Introduction

Nasopharyngeal carcinoma (NPC) is a common malignant tumor in China. With the development of modern imaging and irradiation techniques, the 5-year overall survival (OS) of newly diagnosed NPC patients without distant metastasis has reached 75% after conventional external beam radiotherapy (EBRT) in Asian patients [1]. However, 10-36% of patients had local recurrence [2,3]. EBRT, brachytherapy and stereotactic radiosurgery have been used to re-irradiate locally recurrent disease. Each method has its optimal treatment population [4]. Early data have suggested that the use of IMRT for retreatment of local NPC recurrence is clinically feasible and could produce acceptable disease control. However, treatment-induced adverse effects from re-irradiation with IMRT are not negligible, some of which are quite severe. The aim of this report is to assess the efficacy and late toxicities of 54 patients with locally recurrent NPC treated with a uniform IMRT technique.

Material and Methods

Results

2.2: Toxicities

Acute adverse events: ≥Grade Ⅲ mucositis were observed in 21 patients (38.9%), and 23 patients (42.6%) had GradeⅠ or Ⅱ mucositis.

Late adverse events: The common late normal tissue effects observed before re-irradiation included Grade I xerostomia with the incidence rate of 15.2%, Grade I hearing deficit with the incidence rate of 22.2%, trismus with the incidence rate of 16.7%. After re-irradiation, 33 patients (61.1%) had Grade II trismus, 20 patients (37.4%) had Grade I or II cranial nerve palsy, two patients (3.7%) with Grade Ⅲ cranial nerve palsy; 26 patients (48.1%) had Grade I hearing deficit; 10 patients (18.6%) had Grade II hypomnesis; one patient (1.9%) had Grade I hypopsia.

No SLAE was observed before re-irradiation, after re-irradiation 26 patients (48.1%) developed SLAE, including 17 patients (31.5%) with ulcer or necrosis of the nasopharyngeal mucosa, 11 patients (20.4%) with difficulty in feeding, ten patients (18.5%) with temporal lobe necrosis (TLN), six patients (11.1%) with massive hemorrhage (Figure 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. One patient who had haemorraghe after reirradiation: (a) showed the treatment plan for a patient with recurrent foci invading the poststyloid space and so on, (b) showed the CT scan four months after IMRT of the necrosis of nasopharyngeal mucose, and a cavity continuing from the pharynx was formed after the tissue was expelled, (c) the patient had several attack of haemorraghe, DSA showed the right lingual artery pseudoaneurysm (arrow) five months after IMRT, (d) the patient again had massive haemorraghe six months after IMRT, DSA showed pseudoaneurysm in right internal carotid artery (arrow head), coil (arrow) placed at the lingual artery level for (c).

https://doi.org/10.1371/journal.pone.0073918.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Kaplan-Meier curves showing local progression-free survival (a), overall survival (b)

https://doi.org/10.1371/journal.pone.0073918.g002

Discussion

IMRT techniques are able to deliver optimal dose distributions to the tumor targets and adjacent sensitive organs, thereby offered the prospect of improving therapeutic ratios. The use of IMRT for retreatment of local NPC recurrence is clinically feasible, and was used in several studies [6–10], all of which yielded acceptable disease control. In the initial IMRT study for locally recurrent NPC, the local control rate was 100% at 9 months. In the study from Chua et al [8], after a median dose of 54 Gy of IMRT with or without stereotactic radiosurgery (SRS), 58% of patients had complete regression of primary tumor; one-year loco-regional progression-free, distant metastasis-free and overall survival rates were 56, 90, and 63%, respectively. And in the study from Han et al [7], with a median follow-up time of 25 months, the rates of 2-year locoregional recurrence-free survival and overall survival were 65.8 and 67.4% respectively after 70Gy of IMRT. In the recent study from Qiu et al [9] of 239 NPC patients with local recurrence who were re-irradiated with IMRT between 2001 and 2008, with a mean dose of 70.04 Gy to the GTV, the 5 year overall survival, local progression-free survival were 44.9 and 85.8% respectively. Compared to these study, the 2-year local control rate of 64% in the current study was similar, but the 2-year OS rate of 44.3% seemed inferior. It maybe due to the relatively late rT stage: rT3 was 16.7%, and rT4 ws 63% in our study. Studies from Chua [8] and Han et al [9] all indicated that rT stage was important prognostic factor for survival. In Chua’s study [8], significantly better 1-year local progression-free rate was observed in rT1–3 than rT4 tumor (100 vs. 35%). And the 5-year OS rate were 84.6, 67, 37.9, 32.3% for rT1, rT2, rT3, rT4 respectively in Lu’s study. Advanced T classification, especially T4 disease, may be a strong marker for the underlying locally aggressive biology of the original disease.

Another possible reason maybe the large GTV and PTV herein, the mean GTV and PTV was 59.4 ml and 134 ml respectively, which was much larger than previous study. And GTV was the prognostic factor for survival illustrated by Qiu et al [9], patients with GTV >38 ml had worse prognosis. Larger GTV may be indicative of more aggressive biology, more hypoxic cell in the foci, and more resistant to treatment. In our study PTV was the only adverse prognostic factor in multivariate analysis of OS, but not rT stage or GTV. PTV was not only representative of GTV, but also the direct reflection of the treatment field. And OS of local recurrent NPC depended on not only local control, but also the treatment related side events, which was influenced by treatment field and dose. Local failure and radiation injuries were the two major cause of death in our study. In the multivariate analysis of LPFS, concurrent chemoradiotherapy was the negative prognostic factor, for those treated with concurrent chemoradiotherapy, the LPFS was shorter. This was inconsistent with conventional view. Concurrent chemoradiotherapy did not cause more treatment break or cease in this study. It may be due to the larger proportion of rT3-4 in the concurrent arm, which may affect the local control.

The incidence of SLAE was 48.1% here, which was rather high, and among the patients dead, 37.1% were died because of radiation injuries. Compared to the existing data about IMRT for local recurrent NPC, the side events here was not surprising. As illustrated in Tab. 7, the lowest incidence was from Chua et al [8] with 25% of ≥gradeⅢ events; and the highest was from Qiu et al [9] with 70.3%, whereas it was 35.7% in the data from Han et al [7], and in the external radiotherapy arm from Lee et al [10] the incidence was 75% compared with 8% in the combined-modality arm.

In different data, the calculation of SLAE may be different, this may partly lead to the different incidence of events. Considering that some unique side effects were observed after reirradiation of NPC, such as massive haemorraghe and difficulty in feeding because of various reasons, which were common and sometimes vital in the survival outcome, but were not described and graded in the RTOG side effect grading system, we graded them by CTCAE 3.0, and those ≥gradeⅢ events were also defined as SLAE. This maybe the major difference from other data, and may partly explained the high incidence of adverse events in this study.

The major SLAE was ulcer or necrosis of nasopharyngeal mucose with as high as 31.5% of incidence. This was also the major event in the study from Qiu et al [9], where the incidence was 40.6%; and it was 15.7% in Han et al’s treatment group [7]. As Marx reported [11] irradiation may cause hypoxia, hypovascularity, and hypocellularity and may impair normal collagen synthesis and cell production, which may lead to tissue breakdown and a chronic nonhealing wound. The occurrence of nasopharyngeal necrosis is also related to personal tolerance and radiation methods, especially related to irradiation dose and course. We analyzed the corresponding treatment plans, and found that most of those necrosis mucose were located in the high dose area. And this may be part of the truth of the high incidence of necrosis after second cycle of RT. Ulcer or necrosis of nasopharyngeal mucose was a tough event, even after aggressive treatment combined of internal medicine and surgical approaches such as debridement or sequestrectomy, the consequence was still dissatisfied. The condition was aggravated with subsequent internal carotid artery erosion and difficulty in feeding that leading to exhaustion, which was also the situation in the study from Hua et al [12] and Wu et al [13].

Massive haemorraghe from internal carotid artery was another lethal side effects herein. Seven (13%) patients had massive haemorraghe, and five patients died of it, the seven patients were all combined with nasopharyngeal necrosis. In the data from Han et al [7], six patients were also died of massive epistaxis, though the details were unclear, it was hard to exclude the possibility of the epistaxis due to treatment. Though the exact incidence was not reported in the data from Qiu et al [9], they also concluded that it was important to reduce incidence of nasopharyngeal mucosal necrosis and massive hemorrhage. Second cycle of RT may lead to unrepairable injury to vessels. The subintima fibrous-like substance deposition, the degeneration and decreased elasticity of the media muscle fiber, fibrosis of adventitia, and the foam cells aggregation in vessel walls are all contributing factors to the fragility of vessels after radiotherapies [14]. Vessel dysplasia may be found after radiotherapy with a high probability, which is characterized by thinning wall and deranged structure, making them be at the risk of bleeding as pressure changes or inflammatory injuries occur. Massive haemorraghe often was sudden and unpredictable, thus often lethal, and often the patient cannot wait for treatment before they die.

Another common event was difficulty in feeding, which required nutritional support either by nasal feeding, gastrostomy feeding or total parenteral alimentation. The difficulty in feeding may be caused by trismus, swallowing pain, or nasopharyngeal necrosis, and in patients after second cycle of RT these causes were often confused and hard to be differentiated. 11 patients in this treatment group finally had difficulty in oral feeding, often the normal condition became poor and exhausted, three of them died of cachexia eventually.

The common side effects reported in other reports was TLN, which was the third most severe late events in our study, with the incidence of 18.5%. It was as high as 25.8% in the study from Han et al [7], and 8% in Chua’s study [9]. Patients with TLN usually present with mild symptoms of dizziness, forgetfulness, or temporal lobe epilepsy. Severe and even fatal complications of TLN, including convulsion, intracranial hemorrhage, mass effect causing herniation, and death, had occasionally been seen. In our group, three patients finally died of TLN. The cumulation of radical dose to the lobe from two cycle of RT, especially in those with intracranial invasion may partly lead to this side effect.

SLAE was not only common but sometimes also lethal in this study, 25% of patients finally died of SLAE. In the study from Han et al [7], as much as 34.7% of patients died of treatment related toxicities eventually. Considering the important influence of SLAE on life quality and survival, it was necessary to tell who would benefit from re-IMRT and who may suffer from irrepairable damage. And in view of that some adverse effects were quite distribution-dependent, we also analyzed the relationship between the location of recurrent foci and severe events. Invasion of poststyloid space was predictive of severe events in multivariate analysis, but not GTV or rT stage, which was different from that of Qiu et al [9]. It was reasonable because the major side effects was nasopharyngeal mucosal necrosis and subsequent haemorraghe and difficulty in feeding in our study. Nasopharyngeal mucosal necrosis was most often seen in the lateral wall of nasopharynx especially after high dose of RT. Because this was the first time to analyze the relationship between side events and location of recurrent foci, it was hard to draw conclusion if this was a common phenomenon. But based on this result, it was important to consider the recurrent foci before giving the treatment in order to protect from severe events. And patient with invasion of poststyloid space was prone to have nasopharyngeal mucosal necrosis, it was important for early follow-up and intervention of suspicious clue of side events, and more importantly, to consider what was the more proper treatment for such kind of patient.

In order to decrease the incidence of severe side events, one may lower the total radiation dose, shrink the radiation field, or combine several treatment modalities. As we know, the re-irradiation dose is an important factor for the local control of recurrent NPC. Wang [15] reported the dose–response relationship of re-irradiation to outcomes in the conventional radiotherapy era. The 5 year survival rate was 45% when more than 60 Gy was delivered; however, no patients survived when the dose was less than 60 Gy. Leo et al [16], Chang et al [17] and Leung et al [18] all proposed that RT dose of more than 60 Gy was essential for local control. It seems unreasonable to lower the radiation dose to be less than 60Gy. The dose herein was 68.5Gy, which was similar to the data illustrated in Tab. 7, but even after such high dose, local failure remained the main cause of death, this may partly be explained by the major component of rT3-4 and large tumor in our series, but it also demands some thinking about the optimal dose for disease control. If high dose is necessary, for those patients with predictable high incidence of severe treatment-related events, what is the optimal treatment options?

In the data from Lee et al [10], the combination of intracavitary brachytherapy with external beam radiotherapy compared with external beam RT alone, significantly decreased the incidence of severe side events, which was 8% and 75% respectively, without influencing the local control and overall survival. Maybe for some superficial lesions such as those invading the parapharyngeal space it was a reasonable choice. And also in the data from Chua et al [8], the median dose was 56.8Gy, and radiosurgery was offered to these patients to selectively boost the PTV that received a relatively low dose (40–45 Gy) by IMRT, with a single dose which ranged from 8.5 to 12.5 Gy prescribed to 80% isodose surface. 68% had induction chemotherapy. Their results suggest that a dose between 50 and 60 Gy could yield good control for rT1–3 but not for rT4 NPC, the side effects were less compared to other existing data in their study. Gullu et al [19] illustrated that compared with conformal radiotherapy with or without brachytherapy, robotic stereotactic radiotherapy resulted in lower incidence of serious late toxicities, that was 21% of patients in the SBRT arm and 48% in the CRT arm, fatal complications occurred in three patients (12.5%) of the SBRT arm, and four patients (14.8%) of the CRT arm. Stereotactic body radiotherapy seemed particularly useful when the tumor was large or abutting/encasing critical structure. This maybe another option for some patient, for example, those with tumors invading the intracranial structures.

Though chemotherapy showed no influence on overall survival in the current and previous studies, and the concurrent arm even had shorter LPFS herein, accordingly to the experience from the treatment of primary NPC, one cannot simply exclude the efficience of chemotherapy on the treatment of local recurrent NPC, especially for those with large tumor, but what is the optimal drug and schedule is under discussion. It may be necessary to have a prospective analysis upon the efficience of chemotherapy on recurrent NPC.

In conclusion, the present study demonstrates that IMRT with 70Gy was efficient for local tumor control. However, we observed a high frequency of serious late complications, and some of the complications were lethal. More optimized combination treatment and patient selection are required to achieve excellent local control without significant late morbidities in locally recurrent NPC.

Acknowledgments

The authors thank Dr Tai-Fu Liu and Shao-qin He (Department of Radiation Oncology, Cancer Hospital, Fudan University, Shanghai, China) for critical review and correction of the manuscript.

Disclaimer: This paper has not been presented at any conference, and was not registered as a clinical trial.