Second re-irradiation: a narrative review of the available clinical data

Abstract

Background: Considerable controversy exists about the safety and efficacy of second re-irradiations (three courses of radiotherapy to overlapping volumes). Therefore, all published clinical studies were reviewed.

Material and methods: Contemporary and historical articles were identified. Outcomes such as survival, local control, symptom improvement and side effects were extracted. Contemporary results were grouped by anatomical location of the re-irradiated region in the body.

Results: Most data were derived from central nervous system tumors, pelvic tumors and bone metastases. We could include nine contemporary, retrospective studies with 2–25 patients each. Nearly, all patients were treated with palliative intent. Most of the prescribed re-irradiation regimens were highly individualized and thus difficult to compare. Symptomatic responses were recorded in most patients. In palliatively treated patients with pelvic and bony target volumes, high-grade toxicity was uncommon.

Conclusions: Despite of issues related to study size, length of follow-up and calculation of lifetime cumulative equivalent dose, the available data provide an initial framework for future studies and discussion of dose constraints. Selected dose-fractionation regimens may result in a satisfactory therapeutic ratio even after two previous courses of radiotherapy, if these were well tolerated.

Background

Re-irradiation has a surprisingly long track record [1], probably because clinicians involved in cancer care during the first half of the last century were facing a challenging situation. The armamentarium of effective salvage treatment options was limited, while relapse was a common scenario [2]. During the infancy of radiobiology research, dose–effect relationships were not well defined, neither for permanent tumor control nor for late normal tissue tolerance [3]. Furthermore, external beam treatment machines were not able to deliver a sufficient depth dose without excessive built-up dose in skin and surface-near tissues, while brachytherapy required a set of eligibility criteria, e.g., accessibility and limited tumor stage [4]. Some of the early reports contain brief descriptions of patients who received a third course of radiotherapy to the previously treated region [1]. However, no detailed cumulative doses and outcomes were provided. The present review summarizes historical and contemporary publications, assuming that discussions about a second re-irradiation still create considerable debate, while referrals are expected to increase, because more efficacious systemic therapies extend survival without eliminating the occurrence of skeletal-related events, brain metastases and other indications for radiotherapy [5–7].

Material and methods

Contemporary articles were identified from PubMed (last access 07 September 2017). Historical publications were also searched in PubMed, plus the electronic archives of the British Journal of Radiology, British Medical Journal, Strahlentherapie, Journal of the American Medical Association, California Medicine and its predecessors. In addition, we performed crosschecking of the references from already included publications. The key words 'reirradiation', 're-irradiation', 'repeat radiotherapy', ‘second radiotherapy’, ‘third radiotherapy’, 'radiation retreatment' and 'recurrent AND radiotherapy' were entered. English or German language articles were included if they had analyzed at least two patients. Whatever information a study provided about dose, technique, toxicity and other outcomes is as described in the Results section. Unfortunately, several of the historical studies provided no information about cumulative dose or exposure, other relevant baseline features and outcomes. Contemporary results were grouped by anatomical location of the re-irradiated region in the body. Whenever possible, the cumulative equivalent dose in 2-Gy fractions (EQD2) was calculated, with α/β = 2 Gy for central nervous system tissue, 3 Gy for pelvic organs and 10 Gy for tumors.

Table 1. Published studies that included at least 10 patients (all retrospective, historical data published before 1990 not shown).

Authors	Patients	Treatment concept	Fractionation	Time intervals	Follow-up	Results	Late toxicity grade 3–5
Abusaris et al. [14]	N = 23, different sites and tumor types	Individualized, palliative aim	Variable, e.g., 20 Gy or 32 Gy in 4-Gy fractions	15 and 7 months	Median 7 months	Median OS 7 months, pain response in 71%	1 grade 3 skin reaction
Thibault et al. [17]	N = 24, spinal metastases from different primaries	Individualized, palliative aim	Variable, 24–35 Gy in 2–5 fractions, SBRT	n/a	Median 7 months	n/a^a	No grade 3–5
Katsoulakis et al. [18]	N = 10, spinal metastases from different primaries	Individualized, palliative aim	Variable, e.g., 25 Gy or 30 Gy in five fractions, IG-IMRT	18.5 and 11.5 months	Median 12 months	Median OS 13 months, pain or neurological improvement in 80%, crude LC 80%	No grade 3–5
Jeremic et al. [19]	N = 25, bone metastases from different primaries	Standardized, palliative aim	4-Gy single fraction	n/a and 20 weeks	n/a	Median OS 7 weeks, pain response in 80%	No grade 3–5

SBRT: stereotactic body radiotherapy; OS: overall survival; IG-IMRT: image-guided intensity-modulated radiotherapy; LC: local control.

^a Results not stratified by number of courses (2 vs. 3).

Results

Several historical sources reviewed in [1] have documented that a minority of selected patients, e.g., with brain, head and neck, and gynecological malignancies received more than two courses of radiotherapy with overlapping fields in the decades before World War II. However, no general conclusions can be drawn, because these case studies referred to vastly different scenarios and lacked standardized reporting of baseline data and outcomes. A slightly larger series, although limited to seven patients, was published by Murphy and Schmitz [8]. Seven of their re-irradiated patients with cancer of the uterine cervix received a third course of radiotherapy, either by radium or X-ray technique (unknown dose and area/volume of overlap). The median time interval between first and second re-irradiation was 12 months, while the time elapsed between first-line and recurrence treatment was not reported. Average survival was 50 months from primary therapy and 22 from first re-irradiation. Further outcomes including side effects were not mentioned in this article.

Central nervous system

The brain metastases re-irradiation literature also features historical retrospective data describing three courses of palliative whole-brain radiotherapy (WBRT). The study by Shehata et al. included 81 patients with lung, breast and other primary tumors [9]. Of these, 35 (43%) were retreated, 12 were retreated twice and three patients received even four courses. Fractionation was not reported for these subgroups. Overall, a single dose of 10 Gy was given to 68% of the patients and a short course of less than one week (2–5 fractions) was also commonly prescribed. Time interval, fractionation details and cumulative doses were not reported. A patient with non-small cell lung cancer who received four courses over a 20-month period (total dose 74 Gy, EQD2 not available) experienced a clinical response each time and survived for more than 2 years. A clinical benefit was reported after the first, second and third course in 69, 68 and 50% of the patients, respectively. Mean duration of improvement was 1.8, 2.6 and 1.5 months, respectively. The corresponding figures for mean survival were 102, 166 and 261 days. Cognitive function was not reported. Incidence of selected side effects was reported for all patients combined, rather than stratified by number of courses.

Among several studies mentioning that selected patients with brain metastases who received two treatments with stereotactic radiosurgery (SRS) to the same lesion (salvage SRS for local relapse) also were treated with WBRT at some point in time only the one published by Koffer et al. provided sufficient details [10]. The authors used single fraction gamma knife SRS in all 22 cases (24 re-irradiated lesions). The second SRS sometimes targeted the tumor bed after surgical resection (n = 5). Median time interval between the two SRS treatments was 13 months and mean target size was 2.25 and 3.3 ml, respectively. A mean dose of 18 and 15.5 Gy was administered, respectively. Eight patients had received up-front WBRT before initial SRS (unknown interval and dose, cumulative EQD2 therefore not available). For all 22 patients, median survival after the second SRS was 9 months. The actuarial rate of radiation necrosis at 12 months was 9%. WBRT any time before SRS was not significantly associated with local failure or radiation necrosis after second SRS. However, trends emerged toward higher local failure rates in WBRT patients (37.5% vs. 12.5%, p = .155) and toward increased risk of radiation necrosis (37.5% vs. 6.3%, p = .053). Overall, 75% of the patients treated with three courses eventually developed local failure or radiation necrosis.

Very limited data exist regarding other brain tumors. A Spanish group reported two patients with diffuse intrinsic pontine glioma (DIPG) who received a second re-irradiation course at second tumor progression [11]. The first child received 54 Gy in 30 fractions, 30.6 Gy in 17 fractions approximately 8 months later, and 21.6 Gy in 12 fractions after a time interval of 4–5 months, resulting in a cumulative EQD2 of 101 Gy (α/β  =  2 Gy). Symptomatic and imaging improvement was obtained each time. Tolerance was good. The patient died 4 months after the second re-irradiation. The second child received 39 Gy in 13 fractions followed by anti-angiogenic therapy. This was followed by 20 Gy in 10 fractions approximately 11 months later. Afterwards, irinotecan and rapamycin were given. Eight months later, a second re-irradiation course with 20 Gy in 10 fractions with concomitant temozolomide was administered with good tolerance and rapid clinical response. Survival was 1 year after second re-irradiation. The cumulative EQD2 was equivalent to 89 Gy in this case.

Lung

In a Swedish study of 29 re-irradiated patients, many with lung metastases, four received three courses of SBRT to overlapping volumes, including one patient with four overlapping series [12]. Mean EQD2 to the CTV was 109 Gy, identical to the first two series. Two patients had peripheral lesions, one of whom developed grade 3 dermatitis and the other one grade 3 dyspnea. Of the two patients with central target volumes one developed grade 2 pleural effusion, lung fibrosis and atelectasis (after 8 Gy × 5, 8 Gy × 5 and 10 Gy × 3), and the other one died during stent placement for fibrotic stenosis of the superior vena cava. Three individual courses consisted of 8 Gy × 5 each and the fourth one of 6 Gy × 7. A grade 4 tracheal fistula was diagnosed already after two courses in this case. Local control and survival were reported for all 29 patients who had received overlapping SBRT courses.

Pelvis

Feddock et al. utilized permanent interstitial brachytherapy (¹⁹⁸Au or ¹³¹Cs) for local recurrences of pelvic malignancies, mostly uterine cancer, in 42 patients [13]. Initial treatment included pelvic radiotherapy with or without brachytherapy and in one case brachytherapy alone. Nine patients underwent a second salvage procedure with re-implantation of the same lesion in the vagina. The time intervals were not reported. The median dose was 45 and 40 Gy for the first and second implant, respectively (EQD2 44 and 39 Gy, respectively). The median cumulative lifetime EQD2 was 152 Gy (range 115–172). Only three (33%) of these tumors were still controlled at the time of death or of last follow-up, with a median time to failure of 7.7 months. All nine patients demonstrated soft tissue necrosis, which was persistent beyond 3 months and was symptomatic in only two of them.

Abusaris et al. collected data from a larger series of 23 patients treated palliatively with second re-irradiation [14]. The main goal was pain relief. Fourteen regions were located in the pelvis (mainly rectal cancer) and six in the thoracic wall (Table 1). In contrast to other publications, a consistent policy regarding organs at risk (OAR) was employed. The maximum dose was set as 50% more than the normal, first-line constraint. If the interval to the next radiotherapy course was relatively short, i.e., 6–12 months, only 25% more was allowed, and no increase was permitted with intervals shorter than 6 months. All treatment schedules were recalculated to EQD2 with α/β  =  3 Gy. Re-irradiation techniques and fractionation were highly individualized. For example, brachytherapy and stereotactic radiotherapy were employed. Illustrative cases with pelvic target volumes include 52 Gy in 2-Gy fractions followed by 32 Gy in 4-Gy fractions and 20 Gy in 4-Gy fractions, or 44.65 Gy in 2.35-Gy fractions followed by 30 Gy in 2-Gy fractions and 18 Gy in 6-Gy fractions. The volume of overlap was estimated with different methods, depending on technique and treatment planning system. In three cases, not all necessary data were available to estimate the cumulative lifetime doses. Two patients were lost to follow-up. The median time interval between the courses was 15 and 7 months, respectively. The median follow-up from last radiotherapy was 7 months, identical to median overall survival. Pain reduction after third radiotherapy was seen in 71% of the patients. In eight patients, the authors' dose constraints were exceeded (rectum, bowel, bladder or sacral nerve). No patients experienced grade 4 acute toxicity. Less than 10% each reported acute grade 3 pain or dysuria. No high-grade late toxicity was recorded, except for one grade 3 skin reaction. Therefore, the authors concluded that their method of defining EQD2 constraints is safe. An example provided in the article refers to rectum and a time interval of more than one year each. After exposure to 66 Gy (EQD2, α/β = 3 Gy) in first line, an additional 33 Gy (EQD2, α/β = 3 Gy) can be given in the second course and also in the third one. If one accepts a first-line EQD2max of 60 Gy for bowel, the resulting cumulative EQD2 would be 90 Gy in the second course. For bladder, these authors suggested 110 Gy cumulative EQD2 in the first two courses, meaning that another 36 Gy can be given more than one year later, if a third course becomes necessary.

Four patients with a pathologically proven second local recurrence from prostate cancer were treated in an outpatient magnetic resonance imaging (MRI)-guided setting with a single fraction of 19 Gy focal high-dose-rate brachytherapy by Maenhout et al. [15]. For this second salvage, delineation was performed using choline-PET-CT or gallium-PSMA-PET in combination with multiparametric 3 Tesla MRI in all four patients. All had received 145 Gy using ¹²⁵I as first salvage treatment. The treatment volume was approximately hemi-gland. Whole gland ¹²⁵I brachytherapy was the initial treatment in three cases, intensity-modulated radiotherapy (IMRT) in the remaining patient. Time intervals ranged from 5 to 8 years (first salvage) and 3 to 6 years (second salvage). Disease stage before second salvage was low risk (n = 2) and high risk (n = 2, due to T3), and all had PSA < 6 ng/ml. PSA doubling time was one year or more. Toxicity rates of ≤2 (CTCAE version 4.0) from the previous treatments were required. In all second salvage treatments, the institutional constraints derived from Holly et al. [16] for rectum, bladder and urethra were met. Median treatment volume (gross tumor volume, GTV) was 4.8 cm³ (range, 1.9–6.6 cm³). No margin was applied to the GTV. With a median follow-up of 12 months (range, 6–15), there were two patients with biochemical recurrence as defined by the Phoenix-definition. There were no patients with grade 3 or more toxicity.

Bones (higher RT dose)

Thibault et al. reported their experience in salvaging spinal metastases initially irradiated with stereotactic body radiation therapy (SBRT), who subsequently progressed with imaging-confirmed local tumor progression, and were re-irradiated with a salvage second SBRT course to the same level (n = 56 spinal segments in 40 patients) [17]. In addition, 24 of 56 segments had initially been irradiated with conventional external beam radiation therapy before the first course of SBRT (20–30 Gy in 5–40 fractions). The SBRT doses were 20–30 Gy (first) and 24–35 Gy (second) in 2–5 fractions for the subgroup of patients who received a total of three courses. The time intervals between these treatments were not reported. The median follow-up time after salvage second course SBRT was 6.8 months. No radiation-induced compression fractures have been observed in the 19 non-operated spinal segments after second salvage SBRT. However, many patients had received surgical procedures earlier. Furthermore, no case of radiation-induced myelopathy was observed, nor any cases of other grade ≥3 toxicities. Of note, median cumulative spinal cord EQD2 (α/β = 2 Gy) from all three courses was as low as 73.9 Gy (Pmax), which sometimes was achieved at the cost of reduced target volume coverage. For the thecal sac (cauda equina) in lower spine segments, the respective EQD2 was 80.4 Gy. The median time to progression after second salvage SBRT was only 3 months in the 13 cases with local failure (no separate data for those with three courses). Previous conventional radiotherapy was not among the risk factors for local failure in multivariate analysis. According to the authors, a second course of spine SBRT, most often with 30 Gy in four fractions, for spinal metastases that failed previous SBRT is a feasible and efficacious salvage treatment option (crude local control 77%).

The retrospective analysis by Katsoulakis et al. included 10 patients who received a third course to spinal metastases, always with image-guided (IG)-IMRT [18]. Eight patients underwent surgery prior to the third course. The median time interval between the first and second courses was 18.5 months, range 3–144 months. Two patients had intervals ≤6 months, both had a minimum of 9 months to the next radiotherapy. The initial course was delivered with nine different dose-fractionation regimens. The second course was less variable, 30 Gy in five fractions of 6 Gy (n = 5) and 25 Gy in five fractions of 5 Gy (n = 2) was often prescribed. After a median time interval of 11.5 months, range 2–52 months, a third course was delivered. The median follow-up from this final course was 12 months, range 2–26 months. The third course consisted of 30 Gy in five fractions of 6 Gy (n = 4), 25 Gy in five fractions of 5 Gy (n = 3), 20 Gy in five fractions of 4 Gy (n = 2), and 27 Gy in three fractions of 9 Gy (n = 1). The median spinal cord maximum EQD2 (α/β = 2 Gy) was 70.7 Gy, range 51.9–101.7 Gy. The median overall survival was 13 months. Pain or neurological symptoms were improved in 80% of patients. The crude rate of local control was 80%. The worst degree of acute toxicity was grade 3 in two patients. Late toxicity included two cases of grade 1 dysphagia, one case of grade 1 neuropathy and one case of grade 2 neuropathy. The latter was recorded in a patient who had received a Dmax of 101.7 Gy to the lumbar L3 region.

Bones (low RT dose)

Jeremic et al. investigated the efficacy of a second single dose of 4 Gy for patients with painful bone metastases who had already twice received single fraction treatment (4, 6 or 8 Gy plus 4 Gy) in a small group of 25 patients [19]. Nineteen were responders and six non-responders to two prior single fractions, the latter one being 4 Gy. Median time interval between the second and third course was 20 weeks. The overall response rate was 80%. No significant difference was found between the previous responders and non-responders regarding both complete and overall response rate. However, response duration was longer in the previous responders (8 weeks vs. 2 weeks). No acute or late high-grade toxicity was observed and no pathological fractures or spinal cord compressions were seen. However, median survival was limited to 7 weeks. Palliation until death was achieved in 64% of patients. This experience led the authors to conclude that the third single fraction treatment of 4 Gy was effective and not toxic.

Discussion

Our aim was to review historical and contemporary publications that described the results of a second course of re-irradiation, i.e., three courses of radiotherapy. The historical data were to sparse to draw meaningful conclusions, and were also characterized by a lack of important treatment and outcome parameters. We identified and included nine contemporary studies [10–15,17–19]. These studies had included 2–25 patients. Only one dealt with a clearly curative approach, i.e., brachytherapy for prostate cancer [15]. Most of the prescribed re-irradiation regimens were highly individualized and thus difficult to compare. Overall, the included studies provide preliminary evidence, at best. Without doubt, second re-irradiation has been offered to very few and highly selected patients, given that all studies emanated from institutions with very high patient numbers. These patients consented to a third course of radiotherapy, which likely would not have been the case if they had had experienced severe and/or permanent side effects or short-lived responses before. No firm criteria for decision-making could be derived from the available literature.

The brain metastases data question the benefit of second re-irradiation, at least if two courses of SRS are involved [10]. In eight patients treated with up-front WBRT and two courses of SRS trends emerged toward higher local failure rates in WBRT patients compared to those with two courses of SRS only (37.5% vs. 12.5%, p = .155) and toward increased risk of radiation necrosis (37.5% vs. 6.3%, p = .053). Overall, 75% of the patients treated with three courses eventually developed local failure or radiation necrosis. In other words, a cumulative dose associated with high toxicity was still not sufficient to provide excellent local control. This finding supports the historical argument against re-irradiation, namely that recurrent tumors are not sufficiently radiosensitive to warrant further radiotherapy [20,21].

Comparable trends were seen with high-dose pelvic second re-irradiation [13]. Only three of nine (33%) pelvic tumors were still controlled at the time of death or of last follow-up, with a median time to failure of 7.7 months. All nine patients demonstrated soft tissue necrosis, which was persistent beyond 3 months. Although no grade 3 or higher toxicity occurred, two of four prostate cancer patients in a different study experienced biochemical recurrence [15]. The median follow-up of 12 months is too short to judge efficacy and late side effects. Palliative pelvic radiotherapy, as employed by Abusaris et al. [14], had a reasonable therapeutic ratio. Pain reduction after third radiotherapy for pelvic and extra-pelvic targets combined was seen in 71% of the patients. No patients experienced grade 4 acute toxicity. Less than 10% each reported acute grade 3 pain or dysuria. No high-grade late toxicity was recorded, except for one grade 3 skin reaction.

The bone metastases data support the prescription of a second re-irradiation, even if this recommendation is based on small numbers of patients [17–19]. The Jeremic et al. study employed very low cumulative total doses, maximum 8 + 4+4 Gy, which are not expected to cause any high-grade late toxicity [19]. Most studies of first re-irradiation for bone metastases already utilized far higher cumulative equivalent doses without any safety concerns [22,23]. All three studies of second re-irradiation reported symptom improvement in most patients [17–19]. For spinal metastases, hypofractionated SBRT (25–30 Gy in 4–5 fractions) appears safe and efficacious. No strong recommendation regarding minimum time interval can be made. Individual patients were re-treated after less than 6 months [17,18]. Typically, the spinal cord dose constraints for first re-irradiation were not exceeded in patients re-irradiated twice [24,25]. However, the first and last authors' institutions accept higher cumulative life time doses if the individual patient agrees to the proposed treatment regimen and lower doses cannot be achieved.

The phenomenon of tissue recovery from occult damage after first-line radiotherapy has been well described for several OAR [26–28]. Unfortunately, no such experimental animal studies were performed after re-irradiation. It is therefore intriguing to notice that Abusaris et al. decided to translate the first re-irradiation data to the second re-irradiation setting [14]. In their institution, the maximum dose tolerance was set as 50% more than the normal, first-line constraint. If the interval to the next radiotherapy course was relatively short, i.e., 6–12 months, only 25% more was allowed, and no increase was permitted with intervals shorter than 6 months. The safety of this policy is difficult to judge from the available evidence. However, it might be a starting point from which more, ideally prospectively documented data can be derived. Even in the setting of first re-irradiation, dose constraints are only beginning to emerge [29–31]. The same is true for patient selection criteria and prognostic models [32–34]. Therefore, clinical practice is far from standardized.

Several other limitations of the reviewed studies must also be acknowledged. Different methods were used to estimate the amount of overlap between the three irradiated volumes. Such estimates require advanced methods of image registration to account for long-term anatomical changes, and in the long run also intra- and inter-fraction motion to uncover the true cumulative dose. Most studies had limited follow-up and/or median survival, hampering assessment of long-term toxicity. It has been shown that prospective single-arm and randomized trials are feasible in the re-irradiation setting [23,35–39]. Such trials employing highly conformal techniques are also warranted to strengthen the evidence for second re-irradiations, and might include several different strategies to improve the therapeutic ratio [40–43]. With the development of cumulative dose constraints and practice recommendations regarding suitable minimum time intervals, reluctance to retreat is expected to diminish, and an increasing number of patients will thus be able to benefit from repeated courses of radiotherapy.

Disclosure statement

The authors report no conflicts of interest. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.