Radiation remains a staple therapeutic option for many pediatric and adult malignancies. Given the mutagenic nature of this therapy, patients receiving ionizing radiation in the management of their disease early in life are at risk for future treatment-related complications, including secondary malignancies (1, 2). Survivors of childhood cancer, after exposure to radiation and/or chemotherapy, have a 20.5% incidence of secondary malignancies 30 years after the primary diagnosis (3). Specifically, pediatric patients originally treated with radiation, regardless of chemotherapy or initial disease type, demonstrated a significantly increased risk of developing secondary central nervous system (CNS) neoplasms, including gliomas and meningiomas at median latency periods of 9- and 17-year post-radiation, respectively (4).

Radiation-induced gliomas (RIGs) have long been anecdotally thought to act more aggressively than other de novo gliomas. Indeed, survival data for this disease—which effectively consist of compilations of case reports and retrospective data—have been correspondingly poor with median overall survival (mOS) times of 17 and 9 months for radiation-induced histologic grade 3 and 4 gliomas, respectively (5). Despite prior tissue exposure to radiation and concern for long-term radiation recall pathology, reirradiation has been safely utilized in these patients. In contrast to chemotherapy and surgery, only radiation has been associated with prolonged survival. Both 1- and 2-year overall survival were comparable between this RIG cohort (58.9, 20.5%) and the radiation alone arm in the original Stupp temozolomide (TMZ) trial (50.6, 10.4%), despite the inclusion of grade 3 disease with GB for the RIG group; histologic grade 3 astrocytoma survival was far inferior to what has been seen in the CATNON trial overall cohort (6, 7). For patients with RIG not managed with radiation, 1-year survival has been reported to be less than 20% (5).

In addition, favorable prognostic markers such as isocitrate dehydrogenase (IDH) mutation and methylguanine-O6-methyltransferase (MGMT) promotor methylation are less likely to be found in RIG, particularly the former, supporting unfavorable outcomes for radiation-induced GB (8, 9). More recent reports on high-grade RIGs similarly support these abysmal outcomes with mOS ranging from 6.0 to 8.5 months depending on the original pediatric diagnosis (8).

Despite the disappointing survival outcomes in RIG, reirradiation is a meaningful treatment option for some patients based on retrospective and extrapolated data. It remains unclear which patients are more likely to benefit from this therapy. In this regard, while molecular testing, including genomic sequencing, has emerged in neuro-oncology as necessary testing to establish diagnosis, prognosis, and guide therapeutic options, there remains a paucity of this data for RIGs due to its low incidence albeit with emerging evidence suggestive of a “unifying molecular signature” for this disease (8, 10, 11). Herein, we report two cases of aggressively behaving RIGs, diagnosed as GB, IDH-wildtype, based on histologic and molecular criteria using the revised fourth edition of the World Health Organization (WHO) Classifications of CNS Tumors, with exceptional radiographic responses to radiation therapy. We also provide a detailed description of the histologic examination and molecular evaluation of these tissues to contribute to the limited data reported in the literature and touch upon diagnostic considerations for these neoplasms with the new pathologic classification criteria for gliomas in the WHO Classification of CNS Tumors, fifth edition.

Outcomes for high-grade gliomas, including GB, remain poor with limited recent therapeutic advances beyond radiotherapy, TMZ, and alternating electric tumor-treating fields (7, 13). In contrast to most de novo GBs in which an inciting etiology is unknown, patients with RIGs have the unique unifying factor of clear prior mutagenic agent exposure. Because of this, it has been suggested that these RIGs may share an underlying molecular signature secondary to their pathogenesis (8, 10, 11). Despite the emergence of the next-generation sequencing use in glioma in recent years to better understand disease biology and evaluate patient candidacy for nontraditional therapeutic options, this testing remains an inconsistent practice, and the relative rarity of RIGs has further limited widespread reporting of characteristic alterations. Here, we add to the limited molecular literature on RIGs with two unique clinical cases demonstrating remarkably unusual complete responses to treatment following subtotal resections on the basis of Response Assessment in Neuro-Oncology (RANO) criteria, a phenomenon not previously reported in RIGs and only rarely observed in GB independent of bevacizumab-mediated pseudoresponse (14–18).

To date, the largest studies of high-grade RIGs were published simultaneously in 2021. DeSisto et al. comprehensively evaluated 32 tumors by DNA methylation profiling as well as both DNA and RNA sequencing (19). Interestingly, when compared to a non-RIG control cohort, this study, indeed, suggested that many of these tumors could be subgrouped together; methylation profiling of 25 of 31 evaluable tissues clustered epigenetically with pediatric receptor tyrosine kinase I (pedGBM-RTK1) and H3K27M negative midline high-grade gliomas with an odds ratio of 29.2 for falling into this group compared to non-RIG. This was independent of time from radiation and original pediatric diagnosis. Copy number alterations were more frequently seen in RIGs including Chr 1p loss (40%) and/or 1q gain (50%), Chr 13q loss (40%), Chr 14q loss (40%), PDGFRA amplification (36%), CDKN2A loss (29%), BCOR loss (23%), and CDK4 amplification (19%) with findings in the surface tyrosine kinase receptor gene, PDGFRA, and transcriptional co-repressor with influence on apoptosis, BCOR, trending toward and demonstrating significance, respectively. Somatic non-copy number alterations were most often seen in PDGFRA, TP53, MET, and NF1. Deng MY et al. also conducted genetic and expression profiling of 32 RIGs and similarly found that 91% of tumors had methylation profiling consistent with the pedGBM-RTK1 subtype (8). Further, the most commonly observed alterations were again PDGFRA amplification (53%), CDKN2A/B loss (66%), CDK4 amplification (16%), Chr 1p loss (59%) and/or 1q gain (50%), Chr 13q loss (72%), and Chr 14q loss (45%) as well as Chr 6q loss (56%) and MET amplification (28%). Somatic non-copy number alterations reported included PDGFRA, EGFR, TP53, MET, NTRK2, RAF1, BCOR, and ATRX. TERT promotor and IDH1/2 mutations were distinctly absent along with EGFR amplification and polysomy 7/monosomy 10.

Beyond these studies, most RIG genomic analyses have been reported in collectives of small cohorts. López GY et al. profiled 12 RIGs and noted frequent alterations in TP53 (58%), PDGFRA (50%), CDK4 (33%), and CDKN2A/B (33%) in addition to less consistent means of MAP kinase pathway activation (11). Significantly aneuploid genomes were seen in the 10 high-grade gliomas. Whitehouse et al. conducted a systematic review in 2021 to better characterize the molecular features of previously reported RIGs (10). Thirty-one reports including 102 unique high-grade RIGs with molecular data were summarized. Again, the most common alterations observed were PDGFRA and CDK4 amplification, CDKN2A loss, PDGFRA and TP53 mutations, and chromosomal changes including 1p loss, 1q gain, and 13q loss; the authors concluded that these findings suggested that RIGs were molecularly aligned with the pedGBM-RTK1 subgroup.

The two cases presented here demonstrate molecular similarities with those reported in the literature and further support the unique genomic clustering of RIGs. This genomic clustering raises diagnostic considerations given new histologic and molecular classification criteria for gliomas issued in the recent fifth edition (2021) of the WHO Classification of CNS Tumors. Though many RIGs, including both cases here, possess necrosis and/or microvascular proliferation, which in the context of an IDH-wildtype genomic profile is sufficient to establish a diagnosis of GB based on 2021 classification criteria, they actually possess a molecular signature more consistent with diffuse pediatric-type high-grade glioma (PTHGG), H3-wildtype, IDH-wildtype, and WHO grade 4; this is a newly minted tumor class in the fifth edition of the WHO classifications comprising a subset of pediatric high-grade gliomas. Such findings are further exemplified through our cases and literature reports by a general paucity of known GB disease-defining alterations including polysomy 7/monosomy 10, TERT promotor mutation, or EGFR amplification (20). Indeed, if our cases lacked grade 4 histologic features—vascular proliferation and necrosis—as is sometimes observed in incipient grade 4 gliomas, their histologic and molecular features would be insufficient to establish a diagnosis of GB, IDH-wildtype, based on the 2021 criteria. However, the cases would be compatible on both histologic and molecular grounds with PTHGG, with the caveat that such tumors should not be diagnosed outside pediatric and young adult populations. These findings highlight diagnostic ambiguity in RIG classification with the new criteria. Such unifying molecular events suggest that gliomas with molecular profiles of PTHGG in the context of prior radiation exposure may be best given in their own diagnostic category.

Specifically unique to our cases was the marked response to therapy, something not previously reported. Indeed, survival data for RIGs suggest an aggressive disease with poor outcomes (5, 8, 19). While the role of multiple survival and proliferative pathways has clearly been demonstrated to impact the DNA damage response, nothing unique to the genomic profiles of the two cases provides an indication that either tumor would show increased responsiveness to radiation or TMZ (21–23).

In case 1, mutations were observed in BRCA1 (breast cancer type 1; Q1327R) and XPC (Xeroderma pigmentosum complementation group C; I165V). While pathogenic alterations in BRCA1 impede the repair of double-stranded DNA breaks as acquired by ionizing radiation, and XPC dysfunction has been shown to increase TMZ-induced DNA damage, neither point mutation correlated with meaningful functional alteration of the resultant protein nor loss of heterozygosity was observed (24, 25). Case 2 revealed a complex tetraploid chromosomal pattern, reported in approximately 11% of GB as an early event in IDH-wildtype GB (26). While no specific pathogenic alterations in DNA damage repair genes were observed in this case, the duplication of the genome has been associated with genomic instability in GB, potentially portending increased sensitivity to radiation (26). Interestingly, despite a lack of reports suggesting increased susceptibility to DNA-damaging agents, a subset of RIGs harbors decreased transcriptional expression of DNA repair gene products, thus implicating a potential mechanism of improved therapeutic sensitivity in some RIGs (19). Identification of such lesions would require RNA sequencing and/or proteomics which are not yet routine in practice and have technical limitations.

Importantly, response to therapy was independent of MGMT promotor methylation status as there was discordance of this biomarker between the two cases. In fact, given the patient's initial poor performance status and the lack of MGMT promotor methylation in case 2, TMZ was not offered, indicating that the significant response to therapy was not dependent on this known favorable marker nor the administration of chemotherapy. The perioperative appearance of both tumors and the gross pathologic characterization of tissue were reported to be consistent with the typical appearance of a GB without unique or unexpected features.

Our two cases demonstrate the complexity of managing RIGs and predicting outcomes. Despite remarkable complete responses at the sites of the original disease following early postoperative regrowth, one patient developed early parenchymal necrosis in the setting of prior radiation exposure; the other developed overwhelming leptomeningeal spread that, while also markedly responsive to radiation, ultimately exhausted safe treatment options. Sequencing and microarray evaluation of these cases demonstrate similarities with a clustered analysis of other RIGs and add to the molecular literature while contrasting outcomes. This clearly demonstrates that the future evaluation of the tissue will require additional diagnostic testing to parse out disease biology for the prediction of treatment responses. Given the rarity of RIGs, as well as the currently relatively limited case series in the literature compiling corresponding molecular, treatment, and outcome data, a collaborative multi-institutional clinical and pathological review of RIGs would represent an important opportunity to gain meaningful insight into this understudied disease moving forward.

The datasets presented in this article are not readily available because of ethical and privacy restrictions. Requests to access the datasets should be directed to the corresponding author.

Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

PG, JH, and HR: acquisition of material. PG, JH, SH, and HR: direct patient management/data implementation. PG, JH, DD, SH, RJ, and HR: analysis of data/information and interpretation of data/information. All authors contributed to the article and approved the submitted version.