Reassessing cognitive impairment prevalence in 5-FU chemotherapy: A critical methodological perspective

We read with interest the systematic review and meta-analysis article by D’Souza and colleagues.[1] We commend the authors for undertaking an evidence synthesis on the important topic of prevalence of cancer-related cognitive impairment (CRCI). However, the research question is unclear. While the title suggests prevalence of cognitive impairment, the stated population, intervention, comparison and outcomes (PICO) and search terms do not explicitly reflect this. Instead, the outcome appears to be “cognitive function,” a broad term that could encompass neurocognitive or self-report assessments and measures of severity, prevalence, and incidence. In addition, the title infers that the review focuses exclusively on patients, yet data on survivors are included. Given this broad outcome, it is surprising that only eighteen studies met the inclusion criteria. Our lab identified 52 studies examining CRCI prevalence in patients with breast cancer alone.[2] Furthermore, we could not find a clear definition of cognitive impairment from the included studies or how it was handled in the meta-analysis. The European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC QLQ-C30) was used as a self-report measure in 39% of the included studies. However, this is a quality of life tool that contains only two items related to cognition. These concerns raise questions about the robustness of the search and inclusion/exclusion criteria. A lack of clarity in the study’s outcomes is further reinforced by the review design. For instance, the summary characteristics table includes data unrelated to prevalence, with several entries lacking prevalence percentages altogether. Key components of robust systematic review (SR) methodology, such as Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) and risk of bias assessment, were mentioned in the methods section but were not reported making it difficult to assess the quality of included studies and certainty in evidence, and whether the reported prevalence rate (32%) is a reliable estimate. Clinically, it is unclear why the review focuses on 5-fluorouracil (5-FU); while used commonly across multiple cancer types, it is rarely used in isolation. Consequently, any neurocognitive effects observed are likely moderated by dosage and other treatments administered concurrently. The authors attempted to analyze the impact of dosage in their meta-analysis, but without a subgroup analysis, it is not possible to contrast findings. In addition, the study does not attempt to subgroup data based on chemotherapy combinations, which would provide valuable clinical insights. Given the difficulty in isolating the specific contributions of each treatment to cognitive impairment and the fact that cancer itself contributes to cognitive impairment, we ponder whether the study would have had greater externally validity if it had reported cognitive impairment prevalence due to cancer treatment as a whole. We believe the authors missed an opportunity to gather valuable insights on CRCI prevalence, tailored to patient demographics and treatment timelines. It is well established that factors such as cancer type, sex, age at treatment, hormonal status, assessment method (self-report versus neurocognitive testing), and time since treatment cessation[3,4] influence both prevalence and severity of cognitive impairment. The lack of subgroup analyses considering these critical factors is, in our opinion, a major limitation of the study. In conclusion, while determining the prevalence of cognitive impairments following chemotherapy is undoubtedly clinically relevant, the methodological flaws of this SR significantly weaken its findings. A more rigorous approach, including clear inclusion criteria, appropriate subgroup analyses, a thorough evaluation of evidence certainty, and a comprehensive risk-of-bias assessment, is essential for generating reliable and clinically useful results. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.