Abstract
Background
More than 40% of patients with intracranial ependymoma need a salvage treatment within 5 years after diagnosis, and no standard treatment is available as yet. We report the outcome after first relapse of 64 patients treated within the 2nd AIEOP protocol.
Methods
We considered relapse sites and treatments, that is, various combinations of complete/incomplete surgery, if followed by standard or hypofractionated radiotherapy (RT) ± chemotherapy (CT). Molecular analyses were available for 38/64 samples obtained at first diagnosis. Of the 64 cases, 55 were suitable for subsequent analyses.
Results
The median follow-up was 147 months after diagnosis, 84 months after first relapse, 5-year EFS/OS were 26.2%/30.8% (median EFS/OS 13/32 months) after relapse. For patients with a local relapse (LR), the 5-year cumulative incidence of second LRs was 51.6%, with a 5-year event-specific probability of being LR-free of 40.0%. Tumor site/grade, need for shunting, age above/below 3 years, molecular subgroup at diagnosis, had no influence on outcomes. Due to variation in the RT dose/fractionation used and the subgroup sizes, it was not possible to assess the impact of the different RT modalities. Multivariable analyses identified completion of surgery, the absence of symptoms at relapse, and female sex as prognostically favorable. Tumors with a 1q gain carried a higher cumulative incidence of dissemination after first relapse.
Conclusions
Survival after recurrence was significantly influenced by symptoms and completeness of surgery. Only a homogeneous protocol with well-posed, randomized questions could clarify the numerous issues, orient salvage treatment, and ameliorate prognosis for this group of patients.
Keywords: 1q gain, complete surgery, dissemination, ependymoma relapse, re-irradiation
Key Points.
Patients with ependymoma can have late relapses and no standard treatment exists.
Survival is significantly influenced by symptoms and completeness of surgery.
Only randomized questions could clarify the numerous issues of retreatment.
Importance of the Study.
Our manuscript describes the outcome of 64/160 patients with intracranial ependymoma relapsing after homogeneous treatment delivered according to the 2nd AIEOP protocol. This series had a follow-up over 10 years and relapsing patients were followed for a median of 84 months after first relapse; 17/64 survived after relapse. Complete surgery of relapses was the most important prognostic factor while re-radiotherapy fractionation/total dose delivered, tumor site/grade at diagnosis, need for shunting, age above/below 3 years, and molecular subgroup at diagnosis had no influence on outcome. Multivariable analyses identified completion of surgery, the absence of symptoms at relapse, and female sex as prognostically favorable. We confirm the need for a retreatment protocol in relapsing patients that could answer the many questions still unanswered thus ameliorating patients’ prognosis.
It is only in the last 10 years that the pediatric neuro-oncology community has seen the need for adjuvant treatment after surgery (as complete as possible) for pediatric intracranial ependymoma in all patients.1 In the vast majority of cases, this involves focal radiotherapy (RT).2 The use of chemotherapy (CT) is still debated and object of ongoing or recently completed trials, both in the post-radiation setting and in the event of residual disease after surgery, or at diagnosis for very young children.1,3–7
The 5-year survival probability for children and adolescents with ependymoma is around 50%-70%, with marked differences between cases with and without residual disease after surgery. Complete resection is the most important and widely recognized prognostic factor in this disease.8,9 There is no agreed standard treatment as yet for intracranial ependymomas that relapse, which may happen even after many years.5,10 The literature describes retrospective series often consisting of very few patients who frequently received different treatments at first diagnosis, outside any clinical trials.11,12 The outcomes are therefore not always consistent in supporting the benefit of any given strategy for treating relapses.13–15
This study reports on the outcome after first relapse of 64 patients with intracranial ependymoma who were treated according to the 2nd AIEOP protocol,8 with a median follow-up after diagnosis of 140 months.
Patients and Methods
Patients’ follow-up after first-line treatment was updated (as of March 31, 2021) according to a retrospective review protocol authorized by the ethical board of the national coordinating institution (protocol INT 24-20). Consents had already been obtained from patients and parents at the time of first-line treatment and during active follow-up.
First-line treatment and outcomes were already reported elsewhere.5,8 Briefly, patients with no residual disease after surgery received various treatments based on their disease’s WHO histological grade16; grade II patients received focal RT up to 59.4 Gy; grade III patients also had 4 courses of vincristine, etoposide, and cyclophosphamide (VEC) CT after RT. Patients with incomplete resection received up to 4 courses of VEC after surgery, then underwent second-look surgery whenever feasible, and were given RT up to 59.4 Gy + 8 Gy boost delivered over 2 days to any residual tumor still measurable in the three dimensions.
Radiological and clinical follow-up guidelines were homogeneously adopted, and patients were followed up for the first 5 years after diagnosis.8 Thereafter, each treating center checked their patients’ status according to local habits. In cases that relapsed, however, a whole central nervous system (CNS) radiological assessment with MRI was always performed.
For the purposes of this report, we recorded the time and site of first relapse: local relapse (LR), disseminated (DM) or synchronous (both local and disseminated, LR + DM), and whether any DM disease was represented by a single or multiple nodules. Treatment delivered after first relapse was described as complete or incomplete surgery for sites of both single and disseminated nodules, and any adjuvant RT after surgery, using hypo- or standard fractionation (ie, ≥3 Gy or 1.5-1.8 Gy) at various doses (these main strategies could be variously combined), plus any CT.
All tumor samples were centrally reviewed by the referral pathologist at the time of the patients’ diagnosis. In 95/160 protocol cases, the tumor material underwent methylation profiling, DNA copy-number analysis, and gene fusion detection, performed according to previously reported methods.5,17
All patients’ samples were also histologically reviewed at relapse if surgery was performed. The above-described molecular analyses were not repeated but were available for 38/64 original samples, which had been scored as ≥0.8 according to the methylation profile17 in 29 cases. Values of a classifier score <0.8 (9 samples) were not considered in the subsequent analyses
Statistical Methods
From the overall cohort of 64 patients, 55 were selected for further analysis; 4 were excluded because the parents rejected any treatment after first relapse, and 5 with LR + DM who were given different, not strictly comparable treatments. The characteristics of the patients and tumors were recorded using descriptive statistics in both the overall and the selected cohorts; and the 5 patients with LR + DM were also described separately.
The primary goals of the study were to assess the event-free survival (EFS) and overall survival (OS) after first relapse. Survival and incidence times were calculated from the date of first relapse. EFS and OS curves were estimated using the Kaplan-Meier method, and the curves were compared using the log-rank test. EFS was defined as the time to any further LR/progression, DM, or death, whichever came first. OS was defined as the time to death due to any cause. Time was censored at the latest follow-up for patients still alive.
The probability of remaining LR-free and the incidence of second LRs in locally relapsing patients, and the incidence of DM as a second event in patients with a 1q gain were also investigated. The crude cumulative incidence (CCI) of second LRs and DM as second events were estimated in a competing risk setting, using cumulative incidence estimates, and then compared using the Gray test. Either LR or DM, and death were considered as competing events. LR-free probability was estimated using the Kaplan-Meier method, censoring all events except LR. In all the relevant analyses, concomitant LR and DM were considered as DM, and progression was calculated according to the first episode of progression. The median follow-ups were estimated using the reverse Kaplan-Meier method on the basis of OS data.
Multivariable Cox models, including the presence of symptoms at relapse, completion of surgery, and sex, were also developed to assess these covariates’ association with EFS and OS. The interaction term between symptoms at relapse and sex was included.
The statistical analyses were performed with the SAS (version 9.4) and R software (version 3.6.3).
Results
All the patients included in this analysis had been treated according to the 2nd AIEOP protocol for intracranial ependymoma, which had accrued 160 patients by the time it ended in 2014.8
At the time of this report, 64 patients treated at 15 different Italian centers belonging to the AIEOP (Italian Association of Pediatric Hematology and Oncology) had suffered a tumor recurrence and they formed our “overall” cohort. After excluding 9 patients (4 for refusal of any treatment, 5 for LR + DM with distinctive treatments, see Methods), our “core” cohort consisted of 55 patients whose treatment for relapses could be properly examined.
Table 1 shows the characteristics of the patients and tumors at diagnosis and at first relapse.
Table 1.
Patients and Tumor Characteristics of the Main and of the Overall Cohorts at Diagnosis and First Relapse
| Overall Cohort | Main Cohort | |
|---|---|---|
| Sex | ||
| Female | 17 (26.6%) | 11 (20.0%) |
| Male | 47 (73.4%) | 44 (80.0%) |
| Age at diagnosis | ||
| Median (interquartile range) | 3.90 (1.95; 8.65) | 4.00 (2.20; 8.70) |
| <3 years old | 40 (62.5%) | 36 (65.5%) |
| ≥3 years old | 24 (37.5%) | 19 (34.5%) |
| Tumor site | ||
| Posterior fossa | 50 (78.1%) | 43 (78.2%) |
| Supratentorial | 14 (21.9%) | 12 (21.8%) |
| Grade | ||
| II | 24 (37.5%) | 19 (34.5%) |
| III | 40 (62.5%) | 36 (65.5%) |
| 1q gaina | ||
| Yes | 10 (26.3%) | 7 (21.9%) |
| No | 28 (73.7%) | 25 (78.1%) |
| Methylation profile scoreb | ||
| ≥80% | 29 (76.3%) | 26 (83.9%) |
| <80% | 9 (23.7%) | 5 (16.1%) |
| Posterior fossa molecular subgroupingb | ||
| A | 26 (92.9%) | 21 (91.3%) |
| B | 1 (3.6%) | 1 (4.3%) |
| <80% profile score | 1 (3.6%) | 1 (4.3%) |
| Supratentorial molecular subgroupingb | ||
| RELA | 5 (55.6%) | 4 (50.0%) |
| <80% profile score | 4 (44.4%) | 4 (50.0%) |
| Shunt | ||
| Yes | 31 (48.4%) | 24 (43.6%) |
| No | 33 (51.6%) | 31 (56.4%) |
| No evidence of disease pre-radiotherapy at first relapse | ||
| Yes | 40 (62.5%) | 36 (65.5%) |
| No | 24 (37.5%) | 19 (34.5%) |
| First relapse | ||
| Local relapse | 41 (64.1%) | 38 (69.1%) |
| Dissemination | 18 (28.1%) | 17 (30.9%) |
| Synchronous relapse | 5 (7.8%) | 0 (0.0%) |
| Metastatic nodule at first relapsec | ||
| Single | 12 (52.2%) | 9 (52.9%) |
| Multiple | 11 (47.8%) | 8 (47.1%) |
| Disease-free interval >5 years | ||
| Yes | 10 (15.6%) | 8 (14.5%) |
| No | 54 (84.4%) | 47 (85.5%) |
| Symptoms at second relapse | ||
| Yes | 19 (29.7%) | 15 (27.3%) |
| No | 45 (70.3%) | 40 (72.7%) |
| Surgery completion at second relapse | ||
| Yes | 32 (50.0%) | 32 (58.2%) |
| No | 17 (26.6%) | 15 (27.3%) |
| Not done | 15 (23.4%) | 8 (14.5%) |
| Site of radiotherapyd | ||
| Local | 31 (59.6%) | 31 (66.0%) |
| CSI | 2 (3.8%) | 1 (2.1%) |
| Other | 19 (36.5%) | 15 (31.9%) |
| Radiotherapy scheduled | ||
| Hypofractionated | 26 (50.0%) | 25 (53.2%) |
| Standard fractionation | 25 (48.1%) | 22 (46.8%) |
| Other | 1 (1.9%) | 0 (0.0%) |
| Chemotherapy done | ||
| Yes | 18 (28.1%) | 16 (29.1%) |
| No | 46 (71.9%) | 39 (70.9%) |
Abbreviations: CSI, craniospinal irradiation; DM, disseminated; LR, local relapse.
a Including patients with methylation profile score ≥0.80 and those with gene fusion analysis.
b On available original samples.
c On DM and LR + DM or DM only patients in overall and main cohort, respectively.
d On patients underwent to radiotherapy.
Figure 1 shows the sites of relapse and the patients’ outcomes; Figure 2 shows the Consort flows of all patients splitting patients according to the primitive site of disease (supratentorial [ST]; posterior fossa [PF]) and of relapses.
Fig. 1.
Consort flow chart for local and distant relapses in the core cohort.
Fig. 2.
Consort flows splitting patients according to the primitive site of disease (ST, supratentorial; PF, posterior fossa) and local and distant relapse.
Overall Cohort
The median follow-up was 147 months (range 10-223 months) after diagnosis, and 84 months after first relapse (range 0-191 months). Supplementary Figure 1 shows the OS and EFS experience of the overall series. The median time to relapse was 21 months (range 4-123 months): 24 months (range 2-126 months) for LR; 18 months (range 5-51 months) for DM; and 29 months (range 7-113 months) for LR + DM.
Six patients had a relapse (4 LR and 2 LR + DM at first relapse) more than 5 years after their diagnosis, and 4 of them (LR at first relapse) were alive and disease-free.
Forty-one (64.1%) patients relapsed locally, 18 (28.1%) only at distant sites, and 5 (7.8%) had synchronous relapses. Of the patients with DM, 9 had a single nodule (3 among the LR + DM patients). Of the patients with LR, 12/41 had symptoms prompting MRI, while 6 of the 18 patients with DM had symptoms, and so did 1 of the 5 with LR + DM.
Of the 17 surviving patients, only 4 had residual disease at first RT. Of these 4, 3 experienced LR that in 2 cases was completely resected, one was not operated, and the fourth had LR + DM that were not completely resected.
Only one of the 51/60 patients submitted to re-irradiation developed radionecrosis with worsening of neurological conditions corresponding to consistent imaging for a total of 8 months. The median time from the first to the second RT was 17 months (range 2-124 months). Five locally re-irradiated patients (all treated with a hypofractionated schedule at a dose in the range of 12-16 Gy) had also already received a boost of 8 Gy to residual disease in two daily doses of 4 Gy after 59.4 Gy given at 1.8 Gy/day according to the first-line protocol.8 Another patient’s re-irradiation involved craniospinal irradiation (CSI) at 36 Gy after the same primary schedule. The use of dexamethasone was based on each center’s normal practices, without any bevacizumab support, and for as long as it was deemed necessary.
In all, 18/60 patients treated were given CT, mainly involving oral etoposide, the VEC schedule, temozolomide and cisplatin. This never produced a response when given in cases of evaluable disease. Three/five patients that had received CT after complete resection of LR and RT were alive at last follow-up vs 6/16 not receiving CT, without a significant difference in progression-free survival and OS (data not shown).
Core Cohort
The median follow-up was 147 months (range 10-223 months) after diagnosis, and 96 months after first relapse (range 1-191 months). Forty-seven/fifty-five patients underwent surgery: 37 for LR, and 6 for single foci of DM, achieving complete resections in 26 and 6 cases, respectively. Seven/eight patients with ST tumors and 19/28 with PF ones had complete resections after LR (P = .0156). There was instead no difference in surgical results comparing the patients having complete resection before RT at first diagnosis (19/25 complete resection after LR) and those having residual tumor before first RT (7/12 complete resection after LR) (P = .1640).
Another 4 patients with DM involving multiple sites were also treated surgically.
After surgery, patients with completely resected LRs were given adjuvant treatment in 22/26 cases (the other 4 received no further treatment): 21 patients had RT (followed by CT in 5 cases) using a hypofractionated schedule and doses ranging from 4 Gy/day for 2 days to 4 Gy/day for 4 days; the remaining patient was given CT alone after a complete resection.
Patients with LR and incomplete surgery (11) received RT in 9 cases, delivered in 6 using hypofractionation and doses of between 8 and 35 Gy, and in 3 patients with standard fractionation (range 19.8-35 Gy); the other 2 received CT alone.
The last of the 38 patients with LR was only treated with hypofractionated RT at 8 Gy.
As already mentioned, 6/9 patients with a single nodule of DM underwent total resection, followed in 5 patients by local RT with standard fractionation at doses in the range of 45-59.4 Gy (then CT in 3 of them), and in 1 patient by CSI at a total dose of 36 Gy. The other 3 patients were not operated but were all given hypofractionated RT (range 19.8-42 Gy), and 1 also had CT. Eight/twelve patients who had DM with multiple nodules underwent resection at one or more sites in 4 cases, then received standard fractionated RT (range 36-45 Gy) to all sites (followed by CT in 2 cases). Of the other 4 patients, 1 was treated with CT alone, and 3 had RT (hypofractionated in 1, standard fractionated in 2), followed by CT in 1 case.
The median time to second relapse was 13 months (range 0-68 months).
Of the 37 patients with LR who underwent surgery, 13 were alive a median of 84 months after first relapse, 11 of them disease-free, 1 with a further LR, 1 with DM. Among the other 24 patients, 19 had again experienced LR/progression (one of them salvaged by further therapy), 1 had DM, and the other 4 had LR + DM. The median time to death after relapse was 22 months. The patient re-irradiated without any further surgery relapsed locally and died 35 months later.
Three/nine patients with a single nodule of DM were alive and disease-free at 77, 98, and 156 months after relapse. The other 6 had LR in two cases, DM in 3 and LR + DM in 1; these 6 patients died of their disease a median of 22 months after relapse.
The other 8 patients with DM at first relapse all had further DM within a median of 5 months and died of their disease a median of 9 months after their first relapse.
Synchronous First Relapse Subgroup
Of the 5 LR + DM patients, 2 underwent LR excision, then received CSI at 36 Gy plus a boost to bulky sites; the other 3 had RT to LR and DM with a combination of hypofractionation (for LRs) and standard fractionation. All 5 patients relapsed again (LR in 2 cases, LR + DM in 1, DM in 2) a median of 6 months after first relapse, and all but 1 died of tumor progression a median of 7 months after first relapse.
Clinical and Biological Parameters
Five-year EFS and OS (95% CI) were 26.2% (16.7%-41.4%) and 30.8% (20.3%-46.9%); and the median EFS and OS were 13 and 32 months.
For patients with LR, the 5-year cumulative incidence of second LR was 51.6% (95% CI: 36.8%-72.4%) with a 5-year specific event probability of being LR-free of 40.0% (95% CI: 25.5-62.5).
Five-year estimates obtained with the Kaplan-Meier curves are shown in Supplementary Table 1.
The presence of symptoms at relapse (Figure 3A and B) was significantly associated with both a worse EFS and OS (5-year estimates: 6.7% [95% CI: 0%-44.3%] vs 33.8% [95% CI: 21.6%-52.8%], P = .003; and 7.7% [95% CI: 1.2%-5%] vs 39.6% [95% CI: 26.5-59.1], P = .009, in patients with and without symptoms, respectively).
Fig. 3.
Kaplan-Meier curves for event-free survival and overall survival according to the presence of symptoms at relapse (A and B), extent of resection at relapse (C and D), and sex (E and F) in the core cohort.
Patients amenable to complete resection (Figure 3C and D) with LR or single-nodule DM had a better EFS and OS, with 5-year estimates (95% CI) of 36.3% (22.7%-58.2%, P = .021) and 41.2% (26.7%-63.4%, P = .010), respectively, while there were no survivors among the patients whose surgery was incomplete.
Female sex (Figure 3E and F) was associated with a better outcome, with a 5-year EFS and OS (95% CI) of 46.8% (95% CI: 22.4%-97.7%) vs 20.2% (95% CI: 11.18%-36.5%) (P = .021) and of 71.6% (48.8%-100%) vs 22% (12.3%-39.2%, P = .070) for males, respectively. These differences were attenuated in the overall cohort, where the 5-year EFS and OS (95% CI) were, respectively, 29.8% (13.0%-68.0%) vs 18.9% (10.4%-34.3%), and 52.3% (33.0%-82.8%) vs 20.6% (11.5%-36.9%), with P-values of .124 and .308 in the log-rank test.
When survival was compared for LR and DM (Figure 4A and B), it was better for LR (though the difference was not statistically significant), when patients with LR + DM and those not treated at relapse were included. In the core cohort, 5-year EFS and OS (95% CI) were, respectively, 30.0% (18.1%-49.8%) vs 17.6% (6.3%-49.3%, P = .158), and 34.1% (21.2%-55.0%) vs 23.5% (10.0%-55.4%, P = .128) in LR and DM, respectively. In the overall cohort, 5-year EFS and OS (95% CI) were 27.8% (16.7%-46.5%) vs 13.0% (4.5%-37.5%; P = .106) and 31.6% (19.5%-51.3%) vs 20.9% (9.2%-47.2%, P = .090) in LR and DM, respectively.
Fig. 4.
Kaplan-Meier curves for event-free survival and overall survival according to relapse event (A and B), relapsing more than 5 years from diagnosis (C and D), and crude cumulative incidence curves of local relapse and distant metastasis according to chromosome 1q gain at diagnosis (E and F) in the core cohort.
Due to the very small number of patients relapsing after more than 5 years after diagnosis, and their shorter follow-up (none reaching 5 years), comparing them with patients who relapsed earlier (Figure 4C and D) would be of limited value. We nonetheless noted a better prognosis (here again, not statistically significant) in patients who relapsed later. In particular, the log-rank P-values of the comparisons for EFS in the core cohort and for OS in the overall cohort were .114 and .117, respectively.
Hypofractionation vs standard fractionation of RT in patients with a completely resected LR or single disseminated nodule, hypofractionation vs standard fractionation and total RT doses delivered, tumor site and grade at diagnosis, need for shunting, 1q gain, and age above or below 3 years at diagnosis had no effect on the outcomes investigated after relapse (Supplementary Figures 2 and 3). That said, patients whose tumor had 1q gain at diagnosis (Figure 4E and F) tended to have a higher cumulative incidence of DM after first relapse—as high as 50% at 5 years (95% CI: 24.8%-100%) vs 21.4% (95% CI: 10.3%-44.4%) among patients with no 1q gain in their tumor (P = .139).
CSI was very rarely used (in only 3 cases: 1 with single-site DM, 2 with LR + DM), so it was impossible to ascertain any advantage of this therapy. The 26 patients achieving a complete resection of LR, and the 6 with single-node DM at relapse had much the same EFS and OS, with a median of 20 and 21 months, respectively (data not shown).
In the patients whose tumors had a methylation profile scoring ≥0.80 (n = 29), and those who had undergone gene fusion analysis (n = 3), we observed the following molecular subgrouping: 26 patients had PFA, 1 had PFB, and 5 patients with ST tumors had the RELA type, with a clearly higher prevalence of PFA in the group as a whole. 1q gain was present in 10 samples (6 PF tumors) and chromosome 6q loss in 5 (4 PF tumors). Seven/10 patients with 1q gain (5 of the 6 with PF tumors) and all those with 6q gain, 5 after LR and 2 LR+DM, died.
Besides confirming the relevance of complete surgery, even as an adjusted factor, multivariable analyses (Table 2) showed a significant interaction between sex and the presence of symptoms at relapse. Symptoms always had a negative influence on prognosis in both males and females, but in very different ways. In the absence of symptoms, female sex was a very favorable factor (HRs: 0.13 and 0.16, 95% CIs: 0.03-0.54 and 0.04-0.65, Ps = .010 and .002 in the EFS and OS models, respectively). In the presence of symptoms, the level of risk (of both events and death) increased dramatically in females (HRs: 17.2 and 56.2, 95% CIs: 2.46-119.7 and 6.52-483.5 vs 1.62 and 1.38, 0.79-3.31 and 0.64-2.99, Ps = .008 and .001 in the EFS and OS models, respectively), potentially overcoming the sex gap (HRs: 1.41 and 6.50, 95% CIs: 0.22-8.94 and 0.79-53.6 in the EFS and OS models, respectively).
Table 2.
Results of the Multivariable Cox Models for Event-Free Survival and Overall Survival in Main and Overall Cohorts
| Main Cohort | Overall Cohort | |||||||
|---|---|---|---|---|---|---|---|---|
| Covariates | EFS HR (95% CI) | P | OS HR (95% CI) | P | EFS HR (95% CI) | P | OS HR (95% CI) | P |
| Surgery | .023 | .023 | .001 | <.001 | ||||
| Complete vs Incomplete | 0.60 (0.29-1.25) | 0.50 (0.23-1.11) | 0.54 (0.27-1.08) | 0.52 (0.24-1.12) | ||||
| Not done vs Incomplete | 2.83 (0.86-9.37) | 2.21 (0.64-7.61) | 3.30 (1.28-8.50) | 3.47 (1.32-9.13) | ||||
| Sex | .004 | |||||||
| Female vs Male, with symptoms | 1.41 (0.22-8.94) | 6.50 (0.79-53.6) | 1.44 (0.37-5.57) | 2.91 (0.65-13.01) | ||||
| Female vs Male, without symptoms | 0.13 (0.03-0.54) | 0.16 (0.04-0.65) | 0.21 (0.07-0.57) | 0.21 (0.07-0.61) | ||||
| Symptoms | .008 | .001 | .001 | <.001 | ||||
| Yes vs No, in male | 1.62 (0.79-3.31) | 1.38 (0.64-2.99) | 1.69 (0.86-3.33) | 1.52 (0.74-3.12) | ||||
| Yes vs No, in female | 17.2 (2.46-119.7) | 56.2 (6.52-483.5) | 11.8 (3.01-46.4) | 20.9 (4.53-96.1) |
Abbreviations: CI, confidence interval; EFS, event-free survival; HR, hazard ratio; OS, overall survival; P, Wald test P-value.
The results for the overall cohort largely confirm those obtained for the core cohort (Tables 1 and 2, Supplementary Table 1, Supplementary Figures 4–7). Even the subgroup analysis on the single sites (PF and ST) showed consistent results in terms of associations (data not shown), despite some differences in their magnitudes was observed (ie, need of different prognostic models).
Discussion
Salvage treatment will be needed for 30%-40% of patients with recurrent ependymoma receiving first-line treatment.1,4,5
Given the current lack of a standardized treatment for relapsing intracranial ependymoma, adding an update to the previously published experiences may contribute to the shaping of a common protocol with shared aims.13–15,18
The results reported here concern 64 relapsing patients from a series previously enrolled in a novel first-line protocol, the results of which have been reported since 2016.5,8,19 Indeed, the retrospective nature of the study represents a limitation, given that the treatment of relapsing tumor was not predefined.
The chances of a positive outcome after a tumor has relapsed correlates with the imaging findings before any symptoms become apparent—as we had already demonstrated with a previous report on 50 relapsing patients with a median follow-up of 67 months.19 This was confirmed after a follow-up of more than 10 years, with a significantly better EFS and OS for patients with recurrent disease identified on scheduled MRI, and also by the results of multivariable analyses. This should be borne in mind when drawing up guidelines for the first-line follow-up of patients, and when considering long-term follow-up after any event.10
All patients still alive were in active follow-up at the time of this report because late relapses are not exceptional in ependymoma patients5,10 and salvage is possible.13–15,18
While keeping in mind the small number of occurrences in our series (N = 4; 3 PF, 3 DM, 1 LR + DM, 1q gain in 1), we observed that the presence of symptoms appears to be associated with an especially poor outcome in females, whereas they have a statistically better prognosis than males if asymptomatic at relapse.11 As reported for first-line treatment,8 here again, multivariable analysis showed that females had a significantly better prognosis after treatment for relapse. This was reported by Tsang as well in 2018,20 but we, and the literature as well, do not know the reason for this phenomenon. Nor do we have any explanation as yet for the very poor prognosis associated with the combination of symptoms and female sex with no survivors at 5 years among symptomatic girls whose disease relapsed).
Even after a follow-up of more than 10 years, LR remains the most common mode of tumor recurrence, as was the case in 41/64 (64%) patients in our series. The first option to consider at retreatment, therefore, is surgery—as confirmed in our own and other previously reported series.18,20–22 When complete, this gives patients a prognostic advantage, even in the case of single metastatic locations—as we saw in the 3/6 patients with completely resected single distant lesion who survived after salvage treatment. Complete resection of LR was more frequent in ST tumors and, even if not statistically significant, this gave a prognostic advantage to this patient subgroup (Supplementary Figure 2, panels G and H).
The choice of irradiation modality, fractionation, and total doses was not centrally guided and depended on the experience of the hypofractionated boosts already applied in first-line treatment,8 on upcoming literature remarks,20–25 and on international meetings results.
A hypofractionated RT boost was given as adjuvant or sole treatment at relapse in 22/41 patients with LR in our series. The decision to administer 8 Gy in two fractions of 4 Gy stemmed from our first-line protocol, which adopted the same boost for patients with persistent disease after surgery.8 The total doses, as well as the daily doses of RT with hypo- and standard fractionation, were not homogeneous in our patients, but when we compared those given 8 Gy with hypofractionation with those receiving higher doses, we found no statistically significant differences in their outcome.
The other 9 patients with LR were treated with re-irradiation after surgery using standard fractionation, again in various doses (range 19.8-59.4 Gy), but we found no statistically significant differences correlating doses with outcomes.
Among others, the randomized French study NCT02013297,23 due to end soon, will shed some light on the different doses of stereotactic re-irradiation for ependymoma. It should clarify the differences between two modalities for locally re-irradiating patients suffering from a relapse of an already-irradiated ependymoma (3 fractions of 8 Gy vs 5 fractions of 5 Gy).
The probability of remaining LR-free in our 38/41 locally relapsing patients submitted to further treatment was 40% at 5 years, which is in line with other series,18,24–26 especially considering the quite large number of patients and very long follow-up. For locally relapsing patients, complete surgery and adjuvant RT seem to be the best strategy for the time being, but—like Régnier et al, who recently reported on the French series of re-irradiated children with ependymoma—we could find no dose-response relationship in our population.18 On the other hand, Messahel reported finding that over 45 Gy of re-irradiation achieved a better local control.27
We also identified no advantage of a given RT delivery technique (ie, standard vs hypofractionation) probably also due to the heterogeneity of doses and the small number of patient subgroups.
We cannot comment on the use of CSI for long-term local control because it was never used in cases of LR alone—unlike the series described by Merchant and Bouffet.14,15 We might add, however, that only 6/27 (22%) of our locally relapsing patients further relapsed with some form of distant dissemination (2 DM and 4 LR + DM), so promoting CSI for all locally relapsing patients seems to be dubious, at best. As regards this philosophy of delivering only as much treatment as is necessary, it is also worth mentioning the long-term disease-free survival of 3/6 patients who underwent complete resection and focal RT for a single site of DM. This contrasts with the report from Lobon et al of 8/10 patients relapsing after DM resection and focal re-irradiation.28 It is probably more a matter of small numbers and of shorter follow-up of some published series than of real differences, but the opportunity to limit treatment in some cases remains important because of the young age of these patients at diagnosis—and generally at relapse too.
Of our 8 patients with DM involving multiple sites who were treated not with CSI, but with RT to the multiple sites, none survived after retreatment: their deaths a median 8 months after relapsing are proof of the failure of the strategy we adopted. The experience of St. Jude’s reported by Merchant in 2008 was much more positive, with a more than 50% probability of survival at 5 years among 12 patients with metastatic relapses given CSI at 37.6-41.4 Gy.15 Tsang et al also found that patients with only distant relapses had a better outcome if re-irradiated using CSI—even better than patients with LRs treated with local irradiation alone.20 Their experience was further confirmed a year later when CSI was also administered with positive results to patients with LRs alone.29
The story of our patients with synchronous metastases was very similar to that of patients with DM. It is notable that, out of 5 patients, the only survivor had received CSI after surgery for a LR.
Despite the short time elapsing since their first RT, always at doses reaching at least 59.4 Gy, only one of the patients developed radionecrosis that resolved within 8 months. This goes to show, in these young patients at least, that re-irradiation is a safe option in the majority of cases, even after multiple surgical procedures and high total doses at first-line RT.18,28,30
While molecular subgrouping was of little prognostic value in these relapsed cases, it is worth noting the suggestive evidence of second relapses of tumors with 1q gain involving dissemination, as we had already seen for first relapses.5 While numerous papers have underscored the risk of recurrence associated with this chromosomal marker,31–34 few have detailed the site of relapse at first or further recurrences. Tsang et al, for instance, reported that further treatment failed at all distant sites in 5/6 patients with 1q and a second relapse after retreatment.20 Moreover, as described by Baroni et al at first diagnosis,34 even if within a very small patient group, the bad prognosis for the 5 patients with 6q loss was confirmed, despite LR in 4 cases.
It is important to mention that molecular profiling was unavailable for our patients’ tumors resected at relapse. Longitudinal studies on the molecular profiling of ependymoma relapses are likewise lacking in the literature. We anticipate that consistent molecular profiling of ependymoma recurrences will contribute both to establishing rational novel salvage treatments to test in the relapse setting, and to identifying prognostic biomarkers.
Surgery, alone or followed by CT, for locally relapsing patients, seemed to have little effect in saving our patients: all 5/5 had further LRs, and 4 died of their disease. This is quite different from the recently reported experience of the E-HIT-REZ study,13 in which adjuvant radiation did not ameliorate the outcome of patients with completely resected LRs. The addition of CT after RT for those patients with completely resected LR did not ameliorate final outcome, differently from what reported in the preliminary results of COG phase 3 protocol for first-line ependymoma treatment.7
The limitations of this work are clear due to the heterogeneity of treatments, limited molecular information, lack of long-term neurocognitive data for all surviving patients but nevertheless, we can draw some conclusions.
Among 64 relapsing patients previously treated according to the 2nd AIEOP protocol, there were 17 long-term survivors. The likelihood of survival was affected by the presence or absence of symptoms, and the completeness of surgery for the recurrence, but not by the RT schedule adopted or any inclusion of CT. Females with symptomatic relapses inexplicably had a particularly poor prognosis. There are still more questions than answers regarding salvage treatment for ependymoma. Only a homogeneous protocol with well-posed, possibly randomized questions and more refined biological details may be able to clarify this picture and ameliorate patients’ prognosis (ie, window therapy to facilitate complete resection, indications for CSI in the presence of 1q gain even if LR, re-irradiation technique for balanced tumors, adjuvant CT after RT for completely resected LR and single DM, plus modality/duration of follow-up in first-line and relapse trials).
Supplementary Material
Acknowledgments
The authors would like to acknowledge the support provided by Associazione Bianca Garavaglia Onlus, Busto Arsizio (VA), AIRC (Associazione Italiana per la Ricerca sul Cancro), and Associazione Bimbo Tu, Bologna, Italy. The authors also thank all families and patients who trusted us during treatment and follow-up and relapse strategies and the Italian neuropathologists and surgical pathologists who agreed to case contribution.
Conflict of interest statement. None to declare by any author.
Authorship statement. Experimental design (M.Massimino, F.B., P.M., F.G., P.J., and F.R.B.), implementation of the design (all authors), analysis and interpretation of the data (M.Massimino, F.B., P.M., F.G., P.J., and F.R.B.). All authors have been involved in the writing of the manuscript and approved the final version.
Funding
None.
References
- 1. Pajtler KW, Mack SC, Ramaswamy V, et al. The current consensus on the clinical management of intracranial ependymoma and its distinct molecular variants. Acta Neuropathol. 2017;133(1):5–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Liu AP, Shing MM, Yuen HL, et al. Timing of adjuvant radiotherapy and treatment outcome in childhood ependymoma. Pediatr Blood Cancer. 2014;61(4):606–611. [DOI] [PubMed] [Google Scholar]
- 3. Massimino M, Solero CL, Garrè ML, et al. Second-look surgery for ependymoma: the Italian experience. J Neurosurg Pediatr. 2011;8(3):246–250. [DOI] [PubMed] [Google Scholar]
- 4. Merchant TE, Bendel AE, Sabin ND, et al. Conformal radiation therapy for pediatric ependymoma, chemotherapy for incompletely resected ependymoma, and observation for completely resected, supratentorial ependymoma. J Clin Oncol. 2019;37(12):974–983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Massimino M, Barretta F, Modena P, et al. Second series by the Italian Association of Pediatric Hematology and Oncology of children and adolescents with intracranial ependymoma: an integrated molecular and clinical characterization with a long-term follow-up. Neuro Oncol. 2021;23(5):848–857. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Grundy RG, Wilne SA, Weston CL, et al. ; Children’s Cancer and Leukaemia Group (formerly UKCCSG) Brain Tumour Committee . Primary postoperative chemotherapy without radiotherapy for intracranial ependymoma in children: the UKCCSG/SIOP prospective study. Lancet Oncol. 2007;8(8):696–705. [DOI] [PubMed] [Google Scholar]
- 7. Smith A, Onar-Thomas A, Ellison D, et al. Phase III randomized trial of post-radiation chemotherapy in patients with newly diagnosed ependymoma ages 1 to 21 years. The 19th Symposium on Pediatric Neuro-Oncology. https://app.oxfordabstracts.com/events/1051/program-app/submission/211435. Accessed September 10, 2021.
- 8. Massimino M, Miceli R, Giangaspero F, et al. Final results of the second prospective AIEOP protocol for pediatric intracranial ependymoma. Neuro Oncol. 2016;18(10):1451–1460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Khatua S, Mangum R, Bertrand KC, Zaky W, McCall D, Mack SC. Pediatric ependymoma: current treatment and newer therapeutic insights. Future Oncol. 2018;14(30):3175–3186. [DOI] [PubMed] [Google Scholar]
- 10. Marinoff AE, Ma C, Guo D, et al. Rethinking childhood ependymoma: a retrospective, multi-center analysis reveals poor long-term overall survival. J Neurooncol. 2017;135(1):201–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Ritzmann TA, Rogers HA, Paine SML, et al. A retrospective analysis of recurrent pediatric ependymoma reveals extremely poor survival and ineffectiveness of current treatments across central nervous system locations and molecular subgroups. Pediatr Blood Cancer. 2020;67(9):e28426. [DOI] [PubMed] [Google Scholar]
- 12. Byer L, Kline CN, Coleman C, Allen IE, Whitaker E, Mueller S. A systematic review and meta-analysis of outcomes in pediatric, recurrent ependymoma. J Neurooncol. 2019;144(3):445–452. [DOI] [PubMed] [Google Scholar]
- 13. Adolph JE, Fleischhack G, Mikasch R, et al. Local and systemic therapy of recurrent ependymoma in children and adolescents: short- and long-term results of the E-HIT-REZ 2005 study. Neuro Oncol. 2021;23(6):1012–1023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Bouffet E, Hawkins CE, Ballourah W, et al. Survival benefit for pediatric patients with recurrent ependymoma treated with reirradiation. Int J Radiat Oncol Biol Phys. 2012;83(5):1541–1548. [DOI] [PubMed] [Google Scholar]
- 15. Merchant TE, Boop FA, Kun LE, Sanford RA. A retrospective study of surgery and reirradiation for recurrent ependymoma. Int J Radiat Oncol Biol Phys. 2008;71(1):87–97. [DOI] [PubMed] [Google Scholar]
- 16. Louis DN, Ohgaki H, Wiestler OD, et al. The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol. 2007;114(2):97–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Capper D, Jones DTW, Sill M, et al. DNA methylation-based classification of central nervous system tumours. Nature. 2018;555(7697):469–474. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Régnier E, Laprie A, Ducassou A, et al. Re-irradiation of locally recurrent pediatric intracranial ependymoma: experience of the French society of children’s cancer. Radiother Oncol. 2019;132:1–7. [DOI] [PubMed] [Google Scholar]
- 19. Massimino M, Barretta F, Modena P, et al. Pediatric intracranial ependymoma: correlating signs and symptoms at recurrence with outcome in the second prospective AIEOP protocol follow-up. J Neurooncol. 2018;140(2):457–465. [DOI] [PubMed] [Google Scholar]
- 20. Tsang DS, Burghen E, Klimo P Jr, Boop FA, Ellison DW, Merchant TE. Outcomes after reirradiation for recurrent pediatric intracranial ependymoma. Int J Radiat Oncol Biol Phys. 2018;100(2):507–515. [DOI] [PubMed] [Google Scholar]
- 21. Tsai JW, Manoharan N, Alexandrescu S, et al. Outcomes after first relapse of childhood intracranial ependymoma. Pediatr Blood Cancer. 2021;68(8):e28930. [DOI] [PubMed] [Google Scholar]
- 22. De B, Khakoo Y, Souweidane MM, et al. Patterns of relapse for children with localized intracranial ependymoma. J Neurooncol. 2018;138(2):435–445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. https://clinicaltrials.gov/ct2/show/NCT02013297 (consulted on May 24, 2021).
- 24. Stafford SL, Pollock BE, Foote RL, Gorman DA, Nelson DF, Schomberg PJ. Stereotactic radiosurgery for recurrent ependymoma. Cancer. 2000;88(4):870–875. [PubMed] [Google Scholar]
- 25. Stauder MC, Ni Laack N, Ahmed KA, Link MJ, Schomberg PJ, Pollock BE. Stereotactic radiosurgery for patients with recurrent intracranial ependymomas. J Neurooncol. 2012;108(3):507–512. [DOI] [PubMed] [Google Scholar]
- 26. Hoffman LM, Donson AM, Nakachi I, et al. Molecular sub-group-specific immunophenotypic changes are associated with outcome in recurrent posterior fossa ependymoma. Acta Neuropathol. 2014;127(5):731–745. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Messahel B, Ashley S, Saran F, et al. ; Children’s Cancer Leukaemia Group Brain Tumour Committee . Relapsed intracranial ependymoma in children in the UK: patterns of relapse, survival and therapeutic outcome. Eur J Cancer. 2009;45(10):1815–1823. [DOI] [PubMed] [Google Scholar]
- 28. Lobón MJ, Bautista F, Riet F, et al. Re-irradiation of recurrent pediatric ependymoma: modalities and outcomes: a twenty-year survey. SpringerPlus. 2016;5(1):879. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Tsang DS, Murray L, Ramaswamy V, et al. Craniospinal irradiation as part of re-irradiation for children with recurrent intracranial ependymoma. Neuro Oncol. 2019;21(4):547–557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Liu AK, Foreman NK, Gaspar LE, Trinidad E, Handler MH. Maximally safe resection followed by hypofractionated re-irradiation for locally recurrent ependymoma in children. Pediatr Blood Cancer. 2009;52(7):804–807. [DOI] [PubMed] [Google Scholar]
- 31. Modena P, Lualdi E, Facchinetti F, et al. Identification of tumor-specific molecular signatures in intracranial ependymoma and association with clinical characteristics. J Clin Oncol. 2006;24(33):5223–5233. [DOI] [PubMed] [Google Scholar]
- 32. Kilday JP, Mitra B, Domerg C, et al. Copy number gain of 1q25 predicts poor progression-free survival for pediatric intracranial ependymomas and enables patient risk stratification: a prospective European clinical trial cohort analysis on behalf of the Children’s Cancer Leukaemia Group (CCLG), Societe Francaise d’Oncologie Pediatrique (SFOP), and International Society for Pediatric Oncology (SIOP). Clin Cancer Res. 2012;18(7):2001–2011. [DOI] [PubMed] [Google Scholar]
- 33. Araki A, Chocholous M, Gojo J, et al. Chromosome 1q gain and tenascin-C expression are candidate markers to define different risk groups in pediatric posterior fossa ependymoma. Acta Neuropathol Commun. 2016;4(1):88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Baroni L, Sundaresan L, Heled A, et al. Ultra high-risk PFA ependymoma is characterized by loss of chromosome 6q. Neuro Oncol. 2021;23(8):1360–1370. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.