ORIGINAL PAPER

MMP2 expression is a prognostic marker for primarymelanoma patients

Anand Rotte & Magdalena Martinka & Gang Li

Accepted: 6 May 2012 /Published online: 6 June 2012# International Society for Cellular Oncology 2012

AbstractBackground Matrix metalloproteinase 2 (MMP2) is a colla-genase, which aids tumor growth and invasion by digestingthe extracellular matrix surrounding the tumor tissue. Ourstudy examined MMP2 expression in various stages of mela-noma progression and tested the prognostic significance ofMMP2 expression. We also analyzed the correlation betweenp-Akt status and MMP2 expression in melanoma patients.Methods Tissue microarray (TMA) and immunohistochem-istry were employed to study the expression of MMP2. Atotal of 482 melanoma (330 primary and 152 metastatic)tumor biopsies and 149 nevi biopsies (49 normal and 100dysplastic nevi) were used for the analysis. MMP2 expres-sion was correlated with melanoma progression. Kaplan-Meier survival curve and multivariate Cox regression anal-ysis were applied to verify the prognostic significance ofMMP2 expression. The correlation between MMP2 and p-Akt expression was analyzed in 92 cases which were com-mon in the present and the previous study on p-Aktexpression.Results Strong MMP2 expression is significantly increasedin primary (25 %) and metastatic melanoma (43 %)

compared to normal (5 %) and dysplastic nevi (10 %).Patients with strong MMP2 had significantly poorer survivalcompared to those with negative-to-moderate MMP2 expres-sion. MMP2 expression could predict the patient survivalindependent of tumor thickness and ulceration. Furthermore,in our cohort study MMP2 expression was associated withp-Akt status and patient survival.Conclusions Strong MMP2 staining is associated withworse survival of melanoma patients and is an independentmolecular prognostic factor for primary melanoma.

Keywords MMP2 .Melanoma . p-Akt . Patient survival

1 Introduction

Melanoma, a type of cancer arising from the melanocytes inepidermis of skin, is notorious for its rapid progression,metastasis, and poor prognosis [1–3]. Although it accountsfor only 4 % of skin cancers, melanoma is responsible forover 80 % of deaths from skin cancer and only 14 % ofpatients with metastatic melanoma survive for 5 years [3].Due to its ability to metastasize rapidly, up to 20 % ofpatients develop metastatic tumors and metastatic melanomais highly resistant to conventional radiotherapy and chemo-therapy [4–7]. When diagnosed early, melanoma is curablewith surgical excision and the patients can be managed witha relapse free interval of up to 10 years [8, 9]. The knowl-edge of events that are known to be associated with mela-noma initiation and progression, could be critically useful inthe diagnosis, screening and subsequent management of thepatients [2, 10]. There has been considerable amount ofprogress in the understanding of melanoma biology andthere is now valuable information available on the factorsthat could predict the survival of the patients, like tumor

A. Rotte :G. LiDepartment of Dermatology and Skin Science, Jack Bell ResearchCentre, Vancouver Coastal Health Research Institute,University of British Columbia,Vancouver, BC, Canada

M. MartinkaDepartment of Pathology, Vancouver General Hospital,Vancouver, BC, Canada

G. Li (*)Jack Bell Research Centre,2660 Oak Street,Vancouver, BC, Canada V6H 3Z6e-mail: gangli@mail.ubc.ca

Cell Oncol. (2012) 35:207–216DOI 10.1007/s13402-012-0080-x

thickness and ulceration, mitotic rate and lymph node invasionstatus [9, 11, 12]. However, the knowledge of molecularmarkers that can aid in the identification of high risk patientsand serve as prognostic factors of melanoma still remainselusive. There is thus an urgent need to identify molecularprognostic markers that can independently predict the patientsurvival.

Increasing evidence suggest the role of matrix metallo-proteinase (MMP), a large family of secreted peptidases, intumor invasion and metastasis [13, 14]. At least 26 MMPshave been identified and are classified according to theirstructure and substrate specificity as interstitial collagenases,gelatinases, stromelysins and MT-MMPs [15]. MMP2 andMMP9 comprising the family of gelatinases which digestprimarily type IV collagen, are believed to play a major rolein melanoma progression [15–18]. Previous study by Vaisa-nen et al. has shown that MMP2, but not MMP9, is associatedwith the survival of melanoma patients [19]. However, thepatient number used in that study was relatively small and thesignificance of MMP2 expression in various stages of mela-noma progression has never been studied in detail. Moreover,no studies so far have analyzed if MMP2 expression is aprognostic factor independent of well known indicators formelanoma patient survival, such as tumor thickness and ul-ceration status. In order to convincingly translate the knowl-edge of MMP2 expression in melanoma into clinical practice,there is an utmost need to ascertain if MMP2 expression isparticularly relevant in a specific stage of melanoma and if itcan independently predict the patient survival. Therefore, weinvestigated MMP2 expression in melanocytic lesions at dif-ferent stages and analyzed the correlation of MMP2 expres-sion with various clinicopathologic parameters.

Akt/Protein kinase B, a key regulator of cell survivalpathways, is known to promote tumor cell invasion and me-tastasis [20, 21]. Phosphorylation of Akt was shown to besignificantly increased in metastatic melanoma biopsies com-pared to normal tissue and was shown to be associated withworse patient survival [22]. It was recently reported thatabsence of phosphotase and tensin homolog (PTEN) protein,a negative regulator of Akt phosphorylation, would lead toactivation of the transcription factor SP1, which in turn leadsto upregulation of MMP2 activity [23]. Thus, we furtherexamined the correlation between MMP2 and p-Akt expres-sions in melanoma biopsies and analyzed the combined effectofMMP2 and p-Akt expression in predicting patient outcome.

2 Patients and methods

2.1 Patient specimens and tissue microarray construction

The collection of patient specimens and the construction of thetissue microarray (TMA) have been previously described [22,

24]. Formalin-fixed, paraffin embedded tissues from 49 normalnevi, 100 dysplastic nevi, 403 primary melanomas, and 161metastatic melanomas were used for the present study. Allspecimens were obtained from the 1990 to 2009 archives ofthe Department of Pathology, Vancouver General Hospital. Theuse of human skin tissues was approved by the Clinical Re-search Ethics Board of the University of British Columbia andwas done in accordance with the Declaration of Helsinki guide-lines. The most representative tumor area was carefully selectedand marked on the hematoxylin and eosin stained slides and theTMAswere assembled using a tissue-array instrument (BeecherInstruments, Silver Spring, MD). Tissue cores of 0.6-mm thick-ness were taken in duplicate from each biopsy. Using a Leicamicrotome, multiple 4-μM sections were cut and transferred toadhesive-coated slides using regular histological procedures.One section from each TMAwas routinely stained with hema-toxylin and eosin whereas the remaining sections were stored atroom temperature for immunohistochemical staining.

2.2 Immunohistochemistry

TMA slides were dewaxed at 55°C for 20 min followed bythree 5-min washes with xylene. The tissues were then rehy-drated bywashing the slides for 5-min eachwith 100%, 95%,80 % ethanol and finally with distilled water. The slides werethen heated to 95°C for 30 min in 10 mmol/L sodium citrate(pH 6.0) for antigen retrieval and then treated with 3 %hydrogen peroxide for 1 hour to block the endogenous perox-idase activity. After blocking the slides with the universalblocking serum (Dako Diagnostics, Carpinteria, CA, USA),the sections were incubated overnight with monoclonal mouseanti-MMP2 antibody (1:50 dilution; Biolegend, San Diego,CA, USA) at 4°C. The sections were then incubated for30 min with a biotin-labeled secondary antibody and thenwith streptavidin-peroxidase (Dako Diagnostics). The sam-ples were developed by treatment with 3,3′-diamino-benzi-dine substrate (Vector Laboratories, Burlington, Ontario,Canada) and with hematoxylin to counter stain the nuclei.Negative controls were done by omitting the MMP2 antibodyduring the primary antibody incubation.

2.3 Evaluation of immunostaining

The evaluation of MMP2 staining was done blindly by mi-croscopic examination of the tissue sections by one dermato-pathologist and two other observers simultaneously, using amultiple viewing microscope. MMP2 staining intensity wasscored as 0+, 1+, 2+, 3+ whereas the percentage of MMP2positive cells was scored as 1 (1–25 %), 2 (26–50 %), 3 (51–75 %) and 4 (76–100 %). In the cases of discrepancy betweenduplicated cores the higher score from the two tissue coreswas taken as the final score. The product of intensity andpercentage was taken as the immunoreactive score (IRS)

208 A. Rotte et al.

[25]. Based on IRS, MMP2 staining in the tissue sections wascategorized as negative (IRS 0), weak (IRS 1–4), moderate(IRS 6–8), and strong (IRS 9–12).

The correlation between MMP2 and p-Akt expression wasstudied in 92 cases (54 primary and 38 metastatic melanoma)which were common in the present study and the previouslypublished study on prognostic significance of p-Akt expres-sion in melanoma [22]. Based on the IRS obtained as de-scribed previously [22], p-Akt staining in the tissue sectionswas categorized as negative (0), weak (1–2), moderate (3–4),strong (5–6). The cases were then grouped as, negative-to-moderate staining of both MMP2 and p-Akt, negative-to-moderate staining of MMP2 and strong staining of p-Akt,strong staining of MMP2 and negative-to-moderate stainingof p-Akt, strong staining of bothMMP2 and p-Akt. Out of thefour groups, group with strong staining of MMP2 andnegative-to-moderate staining of p-Akt had only 4 cases withone death event, so it was excluded from the survival analysis.

2.4 Statistical analysis

Correlation between MMP2 and clinicopathologic parameterswas evaluated byKruskal-Wallis test andχ2 test between patientsubgroups. Survival time was calculated from the date of mel-anoma diagnosis to the date of death or last follow-up. Theeffect of MMP2 expression on the overall and disease-specificsurvival was evaluated by Kaplan-Meier analysis and log-ranktest. Additionally, univariate and multivariate Cox proportionalhazards regression models were preformed to estimate the crudehazard ratios (HRs) or adjusted HRs and their 95 % confidentialintervals (CIs). p-value <0.05 was considered as statisticallysignificant. All the statistical analysis were performed usingSPSS version 11.5 (SPSS Inc, Chicago, IL) software.

3 Results

3.1 MMP2 expression positively correlated with melanomaprogression

Among the 713 patient samples collected, 11 normal nevi,28 dysplastic nevi, 73 primary melanomas and 9 metastaticmelanomas were excluded from the study due to loss ofbiopsy cores or insufficient cells present in the cores, and theremaining tissue cores were used for further tissue micro-array analysis. The study thus included 482 melanomacases, of which 281 were male and 201 were female(Table 1). As shown in Fig. 1A, MMP2 staining was re-markably stronger in the melanoma biopsies than nevi cases.Kruskal-Wallis test on the IRS scoring pattern in the patientsamples revealed that MMP2 expression increased signifi-cantly from normal nevi (median IRS score, 4) and dysplas-tic nevi (median IRS score, 4) to primary melanoma

(median IRS score, 6) and to metastatic melanoma (medianIRS score, 8; p<0.001; Fig. 1B). Furthermore, Chi-squaretest (χ2) revealed that the percentage of strong MMP2staining was significantly higher in primary melanoma(25 %) compared to normal nevi (5 %) and dysplastic nevi(10 %), and increased even further in metastatic melanoma(43 %; p<0.001; Fig. 1C). Accordingly, the percentage ofcases with strong MMP2 was significantly higher in AJCCstages III and IV (43 %) compared to stages I and II (25 %;p<0.001; χ2 test; Table 1). Moreover, strong MMP2 ex-pression was also significantly more frequent in patientswith tumor thickness greater than 2.0 mm compared withthose thinner than 2.0 mm (p<0.001; χ2 test; Table 1).However, in our patient cohort, MMP2 expression did notcorrelate with the ulceration status, melanoma subtype andtumor location (Table 1). Interestingly, strong MMP2 ex-pression was significantly more frequent in male patients ascompared to female patients (Table 1).

3.2 Increased MMP2 expression is associated with poorpatient survival

We analyzed if the increased MMP2 expression correlatedwith patient survival. Kaplan-Meier survival curve revealedthat patients with negative-to-moderate MMP2 expressionhad better 5-year survival compared to patients with strongMMP2 expression (Fig. 2A, B). The mean overall survival(47.8 months) in patients with negative-to-moderate MMP2expression was significantly better compared to patients withstrong MMP2 expression (38.2 months) (p01×10-5, log-ranktest). Similarly, mean disease-specific survival was also signif-icantly better in patients who had negative-to-moderate expres-sion of MMP2 compared with strong MMP2 staining group(48.5 months vs. 40.6 months) (p05×10-4, log-rank test). Toinvestigate if theMMP2 expression was associated with patientsurvival at specific stage of the disease, the patient cohort wasdivided into primary and metastatic melanoma accordingly andthe patient survival was analyzed. The primary melanomapatients with strongMMP2 expression had worse mean overall(45.3months) and disease-specific survival (48.3months) com-pared to patients with negative-to-moderate MMP2 expression(55.0 and 56.0 months, p01×10-5 and 2×10-4, respectively,log-rank test) (Fig. 2C, D). The results of Kaplan-Meier anal-ysis were further confirmed by Cox proportional hazards re-gression model (Table 2). However, MMP2 expression did notshow any influence on the survival rate of metastatic melanomapatients (Fig. 2E, F).

3.3 MMP2 expression is an independent prognostic markerfor primary melanoma

To study if the association of MMP2 expression with thepatient survival is dependent on other clinical characteristics

MMP2 Expression in Melanoma 209

of patients, multivariate Cox proportional hazard analysiswas performed. Initial analysis was done in all melanomacases, with age, gender, AJCC staging and MMP2 expres-sion as the variables. The results showed that MMP2

expression was significantly associated with overall survivalof melanoma patients. But the association with the disease-specific survival was not significant (Table 3). To see if asimilar association was seen with the primary melanoma

Table 1 MMP2 staining andclinicopathologic characteristicsof 482 melanoma patients

*Sun-protected sites: trunk, arm,leg and feet; Sun-exposed sites:head and neck†χ2 test

Variables MMP2 staining

Negative-moderate Strong Total P†

All melanoma (n0482)

Age (years)

≤60 161 (66.3 %) 82 (33.7 %) 243 (50.4 %) 0.2

>60 171 (71.5 %) 68 (28.5 %) 239 (49.6 %)

Sex

Male 182 (64.8 %) 99 (35.2 %) 281 (58.3 %) 0.02

Female 150 (74.6 %) 51 (25.4 %) 201 (41.7 %)

AJCC stage

I 166 (81.0 %) 39 (19 %) 205 (42.5 %) 4×10-6

II 80 (64.0 %) 45 (36.0 %) 125 (25.9 %)

III 31 (50.8 %) 30 (49.2 %) 61 (12.7 %)

IV 55 (60.4 %) 36 (39.6 %) 92 (18.9 %)

Primary melanoma (n0330)

Age (years)

≤60 113 (73.4 %) 41 (26.6 %) 154 (46.7 %) 0.6

>60 133 (75.6 %) 43 (24.4 %) 176 (53.3 %)

Sex

Male 128 (71.9 %) 50 (28.1 %) 178 (46.1 %) 0.2

Female 118 (77.6 %) 34 (22.4 %) 152 (53.9 %)

Tumor thickness (mm)

≤2.0 173 (81.2 %) 40 (18.8 %) 213 (64.5 %) 1×10-4

>2.0 73 (62.4 %) 44 (37.6 %) 117 (35.5 %)

Ulceration

Absent 212 (76.0 %) 67 (24.0 %) 239 (84.5 %) 0.2

Present 34 (66.7 %) 17 (33.3 %) 51 (15.5 %)

Site*

Sun-protected 250 (68.9 %) 113 (31.1 %) 363 (75.3 %) 1.0

Sun-exposed 82 (68.9 %) 37 (31.1 %) 119 (24.7 %)

Subtype

Acrolentigous 5 (55.6 %) 4 (44.4 %) 9 (2.7 %) 0.2

Lentigous 65 (83.3 %) 13 (16.7 %) 78 (23.6 %)

Nodular 36 (78.3 %) 10 (21.7 %) 43 (13.9 %)

Superficially spreading 87 (73.7 %) 31 (26.3 %) 118 (31.8 %)

Spindle cell type 8 (66.7 %) 4 (33.3 %) 12 (3.6 %)

Unspecified 45 (67.2 %) 22 (32.8 %) 67 (20.3 %)

Metastatic melanoma (n0152)

Age (years)

≤60 48 (53.9 %) 41 (46.1 %) 89 (58.6 %) 0.4

>60 38 (60.3 %) 25 (39.7 %) 63 (41.4 %)

Sex

Male 54 (52.4 %) 49 (47.6 %) 103 (67.8 %) 0.1

Female 32 (65.3 %) 17 (34.7 %) 49 (32.2 %)

210 A. Rotte et al.

cases, multivariate analysis was performed in these cases.Age, gender, thickness, ulceration and MMP2 expressionwere included as variables. As shown in Table 3, MMP2expression was significantly associated with both overalland disease-specific survival of the melanoma patients.Moreover, the hazard ratio was also comparable to that ofthickness and ulceration, indicating the prognostic signifi-cance of MMP2 expression in the primary melanoma. Sincethe Kaplan-Meier analysis did not show any association ofMMP2 expression in metastatic melanoma, multivariateanalysis was not performed in those cases.

3.4 MMP2 expression positively correlates with p-Aktexpression and influences patient survival

Phosphorylation of Akt has been shown to be associatedwith increased invasion and worse patient survival

[20–22]. Since the tissue microarray we used forMMP2 staining in this study contained the 92 caseswe previously stained for p-Akt, we were able to ana-lyze the correlation between MMP2 and p-Akt expres-sion. The results showed that strong MMP2 wassignificantly correlated with strong p-Akt expression(Fig. 3A). We then analyzed the effect of combinedMMP2 and p-Akt expression in patient survival. Thedata were divided into four groups, negative-to-moderatep-Akt and negative-to-moderate MMP2, negative-to-moderate p-Akt and strong MMP2, strong p-Akt andnegative-to-moderate MMP2, and strong p-Akt andstrong MMP2. However, there were only four cases inthe group with negative-to-moderate p-Akt and strongMMP2, so this group was excluded in the Kaplan-Meiersurvival analysis. Interestingly, among the patients with strongp-Akt expression, patients with negative-to-moderate MMP2

Fig. 1 Expression of MMP2 protein in melanoma. (A) Representativeimages at 100× (upper panel) and 400× (lower panel) magnification ofnormal nevi (NN) and dysplastic nevi (DN) with weak staining, pri-mary melanoma (PM) with moderate MMP2 staining and metastaticmelanoma (MM) with strong MMP2 staining. (B) MMP2 staining is

significantly increased from normal and dysplastic nevi to melanomaby Kruskal-Wallis test. (C) Percentage of strong MMP2 staining issignificantly increased from normal and dysplastic nevi to melanomaby χ2 test. **p<0.01, ***p<0.001

MMP2 Expression in Melanoma 211

expression had better mean overall (40.5 months) and disease-specific (41.7) survival compared to patients with strongMMP2 expression (32.2 and 33.4 months, respectively)(Fig. 3B, C).

4 Discussion

In line with previous studies, strong MMP2 expression issignificantly increased from normal and dysplastic nevi to

Fig. 2 Kaplan-Meier analysis of correlation between MMP2 expres-sion and 5-year survival. (A–B) Overall (A) and disease-specific 5-yearsurvival (B) of all melanoma patients (p<0.0001 and p<0.001 respec-tively, log-rank test). (C–D) Overall (C) and disease-specific 5-year

survival (D) of primary melanoma patients (p<0.0001 and p<0.001respectively, log-rank test). (E–F) Overall (E) and disease-specific 5-year survival (F) of metastatic melanoma patients. Labels at the top ofthe figure apply to all graphs in the same column

212 A. Rotte et al.

primary melanoma and to metastatic melanoma [18, 19, 26].The increased MMP2 expression could help the tumor cellsto invade into the neighbouring tissues and metastasize toother organs. Furthermore, simultaneous digestion of adja-cent tissue is critically important for the tumor growth as itprovides the space for the rapidly dividing cells. Increasedexpression of MMP2 could favor tumor invasion and thishypothesis is supported by our data showing significant

association of strong MMP2 expression and the tumorthickness.

In accordance with the previous studies [19, 26], ourKaplan-Meier and univariate Cox regression analysesshowed that patients with strong MMP2 expression had aworse survival. In addition to confirming the previous stud-ies, our present study goes further into identifying the sub-group of patients where MMP2 expression plays a vital role

Table 2 Univariate Cox Regression Analysis on 5-year Overall and Disease-specific Survival of Melanoma Patients

Variables* Overall survival Disease-specific survival

β† SE HR 95 % CI p β† SE HR 95 % CI p

All melanoma (n0482)

Age 0.063 0.149 1.07 0.79–1.43 0.7 −0.044 0.159 0.96 0.70–1.31 0.8

Sex −0.149 0.153 0.86 0.64–1.16 0.4 −0.114 0.162 0.89 0.65–1.23 0.5

AJCC 1.601 0.153 4.96 3.67–6.70 2×10-25 1.816 0.168 6.15 4.43–8.54 3×10-27

MMP2 0.687 0.151 1.99 1.48–2.61 5×10-6 0.606 0.161 1.83 1.34–2.51 2×10-4

Primary melanoma (n0330)

Age 0.786 0.255 2.20 1.33–3.62 0.002 0.692 0.286 2.00 1.14–3.50 0.02

Sex 0.109 0.267 1.12 0.66–1.88 0.7 0.026 0.234 1.03 0.65–1.63 0.9

Thickness 1.454 0.250 4.28 2.62–6.98 6×10-9 1.746 0.302 5.73 3.17–10.36 8×10-9

Ulceration 1.721 0.271 5.59 3.28–9.51 2×10-10 1.537 0.242 4.65 2.90–7.47 2×10-10

MMP2 1.093 0.236 2.98 1.88–4.73 3×10-6 1.069 0.269 2.91 1.72–4.93 7×10-5

*Coding of variables: Age was coded as 1 (≤60 years), and 2 (>60 years). Sex was coded as 1 (male) and 2 (female). AJCC staging was coded as 1(AJCC stage I and II), and 2 (AJCC stage III and IV). MMP2 was coded as 1 (negative to moderate staining) and 2 (strong staining). Thickness wascoded as 1 (≤2.0 mm) and 2 (>2.0 mm). Ulceration was coded as 1 (absent) and 2 (present)†β: regression coefficient

SE standard error of β, HR hazard ratio, CI confidence interval

Table 3 Multivariate cox regression analysis on 5-year overall and disease-specific survival of melanoma patients

Variables* Overall survival Disease-specific survival

β† SE HR 95 % CI p β† SE HR 95 % CI p

All melanoma cases (n0482)

Age 0.218 0.150 1.24 0.93–1.67 0.1 0.135 0.160 1.14 0.84–1.57 0.4

Sex 0.185 0.158 1.20 0.88–1.64 0.2 0.236 0.167 1.27 0.91–1.76 0.2

AJCC 1.569 0.161 4.80 3.50–6.59 2×10-22 1.821 0.175 6.18 4.38–8.71 3×10-25

MMP2 0.341 0.157 1.41 1.03–1.92 0.03 0.210 0.168 1.23 0.89–1.71 0.2

Primary melanoma cases (n0330)

Age 0.511 0.269 1.67 0.98–2.82 0.06 0.318 0.303 1.37 0.76–2.49 0.3

Sex 0.090 0.242 1.09 0.68–1.76 0.7 0.167 0.276 1.18 0.69–2.03 0.5

Ulceration 0.884 0.281 2.42 1.40–4.20 0.002 0.996 0.314 2.71 1.46–5.01 0.002

Thickness 0.920 0.283 2.51 1.44–4.37 0.001 1.220 0.339 3.39 1.74–6.58 3×10-4

MMP2 0.949 0.241 2.58 1.61–4.14 8×10-5 0.869 0.275 2.38 1.39–4.09 0.002

*Coding of variables: Age was coded as 1 (≤60 years), and 2 (>60 years). Sex was coded as 1 (male) and 2 (female). MMP2 expression was codedas 1 (neg-mod staining) and 2 (strong staining). Thickness was coded as 1 (≤2.00 mm) and 2 (>2.00 mm). Ulceration was coded as 1 (absent) and 2(present)†β: regression coefficient

SE standard error of β, HR hazard ratio, CI confidence interval

MMP2 Expression in Melanoma 213

in survival. Interestingly, strong MMP2 expression wasassociated with a worse survival in primary melanoma but

not in metastatic melanoma cases. At least in theory, it ispossible that primary melanoma with strong MMP2 expres-sion is more invasive and has high metastatic potentialallowing the primary tumor cells to metastasize rapidly andthus resulting in poor survival of the patient. The result pointsto the significance of MMP2 expression in the progression ofprimary melanoma to metastatic melanoma. Metastatic mela-noma patients, on the other hand, have very rapid onset oftumor related complications culminating in death within fewmonths. Apparently, at this stage MMP2 could not have anyfurther influence on the disease progression.

Multivariate Cox regression analysis of overall survivalshowed that MMP2 expression was independent of AJCCstaging of melanoma, age and gender, but when the data wasanalyzed for disease-specific survival, it was found to be adependent variable. Interestingly, multivariate Cox regres-sion analysis of both overall and disease-specific survival inprimary melanoma cases showed that MMP2 is a prognosticfactor independent of age, gender, tumor thickness andulceration status. MMP2 expression was associated withpatient survival only in primary (AJCC stages I and II)and not in metastatic melanoma (AJCC stages III and IV)cases and it could be the reason why multivariate analysis ofall melanoma patients showed disease-specific dependenceof MMP2 expression on AJCC staging. Importantly, MMP2expression had a comparable hazard ratio with that of tumorthickness and ulceration status. Our results thus suggestMMP2 as a very useful molecular marker which could havea high prognostic significance in primary melanoma.

MMP2 expression and activity can be regulated by var-ious signalling pathways and recently it was shown thatdecrease in the activity of phosphotase and tensin homolog(PTEN) correlated with increased MMP2 expression in livercarcinoma [23]. The report however did not correlate direct-ly the p-Akt status and MMP2 expression in the patientsamples. Moreover, the relationship between p-Akt andMMP2 has never been investigated in melanoma patients.The present study not only shows a significant correlationbetween strong p-Akt expression and strong MMP2 expres-sion, but also elucidates the importance of MMP2 expres-sion in the patients with strong p-Akt expression in patientsurvival. Phosphorylated Akt is known to promote cancerprogression and cause the death of the patient by accelerat-ing invasion and metastasis through upregulation of MMP2[20, 23]. Our results showed that the patients with strong p-Akt but negative-to-moderate MMP2 had a better survivalas compared to patients with strong p-Akt and strongMMP2, indicating that a powerful cell survival regulatorlike Akt would still need the downstream MMP2 activationin order to have an effect on the cancer progression. Al-though our present observation had a modest sample size of92 cases, the correlation between p-Akt and MMP2 could beimportant in the light of recent reports on the role of p-Akt

Fig. 3 Correlation of MMP2 and p-Akt expression in melanoma. (A)Positive correlation between p-Akt and MMP2 expression in 52 pri-mary melanoma and 38 metastatic melanoma cases. Melanomas whichhave strong p-Akt expression also have significantly higher percentageof strong MMP2 staining (p00.003, χ2 test). (B–C) Kaplan-Meieranalysis of the correlation between combined MMP2 and p-Akt ex-pression and overall (B) or disease-specific 5-year survival (C) ofmelanoma patients

214 A. Rotte et al.

in the resistance to Braf inhibitor, vemurafinib [27]. Wespeculate that MMP2 being a downstream target of p-Aktcould play a role in vemurafinib resistance and therefore,stronger MMP2 expression in vemurafinib-treated patientsmight provide an early indication of possible resistance tothe drug. However, a larger study on the MMP2 expressionstatus in the vemurafinib-treated patients is needed to con-firm the significance of MMP2 expression in the resistance.

The role of MMPs in melanoma invasion and subsequentmetastasis is frequently reported and several studies haveshown the association of MMP2 with the survival in melano-ma patients, but were unable to analyze the MMP2 expressionin different stages of melanoma possibly due to small size ofthe patient cohort [18, 19]. The objective of our study was notjust to repeat and reconfirm the earlier reports, but to explorethe possible differences between primary and metastatic mel-anoma patients with regards to the association with MMP2expression. To our knowledge, the present study is the first toshow that MMP2 expression is important only in primarymelanoma but not in metastatic melanoma. This finding isparticularly important in the light of failure of inhibitors ofMMP in Phase III studies [28]. Our results show that patientswith metastasis might not have any benefit by inhibition ofMMPs. Nevertheless, we successfully elucidated the prognos-tic significance of MMP2 expression in primary melanomapatients. The immunohistochemical staining of the tissuesamples is a robust but simple technique to perform and cangive the results relatively fast [29]. Therefore, our study couldform a practical approach to identify the high risk primarymelanoma patients who could have a worse prognosis.

In summary, MMP2 expression is significantly increasedin melanoma compared with nevi and it is associated withworse patient survival. MMP2 expression could predict thesurvival of primary melanoma patients independent of tumorthickness and ulceration, and it is associated with p-Aktexpression and patient survival. Taken together, MMP2 ex-pression might serve as a useful prognostic biomarker forprimary melanoma.

Acknowledgements This project is funded by grants from theCanadianInstitutes of Health Research (MOP-93810, MOP-110974 and CCI-117958) and Canadian Dermatology Foundation to G.L.

Conflict of interest The authors state no conflict of interest.

References

1. T.L. Diepgen, V. Mahler, The epidemiology of skin cancer. Br. J.Dermatol. 146(Suppl 61), 1–6 (2002)

2. A.C. Geller, S.M. Swetter, K. Brooks, M.F. Demierre, A.L. Yaroch,Screening, early detection, and trends for melanoma: current status(2000–2006) and future directions. J. Am. Acad. Dermatol. 57(4),555–572 (2007). quiz 573–556

3. A.J. Miller, M.C. Mihm Jr., Melanoma. New Engl. J. Med. 355(1),51–65 (2006)

4. H. Tsao, A.J. Sober, Melanoma treatment update. Dermatol. Clin.23(2), 323–333 (2005)

5. M.B. Lens, T.G. Eisen, Systemic chemotherapy in the treatment ofmalignant melanoma. Expert. Opin. Pharmacother. 4(12), 2205–2211 (2003)

6. C.M. Balch, S.J. Soong, J.E. Gershenwald, J.F. Thompson, D.S.Reintgen, N. Cascinelli et al., Prognostic factors analysis of 17,600melanoma patients: validation of the American Joint Committee onCancer melanoma staging system. J. Clin. Oncol. 19(16), 3622–3634 (2001)

7. A.N. Houghton, D. Polsky, Focus on melanoma. Canc. Cell. 2(4),275–278 (2002)

8. R.M. Turner, K.J. Bell, R.L. Morton, A. Hayen, A.B. Francken, K.Howard et al., Optimizing the frequency of follow-up visits forpatients treated for localized primary cutaneous melanoma. J. Clin.Oncol. 29(35), 4641–4646 (2011)

9. K.Y. Bilimoria, M.V. Raval, D.J. Bentrem, J.D. Wayne, C.M.Balch, C.Y. Ko, National assessment of melanoma care usingformally developed quality indicators. J. Clin. Oncol. 27(32),5445–5451 (2009)

10. A.C. Geller, S.M. Swetter, S. Oliveria, S. Dusza, A.C. Halpern,Reducing mortality in individuals at high risk for advanced mela-noma through education and screening. J. Am. Acad. Dermatol. 65(5), S87–94 (2011)

11. C.M. Balch, J.E. Gershenwald, S.J. Soong, J.F. Thompson, M.B.Atkins, D.R. Byrd et al., Final version of 2009 AJCC melanomastaging and classification. J. Clin. Oncol. 27(36), 6199–6206(2009)

12. J.F. Thompson, S.J. Soong, C.M. Balch, J.E. Gershenwald, S.Ding, D.G. Coit et al., Prognostic significance of mitotic rate inlocalized primary cutaneous melanoma: an analysis of patients inthe multi-institutional American Joint Committee on Cancer mel-anoma staging database. J. Clin. Oncol. 29(16), 2199–2205 (2011)

13. M. Bjorklund, E. Koivunen, Gelatinase-mediated migration and inva-sion of cancer cells. Biochim. Biophys. Acta. 1755(1), 37–69 (2005)

14. E.I. Deryugina, J.P. Quigley, Matrix metalloproteinases and tumormetastasis. Canc. Metastasis Rev. 25(1), 9–34 (2006)

15. C.E. Brinckerhoff, J.L. Rutter, U. Benbow, Interstitial collagenasesas markers of tumor progression. Clin. Canc. Res. 6(12), 4823–4830 (2000)

16. U.B. Hofmann, J.R. Westphal, G.N. Van Muijen, D.J. Ruiter,Matrix metalloproteinases in human melanoma. J. Invest. Derma-tol. 115(3), 337–344 (2000)

17. U.B. Hofmann, J.R. Westphal, A.J. Zendman, J.C. Becker, D.J.Ruiter, G.N. van Muijen, Expression and activation of matrixmetalloproteinase-2 (MMP-2) and its co-localization withmembrane-type 1 matrix metalloproteinase (MT1-MMP) correlatewith melanoma progression. J. Pathol. 191(3), 245–256 (2000)

18. T. Turpeenniemi-Hujanen, Gelatinases (MMP-2 and −9) and theirnatural inhibitors as prognostic indicators in solid cancers. Biochi-mie. 87(3–4), 287–297 (2005)

19. A.H. Vaisanen, M. Kallioinen, T. Turpeenniemi-Hujanen, Compar-ison of the prognostic value of matrix metalloproteinases 2 and 9 incutaneous melanoma. Hum. Pathol. 39(3), 377–385 (2008)

20. S. Shukla, G.T. Maclennan, D.J. Hartman, P. Fu, M.I. Resnick, S.Gupta, Activation of PI3K-Akt signaling pathway promotes pros-tate cancer cell invasion. Int. J. Canc. 121(7), 1424–1432 (2007)

21. V. Vasko, M. Saji, E. Hardy, M. Kruhlak, A. Larin, V. Savchenkoet al., Akt activation and localisation correlate with tumour inva-sion and oncogene expression in thyroid cancer. J. Med. Genet. 41(3), 161–170 (2004)

22. D.L. Dai, M. Martinka, G. Li, Prognostic significance of activatedAkt expression in melanoma: a clinicopathologic study of 292cases. J. Clin. Oncol. 23(7), 1473–1482 (2005)

MMP2 Expression in Melanoma 215

23. K.M. Sze, K.L. Wong, G.K. Chu, J.M. Lee, T.O. Yau, I. Oi-LinNg, Loss of phosphatase and tensin homolog enhances cell invasionand migration through AKT/Sp-1 transcription factor/matrix metal-loproteinase 2 activation in hepatocellular carcinoma and has clini-copathologic significance. Hepatology 53(5), 1558–1569 (2011)

24. H. Lin, R.P. Wong, M. Martinka, G. Li, Loss of SNF5 expressioncorrelates with poor patient survival in melanoma. Clin. Canc. Res.15(20), 6404–6411 (2009)

25. W. Remmele, H.E. Stegner, Recommendation for uniform defini-tion of an immunoreactive score (IRS) for immunohistochemicalestrogen receptor detection (ER-ICA) in breast cancer tissue. Path-ologe. 8(3), 138–140 (1987)

26. A. Vaisanen, P. Kuvaja, M. Kallioinen, T. Turpeenniemi-Hujanen,A prognostic index in skin melanoma through the combination of

matrix metalloproteinase-2, Ki67, and p53. Hum. Pathol. 42(8),1103–1111 (2011)

27. F. Su, W.D. Bradley, Q. Wang, H. Yang, L. Xu, B. Higgins et al.,Resistance to selective BRAF inhibition can be mediated by modestupstream pathway activation. Cancer Res. 72(4), 969–978 (2012)

28. J.A. Sparano, P. Bernardo, P. Stephenson, W.J. Gradishar, J.N.Ingle, S. Zucker et al., Randomized phase III trial of marimastatversus placebo in patients with metastatic breast cancer who haveresponding or stable disease after first-line chemotherapy: EasternCooperative Oncology Group trial E2196. J. Clin. Oncol. 22(23),4683–4690 (2004)

29. R.L. Camp, V. Neumeister, D.L. Rimm, A decade of tissue micro-arrays: progress in the discovery and validation of cancer bio-markers. J. Clin. Oncol. 26(34), 5630–5637 (2008)

216 A. Rotte et al.