• radiation-induced sarcoma;
  • nasopharyngeal carcinoma;
  • radiation therapy;
  • second primary malignancy;
  • treatment modality


  1. Top of page
  2. Abstract


The increasing incidence of radiation-induced sarcoma (RIS) has become a significant problem that can limit long-term survival. The objective of the current study was to analyze the clinicopathologic characteristics, treatment outcomes, and prognostic factors of RIS after radiotherapy for nasopharyngeal carcinoma (NPC).


Institutional electronic medical records of patients with NPC who received definitive radiotherapy between February 1964 and 2003 were reviewed. Fifty-three patients who developed RIS and fulfilled the study criteria were included.


The median follow-up after a diagnosis of RIS was 15.5 months (range, 0.4-90.3 months), and the median latency between radiotherapy for NPC and an RIS diagnosis was 9.3 years (range, 3.2-26.6 years). Fibrosarcoma was the most frequent histologic type observed, followed by osteosarcoma, and malignant fibrous histiocytoma. The 3-year overall survival (OS) rate for 49 patients who received treatment was 32.4%, and the median survival was 21.2 months (95% confidence interval, 8.7-33.8 months). The median OS was 41.3 months, 8.4 months, and 11 months for the complete resection group, the incomplete resection group, and the chemotherapy group, respectively (P<.0001). The only independent predictive factor that was associated with better OS was complete surgical resection.


This retrospective study confirmed the rarity and poor prognosis of RIS in patients with NPC. Complete surgical resection was a significant prognostic factor for survival. The authors concluded that long-term follow-up is necessary for the early detection of RIS in patients with NPC. Cancer 2010. © 2010 American Cancer Society.

It has been recognized for a long time that the modality used to treat the majority of patients with cancer, namely, ionizing radiation, induces sarcomas; indeed, the first report of these sarcomas was published in 1922.1 A poor prognosis for patients with radiation-induced sarcoma (RIS) has been reported by several groups.2-4 Although advancements in cancer treatment have improved patients' survival rates, an increased incidence of RIS also has been observed.5 Consequently, RIS has become a critical problem that can limit long-term survival and hinder quality of life.

Most published studies examining RIS have described tumors that occur after treatment for breast cancer, lymphoma, pelvic cancer, or Ewing sarcoma.5-8 It has been documented that the anatomic site of the sarcoma has important implications for treatment and outcome.9 To date, most RIS studies have included a heterogeneous population of patients with a mixture of primary malignancies and with the subsequent RIS occurring at a variety of sites. Reports detailing RIS in patients who were treated for head and neck cancer are relatively few, particularly in patients with nasopharyngeal carcinoma (NPC).10-12 Studies reporting the therapeutic management and outcome of RIS in patients with NPC have been limited to case reports and a few small case series, making it difficult to evaluate the prevalence and the best treatment strategy for this disease.13-19 The objectives of the current study were to report our institutional experience with a relatively large group of patients who developed RIS after radiotherapy (RT) for NPC and to analyze the clinicopathologic characteristics, treatment modalities, and potential factors that influenced their survival.


  1. Top of page
  2. Abstract

Between February 1964 and 2003, 39,118 patients with NPC received definitive RT at the Cancer Center, Sun Yat-sen University, Guangdong, China. Eligibility for this retrospective study was based on the criteria for RIS published by Cahan et al20 as modified by Arlen et al21: 1) a prior history of RT, 2) the occurrence of sarcoma within the previously irradiated field, 3) histologic confirmation of the sarcomatous nature of the postirradiation lesion, and 4) a latency between irradiation and a second primary sarcoma of at least 3 years. After reviewing the institutional electronic medical records, 53 patients who developed RIS and fulfilled these criteria were included in this study.

The patients' main characteristics at diagnosis of NPC are summarized in Table 1. The median age at diagnosis of NPC was 41 years (range, 13-64 years), and the ratio of men to women was 3.1:1.0. Patients who received treatment before 1986 were staged according to the Changsha staging system, and patients who received treatment after 1986 were restaged according to the 2002 American Joint Committee on Cancer (AJCC) staging system. The radiation techniques that were used in this study have been described in detail in our previous reports.22 RT was delivered with orthovoltage equipment in 8 patients and with megavoltage equipment in 45 patients. The median radiation dose to the nasopharyngeal and neck region was 68 grays (Gy) (range, 54.2-80 Gy) and 51.2 Gy (range, 33.1-78 Gy), respectively. The radiation dose at the site of RIS ranged from 50 Gy to 80 Gy, and the median dose (66 Gy) was selected as the cutoff between low-dose versus high-dose exposure groups. Patients who had received megavoltage radiation had received a higher dose (median, 68 Gy) than patients who had received orthovoltage radiation (median dose, 55.2 Gy; P = .001). The daily fraction size was 2 Gy delivered using Cobalt-60 or 6-megavolt photons in most patients. Four patients received neoadjuvant chemotherapy, which included alkylating agents in 3 of these patients.

Table 1. Patient Characteristics at the Time of Nasopharyngeal Carcinoma Diagnosis (N=53)
CharacteristicNo. of Patients%
  • NPC, nasopharyngeal carcinoma; WHO, World Health Organization; 3DCRT, 3-dimensional conformal radiotherapy; IMRT, intensity-modulated radiotherapy; RIS, radiation-induced sarcoma; Gy, grays.

  • a

    A radiation machine was used to irradiate the RIS region.

Age at NPC diagnosis, y  
Excessive alcohol intake  
Family history of cancer  
Calendar period at NPC diagnosis  
TNM stage  
Histologic type  
 WHO type III53100
 WHO type I and II00
Radiation machinea  
 Orthovoltage x-rays815.1
 Megavoltage x-rays59.4
Radiation technique  
 Conventional radiotherapy53100
Radiation course  
Radiation dose to RIS region, Gy  
Alkylating agents chemotherapy  
Local or regional recurrence  

Response to treatment for RIS was evaluated 3 months after treatment in accordance with established guidelines for patients with solid tumors.23 The cutoff date for the last follow-up was July 31, 2009 for the censored data analysis. Follow-up was calculated from the time of diagnosis of RIS to the date of last contact. Overall survival (OS) was calculated from the date of RIS diagnosis to the date of either death or last follow-up. The Kaplan-Meier method was to analyze OS, the log-rank test was used to examine group differences, and a Cox regression model was used for multivariate analysis. Differences between radiation doses were tested with independent-sample t tests, and differences in proportions were compared by using chi-square tests. All statistical analyses were performed using the SPSS software package (version 13.0; SPSS Inc., Chicago, Ill). P values <.05 were considered statistically significant.


  1. Top of page
  2. Abstract

Characteristics of RIS

We estimated that the crude incidence of RIS after definitive RT for NPC was approximately 0.14%. The median age of patients at the time of RIS diagnosis was 50 years (range, 18-68 years). The locations and histologic subtypes of RIS are presented in Table 2. RIS originated in bone in 9 patients and originated in soft tissue in 44 patients. Of the 8 patients who received orthovoltage radiation, 1 patient developed a bone sarcoma. Similarly, 8 patients developed bone RIS in the group that received megavoltage radiation (P = .589). The clinically determined site of RIS was superficial (cutaneous or subcutaneous) in 14 patients and deep in 39 patients. In the orthovoltage radiation group, only 1 patient (12.5%) developed a deep sarcoma, whereas 38 patients (84.4%) developed deep RIS in the megavoltage radiation group (P < .0001).

Table 2. Locations and Histologic Types of Radiation-Induced Sarcoma (N=53)
CharacteristicNo. of Patients%
Location of disease  
 Nasal cavity and paranasal sinuses1833.9
  Hard palate11.9
  Bucca cavioris11.9
 External auditory11.9
 Soft palate11.9
Histologic type  
 Malignant fibrous histiocytoma713.2
 Malignant neurilermmoma23.8
 Meningeal sarcoma11.9

Nasal cavity and paranasal sinuses were the most common RIS sites followed by neck and gingiva. Fibrosarcoma was the most frequent histologic type, followed by osteosarcoma and malignant fibrous histiocytoma. Grading of the sarcomas revealed that 6 tumors were grade 1, 12 tumors were grade 2, and 35 tumors were grade 3. Lymph node metastasis was observed in 3 patients, and no distant metastasis was observed at the time of RIS diagnosis. According to the 2002 AJCC staging system, 6 patients (11.3%) had stage I disease, 18 patients (34%) had stage II disease, 26 patients (49.1%) had stage III disease, and 3 patients (5.7%) fulfilled the criteria for stage IV disease.

Latency From RT to RIS

The median latency between RT initiation for NPC treatment and the diagnosis of RIS was 9.3 years (range, 3.2-26.6 years). RIS occurred within 5 years of RT in 15.1% of patients, within 10 years in 56.6% of patients, and within 15 years in 86.8% of patients. In patients who had received RT doses >66 Gy to the RIS region, the median latency period was 7.4 years (95% confidence interval [CI], 5.9-8.8 years), whereas in patients who received less than 66 Gy, the median latency was 10.6 years (95% CI, 9.2-12.0 years; P = .006). Disease latency was shorter for patients who had received megavoltage radiation than for those who had received orthovoltage x-rays (P = .039). The median latency was 5.6 years (95% CI, 5.5-5.6 years), 8.5 years (95% CI, 6.5-10.6 years), and 11.7 years (95% CI, 9.9-3.6 years) for patients who had received megavoltage x-rays, Cobalt-60, and orthovoltage x-rays, respectively. Latency did not differ significantly between patients who developed RIS in bone versus RIS in soft tissue (P = .424). Latency also was not influenced by sex, age at the time of RT, smoking, alcohol intake, family history, TNM stage of NPC, radiation course (split or continuous), chemotherapy, or histologic type of sarcoma.

Treatment of RIS

Of the 53 patients with RIS who met the study criteria, 4 patients refused treatment, and 9 patients received palliative chemotherapy alone because they either had nonresectable lesions or rejected surgery. Thirty-nine patients underwent surgery, including 31 patients who underwent surgery only, 4 patients who underwent surgery and received chemotherapy, and 4 patients who underwent surgery and received RT. One patient was received combined RT and chemotherapy without undergoing surgery.

Complete resection was achieved in 30 of 39 patients (76.9%), whereas 9 of 39 patients (23.1%) had macroscopically incomplete surgical resection or gross residual disease. Two patients (1 who underwent complete resection and 1 who underwent incomplete resection) who had radiation encephalopathy did not survive the postoperative period because of hemorrhage and a failure to regain consciousness after general anesthesia. Five patients received RT at a median dose of 60 Gy (range, 56.0-60.3 Gy). Fourteen patients received chemotherapy that involved various regimens, including combined doxorubicin, ifosfamide, dacarbazine, and mesna in 7 patients; combined ifosfamide and cisplatin in 4 patients; combined docetaxel and cisplatin in 2 patients; and high-dose methotrexate in 1 patient. One of these 14 patients died from hemorrhagic shock during chemotherapy.

Follow-Up and Outcome

Because 3 patients did not survive the treatment period, short-term response data were available for 46 of the 49 patient, including 29 patients who achieved a complete response, 10 patients who achieved a partial response, 3 patients who had stable disease, and 4 patients who had progressive disease. The median follow-up after the diagnosis of RIS was 15.5 months (range, 0.4-90.3 months). Of the 49 patients who received treatment, 2 patients (4.1%) developed distant metastasis in bone, and 1 patient (2%) developed a third primary leiomyosarcoma in the maxillary sinus. Thirty-nine patients underwent surgery, and 19 patients (48.7%) developed locally recurrent disease. The interval between curative treatment and presentation of recurrence ranged from 1.5 months to 43.4 months, with a median interval of 6.5 months. The 1-year and 3-year recurrence rates were 44.2% and 57.5%, respectively. Six patients underwent another surgery for their recurrence.

At the time of the current analysis, 3 patients (6.1%) in the original study cohort had failed to return for follow-up after 5.6 months, 25.2 months, and 26.9 months, respectively. Thirteen patients (26.5%) remained alive, including 8 patients who were without disease and 5 patients who had disease progression. Thirty-three patients (67.3%) had died, including 28 who died from progressive disease, 3 who died during the treatment period, 1 who died from a third RIS, and 1 who died from treatment-related complications. The 3-year OS rate for the 49 patients who received treatment was 32.4%, and the median survival was 21.2 months (95% CI, 8.7-33.8 months) (Fig. 1).

thumbnail image

Figure 1. Overall survival is illustrated for patients with radiation-induced sarcoma who received treatment (N = 49).

Download figure to PowerPoint

Prognostic Indicators of Survival

Patients were classified into 3 groups according to treatment: complete surgery (30 patients), incomplete surgery (9 patients), and chemotherapy alone (9 patients). The median OS was 41.3 months (95% CI, 26.9-55.6 months) in the complete surgery group, 8.4 months (95% CI, 3.6-13.1 months) in the incomplete surgery group, and 11.0 months (95% CI, 3.1-18.9 months) in the chemotherapy alone group. The 3-year OS rates were 50.5% for patients in the complete surgery group and 0% for patients in the incomplete surgery and chemotherapy alone groups (P<.0001) (Fig. 2).

thumbnail image

Figure 2. Overall survival is illustrated for patients with nasopharyngeal carcinoma who underwent complete surgical resection, underwent incomplete surgical resection, or received chemotherapy alone (P < .0001).

Download figure to PowerPoint

In addition to treatment modality, the site of RIS, histologic grade, and TNM stage also influenced outcome in our univariate analysis. Univariate analysis also revealed that sex, age, length of latency period, origin of the sarcoma (bone or soft tissue), and histologic type did not affect survival (Table 3). In the multivariate analysis, treatment modality was the only independent predictive factor for OS. Those patients who underwent complete surgical resection had better survival than patients who underwent incomplete surgery or no surgery (P = .043).

Table 3. Analysis of Prognostic Factors for Survival (N=49)
FactorNo. of Patients3-Year OS, %UnivariateMultivariate
  1. OS indicates overall survival; RIS, radiation-induced sarcoma; TNM, tumor, lymph node metastasis classification.

Age at RIS, y    
Latency, y    
Histologic type    
Histologic grade    
Anatomic site    
Site of origin    
 Soft tissue4035.0  
TNM stage group    
Treatment modality   
 Complete surgery3050.5<.0001.043
 Incomplete surgery90  


  1. Top of page
  2. Abstract

The management and treatment outcomes from a large series of patients with RIS after NPC have not been well documented previously. To our knowledge, the series in the current study was the largest published to date from a single institution in which RIS was analyzed in patients with NPC. The patients who successfully underwent complete surgical resection survived significantly longer than the patients who underwent incomplete resection or no surgery. The site of RIS, histologic grade, and TNM classification also influenced outcomes in univariate analysis.

Radiation-induced malignancies in the head and neck are uncommon. When they do occur, squamous cell carcinoma (SCC) is the most frequent histologic type of second primary malignancy.10 RIS is particularly rare, with a reported incidence in adults ranging from 0.03% to 0.8%.6, 24, 25 A dose-dependent response between RT and RIS incidence has been demonstrated for some malignancies.26-28 For example, Kuttesch et al26 reported that the cumulative incidence rate of secondary sarcoma was radiation dose-dependent in children with Ewing sarcoma. For their study, the absolute risk of RIS among patients who had received ≥60 Gy was 4-fold greater than the risk for patients who received doses of 48 to 60 Gy. RT with a full course of 65 to 70 Gy is the main treatment modality for patients with NPC. Therefore, theoretically, patients with NPC would be expected to have a high chance of developing RIS in organs in the irradiated fields. The rarity of reports addressing RIS after NPC treatment can be explained by the various sensitivities of different organs to RT-induced carcinogenesis. In the current study, our estimated incidence of RIS was comparable to that reported by Kirova et al.6 The incidence of 0.14% may be lower than the actual incidence, because some patients did not return for follow-up, and some may have died before the presentation of RIS. In addition, the short follow-up period for many patients who received megavoltage x-rays also may have led to an underestimation of the RIS incidence.

Not all sarcomas that arise after RT are induced by irradiation, but it is impossible to differentiate RIS from primary sarcomas. The lower limit of the latency period for establishing the diagnosis of RIS is controversial and ranges from 6 months to 8 years.18-21, 29 In our study, we used the most common latency criteria of 3 to 4 years as reported by Arlen et al.21 The median latency in our series, 9.3 years, was consistent with data from other studies, in which the median latency ranges from 7.6 years to 22 years.3-11, 29 The shorter period of 7.6 years reported by Kuttesch et al26 may be explained by the younger age of their patients. In addition, the administration of chemotherapy may shorten the interval between RT for the first cancer and the subsequent presentation of RIS. However, we produced no evidence that chemotherapy or age had an impact on the timing of RIS in our series. In a study by Wiklund et al,30 a total radiation dose >36 Gy resulted in a median latency of 7.4 years, whereas a dose <36 Gy resulted in a latency of 14.4 years, supporting a correlation between dose and latency. This purported correlation between radiation dose and latency was confirmed by our results.

Our results are consistent with several previous studies suggesting that there may be a longer latency interval in patients who receive orthovoltage radiation compared with those who received megavoltage radiation.9, 30 Our results suggest that this difference may be caused at least in part by the lower dose given to the patients who received orthovoltage radiation. Our study also suggested that the latency of RIS after megavoltage x-ray therapy was not only shorter than that after orthovoltage x-ray therapy but also was shorter than the latency after Cobalt-60 therapy. However, the limited number of patients and the shorter follow-up of patients who received megavoltage x-rays did not permit us to reach a definitive conclusion in this regard. There were significantly fewer patients with cutaneous or subcutaneous RIS in the megavoltage group than in the orthovoltage group, probably because of a larger skin-sparing effect of megavoltage radiation. In addition, soft tissue adjacent to bone receives a greater absorbed dose with megavoltage radiation than with orthovoltage radiation because of the Compton effect.4

Despite the small number of reports, a wide variety of histologic subtypes has been reported. The most frequent histologic subtypes for de novo sarcomas are malignant fibrous histiocytoma, liposarcoma, and leiomyosarcoma, which are different from those commonly classified as RIS.31 In our study, fibrosarcoma was the most frequent histologic type of RIS followed by osteosarcoma and malignant fibrous histiocytoma. These results contrast with previous studies in which malignant fibrous histiocytoma and osteosarcoma reportedly were the most frequent types.4, 9, 29 The histologic subtype of RIS may be related to the primary tumor type and other risk factors. For instance, Kirova et al6 reported that angiosarcoma was the most frequent subtype of RIS to develop after breast carcinoma treatment by irradiation. In addition to RT, lymphedema has been implicated as a potential causative factor in the development of angiosarcoma.32 Lagrange et al reported that most RIS was high grade, as observed in our study, in which 66% were grade 3.33 Regarding RIS site, the proportion of sarcomas that arose from the maxillary sinus in the current study cohort (18.9%) was consistent with the data reported by Wang et al.17

RIS traditionally has been viewed as an aggressive tumor with a generally poor prognosis. The 3-year OS rate for patients with RIS in our series (32.4%) was within the lower end of the wide range reported previously (8%-60% at 5 years).3-13, 34 The best survival outcome was reported by Tabone et al,35 who reported that the OS and event-free survival rates for 23 patients with radiation-related osteosarcoma at 8 years were 50% and 41%, respectively.

The prognosis for patients with RIS is related to the tumor site, because the site can limit a surgeon's ability to resect the tumor. Inoue et al9 reported that the 5-year survival rate for patients with resectable peripheral lesions (extremities, including proximal femur and hip) was 68.2%, whereas the rate for patients with central lesions (pelvis, head and neck, and ribs) was 27.3%. Mark et al reported the poor survival of patients with RIS, including a 5-year OS rate of 8% for those with sarcomas in the head and neck.34 In addition, the survival rates for patients with RIS in the head and neck was poor relative to the rates for patients with de novo sarcomas in head and neck. In a study by Patel et al,10 the 5-year disease-free survival rate in the RIS group was significantly worse than that in the de novo group (31% vs 54%).

Surgery is the accepted standard treatment for most sarcomas. Our multivariate analysis revealed that complete surgical resection was the only independent predictive factor for better survival, a finding that is consistent with other reports.2, 3, 29 Cha et al29 reported a 5-year survival rate of 54% for patients who achieved a microscopically negative resection. Inoue et al9 also reported a relatively good 5-year survival rate of 62.5% for patients with radical margins. The lower 3-year OS rate of 50.5% in our complete resection group indicates that the prognosis for patients with RIS after NPC is poor, even if they achieve complete resection with negative margins. This relatively poor prognosis may be because of the difficulty of performing surgery in the head and neck region and the lack of adjuvant therapy.

The extent and adequacy of the margin of resection could determine the local recurrence rate and overall survival for sarcomas. Resection of head and neck tumors cannot be performed using conventional wide surgical margins as are used in extremities given the profound risk of causing functional morbidity because of the close proximity of these tumors to critical structures, such as the carotid artery and the skull base. Furthermore, it is difficult to determine surgical margins in the head and neck. Moreover, the risk of surgical complications in a highly irradiated area is elevated because of poor healing and radiation encephalopathy in patients treated for NPC. Sun et al36 reported that 5 patients who developed postirradiation tongue cancer after NPC died from postoperative coma after general anesthesia; 4 of these patients had developed radiation encephalopathy before their RIS diagnosis. Similarly, as our study found, 2 patients with radiation encephalopathy died postoperatively, having suffered hemorrhage and failing to regain consciousness.

It has been reported previously that patients with RIS who underwent surgery plus received chemotherapy had better survival outcomes than patients who either underwent surgery alone or received chemotherapy alone.37 However, only 4 patients underwent surgery plus received chemotherapy in our series. This relatively small portion of cotreated patients may explain at least in part why the survival in our series was worse than that reported elsewhere in the literature. The contribution of adjuvant RT to the treatment of RIS in our series is unknown, because the number of patients who received adjuvant RT was too limited for analysis.

Although a complete resection with negative margins seems to offer the best chance for long-term survival, a substantial portion of patients with locally advanced sarcomas are not able to undergo complete resection. In the current study, we observed that the primary treatment received by patients who had unresectable RIS was palliative chemotherapy, and there were no long-term survivors in that group. Only 1 such patient received combined RT and chemotherapy, and that patient survived for only 4.3 months, which was shorter than the survival in the majority of patients who received chemotherapy alone. Reirradiation of second primary cancers in the head and neck has been reported as feasible in recent years, particularly for SCC.38, 39 However, relative to SCC, sarcoma is less sensitive and sometimes even is resistant to RT and chemotherapy. Furthermore, a high cumulative radiation dose could result in serious organ injury. Thus, currently, the reirradiation treatment modality should remain under careful consideration for patients with RIS, because further treatment may do more harm than good. Further studies with larger samples will be needed to investigate the efficiency of reirradiation.

In recent years, intensity-modulated RT (IMRT) and 3-dimensional conformal RT (3DCRT) have been used increasingly in the treatment of head and neck cancers. Relative to conventional RT, IMRT/3DCRT exposes a greater volume of normal tissues to low doses of radiation. Accordingly, Hall and Wuu40 estimated that IMRT would lead to an increase in the incidence of second malignancies by a factor of 2 compared with conventional RT. Between 2001 and 2003, we used IMRT/3DCRT to treat 248 patients with NPC at our institution, none of whom had developed RIS at the time of this analysis. However, a longer follow-up and a larger series of patients will be needed to determine whether IMRT indeed increases the incidence of second malignancy.

In conclusion, the current retrospective study confirmed the rarity and poor prognosis of RIS after treatment for NPC. Complete resection seems to offer the best chance for long-term survival. Careful and long-term follow-up is necessary for the early detection of RIS in patients with NPC after definitive RT.


  1. Top of page
  2. Abstract