Expression and prognostic significance of matrix metalloproteinases and their tissue inhibitors in primary neuroendocrine carcinoma of the skin☆

Materials and methods 

Immunohistochemistry 

Formalin-fixed, paraffin-embedded tissue specimens from PNCs used for all immunohistochemistry studies were retrieved from the Department of Human Pathology and Oncology, University of Florence. Tissue sections were incubated with primary antibodies to MMP-1, MMP-2, MMP-3, MMP-9, MMP-11, MMP-13, and MMP-14 and TIMP-1, TIMP-2, and TIMP-3 (Table 2).

Table 2. Antibodies used in the immunohistochemical study

	Antibody	Clone	Source	Dilution	Incubation Time	Pretreatment
	MMP-1	41-1E5	Oncogene, Boston, MA	1:400	1 hour	None
	MMP-2	42-5D11	Oncogene	1:10	2 hour	10 mM citrate buffer, MW (Micromed T/T Mega-Mileston Srl., Sorisole (BG), Italy) 35 minutes
	MMP-3	55-2A4	Oncogene	1:20	1 hour	0.5% proteasi XIV, r.t. 15 minutes
	MMP-9	56-2A4	Oncogene	1:10	1 hour	10 mM citrate buffer, MW 35 minutes
	MMP-11	SL3.05	Neomarkers, Fremont, CA	1:200	1 hour	None
	MMP-13	VIIIA2	Neomarkers	1:10	2 hour	10 mM citrate buffer, MW 35 minutes
	MMP-14	113-5B7	Chemicon International, Temecula, CA	1:40	2 hour	10 mM citrate buffer, MW 35 minutes
	TIMP-1	102D1	Oncogene	1:10	2 hour	10 mM citrate buffer, MW 35 minutes
	TIMP-2	T2-N IC3	Oncogene	1:10	2 hour	10 mM citrate buffer, MW 35 minutes
	TIMP-3	136-13H4	Chemicon International	1:50	2 hour	10 mM citrate buffer, MW 35 minutes

The specific MMPs were selected because of previous reports showing their increased expression, and even a prognostic impact for some of them, in several solid tumors, including small-cell neuroendocrine lung cancer. Immunohistochemistry was performed by the standard avidin-biotin-peroxidase complex method (Dako SpA, Milan, Italy) with diaminobenzidine as a chromogen and hematoxylin as a counterstain. The final dilution of each antibody was determined after preliminary serial dilution studies. Human colon adenocarcinoma was used as a positive control. Procedures used as negative controls included omission of the primary antibody from the staining protocol and replacement of the primary antibody with normal rabbit or mouse IgG in appropriate concentrations.

Analysis of immunohistochemical staining was independently performed by 2 investigators (D.M. and A.F.) without knowledge of the clinical and survival data. Immunoreactivities were semiquantitatively evaluated as “negative” (−) when no positive cells were found, “focal” (+) when positive tumor cells accounted for <20% of the total number, “moderate” (++) when 21% to 50% of tumor cells were positive, and “diffuse” (+++) when >50% tumor cells were positive. For statistical analysis, negative/focal stainings were classified as low expression and moderate/diffuse stainings as high expression of the respective antigens. Stromal staining of TIMPs was also assessed in fibroblasts and collagen stroma.

Results 

Patient population 

Clinical details for the study group are given in Table 3.

Table 3. Clinical data in 23 patients with PNC of the skin

	Case	Age (year)	Sex	Site	Number of Lesions	Stage	Type of Progression	Status	Follow-Up (months)
	1	57	M	Right leg	Multiple	Ib	Skin rec; systemic dis	DOD	47
	2	54	F	Back	Single	Ia	—	NED	291
	3	91	M	Neck	Single	II	Skin rec; systemic dis	DOD	5
	4	84	F	Right leg	Multiple	Ib	Skin rec; systemic dis	DOD	13
	5	65	F	Right leg	Single	Ib	—	NED	132
	6	84	F	Left cheek	Single	Ib	N+; systemic dis	DOD	11
	7	83	F	Forehead	Single	Ia	Skin rec; systemic dis	DOD	4
	8	84	M	Right leg	Single	Ib	Skin rec; skin met	DOD	16
	9	91	F	Right cheek	Single	II	Systemic dis	DOD	22
	10	85	F	Lower eyelid	Single	Ib	—	DOC	76
	11	81	M	Right leg	Multiple	Ib	—	NED	83
	12	68	M	Forehead	Multiple	Ib	Skin rec; N+; systemic dis	DOD	10
	13	81	M	Right leg	Single	Ia	—	NED	55
	14	57	F	Left thigh	Multiple	III	—	DOD	4
	15	85	F	Right thigh	Single	Ib	Systemic dis	DOD	24
	16	67	F	Right thigh	Single	Ia	—	NED	42
	17	80	F	Right cheek	Single	Ia	—	NED	37
	18	78	F	Left leg	Single	Ib	Skin rec	AWD	42
	19	89	F	Left knee	Single	Ib	Skin rec	DOC	9
	20	73	F	Left cheek	Single	Ia	Skin rec	AWD	16
	21	78	F	Left cheek	Single	Ia	—	NED	9
	22	89	M	Forehead	Single	Ia	—	NED	10
	23	69	F	Left arm	Multiple	Ib	—	NED	14

Abbreviations: N, lymph node metastasis; skin rec, local recurrence in the skin; skin met, metastasis in the skin or subcutaneous tissue beyond the regional lymph nodes; systemic dis, systemic dissemination; NED, no evidence of disease; AWD, alive with disease; DOD, dead of disease; DOC, dead of other causes.

The study included 7 (17.4%) males and 16 (69.6%) females with an age range of 54 to 91 years (mean, 77 years; median, 81 years). The tumors were located mainly on the lower extremities (n = 11) and the head and neck region (n = 10), with 1 case arising on the upper extremities and 1 case on the back. Fifteen (65.2%) tumors exceeded 2 cm in greatest dimension at the time of diagnosis. Follow-up of patients ranged from 4 to 291 months (mean, 42.2 months; median, 16 months). Disease progression was observed in 13 (56.5%) cases. All patients had been treated with surgery plus adjuvant radiotherapy. Nine patients were alive without evidence of disease, 2 were alive with tumor, 2 had died of other causes, and 10 had died of disseminated PNC. The 2-year survival rate was 54.2%. On histopathologic examination, only 2 cases exhibited surface ulceration. Cytologically, the small-cell type was observed in most cases (n = 20), whereas the tumor cell population comprised an admixture of small and intermediate to large cells in 3 cases. In most cases (n = 21) the architectural pattern was diffuse, whereas in the other 2 cases a mixture of organoid, trabecular, and diffuse patterns was seen. In terms of mitotic activity, 6 cases showed 0 to 5 mitoses per high-power field, 10 cases displayed 6 to 10 mitoses, and 7 cases showed ≥10 mitoses. As expected, the immunohistochemical profile showed diffuse positivity for cytokeratins AE1/AE3 and neuron-specific enolase in all cases. CK20 positivity (paranuclear dot pattern) was detected in 21 (91.3%) cases. Chromogranin A was positive in 18 (78.2%) cases.

Distribution of MMP and TIMP expression by immunohistochemistry 

In the 23 PNC specimens, tumoral immunoreactivity for MMPs was quite variable and heterogeneous within cases (Fig 1).

View Large Image Download to PowerPoint

Fig. 1. Representative immunohistochemistry of MMPs in PNCs of the skin. (A) MMP-1 immunoreactivity is present in the cytoplasm of neoplastic cells, as well as in the deeper epidermal layers. (B) Positive staining for MMP-2 is demonstrated in neoplastic cells and fibroblasts, as well as in endothelial cells around the tumor. (C) Diffuse MMP-3 expression in tumor cells, stromal cells, and the overlying epidermis is present. (D) Neoplastic cells do not express MMP-13, whereas scattered inflammatory cells and stromal cells around the tumor are positive. (E) Immunostaining for MMP-14. (F) Neoplastic cells do not express MMP-9, whereas eccrine glands are strongly positive.

Immunoreactivity was present in the cytoplasm of neoplastic cells. Although positive staining for some MMPs was noted in scattered fibroblasts, vascular endothelial cells, and inflammatory cells, such expression was not specifically evaluated and quantified. Stromal staining was poorly defined and gave the impression of being present predominantly in the extracellular matrix. MMP expression was also seen, at varying frequencies and intensities, within the epidermis and adnexal structures, with a particularly intense and constant immunoreactivity observed in eccrine glands. High (>20%) MMP-1 expression was found in 13 (56.5%) cases. MMP-2 was evidenced in 12 (52.1%) cases, 8 (34.7%) of which showed moderate or diffuse distribution of the immunostaining in neoplastic cells (high expression). MMP-3 was detected in 11 cases (47.8%), with high expression in 9 (39.1%) cases. MMP-9 expression was observed in 3 (13%) cases and MMP-14 expression in 11 (47.8%) cases, all with a moderate to diffuse staining. No immunoreactivity for MMP-11 and MMP-13 was found. Tumoral expression of TIMP-1 was found in 8 (34.7%) cases, with high expression in 3 cases, whereas high tumoral TIMP-3 expression was detected in 21 (91.3%) cases (Fig 2).

View Large Image Download to PowerPoint

Fig. 2. Representative immunohistochemistry of TIMP-1 (A) and TIMP-3 (B) in PNCs of the skin. Note that TIMP-1 expression is observed predominately within neoplastic cells, whereas TIMP-3 immunostaining is located within tumor cells, scattered stromal cells, and the epithelium of eccrine glands.

TIMP-2 expression was observed neither in neoplastic cells nor in stromal cells. Conversely, stromal cell expression of TIMP-1 and TIMP-3 was found in >90% of cases. Table 4 summarizes MMP and TIMP expression for each case.

Table 4. MMPs and TIMPs expression by neoplastic cells in 23 PNCs of the skin

	Case	MMP-1	MMP-2	MMP-3	MMP-9	MMP-14	TIMP-1	TIMP-3
	1	−	−	−	−	−	+	−
	2	−	+	−	+	−	+	−
	3	+++	−	+	−	+++	−	+++
	4	+++	+++	+++	−	−	−	+++
	5	+++	+	++	++	+++	++	+++
	6	+++	−	+++	−	+++	−	+++
	7	+++	−	+++	++	+++	−	+++
	8	+++	+++	+++	−	++	−	+++
	9	+++	+	++	−	++	+	+++
	10	−	−	−	−	−	−	+++
	11	−	+++	−	−	+++	++	+++
	12	+++	−	+	−	−	−	+++
	13	+++	−	−	−	−	−	+++
	14	−	+	++	−	+++	++	+++
	15	+++	++	++	−	+++	−	+++
	16	−	+++	−	−	−	+	+++
	17	+	−	−	−	−	−	++
	18	−	+++	−	−	−	−	+++
	19	++	+++	−	−	−	+	+++
	20	++	+++	−	−	+++	−	+++
	21	+++	−	++	−	+++	−	+++
	22	−	−	−	−	−	−	++
	23	−	−	−	−	−	−	+++
	−, Negative staining; +, focal staining when positive tumor cells accounted for <20% of the total number; ++, moderate staining when 21% to 50% of tumor cells were positive; and +++, diffuse staining when >50% tumor cells were positive.

Statistical analysis failed to identify any statistically significant correlation between MMP/TIMP expression and clinical parameters.

Prognostic factor analysis 

On univariate analysis, the only significant clinical factor that adversely affected survival was status greater than stage I (P = 0.01). High expression of MMP-1 (P = 0.04) and MMP-3 (P = 0.01) were also significant negative prognostic factors (Table 5).

Table 5. Univariate analysis (Cox proportional hazards method) assessing the influence of clinical and IHC variables on survival

	Variable	Cases (%)	Hazard Ratio	95% Confidence Interval	P Value
	Age				0.78
	≤65 years	4 (17.4%)	1.0
	>65 years	19 (82.6%)	1.2	0.2-6.1
	Sex				0.57
	Male	7 (30.4%)	1.0
	Female	16 (69.6%)	0.6	0.1-2.4
	Site				0.22
	Head/neck	10 (43.5%)	1.0
	Other sites	13 (56.5%)	0.4	0.1-1.6
	Number of lesions				0.25
	Single	17 (73.9%)	1.0
	Multiple	6 (26.1%)	2.1	0.5-7.5
	Stage				0.01
	I	20 (87.0%)	1.0
	>I	3 (13.0%)	6.3	1.4-27.1
	MMP-1				0.04
	Low expression	10 (43.5%)	1.0
	High expression	13 (56.5%)	4.9	1.0-23.5
	MMP-2				0.61
	Low expression	15 (65.2%)	1.0
	High expression	8 (34.8%)	0.7	0.1-2.7
	MMP-3				0.01
	Low expression	14 (60.9%)	1.0
	High expression	9 (39.1%)	5.4	1.3-21.6
	MMP-9				0.55
	Low expression	22 (95.7%)	1.0
	High expression	1 (4.3%)	0.04	0.001-1438.0
	MMP-14				0.07
	Low expression	12 (52.2%)	1.0
	High expression	11 (47.8%)	3.4	0.8-13.5
	TIMP-1				0.60
	Low expression	20 (87.0%)	1.0
	High expression	3 (13.0%)	0.5	0.07-4.6
	TIMP-3				0.65
	Low expression	2 (8.7%)	1.0
	High expression	21 (91.3%)	1.6	0.1-13.2

NOTE. Low expression, ≤20% of positive cells; high expression, >20% of positive cells.

Conversely, none of the other MMPs and TIMPs analyzed in this study demonstrated prognostic significance. Patients whose tumors showed high MMP-1 expression had a significantly shorter overall survival time than those with tumors with low expression (P = 0.02, log-rank test; Fig 3).

View Large Image Download to PowerPoint

Fig. 3. Probability of overall survival by MMP-1 expression (low expression versus high expression).

Similarly, patients whose tumors showed high MMP-3 expression had a significantly shorter overall survival time than those with tumors with low expression (P = 0.006, log-rank test; Fig 4).

View Large Image Download to PowerPoint

Fig. 4. Probability of overall survival by MMP-3 expression (low expression versus high expression).

Clinical and immunohistochemistry parameters were also assessed for their prognostic value on survival by multivariate analysis. Only stage (P = 0.02; hazard ratio, 4.4; 95% confidence interval, 1.2 to 15.8) persisted as a significant independent prognostic factor.

Discussion 

The present study is the first report on the immunohistochemical expression of MMPs and TIMPs in PNC of the skin. We focused our evaluation on the expression of these proteins and their inhibitors by tumor cells, and we demonstrated that immunoreactivity for MMPs and TIMPs was quite variable and heterogeneous. Among the different MMPs evaluated, MMP-1, MMP-2, MMP-3, and MMP-14 were expressed in a high proportion of cases, with the frequency of expression ranging from 47.8% to 56.5%. Expression of TIMP-1 and TIMP-3 was found in neoplastic cells in 34.7% and 91.3% of cases, respectively. In stromal cells, staining for both TIMP-1 and TIMP-3 was seen in more than 90% of PNC sections. Our results support the hypothesis that certain members of the MMP family may play a crucial role in PNC development and progression.

A major aim of this study was to assess the prognostic importance of MMP and TIMP expression on survival, while accounting for generally accepted significant clinical factors. In this regard, univariate analysis indicated that stage greater than I (P = 0.01) and high expression of MMP-1 (P = 0.04) and MMP-3 (P = 0.01) resulted in statistically significant factors that adversely affected survival, although multivariate analysis indicted that only stage was an independent prognostic indicator. Thus our results suggest that MMP-1 and MMP-3 may contribute to the aggressive behavior of and influence the invasive and metastatic potential of PNCs by allowing breakdown or proteolytic degradation of components of basement membrane and extracellular matrix. In particular, MMP-1 (also known as interstitial collagenase, mainly produced by fibroblasts and macrophages28, 29) preferentially cleaves fibrillar collagen. In line with our observations in PNCs, MMP-1 expression has been associated with an invasive process in the stromal tissues of breast, esophageal, gastric, colorectal, pancreatic, and lung cancers.30, 31, 32, 33, 34, 35, 36, 37 Furthermore, its expression has been significantly associated to shortened survival in esophageal and colorectal carcinoma.33, 35 MMP-3 (also known as stromelysin-I) degrades a wide spectrum of extracellular matrix components, including laminin, fibronectin, proteoglycans, gelatins, and nonhelical regions of type IV collagens.38 In line with our observations, MMP-3 overexpression has been correlated with poor outcome in patients with esophageal and lung cancer.21, 39

The MMP/TIMP profile has been evaluated in other types of neuroendocrine cells and tumors, including thyroid C cells and medullary thyroid carcinomas,40 normal pancreatic islet cells and islet cell tumors,41 anterior pituitary cells and pituitary adenomas,41 and carcinoids and small-cell lung cancers.21, 42 Interestingly, normal thyroid C cells were found strongly positive for MMP-2, MMP-9, TIMP-1, and TIMP-2, whereas medullary thyroid carcinomas were relatively more weakly stained.40 Similarly, pancreatic islet cells and islet cell tumors were found to be positive for more MMP family members (MMP-1, MMP-2, MMP-3, and MMP-9) and TIMPs (TIMP-1 and TIMP-2), although in general islet cell tumors were weakly stained in comparison with normal islets in the adjacent pancreatic tissue.41 On the basis of these observations, it has been suggested that MMPs/TIMPs can be considered possible markers of neuroendocrine cells and that MMP activity exists in endocrine tumors, although the MMP/TIMP balance may not be the major determinant in endocrine tumor invasion and metastasis.40 At variance with this hypothesis, however, Michael et al21 demonstrated that in small-cell neuroendorine carcinomas of the lung (which share morphologic similarities with PNCs), MMPs, and TIMPs were widely expressed and that increased tumoral expression of MMP-3, MMP-11, and MMP-14 were independent negative prognostic factors for survival, supporting their role in small-cell cancer progression.21

To better clarify the significance of the MMP/TIMP profile in the development of PNCs of the skin, it would be appropriate to evaluate their expression in the normal counterpart of PNCs. However, the histogenesis of PNC is controversial. Originally thought to be of eccrine origin,1 PNC was subsequently believed to originate from Merkel cells, the normal neuroendocrine cell population of the skin, on the basis of the ultrastructural recognition of neuroendocrine granules within the cytoplasm of tumor cells.2 A more recent hypothesis is that pure PNCs and mixed neuroendocrine-epithelial cutaneous carcinomas are of epithelial (epidermal or adnexal) origin, having undergone complete or partial neuroendocrine differentiation.43 In this view, it is important to emphasize that although MMP expression was seen at varying frequencies and intensities within the epidermis and adnexal structures, a particularly strong and constant immunoreactivity was observed in normal eccrine glands.

Among the different techniques that can be used to investigate the MMP/TIMP profile, immunohistochemistry has the advantage of allowing direct correlation with morphology. Furthermore, in immunohistochemical studies mRNA may not be translated, whereas protein expression is functionally more relevant. Another practical advantage is that immunohistochemistry can be performed on paraffin-embedded specimens, and thus it is practical for routine assessment of MMPs and TIMPs in diagnostic practice. However, it has been suggested that immunohistochemistry cannot differentiate between latent MMPs and activated forms.12 Other techniques, including zymography and quenched fluorescent substrate hydrolysis, separate latent and active forms of MMPs, although they do not allow correlation with morphology.37

It is fair to comment that interpreting the results of MMP and TIMP immunohistochemical expression in human tumors is rather complicated. Indeed, tissue localization of specific MMPs and TIMPs in and around the tumor can be heterogeneous, with variable expression within the tumor and adjacent stromal cells. It is not yet clear whether the overexpression of a particular enzyme or inhibitor reflects its functional role in the malignant process or rather is a sign of the host response to the tumor itself.11 In our study, immunolocalization of MMPs and TIMPs was found largely in tumor cells, as well as in fibroblasts and endotelial cells in the adjacent stroma. Our study confirms previous findings and provides further evidence that tumor cells and stromal cells are the major sources of various members of MMP and TIMP families in tumor tissues. Indeed, it is conceivable that dynamic host–tumor interactions modulate MMP/TIMP activity in the progression of human tumors, including PNCs.

In conclusion, the present study has for the first time comprehensively evaluated by immunohistochemistry the MMP/TIMP profile in specimens of human PNC. By univariate analysis, high MMP-1 and MMP-3 expression was shown to be a negative prognostic factor for survival, suggesting that MMP-1 and MMP-3 may contribute to the aggressive behavior of and influence the invasive and metastatic potential of PNCs. There is no doubt that the pathophysiologic roles of specific MMPs and TIMPs in the biology of PNC requires further study; however, if the current results are confirmed in independent datasets, then it is conceivable that future attempts to specifically block MMP-1 and MMP-3 activity may provide a novel means to inhibit invasiveness and distant spread in selected patients with PNC.