A total of 25 patients met the inclusion criteria. The mean age at diagnosis was 13 years (range 2–45 years). The number of male patients was 12, and 13 patients were female.

All patients were examined with brain MRI at 1.5 T (Philips Ingenia) scanner. The MRI sequences were performed were: T2 fast spin-echo [time of repetition (TR)/time of echo (TE), 2,700/100), fluid-attenuated inversion recovery (FLAIR: TR/TE, 9,000/120; inversion time, 2,200 ms), unenhanced T1 spin-echo and at least two planes of contrast-enhanced T1 spin-echo [TR (range)/TE, 600–700/20]. Axial susceptibility-weighted images (TR/TE, 570/30 at 1.5 T) and axial diffusion-weighted images (TR/TE, 8,300/70 at 1.5 T; b 1,000 s/mm², 4 mm thickness).

Various MRI features were evaluated for every case. CT images were reviewed wherever available. All images were independently reviewed by a single neuroradiologist with 12 years of experience in neuroimaging.

MRI assessments included: the location of the mass, presence of solid, and cystic components, their signal intensities on T1- T2-weighted (T1W, T2W), FLAIR, and susceptibility weighted imaging (SWI) images, restricted diffusion within the solid component, enhancement characteristics, presence of calcification, and presence or absence of the “periwinkle sign” or “stellate sign”.

The solid component was characterized as showing high intensity, low intensity, or isointensity relative to grey matter if most (> 90%) of the solid portion had homogeneous signal intensity on T2W images, and heterogeneous if the solid component revealed varying intensities on T2W images. The enhancement pattern of the solid component was defined as homogeneous if enhancement was present in the whole of the tumor volume and heterogeneous, if ring enhancing areas with an irregular rim suggestive of necrosis, were noted within the solid component.

The characteristics of the cystic component assessed were the size of the cyst, signal intensity on T2 and FLAIR relative to cerebrospinal fluid (CSF), presence/absence of blood within the cyst, and presence/absence of wall enhancement.

All the cases were histopathologically confirmed. Histological grade and Mindbomb homolog-1 (Mib) labelling index were noted.

Molecular pathology: testing for ZFTA fusion was available for 8 (32%) of cases. ZFTA fusion was considered positive if L1 cell adhesion molecule (L1CAM) overexpression was positive. Testing for CyclinD1 overexpression was available in 12 cases. Testing for L1CAM and CyclinD1 overexpression was done using immunohistochemistry.

Testing for p53 overexpression was done using immunohistochemistry and was available in 18 cases. Testing for YAP1 was not available.

GTR was considered when no residual tumor was identified on postoperative MRI. Near total resection (NTR) was when > 90% of the tumor was resected and some tumor tissue was left behind. STR was defined as resection of 50–90% of the tumor.

Imaging features including size, location, signal intensity, enhancement, diffusion restriction, and presence of calcification and hemorrhage were described and tabulated. Demographic details, pathological and molecular features, and type of surgical resection were also described, and relevant associations with imaging features were computed and tabulated. All statistical analysis was performed using SPSS software (IBM Corp), version 20.2. P value for categorical variables was calculated using the Chi-square test or Fisher exact test, and for continuous variables was calculated using the Mann-Whitney U test since data is not normally distributed.

Of all the 25 cases in our study, most cases were in the pediatric age group (≤ 18 years of age), with the mean age being 13 years (range 2–45 years). Three patients were below 5 years of age, and 8 patients were teenagers (between 13–19 years of age). Thirteen patients were between 5–12 years of age. A similar age predilection has been documented by other studies [24, 25]. Three patients were in the adult age group. The mean age among ZFTA fused cases was 7 years (range 2–12 years). No sex predilection was observed.

Location: 15 (60%) cases were in the periventricular location, 4 (16%) cases were intraventricular, and 6 (24%) cases were purely cortical. Eleven (44%) of cases were situated in the frontal lobe. A predilection for the left side with 16 (64%) cases was noted.

The typical appearance of STE in the periventricular location is demonstrated in Figure 1.

Intraventricular STE is depicted in Figure 2.

The cortical location of STE is depicted in Figure 3.

Size: the mean maximum axial dimension was 6.5 cm (range 1.5–11 cm). The mean maximum axial dimension for the periventricular location was 8 cm (range 5.5–10.9 cm), for the intraventricular location was 5 cm (range 1–9 cm), and for the cortical location was 4.5 cm (range 2.6–6.5 cm). The mean size of the cystic component in terms of the maximum axial dimension was 4.1 cm and the largest axial dimension of the solid component was 4.6 cm.

Morphology: 19 (76%) cases had a solid and cystic component. The cystic component was hyperintense to CSF on FLAIR, as also observed by Nowak et al. [17], with blood within the cyst in 5 (26%) of cases. The wall of the cystic component showed post-contrast enhancement in 89.5% (n = 17) cases.

The solid component was well-defined, lobulated, and had variable appearances on T1 and T2 weighted sequences, and all cases showed heterogeneous enhancement in the solid component on post contrast sequences and 20 (80%) of cases showed areas of necrosis.

Blooming was seen in 20 cases on susceptibility-weighted MR imaging, out of which 13 cases had CT imaging, and all showed calcification on CT. The “periwinkle sign”, i.e. the flower-like appearance of the tumor with the central area of necrosis and centripetal calcification [25], was observed in 17 (68%) cases. Another pattern of calcification resembling a spoke wheel was observed in these cases, which has been termed the “stellate sign”. The periwinkle sign on MRI and CT is demonstrated in Figures 4, 5, and 6. A significant association was found between the periwinkle/stellate sign and the periventricular location (n = 13, 86.7%, P value = 0.045). It was observed less frequently in the intraventricular location (n = 2, 50%), and purely cortical location (n = 1, 33%).

The demographic and imaging features are summed up in Table 1.

The number of Grade 3 cases was 23 (92%) and 2 (8%) cases were grade 2 tumors. The mean Mib labelling index was 9–11%.

Cases were considered ZFTA fusion positive if L1CAM overexpression was positive. Testing for ZFTA fusion was available for 8 (32%) of cases, out of which 5 (62.5%) cases were ZFTA fusion positive and 3 (37.5%) of cases were negative. Testing for CyclinD1 overexpression was available in 12 cases, out of which 11 were positive and 1 was negative. It is associated with poor prognosis [17]. Histopathology and immunohistochemistry results in a ZFTA fusion positive case are shown in Figure 7.

Of the ZFTA fusion positive cases, 60% (n = 3) were periventricular in location, and 40% (n = 2) were intraventricular in location. None were purely cortical. All 5 ZFTA fused cases were associated with a cystic component, had necrosis, and showed the periwinkle/stellate sign. Figure 8 depicts the Stellate sign in a ZFTA fused case. A significant correlation was found between the presence of ZFTA fusion and the presence of cysts (P value = 0.048). All 5 ZFTA fused cases in this study were WHO grade 3.

Testing for p53 overexpression was available in 18 cases, out of which 12 (66.7%) were positive and 6 (33.3%) were negative. A significant association was found between the presence of p53 overexpression and necrosis (P value = 0.005) with 100% (n = 12) of p53 positive cases showing necrosis, and two p53 negative cases showing necrosis.

The pathological features are summed up in Table 2.

The associations of ZFTA fusion with imaging features are summed up in Table 3.

A total of 64% (n = 16) of cases underwent GTR, 24% (n = 6) cases underwent NTR and 12% (n = 3) of cases underwent STR. It was observed that smaller lesions were found to be more likely to undergo GTR, and larger lesions underwent NTR or STR. Purely cortical lesions were found to have higher rates of GTR, with all 6 of the cortical lesions observed undergoing GTR, while 53% (n = 8) and 50% (n = 2) of periventricular and intraventricular lesions respectively underwent GTR. ZFTA fused cases had higher rates of NTR (60%, n = 3), with 20% (n = 1) of cases undergoing STR and the remaining 20% (n = 1) of cases undergoing GTR.

There was a significant association between the periventricular location of the tumor with presence of a cystic component, (P value = 0.023) with 93% (n = 14) of periventricular tumors in this study showing a cystic component. 50% (n = 2) of intraventricular tumors also showed a cystic component, and 50% (n = 3) of purely cortical tumors had a cystic component.

The imaging features and associations according to intraventricular, cortical, and periventricular locations are summed up in Table 4.