Data from 21 patients treated from April 2010 to December 2021 were analysed. In the same period, we had accrued 94 newly diagnosed patients aged below 18 years with HGG, any histological subtype and extension, not centred in pons.
There were nine females; patients’ median age was 13.4 years (range 1.8–17.8 years). All had received surgery after first imaging. Surgery had consisted of complete tumour removal in 5, partial in 9, plus 2 that also had other locations (encephalic and spinal in one instance each) and were therefore considered metastatic, and 5 had only biopsies. Histological diagnoses originally reported were reclassified according to the 5th WHO classification[8] in 20/21 cases with available material by MA and BP. The revision identified diffuse hemispheric glioma, H3 G34-mutant in 4 cases- one in a patient with Lynch syndrome-, diffuse midline glioma, H3 K27-altered in 6 patients – one also sharing BRAF V600E mutation, diffuse high-grade glioma, H3-wildtype and IDH-wildtype in 8 (one also having constitutional mutation of TP53) and the last case was H3-wildtype, IDH-mutated and with microsatellite instability. Finally a BRAFV600E wild-type anaplastic pleomorphic xantoastrocytoma was diagnosed in one, while for one patient the diagnosis could only be HGG. Primary sites consisted of diffuse cerebral lesions compatible with gliomatosis in 3. Sixteen patients had supratentorial tumours located in basal nuclei in 5, deep frontal in 7 (2 with other nodules in supratentorial compart and spine) and temporo-parietal in 4. The last two patients had a huge brainstem tumour also involving more than half cerebellum in one case and two contiguous lesions in the cerebellar-pontine angle and in the homolateral cerebellar hemisphere in the other one (Table 1).
Table 1
Patient and primary tumour characteristics and treatments
| n (%) |
Sex | |
Female | 9 (42.9) |
Male | 12 (57.1) |
Age (years) | |
Median (1st and 3rd quartile) | 13.43 (8.99–15.76) |
First surgery result | |
Complete resection | 5 (23.8) |
Partial resection | 9 (42.9) |
Partial resection (D+) | 2 (9.5) |
Only biopsy | 5 (23.8) |
2021 WHO tumour classification and H3 gene status | |
Anaplastic pleomorphic xantoastrocytoma | 1 (4.8) |
Diffuse hemispheric glioma, G34-mutant | 4 (19.0) |
Diffuse high-grade glioma, Wildtype | 9 (42.9) |
Diffuse midline glioma, K27-altered | 6 (28.6) |
HGG not better specified | 1 (4.8) |
IDH gene status* | |
Mutated | 1 (11.1) |
Wildtype | 8 (88.9) |
Primary tumour site | |
Cerebral gliomatosis | 3 (14.3) |
Infratentorial | 2 (9.5) |
Double location | 1 (4.8) |
Whole brainstem | 1 (4.8) |
Supratentorial | 16 (76.2) |
Basal nuclei | 5 (23.8) |
Frontal | 5 (23.8) |
Frontal (D+) | 2 (9.5) |
Temporo-parietal | 4 (19.0) |
First radiotherapy site | |
Craniospinal irradiation | 1 (4.8) |
Tumour bed | 15 (71.4) |
Double site | 1 (4.8) |
Whole brain | 4 (19.0) |
First chemotherapy scheme | |
Concomitant and adjuvant temozolomide | 14 (66.7) |
Concomitant and adjuvant nimotuzumab plus vinorelbine | 5 (23.8) |
Sequential high-dose chemotherapy and maintenance | 2 (9.5) |
Abbreviations: WHO, world health organization; D+: dissemination component |
*In diffuse high-grade glioma patients only |
First treatment. All patients had received radiation. Fifteen had RT on tumour bed at 54 Gy (1.8 Gy daily fraction) total dose. Four patients (the three with gliomatosis and one with two supratentorial nodules) had whole brain RT at 39.6 Gy in two cases (1.8 Gy daily fraction), 39 Gy in one case (3 Gy daily fraction) and 34 Gy (1.7 Gy daily fraction) plus a boost up to 48 Gy (2.4 Gy daily fraction) in the last one. The child with the two non-contiguous lesions, aged 1.8 years at diagnosis, received 54 Gy on both nodules. Finally the last patient with metastatic tumour in the spine had craniospinal irradiation (CSI) at 35.2 Gy (1.6 Gy daily fraction) boosting the main tumours up to 54 Gy and the spinal location up to 46 Gy.
All patients also underwent chemotherapy according to the Herby trial[9] with concomitant and adjuvant temozolomide in 14 cases, with concomitant and adjuvant nimotuzumab plus vinorelbine[10] in 5 cases, and with sequential high-dose chemotherapy and maintenance according to a previous institutional protocol in the last 2 cases[11] (Table 1).
Progression, treatment and outcome. Per definition, all the 21 patients had suffered tumour progression. Relapses were locoregional in 12, including 2 patients with gliomatosis and 1 patient with two contiguous lesions: 7 were defined as marginal and 5 in-field. Five patients had local (1 marginal) plus disseminated relapse, including 2 children with already metastatic tumours at diagnosis. Finally, 4 patients had disseminated relapse: 1 encephalic only, 1 in spine only, 2 in both brain and spine. Median PFS and OS were 11.5 months (range 2.5–39.1 months; 1st and 3rd quartile: 8.0-19.5 months) and 23.5 months (range 6.2–59.6 months; 1st and 3rd quartile: 16.4–41.7 months), respectively. One-year PFS and OS were 42.9% (95% CI: 26.2–70.2%), and 76.2% (95% CI: 60.0-96.8%), respectively, and 3-year OS was 30.1% (95% CI: 15.2–59.7%) (Supplementary Fig. 1). At tumour progression, 4 patients had further surgery that was a partial removal in 3 cases and only a biopsy in the fourth; all samples were again centrally reviewed and the same diagnoses made at first surgery were confirmed.
Re-RT sites and further progression after re-RT are illustrated in Fig. 2. The median time gap between first RT and re-RT was 12.4 months was 12.4 months (range: 3.2–43.1 months; 1st and 3rd quartile: 8.6–26.1 months).
The 12 local relapsing patients received local re-RT at doses between 19.8–30.6 Gy (1.8-2 Gy/day); 2 cases had also boosts at 54 Gy (2 Gy/day) and 50.4 Gy (1.8 Gy/day); of 10/12 cases re-evaluated with imaging after re-RT, 6 again had local relapse, 4 local and disseminated.
The 5 patients with local and disseminated relapse after first RT received CSI in 3 cases (range 23.4 Gy-39.6 Gy plus 12 Gy boost in one case, 1.8 Gy/day), whole spine in one case (39.6 Gy plus boost 10.8 Gy, 1.8 Gy/day) and an extended local region re-RT in the last case (30 Gy/day at 2 Gy/day). Subsequent progression in three/4 evaluated further involved local disease in 2 and dissemination in 1, while one child did not progress so far. The 4 with disseminated disease at first relapse received: whole brain RT in one case (30 Gy, 2 Gy/day), CSI into two cases (30.6 Gy on brain and 39.6 on spine and 39.6Gy on brain, all at 1.8 Gy/day), and spine RT in the last one (36 Gy plus a boost at 14.4 Gy, 1.8 Gy/day). They were evaluated for further progression in three cases showing dissemination in two patients and local and disseminated progression in the third one. Twelve patients had other adjuvant treatment during or after re-RT, mainly consisting of procarbazine, lomustine and vincristine (PCV), nimotuzumab plus vinorelbine, oral temozolamide plus etoposide.
The radiological response at first evaluation post-re-RT was SD in 8, PR in 1, pseudoprogression (PsPD) in one, mixed response in one and PD in 10; in those 19/21 patients that had signs and symptoms correlated with disease progression, some benefit could be appreciated in 8 with support therapy reduction or end, stability in 3 and deterioration in 8.
Median time to progression from re-RT was 3.7 months (0.6–16.2 months; 1st and 3rd quartile: 2.1–6.1 months) and survival from re-RT was a median 6.9 months (range: 0.6–17.9 months; 1st and 3rd quartile: 5.1–14.4 months) (Fig. 3).
Three patients received a third irradiation course for 2nd local relapse, 2nd dissemination and local and disseminated progression (after first dissemination).
Eighteen patients died for tumour progression.
Three patients were alive at the time of this report at 19.4, 29.0 and 50.3 months after diagnosis (the first and third after 3 irradiation courses).
Association analyses. First local relapse did not reach statistically significant differences for better PFS (41.7% vs 14.8% at 6 months, P = 0.152) and OS (33.3% vs 20.8% at 1 year, P = 0.211) from re-RT than any distant relapse.
In this selected series of only relapsed patients, complete resection at primary surgery did not affect favourably subsequent outcome: 4/five patients with complete resection had a component of distant relapse (P = 0.038) that appeared at median 8.0 months (range 2.5–11.5) after diagnosis. Two of those 4 patients had a diffuse glioma harbouring H3-G34 mutation; the fifth patient with complete resection at first diagnosis, also having a diffuse glioma with H3-G34 mutation, had synchronous dissemination at second relapse.
Gender, chemotherapy scheme used at first diagnosis, H3 alterations/mutations and the presence of extensive (gliomatosis in 3 children) or contiguous and metastatic disease (one patient with two lesions and two with metastases) were not prognostic for PFS from second RT (PFS2), OS and survival from second RT (Supplementary Fig. 2).
First marginal relapse showed potential association with dissemination after re-RT (P = 0.081).
A longer time span between first RT and re-RT was positively associated with better OS (HR 26 vs 9: 0.69; 95% CI: 0.12–4.04; P = 0.027) and PFS (HR 26 vs 9: 0.27; 95% CI: 0.11–0.66; P = 0.017) from second re-RT.
If patients were grouped according to the response to re-RT (excluding the patient with mixed response), the median OS from re-RT was 14 months for the 10 patients with SD, PR or PsPD versus median 4 months for the 10 with PD (P < 0.001) (Fig. 4).