Major Late Toxicities After Conformal Radiotherapy for Nasopharyngeal Carcinoma—Patient- and Treatment-Related Risk Factors

Int. J. Radiation Oncology Biol. Phys., Vol. 73, No. 4, pp. 1121–1128, 2009 Copyright Ó 2009 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/09/$–see front matter

doi:10.1016/j.ijrobp.2008.05.023

CLINICAL INVESTIGATION

Head and Neck

MAJOR LATE TOXICITIES AFTER CONFORMAL RADIOTHERAPY FOR NASOPHARYNGEAL CARCINOMA—PATIENT- AND TREATMENT-RELATED RISK FACTORS ANNE W. M. LEE, F.R.C.R.,* W. T. NG, F.R.C.R.,* W. M. HUNG, C.M.D., M.SC.,* C. W. CHOI, M.SC.,* RAYMOND TUNG, B.SC.,y Y. H. LING, B.SC.,* PETER T. C. CHENG, B.SC.,* T. K. YAU, F.R.C.R.,* AMY T. Y. CHANG, M.B.B.S.,* SAMUEL K. C. LEUNG, M.SC.,z MICHAEL C. H. LEE, PH.D.,z x AND SOREN M. BENTZEN, PH.D., D.SC. Departments of * Clinical Oncology and z Medical Physics, Pamela Youde Nethersole Eastern Hospital, Hong Kong; y Hong Kong Cancer Fund, Hong Kong; and x Department of Human Oncology, University of Wisconsin School of Medicine and Public Health, Madison, WI Purpose: To retrospectively analyze the factors affecting late toxicity for nasopharyngeal carcinoma. Methods and Materials: Between 1998 and 2003, 422 patients were treated with a conformal technique with 2-Gy daily fractions to a total dose of 70 Gy. Conventional fractionation (5 fractions weekly) was used in 232 patients and accelerated fractionation (6 fractions weekly) in 190 patients. One hundred seventy-one patients were treated with the basic radiotherapy course alone (Group 1), 55 patients had an additional boost of 5 Gy in 2 fractions (Group 2), and 196 patients underwent concurrent cisplatin-based chemotherapy (Group 3). Results: The 5-year overall toxicity rate was significantly greater in Group 3 than in Group 1 (37% vs. 27%, p = 0.009). Although the overall rate in Group 2 was not elevated (28% vs. 27%, p = 0.697), a significant increase in temporal lobe necrosis was observed (4.8% vs. 0%, p = 0.015). Multivariate analyses showed that age and concurrent chemotherapy were significant factors. The hazard ratio of overall toxicity attributed to chemotherapy was 1.99 (95% confidence interval, 1.32–2.99, p = 0.001). The mean radiation dose to the cochlea was another significant factor affecting deafness, with a hazard ratio of 1.03 (95% confidence interval, 1.01–1.05, p = 0.005) per 1-Gy increase. The cochlea that received >50 Gy had a significantly greater deaf rate (Group 1, 18% vs. 7%; and Group 3, 22% vs. 14%). Conclusion: The therapeutic margin for nasopharyngeal carcinoma is extremely narrow, and a significant increase in brain necrosis could result from dose escalation. The significant factors affecting the risk of deafness included age, concurrent chemoradiotherapy, and greater radiation dose to the cochlea. Ó 2009 Elsevier Inc. Nasopharyngeal carcinoma, Late toxicity, Concurrent chemotherapy, Radiation boost.

the standard recommendation, because increasing evidence from randomized trials and meta-analyses has shown that this can significantly improve tumor control (2–5) and overall survival (2–4). Other strategies suggested by retrospective studies have included more aggressive RT with dose escalation using brachytherapy (6) or stereotactic RT (7, 8) and/or accelerated fractionation (9). All these additional strategies could further increase the risk of toxicity (both early and late). Accurate knowledge of the maximal tolerance dose for various organs at risk (OARs) is crucial for determining the risk we can safely take to achieve the greatest possibility of uncomplicated locoregional control. The present

INTRODUCTION Radiotherapy (RT) is the primary treatment modality for nasopharyngeal carcinoma (NPC). Although the classic nonkeratinizing type is radiosensitive, a high dose is required for cure, and a total dose of about 70 Gy is generally recommended (1). With the anatomic proximity of critical structures, the therapeutic margin is notoriously narrow. The importance of using the most conformal technique for maximal protection of normal structures cannot be overemphasized. For most patients who present with extensive locoregional disease, the addition of concurrent chemotherapy has become Reprint requests to: Anne W. M. Lee, F.R.C.R., Department of Clinical Oncology, Pamela Youde Nethersole Eastern Hospital, 3 Lok Man Rd., Chai Wan, Hong Kong. Tel: (852) 2595-4173; Fax: (852) 2904-5216; E-mail: [email protected] Conflict of interest: none. Acknowledgments—We wish to thank the Hong Kong Cancer Fund for supporting the statistical analyses; the Hong Kong Anti-Cancer

Society for supporting the research assistant; Imen Ku, Francis Choi, and Fion Leung for their contribution to the dose–volume data; and Cherine Lau and all medical staff for their contribution to the data entries. Received Feb 4, 2008, and in revised form May 13, 2008. Accepted for publication May 14, 2008. 1121

I. J. Radiation Oncology d Biology d Physics

1122

recommendations for tolerance doses have largely been quoted from the monumental publication by Emami et al. in 1991 (10). Only data on RT alone given using conventional fractionation (1.8–2 Gy daily, 5 d/wk) were considered. Some of the doses recommended were based on available experimental and clinical investigational data, and some were based on the experience of the clinicians involved in the task force. The accurate correlation of dose with late toxicity from patients with dose–volume histogram (DVH) data and long follow-up is grossly lacking. Using a series of consecutive patients with NPC treated by a three-dimensional conformal technique, the present analyses aimed to estimate the relationship between the doses received by the OARs at the nasopharyngeal region and the respective incidence of late toxicity and to assess the increase in risk incurred by additional treatment (including concurrent chemotherapy using a cisplatin-based regimen, radiation dose escalation, and accelerated fractionation). METHODS AND MATERIALS Patient characteristics Included in our study were patients with NPC treated at our center between October 1998 and June 2003 who had completed radical primary RT to a total dose of 70 Gy in 2-Gy daily fractions. If an additional RT boost was given, the total dose was 5 Gy in 2 fractions. No additional boost was given for residual tumor. If chemotherapy was given, at least 1 cycle was given in concurrence with RT. All patients included in the present study had survived for >90 days after beginning RT for evaluation of late toxicity. The data from 422 consecutive eligible patients were analyzed. The patients’ median age was 48 years (range, 17–84); the proportion of males was 72%. The median duration of follow-up was 4.4 years (range, 0.3–7.1). All survivors had been observed for $3 years.

Volume 73, Number 4, 2009

The staging system of the American Joint Committee on Cancer (AJCC) and the International Union against Cancer (UICC) (fifth edition) (11, 12) was used. Of the 422 patients, 63 (15%) had Stage I-IIB and 359 (85%) Stage III-IVB; 60% of the primary tumors were Stage T3-T4. Table 1 lists the patient characteristics, tumor factors, and treatment parameters.

Primary treatment policies Patients with Stage I-IIB disease were treated with RT alone using conventional fractionation of 5 fractions weekly. An additional boost of 5 Gy in 2 fractions was given toward the end of the basic RT course. High-dose-rate brachytherapy was used to deliver this boost to Stage T1-T2a disease and stereotactic RT (SRT) for T2b tumors. Our department policy for patients with Stage III-IVB disease was combined treatment using concurrent chemotherapy and RT with accelerated fractionation (2-Gy daily fractions, 6 fractions weekly). However, starting from 2000, patients were encouraged to participate in the 2 trials organized by the Hong Kong Nasopharyngeal Cancer Study Group. Those with Stage T1-T4N2-N3M0 disease were randomized to RT alone or chemoradiotherapy (CRT) using the Intergroup-0099 regimen (2) of cisplatin in concurrence with RT followed by adjuvant cisplatin and fluorouracil. Patients in both treatment arms underwent RT with conventional fractionation (NPC-9901 trial [5]). Those with Stage T3-T4N0-N1M0 disease were randomized to RT alone with conventional fractionation, RT alone with accelerated fractionation, CRT (using the Intergroup0099 protocol) with conventional fractionation RT, or CRT with accelerated fractionation RT (NPC-9902 trial [13]). With these variations in treatment strategies, patients could be divided into 3 main groups: one hundred seventy-one patients were treated with the basic RT course to 70 Gy alone (Group 1), 55 received an additional boost of 5 Gy (Group 2), and 196 underwent cisplatin-based chemotherapy in a concurrent and/or sequential

Table 1. Patient characteristics, staging, and fractionation Comparison between groups Characteristic Age (y) Mean Range Gender Male Female Stage I-IIB III-IVB T category T1 T2a-T2b T3 T4 Fractionation Conventional Accelerated

Whole series (n = 422)

Group 1 (RT alone; n = 171)

Group 2 (RT + boost*; n = 55)

Group 3 (CRT; n = 196)

50 17–84

52 20–84

50 27–72

48 17–76

305 (72) 117 (28)

117 (68) 54 (32)

37 (67) 18 (33)

151 (77) 45 (23)

63 (15) 359 (85)

14 (8) 157 (92)

49 (89) 6 (11)

0 196 (100)

11 (3) 158 (37) 147 (35) 106 (25)

4 (2) 63 (37) 79 (46) 25 (15)

5 (9) 50 (91) 0 0

2 (1) 45 (23) 68 (35) 81 (41)

232 (55) 190 (45)

82 (48) 89 (52)

55 (100) 0

95 (49) 101 (52)

p 0.010 0.124 < 0.001 < 0.001

< 0.001

Abbreviations: RT = radiotherapy; CRT = chemoradiotherapy; SRT = stereotactic RT. Data presented as numbers of patients, with percentages in parentheses, unless otherwise noted. * Boost by brachytherapy in 22 and SRT in 33 patients.

Nasopharyngeal cancer: factors affecting late toxicities d A. W. M. LEE et al.

sequence (Group 3). The median overall treatment time for the whole series was 46 days (range, 36–57); conventional fractionation was used in 232 patients and accelerated fractionation in 190 patients.

Basic RT course All patients in this study underwent RT with 6-MV photons using three-dimensional conformal techniques throughout the whole course; none were treated with intensity-modulated RT. No differences in the technique were used in the 3 groups. The gross tumor volume was determined by the tumor extent delineated by the imaging and endoscopic findings at presentation. The clinical target volume (CTV) for the 70-Gy dose included the whole nasopharynx and the gross tumor volume with a 5–10-mm margin (if possible). The CTV aimed at 60 Gy covered high-risk local structures (including the parapharyngeal spaces, posterior one-third of the nasal cavities and maxillary sinuses, the pterygoid processes, base of skull, lower one-half of sphenoid sinus, anterior one-half of the clivus, and petrous tips), bilateral retropharyngeal lymph nodes, and upper lymphatic regions. The CTV aimed at 50 Gy covered the remaining potential sites of local infiltration up to the roof of the sphenoid sinus and bilateral cervical lymphatics down to the supraclavicular fossae. The planning target volume was based on the CTV with a 2-mm margin to allow for setup variations. The cervical region was treated with anterior-posterior opposed fields with a 2–3-cm shield to protect the midline structures; the doses to the larynx and esophagus were hence minimal. The important OARs (including the temporal lobes, brainstem, spinal cord, optic nerves and chiasm, pituitary gland, eyeballs, and cochlea) were delineated and the DVH recorded. The mean and maximal doses for these OARs were used for the analyses on the relationship of dose with late toxicity. The specification for critical OARs was that the maximal dose should not exceed (as far as possible) 54 Gy to the brainstem, optic nerves, and chiasm, 48 Gy to spinal cord, 6 Gy to the lens, and 68 Gy to temporal lobes (1% volume). All patients underwent treatment planning with the Helax TMS Treatment Planning System. The treatment plan consisted of four phases. Large fields were used for the first 40 Gy to encompass the whole tumor extent; progressive cone-down and realignment of the radiation beams were then arranged every 10 Gy. The ideal goal was to deliver $95% of the intended dose to 100% of the respective target volumes.

Additional boost Of the 55 patients given an additional boost, 22 received the boost using high-dose-rate intracavitary brachytherapy within 1 week of completion of the basic course of external beam RT. A Rotterdam nasopharyngeal applicator was inserted, 2 fractions of 2.5 Gy (6 h apart) were prescribed at 1 cm from the center of the plane of the 192 Ir source using the microSelectron-HDR machine. Another 33 patients received the boost with fractionated SRT. The treatment was planned using the Radionics Xplan, version 2.0, system and delivered using a 6-MV linear accelerator with a mini-multileaf collimator attached. The planning target volume was defined as the initial gross tumor volume of the primary tumor plus a 2-mm margin. The patients were immobilized with GillThomas-Cosman relocate-able head ring. Two weekly fractions of 2.5 Gy prescribed to the planning target volume were given on the Saturdays of the last 2 weeks of the basic course.

Additional chemotherapy For all 196 patients treated with chemotherapy, single-agent cisplatin (with a scheduled dose of 100 mg/m2 every 3 weeks for 2 to 3

1123

cycles, depending on the fractionation schedule and overall treatment time) was used during the concurrent phase. In addition, 111 patients were given adjuvant chemotherapy using a cisplatin-fluorouracil combination (with a scheduled dose of cisplatin 80 mg/m2 on Day 1 and fluorouracil 1,000 mg/m2/d on Days 1–4 every 4 weeks for three cycles). Induction chemotherapy, using a cisplatin-fluorouracil combination (with a scheduled dose of cisplatin 100 mg/m2 on Day 1 and fluorouracil 1,000 mg/m2/d on Days 1–5 every 3 weeks for 3 cycles), was given to 46 patients, and 33 received a cisplatingemcitabine combination (with a scheduled dose of cisplatin 80 mg/ m2 on Day 1 and gemcitabine 1,250 mg/m2 on Days 1 and 8 every 3 weeks for 3 cycles). The details of the regimens and guidelines for dose modifications have been previously published (2, 14, 15). Altogether 155 patients (79%) in Group 3 completed 6 cycles and 173 (88%) completed 5 or more cycles of chemotherapy. The numbers of cycles given during the concurrent and sequential phases are listed in Table 2.

Assessment and statistical analysis Patients were followed at least every 3 months during the first 3 years and every 6 months thereafter until death. Hormonal function tests (prolactin, cortisol, thyroxin, and thyroid-stimulating hormone) were performed annually. Late toxicity was evaluated prospectively in accordance with the protocols for the NPC-9901 (5) and NPC9902 trials (13) for the 210 patients accrued into the trials and the department guidelines for the remaining patients. The scoring criteria of the Radiation Therapy Oncology Group (16) were initially used during this study period, and the grading was retrospectively converted to that in accordance with the Common Terminology Criteria for Adverse Events, version 3.0 (17) for easy reference with contemporary series. The present analysis was confined to late toxicities at the nasopharyngeal region. The defining events included potentially serious toxicities (e.g., temporal lobe necrosis, cranial neuropathy, damage to the brainstem, spinal cord, or optic chiasm) of all grades, intermediate toxicities (e.g., endocrine dysfunction, soft tissue and bone necrosis) of Grade 2 or greater, and other toxicities (e.g., deafness, trismus, or eyeball damage) of Grade 3 or greater. Toxicities involving normal tissues at the neck or peripheral neuropathy due to chemotherapy were excluded from the present analyses. All events were measured from the date of beginning primary RT. To assess the toxicities attributable solely to the primary treatment, the observations were censored at the beginning of repeat RT for patients who underwent repeat RT for local relapse. The actuarial rates were calculated using the Kaplan-Meier method (18), and the differences were compared with the log–rank test (19). A multivariate analysis of different risk factors was conducted using the Cox proportional hazard model (20). Table 2. Cycles of chemotherapy given to group 3 Chemotherapy cycle None 1 2 $3

Concurrent phase

Sequential phase

11 (7) 75 (38) 110 (56)

7 (4) 6 (3) 8 (4) 175 (89)*

Data presented as number of patients, with percentages in parentheses. * Forty patients with accelerated fractionation had 2 cycles of concurrent chemotherapy and 4 cycles of sequential chemotherapy.

I. J. Radiation Oncology d Biology d Physics

1124

Furthermore, the relationship of radiation dose and chemotherapy with deafness (Grade 3 or greater) was tested by fitting a logistic regression model with a mean dose at the cochlea as a covariate and the effect of chemotherapy as a categorical factor. The dose–response curves using the associated regression coefficients for Groups 1 and 3 were plotted and drawn on the same figure for easy illustration of the differences. All statistical analyses were performed using Statistical Package for Social Sciences.

RESULTS Overall incidence of late toxicities Altogether 122 patients (28.9%) had one or more major late toxicities. The mean latent period was 2.2  1.7 years after the beginning of RT. The maximal severity was Grade 1-2 in 37 (8.8%), Grade 3 in 69 (16.4%), and Grade 4 in 16 (3.8%) patients. Of the 16 with Grade 4 toxicity, 12 (75%) had undergone CRT. The overall actuarial rate at 5 years was similar between Groups 1 and 2 (28% vs. 27%, p = 0.697) but was significantly greater in Group 3 than in Group 1 (37% vs. 27%, p = 0.009; Table 3 and Fig. 1). Accelerated fractionation did not result in a significant increase in the overall rate or damage of different OARs (Table 3, p > 0.23). Table 4 lists the significant factors on multivariate analyses with age, T category (T3-T4 vs. T1-T2), RT boost (yes vs. no), concurrent CRT (yes vs. no), and fractionation (accelerated vs. conventional fractionation) as covariates. Age was a strongly significant factor for overall toxicity. The hazard ratio was 1.04 per year increase (95% confidence interval [CI], 1.02–1.05; p < 0.001). The only significant treatment factor was CRT (hazard ratio, 1.99; 95% CI, 1.32–2.99; p = 0.001). Damage to different structures A total of 4 patients (0.9%) developed temporal lobe necrosis (TLN): 2 had unilateral and 2 bilateral involvement. Of the 4 patients, 3 were asymptomatic, but 1 patient (from Group 3) had Grade 4 severity. None of the patients in Group 1 developed TLN, but the 5-year rate increased to 4.8% in Group 2 (p = 0.015) and 1.3% in Group 3 (p = 0.177). An additional re-

Volume 73, Number 4, 2009

view of Group 2 showed that the 5-year rate was 8.3% among the 33 patients who received a SRT boost but that none of the 22 patients with a brachytherapy boost were affected. Additional analyses of the radiation doses in patients with a SRT boost showed that the mean additional dose to the temporal lobes caused by SRT was 0.08  0.06 Gy. A comparison of the four lobes with radiologic evidence of necrotic changes vs. the other 62 unaffected lobes showed no significant differences in the mean doses attributed to the basic RT course (15.8 vs. 16.2 Gy, p = 0.76) or SRT (0.12 vs. 0.08 Gy, p = 0.26). However, it was impossible to assess the summated doses because different planning systems were used for the two RT methods. Seven patients (1.7%) developed cranial neuropathy. The hypoglossal nerve was affected in 6 and the mandibular branch of the trigeminal nerve in 1 patient. The 5-year rate was significantly greater in Group 3 than in Group 1 (2.5% vs. 0%, p = 0.018). However, the proportion of patients with advanced T-category disease was greatest in Group 3 (Table 1), and the independent effect of chemotherapy could not be assessed by multivariate analyses because the coefficients did not converge. A total of 41 patients (9.7%) developed one or more hypothalamic-pituitary hormonal anomalies that required replacement medication. The hormonal dysfunction included secondary hypothyroidism in 28 (6.6%), secondary hypoadrenalism in 7 (1.7%), and hyperprolactineamia in 8 (1.9%) patients. The 5-year rate increased only slightly from 8.6% in Group 1 to 11.7% in Group 2 (p = 0.42) and 11.1% in Group 3 (p = 0.68). Five patients (1.2%) developed soft-tissue damage. Two had persistent ulcer/necrosis at the nasopharynx and three had pharyngeal damage causing dysphagia. The 5-year rate increased from 0.6% in Group 1 to 2.5% in Group 3; the difference was insignificant (p = 0.22). With exclusion of 21 ears with deafness at presentation, the analyses on hearing impairment included 823 evaluable ears. A total of 81 patients (19.2%) developed hearing loss that required a hearing aid and persisted for >90 days from the beginning of RT: forty-nine had unilateral and 32 bilateral

Table 3. Major late toxicity: 5-year actuarial rate Group 2 (%)

Group 1 (%) Toxicity Temporal lobe necrosis (G$1) Cranial neuropathy (G$1) Endocrine dysfunction (G$2) Soft tissue damage (G$2) Deafness (G$3) Eyeball damage (G$3) Trismus (G$3) Any complication

Group 3 (%)

CF (n = 82) AF (n = 89) All (n = 171) All (CF) (n = 55) CF (n = 95) AF (n = 101) All (n = 196) 0 0 9.3 0 16.7 0 0 21.9

0 0 8.1 1.2 25.9 0 0 31.7

0 0 8.6 0.6 21.5 0 0 27.1

4.8* 0 11.7 0 13.2 0 0 27.7

0 2.2 13.9 1.1 27.3y 1.3 0 38.4*

2.7 2.3 7.0 4.3 24.1 1.2 1.0 34.0

1.3 2.5* 11.1 2.5 26.2 1.3 0.5 37.1*

Abbreviations: CF = conventional fractionation (5 fractions weekly); AF = accelerated fractionation (6 fractions weekly); G = grade of severity according to Common Terminology Criteria for Adverse Events, version 3.0. * Significant increase compared with Group 1 by log–rank test (p < 0.05). y Borderline significance (p = 0.0524).

Nasopharyngeal cancer: factors affecting late toxicities d A. W. M. LEE et al.

Fig. 1. Overall major toxicity rate in different treatment groups.

involvement. The severity of the deafness was Grade 3 in 66 patients (15.6%) and Grade 4 in 15 (3.6%); 11 of the latter were from Group 3. The 5-year rate increased from 21.5% in Group 1 to 26.2% in Group 3, the difference was insignificant using the log–rank test (p = 0.11). However, because patients in Group 1 were older than those in Group 3 (Table 1; p = 0.01), when adjusted for age on multivariate analysis, concurrent chemotherapy was a significant factor (Table 3). The hazard ratio for deafness was 1.90 (95% CI, 1.16–3.09; p = 0.010). One patient had trismus and two developed a cataract (both had exceptionally high doses to the lens for extensive local infiltration into the orbits). None of the patients developed bone necrosis or damage to the optic nerve, optic chiasm, brainstem, or spinal cord. Analyses of radiation dose to OARs The following analyses on the radiation dose to the various OARs were based on patients irradiated to 70 Gy without any additional boost (i.e., Groups 1 and 3). Of the 713 evaluable

1125

ears, 103 had deafness of Grade 3 or greater. Analyses of the correlation with radiation dose using the logistic regression model showed that the mean dose to the cochlea was a significant factor (p = 0.004) affecting the probability of deafness with a coefficient estimate of 0.032. Figure 2 illustrates the dose–response curves for Groups 1 and 3. Multivariate analyses of the independent hazard attributed to the different treatment factors (Table 5) showed that, adjusted for age, the hazard ratio attributed to the mean dose was 1.03 (95% CI, 1.01–1.05; p = 0.005) and that to chemotherapy was 1.89 (95% CI, 1.23–2.88; p = 0.003). Fractionation was not a significant factor (p = 0.41). A comparison of the mean values in Group 1 showed significant differences in the mean doses to the cochlea between the affected and unaffected patients (56 vs. 51 Gy, p = 0.002). The 5-year actuarial deaf rate (Fig. 3) was significantly greater for those with a cochlea dose >50 Gy than for those with lower doses (Group 1, 18% vs. 7%; p = 0.001; and Group 3, 22% vs. 14%; p = 0.041). None of the 342 temporal lobes in Group 1 developed TLN. Their mean dose was 18.0  4.6 Gy, and the maximal dose was 68.2  1.6 Gy. In contrast, 2 (0.5%) of 392 lobes in Group 3 developed TLN. The mean dose in these 2 patients was 22.3 and 37.4 Gy, respectively. The 5-year TLN rate in Group 3 increased from 0 for 264 lobes with a mean dose of #22 Gy to 2.2% for 128 lobes with greater doses (p = 0.032). Of the 367 evaluable patients, 34 had one or more hypothalamic-pituitary hormonal anomalies. The mean dose to the pituitary gland was 55.9 Gy (range, 22.5–73.1). Multivariate analyses (Tables 4 and 5) showed that none of the treatment factors had a statistically significant effect. The hazard ratio attributed to the radiation dose was 1.01 (95% CI, 0.97–1.06). For critical organs without any observable toxicity (brainstem, spinal cord, optic chiasm, and optic nerves), the mean dose and the maximal point dose are listed in Table 6 to give an indication of the doses that were tolerated. However, no additional estimation of the maximal tolerable threshold dose could be performed. DISCUSSION Because of the close anatomic proximity to critical structures, NPC is one of the most difficult and challenging cancer sites for radiation oncologists. Despite some of the potential

Table 4. Multivariate analysis of factors affecting late toxicities (patient-based for whole series) HR (95% CI; p) Factor

Any toxicity

Deafness

Endocrine dysfunction

Age (per year) T category (T3-T4 vs. T1-T2) Chemotherapy (yes vs. no) Boost (yes vs. no) Fractionation (AF vs. CF)

1.04 (1.02–1.05; < 0.001) 0.89 (0.58–1.36) 1.99 (1.32–2.99; 0.001) 1.05 (0.54–2.04) 0.92 (0.61–1.37)

1.05 (1.03–1.07; < 0.001) 0.77 (0.47–1.28) 1.90 (1.16–3.09; 0.010) 0.62 (0.25–1.51) 0.99 (0.61–1.63)

1.01 (0.98–1.04) 1.02 (0.49–2.11) 1.13 (0.56–2.29) 1.17 (0.42–3.27) 0.58 (0.28–1.20)

Abbreviations: HR = hazard ratio; CI = confidence interval; other abbreviations as in Table 3. Multivariate analyses of temporal lobe necrosis, cranial neuropathy, and soft tissue damage: no additional models were fitted because coefficients did not converge.

1126

I. J. Radiation Oncology d Biology d Physics

Volume 73, Number 4, 2009

Fig. 2. Probability of deafness in relation to mean dose at cochlea and effect of chemotherapy.

weaknesses of the present study, including the retrospective data (except for the 210 patients accrued into randomized trials), treatment heterogeneity, and inadequate duration of observation for full assessment of long-term toxicity, our data do have unique strengths for contributing to the knowledge of late radiation damage. The present series of 422 patients (with a median follow-up of 4.4 years, and all survivors observed for $3 years, and 98 patients at risk for assessment at 5 years) give reasonably reliable data for assessing the 5-year actuarial rates. This is one of the largest series irradiated with a uniform three-dimensional conformal technique, providing detailed DVH data on normal structures in the nasopharyngeal region for assessing the dose relationship for various OARs. The present series has also provided an interesting opportunity to assess, not only the tolerance to RT alone, but also the increase in risk incurred with the addition of treatment strategies (including radiation dose escalation, chemotherapy, and acceleration fractionation). Although longer observation to give results at $10 years might give additional Table 5. Multivariate analysis on radiation dose and other factors for deafness and endocrine dysfunction (organ-based for patients without additional boost) Deafness Factor Age (per year) Mean dose to OARs (per Gray) Chemotherapy (yes vs. no) Fractionation (AF vs. CF)

HR (95% CI)

Endocrine dysfunction p

HR (95% CI)

p

1.04 (1.03–1.06) < 0.001 1.00 (0.97–1.03) 0.97 1.03 (1.01–1.05) 0.005 1.01 (0.97–1.06) 0.54 1.89 (1.23–2.88)

0.003 1.08 (0.53–2.20) 0.84

0.84 (0.55–1.27)

0.41

Abbreviations as in Tables 3 and 4.

0.58 (0.28–1.20) 0.14

Fig. 3. Comparison of actuarial deaf rate in patients with/without concurrent chemotherapy and mean cochlea dose >50 Gy vs. <50 Gy.

information, the 5-year rate is a useful starting point for reporting late toxicity. The present findings are important because clinicians should be cautioned of the risk and patients should be duly informed without delay. Compared with patients treated to a total dose of 70 Gy by RT alone (Group 1), the overall toxicity rate was significantly greater in patients given additional cisplatin-based chemotherapy (Group 3): 37% vs. 27% at 5 years (p = 0.009; Fig. 1). Concurrent chemotherapy was the only significant treatment factor affecting the overall toxicity rate on multivariate analyses. The hazard ratio increased by 1.99-fold (Table 4; p = 0.001). This concurs with the preliminary results from the NPC-9901 trial (5), which was the first report showing that CRT using the Intergroup-0099 regimen incurred significantly more late toxicity than RT alone (28% vs. 13% at 3 years; p = 0.024). Ototoxicity, particularly sensorineural hearing loss (SNHL), is a known complication of cisplatin (21). It is hardly surprising that concurrent cisplatin-based CRT incurred a significant 1.9-fold increase in the hazard of deafness (Tables 4 and 5; p # 0.01). This concurred with the observations from the confirmatory trials testing the Intergroup-0099 regimen (5, 13, 22). Prospective assessment of bone conduction hearing thresholds in 115 patients accrued to the trial by Wee et al. (4, 22) showed that the median threshold was significantly poorer in the CRT arm compared with the RT-alone arm, at both lower frequencies (0.5, 1, and 2 kHz: 23 vs. 15; p = 0.03) and a high frequency (4 kHz: 55 vs. 25; p < 0.01). Two previous studies on the factors affecting SNHL after RT for NPC reported conflicting results (23, 24). Both

Nasopharyngeal cancer: factors affecting late toxicities d A. W. M. LEE et al.

1127

Table 6. Radiation dose to critical organs without late damage Mean dose

Maximal dose

OAR

Group 1

Group 3

Group 1

Group 3

Brainstem Spinal cord Optic chiasm Optic nerves

29.1 (22.6-44.0) 18.5 (12.9-28.3) 42.0 (8.2-59.1) 25.9 (6.9-70.6)

31.6 (24.1-43.0) 18.9 (13.6-28.1) 44.3 (8.7-61.8) 28.7 (8.4-67.0)

50.5 (32.9-56.7) 48.7 (44.1-52.4) 52.9 (9.9-62.0) 41.6 (8.7-74.7)

51.3 (40.4-53.9) 48.3 (42.7-51.0) 53.9 (12.8-67.4) 43.7 (14.0-70.2)

Abbreviation: OAR = organ at risk. Data presented as mean values, with ranges in parentheses.

included patients irradiated with a two-dimensional technique, and the described dose to the inner ear was a rough estimation without actual contouring of the OARs and corresponding DVH data. The study by Kwong et al. (23) of 227 ears showed that older age, male gender, and the development of post-RT serous otitis media were associated with a significantly greater risk of SNHL, but that the radiation dose and use of chemotherapy were insignificant. However, only low-dose induction chemotherapy (cisplatin 100–185 mg/m2) had been used for 88 ears, and the radiation doses described were biologic equivalent doses using an a/b ratio of 3 Gy, as the prescribed tumor dose ranged from 2.5 to 3.5 Gy/fraction. In contrast, the study by Grau et al. (24) of 22 ears treated with RT alone showed that those with a cochlea dose >50 Gy had a significantly greater incidence of SNHL; age was not an independent factor. A retrospective study by Bhandare et al. (25) with reconstruction of treatment plans for 325 patients with various head-and-neck cancers showed that patient age and the dose to the cochlea were significant factors for SNHL; the 5-year rate increased to 37% for patients with a median dose to the cochlea >60.5 Gy compared with 3% for those with lower doses. The present analyses on deafness were based on gross clinical observation (taking persistent deafness to the severity of requiring hearing aid $90 days from beginning RT as the defining event). Although additional investigations were not routinely performed to differentiate the exact type of deafness (conductive or sensorineural), the present data give a good reflection of the findings in actual clinical practice. Multivariate analyses showed that, in addition to chemotherapy, patient age and cochlea dose were significant factors affecting deafness (Table 5). The hazard ratio increased by 3% per Gray increase in the mean cochlea dose. This significant dose relationship could be further illustrated by the dose–response curves (Fig. 2) and the actuarial rates (Fig. 3). The 5-year deafness rate increased from 7% among patients with a cochlea dose of #50 Gy by RT alone to 22% among those with greater doses and concurrent use of chemotherapy.

A significant dose relationship was illustrated for TLN. Thus far, no TLN had been observed among our 171 patients treated by RT alone and 22 patients treated with a boost using brachytherapy; however, the TLN rate increased to 8.3% at 5 years among the 33 patients with a boost using SRT, even though the additional prescribed dose was only 5 Gy in two fractions and the additional mean dose to the temporal lobes attributed to SRT was only 0.08  0.06 Gy. This exacerbation of risk concurred with the findings from a study from Stanford University (7) on a SRT boost of 7–15 Gy in a single fraction after external beam RT to 66 Gy. Excellent 5-year local control (98%) was achieved; however, substantial toxicity resulted. The crude incidence of TLN amounted to 12% (10 of 82) at a median follow-up of 3.4 years. For patients treated by CRT, more stringent control of the radiation dose to the temporal lobe is even more important. The findings from the present study showed that the 5-year TLN rate increased from 0% for 264 lobes with a mean dose of #22 Gy to 2.2% for 128 lobes with greater doses (p = 0.032). CONCLUSION The therapeutic margin for treatment of NPC is extremely narrow, and the importance of using the most conformal technique for maximal protection of the normal tissues cannot be overemphasized. The development of intensity-modulated RT is a major advance for improving the physical dose distribution. Although we strive to minimize the dose to the normal tissues, we must also know the maximal limit that we can safely risk to achieve better tumor coverage for patients with extensive locoregional infiltration, particularly tumor affecting/abutting neurologic tissues. Accurate knowledge of the tolerance doses is needed, not only for RT alone with conventional fractionation, but also for more aggressive treatment with CRT and/or accelerated fractionation. The ideal goal of achieving uncomplicated locoregional control for NPC remains a most fascinating challenge.

REFERENCES 1. Lee AWM, Perez CA, Law CK, et al. Nasopharynx. In: Halperin EC, Perez C, Brady L, editors. Principles and practice of radiation oncology. 5th ed. Philadelphia: Lippincott Williams and Wilkins; 2008. p. 820–857.

2. Al-Sarraf M, LeBlanc M, Giri PG, et al. Chemoradiotherapy versus radiotherapy in patients with advanced nasopharyngeal cancer: Phase III randomized Intergroup study 0099. J Clin Oncol 1998;16:1310–1317.

1128

I. J. Radiation Oncology d Biology d Physics

3. Baujat B, Audry H, Bourhis J, et al. Chemotherapy in locally advanced nasopharyngeal carcinoma: An individual patient data meta-analysis of eight randomized trials and 1753 patients. Int J Radiat Oncol Biol Phys 2006;64:47–56. 4. Wee J, Tan EH, Tai BC. Randomized trial of radiotherapy versus concurrent chemoradiotherapy followed by adjuvant chemotherapy in patients with American Joint Committee on Cancer/International Union Against Cancer stage III and IV nasopharyngeal cancer of the endemic variety. J Clin Oncol 2005; 23:6730–6738. 5. Lee AWM, Lau WH, Tung SY, et al. Preliminary results of a randomized study on therapeutic gain by concurrent chemotherapy for regionally-advanced nasopharyngeal carcinoma: NPC-9901 trial by the Hong Kong Nasopharyngeal Cancer Study Group. J Clin Oncol 2005;23:6966–6975. 6. Teo PM, Leung SF, Lee WY, et al. Intracavitary brachytherapy significantly enhances local control of early T-stage nasopharyngeal carcinoma: The existence of a dose-tumor-control relationship above conventional tumoricidal dose. Int J Radiat Oncol Biol Phys 2000;46:445–458. 7. Hara W, Loo BW, Goffinet DR, et al. Excellent local control with stereotactic radiotherapy boost after external beam radiotherapy in patients with nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2008;71:393–400. 8. Chen H, Tsai ST, Wang MS, et al. Experience in fractionated stereotactic body radiation therapy boost for newly diagnosed nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2006;66:1408–1414. 9. Lee AWM, Sze WM, Yau TK, et al. Retrospective analysis on treating nasopharyngeal carcinoma with accelerated fractionation (6 fractions per week) in comparison with conventional fractionation (5 fractions per week): Report on 3-year tumor control and normal tissue toxicity. Radiother Oncol 2001;58: 121–130. 10. Emami B, Lyman J, Brown A, et al. Tolerance of normal tissue to therapeutic irradiation. J Radiat Oncol Biol Phys 1991;21: 109–122. 11. Fleming I, Cooper J, Henson D, for the American Joint Committee on Cancer. AJCC cancer staging manual. 5th ed. Philadelphia: Lippincott-Raven; 1997. p. 31–39. 12. Sobin L, Wittekind C, for the International Union Against Cancer (UICC). TNM classification of malignant tumours. 5th ed. New York: Wiley-Liss; 1997. 13. Lee AWM, Tung SY, Chan AT, et al. Preliminary results of a randomized study (NPC-9902 trial) on therapeutic gain by

Volume 73, Number 4, 2009

14.

15.

16.

17.

18. 19. 20. 21. 22. 23.

24. 25.

concurrent chemotherapy and/or accelerated fractionation for locally-advanced nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 2006;66:142–151. Lee AWM, Yau TK, Wong DH, et al. Treatment of stage IV(AB) nasopharyngeal carcinoma by induction-concurrent chemoradiotherapy and accelerated fractionation. Int J Radiat Oncol Biol Phys 2005;63:1331–1338. Yau TK, Lee AWM, Wong DHM. Induction chemotherapy with cisplatin and gemcitabine followed by accelerated radiotherapy and concurrent cisplatin in patients with stage IV(A-B) nasopharyngeal carcinoma. Head Neck 2006;28: 880–887. Cox JD, Stetz J, Pajak TF. Toxicity criteria of the Radiation Therapy Oncology Group (RTOG) and the European Organization for Research and Treatment of Cancer (EORTC). Int J Radiat Oncol Biol Phys 1995;31:1341–1346. Cancer Therapy Evaluation Program. Common terminology criteria for adverse events, version 3.0. Division of Cancer Treatment and Diagnosis, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, December 12, 2003. Available from: http://ctep.cancer.gov/ forms/CTCAEv3.pdf. Accessed June 30, 2006. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–481. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Br J Cancer 1977;35:1–39. Cox DR. Regression models and life tables. J R Stat Soc B 1972; 34:187–220. Blakley BW, Myers SF. Patterns of hearing loss resulting from cisplatinum therapy. Otolaryngol Head Neck Surg 1993;109: 385–391. Low WK, Toh ST, Wee J, et al. Sensorineural hearing loss after radiotherapy and chemoradiotherapy: A single, blinded, randomized study. J Clin Oncol 2006;24:1904–1909. Kwong D, Wei W, Sham J, et al. Sensorineural hearing loss in patients treated for nasopharyngeal carcinoma: A prospective study of the effect of radiation and cisplatin treatment. Int J Radiat Oncol Biol Phys 1996;36:281–289. Grau C, Moller K, Overgaard M, et al. Sensori-neural hearing loss in patients treated with irradiation for nasopharyngeal carcinoma. Int J Radiat Oncol Biol Phys 1991;21:723–728. Bhandare N, Antonelli P, Morris C, et al. Ototoxicity after radiotherapy for head and neck tumors. Int J Radiat Oncol Biol Phys 2007;67:469–479.