Prospective randomized clinical studies involving reirradiation: update of a systematic review

Except for the landmark bone metastases trial from 2014, the randomized phase 3 trial by Liu et al. [19] represents the largest reirradiation study (n = 200). These researchers confirmed that salvage surgery (endoscopic nasopharyngectomy) is an important option for resectable recurrences, leading to improved 3‑year overall survival compared to intensity-modulated radiotherapy (IMRT; 86% versus 68%, p = 0.0015). Most patients had N0‑1 disease. Seventy-one (71%) of the 100 patients in the IMRT group received cisplatin-based chemotherapy. IMRT was not fully standardized regarding dose per fraction and total dose (60–70 Gy). A proportion of patients (> 30%) had received initial radiotherapy with less advanced techniques that often result in higher doses to critical organs at risk. In line with previous studies, reirradiation caused a relatively high rate of serious toxicity. The most common grade 3 or worse radiation-related late adverse event was pharyngeal mucositis (26% after IMRT). Five (5%) of the patients who underwent surgery and 20% of patients who underwent IMRT died due to late toxic effects specific to radiotherapy. Connecting the toxicity in the reirradiation arm to initial radiotherapy or trial radiotherapy was difficult due to the long-term nature of radiation-related toxicity. However, the fact that high-dose reirradiation may cause fatal toxicity is well-known from the literature [26‐28].

A previous randomized trial in nasopharynx cancer had evaluated de-escalation of the equivalent dose (EQD2) for late responding normal tissues, while maintaining the same EQD2 for tumour cells [10]. This was accomplished by selecting a slightly hypofractionated experimental regimen, which was compared to a conventional regimen with 2‑Gy fractions and longer overall treatment time (Table 1). It was assumed that normal tissue sparing would result in less life-threatening toxicity and thus better overall survival, i.e. a better therapeutic ratio. Indeed, the results showed a 5-year overall survival rate of 44% versus 30%, but the difference was not statistically significant (p = 0.06) in this underpowered study. Serious toxicity was not uncommon and therefore, further EQD2 reduction has now been studied. Hyperfractionation was employed to mitigate toxicity, while maintaining the same overall treatment time [20]. Standard IMRT was identical to the previous study, i.e. 60 Gy in 27 fractions, while hyperfractionated IMRT featured 65 Gy in 54 fractions (2 fractions per day). Overall survival and toxicity were the primary endpoints in this 144-patient trial. Patients were not allowed to have radiation-induced complications grade ≥ 3 before reirradiation. Regarding outcomes, reduced grade 3 or worse late radiation-induced toxicity was observed in the hyperfractionation arm (34% versus 57%, p = 0.02), in line with radiobiological assumptions. Significant differences favouring hyperfractionated radiotherapy were seen in the general quality-of-life domains of global health status, role functioning, and social functioning, and in the symptom burden domains of pain, financial difficulties, and loss of appetite. Furthermore, 3‑year overall survival was better after hyperfractionation (75% versus 55%, p = 0.01). Efficacy was suboptimal, given that 49% of patients in the hyperfractionation group and 46% in the standard once-daily fractionation group had locoregional relapse. A possible strategy would be to moderately increase the dose per fraction in the hyperfractionation regimen and/or to add 1–2 additional days, resulting in dose escalation. However, this carries a risk of increasing fatal toxicity and leaving overall survival unchanged. Alternatively, assuming that not all recurrent tumours are sufficiently sensitive to further irradiation, incorporating other approaches such as drug treatment may be warranted [29].

A French group reported a phase 2 randomized multicentric trial comparing two regimens of reirradiation after salvage surgery: mono-fractionated radiotherapy with concomitant chemotherapy and hyperfractionated radiotherapy with cetuximab as experimental arm [21]. The primary endpoint was the comparison of the number of patients with a treatment interruption for more than 15 days, due to acute toxicity. Thus, size was limited to 60 patients. Inclusion criteria included, e.g., > 6 months between the initial course of radiotherapy and salvage surgery, sufficient healing for beginning reirradiation within 8 weeks of salvage surgery, age 18–75 years, WHO performance status (PS) 0–1, no severe sequelae of initial radiotherapy, and > 50% of recurrent tumor had received ≥ 50 Gy during previous irradiation. One arm employed the so-called Vokes protocol, i.e. 60 Gy in 11 weeks (6 cycles, with each cycle delivering 2 Gy/fraction, 5 days/week, with concomitant hydroxyurea (1.5 g/d orally) and continuous infusion fluorouracil (800 mg/m²/day), with 9‑day rest periods between cycles (split course)). In the radiobiologically more sound hyperfractionated radiotherapy arm, patients received a total dose of 60 Gy in 50 fractions, 1.2 Gy/fraction, 2 fractions/day, 5 days/week during 5 weeks. Cetuximab was initiated one week before radiotherapy at a loading dose of 400 mg/m², followed by weekly 250 mg/m² during radiotherapy. Similar rates of more than 15 days of treatment interruption due to toxicity were reported (n = 1 and 0, respectively, p = 0.49). Overall, toxicities and disease-free survival (DFS) were not different between the two arms. Thus, hyperfractionated reirradiation of 60 Gy/5 weeks with cetuximab was tolerable. The median DFS of 12 months was modest, however approximately 30% of patients remained progression-free at 4 years. Loco-regional recurrence was the main cause of death, again demonstrating that recurrent disease is difficult to control, both at first and subsequent relapse.

In a previous study, 130 patients who underwent salvage surgery were randomly assigned to receive reirradiation (60 Gy) combined with concomitant chemotherapy (5-FU and hydroxyurea) versus no adjuvant treatment [12]. A significant improvement with regard to the primary endpoint of loco-regional tumour control (hazard ratio (HR) 2.7; 95% confidence interval (CI) 1.7–4.5; p < 0.001) and DFS (HR 1.7; 95% CI 1.1–2.5; p = 0.01) was observed in those patients who were assigned to receive postoperative chemo-reirradiation compared to those who underwent surgery alone. However, this benefit in DFS did not translate into a significant improvement of overall survival. The gain in loco-regional tumour control and DFS was achieved at the cost of significantly higher rates of grade 3–4 late side effects (39% versus 10% at 2 years, respectively). Importantly, patients allocated to the wait-and-see arm could receive salvage chemo-reirradiation at the time of loco-regional recurrence after salvage surgery, which was the case in 25% of patients. This type of cross-over reduces the likelihood of improved overall survival. In the newer but smaller hyperfractionation study [21], 2/14 patients in that arm had grade 3–4 toxicity at 2 years (trismus and dysphagia), supporting the concept of hyperfractionation if one proceeds to postoperative reirradiation in selected cases. Based on the results of both studies, a wait-and-see policy may also be considered, in particular when patients suffer from continuous late toxicity from the first course of treatment.

Bergman et al. performed a small, prematurely closed trial of fractionated stereotactic radiosurgery with chemotherapy versus chemotherapy alone for bevacizumab (BEV)-resistant high-grade glioma, largely glioblastoma [22]. Chemotherapy drugs were chosen at the discretion of the treating physician recommended by the tumor board. Patients were stratified by KPS (≤ 80 vs > 80). The primary endpoints were local tumour control at 2 months and PFS. It was assumed that the reirradiation group would have a local control rate of 40% compared to a 10% rate in the comparator group. A total dose of 32 Gy (4 fractions of 8 Gy) was prescribed to a gross tumour volume (GTV; range 3‑186 cc), defined as the T1-weighted contrast-enhancing lesion plus the area of diffusion-weighted imaging seen on the co-registered magnetic resonance imaging scans (MRI). A dose of 24 Gy was prescribed to a clinical target volume (CTV) defined as the area of the new or change in T2-weighted fluid-attenuated inversion recovery abnormality. Treatment was planned via a simultaneous integrated boost (SIB) technique. The dose was prescribed to the highest isodose line encompassing the CTV, which ranged from 50 to 95% of the maximum dose. Planning target volume (PTV) was equal to GTV or CTV, respectively. There was no limit to the maximum target volume. It was assumed that all critical structures had received the maximum point dose during initial external beam radiation treatment. The dose constraints from the American Association of Physicists in Medicine Task Group 101 (TG101) report of stereotactic body radiation therapy for 3‑fraction treatments were used [30]. Coverage to the 32 Gy volume was not compromised based on those constraints. The median time from initial diagnosis to enrollment was 20.5 months (range 7 to 268) and the median number of prior recurrences was 3 (range 2 to 6). The reirradiation group had an improved median PFS compared to BEV-based chemotherapy alone (5.1 months, 95% CI 4.1–6.2 vs 1.8 months, 95% CI 1.2–2.8; p < 0.001). The reirradiation group had a better overall survival compared to the BEV/chemotherapy only group, but this difference was not statistically significant (median overall survival: 7.2 months [95% CI 6.1–8.1] vs 4.8 months [95% CI 1.7–7.6]; p = 0.11). Reirradiation toxicity grade 3 included one patient each with headache, nausea/vomiting, new onset weakness, intratumoural hemorrhage, and seizure (no grade 4 or 5 toxicities). There were no documented cases of radionecrosis.

Lower reirradiation EQD2 was employed in the German ERGO2 trial, a randomized trial of calorie-restricted ketogenic diet and fasting in addition to reirradiation for malignant glioma, largely glioblastoma [23]. It included 50 patients and required KPS ≥ 60 and interval at least 6 months, among other criteria. Most patients had 5 or 10 fractions of reirradiation (5 fractions of 4 Gy (mainly), 10 fractions of 3.5 Gy). The authors assumed PFS at 6 months to increase from 0% to 30%. No significant difference was observed in PFS at 6 months: 20% vs. 16%, p = 0.7. Similar median overall survival was achived (10.9 vs. 9.5 months). During the dietray intervention phase until day 12, 9 adverse events (experimental arm: 4, standard diet (SD): 5) were reported. Three patients suffered from epileptic seizures. The other adverse events were headache, nausea, or possible epileptic seizures with short-lasting aphasia, which could not be categorized by the description from the patient. From day 12 until the first MRI follow-up after 1 month, 11 adverse events (experimental arm: 5, SD: 6) were reported, the majority of which were epileptic seizures. The publication did not comment on radionecrosis. In summary, ERGO2 demonstrated that this dietary intervention can be safely applied to patients with recurrent glioma.

The largest (n = 170) and thus most important recent trial was NRG Oncology/RTOG1205, a randomized phase 2 trial to determine whether BEV plus reirradiation (experimental arm) would improve survival (primary endpoint) compared with BEV alone (control arm) [24]. Patients were stratified by age (< 50 years vs ≥ 50 years), KPS (60 vs 70–80 vs 90–100), and recent re-resection. Inclusion criteria were modified after slow accrual to allow for patient enrollment with up to three relapses, a KPS of ≥ 60, and recurrent tumors ≤ 6 cm. Multifocal recurrence was no longer excluded, provided that the composite tumor volume was ≤ 6 cm. Reirradiation dose was 35 Gy in 10 fractions, using 3D conformal technique, IMRT or protons. The protocol defined certain organ-at-risk doses (planning organ-at-risk volumes, 3 mm margin) and acceptable variations. Optic nerves and chiasm (D0.3cc) were constrained to 20 Gy (acceptable variation: 25 Gy). The respective figures were 24 and 30 Gy for brain stem (D0.3cc). GTV was defined as enhancing tumour using computed tomography and/or MRI or postoperative resection cavity if no residual enhancing tumour was noted. A PTV expansion of at least 3 mm was used. BEV was administered at a dose of 10 mg/kg every 2 weeks until disease progression. Patients randomly assigned to the BEV and reirradiation arm received an initial induction BEV dose (day 1) followed by concurrent BEV and radiation at the next dose (day 14), and then once every 14 days until disease progression. The median survival for the control arm was 9.7 months (80% CI, 9.0 to 11.2) and 10.1 months (95% CI, 9.5 to 11.3) for the experimental arm (HR, 0.98; 80% CI, 0.79 to 1.23, p = 0.46). Twelve patients on the BEV arm received reirradiation as salvage therapy. Furthermore, some imbalances in baseline characteristics favoured the BEV arm. The only notable survival difference between arms was noted for the KPS 90–100 subgroup, in which the BEV/reirradiation arm showed improved survival (HR, 0.67; 95% CI, 0.40 to 1.13; p = 0.13). The median PFS for the control versus reirradiation arms was 3.8 versus 7.1 months, respectively (HR, 0.73; 95% CI, 0.53 to 1.00; p = 0.05). No delayed grade 3 or worse treatment-related central nervous system adverse events were reported. The authors discussed that optimal treatment for patients with recurrent glioblastoma remains controversial in the absence of improved survival. Their study confirmed meaningful improvement in PFS, including the 6‑month PFS rate, which patients may consider clinically beneficial. Quality-of-life was not evaluated. Treatment was safe and well-tolerated with no delayed brain toxicities. Therefore, reirradiation (the study regimen or fewer stereotactic fractions) remains a reasonable option especially for patients with small volume of recurrence and good KPS, as also reflected in a recent guideline [31].

Schouten et al. reported a prematurely closed randomized phase 2 study of reirradiation and hyperthermia versus reirradiation and hyperthermia plus chemotherapy for locally recurrent breast cancer in previously irradiated areas not suitable for resection (n = 49, planned: 104, slow accrual) [25]. Concurrent hormonal therapy was allowed. Patients were stratified by size of recurrence (> 5 cm or ≤ 5 cm) and time interval between primary breast cancer and first recurrence (> 3 years or ≤ 3 years). The authors tried to detect an increase in the local control rate after 1 year from 54% in the standard treatment arm to 69% in the experimental arm (corresponding to a HR of 0.6). Originally, 32 Gy was given in 8 fractions of 4 Gy in 4 weeks, at 2 fractions per week (3 days in between the fractions). After January 2015, the radiotherapy schedule was changed to 46 Gy in 23 fractions of 2 Gy, at 5 fractions per week. Local microwave hyperthermia was delivered once a week, starting within 1 h after radiotherapy. Patients receiving the 32-Gy schedule were given four sessions of hyperthermia and patients receiving the 46-Gy schedule were given five sessions of hyperthermia, a modality with long track record [32, 33]. Patients were treated with weekly cisplatin 40 mg/m² given intravenously for 4 courses, concurrent with hyperthermia. Approximately half of the patients had already been unsuccessfully treated for the current relapse with surgery, chemotherapy, hormonal therapy or trastuzumab. Local (infield) progression-free rate at 1 year was high in both arms, 81.5% in the standard arm and 88% in the combined arm. About 60% per arm achieved a complete response. With or without cisplatin, most patients had subsequent local control until last follow-up or death. No significant difference regarding any endpoint was observed. One patient in the standard arm died due to a necrotizing thoracic wall defect in the radiated area 3 months after treatment. We could not identify randomized trials on repeat breast-conserving surgery with reirradiation, a concept gaining increasing acceptance in well-selected patients [34].