Introduction

Giant cell tumor of bone (GCTB) is a rare primary bone tumor that typically occurs in the meta-epiphyseal region of long bones, predominantly around the knee joint (Campanacci et al. 1987; Freyschmidt et al. 1998). Usually of benign character, it arises after skeletal maturity with a peak incidence in the third and fourth decade of life and a slight female predilection (Balke et al. 2008; Campanacci 1990; Carrasco and Murray 1989; Freyschmidt et al. 1998; Larsson et al. 1975; Salzer-Kuntschik 1998; Schwartz 1998).

The biological behavior of GCTB varies from indolent and static tumors to locally aggressive lesions with extensive bony destruction, penetration/destruction of the cortex and extensive soft-tissue expansion (Enneking 1986). In standard X-rays GCTB present as lucent lesions without matrix calcifications eccentrically located within the meta-epiphyseal region of the bone (Freyschmidt et al. 1998) usually stage II or III according to the three stage classification system of Campanacci (Campanacci 1990; Campanacci et al. 1987) and Enneking (Enneking 1983).

After intralesional procedures the local recurrence rate varies from 10 to 40% (Blackley et al. 1999; Campanacci et al. 1987; Goldenberg et al. 1970; Lausten et al. 1996; Malek et al. 2006). The treatment of choice is intralesional curettage, additional burring with a high-speed air drill and bone cement (polymethyl methacrylate—PMMA) packing; if applicable a chemical adjunct such as hydrogen peroxide (H2O2) or phenol can be used. With this combination the recurrence rate can be reduced by the factor 28 compared to curettage without adjunct (Balke et al. 2008). Local recurrence is accompanied by an increased risk of so called “benign” pulmonary metastases (Balke et al. 2008; Bertoni et al. 1985, 1988; Campanacci et al. 1987; Cheng and Johnston 1997; Dominkus et al. 2006; Goldenberg et al. 1970; Kay et al. 1994; Maloney et al. 1989; McDonald et al. 1986; Osaka et al. 1997; Rock et al. 1984; Sanjay and Kadhi 1998; Siebenrock et al. 1998; Tubbs et al. 1992). As recently published the recurrence rate varies by anatomical localization with the highest recurrence rate in the distal radius, followed by distal tibia and proximal femur (Balke et al. 2008).

Although the recurrence rate of GCTB is relatively high exact data on how to treat the recurrent cases is lacking. The possible surgical procedures range from repeated intralesional curettage to wide resection. This paper focuses on the treatment of 67 patients with recurrent GCTB treated from 1980 to 2007 at the authors institution and gives advices about the sufficiency of the above mentioned treatment options.

Patients and methods

This is a level II retrospective comparative study based on revision of surgical protocols and patient records. The last follow up was done via personal contact or phone call.

Two hundred and fourteen patients with histologically certified GCTB were treated at the authors department from 1980 to 2007. Sixty-seven patients with at least one local recurrence have been included in this study. Twenty-five of these patients received their first surgical treatment at our hospital, 42 were referred from other hospitals due to local recurrence. Thirty-two patients were male, 35 were female. The mean age at first diagnosis was 29.6 (15.5–63.1) years. The mean follow-up was 77.3 (13.2–267.2) months. The mean time until first local recurrence was 22.0 (1.6–172.2) months, the mean amount of recurrences per patient was 1.4 (1.0–4.0). Forty-six patients showed one local recurrence, 14 patients had two, 4 patients had three and 2 patients had four recurrences. In one patient the exact number of recurrences is not documented. The mean follow up after the most recent operation was 45.3 (1.4–208.6) months.

The vast majority (50.8%) of recurrent GCTB was localized around the knee joint (20 × distal femur, 11 × proximal tibia, 2 × proximal fibula, 1 × patella), followed by distal radius (n = 8). The other 25 tumors were distributed as indicated in Fig. 1.

Fig. 1
figure 1

Distribution of local recurrences according to anatomical site. The black numbers indicate the numbers of the recurrent cases (n = 67), the gray numbers the overall appearance at this localization of our whole patient collective (n = 214)

Full size image

At first diagnosis nine patients presented with a stage II, 33 with a stage III, and none with a stage I lesion according to Campanacci and Enneking (Campanacci et al. 1987; Enneking 1983). In 25 patients the radiological findings were not available. Thirty-three patients presented with a soft-tissue component at first diagnosis and 20 without. In 14 patients information on soft-tissue extension was lacking.

The recurrences were detected by standard X-rays (Fig. 2a) and magnetic resonance imaging (MRI) (Fig. 3) and confirmed by fresh frozen sections during surgery.

Fig. 2
figure 2

Recurrent giant cell tumor of bone treated with a tumorprosthesis of the distal femur. Plain radiograph showing local recurrence after cementation and plate fixation (a) treated by wide resection and implantation of a Mutars® tumorprosthesis of the right distal femur (b) due to the massive bony destruction and soft tissue extension

Full size image
Fig. 3
figure 3

Magnetic resonance imaging of recurrent GCTB with massive soft-tissue component Sagittal T2 (a) and transversal fat-saturated T1 (b) sequences of a recurrent GCTB after cementation and plate fixation with massive bony destruction and soft tissue extension. For X-ray see Fig. 2a

Full size image

The most common primary treatment had been intralesional curettage without any adjunct (n = 30), followed by curettage and PMMA packing (n = 18), and curettage plus PMMA plus burring (n = 13). Three patients had been treated with intralesional tumor resection, two with curettage plus burring without PMMA (Table 1).

Table 1 Re-recurrence rate according to surgical procedure
Full size table

The most common surgical procedure for the recurrence was curettage plus PMMA plus burring (n = 23), followed by curettage plus PMMA without burring (n = 14), wide resection (n = 11), curettage without adjunct (n = 9) and intralesional resection (n = 8). One patient received curettage plus burring without PMMA. In one patient the surgical procedure of the local recurrence was not documented. If cementation had been performed in the primary treatment the cement was totally removed before further curettage.

The re-recurrence rate was evaluated with regard to the recurrence operation, the anatomical localization and the outcome at final follow-up.

All statistical analysis was performed using chi-square tests for the likelihood ratio and ANOVA for determining the statistical significance (significance level P < 0.05, 95% confidence interval).

Results

Forty-five and a half percentage (30 out of 66 patients) of local recurrences occurred within the first year, 69.7% (46 out of 66 patients) within the first 2 years. From the third to fifth year, 27.3% (18 out of 66 patients) developed local recurrence and two patients developed local recurrence after more than 10 years. Ninety-seven percentage (64 out of 66 patients) of recurrences occurred within the first 5 years. In one patient the exact date of the recurrence was not documented.

If after intralesional procedures (curettage or intralesional resection) no adjunct had been used, the re-recurrence rate was 58.8% (10 out of 17 patients) and decreased to 26.3% (10 out of 38 patients) if at least one adjunct (PMMA, burring or both) was used (Table 1). The likelihood of re-recurrence was reduced by the factor 5.257 which was statistically significant (P = 0.020) (Table 2). The use of PMMA alone reduces the re-recurrence rate to 35.7% (5 out of 14 patients) by the factor 1.695 but without reaching statistical significance (P = 0.213). The best results of the intralesional procedures concerning re-recurrence were achieved by a combination of all adjuncts (PMMA + burring) with a reduction to 21.7% (5 out of 23 patients) (Table 1). Compared to intralesional procedures without adjuncts the likelihood of re-recurrence was reduced by the factor 5.508 which was clearly significant (P = 0.016) (Table 2). The groups with hydrogen peroxide or burring alone were too small so that statistical analysis was not possible (Table 1). In case of wide resection no re-recurrence occurred.

Table 2 Statistical analysis of the different treatment regimen of GCTB
Full size table

The re-recurrence rate according to the localization of the tumor (Fig. 1) was highest in the distal radius (8 out of 9 patients), followed by distal tibia (5 out of 11 patients), proximal femur (5 out of 14 patients), distal femur (20 out of 58 patients), and proximal tibia (11 out of 42 patients).

In GCTB of the distal radius major surgical procedures commonly became necessary. One resulted in allograft-reconstruction with an external orthosis, one in transposition of the ulna to the radius, one in tumor prosthesis, three in arthrodeses. Only two could be treated by repeated curettage, PMMA packing and subchondral bone graft, one with an additional intralesional resection of a soft tissue recurrence (Table 3). The worst outcome was in recurrent GCTB of the spine; two out of three patients suffered from incomplete paraplegia and one died due to excessive bleeding (Table 3).

Table 3 Overview of the whole patient collective
Full size table

Twenty out of 67 patients (29.9%) suffered from a second recurrence. Three were treated with curettage and cementing (no re-recurrence), three with curettage plus burring and cementing (two re-recurrences), eight with intralesional resection (three re-recurrences) and five with wide resection (no re-recurrence). In one case the treatment was not documented. Out of the six patients (9.0%) who suffered from a third local recurrence two were treated with curettage plus burring and cementing (no re-recurrence) and two with wide resection (no re-recurrence). Two patients (3.0%) experienced a fourth local recurrence, one after intralesional resection, in the second patient the surgical procedure was not documented. The first was finally cured by wide resection.

Seven patients (10.5%) developed pulmonary metastases. Five of them associated with the first local recurrence, one after the second and one after the third recurrence.

At last follow up after a mean of 77.3 (13.2–267.2) months, 60 patients were free of disease and five were alive with disease and stable pulmonary metastases. Despite high dose chemotherapy one male died due to intestinal perforation after developing multiple metastases (lung, soft tissue of upper arm, chest, tongue, brain and small intestine) after third recurrence. One female died in 1982 due to uncontrollable bleeding after resection of second recurrence of a GCTB of the thoracic spine.

Fourteen patients (20.9%) received an endoprosthesis after a mean of 71.1 (13.6–176.5) months (Fig. 2b), 12 due to tumor recurrence, 2 due to secondary arthritis after intralesional procedures, 6 and 7 years after the first surgery (Table 3). Five of these tumors were localized in the distal femur; four were reconstructed with a distal femur prosthesis after wide resection of the first local recurrence, one received a prosthesis due to secondary arthritis after curettage, burring and cementing (Table 3). Three of the tumors affected the proximal femur and were treated by a tumorprosthesis; once after wide (after first recurrence) and twice after intralesional resection (one after the first, one after the second recurrence). Two patients received a prosthesis of the proximal tibia, one after wide resection of the first recurrence, one due to secondary arthritis after curettage, cementing and subchondral bone grafting. The other tumors finally being treated by a tumorprosthesis were localized in the distal radius (wide resection of first recurrence), distal and proximal humerus (both after wide resection of second recurrence) and patella with extension into the joint after an insufficient intralesional procedure (Table 3).

Discussion

Our data shows that almost 70% of local recurrences of GCTB occur within the first 2 years after treatment which is in accordance to the literature (Malek et al. 2006; Turcotte et al. 2002; Vult von Steyern et al. 2006). Therefore we recommend short term controls within this period, e.g. every 3 months. If the patient presents with newly occurred pain or swelling or the standard X-rays show any suspicious findings an MRI should be performed.

A previously published study by the “Scandinavian sarcoma group” with a total of 19 recurrent GCTB reports that recurrent tumors of the long bones can be successfully treated by further intralesional curettage and cementing (Vult von Steyern et al. 2006). An older study by O’Donnel et al. (O’Donnell et al. 1994) reported the treatment of 15 recurrent cases. Five were treated by further curettage and cementing and ten by resection. Because in both studies only a few patients with recurrent GCTB were included statistical analysis was not possible.

Due to the high patient count this is the first study that precisely analyzes the different treatment modalities and the statistical significances. Intralesional procedures without any adjuncts such as burring or PMMA result in an unacceptably high re-recurrence rate of about 58.8%. The additional use of at least one adjunct reduces this rate to 26.3%. The additional use of PMMA as single adjunct reduces the likelihood of re-recurrence by the factor 1.695–35.7% compared to intralesional procedures without adjuncts but did not reach statistical significance (P = 0.213). As recently proven for primary treatment (Balke et al. 2008) the best results concerning re-recurrence can be achieved by a combination of all adjuncts (PMMA + burring) with a recurrence rate of 11.9%. We proof here that even the recurrent cases further curettage, burring of the cavity with a high-speed air drill and cementing significantly reduces the likelihood of re-recurrence to 21.7% by the factor 5.508 (P = 0.016). This rate is an acceptable control rate for a recurrent, usually benign lesion. Unfortunately the groups treated with hydrogen peroxide were too small, so that significance could not be statistically analyzed.

In case of recurrence after cementing we generally removed the cement completely before further curettage. The removal of the cement is sometimes difficult and may cause further destruction of the bone and especially the cartilage. In early recurrences within the first 2 years the tumor often recurred in multiple localizations surrounding the cement and not just at the sites visible on the MRI. Thus we recommend this procedure although it might seem like an over treatment and evidence data is lacking.

Although in case of wide resection no re-recurrence occurred we do not recommend this treatment in general. As long as adjuncts are used the re-recurrences are in most cases sufficiently controllable with further intralesional curettage (Wang et al. 2005). In contrast to wide resection followed by arthrodesis or tumor prosthesis, curettage and cementing mostly result in almost normal function (Oda et al. 1998; Turcotte 2006; Vult von Steyern et al. 2006). With intralesional procedures the recurrence rate in the distal radius was very high in our patient collective which is in contrast to previous reports (Ozalp et al. 2006). It may be explained by insufficient primary treatment. Despite the aggressiveness of the tumors (all grade III, six with a soft tissue component) the initial procedure was intralesional curettage in all cases, bone cement packing was performed in five cases only. There is still no general recommendation for the treatment of GCTB at this special anatomical site (Bianchi et al. 2005; Cheng et al. 2001; Intuwongse 1998; Leung and Chan 1986; Sheth et al. 1995). Due to the high complication rate in recurrent giant cell tumors of the spine, a marginal or even wide primary surgery according to Enneking (Enneking et al. 1980) may be justified for this localization (Chen et al. 2004; Gille et al. 2005; Gruber et al. 1999; Ozaki et al. 2002). Some authors recommend spondylectomy as an aggressive operative treatment to prevent local recurrences (Kawahara et al. 1998; Samartzis et al. 2008) especially because of the difficulty to use conventional treatment options with curettage and cementing in these localizations (Matsumoto et al. 2007). According to Fiedler (Fidler 2001) the en bloc approach is the safest technique. Nowadays we perform preoperative embolisation of giant cell tumors involving the spine and sacrum with the aim to decrease the rate of sometimes excessive bleeding and reduce complications.

There are no reliable numbers in the literature about the use of endoprostheses in recurrent GCTB. In our patient collective 14 out of 67 patients (20.9%) finally received an endoprosthesis which in our opinion is a relatively high rate for a benign skeletal lesion. Only two patients received the prosthesis due to secondary arthritis after 6 and 7 years. In both cases the tumor had been located around the knee and was formerly treated by curettage and cementing (distal femur) or curettage and bone grafting (proximal tibia). Giant cell tumors often reach very close to the articulating cartilage. Thus especially in the weight-bearing lower extremity secondary arthritis is often unpreventable. It is still not clear if the performance of a subchondral bone graft in combination with cementing is able to reduce cartilage destruction (Frassica et al. 1990; Suzuki et al. 2007; Szalay et al. 2006). As recently published there is no increased local recurrence rate caused by a subchondral bone graft (Balke et al. 2008). The other 12 patients received the tumorprosthesis due to the extensive local recurrence and were predominantly located in the lower extremities (four distal femur, three proximal femur, one proximal tibia, one patella due to massive invasion of the joint). Three were cited in the upper extremity (distal radius, proximal and distal humerus).

In contrast to the “Scandinavian sarcoma group” study (Vult von Steyern et al. 2006) in which at final follow-up all patients were free from disease and showed a good function, our study proofs that recurrent GCTB can become a severe disease. The rate of pulmonary metastases in GCTB is generally low with approximately 2–3% (Balke et al. 2008; Bertoni et al. 1985, 1988; Cheng and Johnston 1997; Goldenberg et al. 1970; Kay et al. 1994; Masui et al. 1998; Osaka et al. 1997; Rock et al. 1984; Sanjay and Kadhi 1998; Siebenrock et al. 1998; Tubbs et al. 1992), but the literature indicates that local recurrence is accompanied by an increased risk of pulmonary metastases (Bertoni et al. 1985; Cheng and Johnston 1997; Kay et al. 1994; Lausten et al. 1996; Masui et al. 1998; Rock et al. 1984; Sanjay and Kadhi 1998; Tubbs et al. 1992). In fact in our study 10% of recurrent GCTB developed possibly life-threatening metastases. Therefore every patient with a local recurrence of a GCTB should be examined with a CT-scan of the chest.

Conclusion

Recurrent GCTB can be treated by further curettage with an acceptable re-recurrence rate as long as burring and cementing is used. Rate of the recurrent cases who finally end up in an implantation of an endoprosthesis is 20.9%, which is a remarkable rate for a benign skeletal lesion. In contrast to non-recurrent GCTB the rate of pulmonary metastases in the recurrent cases is high (10%). Thus recurrent GCTB has to be considered a severe disease. Although the outcome in recurrent GCTB of the long bones is generally good, this does not apply to giant cell tumors of the spine.