Association between syndecan-1 expression and clinical outcome in squamous cell carcinoma of the head and neck.

Syndecans are a family of cell-surface heparan sulphate proteoglycans which are involved in cell-matrix interactions and growth factor binding. Syndecan-1 binds basic fibroblast growth factor (bFGF) and several components of the extracellular matrix. Syndecan-1 expression is induced during keratinocyte differentiation and reduced during the formation of squamous cell carcinomas (SCCs). The purpose of this study was to examine the association of syndecan-1 expression with prognostic factors and clinical outcome in SCC of the head and neck. Frozen sections of 29 primary SCCs were analysed for syndecan-1 expression using immunohistochemical methods. Intermediate or strong staining for syndecan-1 was associated with a smaller primary tumour size (P = 0.0005) and higher histological grade of differentiation (P = 0.006) than negative or weakly positive staining. In a univariate analysis, syndecan-1-positive tumours were associated with higher overall (P = 0.001) and recurrence-free survival (P = 0.003) than those tumours with no or little syndecan-1 expression. The results suggest that syndecan-1 could be an important prognostic factor of SCC of the head and neck. Further studies on the prognostic significance of syndecan-1 expression in SCCs are warranted. ImagesFigure 1

Cell adhesion molecules, such as integrins, cadherins and cell-surface proteoglycans, are involved in the regulation of cell differentiation, proliferation, morphology and migration (Gallagher, 1989;Ruoslahti, 1991;Takeichi, 1991;Bernfield et al., 1992). In the past few years it has become evident that this group of molecules working in concert is essential for the maintenance of normal cellular functions. The development of malignant epithelial tumours is associated with reduced intercellular adhesion, disturbed differentiation and changes in the composition of the basement membrane (Weinstein et al., 1976;Liotta et al., 1986), suggesting that the expression and function of cell adhesion molecules could also change during malignant transformation. Indeed, alterations in the expression of many cell adhesion molecules associated with the development of carcinomas have been reported (Virtanen et al., 1990;Inki et al., 1991;Navarro et al., 1991;Schipper et al., 1991).
Syndecans are a family of cell-surface heparan sulphate proteoglycans (HSPGs) which have been suggested to participate in cell-cell and cell-matrix interactions and the binding of growth factors (reviewed by Jalkanen et al., 1991Jalkanen et al., , 1993Bernfield et al., 1992). The primary structure of the core proteins of four different syndecans, syndecan-l to syndecan-4, is known to date (reviewed by Bernfield et al., 1992). All syndecans comprise an extracellular domain containing covalently linked heparan>sulphate chains, a transmembrane domain and a relatively short cytoplasmic domain Bernfield et al., 1992). Syndecan-l has been shown to bind several extracellular matrix molecules via its heparan sulphate chains, including fibrillar collagens, fibronectin, thrombospondin and tenascin (Sun et al., 1989;Elenius et al., 1990;Salmivirta et al., 1991). The localisation of syndecan-l on the surfaces of stratified epithelial cells (Hayashi et al., 1987), as well as its expression in mesenchymal cell aggregates during organ development (Vainio et al., 1989), suggests that syndecan-I could also play a role in cell-cell adhesion. Furthermore, syndecan-l can bind bFGF (Kiefer et al., 1990;Elenius et al., 1992) and present the growth factor to the signalling tyrosine kinase receptor system . Via binding bFGF, syndecan-l could be involved in the regulation of the angiogenic activities of bFGF (Folkman & Shing, 1992) and, in turn, neovascularisation during the development of malignant neoplasia.
During organ development, syndecan-I production is induced concomitantly with epithelial-mesenchymal interactions in a developmentally highly regulated fashion (Thesleff et al., 1988;Sutherland et al., 1991;Vainio & Thesleff, 1992). In adult tissues, syndecan-I expression is almost entirely restricted to epithelial tissues, stratified epithelia containing the most abundant expression (Hayashi et al., 1987;Saunders et al., 1989). In stratified epithelia, syndecan-l is localised over the entire surface of keratinocytes, especially in the suprabasal cell layers, whereas the basal cell layer shows modest expression (Hayashi et al., 1987;Inki et al., 1991).
Syndecan-I expression is induced during the calcium-induced differentiation of cultured keratinocytes (Sanderson et al., 1992;Inki et al., 1994), suggesting that syndecan-l may be involved in keratinocyte differentiation. However, during malignant transformation of keratinocytes syndecan-l expression is progressively lost from both premalignant epithelial lesions and SCCs (Inki et al., 1991(Inki et al., , 1992a(Inki et al., , 1993. In SCCs, syndecan-I is localised immunohistochemically in keratinising cells of the horn pearls in well-differentiated tumours, whereas poorly differentiated SCCs are devoid of syndecan-l (In.ki et al., 1992a(In.ki et al., , 1994. In premalignant lesions of stratified epithelia, syndecan-I is lost from the basally located, atypical cell layers (Inki et al., 1991). Loss of syndecan-l from malignantly transformed cells could be one mechanism by which tumour cells loosen their attachment to each other and to the extracellular matrix and become non-responsive to the signals coming from their microenvironment.
The altered expression of syndecan-1 in SCCs and premalignant lesions of stratified epithelia suggests that syndecan-I could have prognostic value in the determination of the clinical outcome of these lesions. This has not been studied previously. The purpose of this study was to analyse syndecan-I expression immunohistochemically in frozen sections of primary SCCs of the head and neck as well as the relationship between syndecan-I expression and the clinical outcome in patients with SCC of the head and neck.

Patients
Twenty-nine patients histologically diagnosed as having squamous cell carcinoma of the head and neck between 1988 and 1991, and treated in Turku University Central Hospital, P. INKI et al. were included in the study. The diagnosis and analysis of syndecan-1 expression were made from biopsy specimens, which were taken before treatment and immediately snap frozen in liquid nitrogen. All patients with frozen tissue available were included in the study. The patient characteristics, treatment given and follow-up status are shown in Table  I. Twenty-two (76%) of the patients were male and seven (24%) female, and the median age at the time of diagnosis was 65 years (range 38 to 93 years). The WHO performance status was determined as described by Miller et al. (1981). All patients underwent treatment planned to be curative; three patients were treated with radical surgery, nine with radical radiotherapy and 17 with the combination of radical surgery and radiotherapy (Table I). The radiation dose varied from 50 to 72 Gy, with 2 Gy given daily (10 Gy per week). Staging was done according to the UICC TNM classification (1987). The patients have been followed up for a median of 25 months after the diagnosis (range 13-41 months) if still living or until death (17 patients). The overall 2 year survival rate was 42%.

Histology
All histological material obtained from the tumours was re-examined and divided into three histologic grades of differentiation according to the WHO classification (Shamugaratnam & Sobin, 1978) by one pathologist. H&E-and van Gieson-stained 4;Lm sections were used.
Staining of syndecan-J Syndecan-l was localised in tissue sections using a polyclonal affinity-purified rabbit antibody (anti-PI17) described pre-viously (Inki et al., 1994). Anti-P117 was raised against a synthetic peptide corresponding to the cytoplasmic sequence of human syndecan-l and subsequently purified using a cyanogen bromide (CNBr)-activated affinity column coupled with the peptide (Inki et al., 1994). Anti-PI 17 was shown to react specifically with syndecan-l in Western blot and immunohistochemical assays (Inki et al., 1994). As an internal standard, sections from histologically normal tissue from buccal mucosa and tongue were included in each staining of SCCs (protocols approved by the Joint Committee on Ethics of the Turku University and the University Central Hospital of Turku). Tissue samples were cut into 5 Im sections with a cryomicrotome and the sections were stored at -70°C until use.
The sections were stained using anti-PI 17 and the avidin-biotin immunoperoxidase (ABC) technique. After fixing with acetone for 5 min at 4°C, the slides were incubated with 2% normal goat serum in 0.01 M Trisbuffered saline, pH 8.0 (TBS), for 30 min at room temperature (RT), followed by anti-P117 (5 gml-1) in 1% (w/v) BSA-TBS (Sigma, St Louis, MO, USA) overnight at 4°C. Normal rabbit IgG (Sigma), incubated in the same way, was used as a negative control. After washing, the slides were incubated with Vectastain biotinylated anti-rabbit IgG for 30 mmn at RT (Vector Laboratories, Burlingame, CA, USA), followed by avidin DH-biotinylated horseradish peroxidase H mixture for 30 min at RT (Vector Laboratories). For colour reaction, the slides were incubated with 0.02% hydrogen peroxide, 0.68 mg ml-' imidazole and 0.1% (w/v) diaminobenzdine tetrahydrochloride in 0.1 M Tris buffer, pH 7.2, for 5 min, counterstained with haematoxylin and mounted. Staining intensity for syndecan-l was classified as: -, negative; +, weak staining of tumour cells; +,

Statistical analysis
Survival analysis was done using a BMDP computer program (BMDP Statistical Software, Department of Biomathematics, University of California Press, Los Angeles, CA, USA). Survival was estimated with the product-limit method, and comparison of survival between groups was done using the log-rank test (BMDP 1L). All P-values are two-tailed. Frequency tables were analysed with the chi-square test or Fisher's exact test.

Results
Expression of syndecan-l in SCCs of the head and neck Syndecan-l was localised in frozen sections of SCCs of the head and neck using anti-PI 17 antibody and immunohistochemical methods. Anti-PI 17 has previously been shown to react specifically with the cytoplasmic domain of human syndecan-l core protein (Inki et al., 1994). Staining intensity was compared with staining of normal epithelia from corresponding locations, which showed strong staining of keratinocyte cell surfaces (Figure la). Seven (24%) SCCs showed negative (Figure lb) and ten (34%) weakly positive staining for syndecan-l (Table I) Intermediate or strong staining for syndecan-I was associated with small primary tumour size (P = 0.0005, Table  II). However, no association between syndecan-l expression a and the presence of cervical nodal metastases was found (Table IL). Tumours that expressed syndecan-l (+ or + +) were more often well differentiated than those that did not express it (P =0.006, Table II). Syndecan-l expression did not correlate significantly with age at diagnosis, WHO performance status or gender (Table II).

Relationship ofsyndecan-J expression with clinical outcome
In a univariate analysis, syndecan-l staining intensity correlated well with overall survival and patients with a syn-TabMe H Correlation of syndecan-l expression with six clinicopathological factors in SCCs of the head and neck  decan-1-positive tumour had a more favoura The difference in survival wassignicant both staining groups were compared with each oth and when tumours with strong syndecan-I ex or +) grouped together were compared wit negative (or ±) tumours ( Figure 2a, P = year survival rates for syndecan-I + +, + tumours were 80%, 80%, 30% and 0% respe year survival rate for tumours expressng synx + +, n=17) was 81 %, whereas that of tumours with no or weak expression (or 4 only 18%. Similarly, patients with a tumour w expression had a higher recurrece-free surviF with a tumour with no or lttle syndecan-I exp survival 63% vs 18%, P=0.003, Figure 2b).

Dis dss
We have previously shown alterations in the syndecan-I during malignant transformation.
overall reduction of expresson found in tissue SCCs and adenocarcinomas (Inki et al., 1991, togther with biochemical changes such as  (Ink et al., 1992b). Furthermore, transfection of malignant, pression (+ + fusiform carcioma cells with syndecan-l restores their h syndecan-l-epithelial morphology and reduces their tumorigenicty in 0.001). The 2 nude mice (Lepp et al., 1992). These results suggest that ±, andsyndecan-l is involved in the regulation of cell morphology, ctively. The 2 adhesion and differentiation, and that loss of syndecan-l decan-I (+ or from transformed cells couldbe associated with uncontrolled patients with proliferation, reducd adhesion and disurbed differentiation , n = 12) was of tumour cells The present study is the first one to describe ith syndecan-l the prognosticsignicance of syndecan-l expresson in maligval than those nant tumours. Syndecan-l expression was shown to correlate ression (2 year with the clinical outcome of SCC of the head and neck, with syndecan-l-positive tumours being associated with a more favourable prognosis. Syndecan-I expression showed a highly signiant association with both overall survival and recurrence-free survival. A staistically significnt correlation between syndecan-I expression and survival was observed when all four saining groups were compared with each other expression of (P =0.005), and the lowest P-value was obtained when These include negative or weakly positive syndecan-l expression (or ±) sections from was com with positive expression (+ or + +) 1992a, 1994), (P =0.001). altered glyco-Syndecan-l has previously been shown to be induced during normal differentiation of keratinocytes (Sanderson et al., 1992;mnuci et al., 1994). Furthermore, syndecan-I is suppressed in malignant, poorly differentiated SCCs (Inci et al., 1991(Inci et al., , 1992a(Inci et al., , 1994, wheas positive expression is found in well-differentiated tumours (Inki et al., 1992a(Inki et al., , 1993, suggesta ing that syndecan-l is involved in keratinocyte differentiation in both normal and malignant cells. In line with these findings, we found a staisically signnt correlation + or ++ between syndecan-l expression and histological differen tiation in SCC of the head and neck. This result provides further support for the yet unidentified role of syndecan-l in keratinocyte differentiation, which is lost from aplasc cells. However, the role of syndecan-l may not be limited to cell differentiation, because the histological grade alone does not have prognostic value in a univariate analysis in the present series (P = 0.45). The poor prognostic value of the histological grade may be related to the limited size of the 118% seres or to the subjective nature of histological grading of SCC (Sorensen et al., 1989). However, the prognostic value of the histological grade in SCC of the head and neck is currently not estabished. Although many studies have dem-40 50 onstrated its prognostic s a (Wirnik et al., 1991), its value has also been disputed by others (Dreyfuss & Clark, 1991;Bungaard et al., 1992;Trueson et al., 1992). Our study showed a highly sificat association between syndecan-I ecpression and the prinary tumour size, which has prognstic value in a univariate survival analysis in the present series (P= 0.007) and many others (Bundgaard et al., b 1992;Cerezo et al., 1992). In addition, syndecan-l expression was associated with the patients' WHO performance status.
Although a multivariate analysis was not carried out because of the lmited size of the seres, in addition to syndecan-l expresson only T-stage (P = 0.007) and WHO performance % status (P = 0.003) showed statistclly sin t association with survival (data not shown). This suggests that syndecan-l expresson may be one of the most important prognostic factors in SCC of the head and neck.