This study is the first to provide a comprehensive description of baseline NCF, including variability of cognitive profiles and the incidence of impairments, in a large representative cohort of patients with LGG who are eligible for PRT because of their favorable prognosis. These patients are considered to have high chances for long term survival and to benefit most from the brain and NCF sparing capabilities of PRT compared to photon RT. The LGG patients in this study were selected for PRT following three nationally defined eligibility criteria indicating favorable prognosis [21]. However, neurocognitive status is not an ingredient of these criteria, and our data show that not all selected patients with favorable clinical characteristics have a favorable NCF profile.

We found that there was substantial variability in NCF profiles ranging from completely intact to overall impaired. The average composite total and domain NCF scores were significantly lower than the norm-scores of the healthy population, indicating the effects of tumor and previous treatment. However, when assessed individually, the majority of the patients (66.2%) had a favorable NCF profile without cognitive domain impairments. On the other end of the spectrum, there was a subgroup of patients (11.3%) performing on an overall impaired level. Cognitive impairments were most frequently observed in the memory domains. Relevant clinical and radiological covariates associated with impaired NCF identified in this study are a treatment history with Wait & Scan policy, left temporal or occipital lobe tumor involvement, tumor laterality, tumor volume, neurological function as assessed by NANO and the radiological appearance of the surrounding brain as assessed by Fazekas and GCA score. Wait & Scan policy is a strategy considered in patients with more favorable features, applied in 35.8% of patients in our cohort. At start of PRT, these patients did not perform worse than their counterparts, indicating that deferring the neuro-toxic effects of radio-chemotherapy did not come at the expense of a more unfavorable NCF in these patients. However, the changes in function over the Wait & Scan trajectory were not measured.

We found that working memory and verbal memory were most severely affected prior to PRT, in line with findings of Lemaitre et al. [12] of neurocognitive performance three months post-surgery. This may be a finding typically seen in patients with left-sided, mostly temporal LGG, in the early post-surgery phase that might improve over time [19]. Our finding of a relation with verbal memory between left temporal lobe tumor involvement and immediate post-surgery referral (without Wait & Scan policy) supports this postulation.

We aimed to identify the subgroup of patients performing overall on an unfavorable, impaired level. However, in literature no uniform definition for NCF impairment is used [23]. We compared 3 commonly applied definitions, and observed a very large difference in number of patients subsequently labeled as impaired. With definition 1 (total composite Z score < -1.0), a relatively small subset of patients (11.3%) was selected, that overall displayed the lowest performance and more diffuse NCF damage. Definition 2 (ICCTF criterium whole test battery) classified a much larger subset of patients (64.2%) as impaired. Definition 3 (ICCTF criterium on core test set), selected an intermediate number of patients (39.1%). Consequently, depending on the definition of NCF impairment applied, different associated clinical and radiological covariates were found. Definition 1 can be regarded as a rather crude method for LGG patients, especially when appraising the large heterogeneity of expressed NCF profiles. Definition 2 is not a very practical as extended test batteries are not likely being conducted in many other centers and may this method may provide an overestimation. In our opinion, definition 3 is the preferred method as it is a balanced alternative.

In this cohort of IDH mutated LGG patients, NCF impairment was not related to LGG subtype, nor to epilepsy associated factors, although these are considered important covariates for NCF [16]. Depending on the definition of NCF impairment used tumor size (CTV as surrogate), right-sided LGG location and neurological function (NANO score), were relevant factors. A new interesting finding in our study is the association between NCF impairment and the radiological aspect of the brain (Fazekas and GCA score). Despite the young age and limited comorbidities of patients, the prevalence of any sign of white matter damage or atrophy was relatively high: 24.5% and 38.4% respectively. These radiological indices could be a reflection of the patient’s vulnerability to neurotoxicity and/or the ability to respond and adapt to damage.

Furthermore, overall limited associations between tumor location and cognitive domain scores were found, for which the slow growth of LGG, the plasticity and adaptation of the brain over years in response, as well as the functional organization of the brain with underlying neuronal networks are likely explanations [8, 26, 27]. However, we still found associations between: (1) the left temporal lobe and verbal memory; and (2) the occipital lobe and processing speed. The first finding is in line with other studies [19, 28]. Visual field deficits could be a good explanation for the second finding. In our study processing speed is evaluated by primarily visual tasks, and patients with visual field deficits (NANO visual field sub-score 1 or higher, n = 8) displayed significantly lower processing speed (mean composite Z score − 0.56 versus 0.09, p = 0.05), and patients with occipital tumors were more likely to have visual field deficits.

The strength of this study is the large, homogeneous sample size and the very high compliance rate with the NPA of our prospective registration program (net compliance rate 94%), reflecting the relevance for patients and caregivers. Therefore, these findings can be regarded as real-world data. However, it is important to note that the eligibility criteria for PRT as well as the timing of RT treatment in the disease course of LGG patients are variable between countries.

A limitation of our study is the lack of NPA pre-surgery and during Wait & Scan trajectory. Therefore, we cannot report on the pre-diagnosis functioning of the patients or the duration of impairments. The qualitative scoring of tumor lobe involvement that was used in this study, is a practical method that can be easily adopted in clinical practice. However, as this is a very crude way to assess tumor involvement, a voxel-based methodology might be more sensitive to detect correlations between tumor location and specific NCF impairments [28]. In this study we did not measure pre-operative tumor volumes. In the patients with Wait & Scan policy and/or multiple resections it is not straight forward to measure a pre-surgery volume. Therefore, we used the PRT CTV as a surrogate in this study.

Our baseline data show that overall patients considered eligible for PRT have a favorable NCF profile, but also that a subgroup has NCF impairment. It is yet unknown what the effects of PRT will be for this group, differentiated across impairments levels or cognitive domains. To date, there is still very limited knowledge on the relation between RT dose and NCF changes over time [16, 17, 29]. Therefore, well-organized clinical registration studies to obtain high-quality longitudinal NCF data after PRT are inevitable to move forward [25]. This knowledge can drive future RT technique optimization, and might have an important value for improving clinical outcome [30, 31].