Measurement of invasive potential provides an accurate prognostic marker for giant cell tumour of tendon sheath

M E A S U R E M E N T OF I N V A S I V E P O T E N T I A L P R O V I D E S A N ACCURATE PROGNOSTIC MARKER FOR GIANT CELL T U M O U R OF T E N D O N S H E A T H R. GROVER, A. O. GROBBELAAR, P. I. RICHMAN and P. J. SMITH

From the Departments of Plastic Surgery and Histopathology, Mount Vernon Hospital, Northwood, UK Giant cell turnouts of tendon sheath vary from solitary nodules to a multinodular variety that exhibits local infiltration. Recent advances in molecular oncology have defined a gene, nm23, expressed in normal cells which is responsible for inhibiting infiltration. The aim of this study was to investigate the expression of nm23 in a series of 52 giant cell tumours using immunohistochemistry and to assess its prognostic potential, nm23 expression was absent in 21% of tumours and this was associated with a highly significant risk of local recurrence (P<0.0001). Multivariate analysis of outcome showed rim23 expression to be more reliable than other clinicopathological parameters for predicting outcome. This immunohistochemieal test for nm23 is easily performed on standard paraffin sections and is recommended as an accurate prognostic marker for giant cell tumours of tendon sheath.

Journal of Hand Surgery (British and European Volume, 1998) 23B: 6:728-734 The giant cell tumour of tendon sheath has been classified as being of fibrohistiocytic origin and is remarkable for its tendency to show a range of histological appearances from tumours that are almost purely fibrous to those that are mostly histiocytic. Because of the varied morphology this disease spectrum has been described by a number of terms. The term pigmented villonodular synovitis is largely a descriptive phrase for those less well circumscribed giant cell tumours that grow more diffusely and incorporate haemosiderin pigment. This condition has been thought to have an inflammatory aetiology (Nielsen and Kiaer, 1989) whilst more recent reports including deoxyribonucleic acid (DNA) analysis by flow cytometry support a truly neoplastic origin (Abdul-Karim et al. 1992). This diversity is also reflected in the clinical behaviour of giant cell tumours, with the majority manifesting as slow growing, firm, non-tender solitary masses although a linked multinodular variety also exists. A significant number demonstrate local infiltration into the tendon sheath and erosion of bone (Fig 1). This group is particularly troublesome with recurrences occurring even after seemingly curative surgery. Recent research into prognosis in neoplastic disease has focused on genes responsible for malignant progression and invasive potential. The process of local invasion and metastasis relies on the loss of intercellular recognition and contact inhibition to growth. The nm23 gene has recently been described as an antimetastasis and anti-invasion gene, whose product is a crucial intermediary in the cellular control pathway for local recognition (Yamaguchi et al. 1993). immunohistochemical evaluation of nm23 expression has proved to be a useful prognostic marker of invasion and metastasis in cutaneous melanoma (Xerri et al. 1994). The aim of this study was to investigate the prognostic potential of both clinical parameters and nm23 expression in a series of giant cell tumours of tendon sheath affecting the hand.

PATIENTS A N D M E T H O D S Patients and specimens

Paraffin embedded tumour specimens from 52 patients were obtained from the histopathology archive at Mount

Fig 1

728

Radiograph of giant cell tumour of tendon sheath illustrating bone erosion

GIANT CELL TUMOUR OF TENDON SHEATH

Vernon Hospital. All specimens were obtained at the time of primary surgical treatment that was performed by the senior author (PJS). Data was collected on the site of the tumour (finger/thumb or hand), nodularity of the specimen (single or multiple) and the presence of bone erosion. Follow-up was obtained on all patients over a median period of 79 months (range, 7-174 months) with documentation of any recurrence during that time. lmmunohistochemical detection of nm23 ran23 expression was measured using immunohisto-

chemistry with a monoclonal antibody directed at the cytoplasmic protein as described by Yamaguchi et al. (1993). Four micrometre-thick sections of paraffin embedded giant cell tumour were cut and de-waxed. Endogenous peroxidase activity was blocked by incubation with 1% hydrogen peroxide in methanol for 30 minutes. Sections were pre-treated with 0.1% trypsin solution in distilled water (pH adjusted to 7.8 by addition of 0.1 M sodium hydroxide solution) for 10 minutes at 37°C. Non-specific staining was then reduced by pre-incubation with normal swine serum (Dako Ltd, UK) diluted 1:5 in Tris buffered saline (TBS) for 10 minutes at room temperature. The sections were then incubated with mouse antihuman monoclonal antibody to nm23 at room temperature for 90 minutes. After washing with TBS the sections were incubated with biotinylated rat antimouse immunoglobulin (Dako Ltd, UK) at room temperature for 60 minutes. After another wash with TBS, sections were covered with a 1% solution of avidin-biotinperoxidase complex (Dako Ltd, UK) in TBS at room temperature for 30 minutes. The antibody was visualized by reaction with 3-3'-diaminobenzene tetrahydrochloride and hydrogen peroxide in 0.5 mM Tris buffer (pH 7.2). Slides were lightly counterstained with haemotoxylin. Negative controls were carried out by omitting the primary nm23 antibody whilst human breast carcinoma acted as the positive control. Expression of nm23 was graded as positive or negative by a senior pathologist (PIR) independent of any knowledge of the clinical follow-up. Statistical analysis

729

RESULTS Clinicopathological correlations The median age of the patients in the study was 37 years (range, 9-73) with a sex distribution of 33 female: 19 male. The sites of the tumours, the number of nodules and the presence of bone erosion are given in Table 1. Eight patients in the study had a recurrence, all of whom were treated by further surgery. There was a borderline association between the site of tumour and higher rates of recurrence in the thumb (P=0.07). However, there was a statistically significant association with the number of tumour nodules, with higher rates of recurrence in patients with two or more nodules (P=0.00t8) (Fig 2). The presence of bone erosion was a significant factor in predicting the recurrence of giant cell tumours (P<0.0001) (Fig 3). However, as four out of the eight recurrences in the series occurred without bone erosion, this factor alone would not have predicted these. Clinical significance of nm23 expression Immunohistochemical detection of nm23 was observed in 41 tumours (79%) (Fig 4) with the remaining 11 patients showing no cytoplasmic staining (Fig 5). The prognostic significance of nm23 expression was investigated by dividing patients into two groups according to nm23 positivity. Recurrence curves for each group were plotted using the Kaplan-Meier method (Fig 6). Patients whose original tumour was nm23 negative showed significantly worse outcome (P<0.0001) supporting the role of this gene as an inhibitor of infiltration. In order to determine whether nm23 expression maintained an independent prognostic association with outcome, recurrence data was analysed using the Cox's proportionM hazards model taking into account the site of the tumour, the number of nodules and the presence of bone erosion. The results of this analysis, ranked in order of prognostic significance, are given in Table 2. Expression of nm23

Table l--Clinicopathological parameters and clinical outcome in patients with giant cell tumour of tendon sheath Outcome o f jbllow up

o

Correlation of rim23 expression with clinical outcome was performed by plotting recurrence curves for patients using the Kaplan Meier method for analysis of censored data (Kaplan and Meier, 1958). The significance of any observed difference was calculated by the Log-Rank test (Peto et al. 1977). Assessment of nm23 as an independent prognostic marker was performed by multivariate analysis applying Cox's proportional hazards model (Cox, 1972) to compare outcome with the site of the tumour, the number of nodules and the presence of bone erosion.

No Recurrence

Recurrence

Total

Hand

30 4 10

3 4 1

33 8 11

Nodules Single Multiple

35 9

4 4

39 13

Bone erosion Present Absent

3 41

4 4

7 45

Site

Finger Thumb

730

THE JOURNALOF HANDSURGERYVOL.23BNo. 6 DECEMBER1998 1

1 ¢,,,I

0.9 ~a

0.8

I,,,,,

=eo

0.7

o

0.5

.~ az

0.4 0.3 0.2 0.1

t~

=

o.6

~,,

~a

Single nodules (n=39)

-~

Multiple nodules (n=13)

,

i

i

20

40

60

i

i

i

i

Bone erosion absent (n=45)

0.8 0.7 0.6

Bone erosion present (n=7)

0.5

,~ ,~

0

0.9

i

i

i

80 100 120 140 160 180 200 Time (months)

Fig 2 Kaplan-Meier recurrence curve of patients with giant cell turnout of tendon sheath stratified according to number of nodules. Patients with multiple nodules showed significantly higher rates of local recurrence (P = 0.0018).

0.4 0.3 0.2 0.1 0 0

20

i

i

40

60

i

~

i

i

i

i

Time (months)

Fig 3 Kaplan-Meier recurrence curve of patients stratified according to the presence of bone erosion• Patients with bone erosion showed significantly higher rates of local recurrence (P = 0.0001).

@v,.

Fig 4

i

80 100 120 140 160 180 200

v•

i

Photomicrograph illustrating a section of giant cell tumour of tendon sheath with positive nm23 expression identified by immunohistochemistry (original magnification x90).

Fig 5 Photomicrograph illustrating a section of giant cell tumour of tendon sheath with no nm23 expression identified by immunohistochemistry (original magnification xg0).

maintained an independent association with clinical o u t c o m e and proved to be the m o s t accurate prognostic marker in this study, over and above all other clinicopathological parameters.

by surgery although recurrence rates between 5% and 50% have been reported (Glowacki and Weiss, 1995; Sirinivasa R a o and Vigorita, 1984). The c o m m o n m o d e o f presentation is with a firm, non-tender slowly growing mass fixed to deeper structures. It occurs usually on the flexor aspect, although attachment to the extensor apparatus is well d o c u m e n t e d (Hansen et al. 1988). F r a n k metastasis has not been reported, although local infiltration is a relatively c o m m o n feature. O u r series confirmed the female p r e p o n d e r a n c e that has been reported in similar series (Glowacki and Weiss, 1995). In addition, the observed recurrence rate o f 15% fell within the published range (Glowacki and Weiss, 1995; Sirinivasa R a o and Vigorita, 1984). A definite association was observed between clinical o u t c o m e and the n u m b e r o f nodules as well as the presence o f bone erosion. However, neither o f these parameters were

DISCUSSION Giant cell t u m o u r o f t e n d o n sheath is the second c o m m o n e s t soft tissue mass presenting in the h a n d after ganglion cysts (Glowacki and Weiss, 1995). A l t h o u g h there is some speculation a b o u t the aetiology o f this disease, the evidence appears to support a neoplastic rather than inflammatory process. Flow cytometry o f D N A extracted from giant cell t u m o u r specimens has revealed a significant n u m b e r o f aneuploid cases that strongly favours a genetic cause for neoplastic transformation ( A b d u l - K a r i m et al. 1992). The mainstay o f treatment is

GIANT CELL TUMOUR OF TENDON SHEATH

0.9

731

Table 2--Multivariate analysis of recurrence using Cox's proportional hazards model with parameters ranked in order of prognostic significance

nm23 positive (n=41)

0.8

0.7 0.6 0.5

nm23 negative (n= 11)

Parameter

Probability

nm23

P=0.04 P=0.08 (NS) P=0.16(NS) P=0.62 (NS)

Site Number of nodules / Bone invasion

0.3 0.2 0.

0

i

i

L

20

40

60

i

i

i

i

i

i

i

80 100 120 140 160 180 2OO Time (months)

Fig 6

Kaplan-Meier recurrence curve of patients stratified according to nm23 expression. Patients without expression of this antiinvasion gene showed significantlyhigher rates of local recurrence (P = 0.0001).

specific enough to predict more than half of the total number of recurrences in the series. The nm23 gene codes for a cytoplasmic protein, which is a crucial intermediary in the pathway for control of local invasion and subsequent metastasis (Yamaguchi et al. 1993). Down-regulation of this gene has therefore been associated with the prediction of invasive potential and metastasis in cutaneous melanoma (Xerri et al. 1994). Expression of this gene in giant cell tumours of tendon sheath revealed 79% to be immunopositive. The remaining 21%, which on theoretical grounds would have a greater invasive potential, were indeed found to be associated with a significantly higher rate of recurrence after surgery. In clinical practice, evaluation of nm23 expression can be performed on standard paraffin embedded tumour specimens and the technique of immunohistochemistry is reliable enough to be carried out in a routine histopathology department (Xerri et al. 1994). The finding of down-regulation of the nrn23 gene in giant cell tumours of tendon sheath also provides further evidence of their neoplastic origin. Although this gene has been found to be related to invasive and metastatic potential in m a n y human malignancies, there are as yet no reports of its down-regulation in inflammatory conditions.

Acknowledgements The authors wish to thank the Restoration of Appearance and Function Trust (RAFT) and the Cancer Research Campaign for generously supporting this work.

References Abdul-Karim FW, EI-Naggar AK, Joyce MJ, Makley JT, Carter JR (1992). Diffuse and localized tenosynovial giant cell tumour and pigmented villonodular synovitis: a clinicopathologic and flow cytometric DNA analysis. Human Pathology, 23: 729~35. Cox DR (1972). Regression model and life tables. Journal of the Royal Statistical Society, 34:187 220. Glowacki KA, Weiss AP (1995). Giant cell tumors of tendon sheath. Hand Clinics, 11: 245~53. Hansen R Nielsen PT, Wahlin AB (1988). Pigmented villonodular synovitis of the extensor tendon sheaths in a child. Journal of Hand Surgery, 13B: 313 314. Kaplan EL, Meier P (1958). Nonparametric estimation of incomplete observations. Journal of the American Statistical Association, 53: 457481. Nielsen AL, Kiaer T (1989). Malignant giant cell tumour of synovium and locally destructive pigmented villonodular synovitis. Human Pathology, 20: 765-771. Peto R, Pike MC, Armitage P e t al. (1977). Design and analysis of clinical trials requiring prolonged observation of each patient: analysis and examples. British Journal of Cancer, 35: 457481. Sirinivasa Rao A, Vigorita VJ (1984). Pigmented villonodular synovitis. Jourval of Bone and Joint Surgery, 64A: 76 94. Yamaguchi A, Urano T, Fushida Set al. (1993). Inverse association of nrn23-Hl expression by colorectal cancer with liver metastasis. British Journal of Cancel; 68: 1020-1024. Xerri L, Grob J J, Battyani Z (1994). NM23 expression in metastasis of malignant melanoma is a predictive prognostic parameter correlated with survival. British Journal of Cancer, 70:1224 1228.

Received. 20 February 1998 Accepted after revision.23July 1998 Mr R. Grover, Department of Plastic Surgery,Mount Vernon Hospital, Northwood, MiddlesexHA6 2RN, UK. ~31998 The British Societyfor Surgery of the Hand