Cognition and Anesthesia Exposure in Adolescent and Young Adult Retinoblastoma Survivors
Myron Yaster MD and Lena Sun MD
Retinoblastoma (RB), the most common intraocular malignancy of early childhood.1 It requires intensive surveillance and treatment, often necessitating repeated general anesthetics for examinations and therapeutic procedures. As discussed in previous PAADs here and here, concerns persist regarding the long-term neurodevelopmental effects of such exposures, particularly during early childhood.2,3 In today’s PAAD Belson et al.4 evaluated the relationship between cumulative anesthesia exposure and cognitive outcomes in adolescent and young adult (AYA) retinoblastoma survivors compared with matched healthy controls. I’ve asked Drs. Randy Flick, Stephen Gleich, and Lena Sun, who know much more about this issue than I ever will to help review this article. Myron Yaster MD
And one more thing: A good review of retinoblastoma and the myriad therapies used in treating it can be found in the OpenAnesthesia article by Drs. Young May Chao and Tessa Mandler here.
Original article
Belson PJ, Berry JL, Reid MW, Pike NA. Cognition and Anesthesia Exposure in Adolescent and Young Adult Retinoblastoma Survivors. Paediatr Anaesth. 2026 May;36(5):549-559. doi: 10.1002/pan.70152. Epub 2026 Feb 20. PMID: 41721443.
The study included 98 RB survivors and 97 healthy controls aged 14–26 years, matched for age, sex, and ethnicity. Cognitive function was assessed using the Montreal Cognitive Assessment (MoCA), a widely used 10-to-15-minute screening tool designed to detect mild cognitive impairment and early dementia MoCA website.5 It evaluates multiple cognitive domains including memory, executive function, and attention. The maximum score is 30 and scores > 26 are generally considered normal.
Anesthesia exposure was quantified as the number of events and estimated cumulative duration, with additional analysis of exposures occurring before age three. Statistical analyses included group comparisons, correlation testing, and multivariable regression adjusting for demographic and socioeconomic variables.
OK, what did they find? Obviously, retinoblastoma survivors experienced substantially greater anesthesia exposure than controls and patients with bilateral disease and those receiving chemotherapy or radiation had significantly greater exposure, reflecting higher disease burden and treatment complexity. Cognitive outcomes demonstrated modest but statistically significant differences. Retinoblastoma survivors had lower mean MoCA total scores compared with controls (26.03 vs 26.84; p=0.02), with deficits most pronounced in naming, attention, language, and abstraction domains. Other domains, including visuospatial function, delayed recall, and orientation, were not significantly different. Importantly, mean scores for both groups remained within the normal range, suggesting preserved global cognition at the population level.
Multivariable regression modeling provided further insight. After adjusting for confounders, each additional anesthesia exposure during childhood was associated with a small decrease in MoCA score (β=−0.04, p=0.005). Among RB survivors, both early (<3 years) and later exposures were similarly associated with reduced cognitive scores (β≈−0.07, p<0.01 for both), suggesting that cumulative exposure, rather than timing alone, is relevant. The magnitude of effect was small; even multiple exposures resulted in minimal reductions unlikely to be clinically meaningful for most individuals. Further, socioeconomic factors exerted a stronger influence on cognitive outcomes than anesthesia exposure. Lower socioeconomic status and use of public insurance were independently associated with significantly lower MoCA scores.
Ultimately, the observed cognitive differences in this study were subtle and domain-specific, primarily affecting higher-order functions such as language and abstract reasoning. And, the study has several limitations, including its single-center design, modest sample size, and most importantly, reliance on MoCA as a screening rather than comprehensive neuropsychological assessment.
We think these findings support the overall safety of necessary repeated anesthesia in pediatric oncology while highlighting the importance of addressing broader social determinants of neurodevelopment.
From Stephen Gleich, MD and Randall Flick, MD, MPH
The question of whether early exposure to general anesthesia affects long-term neurocognitive outcomes has been studied extensively for nearly two decades. Much of the early literature arose from analyses of existing cohorts that were originally assembled for other purposes. The classic example being the Wilder study, which repurposed a birth cohort designed to determine the frequency of learning disabilities rather than anesthetic exposure. This context continues to shape the field, including the study by Belson et al.
With all studies based on pre-existing cohorts, interpretation depends heavily on several methodological considerations: the choice of outcome measure, the precision of exposure assessment, and the ability to address confounding. Differences across studies—including the Belson study highlighted in this PAAD—likely reflect variability in these domains as much as true differences in anesthetic effect.
Within this framework, the modest cognitive differences reported by Belson et al. are best interpreted with caution. Observational studies in this area are inherently complex, and it is often difficult to fully disentangle the effects of anesthesia from those of the underlying condition and its treatment. At the same time, detecting subtle neurocognitive changes presents its own challenges. Broad or screening-based outcome measures may not be sufficiently sensitive to capture domain-specific effects. Recent work has consistently shown that while global measures such as IQ or academic achievement are typically reassuring, more detailed neuropsychological testing reveals small but reproducible differences in specific domains. Even large rigorous studies such as MASK, PANDA, and GAS found no differences in global IQ while identifying modest changes in behavior or executive function. In this context, the findings by the authos provide useful information. However, the use of the Montreal Cognitive Assessment—a brief screening tool not validated in adolescents—makes the interpretation of the results uncertain.
A broader challenge, common to nearly all studies in this field, is confounding by indication. Children who require repeated anesthesia often do so because of significant underlying medical conditions and treatments, each of which may independently influence neurodevelopment. In the Belson study, adolescents, and young adults with retinoblastoma—many of whom have undergone chemotherapy, radiation, and prolonged care—are compared with healthy controls. While this comparison is practical and clinically meaningful, it also introduces differences between groups that are difficult to fully account for and that almost certainly influence cognitive outcomes.
The study design does have important strengths, including relatively detailed characterization of repeated anesthesia exposure among retinoblastoma survivors. At the same time, exposure assessment differs between groups, with greater reliance on self-report in controls, and there is limited matching on factors known to influence cognition. These are common and understandable challenges in studies of this type.
Taken together, the findings are consistent with a broader pattern in the literature: small differences in cognitive performance associated with anesthesia exposure that are statistically detectable but of uncertain clinical significance and difficult to attribute definitively. Studies such as this are valuable in extending our understanding, particularly in populations with substantial exposure, while also highlighting the complexity of drawing causal inference. For a more detailed discussion of attributing causation in observational studies, the reader may wish to review the editorial: A Users’ Guide to Interpreting Observational Studies of Pediatric Anesthetic Neurotoxicity: the Lessons of Sir Bradford Hill (Anesthesiology 2012).
In spite of the major limitation in the study related to the instrument used for outcome assessment, the study does offer the opportunity to consider the larger context of the field. Ongoing interest is driven largely by robust animal data demonstrating anesthesia-related neurotoxicity. In contrast, human studies, especially those with more rigorous designs—have generally been reassuring, showing no clear or clinically meaningful deficits in global cognitive outcomes. This does not exclude the possibility of subtle effects but suggests that any such effects are likely small and must be weighed against the clear benefits of necessary procedures.
What was once viewed as a pressing public health concern has evolved into a more nuanced and carefully studied question. Contributions such as the study by Belson et al. are useful as they now focus on more specific populations that may be at risk, and remind us that challenges remain.
Send your thoughts and comments to Myron (myasterster@gmail.com ) and he will post in a Friday reader response.
References
1. Dimaras H, Corson TW. Retinoblastoma, the visible CNS tumor: A review. J Neurosci Res 2019;97(1):29–44. (In eng). DOI: 10.1002/jnr.24213.
2. Xin A, Grobler A, Bell G, et al. Neurodevelopmental Outcomes after Multiple General Anesthetic Exposures before 5 Years of Age: A Cohort Study. Anesthesiology 2025;142(2):308–319. (In eng). DOI: 10.1097/aln.0000000000005293.
3. Warner DO, Flick RP, Sprung J, Wilder RT. Anesthesia and Babies’ Brains: Lessons from the Lounge. Anesthesiology 2026;144(2):453–456. (In eng). DOI: 10.1097/aln.0000000000005808.
4. Belson PJ, Berry JL, Reid MW, Pike NA. Cognition and Anesthesia Exposure in Adolescent and Young Adult Retinoblastoma Survivors. Paediatric anaesthesia 2026;36(5):549–559. (In eng). DOI: 10.1002/pan.70152.
5. Nasreddine ZS, Phillips NA, Bédirian V, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. Journal of the American Geriatrics Society 2005;53(4):695–9. (In eng). DOI: 10.1111/j.1532-5415.2005.53221.x.