A 63-year-old man with no significant past medical history presented to the emergency department with a 1-week history of confusion, left-sided hemiparesis, and right-sided hemianopia. A CT scan of the head showed an isodense right temporo-occipital intrinsic lesion with surrounding edema and a significant mass effect. An MRI confirmed a 6.5 cm mass in the right temporo-parietal region with homogeneous contrast enhancement, diffusion restriction, and hyperintensity on T2/FLAIR, with 14 mm of leftward midline shift and marked perilesional edema causing ongoing mass effect (Figures 1A–E). A CT scan of the chest, abdomen, and pelvis (CAP) did not show any primary tumor. Given the main differential diagnoses discussed at the neuro-oncology multidisciplinary meeting (primary CNS lymphoma versus high-grade glioma), the patient underwent an image-guided burr hole biopsy, which, alongside the intraoperative smear findings, supported a diagnosis of GCA. Given the presence of a glial tumor with a significant mass effect in an otherwise healthy patient, tumor resection was undertaken. A trans-sulcal, minimally invasive parafascicular approach (MIPS) was performed with the aid of preoperative tractography, intraoperative neuromonitoring, ultrasound, and 5-ALA guidance. (Figures 1F,G). An extradural slip electrode was inserted, and baseline somatosensory, motor-evoked, and visual-evoked potentials (SSEPs, MEPSs, and VEPs) were measured. The macroscopic appearance of the tumor was pale grey, and it fluoresced avidly under ultraviolet light. The superior margin of the resection was limited by a drop in VEPs to 20% of the baseline. The extent of the resection was guided anteriorly by fluorescence, superiorly by the lateral ventricle, and inferiorly by the temporo-occipital arachnoid and the limit of fluorescence (Figure 1G). A post-operative MRI scan showed no residual enhancing tumor at the T1 contrast sequences, consistent with gross total resection (Figure 1H). Following surgery, the patient’s recovery was slow, with reduced mobility and a requirement for assistance with activities of daily living. The patient was treated with dexamethasone 2 mg once daily for tumor-associated cerebral edema, which was later converted to prednisolone due to a steroid-related rash. During disease progression, escalation of steroid therapy (to up to 85 mg of prednisolone daily) was associated with transient symptomatic improvement, although this effect diminished over time. The patient received the standard Stupp protocol of chemoradiotherapy, comprising 60 Gy in 30 fractions with concurrent temozolomide (145 mg daily), followed by five cycles of adjuvant temozolomide. He was also enrolled in the TRIDENT trial and received tumor-treating fields (Optune), which were discontinued approximately 3 months after completion of chemoradiotherapy due to radiological progression. Early post-treatment imaging, performed within weeks of completing chemoradiotherapy, raised the possibility of pseudoprogression, as the changes occurred within the radiotherapy field and within the expected temporal window. However, on short-interval follow-up imaging, there was progressive solid nodular enhancement with diffusion restriction and increasing lesion size, fulfilling RANO criteria for true disease progression, as confirmed at neuro-oncology multidisciplinary review. The recurrence was centered at the margins of the resection cavity, with subsequent ependymal extension and, over the following months, development of leptomeningeal dissemination involving the upper cervical spinal cord and cranial nerve complexes was observed. In the context of ongoing clinical and radiological deterioration, re-irradiation was not pursued. Following further progression, it was agreed that no further active oncological treatment would be offered, as this was unlikely to improve quality of life. The patient was transitioned to specialist palliative care and died within approximately 1 month of cessation of active treatment and 11 months post-diagnosis.

A 54-year-old man presented to his local Accident & Emergency Department with symptoms indicative of increased intracranial pressure, including a 2–3 weeks history of fatigue, unsteadiness, vomiting, confusion, and intermittent headaches with minimal response to routine analgesics. An MRI of the head revealed a large, heterogeneous space-occupying lesion centered in the right frontal lobe, which was causing a significant mass effect on the right ventricle and supratentorial brain. The lesion extended across the midline to the left side, resulting in a 9.2-mm midline shift, brain herniation, and brainstem crowding (Figures 3A–D,H). The patient was commenced on dexamethasone 4 mg twice daily for 3 days pre-operatively to reduce tumor-associated vasogenic edema and optimize neurological status. Following surgical resection of the contrast-enhancing portion of the tumor, the patient’s symptoms improved, and he was discharged on postoperative day 3 without any focal neurological deficits (Figure 3G). Following surgical resection, the patient received standard Stupp protocol chemoradiotherapy (60 Gy in 30 fractions with concurrent temozolomide), followed by adjuvant temozolomide administered over three cycles (150–200 mg/m² on days 1–5 of a 28-day cycle). Interval imaging approximately 4–5 months post-operatively demonstrated stable appearances at the primary resection site but revealed a new enhancing lesion in the left cerebellar dentate region. Temozolomide was therefore discontinued. The case was discussed at a multidisciplinary team meeting, and the lesion was deemed unamenable to surgical resection. The patient underwent focal re-irradiation to the posterior fossa (40 Gy in 15 fractions). The patient was also assessed for clinical trial enrollment but was not eligible due to radiological disease progression in a high-risk location with concerns about fourth ventricular compression. Following re-irradiation, the patient experienced a gradual clinical decline with increasing fatigue and worsening mobility but remains alive and under ongoing clinical follow-up at the time of writing (approximately 1 year after initial presentation).

GCA is a rare but recognized morphological variant within IDH-wildtype glioblastomas, as noted in the 5th edition of the WHO Classification of CNS Tumors (2021) [1]. Large tumor cells with abundant PAS-positive, diastase-resistant granular cytoplasm may either be scattered throughout the tumor or dominate the histological picture, sometimes obscuring conventional astrocytic features [2]. The granular cell phenotype reflects lysosomal accumulation and can mimic histiocytic cells or macrophages, particularly in the presence of perivascular inflammation, raising the possibility of various histiocytoses (e.g., CNS Rosai-Dorfman disease or Erdheim-Chester disease), a brain infarct, or demyelinating processes [9]. In our first case, bland granular morphology predominated, with no obvious conventional high-grade glioma cells. In contrast, our second case exhibited a biphasic appearance, with granular cells and classic high-grade astrocytic areas, a pattern seen in approximately half of the cases reported in the literature [2, 3, 5]. This morphologic spectrum reinforces the concept that GCA is a histological variant of IDH-wildtype glioblastoma, rather than a distinct clinicopathological entity. Our cases exemplify both extremes of this spectrum and highlight the diagnostic challenges posed by granular cell-dominant tumors, particularly when working with limited biopsy material or intraoperative smears.

Immunohistochemically, GCAs typically express GFAP and S100, consistent with astrocytic differentiation; however, they often show atypical patterns that can complicate diagnostic interpretation. Olig2, a key transcription factor associated with glial lineage commitment, is frequently reduced or absent in granular cell variants [5]. In our first case, granular cell morphology was accompanied by strong GFAP and S100 expression, focal EMA reactivity, and only sparse Olig2 staining; these features mirror those reported by Brat et al. and, more recently, by Vizcaíno et al. [2, 5]. The tumor was negative for CD68 and HLA-DR, arguing against a histiocytic lesion. Interestingly, the WHO notes that granular tumor cells may occasionally express macrophage markers, such as CD68, likely reflecting their abundant lysosomal content [1]. However, expression of CD163 (a more specific marker of the monocyte–macrophage lineage) is typically absent, supporting a non-histiocytic origin. Our second case showed a similar immunoprofile with slightly more robust GFAP expression and a higher Ki-67 labeling index (∼20%), consistent with its overt high-grade glial morphology.

A lack of correlation between histological grade or Ki-67 indices and clinical outcome in GCAs has been reported previously: in one series, no significant association was found between survival and Ki-67 index [5]. Our findings reinforce this and caution against using morphological features or proliferation index as evidence of indolent behavior in these tumors. Case 1 illustrates this particularly well: although the tumor lacked mitotic activity, necrosis, and definite microvascular proliferation, the integrated histomolecular profile was entirely in keeping with glioblastoma, IDH-wildtype, CNS WHO grade 4. In this setting, vascular prominence alone should not be equated with microvascular proliferation or grade 4 designation, as similar appearances may be encountered in other tumor types, including pilocytic astrocytoma, high-grade astrocytoma with piloid features, pleomorphic xanthoastrocytoma, and ependymoma, and therefore require interpretation in the full clinicopathological and molecular context.

To date, molecular studies consistently support the inclusion of GCAs within the spectrum of IDH-wildtype glioblastomas [1, 5]. While methylation-based classification is now integrated into the diagnostic algorithm of the WHO CNS 5th edition, this has not previously been reported for granular cell variants. Brain classifier v12.5 and v12.8 (DKFZ) placed both of our reported tumors firmly within the mesenchymal subtype of IDH-wildtype glioblastoma with calibrated scores of 0.99 and 0.98, respectively. The genomic profiles included canonical glioblastoma alterations, such as gain of chromosome 7 and loss of chromosome 10 (+7/–10), homozygous deletion of CDKN2A/B, and additional changes, including a TP53 variant and amplification of the EGFR, CDK4, and DDIT3 genes (Case 1), and a TERT promoter mutation (Case 2). These findings mirror those in previous case reports and case series, in which TERT promoter mutations and +7/–10 copy-number changes were seen in the majority of cases [4, 5]. Interestingly, NF1 alterations were identified in both cases; each was predicted to result in a premature stop codon and a truncated, non-functional protein. In Case 1, this was accompanied by additional NF1 loss. These findings are consistent with the reported enrichment of NF1 alterations within the mesenchymal subclass of glioblastoma [10]. Neither tumor harbored IDH1/2 or ATRX mutations, and both lacked MGMT promoter methylation, correlating with a poor response to alkylating agents and reduced overall survival [11].

As previously mentioned, survival in GCAs appears to be poor regardless of histological characteristics. Schittenhelm and Psaras reviewed 59 cases and reported median survival times of 9–11 months, regardless of the presence of high-grade features on histology [12]. Vizcaíno et al. similarly reported a mean overall survival of just 11.3 months in a cohort of 39 patients with no survival correlation to grade, Ki-67 index, or extent of granular cell change [5]. These observations suggest that grading based solely on histology may underestimate the clinical aggressiveness in GCAs. Our cases further support this, particularly Case 1, which lacked overt high-grade features, yet it demonstrated a full molecular profile of glioblastoma with multiple gene amplifications and recurred rapidly despite all therapeutic efforts (resulting in an overall survival of 11 months).

This raises the important question of how survival outcomes might be improved for this tumor variant. From a surgical perspective, optimization of the extent of resection remains a key consideration, and enhanced intraoperative visualization could play a role in achieving this goal. To date, no published data are available regarding the use of 5-ALA fluorescence in GCAs. In Case 1, intraoperative 5-ALA administration resulted in bright, homogeneous fluorescence, a pattern more commonly seen in CNS WHO grade 4 gliomas and less frequently in lower-grade gliomas [13, 14]. However, 5-ALA should not be interpreted as a tool for formal tumor grading. Rather, its established role is as an adjunct for intraoperative visualization of tumor tissue and optimization of the extent of resection.

Regarding adjuvant therapy, temozolomide did not appear to halt tumor progression in either case, likely reflecting the unmethylated MGMT promoter status. Nevertheless, a growing number of clinical trials across Europe targeting progressive or recurrent IDH-wildtype glioblastomas may offer potential opportunities for personalized treatment in the near future. For instance, patients with biallelic NF1 loss may be eligible for MEK inhibitor–based trials, such as the 5G-RUBY study (as in Case 1) [15], while those with homozygous MTAP deletion may benefit from emerging PRMT5 inhibitor trials (as in Case 2) [16].