Chen Xiaofang, Chen Yiming, Bian Xiaohe, Ji Tong

Department of Otolaryngology, Second Affiliated Hospital of Fujian Medical University, Quanzhou 362000, Fujian, China (Chen Xiaofang); Department of Oral and Maxillofacial-Head and Neck Oncology, Ninth People's Hospital, School of Medicine, Shanghai Key Laboratory of Stomatology & Shanghai Research Institute of Stomatology, National Clinical Research Center of Stomatology, Shanghai Jiao Tong University, Shanghai 200011, China (Chen Yiming, Bian Xiaohe, Ji Tong).

[Abstract] Objective: To investigate factors that benefit patients with head and neck osteosarcoma (HNOS) before death. Methods: Patients admitted in our department from 2007 to 2015 who died during follow-up after surgery for HNOS were selected. Information concerning demographics, disease, treatment and follow-up were collected and analyzed. Results: A total of 77 patients died at a median age of 42 years old, of whom 33 were male and 44 female. Thirty-nine tumors were identified in the maxilla and skull base region while 38 in the mandible. Positive surgical margins were observed in 7 cases. Twenty patients received adjuvant chemotherapy, and 36 received postoperative radiotherapy. Local recurrence, as an initial cause of death, occurred in 25 cases, whilst metastasis occurred in the rest. Conclusion: Patients with HNOS benefit from negative surgical margins and adjuvant chemotherapy before death. Patients above 30 years old have longer disease-free survival, and patients with tumors of low pathological grade have longer overall survival. Tumor size and postoperative radiotherapy are negative prognostic factors. Frequent follow-up should be conducted for at least 48 postoperative months.

[Key words] Head and neck osteosarcoma; Cancer specific death; Local recurrence; Metastases

Introduction

Osteosarcoma, a condition which mainly affects teenagers and young adults, is the most common malignant bone tumor that often occurs in the extremities. It has an annual incidence of 1/100 000 in general population[1-2]. Head and neck osteosarcoma (HNOS), which tends to occur in the third or fourth decade, accounts for less than 10% of all osteosarcomas[2-3].

Due to the low annual incidence of HNOS, treatment strategies for HNOS are mostly based on retrospective studies or meta-analysis conducted by large research centers or national institutes[4-5]. Surgery is regarded as the main modality in the management of HNOS[6]. Adjuvant chemotherapy is applied with the purpose of reducing the risk of metastases, while its effect, to some extent, has been proven by some retrospective studies[2]. Adjuvant postoperative radiotherapy was conducted for cases with close or positive surgical margins, although the effect of radiotherapy remains uncertain[6-7].

Survival data of HNOS varies from different research institutes, the 5-year overall survival (OS) rates in the last two decades range from 59.7% to 64.96%[3-4]. Local recurrence (LR) and metastases (MT) of tumor lead to cancer specific death of patients with HNOS[2-4]. MT has been reported in 10%～20% of patients with HNOS[8-10], while LR predominates in HNOS with a reported incidence of 17%～70%[11-12].

Previous publications regarding the prognosis and treatment strategies of HNOS seldom focused on cancer specific death of HNOS patients[1-12]. A large number of HNOS patients were enrolled in our institute annually. Information relating to demographics, disease, treatment and follow-up was collected. This paper mainly focuses on patients, diagnosed as HNOS and treated in our institute from 2007 to 2015, who died of HNOS during the follow-up period.

Methods

A retrospective study was conducted in patients diagnosed as HNOS and who received surgery in the Department of Oral & Maxillofacial-Head & Neck Oncology in our institute from 2007 to 2015. All patients had signed informed consent forms to authorize the use of personal information in the current study. Patients who died during the follow-up period were selected and analyzed. Information related to demographics, disease, treatment and follow-up were collected for further study. All dead patients were divided into 3 age groups: ≤30 years old, 31～60 years old, and >60 years old. According to surgical samples and image materials, the longest diameter of the individual tumor was detected. Those tumors were then divided into 2 groups using the median value of the longest diameter as a border. The pathological standards referred to the WHO Classification of Tumors of Soft Tissue and Bone (Version 2013). All pathological samples had been checked by two experienced pathologists.

As for routine follow-up, patients were informed to visit the outpatient department for clinical examination every 3 months during the first 12 postoperative months (PMs). The frequency was every 4 months within the second 12 PMs, every 5～6 months for the third 12 PMs, and every 6～8 months for the rest of follow-up period. Enhanced CT scans were performed for the head and neck region as well as for the chest every 5～8 months. During the follow-up period, first LR or first MT observed clinically was defined as initial death cause.

Disease-free survival (DFS), OS, LR and tumor MT were statically analyzed. Characteristics like gender, age, tumor site, tumor size, surgical margin, pathological grade and treatment were reviewed to analyze OS and DFS. Log-rank test was used to check the difference.P<0.05 was considered statistically significant.

Results

A total of 183 HNOS patients were treated during the mentioned time in our institute. A total of 157 were followed up. Seventy-seven patients died during the follow-up period. According to follow-up records, all deaths were cancer specific. Of the 77 dead patients, 33 were male and 44 were female. The median age was 42 years old (Min: 9 years old, Max: 78 years old) (Table 1).

Table1.InformationonDemographics,DiseaseandTreatmentofPatientsDiedofCancer

VariableNAge (years) ≤3023 31～6041 >6013Gender Male33

(Table 1 continues on next page)

(Continued from previous page)

VariableN Female44Tumor site Maxilla and skull base39 Mandible38Pathological Subtype Conventional54 Secondary14 Small cell3 Low grade central2 Unknown4Pathological grade I (well differentiated)2 III (poorly differentiated)71 Unknown4Tumor size (cm) ≤451 >426Surgical margin Negative70 Positive7Treatment strategy Surgery only31 Surgery+CT8 Surgery+CT+RT12 Surgery+RT26CT given Yes20 No57RT given Yes36 No41 Total77

CT indicates chemotherapy;RT indicates radiotherapy

As is shown in table 1, tumors were observed in the maxilla and skull base in 39 cases and in the mandible in 38 cases. The median value of tumor size (as indicated by the longest diameter) was 4 cm according to scans and surgical tumor samples. Tumors≤4 cm were seen in 51 cases, while tumors>4 cm were seen in 26 cases. HNOS in 2 cases were reported as grade I; 71 cases were reported as grade III, and 4 cases were of an unknown grade according to pathological results.

As for treatment, 31 cases underwent surgery only, 8 underwent surgery and adjuvant chemotherapy, 26 underwent surgery and postoperative radiotherapy, and 12 underwent surgery followed by adjuvant chemoradiotherapy. Seventy cases had negative surgical margins, 7 cases had positive surgical margins. No case had close margin (Table 1).

During the follow-up period, the median value of DFS was 9.98 months (Min: 0.22 months, Max: 51.18 months, Table 2), the median value of OS was 14.06 months (Min: 2.25 months, Max: 57.51 months, Table 3). MT, as the initial cause of death, occurred in 52 cases. It often occurred in the first 12 PMs (31, 59.62%, Table 4). The most frequently seen predilection site of MT was the lungs, and MT sometimes occurred in multiple sites (Table 5). LR, as the initial cause of death, was detected in 25 cases. Of these, 18 (72.00%) cases experienced LR throughout the first 12 PMs, while 1 case in the fourth 12 PMs (Table 6).

Table2.DFSof77Patients

VariableMedian value (Month)Max value (Month)Min value (Month)PAge (years)0.026 ≤304.7532.010.32 31～6011.3051.180.22 >6010.7649.070.71Gender0.969 Male8.3151.180.35 Female10.3449.070.22Tumor site0.046 Maxilla and skull base7.2147.450.22 Mandible10.7851.180.35Pathological grade0.064 I (well differentiated)30.7547.4514.06 III (poorly differentiated)9.3949.070.22 Unknown19.7751.184.56

(Table 2 continues on next page)

(Continued from previous page)

VariableMedian value (Month)Max value (Month)Min value (Month)PTumor size (cm)0.440 ≤49.3949.070.22 >410.7851.180.34Surgical margin0.024 Negative10.7351.180.35 Positive4.5614.060.22Treatment strategy0.167 Surgery only8.1743.020.22 Surgery+CT11.8347.453.08 Surgery+RT10.9749.070.68 Surgery+CT+RT11.3551.181.12CT given0.088 Yes11.4951.181.12 No8.5749.070.22RT given0.169 Yes11.2051.180.68 No9.3947.450.22Initial death cause0.550 LR9.9244.310.35 MT10.6751.180.22 Total9.9851.180.22-

CT indicates chemotherapy;RT indicates radiotherapy

Table3.OSof77Patients

VariableMedian value (Month)Max value (Month)Min value (Month)PAge (years)0.640 ≤3011.1647.162.61 31～6016.6857.152.25 >6020.7657.514.72Gender0.705 Male13.3457.513.07 Female15.8649.272.25Tumor site0.051 Maxilla and skull base12.4157.512.25 Mandible16.9857.154.35Pathological grade0.060 I (well differentiated)36.2857.5115.04 III (poorly differentiated)13.6949.272.25 Unknown27.7457.159.58Tumor size (cm)0.224 ≤413.6949.272.25 >414.9257.512.61

(Table 3 continues on next page)

(Continued from previous page)

VariableMedian value (Month)Max value (Month)Min value (Month)PSurgical margin0.001 Negative15.2957.512.61 Positive9.7815.042.25Treatment strategy0.034 Surgery only12.3048.052.25 Surgery+CT21.5357.518.43 Surgery+RT15.8649.072.61 Surgery+CT+RT28.7557.467.78CT given0.008 Yes22.3457.517.78 No13.0049.072.25RT given0.145 Yes17.1857.462.61 No13.0057.512.25Initial death cause0.939 LR14.9257.512.61 MT13.0049.272.25 Total14.0657.512.25-

CT indicates chemotherapy;RT indicates radiotherapy

Table4.FrequencyofMTafterOperation

VariableMonths after operation0～1213～2425～3637～4849～60Age (years) ≤301412-- 31～60139211 >6042-21Gender Male135232 Female187311Tumor site Maxilla and skull base21721- Mandible105222Pathological grade I (well differentiated)-1-1- III (poorly differentiated)3010321 Unknown111-1Tumor size (cm) ≤4238311 >484121Surgical margin Negative2611432 Positive51---

(Table 4 continues on next page)

(Continued from previous page)

VariableMonths after operation0～1213～2425～3637～4849～60Treatment strategy Surgery only10321- Surgery+CT23-1- Surgery+CT+RT612-1 Surgery+RT135-11CT given Yes84211 No238221RT given Yes196212 No12622-Total3112432

CT indicates chemotherapy；RT indicates radiotherapy

Table5.FrequencyofLRafterOperation

VariableMonths after operation0～1213～2425～3637～4849～60Age (years) ≤3051--- 31～609321- >604----Gender Male92--- Female9311-Tumor site Maxilla and skull base71--- Mandible11411-Pathological grade I (well differentiated)----- III (poorly differentiated)18511- Unknown-----Tumor size (cm) ≤410311- >482---Surgical margin Negative17511- Positive1----Treatment strategy Surgery only132--- Surgery+CT2----

(Table 5 continues on next page)

(Continued from previous page)

VariableMonths after operation0～1213～2425～3637～4849～60 Surgery+CT+RT1--1- Surgery+RT231--CT given Yes3--1- No1551--RT given Yes3311- No152---Total18511-

CT indicates chemotherapy;RT indicates radiotherapy

Table6.PredilectionSitesofMT

Predilection siteN Lung41 Liver9 Kidney5 Bone2 Brain15

Sometimes malignancies metastasize to multiple sites

Kaplan-Meier (KM) graphs were made for factors such as the following: age, gender, tumor site, tumor size, pathological grade, surgical margin, postoperative radiotherapy, chemotherapy, treatment strategy and initial death cause. Log-rank test was also done to calculate the difference in KM graphs.

According to log-rank results, factors that affected DFS included age (P=0.026, Table 2), tumor site (P=0.046, Table 2) and surgical margins (P=0.024, Table 2). Pathological grade (P=0.064, Table 2) and adjuvant chemotherapy (P=0.088, Table 2) might also affect DFS.

Factors which affected OS included surgical margins (P=0.001, Table 3), treatment strategy (P=0.034, Table 3) and adjuvant chemotherapy (P=0.008, Table 3). Differences among tumor sites (P=0.051, Table 3) and pathological grades (P=0.060, Table 3) were not statistically significant in OS, though thePvalue was close to the standard (P<0.05 was considered statistically significant).

Discussion

Osteosarcoma is an uncommon bone malignancy in the head and neck region, which has a unique biological and clinical behavior compared with osteosarcoma in the extremities[13]. HNOS mainly occurs in the mandibular region, most of them are of the pathological grade III, while MT is the main cause for most cancer specific deaths[3].

According to the previous studies, most of HNOS cases and cancer specific deaths occurred in patients above 30 years old[3-4]. In our study, 41 patients aged between 31 and 60 years old died at the median DFS of 11.30 months which was longer than the other two age groups (P=0.026, Table 2). Although it had been reported that HNOS patients under 30 years old had comparatively low pathological grade and better general outcomes[3-4, 14], the DFS in this group was the shortest (Table 2). Patients above 60 years old had the longest median value of OS (20.76 months), but there was no statistically significant difference among age groups (P=0.640, Table 3). It might be that the disease characteristics vary a lot among different age groups. Once there was MT or LR, young HNOS patients were associated with shorter survival term.

As reported, site and pathological grade might affect the general outcome of HNOS patients. Patients with tumors in the mandibular region and tumors of low pathological grade had better prognosis[3-4,11,14]. Our studies also revealed that these factors affected the prognosis in patients who subsequently died of HNOS. DFS and OS of patients with osteosarcoma of the mandible were 10.78 months (P=0.046, Table 2) and 16.98 months (P=0.051, Table 3) respectively. Patients with tumors of pathological grade I experienced a DFS of 30.75 months (P=0.064, Table 2) and an OS of 36.28 months (P=0.060, Table 3) which were longer than those of others, although the results were not statistically significant. In previous studies, tumor size was reported as another prognostic factor which may affect general outcomes of HNOS patients[4]. In this study, however, there were no statistically significant differences in DFS and OS between patients with large tumors (>4 cm,P=0.440, Table 2) and those with small tumors (≤4 cm,P=0.224, Table 3). Once MT or LR was established, the survival was similar for patients with tumors of different pathological grades.

Surgery is the main treatment approach for HNOS, and the surgical margin status has a strong impact on prognosis. Positive surgical margins is regarded as an important cause of LR or MT[3-4,11-14]. In this study, the median values of DFS (10.73 months,P=0.024, Table 2) and OS (15.29 months,P=0.001, Table 3) of patients with negative surgical margins were almost twice as long as patients with positive surgical margins (Table 2, 3), suggesting that negative surgical margin was beneficial to patients with HNOS.

Previous studies did not elucidate the role that postoperative radiotherapy plays in the treatment of HNOS. Large retrospective studies conducted by the National Cancer Data Base (NCDB) reported that postoperative radiotherapy had not shown any special effect on the prognosis of HNOS[4], while Chen et al. found that postoperative radiotherapy had an effect on the local control of HNOS[3]. Oertel et al. also reported that it was possible to achieve long-term control with adequate radiation doses[15]. In our study, there were no statistically significant differences between patients with postoperative radiotherapy and those without postoperative radiotherapy in DFS (P=0.169, Table 2) or OS (P=0.145, Table 3).

The impact of adjuvant chemotherapy is the most controversial issue in the management of HNOS. Some retrospective studies found an improvement in survival conditions with adjuvant chemotherapy[3-4,14], while a meta-analysis had provided conflicting conclusions for it[16,17]. Our study showed that patients who had received adjuvant chemotherapy experienced a longer DFS (11.49 months,P=0.088, Table 2), and OS (22.34 months,P=0.008, Table 3) which might suggest that adjuvant chemotherapy is needed in the treatment of HNOS patients.

As for the treatment strategy of HNOS, surgery alone is not recommended nowadays. Surgery combined with adjuvant therapies such as adjuvant chemotherapy and postoperative radiotherapy helps patients achieve better outcomes[3-4,14,18]. In our study, treatment options with adjuvant chemotherapy and postoperative radiotherapy also benefited the patients in OS (P=0.034, Table 3).

It is confirmed that MT and LR are two main causes for death due to osteosarcoma, no matter if it occurs in the extremities or in the head and neck region. Rarely in literature the detailed information, such as time and exact site, has been described[3-4,19,20]. Table 4 lists details regarding MT in all patients. Thirty-one cases experienced MT in the first 12 PMs, while the remaining 2 experienced it in the fifth 12 PMs. The lungs seemed to be the most likely predilection site of MT, while the bones outside the head and neck region were the least likely predilection sites of MT. Sometimes malignancies metastasize to multiple sites (Table 6). In table 5, the time to LR was also listed: LR in 18 cases occurred during the first 12 PMs, and in the remaining 5 occurred in the second to fourth 12 PMs. There were no statistically significant differences among patients with different initial causes of death in DFS (P=0.550, Table 2) and OS (P=0.939, Table 3). According to the above-mentioned information, strict and frequent follow-ups should be conducted for at least 48 PMs.

Conclusion

For patients died of HNOS, surgical excision with negative margins is crucial to better outcomes. Patients above 30 years old experienced longer DFS than younger patients. Improved survival had been noted for patients receiving adjuvant chemotherapy as well as for those with tumors with a low pathological grade. Factors like tumor size and postoperative radiotherapy had been identified as negative prognostic factors. Strict and frequent follow-up should be conducted for at least 48 PMs, and life-long follow-up is recommended.