To the best of our knowledge, in this study, we reported the largest number of patients with pure soft tissue recurrence of GCTB in a single center. Meanwhile, we testified predictive value of recurrence types and a modified classification system for this soft tissue recurrence. Although our results showed that types of recurrence may not affect number of soft tissue recurrence times, we still believe that this new classification would be helpful for surgeons to deeply assess and chose a better examination tool to detect these lesions. Our results also revealed the effectiveness of ultrasonography in detection of soft tissue recurrence of GCTB.

An accurate soft tissue recurrence rate of GCTB has not been revealed. GCTB has a high local recurrence rate [8], distant metastasis and malignant transformation could also be observed in patients with GCTB [9]. Previous studies have recognized the phenomenon of soft tissue recurrence after primary operation for GCTB, but their samples were small or mixed with local recurrence in bone to analyze. Thus, an accurate incidence rate and risk factors could not be evaluated exactly. Previously, 17 cases with soft tissue recurrence were reported by Cooper et al. [6] among 1100 patients with GCTB, the rate of soft tissue recurrence was 1.5%, which was similar with our results. A rate of 3.1% (4 patients among 129 GCTB cases) was reported by Park et al., [10] and 2.1% (6 soft tissue recurrent cases in 291 patients with GCTB) by Xu et al. 7] Another 3 cases were described by Ehara et al. [11] and 4 cases by Lee et al. [12] A concern about extending these incidence rate to other centers is that these cases were mainly detected by radiography imaging when ossification or mineralization could be found on X-ray or CT. While some candidates without ossification may have been missed.

Many methods could be used to detect recurrence of soft tissue in GCTB. If recurrent soft tissue mass located at a body region with less muscle or fat tissue, like the forearm, it could be detected by palpation and an ultrasound examination would be enough. Detection of this lesion by plain radiograph was difficult, even in some cases with peripheral mineralization [13]. In this situation, CT or MRI may be a preferred method. Inhomogeneous low to intermediate or high signal intensity could be observed in recurrent soft tissue lesions on different weight or fat-suppressed MRI series [7, 10]. But MRI was less sensible than CT in observing ossification. In our results, 27 of all 57 recurrent lesions were firstly detected by ultrasonography, followed by CT and X-ray. This may be explained by making full use of ultrasound examination during follow-up at our department. The results indicated the effectiveness of ultrasonography in distinguishing soft tissue recurrence of GCTB in clinical practice, when recurrent mass was not obvious in radiograph or other radiological exsminstions were postponed. Further research could focus on the ultrasound imaging characteristics of soft tissue recurrent lesions of GCTB, and compare the sensitivity of ultrasonography with X-ray or CT examination in these patients, so as to provide a simple, fast and economical examination method for clinicians to detect such recurrence as soon as possible.

The original classification for soft tissue recurrent lesions introduced be Xu et al. [7] was proposed by a combination of MRI and plain radiograph. During practical application, we found that in spite of its simplicity and practicability, this classification may not be detailed enough to distinguish some cases with no typical eggshell like peripheral rim or overall central ossification. Therefore, we modified the original classification into 5 types based on features of recurrent soft lesions on CT. In our system, the original type I and II were further divided into more detailed types in order to provide surgeons a better guidance. However, our results showed that neither the original nor our modified classification system could contribute to number of recurrence. On the other hand, size of soft tissue recurrence lesion was not taken into consideration in the original and our modified classification, this may also contribute to the poor predictive effect of these two classification systems. Therefore, further studies are needed to develop a more refined classification system with better practicability and accuracy.

Due to various sizes, locations and extending to other surrounding tissues such as nerves, there was no one operation method could be appropriate for all recurrent patients. Adjuvant techniques, like burring or cementation, combined with intralesional curettage could help to reduce recurrence rate of GCTB in bone [14]. But this may not be suitable for soft tissue recurrence, since some lesions have no eggshell like ossification rims. Marginal excision remains mainstay surgical treatment for soft tissue recurrence. In the study of Xu et al., [7] all six patients with recurrent mass in soft tissue had received a marginal resection and none of them suffered a re-recurrence again. En bloc resection technique was also applied to reduce recurrence of primary GCTB with soft tissue extension. Generally, en bloc resection was deemed as a more effective manner than curettage in reducing recurrence rate. Therefore, if local recurrence was observed in a patient following an en bloc resection, this GCTB could be indicated as an aggressive type and a careful follow-up was recommended [15].

Except surgical treatment, effectiveness of adjuvant treatment with denosumab for soft tissue recurrence of GCTB was unclear. Similar to its bone recurrence counterparts, soft tissue recurrence could also develop to peripheral or intralesional mineralization, and denosumab may improve this procedure. Denosumab therapy for GCTB has been applied for years and its benefits and good therapeutic effect have been reported [16]. However, up to date, few studies have reported outcomes with denosumab application in soft tissue recurrence of GCTB. Akaike K et al. [13] and Suzuki T et al. [17] respectively reported one patient with a satisfactory outcome using denosumab. Among our 31 cases, 7 of them had a history of denosumab application, (Figs. 6, 7, 8) but the dosage and time of denosumab usage were not consistent, which made us have not conducted an effectiveness assessment of denosumab for soft tissue recurrence treatment. Among the 7 patients who received at least one dosage of denosumab, one of them had a denosumab treatment for two months before primary surgery and had two times of recurrence. Denosumab was applied in 3 patients after detection of the first recurrence, and then a reoperation was performed. One of these 3 patients underwent two times of re-recurrence, the other 2 patients had no a twice soft tissue recurrence. When a twice re-recurrence was detected, 3 other patients received denosumab treatment, one of them was still under chemotherapy for pre-existing lung metastasis, the other two had no recurrence at final follow-up. This result showed the various therapeutic effectiveness for soft tissue recurrence of GCTB in different patients. This may also be an indicator for future researches to investigate whether soft tissue recurrent patients would be benefit from adjuvant treatment with denosumab.

Preoperative images of another 22 years old female patient with GCTB located in distal radius. a–d Osteolytic destruction in right distal radius were revealed by X-ray (a, b), CT (c) and MRI (d). e–g Ossification could be observed on X-ray radiography (e, f) and CT (g) after application of denosumab for 2 months

Although denosumab and widely resection had been applied, soft tissue recurrence still occurred after primary surgery. a Specimen of GCTB in distal radius with wide resection. b Post-operative plain radiography showing a wrist fusion with autogenous iliac bone graft and internal fixation were performed. c Hypoechoic mass could be detected around the original operation site by ultrasound. d Recurrent lesion could be obviously observed on CT. e: Specimens of excised soft tissue recurrence lesions

Re-recurrence of soft tissue occurred in the 22 years old female patient with GCTB located in distal radius. a and b A hypoechoic mass was detected 5 months after excision of the first recurrence. c CT examination before operation for the twice recurrence of soft tissue. d The resected soft tissue lesion

Location of GCTB may influence its recurrence. GCTB at distal radius showed a higher recurrence rate [1, 18]. However, whether the same is true for soft tissue recurrence remains unclear. In our cases, primary GCTBs of 8 patients with soft tissue recurrences located in radius, which accounted the most among all 31 patients (25.81%). But owing to the small number of patients in our study, we did not conduct a comparation between different locations. In another study about risk factors for local recurrence after surgical treatment of GCTB, Li D et al. [19] revealed that surgical methods, but not tumor location, pathologic fracture, age that younger than 30 years, gender and the Campanacci grade, would impact recurrence. However, we also failed to analyze primary surgical methods in our study for that some patients had received their primary surgeries in other institutions.

Campanacci grade may predict recurrence of GCTB [20]. However, our results have not validated it as a risk factor for soft tissue recurrence, and this was consistent with results of a previous multi-center cooperative research including 110 recurrent patients of GCTB [21]. The reasons may lie in that soft tissue recurrence is mainly caused by implantation during operation, while the Campanacci grade is a classification system used to reflect aggressiveness of GCTB. Thus, the Campanacci grade is more suitable for assessing recurrence in bone which may be mainly caused by residual components after non-suitable curettage or pathological fracture, rather than soft tissue recurrence which may be caused by tumor implantation during primary surgery. But the exact mechanism of soft tissue recurrence is not yet clear. Interestingly, a pathology report of one patient in our study indicated tumor embolus in a peripheral blood vessel surrounding the primary GCTB. This may indicate another underlying mechanism for soft tissue recurrence or metastasis of GCTB. However, this hypothesis needs more pathological evidence to validate.

Few researches have focused on outcomes after reoperation for recurrent GCTB, especially in soft tissue recurrence [16, 22]. Repeated recurrence of GCTB has an association with malignant transformation, and would lead to a higher incidence of metastasis to lung [21]. Early local recurrence was considered as a risk factor causing repeat recurrence in bone [21]. A median time of 13.5 months of local recurrence after GCTB resection to malignant transformation was reported by Tsukamoto, S et al. [15]. Our results showed a rate of 61.29% of patients underwent soft tissue recurrence more than once. Interval period of the first, twice and third time of recurrence gradually shortened as presented in the results section. MSTS score of these 7 patients with 3 times recurrence was slightly lower compared with the overall scoring. Among them, lung nodules were observed in 4 patients. One patient with twice recurrence died after lung metastasis. One patient with once recurrence suffered lower limb amputation at hip joint. All these indicated that soft tissue recurrence of GCTB should be given enough attention to.

This study contained some obvious disadvantages. First of all, impact of primary surgical manners of GCTB on soft tissue recurrence were not evaluated, since 15 patients received their first operations with various methods in other institutions. Experience of surgeons may not be comparable. However, no relationship between primary surgery manner and recurrence of GCTB has been reported before [21]. On the other hand, complication incidence was not compared between patients with different number of recurrences. Owing to the small number of samples and various of originate GCTB locations, we believe that it would be suitable to calculate and compare complications along with soft tissue recurrence with a larger number of patients from multi centers. Last but not the least, signal intensity in MRI was not extensively studied in this research, because we classified soft tissue recurrence mainly based on CT images. A further study concerning this aspect could be performed to reveal association between MRI presentation and prognosis of GCTB with soft tissue recurrence.