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
Yun-ming Tian MD,
Department of Radiation Oncology, Hui Zhou Municipal Centre Hospital, China
Department of Radiation Oncology, Sun Yat-sen University Cancer Centre, State Key Laboratory of Oncology in South China, Collaborative Innovation Centre of Cancer Medicine, Guangzhou, China
Corresponding author: Fei Han, PhD, Department of Radiation Oncology, Sun Yat-sen University Cancer Centre, State Key Laboratory of Oncology in South China, Collaborative Innovation Centre of Cancer Medicine, No. 651 Dongfeng East Rd, Guangzhou, China; Fax: (011) 86-20-87343372; firstname.lastname@example.org
The optimal model of total dose and fraction size for patients with locally recurrent nasopharyngeal carcinoma treated with intensity-modulated radiotherapy (IMRT) remains unclear. The authors designed a randomized phase 2 clinical trial to investigate the efficacy of 2 different models, with the objective of determining an optimal model.
Between January 2003 and December 2007, a total of 117 patients with locally recurrent nonmetastatic nasopharyngeal carcinoma were randomized to 2 different models of total dose and fraction size: group A (59 patients) received 60 gray in 27 fractions and group B (58 patients) received 68 gray in 34 fractions. Both groups received 5 daily fractions per week. All patients received IMRT alone.
The median follow-up was 25.0 months. The 5-year overall survival in group A was higher than that in group B (44.2% vs 30.3%; P =.06), and the local failure-free survival in group A was slightly lower than that in group B (63.7% vs 71.0%; P =.41). Severe late complications were the main cause of death. The incidences of mucosal necrosis and massive hemorrhage in patients in group B were significantly higher than those among patients in group A at 50.8% versus 28.8% (P =.02) and 31.0% versus 18.6% (P =.12), respectively. Tumor volume (P<.01) and model of total dose and fraction size (P =.03) were found to be significant factors for mucosal necrosis and massive hemorrhage.
Nasopharyngeal carcinoma (NPC) is one of the most common head and neck malignancies in southern China.[1, 2] Although excellent control has been achieved with the development of irradiation techniques and their combination with chemotherapy, local recurrence remains a major cause of treatment failure in patients with advanced NPC, at a rate of approximately 8% to 10%.[3, 4]
The treatment of patients with locally recurrent NPC who have undergone previous full-dose radiotherapy (RT) remains a challenge. However, a significant percentage of patients can achieve long-term survival after potentially curative salvage therapy including RT and surgery. For tumors that are localized to the nasopharynx or are of small volume, salvage treatments including surgery, brachytherapy, and stereotactic radiosurgery may improve the patient's quality of life. Unfortunately, most locally recurrent NPCs are found to be extensive at the time of diagnosis and external beam RT is often the only potentially curative salvage treatment.[5-7] With conventional RT, it is difficult to provide a radical dose to the tumor due to the presence of critical structures in the vicinity; as a consequence, the clinical outcome for these patients is poor, with a 5-year overall survival (OS) rate of only 8% to 36% and an incidence of severe late complications of 26% to 80%.[8, 9] The use of intensity-modulated RT (IMRT) can improve the balance between target coverage and the sparing of adjacent organs, and may become the principal technique for the treatment of patients with locally recurrent NPC.[10-12]
Total dose and fraction size are important prognostic factors for survival, although to our knowledge the optimal model has not been well determined to date. According to retrospective analyses of patients with locally recurrent NPC, a dose of 60 gray (Gy) to 70 Gy is considered radical for reirradiation.[13-18] Fraction size is also considered to be a factor associated with local control.[19, 20] It is therefore worthwhile to investigate the feasibility of decreasing the total dose and increasing the fraction size with the objective of achieving a better balance between local control and severe late complications. Based on our experience with conventional RT, we conducted a randomized trial to investigate the efficacy of 2 different models of total dose and fraction size to improve the clinical treatment benefit by reirradiation using IMRT.
MATERIALS AND METHODS
Patients were eligible for the current study if they met the following inclusion criteria: 1) histologically confirmed locally recurrent NPC or NPC diagnosed by clinical symptoms and radiological findings in those patients with disease located in the skull base or intracranial cavity that was inaccessible for biopsy; 2) no evidence of distant metastases at the time of diagnosis; 3) an interval of >6 months between the end of primary RT and disease recurrence; and 4) a Karnofsky performance status score of at least 70. Our exclusion criteria included previous chemotherapy, RT, or definitive surgery after the diagnosis of locally recurrent NPC. We also excluded patients who had another active cancer in addition to NPC or who had unstable cardiac or renal disease that required treatment. All participants provided written informed consent. Our protocol was approved by the institutional ethics committee.
The pretreatment evaluation involved a complete medical history and physical examination. Complete blood count and renal and liver function tests were required. Magnetic resonance imaging (MRI) of the head and neck was performed for the staging evaluations. Chest radiography, abdominal sonography, and bone scanning were performed to detect distant metastases. All patients with recurrent NPC were restaged according to the 2002 classification system of the American Joint Committee on Cancer.
Target volumes were delineated according to our institutional treatment protocol. The gross tumor volume (GTV) on the primary site (GTV-nx) and neck (GTV-nd) included all disease observed on MRI. The clinical target volumes (CTV) were designed to encompass microscopic disease including the GTV plus a 1-cm to 1.5-cm margin and a smaller margins (<3 mm) allowed near critical intracranial structures or the spinal cord. The CTV also included the entire nasopharynx and the lymph node-positive involved regions. No elective reirradiation of uninvolved regional lymph nodes was performed. Additional 1.4-mm to 2.8-mm margins were added to the CTV and GTV to create the planning target volume for setup variability and internal motion.
Patients were randomized to 1 of 2 different models of total dose and fraction size. The control group (group B) was prescribed 68 Gy in 34 fractions to GTV-nx; the trial group (group A) was prescribed 60 Gy in 27 fractions to GTV-nx. Both were given 5 daily fractions per week. All patients underwent a full course of IMRT (Nomos Corporation, Pittsburgh, Pa) with 6-megavolt x-rays generated using a Clinac-600C linear accelerator (Varian Medical Systems, Palo Alto, Calif).
Patient Assessment and Follow-Up
During treatment, we assessed toxic effects and tumor regression weekly. After the completion of treatment, patients were evaluated at least once every 3 months during the first 3 years and every 6 months thereafter until death. MRI of the head and neck, chest radiography, and abdominal sonography were performed annually. Severe radiation toxicities secondary to treatment were assessed and scored according to the radiation morbidity scoring criteria of the Radiation Therapy Oncology Group. The date of the last follow-up was December 31, 2012.
Based on previous data, the 5-year OS rate for group B was expected to be 30%, with 80% power to detect a 23% difference between the 2 groups (2-sided P<.05). This power level, with an expected loss of 5% of patients during follow-up, means that this study required a sample size of at least 114 patients (57 patients per group). The randomization code was developed using a computerized random number generator. Patients were assigned randomly using blocks of 4 based on 1:1 treatment allocation.
The primary endpoint of the study was the OS of groups A and B. Secondary endpoints were local failure-free survival (LFFS), distant failure-free survival (DFFS), progression-free survival (PFS), and the incidence of severe late complications (grade 3/4). OS was defined as the time between the date of randomization and death from any cause. PFS was defined as the time between the date of randomization and the first failure at any site. For LFFS and DFFS analyses, the latencies to the first local or distant failure, respectively, were recorded.
The analyses were performed on an intention-to-treat basis. The Kaplan-Meier product-limit method was used for the calculation of OS, PFS, LFFS, and DFFS. Incidences of severe late complications were compared using the chi-square test (or Fisher exact test, if indicated). The statistical significance of differences between survival curves was analyzed using the log-rank test. We calculated hazard ratios (HRs) using the Cox proportional hazards model. Covariates included host factors (sex and age), tumor factors (recurrent T classification, disease-free interval, and tumor volume), and model of total dose and fraction size (treatment group). Two-tailed P values of <.05 were considered to be statistically significant.
Figure 1 shows the trial profile. Between January 2003 and December 2007, a total of 117 patients were included in this analysis (59 patients in group A and 58 patients in group B). The 2 groups were well balanced with regard to patient characteristics and tumor factors (Table 1).
Table 1. Clinical Characteristics
Group B (n=58)
Abbreviations: 2D-RT, 2-dimensional radiotherapy; 3D-CRT, 3-dimensional conformal radiotherapy; AJCC, American Joint Committee on Cancer; DFI, disease-free interval (time from the end of the first course of radiotherapy to diagnosis of recurrence); Gy, gray; IMRT, intensity-modulated radiotherapy; KPS, Karnofsky performance status; WHO, World Health Organization.
The diagnosis was based on radiological findings and clinical symptoms.
The minimum, mean, and maximum doses to the nasopharyngeal GTV were 53.1 Gy, 65.4 Gy, and 71.7 Gy, respectively, in group A and 60.0 Gy, 73.1 Gy, and 79.6 Gy, respectively, in group B. Details are summarized in Table 2.
Table 2. Reirradiation With IMRT: Doses to Targets and Critical Structures
Group A Median Dose (Range), Gy
Group B Median Dose (Range), Gy
Abbreviations: BS, brain stem; CTV, clinical target volume; D95, dose encompassing 95% of the gross tumor volume; GTV, gross tumor volume; Gy, gray; IMRT, intensity-modulated radiotherapy; TL, temporal lobe; V95, percentage of the GTV receiving 95% of the prescribed dose.
Parotids (left), mean
Parotids (right), mean
Patient Survival Outcomes
Seven patients (6.0%) were lost to follow-up between 6.0 and 36.0 months. The median follow-up was 25.0 months (range, 6 months-118 months).
The 3-year and 5-year OS rates were 47.7% and 37.3%, respectively, for all patients in the study. Survival in group A was higher than that in group B (57.4% vs 38.0% and 44.2% vs 30.3%, respectively; P =.06). The HR for patients in group B was 1.52 (95% confidence interval [95% CI], 0.97-2.38). The 5-year LFFS rate in patients in group A was slightly lower than that in group B (63.7% vs 71.0%; P =.41). The 5-year DFFS and PFS rates were similar in the 2 groups (86.8% vs 81.1% [P =.34] and 56.8% vs 55.2% [P =.58], respectively). Survival curves are shown in Figure 2.
Patterns of Treatment Failure and Causes of Death
Local failure after the completion of RT was confirmed in 31 patients (26.4%), comprising 18 patients in group A and 13 patients in group B. Fourteen patients (12.0%) developed distant metastases; 2 of these individuals also had local failure.
Seventy-nine patients died, including 5 deaths that were unrelated to the cancer or its treatment (from cardiac disease, digestive disease, and intracranial infection). The main causes of death were radiation-related late complications and disease progression. Among patients in group A, disease progression was the major cause of death (51.4%) followed by severe late complications (40.0%); in patients in group B, the major cause of death was severe late complications (54.5%), especially nasopharyngeal mucosal necrosis and massive hemorrhage, followed by disease progression (40.9%). Causes of death are summarized in Table 3.
After their primary course of RT, the majority of patients experienced mild to moderate xerostomia, neck fibrosis, and trismus; 3 patients also developed temporal lobe necrosis and 4 patients had cranial neuropathy. In the current study, we recorded only new complications or complications that became more severe. None of the patients had their treatment interrupted due to severe acute toxicity. Five of 59 patients in group A (8.5%) experienced grade 3 acute mucositis, compared with 13.7% of patients in group B. The most common late complications were xerostomia, hearing loss, trismus, mucosal necrosis, massive hemorrhage, temporal lobe necrosis, and cranial nerve palsy. Mucosal necrosis and massive hemorrhage were the main causes of death. The incidences of mucosal necrosis and massive hemorrhage among patients in group B were significantly higher than those in group A, at 50.8% versus 28.8% (HR, 2.3; 95% CI, 1.1-5.0 [P =.02]) and 31.0% versus 18.6% (HR, 1.7; 95% CI, 0.7-4.3 [P =.12]), respectively. However, there were no statistically significant differences with regard to the other complications. The incidences of complications are summarized in Table 4.
Table 4. Late Severe Complications in 117 Cases of Locally Recurrent NPC
The following covariates were examined in our statistical models: sex (male vs female), age (>45 years vs ≤45 years), rT classification (rT3/4 vs rT1/2), disease-free interval (≤36 months vs >36 months), tumor volume (>26 cm3 vs ≤26 cm3), and model of total dose and fraction size (68 Gy/34 fractions vs 60 Gy/27 fractions). Table 5 shows the results of multivariate analyses for OS, LFFS, and DFFS. The statistically significant negative prognostic factors for OS were age >45 years, rT3/4 disease, and a tumor volume >26 cm3. The 5-year OS rates for patients with rT1/2, rT3, and rT4 disease were 66.8%, 41.3%, and 15.8% (P<.01), respectively. Patients with a tumor volume >26 cm3 had a poorer OS compared with those with a tumor volume ≤26 cm3 (61.0% vs 15.9%; P = .05). rT classification was the only significant negative prognostic factor for LFFS, DFFS, and PFS. The 5-year LFFS rates for patients with rT1/2, rT3, and rT4 disease were 84.8%, 66.5%, and 56.9%, respectively.
Table 5. Multivariate Analysis of OS, LFFS, and DFFS
The results of the current study indicate that tumor volume (P<.01) and model of total dose and fraction size (P =.03) were significant negative factors associated with mucosal necrosis. The incidence of mucosal necrosis among patients with a tumor volume >26 cm3 was 53.1%, which was significantly greater than that noted among patients with a tumor volume ≤26 cm3, at 22.6%. The HR in patients with a tumor volume >26 cm3 was 3.7 (95% CI, 1.6-8.4). Tumor volume was also found to be a significant independent negative prognostic factor for massive hemorrhage, at 35.9% with a tumor volume >26 cm3 and 11.3% with a tumor volume ≤26 cm3 (P<.01). The HR in patients with a tumor volume >26 cm3 was 4.6 (95% CI, 1.7-11.8). There were no significant independent factors found to be associated with temporal lobe necrosis or cranial nerve palsy.
With its favorable balance between target coverage and sparing of adjacent organs, IMRT is the principal radiation technique used for the treatment of patients with recurrent NPC. Hsiung et al reported that the average maximal dose to the brainstem was reduced from 30.9% of the prescribed dose with 3-dimensional conformal RT to 15.3% and 14.7% with 5-field and 7-field IMRT, respectively. In a study by Lu et al, a local control rate of 100% was achieved with IMRT in 49 patients without any severe late complications after a median follow-up of 9 months. Chua et al reported a 1-year local control rate of 56% and an OS rate of 63% in 31 patients after IMRT, with a 1-year locoregional control rate of 100% for patients with rT1 to rT3 disease. In the current study, although advanced tumors accounted for 78% of the cases, the 5-year OS rate was approximately 37.3%, which confirms the importance of IMRT in locally recurrent NPC.
The model of total dose and fraction size used is an important factor in the efficacy of RT,[13-16, 21] although the optimal model remains unclear and varies between centers. The sensitivity of a recurrent tumor to radiation may be considerably reduced due to the presence of radioresistant cell clones and hypoxia caused by the decreased blood supply. Thus, such tumors cannot be well controlled by a low dose of RT, but serious late complications are common after a high dose, and both of these situations lead to failure of reirradiation. Determining the optimal model of total dose and fraction size is therefore of great importance in improving the survival of patients with recurrent NPC.
The total dose delivered was based on experience with conventional RT. According to the dose response relationship for recurrent NPC reported by Wang et al, the 5-year OS rate was approximately 45% when >60 Gy was delivered to the tumor; no patient survived for >5 years when the dose was <60 Gy, indicating that the total dose was a significant positive factor for survival. In a retrospective analysis of 654 patients with recurrent NPC reported by Lee et al, the 5-year local control rate for those with early-stage disease was 40%, 35%, and 14%, respectively, when a biological effective dose (BED) of the second course was given at >70 Gy, 60 to 70 Gy, or <60 Gy. When compared with the group given 60 to 70 BEDl0 (α/β = 10), the rate of local control in those patients treated with <60 Gy was definitely inferior (P =.00l), but the superior local control rate among patients who received >70 Gy failed to reach statistical significance (P =.229). The HR for local failure decreased by 1.7% per BED (1 Gy) in the second course. However, the survival benefit provided by a high dose of conventional RT remains uncertain because of the occurrence of severe late complications. In a study of 176 patients with recurrent NPC reported by Teo et al, the 5-year local control and OS rates were 15.2% and 7.6%, respectively, when the radiation dose was >60 Gy, but this was accompanied by a high incidence of serious late complications including trismus (69.9%) and temporal lobe necrosis (20.4%), indicating that the severe complications caused by high-dose reirradiation could outweigh the potential for prolongation of life. In a study of 86 patients treated with 3-dimensional conformal RT, Zheng et al reported a 5-year actuarial local control rate and an OS rate of 71% and 40%, respectively, and the incidence of grade 3 and grade 4 late toxicities was as high as 100% and 49%, respectively, when the mean dose to the tumor target was >68 Gy; these toxicities were becoming the main cause of death. In the current study, excellent local control was achieved when the total dose was >60 Gy; however, the incidence of late complications including mucosal necrosis and massive hemorrhage increased when the dose was increased from 60 Gy to 68 Gy, significantly increasing the HR for death.
Fraction size is another factor associated with local control of disease. Superior outcomes with an increase in fraction size have been reported in the literature for patients with head and neck cancer.[19, 20] Le et al reported that the 5-year local control rate for patients with T2 glottic carcinoma was 100% when a fraction size of >2.2 Gy was delivered, which was significantly higher than that of 44% reported with a fraction size of <1.8 Gy. Furthermore, in the study by Schwaibold et al, a fraction size of <2.0 Gy was found to result in inferior tumor control. However, the data regarding fraction size related to locally recurrent NPC appear to be limited.[14, 18] It is interesting to note that, in the study by Lee et al, correlation with BED for late tissues suggested that the impact of the first course of RT was more predominant. The risk of late complications increased by 4.2% per BED3 (α/β = 3) of the initial course of RT (P = .03), whereas the corresponding increase was only 1.2% per BED3 of the second course (P = .08). In our previous retrospective report, the 5-year OS rate was 50% in the group treated with fractionation of >2.3 Gy, and 38.2% among patients in the subgroup treated with ≤2.3 Gy (P =.011). These outcomes indicate that a slightly higher dose per fraction of the second course may be beneficial in the reirradiation of patients with locally recurrent NPC by improving the local control of disease with an acceptable complication rate. The results of the current study indicate that an increase in fraction size to 2.2 Gy can to some extent compensate for a smaller total dose of 60 Gy, achieving local control similar to that observed with a dose of 68 Gy. Importantly, a greater survival benefit was achieved with the lower total dose due to the decrease in the rate of severe late complications.
Although excellent local control can be achieved among patients with recurrent NPC after IMRT via the model using a decrease in total dose and an increase in fraction size, severe late complications still remain a challenge. Mucosal necrosis is common and can seriously reduce quality of life or lead to death due to massive hemorrhage when the internal carotid artery is involved. However, to our knowledge, the mechanism of mucosal necrosis after irradiation is not well documented. As reported by Marx, irradiation may cause hypoxia, hypovascularity, and hypocellularity and may lead to tissue injury and a chronic nonhealing wound. Furthermore, the treatment of mucosa necrosis is very complicated, and to our knowledge no effective and safe method currently exists. Although the study by Hua et al indicated the endoscopic surgery can to some extent reduce symptoms and improve quality of life for patients with early-stage disease, it is limited by the high risk of massive hemorrhage and other serious complications due to the nonhealing wounds. Therefore, the investigation of the risk factors associated with mucosal necrosis to reduce its incidence was significant. In the current study, total dose and tumor volume were associated with the occurrence of mucosal necrosis; the incidence in patients with a tumor volume of >26 cm3 was 53.1%, and greater than one-half of these patients developed massive hemorrhage. The incidences of mucosal necrosis and massive hemorrhage were 50.8% and 31.0%, respectively, among patients in group B (treated with 68 Gy), which were significantly higher than those noted in group A (treated with 60 Gy) at 28.8% and 18.6%, respectively. The high incidence of mucosal necrosis may be related to the following factors: 1) IMRT achieves a more favorable balance between target coverage and the sparing of adjacent organs and therefore the dose to the tumor target, including the nasopharyngeal mucosa, is significantly higher than in patients treated with conventional RT; and 2) the hypovascularity and poor immunity caused by reirradiation lead to a high risk of infections and nonhealing injuries. It is therefore crucial to strike a balance between disease control and potential severe late complications, especially among patients with bulky tumors, and an appropriate decrease in the total radiation dose may be effective. Furthermore, with the advances in comprehensive treatment, chemotherapy has played an important role in the treatment of patients with primary advanced NPC, with the aim of improving local control and decreasing the rate of distant metastases. However, to our knowledge, only a few retrospective analyses with small sample sizes have been performed to date to investigate the feasibility of chemotherapy, and its role in patients with locally recurrent NPC remains uncertain.[27, 28] To address this issue, a randomized trial has been conducted in the study institution to compare IMRT alone with IMRT plus concurrent chemotherapy. Those investigations may further improve the therapeutic ratio of locally recurrent NPC.
Reirradiation with IMRT achieves excellent disease control and long-term survival in patients with locally recurrent NPC. Appropriately decreasing the total dose and increasing the fraction size can achieve local control rates similar to those observed with a higher dose. Furthermore, it can significantly reduce the incidence of severe, potentially fatal complications including mucosal necrosis and massive hemorrhage, thereby improving the therapeutic ratio. More effective methods including combination chemotherapy should be investigated further.