normative mean on all cognitive measures. Logistic regression was used to identify variables associated with better‐performing group membership. Results Approximately 25% of the sample met the criteria for the better‐performing group. In unadjusted analyses, RT type (whole brain irradiation and partial brain irradiation), sedating medications, and fatigue were independently associated with cognition. Sociodemographic and other clinical characteristics were not significant. In adjusted analyses, only fatigue remained significantly associated with group membership (OR = 1.05, 95% CI = 1.01‐1.09, P = .009). Conclusions There is a subgroup of survivors with minimal long‐term cognitive deficits despite undergoing a full course of brain RT as part of cancer treatment. Lower fatigue had the strongest association with better cognitive performance. Interventions targeting cancer‐related fatigue may help buffer the neurotoxic effects of brain RT.

Prediction Models for Radiation-Induced Neurocognitive Decline in Adult Patients With Primary or Secondary Brain Tumors: A Systematic Review

“…We further excluded 106 publications based on the exclusion criteria. Finally, 23 studies describing the development of prediction models were included, of which 16 ( Gregor et al, 1996 ; Blay et al, 1998 ; Klein et al, 2002 ; Kaleita et al, 2004 ; van Beek et al, 2007 ; Wang et al, 2010 ; Starke et al, 2011 ; Gondi et al, 2012 ; Kangas et al, 2012 ; Chapman et al, 2016 ; Wong et al, 2019 ; Dutz et al, 2020 ; Gui et al, 2020 ; Tibbs et al, 2020 ; Langegård et al, 2021 ; Zamanipoor Najafabadi et al, 2021 ) studies included patients with primary and 7 ( Wolfson et al, 2011 ; Gondi et al, 2013 ; Nakazaki and Kano, 2013 ; Chen et al, 2017 ; Yamamoto et al, 2017 ; Gui et al, 2019 ; Brown et al, 2020 ) with metastatic brain tumors. No external validation studies were identified.…”

Prediction Models for Radiation-Induced Neurocognitive Decline in Adult Patients with Primary or Secondary Brain Tumors: A Systematic Review

“…We further excluded 106 publications based on the exclusion criteria. Finally, 23 studies describing the development of prediction models were included, of which 16 [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] studies included patients with primary and 7 [35][36][37][38][39][40][41] with metastatic brain tumors. No external validation studies were identi ed.…”

“…Regarding the data source, retrospective cohort was the most popular study design used for nine studies [19,20,23,27,31,33,34,37,39], followed by prospective cohort (seven studies) [21,22,25,26,28,30,38], and prospective trials (seven studies) [24,29,32,35,36,40,41]. The median sample size of the cohorts was 81 (IQR=54-266) and the median incidence of neurocognitive toxicity for studies with binary outcome was 26% (IQR=14%-73%).…”

“…The median sample size of the cohorts was 81 (IQR=54-266) and the median incidence of neurocognitive toxicity for studies with binary outcome was 26% (IQR=14%-73%). Three regression analyses were applied for model development: linear regression in 11 studies (48%) [21-24, 27, 28, 30-32, 34, 40], logistic regression in eight studies (35%) [19,25,26,29,33,35,36,39], and Cox regression in four studies (17%) [20,37,38,41]. The intercept of the model was reported in two studies [22,31].…”

“…The median number of signi cant predictors in prediction models was 2 (IQR=1-3). Eight models received a low score for concerns about applicability, which indicates that they are well aligned with the review question [21,29,30,33,[35][36][37][38].…”