Introduction

Giant cell tumor of the bone (GCTB) is a benign but locally aggressive tumor [1–5]. Giant cell tumor of the spine (GCTS) only accounts for 3–6 % of all GCTB [6]. Surgical resection is the treatment of choice for GCTS, but the postoperative recurrence rate is as high as 30–50 % when it arises in the spine [7]. Severe functional sequelae have restricted its application in spine procedures for fear that adjacent critical structures may be functionally compromised [2, 8]. Worse, some patients of primary or recurrent GCTS may present with a tumor not suitable for surgical procedure at all.

Obstacles in GCTS surgery have brought about novel ideas of non-surgical procedures (selective arterial embolization, radiotherapy, bisphosphonate therapy), of which therapeutic radiotherapy proves one classic method of conservative management, yet reports on the results of therapeutic RT varied from failure to success [1] and there has been no published systemic review as yet.

Therefore, in this study, we did a systemic review to comprehensively access the value of therapeutic RT, hopefully to shed new light on therapeutic modalities of GCTS patients not suitable for surgical resection.

Methods and materials

Inclusion and exclusion criteria were defined before literature search. The inclusion criteria were as follows: (1) pathologically confirmed GCTS; (2) GCTS not suitable for surgical resection; (3) GCTS patients that received therapeutic RT; (4) the prognosis and follow-up time were decisively reported. The exclusion criteria were as follows: (1) GCTS patients that received surgical resection followed by adjuvant RT; (2) non-specific follow-up time; (3) vague results of RT. The database of Medline and EmBase were the source of our search. The search terms were “radiotherapy”, “radiation” and “giant cell tumor”.

Information of the included patients has been discreetly recorded, including basic information (age, gender, publication time), status prior to RT (tumor site, recurrence or primary tumor, previous treatment, neurological status), dosage, detailed prognosis, and follow-up time. The prognosis indicators were cautiously defined as follows. The response to RT was defined as lack of disease progression after RT. Local failure (LF) is defined as evidence of tumor progression at final follow-up; local control (LC) is defined as confirmed lack of disease progression on the date of observation irrespective of distant metastasis; the LC rate is detailed as 1 and 2 years and overall LC rate; follow-up time is defined as the time interval between the onset of RT to the time of LF or loss to follow-up. Two investigators independently searched and extracted the data. Disagreements were resolved by discussion. We have contacted the authors for details of any ambiguous information.

For statistical analysis, quantitative data are described by the mean, range, median and confidence interval; qualitative data are described as counts and percentages. The univariate analyses and Chi-square test of various patient factors were performed to identify possible variables that could predict RT outcomes. The LC and overall survival (OS) rate was estimated by the Kaplan–Meier method, and differences were analyzed by log-rank test. Factors with p values 0.05 were considered statistically significant. All statistical computation has been performed using SPSS statistics version 20.

Results

A flow chart describing the procedure of study selection is shown in Fig. 1. The search yielded 66 articles of retrospective case series. Eventually 42 patients from 13 articles were included in our systemic review: 4 articles describe 1 patient; 4 articles describe 2 patients; 5 articles describe 3, 4, 5, 6, 12 patients, respectively. The year of publication ranges from 1980 to 2012. Of the 42 patients, 18 were males and 22 were females, with a mean age of 34.1 ± 14.8 years (median 31.5; range 13–66). All patients received RT from megavoltage machines. Patient data were shown in Table 1.

Fig. 1
figure 1

Selection strategy diagram

Full size image
Table 1 Data on the patients included in this study
Full size table

As for the site of tumor, cervical tumors comprise 9 cases, thoracic tumors 9, lumbar tumors 5 and sacral tumors 19. Therefore, GCTS that occurs in the mobile spine comprise 54.8 % of all cases and sacral tumors 45.2 %. Notably, there are 4 patients that suffered from multiple vertebral tumors, all of which occurred at the mobile spine (Table 2).

Table 2 Relationship between dosage and local control rate of GCT
Full size table

Prognosis of therapeutic radiotherapy

The mean follow-up period was 82.2 months (median 38.0; range 1–348). Of the 42 cases, 13 cases (31 %) reported follow-up of more than 10 years (120 months). All patients had response to therapeutic RT. Nine patients developed LF. The mean progression-free duration for LF group was 11.3 months (median 6.0; range 1–30). Of the LF patients, six patients had LF in less than 1 year (6/9, 67 %) and 8 patients in less than 2 years (8/9, 89 %). Four LF cases occur in the sacrum, and 5 in the mobile spine. Four patients developed distant metastasis, one died without LF of the original tumor [16]. Four patients had post-RT neurological complications. The results suggested a response rate of 100 % (42/42), 1-year LC rate of 85.7 % (36/42), 2-year LC of 81.0 % (34/42), overall LC rate of 79 % (33/42). Distant-metastasis-free-survival rate is 90.5 % (38/42). The overall survival rate was 97.6 % (41/42). The Kaplan–Meier graph demonstrating the overall LC and OS has been exhibited as Figs. 2 and 3.

Fig. 2
figure 2

Kaplan–Meier function curve of local control

Full size image
Fig. 3
figure 3

Kaplan–Meier function curve of overall survival

Full size image

Relationship between recurrence status and local control

We also did a correlation between recurrence status of GCTS patients and their overall LC. 36 of the GCTS patients had primary or initially recurrent GCTS, 7 of whom had LF (19.4 %); 6 of the GCTS patients had repeated recurrence, and 3 of them had LF (50 %). Chi-square test showed the overall LC rate of the repeatedly recurrent GCTS patient was significantly lower than that of initially plus primary GCTS patient (p = 0.028).

Relationship between dosage and local control

In terms of dosage, a variety of doses have been applied for therapeutic RT, with the mean dose of 45 Gy (median 45, range 21–65). The most prevalent dose was 45 Gy (13 cases), followed by 40 Gy (8 cases), and 50 Gy (6 cases). The relationship between these dosage groups and the corresponding overall LC rate differences has been exhibited below.

Discussion

Giant cell tumor of bone is one of the most common primary bone tumors with potentially aggressive biological behavior. GCTS only accounts for 2–8 % of GCTB [22]. Although surgery, especially widely en bloc resection in recent years, remains the treatment of choice for GCTS, functional deficits usually ensue, which has significantly restricted its use [18, 23]. Marginal resection, followed by reconstruction with bone graft or cement, proved to result in high rates of recurrence [24, 25]. More intractable are conditions where tumors are not readily operable. For GCTS not suitable for surgical resection, therapeutic radiotherapy (RT) stands out as the major conservative management [15, 18]. RT was formerly applied as a postoperative adjuvant treatment modality [23, 26]. Recently, GCTB has been evidenced to be highly radiosensitive to either therapeutic or adjuvant RT. Reports have shown that radiotherapy is effective in achieving LC with minimal long-term side effects [16], ranging from 69 to 90 % according to both earlier and recent studies [27, 28]. An abundance of case series have been published concerned with the value of therapeutic RT, yet little has been done on the comprehensive study of therapeutic RT, and thus we did the systemic review to assess the prognosis and especially the LC of therapeutic RT.

Our study yielded perfect response rate (100 %). This shows therapeutic RT, even if surgery proves not feasible, has the capacity for instant improvement of clinical symptoms of GCTS patients. Our study yielded a relatively high 1-year LC rate of 85.4 %, 2-year LC rate of 80.2 %, and overall LC rate of 79 %. Indeed, local control is especially important in vertebral tumors, since the possibility of salvage therapy after local recurrence is remote [29]. These results, including response and LC rates, coherently demonstrate that therapeutic RT is an effective modality for GCTS patients not suitable for surgical resection, which successfully serves as a last resort in the line of conservative management.

The high local control and OS rate can be partially explained by the fact that every patient in our series received RT from megavoltage machines. In recent studies, the rate of disease progression after treatment with megavoltage radiation ranged from 7 to 30 %, a significant reduction compared to that of orthovoltage era before the 1950s, when LF was reported to be as high as 63 % [10, 12, 26, 30–32].

Also, megavoltage proves to be safe. Compared to orthovoltage, it enriches the ray in the tumor location such that the total applied dosage becomes smaller. Arnab et al. [16] did a pooled study of 37 GCTB patients that received RT from megavoltage machines followed up for more than 9 years, and discovered one patient who had post-RT sarcomatous transformation, yet no malignant transformation found on the spine. In contrast, data on treatment with orthovoltage radiation alone have indicated rates of malignant transformation of as high as 33.3 % [26]. In our study, no patient was found to have malignant transformation during follow-up. These findings suggest that improvement of RT techniques significantly boost the safety of therapeutic RT.

Previous studies reported that repeated recurrence was a negative prognostic factor for RT. Jimmy et al. [18] demonstrated 6 local failures occurring in the group of the 12 repeatedly recurrent patients, resulting in a trend toward a lower LC probability for those patients treated for recurrent disease. Turcotte and colleagues [33] also described referral for recurrent tumor as a significant risk factor for increased local recurrence. In this study, the overall LC rate of the repeatedly recurrent GCTS patient was significantly lower than that of initially recurrent and primary GCTS patient (p = 0.028). These results supported that repeatedly recurrent patient does have worse prognosis.

From our result, 67 and 89 % of the LF group had LF within 1 and 2 years, respectively, and 11 % of the LF group had LF after 2 years. It demonstrated that GCTS patients were prone to have long-term LC provided no evidence of LF was found in the initial 2 years after therapeutic RT. The fact suggests the trend that, with time gone by, the LF rate of GCTS would become increasingly low.

According to our study, there is no significant relationship between dosage and overall LC. This result is in coherence with many previous studies that failed to find clear-cut relationship between LC and dosage [18, 29, 34–36]. However, therapeutic RT for GCTS can still pose a threat of malignancy risk. Concerns have been raised on parallel increase in malignancy occurrence with escalation of doses, and some suggested that dosage be less than 45 Gy [34]. Thereby a range of 35–45 Gy has been recommended for current management of GCTS patients [29]. However, it can be concluded from the study that no relationship is found between dosage and malignant transformation, and further study is needed to ascertain their relationship. Four patients had neurological complications, reporting pain, residual spasticity and parasympathetic dysfunction. This suggests that precaution is necessary in the RT procedure to avoid neurological damage.

We observed comparatively short LC duration for those with LF, with mean duration of 11.3 months (median 6.0; range 1–30). One of the patients of LF (case #1), who exhibited signs of LF 2.5 years after RT, and 4 years following RT the patient showed signs of tumor involution without any additional therapy. Some investigations discovered tumor expansion of the initial period (lasting 2–6 months) after RT, followed by a period of regeneration and repair [10, 37, 38]. According to our definition, the case was regarded as LF. Considering the short LC duration of the LF group, and especially the disease course of case #1, it can be inferred that some of the LF patients of our study may have opportunity for disease control if the follow-up time were extended enough.

Four patients were found to have lung metastasis in our series. Historically 3–6 % of pulmonary metastasis of GCT was found [16, 23, 36], a rate that is comparable to case series of surgery alone (5 %) for treatment of GCT [12, 39], suggesting that lung metastasis may be an integral part of GCT pathogenesis. There are four patients (9.5 %) developing lung metastasis during the length of follow-up time, and all patients with lung metastasis were from sacrum. It could be that in our series, patients with unresectable GCT presented at the late stage of the disease, and thus lung metastasis could be more likely to occur. Also, it has been reported that lung metastasis from GCTS was 13.7 % [40], a rate higher than that of GCT of the extremities, suggesting that lung metastasis may be more likely to occur for GCTS patients. Nevertheless, lung metastasis from GCTS after RT may require further investigation.

Apart from RT, there have been various other forms of treatment for inoperable GCTS. Bisphosphonates have been used as sole treatment for GCTS and yield good results [41]. For sacral GCT, serial arterial embolization has achieved success in some case series [42, 43]. Denosumab, a fully human monoclonal antibody to receptor activator of nuclear factor kappa B ligand (RANKL) inhibitor, proved to be highly effective in treatment of unresectable GCT [44], and a recent clinical trial has demonstrated that denosumab would reduce the need for morbid surgery, with safety profiles within normal limits [45]. Further studies are needed to compare the therapeutic effects of the various modalities for treatment of unresectable GCTS.

Limitation

Although this is a novel and large systemic review of GCTS treated by therapeutic RT, it has some limitations. First, the sample size of the study is small, which may decrease the power of the statistics. Second, since the follow-up time of some patients were short, 1-year and 2-year LC rate were calculated in compensation to achieve higher statistical power.

Conclusion

Therapeutic RT could provide a satisfactory prognosis for GCTS patients in the study, and can be an alternative treatment modality for GCTS patients not suitable for surgery. This study demonstrated that repeated recurrence was a negative factor for GCTS prognosis in this treatment modality. Further well-designed randomized control trials are needed to determine the value of therapeutic RT in unresectable GCTS.