Reirradiation with intensity-modulated radiotherapy for locally recurrent T3 to T4 nasopharyngeal carcinoma

Transcription

1 ORIGINAL ARTICLE Reirradiation with intensity-modulated radiotherapy for locally recurrent T3 to T4 nasopharyngeal carcinoma Oscar S. H. Chan, FRCR, 1 Henry C. K. Sze, FRCR, 1 Michael C. H. Lee, PhD, 2 Lucy L. K. Chan, BSc, 1 Amy T. Y. Chang, FRCR, 1 Sarah W. M. Lee, FRCR, 1 Wai Man Hung, MSc, 1 Anne W. M. Lee, MD, 3 Wai Tong Ng, MD 1 * 1 Department of Clinical Oncology, Pamela Youde Nethersole Eastern Hospital, Hong Kong, 2 Department of Medical Physics, Pamela Youde Nethersole Eastern Hospital, Hong Kong, 3 Department of Clinical Oncology, University of Hong Kong, Hong Kong. Accepted 21 October 2016 Published online 29 November 2016 in Wiley Online Library (wileyonlinelibrary.com). DOI /hed ABSTRACT: Background. The purpose of this study was to assess the efficacy and toxicities of reirradiation using intensity-modulated radiotherapy (IMRT) in patients with locally advanced recurrent nasopharyngeal carcinoma (NPC). Methods. Thirty-eight patients with consecutive rt3 to rt4 NPC treated between 2005 and 2013 were retrospectively analyzed. Results. The 3-year overall survival (OS), progression-free survival (PFS), and local control rate were 47.2%, 17.5%, and 44.3%, respectively. Gross target volume (GTV) D 95, GTV D 50, and age were all important prognostic factors for OS and PFS, but only GTV D 95 was an important determinant for local control. A total of 73.7% patients experienced 1 grade 3 late toxicities and 3 patients died of massive epistaxis. Temporal lobe necrosis (TLN) developed sooner with a higher total biological equivalent dose. Conclusion. Adequate tumor dose coverage was important for treating rt3 to rt4 NPC. Although late complications were common, treatment-related mortality was solely vascular in nature. Dose constraints of neurologic structures for reirradiation should be revised with the latest information on late toxicities. VC 2016 Wiley Periodicals, Inc. Head Neck 39: , 2017 KEY WORDS: reirradiation, local recurrence, nasopharyngeal carcinoma, intensity-modulated radiotherapy, toxicities INTRODUCTION With the advancement in treatment strategy and application of intensity-modulated radiotherapy (IMRT), local recurrence is no longer a major failure after the primary treatment of nasopharyngeal carcinoma (NPC). However, it inevitably happens in some patients, especially those with advanced disease, and remains a very challenging problem to tackle. In modern series using IMRT, the reported local failure rate is 5% to 15% overall, and is much higher at 15% to 45% for T4 disease. 1 7 Aggressive salvage treatment with a curative intent is advocated for locally recurrent NPC because many patients could still achieve long-term survival. 8 Both nasopharyngectomy and reirradiation are the mainstay of therapy, but surgical salvage is only applicable for early-stage recurrence and rt3 to rt4 diseases are not suitable candidates. For advanced local-recurrence, reirradiation is the major treatment option and might represent a chance of cure, although it poses a much higher risk of late complications. 9 Because very advanced treatment modality with promising potential, like proton therapy, is yet to be explored and often not available, IMRT remains the best possible mode of reirradiation for locally recurrent NPC. Previous studies *Corresponding author: W. T. Ng, Department of Clinical Oncology, Pamela Youde Nethersole Eastern Hospital, 3 Lok Man Road, Chai Wan, Hong Kong. ngwt1@ha.org.hk of IMRT often included patients with potentially resectable rt1 to rt2 diseases, but its actual efficacy in treating advanced rt3 to rt4 diseases is not known. This study reports the treatment outcome for locally advanced rt3 to rt4 diseases using IMRT and attempts to identify key factors affecting treatment outcomes as well as the occurrence of potentially serious neurologic complications. MATERIALS AND METHODS Patient selection All patients with suspected local recurrences would undergo fiberoptic nasopharyngoscopy and biopsy. For those without histological confirmation, the diagnosis of recurrence was made by unequivocal radiological progression in MRI. This would be further supplemented by positron emission tomography/ct (PET/CT) and Epstein Barr virus-dna in uncertain situations. Consensus would be sought for dubious findings in multidisciplinary meetings with diagnostic radiologists and nuclear medicine specialists. Those confirmed recurrences would undergo, apart from MRI of the nasopharynx and neck, CT of the thorax and abdomen, as well as a bone scan as staging workup. PET/CT was recommended but not mandatory. Patients without distant failure would be considered for curative treatment (see Figure 1). Those patients with rt1 to rt2 disease would be assessed by surgeons for nasopharyngectomy. For patients who refused surgery or had inoperable disease, reirradiation HEAD & NECK DOI /HED MARCH

2 CHAN ET AL. FIGURE 1. Patients profile showing the number of patients in different groups. RT, radiotherapy. with IMRT would be given if they were medically fit and free from severe (grade 3) late radiotherapy (RT) complications (except xerostomia and hearing loss). If organ functions allowed, 4 to 6 cycles of cisplatin-based induction chemotherapy was usually given before reirradiation to try to downsize the tumor. This study included all the patients with rt3 to rt4 NPC who received IMRT for reirradiation according to the above criteria between January 2005 and December 2013 (see Figure 1). Approval from the institutional ethics committee for a retrospective chart review was obtained. Radiotherapy planning and intensity-modulated radiotherapy technique Patients were immobilized with thermoplastic casts from head to shoulder while in the supine position. A contrast planning CT scan was performed from vertex to 2-cm below the clavicle at 3-mm slice intervals. Diagnostic MRIs and PET/CTs (if available) would be fused with the planning CT images for gross tumor volume (GTV) delineation. No clinical target volume was delineated and the planning target volume (PTV) was produced by expanding the GTV with a 10-mm margin. A tighter margin (minimum 2 mm) would be used if the GTV was in close contact with vital neurologic structures. Organs-at-risk (OAR), including the brain stem, temporal lobe, and optic chiasm, were also contoured. Planning organ at risk volumes were produced for the spinal cord (3-mm margin), temporal lobe, brain stem, and optic chiasm (all 1-mm margins). The PTV would be prescribed 64.8 Gy (1.2 Gy b.i.d.; biologic effective dose [BED] calculated at a/b 510 Gy; ie, BED Gy), 60 Gy, or 50 Gy (both in 2 Gy daily fractions, BED and 60 Gy, respectively) upon the oncologists discretions and availability of Linac time (BED 5 D(11d/(a/b)), where D 5 total dose and d 5 fractional dose). Inverse planning was performed using the Varian Eclipse treatment planning system (Varian Medical Systems, Palo Alto, CA). The objective was to give at least 95% prescribed dose to the entire PTV as long as the OAR constraint permitted. The critical OAR (mainly neurologic structures) were allowed to receive a cumulative lifetime physical dose up to 130% of the single-course maximum tolerance dose. The dose limits of most important OAR are shown in Table 1. The optic chiasm, brain stem, and spinal cord had the highest priority, followed by the GTV and then other less critical OAR, such as temporal lobes and optic nerves. The treatment usually comprised of 8 to 10 coplanar fields, supplemented with another 1 to 2 noncoplanar vertex fields when necessary, delivered with 6 MV photon beams modulated by the sliding window technique with the Varian 120-leaf Millennium multileaf collimation. Setup accuracy was ascertained by weekly megavoltage portal imaging before This was changed to daily on-board kv imaging afterward. Follow-up assessment Patients were assessed weekly during the course of RT, and then every 3 months after RT completion in the first 2 years. Endoscopy was routinely performed and biopsy would be taken whenever necessary. MRI was performed 12 weeks after RT completion and then at least yearly. Patients would also be assessed by otorhinolaryngologists and dental surgeons at regular intervals. Major late toxicities were scored according to the Common Terminology Criteria for Adverse Events version 3.0. TABLE 1. Lifetime physical dose limits of most important organs at risk for re-treatment radiotherapy planning. OAR Absolute dose limits Brainstem Spinal cord Optic chiasm Desirable dose limits Optic nerve Temporal lobe Brachial plexus Abbreviation: OAR, organs at risk. Dose limits D 1% 78 Gy D 1cc 65 Gy 78 Gy 78 Gy D 1cc 84.5 Gy D 1cc 85.8 Gy 534 HEAD & NECK DOI /HED MARCH 2017

3 INTENSITY-MODULATED RADIOTHERAPY FOR RECURRENT NPC TABLE 2. Demographics Demographics of individual treatment subgroups. Data analysis and statistical method No. of patients 5 38 Median age, y (range) 50 (37 78) Sex M:F 31:7 T classification on first diagnosis T2 7 (18.4%) T3 21 (55.3%) T4 10 (26.3%) N classification on first diagnosis N0 2 (5.3%) N1 8 (21.1%) N2 23 (60.5%) N3 5 (13.1%) Group stage on first diagnosis II 4 (10.5%) III 20 (52.6%) IVA/IVB 14 (36.9%) RT technique in first course treatment 2D 2 (5.3%) 3D 13 (34.2%) IMRT 23 (60.5%) Induction/adjuvant chemotherapy 30 (79%) in first treatment course Concurrent chemotherapy in first 28 (74%) treatment course Median time from diagnosis to 4.2 y ( ) first relapse (range) Performance status at the time of relapse 0 3 (7.9%) 1 31 (81.6%) 2 4 (10.5%) rt classification, UICC/AJCC 7th edition* rt3 23 (60.5%) rt4 15 (39.5%) rn classification, UICC/AJCC 7th edition* rn0 25 (65.8%) rn1 11 (28.9%) rn2 1 (2.6%) rn3 1 (2.6%) Recurrent group stage, UICC/AJCC 7th edition* III 23 (60.5%) IVA/IVB 15 (39.5%) Induction chemotherapy scheme before second course of RT PF/JF 10 (26.3%) PG/JG 16 (42.1%) TPF 8 (21.1%) PPe 1 (2.6%) No chemotherapy 3 (7.9%) Fractionation Hyperfractionation 13 (34.2%) Conventional fractionation 25 (65.8%) Median total dose in EQD2 (range) 128 ( ) Abbreviations: RT, radiotherapy; IMRT, intensity-modulated radiotherapy; UICC/AJCC, Union for International Cancer Control/American Joint Committee on Cancer; P, cisplatin; F, 5-fluorouracil; J, carboplatin; G, gemcitabine; T, docetaxel; Pe, pemetrexed; EQD, equivalent dose in 2 Gy fractions. * Patients treated before 2010 were restaged according to the UICC/AJCC 7th edition. The data cutoff date was October 9, The time to locoregional progression, distant metastases, and death was measured from the date of commencement of any therapy (either chemotherapy or RT). The time to RTrelated late toxicity developments was measured from the date of reirradiation commencement. The competing risk method is used to estimate the rate of local control and regional control, whereas actuarial estimates of the progression-free survival (PFS) and overall survival (OS) were obtained by Kaplan Meier method. Univariate analyses were conducted using log-rank tests on age, sex, rt classification, rn classification, tumor volume, median primary tumor dose (GTV D 50 ), dose coverage (by GTV D 95 ), fractionation, and time to relapse to study their significance on OS, PFS, and local control rates. The association of late toxicities and fractionation scheme (conventional vs b.i.d. hyperfractionation) was studied using the Pearson chi-square test. Apart from hearing loss, temporal lobe necrosis (TLN) was an important radiation-induced neurologic toxicity observed in our cohort. The anatomic position of temporal lobes and their proximity to the primary tumor made them very difficult to spare; and they were given lower priority in treatment planning as TLN is relatively less catastrophic than an injury to the brainstem, spinal cord, or optic chiasm. However, as TLN can significantly affect patients quality of life (QOL), we looked at the pattern of TLN development, both symptomatic and revealed only by MRI, with total BED, time gap between the first and re-treatment RT, and the time to TLN development to study their relationships. RESULTS Between January 2005 and December 2013, local recurrences were detected in 85 patients (see Figure 1). This study included the 38 patients who had rt3 to rt4 diseases and received reirradiation with IMRT. Their median time from first diagnosis to recurrence was 4.2 years. Hyperfractionation with 1.2 Gy b.i.d. to a total of 64.8 Gy was used in 13 patients and the remaining patients received 50 to 60 Gy in conventional 2 Gy per daily fraction. Thirty-five patients (92.1%) received induction chemotherapy before reirradiation and 8 of them also received concurrent chemotherapy. The demographics of the 38 patients are tabulated in Table 2. Dosimetric data Dosimetric data for rt3 and rt4 stages are shown in Table 3. Underdosage in most of the treatment plans was inevitable because of the prior RT exposure and the close proximity to the critical neurologic structures. The degree of underdosage (reflected by GTV D 95 ) was more severe with rt4 disease. On average, D 95 was around 80% of the intended dose for rt3, but only amounted to <60% of the intended dose for rt4. Outcome after reirradiation Of the 23 patients with rt3 and the 15 patients with rt4, 15 of the patients with rt3 (65.2%) and 8 of the patients with rt4 (53.3%) had further local tumor progression after reirradiation. Distant metastases occurred in 8 patients (21.1%). For the 13 patients with recurrent nodal disease, 6 of them had no further nodal tumor progression after reirradiation. At the time of analysis, 16 patients were still alive and their median follow-up time was 47.8 months (range, months). The 3-year HEAD & NECK DOI /HED MARCH

4 CHAN ET AL. TABLE 3. Summary of dosimetric data (average and range) according to recurrent T classification. rt3 rt4 p value Intended dose, Gy No. of patients GTV, cc 29.5 ( ) 55.3 ( ).03 GTV D 95, in BED 60.3 ( ) 38.9 ( ).001 GTV D 50, in BED 73.3 ( ) 64.3 ( ) <.001 GTV D min, in BED 40.2 ( ) 17.6 ( ).001 Abbreviations: Gy, Gray; GTV, gross tumor volume; GTV D 95, dose received by 95% of the GTV; BED, biologically effective dose; GTV D 50, dose received by 50% of the GTV; GTV D min, minimum dose to the GTV. OS, PFS, local control, and regional control rates for the whole group were 47.5%, 17.5%, 44.3%, and 76.4%, respectively (see Figure 2). Results for the rt3 and rt4 subgroups are also shown in Figure 2. Prognostic factors The results of univariate analyses are shown in Table 4. Among all the factors, tumor dose coverage, as reflected by GTV D 95, was the only important determinant in local control rate, whereas GTV D 95, GTV D 50, and age were all important prognostic factors for OS and PFS. Major late toxicities associated with reirradiation There was no visual loss or spinal cord/brainstem damage in our series, but other severe late toxicities that impaired patients QOL and normal organ functions were common. Twenty-eight patients (73.7%) experienced at least 1 grade 3 late toxicity, and 18 (47.4%) of them had 2 or more grade 3 late toxicities by the third year post- RT (more details in Table 5). At the time of analysis, only 2 of the 16 patients still alive were free of any late grade 3 complications. The actuarial rates of developing grade 3 hearing loss, TLN, soft tissue necrosis, dysphagia, massive nasal bleeding (including pseudoaneurysm formation), and trismus at the third year post-rt were 47.5%, 24.2%, 23.1%, 24.0%, 20.2%, and 10.2%, respectively. Three patients (7.9%) died of massive epistaxis attributed to reirradiation. Logistic regression did not reveal any statistical association between the fractionation scheme (conventional vs hyperfractionation) and the development of grade 3 toxicities. The hazard ratio was 6.75 (range, ; p 5.09). Massive nasal bleeding Of the 6 patients with massive nasal bleeding, 4 underwent angiography, whereas 2 patients died upon arrival to the hospital. The bleeding source of 1 patient cannot be confirmed even after angiography. For the remaining 3 patients, the bleeding sites were all located at the upper cervical internal carotid artery and petrous internal carotid artery. One bleeding site was inside and one was abutting the GTV. The risk of these toxicities is considered unavoidable. Pattern of temporal lobe necrosis The lifetime BED_R 1cc to the temporal lobe was calculated for this analysis. The a/b ratio was 2.5 Gy, a correction FIGURE 2. Treatment outcomes for the whole group (A) and according to the recurrent T classification (B); 1 5 overall survival (OS), 2 5 progression-free survival (PFS), 3 5 local control rate, 4 5 regional control rate. 536 HEAD & NECK DOI /HED MARCH 2017

5 INTENSITY-MODULATED RADIOTHERAPY FOR RECURRENT NPC TABLE 4. Univariate analyses of selected prognostic factors on treatment outcomes. OS p value PFS p value local control rate p value Age, y 1.07 ( ) ( ) ( ).30 Sex, F vs M 1.26 ( ) ( ) ( ).40 rt classification, rt4 vs rt ( ) ( ) ( ).81 rn classification negative vs positive 1.07 ( ) ( ).025 Not tested Intended tumor dose, 50 vs 60 Gy 2.25 ( ) ( ) ( ).44 GTV 1.01 ( ) ( ) ( ).71 GTV D 95, in BED 0.98 ( ) ( ) ( ).008 GTV D 50, in BED 0.94 ( ) ( ) ( ).10 GTV D min, in BED 0.98 ( ) ( ) ( ).76 Hyperfractionation vs 0.69 ( ) ( ) ( ).53 conventional fractionation Time to relapse, <4yvs4 y 1.06 ( ) ( ) ( ).55 Abbreviations: OS, overall survival; PFS, progression-free survival; GTV, gross tumor volume; GTV D 95, dose received by 95% of the GTV; BED, biologically effective dose; GTV D 50, dose received by 50% of the GTV; GTV D min, minimum dose to the GTV. factor h m was applied for hyperfractionation to account for the incomplete sublethal damage between 2 daily fractions, this corresponded to a repair half-life of 4 hours. The maximum dose contributed to the relevant temporal lobe region from the primary course was taken to be 70 Gy, the full tumor dose, in 35 daily fractions because of the tumor proximity to the temporal lobes and the unavailability of 3D primary course dose data (some delivered over 20 years ago). This amounted to a BED of 126 Gy from the first course. A total of 4 TLNs were observed in 4 patients, 2 of them received hyperfractionation reirradiation, and 3 of them had TLNs on the right side. The scatter plot of BED_R 1cc against the time gap between the 2 treatment courses is shown in Figure 3A, with the left and right temporal lobes plotted as separate data points. The BED from reirradiation ranged from 7.4 Gy to Gy, and BED_R 1cc ranged from Gy to Gy. The minimum treatment gap was 10 months in our cohort, and no relationship was observed between the treatment gap and onset BED_R 1cc for TLN. BED_R 1cc versus the time of TLN development (or last follow-up) after reirradiation was plotted in Figure 3B. There seems to be an inverse relationship between BED_R 1cc and the time of TLN manifestation. Those who received a BED_R 1cc of 220 to 250 Gy all had TLN in the third year after reirradiation. The patient who received 164 Gy had TLN in the sixth year; however, no TLN was observed for less than 150 Gy BED_R 1cc up to even the ninth year after reirradiation. TABLE 5. Major late complications. Late complications, grade 3 rt3 rt4 (No. of patients 5 23) (No. of patients 5 15) Hearing loss 10 (43.5%) 7 (46.7%) TLN 3 (13.0%) 1 (6.7%) Soft tissue necrosis 5 (21.7%) 3 (20.0%) Dysphagia 6 (26.1%) 4 (26.7%) Massive nasal bleeding 4 (17.4%) 2 (13.3%) Trismus 3 (13.0%) 0 (0.0%) Abbreviation: TLN, temporal lobe necrosis. DISCUSSION Most of the previous studies on NPC reirradiation by IMRT included both early and advanced rt diseases, and demonstrated that rt classification is a most important prognostic factors. However, salvage surgery is a major treatment option for patients with operable earlystage disease to avoid cumulative morbidities from 2 RT courses. A recent study by You et al 17 suggested that salvage endoscopic nasopharyngectomy yielded better survival and QOL benefits while minimizing treatmentrelated complications for limited recurrent disease when compared to IMRT. By excluding early-stage recurrences, this series focused on the effectiveness of reirradiation to the patients for whom reirradiation was the only feasible option of cure. The intensity of primary treatment that our cohort had received is remarkable. The majority of them had 3D conformal RT or IMRT for the primary treatment and >70% of them had concurrent chemotherapy during the primary RT. Tumor clones that survived after such intensive therapy are probably more radioresistant or chemoresistant intrinsically. This phenomenon has been demonstrated in other head and neck cancers. Choe et al 18 reported a series of 166 patients with recurrent head and neck cancer treated by concurrent chemotherapy reirradiation and demonstrated that survival was reduced significantly if patients received chemotherapy irradiation previously (2-year OS rate: 10.8% vs 28.4%). Despite all these difficulties, about 40% of our patients with rt3 to rt4 disease could have their local disease under control after reirradiation. By eliminating patients with early-stage recurrent diseases, we found that rt classification and GTV were no longer prognostically significant in OS, PFS, and local control rate. This is different from what previous studies have suggested. Tian et al 19 showed that GTV >26 cm 3 led to a lower OS (5-year OS rate 16% vs 61%; p 5.05), similarly for GTV 19 cm 3 in the series by Karam et al 16 (5-year OS rate 35% vs 88%; p 5.07; local control rate 31% vs 75%; p 5.01). In the study by Xiao et al 20 on 291 patients with locally recurrent NPC, the 5-year OS rates were 63.1% and 20.8% for tumor volume <22 cm 3 and 22 cm 3, respectively, the distant metastasis rate was also higher, and there were more radiation-induced HEAD & NECK DOI /HED MARCH

6 CHAN ET AL. FIGURE 3. (A) Scatter plot of the maximum total biologic equivalent dose (BED) to the temporal lobe and the time gap between the first and retreatment radiotherapy, a longer time gap did not lead to a higher total tolerable dose. (B) Scatter plot of maximum total BED to temporal lobe and the time to temporal lobe necrosis (TLN) development (or last follow-up) showing an inverse relationship between the two. toxicities that led to half of the deaths. The most distinct prognostic factor in our series was tumor dose coverage as reflected by GTV D 95, which was the only important determinant in local control rate, whereas GTV D 95, GTV D 50 (median dose), and age were important prognostic factors for OS and PFS. With the current prescription level, a better treatment outcome can be achieved with a better target dose coverage (a higher D 95 ), as long as the OAR complication is tolerable. However, one should be cautious that the lack of prognostic significance in the GTV and rt classification in our series may be due to the small sample size or a plateau of prognostic effect beyond a certain GTV threshold. The current study on locally advanced recurrent NPC is limited by both the small number of patients and the heterogeneity in the prescription level and fractionation scheme. This could have negatively impacted the statistical significance of the tests performed. Only univariate analysis was performed because of the number of patients studied. Besides, important issues like the effect of chemotherapy and fractionation scheme cannot be elucidated. QOL data were also not collected for this study because of its retrospective nature. Despite the use of IMRT, the risk of any grade 3 complications was over 70% and almost all patients still alive at the time of analysis had some grade 3 or more late complications. These toxicities were inevitable because the tumor was encasing, embedded in, or surrounded by the normal structures. Although the rates of late toxicity in our cohort were comparable with other reirradiation series, 9 one must understand the complication rates in all similar studies could be underestimated as 538 HEAD & NECK DOI /HED MARCH 2017

7 INTENSITY-MODULATED RADIOTHERAPY FOR RECURRENT NPC many late toxicities can take years to develop but a significant proportion of patients died relatively early after reirradiation. The 8% treatment mortality rate in our series was, however, relatively low compared with the 9% to 35% reported in many other IMRT series. 11,19,21,22 Because all our mortality was vascular in nature, there is a paramount need to study the vascular 23 and mucosal tolerance, their complication detection, and to develop ways to tackle or prevent such tragic events. The pattern and magnitude of organ recovery after RT remains unclear, making it very difficult to determine the tolerance dose for the second RT course. Various sets of dose constraints have been mentioned in the literature for important central nervous system (CNS) structures, including the brainstem, spinal cord, optic chiasm, and temporal lobes. In the series by Chua et al, 12 the constraints (dose limit and volume about limit) to critical structures were: 10 Gy and 10% for the brainstem and temporal lobe; 4 Gy and 5% for the spinal cord; and 8 Gy and 5% for the optic nerve and chiasm. Qiu et al 21 reported the use of much more generous dose constraints: 50 Gy for the brainstem and temporal lobes; 40 Gy for the spinal cord; and 54 Gy for the optic nerve and optic chiasm, regardless of the dose delivered in the first RT course. In the series by Tian et al, 24 50% dose tolerance recovery of the CNS structures was assumed if the treatment gap was at least 1 year. A recent study by Jones and Grant 25 proposed a formula, taking into account the treatment gap, to estimate the remaining tolerance dose. In our series, a cumulative lifetime physical dose up to 130% of the single course maximum tolerance dose of critical OAR (ie, 30% recovery from the first RT course) was allowed and seemed to be safe. Despite such a wide variation in dose constraints used by many groups, severe late toxicities with CNS structures were rather uncommon. The development of more accurate methods incorporating relevant factors, like treatment gap, dose, volume, etc., to determine the remaining tolerable dose for re-treatment RT is definitely warranted to strike the best balance between achieving the tumor dose coverage and the risk of late toxicities. This study revealed that our treatment-related morbidity rates were comparable to many others, and that there were no late toxicities with the brain stem, spinal cord, and optic chiasm in our cohort. It seemed that our current CNS dose constraints of the lifetime physical dose at 130% of the single-course maximum dose might be a bit too conservative. The use of lifetime BED at 130% of the single-course maximum BED, as suggested by Lee et al, 9,26 will be explored to strike a better balance between treatment performance and risk of complications. For the temporal lobes, our pattern suggested that a lifetime BED 1cc of 150 Gy might be safe and a longer gap between the first and re-treatment RT did not lead to a higher tolerance dose, concurring with the observation by Mayer and Sminia. 27 Although there were only 4 data points with TLN, all other data points without TLN also contributed to this TLN development pattern and its significance should not be underestimated. A more accurate picture will emerge in a few years with more follow-up data, especially the 2 temporal lobes that received 175 Gy and 210 Gy lifetime BED 1cc. With the current data, we postulate that a lifetime BED 1cc of 200 Gy should be acceptable after balancing the risk of TLN and tumor dose coverage. Assuming a full 70 Gy physical dose has been delivered in the primary RT course, the tolerable reirradiation BED 1cc will then be 74 Gy, equivalent to a 45.9 Gy physical dose in 30 fractions. By using less conservative OAR dose constraints, a better target dose coverage could be achieved with the potential to improve cure. We also believe that other advanced RT modalities with better dose-sculpting capability should also be explored. In particular, the application of intensity-modulated proton therapy is being actively studied, as some dosimetric and clinical studies have shown its potential in reirradiation CONCLUSION The use of IMRT in treating patients with recurrent rt3 to rt4 NPC could achieve reasonable local control and a certain proportion of these patients could enjoy long-term survival. Adequate tumor dose coverage is crucial for a favorable treatment outcome. Severe late complications after reirradiation were common but the rate of CNS toxicity and treatment-related deaths were acceptable. Neurologic dose constraints for reirradiation should be cautiously reviewed and revised based on clinical data to balance the tumor dose coverage and risk of late toxicities. REFERENCES 1. Lee AW, Ng WT, Chan LL, et al. Evolution of treatment for nasopharyngeal cancer success and setback in the intensity-modulated radiotherapy era. Radiother Oncol 2014;110: Sun X, Su S, Chen C, et al. Long-term outcomes of intensity-modulated radiotherapy for 868 patients with nasopharyngeal carcinoma: an analysis of survival and treatment toxicities. Radiother Oncol 2014;110: Setton J, Han J, Kannarunimit D, et al. Long-term patterns of relapse and survival following definitive intensity-modulated radiotherapy for nonendemic nasopharyngeal carcinoma. Oral Oncol 2016;53: Ng WT, Lee MC, Chang AT, et al. The impact of dosimetric inadequacy on treatment outcome of nasopharyngeal carcinoma with IMRT. Oral Oncol 2014;50: Lee AW, Ma BB, Ng WT, Chan AT. Management of nasopharyngeal carcinoma: current practice and future perspective. JClinOncol2015;33: Kong FF, Ying H, Du CR, Huang S, Zhou JJ, Hu CS. Effectiveness and toxicities of intensity-modulated radiation therapy for patients with T4 nasopharyngeal carcinoma. PLoS One 2014;9:e Cao CN, Luo JW, Gao L, et al. Clinical outcomes and patterns of failure after intensity-modulated radiotherapy for T4 nasopharyngeal carcinoma. Oral Oncol 2013;49: Yu KH, Leung SF, Tung SY, et al. Survival outcome of patients with nasopharyngeal carcinoma with first local failure: a study by the Hong Kong Nasopharyngeal Carcinoma Study Group. Head Neck 2005;27: Lee AW, Fee WE Jr, Ng WT, Chan LK. Nasopharyngeal carcinoma: salvage of local recurrence. Oral Oncol 2012;48: Roeder F, Zwicker F, Saleh Ebrahimi L, et al. Intensity modulated or fractionated stereotactic reirradiation in patients with recurrent nasopharyngeal cancer. Radiat Oncol 2011;6: Han F, Zhao C, Huang SM, et al. Long-term outcomes and prognostic factors of re-irradiation for locally recurrent nasopharyngeal carcinoma using intensity-modulated radiotherapy. Clin Oncol (R Coll Radiol) 2012;24: Chua DT, Sham JS, Leung LH, Au GK. Re-irradiation of nasopharyngeal carcinoma with intensity-modulated radiotherapy. Radiother Oncol 2005; 77: Lu TX, Mai WY, Teh BS, et al. Initial experience using intensitymodulated radiotherapy for recurrent nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2004;58: Tian YM, Tian YH, Zeng L, et al. Prognostic model for survival of local recurrent nasopharyngeal carcinoma with intensity-modulated radiotherapy. Br J Cancer 2014;110: Hua YJ, Han F, Lu LX, et al. Long-term treatment outcome of recurrent nasopharyngeal carcinoma treated with salvage intensity modulated radiotherapy. Eur J Cancer 2012;48: HEAD & NECK DOI /HED MARCH

8 CHAN ET AL. 16. Karam I, Huang SH, McNiven A, et al. Outcomes after reirradiation for recurrent nasopharyngeal carcinoma: North American experience. Head Neck 2016;38 Suppl 1:E1102 E You R, Zou X, Hua YJ, et al. Salvage endoscopic nasopharyngectomy is superior to intensity-modulated radiation therapy for local recurrence of selected T1 T3 nasopharyngeal carcinoma a case-matched comparison. Radiother Oncol 2015;115: Choe KS, Haraf DJ, Solanki A, et al. Prior chemoradiotherapy adversely impacts outcomes of recurrent and second primary head and neck cancer treated with concurrent chemotherapy and reirradiation. Cancer 2011;117: Tian YM, Zhao C, Guo Y, et al. Effect of total dose and fraction size on survival of patients with locally recurrent nasopharyngeal carcinoma treated with intensity-modulated radiotherapy: a phase 2, single-center, randomized controlled trial. Cancer 2014;120: Xiao W, Liu S, Tian Y, et al. Prognostic significance of tumor volume in locally recurrent nasopharyngeal carcinoma treated with salvage intensitymodulated radiotherapy. PLoS One 2015;10:e Qiu S, Lin S, Tham IW, Pan J, Lu J, Lu JJ. Intensity-modulated radiation therapy in the salvage of locally recurrent nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2012;83: Chen HY, Ma XM, Ye M, Hou YL, Xie HY, Bai YR. Effectiveness and toxicities of intensity-modulated radiotherapy for patients with locally recurrent nasopharyngeal carcinoma. PLoS One 2013;8:e McDonald MW, Moore MG, Johnstone PA. Risk of carotid blowout after reirradiation of the head and neck: a systematic review. Int J Radiat Oncol Biol Phys 2012;82: Tian YM, Guan Y, Xiao WW, et al. Long-term survival and late complications in intensity-modulated radiotherapy of locally recurrent T1 to T2 nasopharyngeal carcinoma. Head Neck 2016;38: Jones B, Grant W. Retreatment of central nervous system tumours. Clin Oncol (R Coll Radiol) 2014;26: Lee AW, Foo W, Law SC, et al. Total biological effect on late reactive tissues following reirradiation for recurrent nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2000;46: Mayer R, Sminia P. Reirradiation tolerance of the human brain. Int J Radiat Oncol Biol Phys 2008;70: Liu SW, Li JM, Chang JY, et al. A treatment planning comparison between proton beam therapy and intensity-modulated x-ray therapy for recurrent nasopharyngeal carcinoma. J Xray Sci Technol 2010;18: Lin R, Slater JD, Yonemoto LT, et al. Nasopharyngeal carcinoma: repeat treatment with conformal proton therapy dose-volume histogram analysis. Radiology 1999;213: Lewis GD, Holliday EB, Kocak Uzel E, et al. Intensity-modulated proton therapy for nasopharyngeal carcinoma: decreased radiation dose to normal structures and encouraging clinical outcomes. Head Neck 2016;38 Suppl 1: E1886 E HEAD & NECK DOI /HED MARCH 2017