Histopathological and molecular characteristics of IDH-wildtype glioblastoma without contrast enhancement: Implications for clinical outcomes

Abstract

Background

Glioblastoma (GB) heterogeneity poses substantial challenges for diagnosis and treatment. Isocitrate dehydrogenase (IDH)-wildtype GB may lack contrast enhancement on MRI and exhibit a “low-grade radiologic appearance” (non-contrast-enhancing [CE] GB), a phenomenon with unclear clinical implications. This study investigates the histopathological and molecular differences and survival outcomes between CE and non-CE GB.

Methods

This retrospective study at Heidelberg University Hospital analyzed 457 IDH-wildtype GB cases (09/2009–01/2021). Contrast enhancement on preoperative MRI was volumetrically assessed, classifying tumors as non-CE/CE GB using a 1 cm³ cutoff. Molecular and histopathological features, including microvascular proliferation, necrosis, and overall survival (OS), were compared between the groups.

Results

Of the initial cohort, 352 (77%) patients met the inclusion criteria, with 44 (12.5%) non-CE and 308 (87.5%) CE GB. The histopathological assessment revealed that non-CE GB was less likely to present traditional hallmarks of GB, such as microvascular proliferation (39% vs. 94%) and necrosis (25% vs. 92%) (P < .001). In the non-CE group, 24 patients (55%) were diagnosed as molecular GB, compared to only 8 patients (3%) in the CE group (P < .001). A significant difference was observed in Ki-67 levels, with non-CE GBs having a lower mean Ki-67 index of 18% ± 12% compared to 26% ± 13% in CE tumors (P < .001). The median OS was 27.2 months (95% CI, 19.8–NA) for non-CE and 14.7 months (95% CI, 13.2–17.1) for CE GB (P = .0049).

Conclusions

IDH-wildtype GBs without contrast enhancement are often diagnosed based on molecular criteria due to less frequent histopathological hallmarks and are associated with prolonged OS.

Key Points.

• Non-contrast-enhancing (CE) glioblastomas (GBs) show a more indolent clinical course with longer overall survival.

• Non-CE GBs often lack traditional histopathological hallmarks, such as microvascular proliferation or necrosis.

• Non-CE GBs demonstrate a lower proliferation index on histopathology (Ki-67).

Importance of the Study.

Our study demonstrates that isocitrate dehydrogenase (IDH)-wildtype glioblastoma (GB) that exhibits a “low-grade radiologic appearance” on MRI without contrast enhancement is less likely to present traditional hallmarks of GB on histopathological assessment. Hence, it is more likely to be diagnosed on purely molecular criteria, as recently introduced in the World Health Organization classification, and shows a more indolent clinical course with more prolonged overall survival.

Glioblastomas (GBs) are the most common and aggressive form of adult brain tumors. Despite advancements in understanding the pathophysiology and the advent of novel treatments, the outlook remains grim, with a median survival of 15 months.¹ Diagnostic processes rely on histopathological examination and molecular genetic analysis. The 2021 World Health Organization (WHO) classification permits the diagnosis of GB based on specific molecular criteria, even without traditional histopathological hallmarks such as microvascular proliferation and necrosis.²

MRI plays a crucial role in the initial diagnosis, staging, and response assessment of GB patients, offering superior soft tissue contrast and sensitivity compared to CT scans.³ Typically, GBs exhibit a T1-weighted (T1w) hypo- to isointense signal with central heterogeneity indicative of necrosis or intratumoral hemorrhage, a T2-w hyperintense signal, and irregular contrast enhancement patterns reflecting increased blood–brain barrier permeability.^4,5 GBs without contrast enhancement are less frequent, and existing studies present conflicting findings on the survival benefits associated with these GBs exhibiting low-grade imaging features. While Tesileanu et al. did not demonstrate an overall survival (OS) benefit for GB that lacks contrast enhancement,⁶ a recent large-scale study by the RANO resect group demonstrated superior OS as confirmed in propensity score analyses by matching IDH-wildtype GB with contrast enhancement those without contrast enhancement.⁷

In this single-center, retrospective study, we aim to explore whether IDH-wildtype GBs with low-grade radiologic appearance (absence of contrast enhancement) on preoperative MRI have a distinct histopathological or molecular phenotype and a more indolent clinical course compared to tumors with classical radiologic appearance and presence of contrast enhancement.

Materials and Methods

Dataset

The Ethics Committee of the University of Heidelberg provided ethical approval for this study, issuing an exemption from the need for informed consent (Approval Numbers: S-320/2012 and S-784/2018). This research incorporated a cohort of consecutive patients diagnosed with IDH-wildtype (IDH-wt) glioma at the Heidelberg University Hospital from 09/2009 to 01/2021. These patients underwent preoperative MRI scans at the Heidelberg University Hospital’s Department of Neuroradiology, Heidelberg, Germany, resulting in a sample size of 457 individuals. OS, clinical treatment details, and data on initial clinical symptoms were extracted from the patient’s medical records. When a patient’s death date was not recorded—owing to ongoing survival or loss to follow-up—the most recent medical interaction date was noted, and the patient’s data were subsequently treated with censoring.

Histopathology and DNA Methylation Profiling

Following the WHO 2021 criteria, an IDH-wt glioma was only classified as a GB if (a) microvascular proliferation and/or necrosis was described in the histological sample or (b) specific genetic alterations, namely mutation of the telomerase reverse transcriptase promoter (TERTp) or amplification of the epidermal growth factor receptor (EGFR), or a chromosomal pattern characterized by the simultaneous gain of chromosome 7 and loss of chromosome 10 (+7/−10) were present, so-called molecular GB.² Histological parameters, including microvascular proliferation, necrosis, and the Ki-67 proliferation index, were derived from the primary histopathological report. Unclear cases were re-evaluated by a board-certified neuropathologist with several years of experience. The Ki-67 index was assessed through immunohistochemistry utilizing a monoclonal mouse anti-human Ki-67 antibody. Genomic DNA was extracted from tumor tissues and analyzed to map DNA methylation profiles throughout the genome using either Illumina 450k or EPIC platforms, incorporating, bespoke methods, as elaborated in previous studies.^8,9 To categorize the samples further, the classification schema for brain tumors established by the German Cancer Research Center (Deutsches Krebsforschungszentrum, DKFZ), version v12.5, was applied.¹⁰

MRI Examinations and Evaluation of Contrast Enhancement

Data acquisition was conducted using 3 Tesla MRI systems, including Magnetom Verio, TIM Trio, or Skyra models from Siemens Healthcare, equipped with a 12-channel head-matrix coil. Our study protocol conformed to internationally accepted standards for medical imaging,¹¹ encompassing the capture of T1w scans before (T1) and after (cT1) the infusion of a gadoterate meglumine solution (0.1 mmol/kg, Dotarem). Additionally, 2D fluid-attenuated inversion recovery (FLAIR) and T2w images have been obtained.

To determine the tumor volume, we performed automatic tumor segmentation utilizing a nnUNET-modified version of the HD-GLIO, to delineate various tumor zones, including necrotic, contrast-enhancing (CE) precisely, and T2/FLAIR hyperintense regions without enhancement.^12,13 MF, a neuroradiology fellow with 6 years of experience, verified the results of this automated segmentation process. Tumors that exhibited contrast enhancement of 1 cm³ or less were categorized as non-contrast-enhancing (non-CE) tumors based on the premise that enhancements below 1 cm³ are challenging to quantify accurately and that prior studies have not demonstrated significant differences in outcomes between patients with no contrast enhancement and those with enhancements ranging from 0 to 1 cm³, after adjusting for the presence of non-CE tumors.^14,15

Statistical Analysis

Data analysis was performed using R software (Version 4.2.2, provided by the R Foundation for Statistical Computing). The analysis used mean values and their SDs for continuous variables. Categorical variables, on the other hand, were presented as frequencies and percentages. The comparison of continuous variables was facilitated by applying the Wilcoxon–Mann–Whitney test. In contrast, categorical variables were examined using chi-squared tests.

This study determined statistical significance by setting the P-value cutoff at less than .05 for all conducted analyses.

Results

Patient Cohort

In total, 457 adult patients with newly diagnosed IDH-wt glioma were initially considered for this study. Of these, 352 (77.0%) patients, aged between 28 and 86 years, fulfilled the inclusion criteria and were included in further analysis. To maintain substantial tissue volume and enhance the significance of histopathological findings, n = 64 patients who underwent stereotactic biopsy and insufficient representative tissue could be obtained were omitted from the study. An additional exclusion criterion was applied to n = 15 patients whose histopathological samples showed only an infiltration zone and no bulk tumor. A total of n = 16 patients were omitted from the study due to genetic profiles indicative of alternative glioma entities. Additionally, n = 10 patients, despite comprehensive characterization, did not conform to the predefined classification schema (Not Elsewhere Classified).¹⁶ Within the cohort of n = 352 patients ultimately enrolled, n = 308 patients (87.5%) presented with CE tumors, whereas n = 44 patients (12.5%) exhibited non-CE tumors Figure 1. Figure 2 illustrates the FLAIR and post-contrast T1 sequences for 2 representative patients from each study group.

Figure 1.

Flowchart depicting patient selection and exclusion criteria for the study cohort of IDH-wt glioma. Initially, 457 IDH-wt glioma patients were identified, with successive exclusions for stereotactic biopsies, infiltration zone presence without bulk tumor, and non-conforming genetic profiles, resulting in a final cohort of 352 patients diagnosed with GB. Based on MRI evaluations, this final group is further subdivided into those with contrast-enhancing (CE) tumors and non-contrast-enhancing (non-CE) tumors. IDH = isocitrate dehydrogenase; GB = glioblastoma; NEC = Not Elsewhere Classified. *With insufficient obtained tissue.

Figure 2.

Comparative MRI scans demonstrating the radiological features of contrast-enhancing (CE) and non-CE GB. The top row displays FLAIR images, highlighting hyperintense signals characteristic of the tumors. In contrast, the bottom row shows T1-weighted post-contrast (T1c) images, illustrating the presence of contrast enhancement in the CE tumor and its absence in the non-CE tumor. FLAIR = fluid-attenuated inversion recovery; GB = glioblastoma.

An analysis of the initial clinical symptoms revealed 2 significant differences between the groups, noting that individual patients could present with more than 1 symptom. Headaches were reported more frequently in the CE group, with n = 40 cases (12%), compared to the non-CE group, where no cases were observed (P = .048). Conversely, seizures were more frequently reported as an initial symptom in the non-CE group with n = 25 (59%) compared to the CE group n = 72 (23%) (P < .001). No significant differences were observed between the groups concerning other primary symptoms, including hemiparesis, cognitive or personality changes, or impairments in speech, vision, or sensory function (P > .09). A detailed breakdown of the initial symptoms in both groups is provided in Supplementary Table 1.

Molecular Differences and Treatment Variations

In the overall cohort, n = 32 cases (9%) were diagnosed as molecular GB (mol-GB). In the non-CE group, n = 24 patients (55%) were diagnosed as mol-GB, in contrast to the CE group, where only n = 8 patients (3%) received this diagnosis, indicating a statistically significant difference (P < .001). Among the n = 32 mol-GB cases, n = 4 (12%) exhibited an isolated TERTp mutation without accompanying EGFR amplification or +7/−10. The prevalence of isolated TERTp mutation was similar between the CE and non-CE groups, with n = 1 case (12%) in the CE group and n = 3 cases (12%) in the non-CE group (P > .99). Other genetic aberrations identified in these 4 isolated TERTp-mutated tumors are presented in Supplementary Table 2. Methylation of the MGMT promoter was observed in 16 patients (36%) in the non-CE group and 140 patients (45%) in the CE group, a difference that was not statistically significant (P = .26). Furthermore, no significant difference was observed in the distribution of methylation classes (P = .92), with the RTK II subgroup being the most prevalent in both cohorts. Additionally, tumors in the non-CE group were notably smaller, with an average preoperative volume of 29 ± 27 cm³, compared to the CE group’s average tumor volume of 95 ± 59 cm³ (P < .001). Radiochemotherapy emerged as the predominant treatment modality for both groups, with 30 patients (68%) in the non-CE group and 205 tumors (67%) in the CE group undergoing this treatment showing no significant differences in treatment approaches (P = 0.85). There was no statistically significant difference in second-line therapy between the CE and non-CE groups (P = 0.07). In both groups, the most frequently administered treatment was combination therapy with etoposide and lomustine (CCNU). An analysis of tumor localization across the different brain lobes revealed no significant differences between the 2 groups (P = 89). The most frequently affected lobe was the temporal lobe, followed by the frontal lobe in both cohorts. Surgical interventions varied between the groups: in the non-CE group, 13 patients (30%) underwent biopsy, 21 (48%) had a subtotal resection (STR), and 10 (23%) had a total resection (GTR); in contrast, in the CE group, 23 patients (7%) underwent biopsy, 194 (63%) had an STR and 90 (29%) had a GTR, showing significant differences in surgery approaches (P < .001) as detailed in Table 1.

Table 1.

Demographic and clinical features of study cohort of patients with newly diagnosed IDH-wt GB with and without contrast enhancement on preoperative MRI. Bold values indicate statistically significant results.

Parameter		Whole dataset	CE	Non-CE *	P-value (CE vs. non-CE)
Total no. of patients [%]		352	308(88)	44(13)	NA
Sex [n (%)]	Female	154 (44)	131 (43)	23(52)	.22
Age [yr]	Mean, STD	63 ± 11	63 ± 11	61 ± 11	.11
Preoperative ECOG [n (%)]	ECOG ≤ 2	335 (95)	293 (95)	42 (95)	.93
GB [n (%)]	Histopathological Molecular	320 (91) 32 (9)	300 (97) 8 (3)	20 (45) 24 (55)	<.001
Methylation subclass [n (%)]	RTK I RTK II MES Other Missing	76 (22) 140 (40) 111 (32) 21 (6) 4 (1)	68 (22) 122 (40) 97 (31) 18 (6) 3 (1)	8 (18) 18 (41) 14 (32) 3 (7) 1 (2)	.92
MGMT status [n (%)]	Methylated Unmethylated	156 (44) 196 (56)	140 (45) 168 (55)	16 (36) 28 (64)	.26
First-line therapy [n (%)]	Radiochemo therapy Radiotherapy alone Chemotherapy alone Resection alone BSC Missing	235 (67) 44 (12) 16 (5) 5 (1) 3 (1) 49 (14)	205 (67) 38 (12) 13 (4) 5 (2) 3 (1) 44 (14)	30 (68) 6 (14) 3 (7) 0 0 5 (11)	.85
Second-line therapy [n (%)]	Etoposid and CCNU Bevacizumab and CCNU CCNU alone Bevacizumab alone TMZ alone Radiochemo therapy Radiotherapy alone Resection alone BSC Other Missing	97 (28) 7(2) 19 (5) 7 (2) 28 (8) 9 (3) 15 (4) 3 (1) 2 (1) 23 (7) 142 (40)	87 (28) 5 (2) 18 (6) 5 (2) 19 (6) 8 (3) 14 (5) 3 (1) 2 (1) 22 (7) 125 (41)	10 (23) 2 (5) 1 (2) 2 (5) 9 (20) 1 (2) 1 (2) 0 (0) 0 (0) 1 (2) 17 (39)	.07
Localization [n (%)]	Frontal lobe Parietal lobe Temporal lobe Occipital lobe Insula	113 (32) 53 (15) 148 (42) 27 (8) 11 (3)	97 (31) 47 (15) 129 (42) 25 (8) 10 (3)	16 (36) 6 (14) 19 (43) 2 (5) 1 (2)	.89
EOR [n (%)]	Biopsy STR GTR Missing	36 (10) 215 (61) 100 (28) 1(0)	23 (7) 194 (63) 90 (29) 1 (0)	13 (30) 21 (48) 10 (23) 0 (0)	<.001
Tumor volumes [cm3 (mean)]	Pre-OP CE Pre-OP non-CE Whole tumor	20 ± 20 57 ± 43 87 ± 60	23 ± 20 61 ± 44 95 ± 59	0.24 ± 0.29 29 ± 27 29 ± 27	<.001

Abbreviations: BSC = best supportive care; CCNU = lomustine; CE = contrast-enhancing; EOR = extent of resection; GB = glioblastoma; GTR = gross total resection; MES = mesenchymal subclass; RTK I = Receptor tyrosine kinase I subclass; STR = subtotal resection.

* Using a cutoff of 1 cm³.

Contrast Enhancement Impact on Survival

The survival outcomes differed significantly between the groups, with the non-CE group exhibiting a median OS of 27.2 months (95% CI, 19.8–NA), compared to 14.7 months (95% CI, 13.2–17.1) in the CE group (P = .0049), illustrated in a Kaplan–Meier curve in Figure 3. Furthermore, a comprehensive multivariate Cox regression analysis, which adjusted for variables such as age, preoperative Eastern Cooperative Oncology Group (ECOG) performance status, and extent of resection (EOR) revealed an influence of contrast enhancement on OS, with a hazard ratio of 0.54 (95% CI, 0.34–0.87, P = .001). The detailed outcomes of this analysis are delineated in Figure 4.

Figure 3.

Kaplan–Meier survival curves comparing the overall survival probabilities between patients with contrast-enhancing (CE) and non-CE GBs over time. The shaded areas represent the 95% confidence intervals. GB = glioblastoma.

Figure 4.

Multivariate Cox regression analysis of prognostic factors affecting overall survival (OS) in the study cohort.

Histopathological Differences

Histological evaluations revealed differences in the 2 groups’ GB-related histopathological features, specifically microvascular proliferation and necrosis. Microvascular proliferation was observed in 291 (94%) patients within the CE group, in contrast to 17 (39%) in the non-CE cohort (P < .001). In a similar vein, necrosis was present in 282 (92%) tumors from the CE group, as opposed to 11 (25%) in the non-CE group (P < 0.001), as depicted in Figure 5A and summarized in Table 2. Additionally, a significant disparity was noted in the proliferation marker Ki-67 index between the groups; the CE group exhibited a mean Ki-67 index level of 26% ± 13%, whereas the non-CE group had a lower mean level of 18% ± 12% (P < .001), as shown in Figure 5B.

Figure 5.

(A) Proportion of contrast-enhancing (CE) and non-CE glioblastomas (GBs) with and without microvascular proliferation (left) and necrosis (right), showing significant differences in histopathological features between the 2 groups (P < .001). (B) Box plot of Ki-67 proliferation index levels in CE and non-CE tumors, indicating a significantly lower Ki-67 index in non-CE GBs (P < .001).

Table 2.

Comparative analysis of the Ki-67 index, microvascular proliferation, and necrosis in the entire study data and among CE and non-CE subgroups. Bold values indicate statistically significant results.

Parameter		Whole dataset	CE	Non-CE	P-value (CE vs. non-CE)
Ki-67 [%]	Mean, STD	25 ± 13	26 ± 13	18 ± 12	<.001
Microvascular proliferation [n (%)]	Present Not present	308 (88) 44 (12)	291 (94) 17 (6)	17 (39) 27 (61)	<.001
Necrosis [n (%)]	Present Not present	293 (83) 59 (17)	282 (92) 26 (8)	11 (25) 33 (75)	<.001

Abbreviations: CE = contrast-enhancing.

Discussion

Due to its superior soft tissue contrast capabilities, MRI is pivotal in GB patients’ diagnostic and therapeutic assessment processes. As the most malignant form of brain neoplasms, GBs characteristically present on MRI with features indicative of necrosis and variable contrast enhancement. However, instances of GBs that do not exhibit contrast enhancement have been sporadically reported. The prognostic parity between non-CE tumors and their CE counterparts remains ambiguous. This retrospective investigation endeavors to elucidate the survival outcomes and histopathological distinctions between CE and non-CE GBs, thereby contributing to a more nuanced understanding of their clinical implications.

Our investigation observed an extended OS in patients with non-CE GB compared to those with CE tumors. Additionally, multivariate Cox regression analysis, adjusted for clinical parameters such as preoperative ECOG performance status, and the EOR, identified the presence or absence of contrast enhancement as an independent significant prognostic indicator for OS. The existing literature presents ambiguous findings regarding survival differences between CE and non-CE GBs. In a recent multicentre study, Karschnia et al.⁷ reported that GBs displaying a “low-grade radiologic appearance” on imaging have a more favorable prognosis than those with typical high-grade features. They also highlighted the prognostic importance of postoperative residual tumor volume in such GBs—conversely, Tesileanu et al.⁶ found no significant differences in OS between non-CE and CE GBs. Notably, their study included a substantial proportion of patients diagnosed with gliomatosis cerebri, defined as a diffuse lesion involving at least 3 lobes, and observed a lower resection rate in non-CE GBs compared to CE GBs 83%. Given the recent findings by Karschnia et al. regarding the prognostic value of surgical resection, these baseline disparities might align the prognosis of non-CE GBs more closely with that of CE GBs. In our study, a higher percentage of non-CE patients underwent biopsy rather than GTR, compared to CE patients, which could introduce a bias in the observed OS in the non-CE group. Despite this, the non-CE group demonstrated significantly longer survival in our analysis.

In our cohort, patients in the CE group were significantly more likely to present with headache as their initial symptom. In contrast, seizures were more frequently reported in the non-CE group. No significant differences were observed between the groups regarding other initial symptoms, such as cognitive deficits. A recently published study by Bruhn et al.¹⁷ investigated the prognostic value of presenting symptoms in GB patients using a cohort of n = 1458 from the Swedish Brain Tumor Register. The study found that patients presenting with seizures or headaches had significantly longer OS than those without these symptoms; however, this association did not persist in multivariate analysis. Similarly, a study by Rilinger et al.¹⁸ reported that clinical seizures were associated with improved OS in patients with high-grade gliomas at diagnosis and during treatment. The mechanisms underlying this potential survival benefit remain unclear and may be influenced by earlier diagnosis, genetic predisposition, or other factors. It has been suggested that seizures could be linked to improved survival through cancer-independent mechanisms, such as increased medical surveillance, more frequent follow-ups, or potential yet unidentified benefits of antiseizure medication use. Importantly, while epileptic activity may have prognostic implications, seizures can significantly impact quality of life. Further research is warranted to elucidate the relationship between seizure activity, disease progression, and survival in GB. Microvascular proliferation and necrosis are 2 histopathological hallmarks of GBs. Our analysis showed these characteristics were markedly less prevalent in non-CE GBs than in their CE counterparts. This observation aligns with other studies that associate the absence of these features with extended OS in non-CE tumors. The role of angiogenesis in the progression of high-grade gliomas was initially demonstrated by Brem et al., who used a rabbit cornea assay to study the effects of a GB fragment on blood vessel formation.¹⁹ Subsequently, the discovery of vascular endothelial growth factor in hypoxic, pseudo-palisading cells near necrotic zones and proliferative vessels further elucidated the mechanism of angiogenesis in gliomas.²⁰ This vascular characteristic has been hypothesized to be crucial for glioma growth, with numerous studies confirming the link between angiogenic activity and prognosis.^21–23 Further supporting this, a 2021 French clinical study found that GB patients with less aggressive histological profiles, characterized by the absence of anaplasia, heightened mitotic activity, necrosis, or vascular proliferation, exhibited remarkably favorable outcomes. Specifically, the median OS reached 88 months when only a TERTp mutation was present, highlighting the prognostic significance of these histological factors.²⁴ In their recent correspondence, Priesterbach-Ackley et al.²⁵ also reported a survival advantage in mol-GB with isolated TERTp mutations compared to those with additional molecular alterations, such as EGFR amplification or +7/−10 chromosomal changes. In our cohort, while we observed a higher proportion of mol-GB cases in the non-CE group, we found no significant difference in the prevalence of isolated TERTp mutations between the CE and non-CE groups. However, it is essential to note that our analysis included only 4 patients with isolated TERTp mutations, limiting the statistical power to detect significant differences.

Ki-67, a nuclear protein, is a marker of cellular proliferation, reflecting the specific growth phase of the cell it marks.²⁶ Our analysis noted a significantly lower average Ki-67 index in non-CE GBs. The relationship between Ki-67 expression in tumor tissues and patient survival has been established across various cancer types.^27–29 For gliomas, the immunohistochemical quantification of the Ki-67 index is the predominant method for assessing cellular proliferation within the clinical diagnostic framework. This index tends to rise in correlation with increasing WHO tumor grades and elevated Ki-67 levels have been linked to decreased OS in patients with lower-grade gliomas and ependymomas.^30–32 However, the prognostic significance of the Ki-67 index in GB patients remains contentious. Studies have yielded mixed results; some researchers have found that higher Ki-67 levels correlate with better OS outcomes,³³ while others have observed that increased Ki-67 levels indicate poorer OS.³⁴ Additionally, there are reports where no significant association between Ki-67 expression and OS was found,³⁵ further complicating the understanding of its prognostic value in GBs. This variability underscores the complex role of cellular proliferation as a biomarker in the prognosis of GB and suggests the need for further research to clarify its implications.

While providing valuable insights into the prognostic differences between non-CE and CE GBs, this study has several limitations that merit consideration. First, the retrospective design inherently limits our ability to establish causality between observed outcomes and contrast enhancement characteristics. Additionally, the single-center nature of the study may restrict the generalizability of the findings to other populations or settings, as institutional practices and patient demographics may influence results. The study also faces limitations in histopathological evaluation. While using WHO criteria for GB classification is a strength, relying on archived histopathological data without re-evaluation might lead to misclassification or oversimplification of tumor characteristics. The intratumoral heterogeneity of GBs presents a significant diagnostic and therapeutic challenge.³⁶ In our cohort, a subset of 8 patients exhibited contrast enhancement on MRI but lacked high-grade histopathological features. While this could suggest a potential sampling bias, all samples were obtained from CE tumor regions, which are generally considered the most aggressive parts of the tumor.^37,38 Given this, the classification as molecular GBs is unlikely to be attributable to the under-sampling of high-grade components. The analysis of Ki-67 was constrained by the variability in assessment techniques and interobserver variability, which could affect the reliability of proliferation data. Furthermore, a limitation of this study is the relatively low number of non-CE GBs, which, although expected, limits the statistical power of our analyses in this subgroup. Additionally, a considerable proportion of OS data points for the non-CE group are censored, which may impact the robustness of our conclusion that these patients exhibit better OS. Moreover, the study’s reliance on OS as the primary endpoint, without considering progression-free survival or quality of life, provides a limited view of the clinical implications of contrast enhancement in GB. Future prospective studies with multicenter involvement, standardized histopathological assessment protocols, and the inclusion of broader clinical endpoints are recommended to validate and expand upon the findings of this study.

In conclusion, IDH-wt GBs without contrast enhancement are often diagnosed based on molecular criteria due to less frequent histopathological hallmarks and are associated with prolonged OS.

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://academic.oup.com/neuro-oncology).

Conflict of interest statement

None declared.

Funding

M.F.-D is funded by the AI Health Innovation Cluster of the Heidelberg-Mannheim Health and Life Science Alliance. G.B. and M.S. are funded by the Physician-Scientist Program of Heidelberg University, Faculty of Medicine. M.A.M. is funded by Else Kröner Research College for Young Scientist (funding number: 2023_EKFK.02). We acknowledge funding from the DFG as part of the Priority Programme 2177 Radiomics: Next Generation of Biomedical Imaging (project identifier: 428223917) and Collaborative Research Center 1389 (UNITE Glioblastoma – project identifier: 404521405). PV is funded through an Else Kröner Clinician Scientist Endowed Professorship by the Else Kröner Fresenius Foundation (reference number: 2022_EKCS.17).

Authorship statement

Study design by P.V., M.F.-D, F.S., and R.B. Data processing and statistical analysis by M.F.-D and R.B. Data collection by M.F.-D, R.B., G.B., M.S., M.A.M., T.K., W.W., M.B., and F.S. Data interpretation by P.V., M.F.-D., F.S., and R.B. First Manuscript draft by M.F.-D. All authors critically revised the manuscript for important intellectual content and approved the final version.

Data availability

The MRI scans and tissue samples used in this study, obtained from Heidelberg University Hospital, are not publicly available due to institutional and ethical regulations.