Introduction

Survival outcomes for patients with intracranial gliomas are significantly influenced by the molecular characteristics outlined in the latest World Health Organization (WHO) 2021 classification1. A noteworthy concern has emerged from several prior studies that did not integrate the recently established glioma classification into their analyses. This omission may lead to misleading conclusions concerning contemporary glioma diagnoses and treatment outcomes. It has been highlighted that changes in standard-of-care treatment protocols and molecular stratification have resulted in a shortage of up-to-date survival data for glioma patients. One of the largest multi-institutional cohort studies indicated that over 25% of gliomas underwent a revision in their final histopathologic classification based on molecular criteria2. The management of WHO grade 2 to grade 4 gliomas primarily involves surgical resection followed by radiotherapy or radiochemotherapy3,4. Tumor Treating Fields (TTFields) therapy may enhance treatment outcomes, particularly in supratentorial WHO grade 4 astrocytoma (AST) and glioblastoma (GBM)4,5. An expanding array of targeted therapy agents, biological treatments, and immunotherapeutic strategies is currently being evaluated, particularly for patients with lower-grade, IDH-mutated tumors6,7. Numerous contemporary theories link treatment outcomes to the size of target volumes, the radiation total dose delivered within brain tissue, radiotherapy dose, and their relationship to radiation-induced leukopenia/lymphopenia (RIL) in circulating leukocytes and lymphocytes8. The inclusion of chemotherapy can further influence this relationship.

A recent analysis evaluated the effects of RIL on the survival of GBM patients, considering different censoring time frames: at a specific time point versus across a defined time range9. The findings indicated a mean overall pooled incidence of lymphopenia at 31.8%, with 11.8% and 39.9% reported for time-point and time-range studies, respectively. Furthermore, lymphopenia was found to be associated with an elevated risk of mortality, yielding a pooled hazard ratio (HR) of 1.78 (95% Confidence Interval [CI] 1.46–2.17, p < 0.00001) for time-range studies and a pooled HR of 1.38 (95% CI 1.24–1.55, p < 0.00001) for time-point studies9.

We aimed to evaluate treatment outcomes in the WHO2021- CNS-5 reclassified real-world multi-institutional cohort of WHO G2–G4 glioma patients. The prognostic value of RIL and other specific clinical and treatment-related factors were assessed.

Material and methods

This study was conducted in four radiation therapy institutions in Poland and France. The institutional databases were searched retrospectively for patients with intracranial gliomas verified after surgical resection with the following inclusion criteria:

  1. (a)

    Age > 18 years old;

  2. (b)

    WHO G2-G4 brain gliomas diagnosed according to WHO 2021 classification;

  3. (c)

    Complete molecular information on IDH1/2 mutation and 1p19q codeletion;

  4. (d)

    MGMT promoter methylation status for WHO G4 gliomas;

  5. (e)

    Adjuvant conventional (fraction dose of 1.8–2.0 Gy) or hypofractionated radiation therapy (> 2 Gy per fraction);

  6. (f)

    Concomitant and/or adjuvant temozolomide (TMZ) or adjuvant PCV (procarbazine, lomustine, vincristine) chemotherapy was allowed;

  7. (g)

    At least 6 months follow-up time from completion of radio/radiochemotherapy.

Primary endpoints in this cohort were OS and PFS, providing a dataset aligned with the newest WHO 2021 glioma classification.

We also aimed to evaluate well-known and find new clinical and dosimetric prognostic factors for OS and PFS. As potential predictors, we selected:

  1. 1.

    Clinicopathological factors: sex, age, glioma grade, histopathological subtype, IDH1/2 mutation, 1p19q codeletion, MGMT promoter methylation, chemotherapy addition to radiotherapy, ongoing corticosteroid therapy at the start of radio(chemo)therapy, leukopenia and lymphopenia grade.

  2. 2.

    Radiotherapy-related variables: total dose in EQD2, conventional vs. hypofractionated approach, and CTV (clinical target volume).

  3. 3.

    Dosimetric related variables: total brain volume (TBV), uninvolved brain volume (UBV, defined as the TBV minus the planning target volume [PTV]), V5Gy (%), V10Gy (%), V15Gy (%), V20Gy (%), V25Gy (%), V30Gy (%), V40Gy (%), V50Gy (%), mean brain dose (MBD), median brain dose (MedBD), and whole brain minimal dose (WBMD) (Fig. 1).

Fig. 1
figure 1

Flowchart for the presented study.

Full size image

Baseline white blood cell count (WBC) and absolute lymphocyte count (ALC) and their values during chemoradiotherapy were collected weekly for six weeks (depending on radiotherapy length) and in the first and second months after radiotherapy. The highest grades of leukopenia and lymphopenia, according to CTCAE v5.0, were additionally evaluated (Supplementary Table 1).

Patients with at least 6 months of follow-up were included. For patients diagnosed before the publication of the WHO 2021 glioma classification, a molecular analysis of IDH1/2 and 1p19q codeletion was performed prospectively from the paraffin blocks. Our previous study described the determination of 1p19q codeletion by FISH and IDH 1/2 mutation with sequencing10.

OS was counted from the beginning of radio(chemo)therapy to death or was censored to the date of the last follow-up. PFS was counted from the beginning of radio(chemo)therapy to radiological progression diagnosed with MRI/CT, death, or was censored to the date of last follow-up.

The study was conducted following the latest version of the Declaration of Helsinki. The Medical University of Lublin Ethics Committee approved the study (approval no. KE-0254/349/2018). The requirement for informed consent was waived for the retrospective part of the treatment by the Medical University of Lublin Ethics Committee, while written informed consent was obtained from patients included in the prospective evaluation of selected molecular variables. We have followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines, and the checklist is presented in Supplementary Materials 1.

Statistical analysis

Data preparation and analysis were performed in Python using the libraries scikit-learn11 and statsmodels12. The collected dataset contained missing values (< 5%). Two imputation methods were applied to minimize the impact of missing data: k-nearest neighbors (kNN) imputation and iterative imputation. This method was applied to hematological parameters- WBC and ALC- measured over consecutive weeks. Survival analysis was conducted using the lifelines library13,14. The study included non-parametric methods (Kaplan–Meier) for describing survival functions across groups and Cox proportional hazards regression to identify factors influencing event risk. Proportional hazards assessment, confidence interval estimation, and survival plot visualization were also performed. Data censoring was conducted following standard practices for survival analysis, separately for PFS and OS. In the absence of event information, the censoring time was defined as the date of data retrieval from the medical information systems of the respective healthcare institutions responsible for patient care.

In specific analyses, an approach based on comparing survival functions across groups defined by varying threshold values was applied to determine a risk predictor’s optimal cutoff value for time-to-event data. Patients were divided into two groups for each potential cutoff of a continuous variable (e.g., estimated risk), and their survival distributions were compared using the log-rank test. The threshold that yielded the most statistically significant difference in survival between the groups (i.e., the lowest p-value of the log-rank test) was selected as optimal15,16. The implemented function was developed in Python, and through integration with the lifeline’s library.

Principal component analysis with multiple factor analysis (PCA-MFA) was performed. The first PCA included MGMT promoter methylation, 1p/19q codeletion, IDH1/2 mutation, glioma grade, gross tumor volume (GTV), CTV, concomitant chemotherapy, ongoing steroid therapy, WBC nadir, ALC nadir, PFS (months), and OS (months). The second PCA assessed relationships between dosimetric parameters (e.g., V5Gy–V50Gy) and hematologic measurements during and after radiotherapy. Only patients with progression or death events were included; thus, PFS and OS represent event times.

Results

The cohort included 179 glioma patients (45 WHO G2, 51 G3, 83 G4). None received 3D-CRT; all were treated with IMRT (55.9%) or VMAT (44.1%). The predominant regimen was 60 Gy in 30 fractions. Median PTV was 264.4 mL. Chemotherapy was administered in 74.9% of cases. The diagnosis of AST and OGD exhibited similar frequencies among WHO G2 and G3 gliomas (14.0% and 11.2%, and 17.3% and 11.2%, respectively). Molecular alterations including IDH1/2, 1p/19q codeletion, and MGMT promoter methylation are detailed in Table 1.

Table 1 Basic clinicopathological characteristics of patients and their treatment.
Full size table

Progression-free and overall survival by sex and age

Male patients had significantly longer PFS than females (median 6.4 vs. 1.7 years, p = 0.018), with no difference in OS (median 8.0 vs. 6.5 years, p = 0.175) (Supplementary Fig. 1). Among patients with WHO G2/G3 gliomas, age < 40 vs. ≥ 40 years had no significant impact on PFS or OS (median 7.7 vs. 8.0 years, p = 0.74 and 7.9 vs. 8.6 years, p = 0.40; respectively) (Supplementary Fig. 2).

Survival by WHO grade

WHO G2 gliomas showed the most favorable prognosis, with median PFS and OS not reached; 10-year PFS and OS were 68% (95% CI 51–79%) and 70% (95% CI 53–82%), respectively. For WHO G3 gliomas, median PFS and OS were 5.9 years (95% CI 3.7–6.8) and 6.0 years (95% CI 3.8–7.2), with corresponding 10-year rates of 24% and 23%. WHO G4 gliomas demonstrated the poorest outcomes (PFS: 1.0 year, 95% CI 0.7–1.2; OS: 2.5 years, 95% CI 1.7–6.3). PFS and OS differences across all grades were statistically significant (PFS: p < 0.0001; OS: WHO G2 vs. G3 p < 0.0001, G2 vs. G4 p < 0.0001, G3 vs. G4 p = 0.0436) (Fig. 2, Table 2).

Fig. 2
figure 2

Kaplan–Meier progression-free survival and overall survival regarding WHO 2021 subtype and grade.

Full size image
Table 2 The PFS and OS for WHO 2021 classified gliomas.
Full size table

Impact of molecular markers

IDH1/2 mutation was associated with significantly improved PFS (7.7 vs. 1.0 years, p < 0.0001) and OS (8.2 vs. 2.5 years, p < 0.0001) (Fig. 3A,B).

Fig. 3
figure 3

Kaplan–Meier progression-free survival and overall survival regarding IDH1/2 mutation status (A,B), 1p191 codeletion(C,D).

Full size image

In the total cohort, 1p/19q codeletion predicted longer PFS (7.7 vs. 1.6 years, p = 0.0002), but not OS (8.2 vs. 6.0 years, p = 0.0792). Among WHO G2/G3 gliomas, 1p/19q status was not prognostic for PFS or OS (both p > 0.5) (Fig. 3C,D, Supplementary Fig. 3).

In GBM patients, MGMT promoter methylation did not significantly impact PFS or OS (0.9 vs. 1.0 years, p = 0.23; NR vs. 2.5 years, p = 0.11) (Supplementary Fig. 4).

Treatment-related factors

Among WHO G2 gliomas, chemotherapy was associated with worse outcomes (PFS: NR vs. 7.7 years, p = 0.0447; OS: NR vs. NR, p = 0.0317) (Fig. 4A,B). In contrast, for WHO G3 gliomas, chemotherapy significantly improved PFS (6.8 vs. 3.6 years, p = 0.0014) and OS (6.9 vs. 3.9 years, p = 0.0024) (Fig. 4C,D).

Fig. 4
figure 4

Kaplan–Meier progression-free survival and overall survival in patients with WHO G2 gliomas (A,B) and WHO G3 gliomas (C,D) according to chemotherapy addition to radiotherapy.

Full size image

Steroid use at the beginning of radiochemotherapy was linked to worse PFS (0.9 vs. 3.8 years, p = 0.034), with no significant OS difference (7.2 vs. 7.0 years, p = 0.18) (Supplementary Fig. 5).

Hematologic markers

Patients with leukopenia grade 0–2 within two months after radiochemotherapy had significantly longer PFS (3.6 vs. 1.2 years, p = 0.02) and OS (7.2 vs. 3.2 years, p = 0.04) than those with grade 3–4 toxicity (Fig. 5A,B).

Fig. 5
figure 5

Kaplan–Meier progression-free survival and overall survival regarding the leukopenia grade 0–2 vs. 3–4 (A,B) and ALC nadir (C,D) in the total cohort and.

Full size image

Similarly, patients with ALC > 1 × 103/mm3 had significantly longer OS (8.7 vs. 5.3 years, p = 0.0043) and PFS (6.8 vs. 1.4 years, p < 0.001) (Fig. 5C,D).

Radiotherapy dose and volume

No significant differences in OS or PFS were found between patients receiving EQD2 doses < 54 Gy vs. ≥ 54 Gy in WHO G2/G3 gliomas (OS: 8.2 vs. 7.9 years, p = 0.74; PFS: 8.2 vs. 7.6 years, p = 0.56) (Fig. 6). Similarly, no significant OS or PFS differences were observed between conventional and hypofractionated radiotherapy in WHO G4 gliomas (OS: 2.2 vs. 2.9 years, p = 0.39; PFS: 1.0 vs. 0.7 years, p = 0.14) (Supplementary Fig. 6).

Fig. 6
figure 6

Kaplan–Meier progression-free survival and overall survival regarding EQD2 dose administered (< 54 Gy vs. ≥ 54 Gy) in patients with WHO G2 and G3 gliomas.

Full size image

For WHO G4 gliomas, a CTV cut-off of 127 cm3 was prognostic for OS (1.7 vs. 3.2 years, p = 0.012), but not for PFS (0.7 vs. 1.1 years, p = 0.31). No such associations were observed in WHO G2/G3 gliomas (Fig. 7).

Fig. 7
figure 7

Kaplan–Meier progression-free survival and overall survival with the optimal CTV cut-off value of 127 cm3 for patients with WHO G4 gliomas.

Full size image

Multifactorial analysis

PCA-MFA was performed to explore the correlations among clinical, molecular, treatment-related, dosimetric, and hematologic variables. In the first analysis (Fig. 8A), the first two components explained 50.5% of the total variance (PC1: 33.1%, PC2: 17.4%). Glioma grade, molecular alterations (IDH mutation, 1p/19q codeletion, MGMT methylation), concomitant chemotherapy use, hematologic nadirs, and steroid use were major contributors. PFS and OS were aligned primarily with PC1, indicating that worse hematologic parameters and steroid use were associated with inferior survival outcomes. In the second analysis (Fig. 8B), the first two components explained 52.9% of the variance (PC1: 31.2%, PC2: 21.7%). Strong clustering of dosimetric variables (V5Gy–V50Gy) was observed, with moderate associations with hematologic parameters. PFS and OS showed weaker.

Fig. 8
figure 8

Principal component analysis (PCA-MFA) of clinical, molecular, treatment-related, and hematologic variables (A) and PCA-MFA of dosimetric parameters and hematologic variables (B) and their association with progression-free survival (PFS) and overall survival (OS).

Full size image

Discussion

This study presents the most extensive analyses of patients with WHO 2021-classified gliomas (grades 2–4), demonstrating clear survival differences across grades: median OS was not reached for G2 gliomas (10-year OS: 70%), 6 years for G3, and 2.5 years for G4. Prognostic factors varied by grade. In G2–3 gliomas, IDH1/2 mutations and 1p/19q codeletion were associated with improved survival, while MGMT methylation showed no significant impact in GBM.

Survival outcomes were influenced by glioma grade, treatment intensity, and hematologic toxicity. Patients with milder leukopenia or ALC > 1 × 103/mm3 had significantly longer OS and PFS. While prior studies focused on isolated hematologic markers, our PCA integrated molecular, clinical, and hematologic factors, highlighting their interdependence. Higher-grade gliomas were linked to more frequent chemotherapy use, contributing to hematologic decline. Steroid use correlated with poorer OS, likely reflecting worse clinical status. Additionally, larger CTV predicted shorter OS in WHO G4 gliomas. A second PCA confirmed strong clustering of dosimetric parameters, with only modest associations with hematologic toxicity.

Our analysis fulfills the gap in existing literature, which relies on the previous WHO 2016 glioma classification17. It inadequately addressed treatment outcomes in the context of contemporary radiotherapy/radiochemotherapy. Modern radiotherapy approaches incorporate adaptive strategies18, hypo- and hyperfractionation19,20,21, dose intensification through simultaneous integrated boost22,23, target volume reduction24,25,26, and more conformal techniques27.

Many gliomas initially classified under WHO 2016 would be redefined under the WHO 2021 system, which emphasizes molecular over histological criteria28. Re-analyses of RTOG 0424 and 9802 revealed that 32.5% and 24% of WHO G2 cases were reclassified as IDH-wildtype29,30, while 33% of patients in RTOG 9813 shifted from WHO G3 to G431.

We included WHO G2 gliomas and reclassified all cases per WHO 2021 using updated molecular testing of archival samples. While OS did not differ by sex, PFS was shorter in females, contrary to reports suggesting better OS in women with G2/G4 gliomas32,33. A Swedish registry study found marginal OS benefit only in women undergoing radical surgery34. Age ≥ 40 years was not prognostic for PFS or OS in G2/G3 gliomas, consistent with prior OGD analyses35. However, age as a continuous variable predicts OS in IDH-mutant gliomas (HR 1.03, p = 0.018)30, and OS is poorer in GBM patients ≥ 60 years (HR 1.56, p = 0.0082)36, though our cohort included too few older patients for subgroup analysis.

Patients with WHO G2 AST and OGD in our study had the most favorable survival, with mOS and mPFS not reached. In updated RTOG 0424, mOS was 9.4 years and 8.8 years and mPFS was 8.1 years and 7.5 years for G2 OGD and AST, respectively29. RTOG 9802 reported mOS of 13.9 years for OGD and 6.9 years for AST; mPFS was 10.2 years and 3.9 years, respectively30. For G3 OGD and AST, our mOS was 6.9 and 5.3 years, compared to RTOG 9813 values of NR and 8.7 years, respectively, noting BCNU/CCNU use in RTOG31. Our mPFS for G3 IDH-mutant gliomas (5.8 years AST, 6.4 years OGD) was consistent with RTOG 981331. For GBM, our mOS and mPFS were 2.5 years and 1.0 year, comparable to RTOG 9813 (mOS 1.9, mPFS 0.6 years)31, Guo et al. (mOS 15.6 months, 95% CI 12.5–21.9)36, and molecularly reclassified GBMs in RTOG 0424/9802 (mOS 2.3 and 1.9 years; mPFS 1.0 and 0.7 years)29,30.

Our study demonstrated the positive prognostic significance of IDH1/2 mutations for OS and PFS. These findings are consistent with the results reported in RTOG0424 and RTOG9802, which indicated HR of 0.2 and 0.42 for OS, along with HRs of 0.29 and 0.31 for PFS, respectively29,30. A recently published analysis from RTOG9813 further corroborated these results, showing that patients with IDH1/2 mutations experienced a longer OS (8.7 years vs. 1.9 years; HR = 0.34) and PFS (5.6 years vs. 0.6 years; HR = 0.41;) compared to those with wild-type (wt) status31. Additionally, our findings indicated that patients with 1p19q codeletion exhibited longer PFS and OS, with median values of 7.7 years versus 1.6 years (p = 0.0002) and 8.2 years versus 6.0 years (p = 0.0792), respectively. However, in the context of WHO grade 2/3 gliomas, the status of 1p19q codeletion did not demonstrate prognostic value for PFS and OS, as indicated by median values of 7.7 years versus 7.7 years (p = 0.57) and 8.3 years versus 8.2 years (p = 0.74). The RTOG 0424 trial yielded similar findings; in the overall cohort, patients with 1p19q codeletion exhibited significantly longer OS, with an HR of 0.36, and PFS, with an HR of 0.49. It is noteworthy that univariable analysis (UVA) did not reveal significant differences in OS and PFS between WHO G2 codeleted and non-codeleted gliomas, which aligns with our findings29. Furthermore, the multivariable analysis (MVA) of the RTOG 9802 trial indicated that the presence of 1p19q codeletion was associated with improved OS, with an HR of 0.27, and PFS, with an HR of 0.37, in the total cohort. The UVA for WHO G2 gliomas specifically revealed a more favorable OS, with an HR of 0.31, and PFS, with an HR of 0.46 for patients with codeletion30. There are currently no results from the RTOG 9813 trial regarding the role of codeletion in WHO G3 gliomas, as most participants (96%) had intact 1p/19q status31. These results underscore the necessity for further exploration of the prognostic significance of the 1p19q codeletion, specifically within the population of patients diagnosed with WHO grade 2/3 gliomas.

We analyzed MGMT promoter methylation status exclusively in patients diagnosed with GBM. Previous studies have indicated that MGMT does not provide significant additional value to the WHO subgrouping beyond what IDH and 1p/19q status offer in LGGs29. Our analysis revealed no significant differences in PFS or OS within this cohort. Likewise, a subgroup analysis from the RTOG 0424 trial assessed MGMT methylation status within the GBM subgroup and found no significant difference in OS. The median OS for GBM patients with MGMT methylation was 3.8 years, compared to 2.6 years for those with unmethylated status29. Similarly, in the RTOG 9802 trial, MGMT promoter methylation status did not demonstrate significant differences in OS or PFS among GBMs30. Within the RTOG 9813 trial, survival analyses were not performed for the GBM cohort due to the limited number of GBM patients with MGMT methylation (n = 3). For the non-methylated GBMs, the median OS and median PFS were 1.5 years and 0.5 years, respectively31. However, more recent analyses in WHO-2021-compliant or IDH-wildtype GBM cohorts underline the dual role of MGMT as both a prognostic and predictive biomarker, particularly in relation to benefit from alkylating chemotherapy. Hegi et al. demonstrated no survival advantage from temozolomide in truly unmethylated elderly GBM patients (CE.6, Nordic/NOA-08), while Ghimire et al. confirmed the negative prognostic significance of unmethylated MGMT in inoperable IDH-wildtype GBM37,38. Likewise, a large NCDB-based analysis by Pham et al. showed consistent survival differences according to MGMT status in IDH-wildtype GBMs39. These findings indicate that MGMT methylation retains important prognostic and predictive value in contemporary GBM cohorts, Guo et al. reported that patients with non-methylated GBMs experienced significantly worse mOS across all GBM cases (including molecularly and histologically diagnosed tumors), with an HR of 0.61. However, no precise analysis was explicitly conducted for molecularly classified GBMs36. In a post hoc analysis of the CANTON 3 trial, which included only GBM patients, the addition of temozolomide to radiotherapy did not show a significant effect on either OS or PFS. While MGMT promoter methylation was found to be prognostic for OS, it did not predict the outcome of TMZ treatment for either OS or PFS40.

We have found that patients with WHO G2 gliomas had significantly worse PFS and OS when chemotherapy was added to radiotherapy (median NR vs. 7.7, p = 0.0447 and NR for both, p = 0.0317, respectively). Notably, both recently updated studies—RTOG 0424 and RTOG 9802—focused exclusively on high-risk patient populations identified by criteria such as incomplete resection or age equal to or greater than 40 years29,30. The findings from RTOG 0424 indicated that adding TMZ chemotherapy to radiotherapy significantly enhanced survival outcomes. Similarly, RTOG 9802 reported improved PFS and OS among patients with WHO G2 AST and OGD30. Given these insights, the findings of our study may appear counterintuitive; however, it is essential to clarify that our WHO G2 cohort is not selectively defined and is not solely based on high-risk characteristics. It is plausible to posit that chemotherapy does not augment survival outcomes in unselected cases of WHO G2 gliomas, particularly in light of recent trial results suggesting no advantage from chemotherapy in WHO G2 OGD35. The specific group of WHO G2 gliomas is likely to benefit from combined chemotherapy and radiotherapy and appears to comprise individuals with defined high-risk features. Further validation through prospective trials would be necessary to ascertain whether this patient cohort should be expanded. Conversely, our results indicate a definitive benefit in OS and PFS with the addition of chemotherapy to radiotherapy for WHO G3 gliomas. A study by Allwohn et al. revealed through MVA that radiochemotherapy with TMZ or PCV for WHO G3 gliomas led to prolonged PFS (p = 0.02, HR 0.09)35. Additionally, findings from the RTOG 9813 trial illustrated that patient with WHO G3 gliomas, predominantly AST, who were treated with radiotherapy and TMZ, did not experience differences in OS or PFS compared to those receiving radiotherapy with BCNU/CCNU31. This trial did not compare radiotherapy with the combination of chemoradiotherapy31. The results of the CANTON trial for WHO G3 AST indicated a mOS of 114.4 months with any TMZ combination (concomitant, concomitant and adjuvant, adjuvant alone), vs. 68.2 months in the radiotherapy-only cohort. Furthermore, the mPFS was 77.0 months vs. 34.2 months for patients subjected solely to radiotherapy41.

These findings accentuate the growing importance of molecular testing within contemporary clinical practice, thereby steering therapeutic approaches toward a more personalized treatment framework.

Research has established that large clinical target volumes are associated with significant lymphopenia, which in turn correlates with poorer outcomes, increased mortality, and higher recurrence rates in various primary tumors42,43,44,45. RIL-related prognostic factors were identified for high-grade glioma (HGG)46. They include large planning target volumes (PTVs), brain V25 exceeding 40 Gy, whole brain mean dose greater than 34 Gy, elevated whole brain minimum doses, and treatment techniques such as IMRT versus 3DCRT, as well as the type of radiation modality utilized, specifically photons compared to protons46,47,48,49,50,51.

Our cohort identified a CTV cut-off value of 127 cm3 as a predictor of OS, although it was not a predictor of PFS for patients diagnosed with WHO Grade 4 gliomas (mOS: 1.7 years vs. 3.2 years, p = 0.012; mPFS: 0.7 years vs. 1.1 years, p = 0.31). In the study by Flores et al., a significantly poorer OS was noted among patients with tumors measuring ≥ 60 cm3 and/or ≥ 2000 mm2, as determined by preoperative T2-FLAIR imaging52. Cherlow et al. reported a significant association between tumor size and OS, highlighting that patients with tumors larger than 5 cm in diameter experienced inferior survival outcomes53. Chapman et al. established a substantial relationship between PTV and OS in high-grade gliomas. Their findings indicated that for non-stereotactic techniques, PTV volumes exceeding 131 cm3 were significantly associated with decreased OS54. Guram et al. found no significant differences in OS based on PTV size when margins of 0.4 cm or 1 cm were added to the gross tumor volume, as opposed to standard margins of 2–3 cm in HGG patients55. Our study specifically addressed the CTV, which, according to existing literature, demonstrates a more accurate correlation with OS than PTV. Nevertheless, these findings are not substantiated by Level 1 evidence and necessitate further validation. Furthermore, Liu et al. investigated the relationship between CTV size and survival outcomes, including OS and PFS. Their study divided glioma patients into two groups: one wherein CTV was defined according to European Organisation for Research and Treatment of Cancer (EORTC) guidelines (which included the tumor cavity plus peritumoral edema with a 2 cm margin), and another wherein CTV was defined without including edema, applying only a 2 cm margin to the resection cavity. Following a median follow-up period of 26.4 months, no significant differences in OS or PFS were observed between the two groups (p = 0.418 and p = 0.388, respectively) among the 118 patients involved in the study56. It was also demonstrated that in patients with GBM, incorporating T2-FLAIR alterations into the gross tumor volume (GTV) resulted in promising outcomes, enabling smaller CTV margins and impressive survival rates without marginal recurrences. Additionally, it did not increase the volume of irradiated brain tissue, leading to a potential reduction in toxicity, including RIL57. Treatment-related lymphopenia was linked to reduced survival rates in patients with gliomas treated with chemoradiotherapy58,59. Leukocyte subpopulations from both lymphoid and myeloid lineages contribute significantly to antitumor immunity. And previous work proposed even an nomogram for radiation-induced lymphopenia incorporating risk factors demonstrated fair predictive accuracy60. Although radiation-induced lymphopenia has been widely characterized, the effects of irradiation on other leukocyte subsets, including their differential radiosensitivity and distinct recovery kinetics, are less well understood. Emerging modeling approaches integrating experimental and physiological data allow for detailed prediction of leukocyte depletion and repopulation dynamics after radiotherapy. These models provide a framework to understand radiation-induced leukopenia beyond lymphocyte loss better and may guide the development of tailored radiotherapy-immunotherapy strategies61.

A major strength of our study is its multidimensional approach, combining clinical, molecular, hematologic, and dosimetric factors. By integrating PCA with clinical outcomes, we offer a broader and more comprehensive view of prognostic interactions that previous studies based only on univariate or isolated multivariate analyses could not fully capture29,30,31.

This study has the inherent limitations of performing a retrospective analysis of treatment results with prospective molecular reclassification. Additionally, due to the small sample size of specific subgroups, some studies of well-known prognostic factors like Karnofsky performance status (KPS), resection status, or differences between different chemotherapy approaches were not conducted or did not have sufficient statistical power, requiring further validation in larger cohorts. This includes analyses evaluating the prognostic significance and treatment outcomes of the WHO2021-defined glioma subtypes and new emerging molecular factors. Furthermore, extent of resection—although recognized as a major prognostic factor—was not consistently available across centers, preventing reliable stratification into gross total, subtotal, or biopsy-only categories. In addition, the proportion of missing IDH results was relatively high due to incomplete molecular testing at initial diagnosis in some centers, which may have introduced bias and reduced the accuracy of subgroup analyses. Moreover, hematologic parameters are influenced by many factors, which we acknowledge as a limitation.

Conclusions

This multicenter study provides RWE that molecular, clinical, hematologic, and dosimetric factors significantly impact survival in WHO 2021-classified WHO G2-4 gliomas, emphasizing the need for their integration into future prospective trials and personalized treatment strategies.