Glioblastoma multiforme (GBM) is one of the most frequent and aggressive primary tumors in the Central Nervous System (CNS), representing more than 60% of all brain tumors in adults (1,2).

In over 90% of cases GBM occurs de novo (primary GBM) without evidence of a less malignant precursor and usually grows more rapidly and has a worse prognosis than secondary GBM, developing from lower grade astrocytoma or oligodendrogliomas.

GBM remains incurable with a poor prognosis both for limited therapeutic alternatives and for high risk of progression or recurrence. The median overall survival (mOS) is about 15 months with 2- and 1-year survival rate respectively of 27 and 41%; only GBM associated with methylated methylguanine-DNA-methyltransferase (MGMT) gene reach a mOS about 24 months; however, less than 10% of patients survives at 5 years (3-5).

In newly diagnosed GBM, the standard of care (SoC) is surgical resection followed by concomitant chemoradiotherapy (CRT) with temozolomide (6) and subsequently adjuvant chemotherapy (5,7-10). However, relapse rate remains poor, with most relapses occurring around 6-9 months after primary treatment (5,11).

In case of recurrence, treatment options are scarce and include re-surgery, re-irradiation, traditional chemotherapeutic drugs, alone or in combination, such as nitrosourea compounds like fotemustine or CCNU (lomustine), temozolomide rechallenge, or antiangiogenic drugs such as bevacizumab, all with limited efficacy (9,12). In addition, immune-checkpoint inhibitors, such as nivolumab or pembrolizumab, have shown poor results. There is no uniform consensus on standard treatment and even the guidelines fail to facilitate therapeutic decision in recurrent setting; therefore, enrollment in clinical trials is recommended. In several countries, Lomustine is usually used in second line after temozolomide failure, with mOS range of 8.6-9.8 months and median progression-free survival (mPFS) of 1.5-2.7 months (13,14).

Fotemustine is the only third generation of nitrosourea available In Italy, and has achieved encouraging results, though with low-certainty evidence (15).

Regorafenib is a small-molecule multi-kinase inhibitor already approved in second-line therapy for advanced hepatocellular carcinoma (HCC) (16,17) and in third-line treatment both for advanced colorectal cancer (CRC) (18-20) and gastrointestinal stromal tumors (GISTs) (20). Since 2019, regorafenib has been approved by Agenzia Italiana del Farmaco (AIFA)in recurrent glioblastoma as second line therapy, following the results of a phase 2 trial by Lombardi et al (21) (REGOMA). REGOMA is a randomised, comparative, multicenter phase 2 trial designed to evaluate the efficacy of regorafenib in this setting, compared to lomustine. In this trial, 119 patients were randomized to receive either regorafenib or lomustine. At the median follow-up of 15.4 months, OS, primary endpoint of the study, was greatly improved in the regorafenib group vs. SoC cohort (7.4 vs. 5.6, respectively). The disease control rate (DCR) was 44% in the regorafenib arm and 20% in the lomustine control arm. Because of this, regorafenib has been approved in Italy and, while sometimes used in other countries on a single patient basis as an off label treatment, it is conspicuously absent from European Association of Neuro-Oncology (EANO) (9).

Although the REGOMA trial has brought new hope in patients with GBM, it presents some critical issues related to both the absence of a phase 3 study, necessary to confirm the results obtained, and to the mOS of patients treated with lomustine found to be inferior to what is known in literature in the same population (8,6-9,3 months). This underlines the importance of stratifying GBM based on key molecular alterations and/or specific prognostic factors.

Moreover, elderly GBMs often present with a dismal prognosis, with survival around 6 months, and a limited response to treatments. Several molecular features are being investigated and different prognostic assessment including age, performance status (PS), disease burden, comorbidities and other factors have been proposed to better predict outcomes and prognosis (22).

We report our findings based on a retrospective analysis of a cohort of 56 patients >60 years treated in two Units in South Italy, Azienda Ospedaliera Universitaria Luigi Vanvitelli (Naples) and Ospedale Civile San Giovanni di Dio (Frattamaggiore, Naples). It is one of the few studies following REGOMA trial to study regorafenib in a real-life environment and one of the few to do so with a homogeneous ITT population: all patients were diagnosed with recurrent glioblastoma and treatment was initiated in a second line setting. We chose to focus on elderly patients, defined as patients >60 years, since median age of diagnosis of glioblastoma is 64 years, with many diagnoses being made in 70 years or older patients (23), and seeing how for these patients is harder not only to participate in clinical trials but also to receive SoC therapy, due to worsening clinical conditions, increasing comorbidities, and reduced social network.

To our knowledge, this was one of the few analyses following REGOMA to study regorafenibas treatment at first recurrence in glioblastoma patients in a real-life setting and possibly the only one to focus on an older population. Median age of patients treated with regorafenib in REGOMA was 54.8 years (46.8-61.3), whereas in our project median age reached 68 years. (60-79). Although appearing small, our sample size is comparable to other publications in the same setting due to the relative incidence and prognosis of glioblastoma. Thus, we believe our results to be representative of a real life population.

Our results were mostly comparable to those of REGOMA, with a DCR of 48% (vs. 44%), although our patients only showed SD or PR with no CR.CR were instead reported in REGOMA trial in 2% of patients. It must be noted that neither a subsequent prospective study by Lombardi et al (28) published in 2021 nor a 2019 bicentric retrospective analysis by Tzaridis et al (29) showed evidence of CR; indeed, a 2019 retrospective analysis by Kebir et al (30) on six high grade astrocytomas showed a DCR of 0%.

mOS was slightly lower in our findings, 6.8 months vs. 7.4 months, easily due to an older and more comorbid population than that of the REGOMA trial and, of course, to the characteristics of a real-life retrospective setting, that allows for less stringent inclusion criteria. Anyhow, results were largely superimposable, confirming the benefit of regorafenib on OS on the ITT.

At the same time, mPFS was then doubled in our study, 4.0 months vs. 2.0 months. Kebir et al (30) also showed a higher mPFS (3.5 months) compared to REGOMA. As in that case, our discordant results may be due to the nature of the study: being a bicentric retrospective analysis, MRI time schedule was easily dependent on investigators choices and variable on a case-to-case basis, thus determining a formally higher result.

As reported above, mOS was found statistically superior in patients with methylated MGMT promoter (mOS 7.7 months vs. 5.6 months), with no difference in mPFS (mPFS 3.1 months vs. 4.1 months). MGMT promoter methylation has been since long identified as a predictive factor of increased survival from alkylating agents (9,31). Unfortunately, no benefit has yet been identified in patients treated with regorafenib (9,28). While our data need to be evaluated in other studies, especially prospective trials with larger populations, one must consider the possibility of the two subgroups being unbalanced for confounder factors, determining such a result. Indeed, while the proportion of patient in the two subgroups that completed at least 6 cycles of temozolomide is quite similar (50% in the unmethylated group vs. 62.5% in the methylated group), there was a higher percentage of patient that underwent radical surgery at diagnosis in the latter group (77% vs. 41%).

Age did not appear to be a prognostic factor, with no difference in mPFS in the two subgroup analysis performed. This is consistent with previous results from Lombardi et al (28) real-life trial and may help considering older patients with good performance status for treatment with regorafenib, irrespective of age. Furthermore, if we compare our results to those of Lombardi regarding ORR, they are largely superimposable. DCR reached 48% in our study compared to 46% in the latter, whereas ORR resulted 9 to 7.4% (28).

Our study may help in referring older patients with good PS to treatment with regorafenib, irrespective of age, especially since our population, although significantly older (median age of 68 years vs.55 years of Lombardi trial or 54.8 years of REGOMA trial), presented similar survival and control rates.

Survival was also analyzed according to IDH status, but our results cannot be generalized due to the strong unbalance between the two groups, with only 3 patients reporting a mutation in IDH (data similar to that of REGOMA trial with only 2 patients in the regorafenib arm and 0 in the lomustine arm). Survival was prolonged in the IDH mutant population, with a mPFS of 6.1 months and mOS of 11.5 months vs. respectively 3.1 and 6.1 months in the IDH wild type group (mOS p 0.041; mPFS p 0.009). Mutations in IDH have been generally reported in so-called secondary glioblastomas (32) and are an important prognostic factors associated with longer OS, although, as specified before, the new WHO system does not allow for IDH mutant glioblastoma but classifies them into astrocytomas (grade 2 to 4) or oligodendrogliomas (grade 2 or 3) (24).

Regorafenib was overall well tolerated in our population, with mainly grade 1 and 2 AEs. Only 25% of the ITT (14 patients) did not report any toxicities. 56% of the patients in Regorafenib arm developed at least one grade 3-4 AE in the REGOMA trial and 90% ported at least one all-grade drug-related toxicity in a subsequent study by Lombardi et al (28): with this in mind, one must consider the possibility of low accuracy in toxicity reports for this study. Even adjusting for this eventuality, safety profile was comparable to other trials involving regorafenib, both in glioblastoma and in other setting, thus advocating for its use in older but medically fit patients as a real therapeutic alternative.

The only grade 3 reported AEs were HFSR, piastrinopenia, fatigue, voice changes, proteinuria, and aminotransferase elevation. HFSR accounted for 7% of overall AEs (11% of the ITT, 6 patients) with only 2 patients showing grade 3 AEs. Incidence is thus lower than what was reported in REGOMA trial (grade 1-2, 22%, grade 3, 10%) or other trials in GBM (29), which was even lower than data from CORRECT (33) and RESORCE (34) trials (overall HFS rate 47% in the CORRECT trial and 53% in the RESORCE trial). All grade thrombocytopenia was reported by 13% of the cohort, with 5 grade 2 AEs and 2 grade 3 AEs, a result slightly more favorable than that of REGOMA trial, with 20% of grade 1-2 and 2% (1 patient) grade 3 thrombocytopenia and similar to the subsequent analysis by Lombardi et al (28).

AST elevation and hyperbilirubinemia were also among the most common AEs, with 11% manifesting AST elevation and 7% increase in bilirubinemia, similar to what was expected based on prior studies (21,34).

As already exposed, one of our main limitations is the retrospective nature of our study, determining a higher risk of incomplete data, information and recall bias, and the small population. However, our results are fairly superimposable to those of the available literature. This helps generating a framework in which regorafenib is a valid approach even in elderly patients due to both survival rates and toxicity profile being similar to those of a younger population. In any case, more phase 3 trials are needed to unravel the question of whether regorafenib is definitely superior to lomustine and define the best strategy at recurrence. An observational prospective study (REGOMA-Oss, [NCT04810182]) is already ongoing and will analyze the role of regorafenib in recurrent GBM in real world patients.

Unfortunately, predictive biomarkers of response to regorafenib are not yet available. In a recent study by Santangelo et al (35) based on patients from REGOMA trial, a group of 5 RNA biomarkers (HIF1a and CDKN1A mRNA, miR-93-5p, miR-3607-3p and miR301a-3p) identified a favorable subgroup of patients. These findings, given the relatively small population and the study design, must be validated in larger and in prospective trials (35).

Nevertheless, new studies are already exploring other strategies for regorafenib. GBM AGILE trial an international, seamless Phase II/III response adaptive randomization platform trial designed to evaluate multiple therapies, with regorafenib being used both at first diagnosis after concomitant CRT with temozolomide or at first recurrence. A phase II trial is evaluating regorafenib use in bevacizumab refractory high-grade gliomas (not only GBM but also gliosarcoma, small cell glioblastoma etc. can be included) [NCT04051606], while another phase II basket trial is investigating the association of regorafenib and nivolumab in several tumor types [NCT04704154].