Abstract
Some authors attribute recurrences of giant cell tumours to biological factors which are only expressed in some tumours. Grover et al. (1998) suggested that the risk for recurrence is associated with the down-regulation of the nm23-H1 gene. We reviewed the charts of the 154 patients operated on for giant cell tumours of the tendon sheath and selected a group of patients with recurrence (ten cases) and a group of patients who did not have a recurrence after a minimum follow-up of 3 years (13 cases). Immunohistochemical detection of nm23-H1 was performed blindly of the clinical outcome on the paraffin-embedded specimens of these patients and no correlation was found between nm23-H1 expression and the risk for recurrence.
INTRODUCTION
Is has long been thought that recurrence after surgery for giant cell tumour of the tendon sheath (GCTTS) is due to inadequate excision. Recently, some publications have suggested that recurrences may be attributed to intrinsic biological factors which are expressed only in some tumours (Al-Qattan, 2001; Grover et al., 1998; Loréa et al., 2001), and the risk for recurrence seems to be associated with the multinodularity of the tumour and its ability to invade bone or joints. Grover et al. (1998) have shown that the probability of recurrence is associated with the down-regulation of the nm23-H1 gene which codes for a cytoplasmic protein involved in the control of normal development and cellular differentiation. Others have since proposed the use of this test as a predictor of recurrence (Al-Qattan, 2001; Loréa et al., 2001). The aim of this study was to evaluate the efficiency of the nm23-H1 detection in predicting recurrences in a comparative retrospective study.
METHODS
Patient selection
We identified the 154 patients from our computerized database who had undergone surgery for GCTTS of the fingers between 1987 and 1999 at our institution. These charts were reviewed in order to select a group of patients who had a recurrence (Group I) and a group who did not have a recurrence after a minimum follow-up of 3 years (Group II). All the patients were operated on by five different senior surgeons with expertise in hand surgery.
Group I included 10 patients (two men and eight women) with a mean age of 42 (range, 16–71) years at the time of surgery. The thumb was involved in three cases, the index finger in three, the middle finger in two, and the ring and little fingers in one each. The mean number of recurrences was 1.6 (range, 1–3). Recurrences occurred at a mean of 11 (range, 2–23) months after surgery.
Group II included 13 patients (five men and eight women) with a mean age of 46 (range, 27–51) years at the time of surgery. The thumb was involved in two cases, the index finger in three, the middle finger in three and the ring and little fingers in two and three cases, respectively. The absence of recurrence was assessed by clinical examination after a mean follow-up of 43 (range, 36–55) months.
Immunohistochemical detection of nm23-H1
Immunohistochemical detection of nm23-H1 was performed by a pathologist (HVDW), who was blinded to the clinical outcome. Paraffin-embedded specimens and a monoclonal antibody directed at the cytoplasmic protein were used (Grover et al., 1998; Yamagushi et al., 1993). Four micrometre-thick sections of the paraffin-embedded specimens were cut and dewaxed. Endogenous peroxidase activity was blocked by incubation with 1% hydrogen peroxide in methanol for 30 minutes. Non-specific staining was then reduced by pre-incubation with normal horse serum (Vector PK6102) diluted 1/50 in Tris-buffered saline (TBS) for 20 minutes at room temperature. The sections were then incubated with mouse anti-human monoclonal antibody to nm23- H1 (Novo Castra) diluted 1/500 at room temperature for overnight. After washing with TBS the sections were incubated with biotinylated antimouse immunoglobulin (Vector PK6102) at room temperature for 30 minutes. After another wash with TBS, sections were covered with an avidin–biotin–peroxidase complex (Vector PK6102) in TBS at room temperature for 30 minutes. The antibody was visualized by reaction with 3-3′- diaminobenzene tetrahydrochloride (DAKO K3468) and hydrogen peroxide in 0:5mM Tris buffer (pH 7.2). Slides were slightly counterstained with haematoxylin. Negative controls were carried out by omitting the primary nm23-H1 antibody whilst a human breast carcinoma specimen acted as the positive control. All the sections were prepared at the same time and the immunohistochemical staining of all the specimens was performed simultaneously.
RESULTS
Peroperative evaluation (Table 1)
In Group I (recurrence), bone invasion was noted during the primary surgery in three cases, joint involvement in three and extension under the flexor and extensor tendons in six and two cases, respectively. Adherence to the neurovascular pedicle was observed in two cases and multinodularity in two cases. In Group II (no recurrence) there was neither bone invasion, multinodularity nor adherence to the neurovascular pedicle. Joint involvement occurred in one case and extension under the flexor and extensor tendons in two cases each (Table 1).
Immunohistochemical data
Nm23-H1 staining was not an all or nothing phenomenon: Nm23-H1 labelled most mononuclear spindle cells and most giant cells, but all specimens showed some giant cells which were strictly nm23-H1 negative. A discrete variability in the degree of expression was noted. The average signal intensity of staining was scored semiquantitatively as negative, weakly positive or positive.
When analysing the results of the staining intensity (Table 2), we did not find any difference between Groups I and II. Moreover, the only two really negative tumours were in group II (no recurrence).
The histological slides of the tumours with recurrence were evaluated and no significant differences in the microscopic characteristics or immunohistochemical staining patterns were observed between the primary and recurrent tumours.
DISCUSSION
The variety of names given to GCTTS attests that its cause and pathogenesis is unclear. Pathological studies of GCTTS reach conflicting conclusions and none has been able to clearly demonstrate any neoplastic, trauma-induced, immune-mediated, lipid metabolic process or inflammatory pathogenesis (O’Connell et al., 1995; Regezi et al., 1987; Ushijima et al., 1986; Vogrincic et al., 1997).
Treatment of GCTTS is by local excision and the reported recurrence rate varies from 4% to 44% (Al-Qattan, 2001; Grover et al., 1998; Marcucci and Foucher, 1991; Moore et al., 1984; Reilly et al., 1999). Cases of malignant GCTTS have been reported, but these lesions either fall into the category of clear cell sarcomas of tendons or are probably misdiagnosed malignant fibrous histiocytomas (Moore et al., 1984; Nielsen and Klaer, 1989; Noordanus et al., 1995).
Unfortunately, there are few identified clinical risk factors for recurrence, and risk factors identified by some workers are refuted by others. Amongst the proposed risk factors, mitotic activity, cellularity, multinodularity and bone or joint invasion are the most frequently proposed (Al-Qattan, 2001; Loréa et al., 2001; Monagan et al., 2001; Reilly et al., 1999; Uriburu and Levy, 1998).
Recently, Grover et al. (1998) have proposed that recurrent tumours have lost some of their properties of intercellular recognition and contact growth inhibition. In their retrospective study, recurrences were shown to be associated with the loss of expression of the nm23-H1 gene: patients with nm23-H1-negative tumours had a much higher recurrence rate than patients with nm23- H1-positive tumours. Out of the 52 tumours reviewed by Grover et al., eight had recurrence and 11 were nm23-H1 negative. Although the number of false-negative and false-positive tumours is not given, the association between nm23-H1 negativity and recurrence was highly significant. Moreover, after consideration of the site of the tumour, the number of nodules and the presence of bone erosion, nm23-H1 was shown to be an independent prognostic factor (P = 0.04).
The first nm23 gene was identified as a potential metastasis inhibitor and subsequent studies have shown that nm23 is also involved in tumour pathogenesis (Postel, 1998). At present, five nm23 genes (named from H-1 to H5) have been identified in humans (Lombardi et al., 2000). The proteins that they produce are implicated in cell proliferation, differentiation and cancer and probably alter the promoter DNA structure of other genes, thus regulating their expression. nm23- H1 is under expressed in breast carcinoma, metastatic melanoma and colorectal carcinoma and nm23-H2 is under expressed in colorectal carcinomas. Nm23 expression is not presently used in clinical practice for these tumours (Lombardi et al., 2000; Postel, 1998; Tseng et al., 2001; Yamagushi et al., 1993).
Neither the subtype of the gene investigated nor the commercial source of the antibody used are mentioned by Grover et al. (1998), but from the articles referred to in their paper, it is deduced that it was nm23-H1 gene expression which was labelled. Using the anti-nm23-H1 monoclonal antibody and the technique detailed in the reference quoted by Grover et al. (Yamagushi et al., 1993), we were unable to find any significant relationship between the risk of recurrence and the expression of the nm23-H1 protein.
It should be emphasized that although Grover et al. (1998) analysed the staining results simply as positive or negative, we found that only a few tumours were entirely positive or negative, and that the majority were heterogeneous with variable numbers of positive and negative cells. Moreover, there were no differences in staining between the primary and recurrent specimens which indicates that, unlike other tumours including chondromas, no evolution in the morphology of the tumour occurred with time. We looked for explanations the discrepancy between our results and those of Grover et al. (1998), but the technique used in both studies was identical and is routinely used in immunohistochemical detection: it is used daily in our laboratory.
We did however identify two possible biases in our study. First, the number of surgeons involved raises the possibility that all the tumours were not removed with the same accuracy and that some tumours in the recurrent group would not have recurred if they had been excised by another surgeon. However, all the operations were performed by experienced senior hand surgeons for whom hand surgery accounted for more than 90% of their practice. The second possible bias is the selection of the non-recurrent group. However, although late recurrences may occur (Weinzweig and Watson, 1997), recurrences of GCTTS after 3 years are exceptional, as demonstrated by the mean time for recurrence in this series of 11 months. Because there is no correlation at all between nm23-H1 expression and recurrence in our study, this bias could only slightly affect our results and should not affect our conclusion.
Although some clinical and histological markers may be different for recurrent and non-recurrent GCTTS, none is able to predict the risk for recurrence accurately. The relationships between some peroperative findings (bone invasion, joint involvement, and multinodularity) or histological parameters (mitotic rate) and the risk for recurrence have already been observed (Al-Qattan, 2001; Loréa et al., 2001; Monagan et al., 2001; Reilly et al., 1999; Uriburu and Levy, 1998), and the fact that this study confirms these previously described risk factors validates its design and increases our confidence in our conclusion that there is no correlation between nm23-H1 expression and recurrence. Nevertheless, the possibility of two groups of GCTTS remains, with recurrences occurring in one because of intrinsic characteristics of the tumour rather than because of incomplete excision. Peroperative risk factors such as bone invasion may either be interpreted as demonstrating aggressive clinical behaviour of a subgroup of the tumour, or rendering complete excision impossible.
Footnotes
Acknowledgements
The authors would like to acknowledge Drs F. Lemeunier, Y. Baeten and N. Chahidi for their contribution to this study.