In the last decades combined chemoradiation has emerged as the standard of care in patients suffering from primary advanced nasopharyngeal cancer [1–6]. Local control and overall survival have further improved through the introduction of modern radiotherapy techniques which allowed dose escalation up to and beyond 70 Gy in conventional fractionation with an improved toxicity profile [24–28]. However, isolated local recurrence still remains an issue in about 10% of the patients and appears even more challenging to treat than in the past because of the intensified prior treatment in many patients. Currently available treatment options for recurrent nasopharyngeal cancer include surgery , chemotherapy  and various techniques of reirradiation like radiosurgery, fractionated stereotactic radiotherapy (FSRT), brachytherapy using mould or gold grain techniques and fractionated external beam radiotherapy [10–12, 31]. Most of these options, namely surgery, radiosurgery and brachytherapy techniques yield excellent results in highly specialized, experienced centers, but are usually restricted to small volume recurrences confined to the nasopharynx and adjacent soft tissues . In contrast, systemic treatment alone hardly results in long term survivors, and therefore it is usually restricted to patients with metastatic disease as a palliative treatment option [9, 30]. For the remaining patient group, suffering from isolated but locally advanced recurrent lesions, external beam irradiation has been investigated using 2D or 3D treatment techniques, although this approach seems to be difficult due to the large numbers of important structures situated in the vicinity of a region that was already irradiated to a high dose during the primary treatment . However, substantial local control rates have been reported by several groups, especially using modern 3D-RT techniques, but were commonly accompanied by high incidences of severe late toxicity [13–15]. For example, Zheng et al.  reported a 5-year-local control rate of 71% after treating recurrent nasopharyngeal cancer with 3D-RT up to 70 Gy, but grade 3 toxicitiy was found in all patients with half of them suffering from grade 4/5 side effects.
Intensity-modulated radiotherapy has been shown to yield superior dose distribution and sparing of organs at risk compared to 3D-conformal radiotherapy in many sites of the body. In the nasopharyngeal region, Hsiung et al.  also could show superior target coverage with less dose to organs at risk, especially brainstem and eyes, comparing IMRT with 5-field 3D-conformal RT used for boost or salvage irradiation. Excellent local control and overall survival rates beyond 90% have been reported by several groups using IMRT in the treatment of primary nasopharyngeal cancer with acceptable toxicity [24–28]. In our study, with the majority of patients suffering from locally advanced recurrences treated with moderate dose IMRT (median 50.4 Gy), we found a complete response rate of 41%, with 1-and 2-year local control rates of 76% and 69%, respectively. The corresponding overall survival rates decreased from 82% after one year to 44% after two years, probably due to the relatively high rate of distant metastases occurring during the follow-up period. It is possible that this patient group belongs to a subset of patients with primarily unfavourable biological tumor properties including the tendency for early metastasis. The incidence of severe late complications was 29%, which is lower than in most of the studies using 2D- or 3D-conformal radiotherapy [13–15], especially in terms of neurological side effects like temporal lobe necrosis, brain stem damage or cranial neuropathy. But despite the theoretically advantages of IMRT, only few clinical reports exist about the use of IMRT for recurrent nasopharyngeal cancer in the literature. For example Lu et al.  reported about 49 patients treated with high dose IMRT (68-70 Gy) which resulted in a 100% local control rate after a median follow up of 9 months. However, the incidence of late toxicities was not reported probably due to the very short follow-up time, excluding definite conclusions about the influence of late toxicity on the overall outcome. Chua et al.  reported on 31 cases treated with moderate dose IMRT (median dose 54 Gy) with very similar patient characteristics compared to our cohort considering age, gender, rT stage, time interval between the radiation courses and primary treatment. They found similar results, with a complete response rate of 58%, 1-year rates of locoregional control and survival of 56% and 63% respectively and a severe late toxicity rate of 19%, suggesting good short-term control with acceptable incidence of late side effects for this treatment concept. In a recent update, their initial results transferred into a 5-year local control rate of 27%-43% depending on rT stage, indicating reasonable long-term control and survival in a substantial proportion of patients in this unfavourable group with an acceptable toxicity profile .
Several prognostic factors have been discussed for outcome after reirradiation treatment of locally recurrent nasopharyngeal cancer, including age, performance score, histology, rT stage, tumor volume, time interval between radiotherapy courses, prior local failures, reirradiation dose and even EGFR-status [7, 9, 15, 34–39]. As our study included only a small number of patients, conclusions considering prognostic factors should be drawn with caution, given the known limitations of univariate analyses in small cohorts. For example, it cannot be ruled out that differences in outcome according to treatment factors, for example radiation dose or simultaneous application of systemic therapy were biased by a tendency to intensified treatment in less advanced lesions. However, despite the small number of patients in our study we found an impact of rT stage, tumor response, reirradiation dose > 50 Gy and simultaneous use of chemotherapy for local control and/or overall survival, whereas time interval between the RT courses, age, gender and performance score showed no prognostic value in our analysis.
The most consistent prognostic factor being reported is rT stage [7, 9, 15, 35, 36]. Especially patients with invasion of intracranial structures (rT4) had a particularly poor local control and overall survival in most of the series. In our cohort, all local recurrences were seen in patients with rT4 stage resulting in 1- and 2-year local control rates of only 67% and 56% and a 2-year overall survival of only 28%. Chua et al  reported an even more distinguished difference in their cohort with a 1-y-locoregional control rate of only 35% in patients with rT4 stage compared to 100% in rT1-3 stage patients, but in contrast to our study, rT stage had no prognostic value for overall survival in their series. The poor outcome of patients with rT4 stage with respect to local control and overall survival is possibly related to different factors. The decreased local control may be caused by suboptimal target coverage due to the constraints of the nearby critical structures and the tendency to lower overall dose prescription in heavily pre-treated patients with advanced lesions in the fear of excess toxicity. However, although the difference in overall survival could obviously be at least partly attributed to uncontrolled local disease, these patients could also have a higher risk of regional and distant failure per se. In our study only one out of five patients with rT1-3 stage, but 5 out of 12 patients with rT4 stage developed distant metastasis after reirradiation.
The second consistently reported prognostic factor is reirradiation dose [9, 15, 35, 38, 39]. Several authors described improvements in outcome for reirradiation doses above 60 Gy [9, 15, 35], but in some of these series high local control did not transfer into improved survival and was rather accompanied by high rates of severe complications probably responsible at least in part for this difference. For example Zheng et al.  achieved an excellent 5-year local control rate of 71% in a cohort of less advanced lesions, whereas the 5-year overall survival rate was only 40%. Eleven of their 86 patients died without signs of disease progression but showed severe complications. In contrast, Chang et al.  observed that a dose > 50 Gy already yielded better survival in their series. In our cohort, doses above 50 Gy were associated with improved local control and overall survival, but this result could be influenced by a tendency to apply higher doses in patients with less advanced rT stage. However, for patients suffering from rT1-3 tumors, reirradiation doses of 50-60 Gy resulted in excellent short term local control and overall survival in our and other reported series , while in rT4 stage patients doses of about 50 Gy seem to have a palliative value only in most cases. However, whether further dose escalation in those advanced patients will improve outcome remains uncertain. According to our experience, sparing of adjacent organs at risk can be difficult in rT4 patients even with the use of IMRT and further dose escalation would probably distinctly increase late toxicity. One way to improve outcome could be the use of newer radiotherapy techniques like protons or heavy ions, which could allow a superior dose distribution and sparing of normal tissues. For example, Taheri-Kadhoda et al.  showed a superior dose distribution in the nasopharyngeal area with respect to target coverage and dose to organs at risk with 3-field intensity modulated proton therapy compared to 9-field step and shoot photon IMRT.
Another possible way to improve outcome could be combined modality treatment using induction and/or concurrent systemic therapy. In our series, we found a trend to improved local control and a significant improved overall survival in patients receiving concurrent systemic treatment. Another series reported by Chua et al.  showed a 1-year local control rate of 75% in advanced recurrences after induction chemotherapy with cisplatin/gemcitabine followed by reirradiation with IMRT. While induction chemotherapy could not only allow better target coverage and sparing of adjacent organs at risk through tumor shrinkage, it may also be used to delay the second course of RT in patients who relapse shortly after primary treatment . Concurrent chemotherapy has been shown to improve response, local control and overall survival compared to radiotherapy alone in the treatment of primary nasopharyngeal cancer in many series and is now accepted as the standard of care for advanced primary lesions [1–6]. In recurrent nasopharyngeal cancer it could also improve the rate of complete responses, which had a significant impact on local control and overall survival in our series. Patients with complete response showed a 2-year local control rate of 100% and a 2-year OS rate of 86% compared to 45% and 12% in the group with residual disease. Therefore combined modality approaches including induction and/or concurrent systemic treatment could not only be used to further improve outcome especially in advanced (rT4) recurrent lesions but also for patient selection processes.
In conclusion, reirradiation using IMRT for local recurrences of nasopharyngeal cancer with moderate doses yields excellent results in terms of local control and overall survival in rT1-3 lesions. Acute and late toxicity seems to be reduced compared to the results published with 2D- or 3D conformal treatment approaches. However, outcome in locally advanced recurrent lesions (rT4) is still limited and treatment with doses in the range of 50 Gy has to be considered palliative in the majority of cases. Therefore close monitoring of patients after primary treatment of nasopharyngeal cancer should be mandatory in order to detect local recurrences early enough to offer salvage options with curative intent. Combined modality approaches or newer radiation technologies like protons or heavy ions should be further investigated especially considering advanced recurrent lesions.