Early Detection and Diagnosis
Serum matrix metalloproteinase-9 in head and neck squamous cell carcinomais a prognostic marker
Article first published online: 7 APR 2005
Copyright © 2005 Wiley-Liss, Inc.
International Journal of Cancer
Volume 116, Issue 3, pages 422–427, 1 September 2005
How to Cite
Ruokolainen, H., Pääkkö, P. and Turpeenniemi-Hujanen, T. (2005), Serum matrix metalloproteinase-9 in head and neck squamous cell carcinomais a prognostic marker. Int. J. Cancer, 116: 422–427. doi: 10.1002/ijc.21092
- Issue published online: 10 JUN 2005
- Article first published online: 7 APR 2005
- Manuscript Accepted: 21 DEC 2004
- Manuscript Received: 13 SEP 2004
- Cancer Society of Northern Finland
- Finnish Cancer Society
- head and neck squamous cell carcinoma;
- prognosis tumor marker;
- cause-specific survival;
- relapse-free survival
The aim of this study was to determine whether serum matrix metalloproteinase-9 (MMP-9) could predict cause-specific and relapse-free survival in patients with squamous cell carcinoma of head and neck. Furthermore, this study was designed to investigate whether there is an association between MMP-9 immunohistochemical staining and serum MMP-9 levels. Pretreatment serum levels of MMP-9 were quantitatively measured by ELISA assay in 67 patients presenting with a primary head and neck squamous cell carcinoma. The results were compared with the corresponding immunohistochemical staining results, clinical data and the patients' outcome. The follow-up time for all of the patients was at least 5 years. There was a statistically significant correlation between circulating MMP-9 and MMP-9 immunohistochemical staining in the corresponding tumors (p = 0.028). The cause-specific and relapse-free survival rates were clearly lower among patients with high MMP-9 serum levels (> 73 ng/ml). The 5-year cause-specific survival-rate was 40% in a patient group with high serum MMP-9, and 69% for patients with a low MMP-9 level (p = 0.027). In the same follow-up period, the cumulative relapse-free survival rate was 36% in patients presenting with a high serum MMP-9 and 66% in those with a low MMP-9 level. No correlation was found between MMP-9 serum levels and the traditional clinical or histopathologic factors. The results suggest for the first time that pretreatment serum MMP-9 level could serve as a prognostic factor in head and neck squamous cell carcinoma. © 2005 Wiley-Liss, Inc.
Over 90% of head and neck tumors are squamous cell carcinomas (HNSCCs). Squamous cell carcinoma of the head and neck is one of the most common cancers, constituting 4% of all newly diagnosed cancers in the United States and 5% in the United Kingdom. Worldwide, more than 500,000 new cases are diagnosed annually.1, 2 HNSCC tumors have the capacity to invade adjacent tissues and metastasize locoregionally. The major cause of death among these patients is a local-regional recurrence. Prognosis depends mainly on the stage of the tumor, with the most important factor being the presence of lymph node metastases.3 The prognosis may also vary greatly in each stage group, suggesting that some other, possibly biologic, factors may also play a role in the progression of HNSCC. Evaluation of the metastatic potential of the primary tumor could enable more correct prognosis assessment by providing information about the propensity of a tumor to disseminate. Useful markers associated with biologic aggressiveness could also be valuable in developing new treatment strategies in HNSCC. Therefore, especially prognostic serum biomarkers that are easily detectable from serum pre- or postoperatively could also be valuable in the follow-up of the patients.
Matrix metalloproteinases (MMPs) are a large group of zinc-dependent endopeptidases.4 Gelatinase A (MMP-2) and gelatinase B (MMP-9) are the largest members of this gene family. They are able to degrade connective tissue, among other substrates the basement membrane collagen, which appears to play a key role in tumor cell invasion and in the process of metastasis.4, 5, 6, 7, 8 Gelatinases have also been associated with tumor angiogenesis, especially gelatinase B.9, 10 Increased expression of MMP-2 and MMP-9 immunoreactive protein in various cancer tissues has been found to be associated with shortened survival and shortened relapse-free times in patients suffering from these cancers.11, 12, 13, 14 Recently, increased serum MMP-9 and MMP-2 levels have been reported to be practical in predicting disease prognosis in some types of tumors.15, 16, 17, 18
Several studies have shown that gelatinases are expressed in head and neck carcinoma cells, and gelatinases may take part in the progression and invasion of these tumors.19, 20, 21 We have recently found that MMP-9 positivity in immunohistochemical staining is associated with unfavorable prognosis in head and neck squamous cell carcinoma.22 Some studies have suggested an association between gelatinase mRNA, immunoreactive protein or enzyme activity with invasion or lymph node metastasis in some tumor types according to their malignancy category.23, 24, 25 However, the data relating to the significance of circulating MMP-9 in patients with HNSCC as a prognostic marker have been less studied. The present study was conducted to measure preoperative or pretreatment serum MMP-9 levels in patients with head and neck squamous cell carcinoma and to correlate them with clinical behavior and disease outcome as well as to study whether the circulating MMP-9 levels would correlate to MMP-9 immunohistochemical staining in the corresponding primary tumor or to the tumor burden.
Material and methods
The patients referred to Oulu University Hospital for treatment of a primary squamous cell head and neck carcinoma between the years 1994 and 1996 were recruited consecutively to the study and they were followed for a minimum of 5 years. There were no dropouts from the follow-up. Serum samples were taken during the first visit to the hospital and they were taken before the operation or radiotherapy. Serum samples were stored frozen in −70°C until used for the study. In our series, there were 67 cases where both paraffin-embedded tissue samples from the primary tumor and pretreatment serum samples were available. The study protocol was approved by the ethical committee of Oulu University in March 1994 and renewed by the ethical committee of Oulu University Hospital in July 2002. As control values, the serum sample were taken from 44 healthy volunteers (27 female, 17 male) without history of a known malignant disease or recent surgery.
The mean age of the patients was 64 years (minimum, 28 years; maximum, 88). The stage of the disease, tumor size and lymph node involvement were determined according to the International Union Against Cancer TNM classification.26 The histologic grade of the tumors was reviewed and classified according to the World Health Organization Classification of head and neck tumors.27
The treatment strategies for patients in this series were carried out according to the local protocol for treatment, and the treatment line depended on the stage of the tumor. Radical surgical operation with postoperative radiotherapy (50–60 Gy) was the treatment for 23 of the 67 patients; 18 patients were treated by a radical surgical operation without other treatments and 9 patients were inoperable and received only radiotherapy (50–64 Gy). Fifteen patients received preoperative radiotherapy (50 Gy) treatment before the surgical operation. No patients were treated with adjuvant chemotherapy. Two of the patients had an advanced carcinoma and received only palliative treatment.
ELISA quantification of MMP-9
ELISA was performed on 8-well EIA microtiter plates (Corning, New York, NY) using the standard protocol.16 When measuring the total MMP-9 (free and complexed forms of MMP-9), monoclonal anti-MMP-9 was used (code GE-213; Diabor, Oulu, Finland). The antibody recognizes both the free MMP-9 and that bound to its inhibitor.28 The antibody was coated onto microtiter plates before adding the samples and standards. The bound analytes were recognized with a polyclonal anti-MMP-9 antibody produced in chicken against MMP-9. A second antichicken antibody was added and OPD solution (P-1526; Sigma, Steinheim, Germany) was used to visualize the peroxidase label. The color formation was measured on 450 nm (Anthos Reader 2001) and calculations were done using a specific Windows-based control and evaluation software for Rosys Anthos Microplate readers (Anthos Labtec Instruments, Wals, Austria). Between each step of the procedure, the wells were thoroughly washed with 0.05% Tween-phosphate-buffered saline.
Immunohistochemical staining and evaluation of MMP-9 immunostaining
Immunohistochemical staining procedure was carried out using a standard method described previously.29 Paraffin-embedded sections were stained using the streptavidin-biotin-immunoperoxidase technique by using a mouse monoclonal antibody (code GE-231; Diabor) for MMP-9 as a primary antibody. The antibody recognizes the amino terminal end of the latent MMP-9 both as a free enzyme and when it is in a complex with the tissue inhibitor of metalloproteinase. The specificity has been shown by Nikkari et al.28 using Western blot analysis. Before incubation with the primary antibody, the paraffin sections were dewaxed, hydrated, pepsinized and treated with 3% hydrogen peroxidase to block the endogenous peroxidase activity. The nonspecific binding was blocked with goat serum.
The immunohistochemical staining was continued according to the manufacturer's protocol using Histostain-bulk kit (Zymed, San Francisco, CA), which uses a biotinylated antimouse IgG as a second antibody and a streptavidin conjugate as a linkage protein. The final reaction was visualized by using aminoethyl carbazol substrate kit (AEC-kit; Zymed), and the sections were counterstained with hematoxylin, dehydrated and mounted. Each set of staining always included a separate known positive control and a negative control, where the primary antibody for MMP-9 was replaced with mouse nonimmuno IgG.
The slides were analyzed by 2 independent observers blinded from clinical data and the immunoreactivity in the malignant cells in each section was graded from 0 to 3 according to the extent of positive staining. The case was considered positive when > 1% of the tumor cells showed a positive staining. Weak positivity was marked as + (1% < cells with a positive reaction < 25%) and moderate positivity was marked as ++ (25% < cells with a positive reaction < 50%). Cases showing a positive staining in > 50% of the tumor cells were considered extensively positive for MMP-9 (+++).
All statistical analyses were performed by using the SPSS software system for Windows (version 10.0; SPSS, Chicago, IL). The correlations of tumor stage, TNM classification, histologic grade and the primary anatomical site were analyzed separately with the Fisher's exact test to the MMP-9 immunoreactive protein level measured by ELISA. To compare the difference of the mean between groups, the Mann-Whitney U-test was used. The cause-specific and relapse-free survival rates were analyzed using the Kaplan-Meier method, and the statistical differences in survival among subgroups were compared by a log-rank test.30 The cause-specific survival was defined as the time from the date of diagnosis to the date of death due to head and neck carcinoma. The cases were censored on the date of the last control or at the time of death to another disease. Similarly, the relapse-free survival was calculated from the date of diagnosis to the date of relapse, and the case was censored at the time of the last follow-up visit. To assess the independent prognostic value of MMP-9 serum level, we used Cox proportional hazard regression analysis that included MMP-9 and stage or grade, which were associated with the survival rate.
MMP-9 serum levels
The pretreatment serum sample was available from 67 patients. The median serum level for MMP-9 immunoreactive protein was 89.9 ng/ml and it varied from below the detectable level to 325.9 ng/ml. The mean value was 115.2 ng/ml. In analyses, the high level was defined as values > 73.0 ng/ml. This cutoff value was found when the effect of different cutoff values to log-rank analysis was studied. This value was more powerful in differentiating the cases according to their prognosis than the median or the mean levels would have been. In contrast, the MMP-9 serum levels in the 44 healthy controls ranged from 13.5 to 153.6 ng/ml (median, 55.6 ng/ml; mean, 69.7 ng/ml). MMP-9 serum levels were significantly higher in the HNSCC patients than in healthy controls (p < 0.001, Mann-Whitney test).
Lack of association between MMP-9 serum levels and baseline clinical data
No correlation was found between serum MMP-9 and the stage of the disease (p = 0.35, Fisher's exact test; Table I). Thirty-six out of 49 (74%) patients presenting with an MMP-9-positive primary tumor with stage II–IV disease had a circulating MMP-9 level above 73 ng/ml, whereas 50% of the patients with stage I disease had a high level of circulating MMP-9. This difference was not statistically significant (p = 0.23, Fisher's exact test). In particular, we could not find any association between circulating MMP-9 and tumor size, cervical lymph node involvement, or anatomical region of the primary tumor (p = 0.43, 0.57 and 0.40, respectively, Fisher's exact test). Additionally, no correlation was seen between the MMP-9 serum levels and the grade of the tumor (p = 0.52, Fisher's exact test) or the patients' age or sex (p = 0.43 and 0.56, respectively, Fisher's exact test; Table I). Among the groups of patients with different treatment, no association was found with circulating MMP-9 levels (p = 0.50, Fisher's exact test).
|Patient characteristics||n||MMP-9 low,1n (%)||MMP-9 high,2n (%)||p|
|All patients||67||22 (33%)||45 (67%)|
|Male||45||15 (33%)||30 (67%)||0.56|
|Female||22||7 (32%)||15 (68%)|
|28–50||9||3 (33%)||6 (67%)||0.43|
|51–65||25||9 (36%)||16 (64%)|
|66–75||23||8 (35%)||15 (65%)|
|76–88||10||2 (20%)||8 (80%)|
|Oral cavity||31||11 (36%)||20 (64%)||0.40|
|Larynx||21||7 (33%)||14 (67%)|
|Pharynx||9||3 (33%)||6 (67%)|
|Others||6||1 (17%)||5 (83%)|
|Grade 1||14||5 (36%)||9 (64%)||0.52|
|Grade 2||42||13 (31%)||29 (69%)|
|Grade 3||11||4 (36%)||7 (64%)|
|T1–2||34||12 (35%)||22 (65%)||0.43|
|T3–4||33||10 (30%)||23 (70%)|
|N0||37||12 (32%)||25 (68%)||0.57|
|N+||30||10 (33%)||20 (67%)|
|I||8||3 (38%)||5 (62%)||0.35|
|II||14||3 (21%)||11 (79%)|
|III||21||10 (48%)||11 (52%)|
|IV||24||6 (25%)||18 (75%)|
Correlation between circulating MMP-9 and MMP-9 protein in primary tumors
There was a correlation between MMP-9 immunohistochemical staining and the serum level of the MMP-9 immunoreactive protein. In immunohistochemical staining, the MMP-9 immunoreactive protein was prominent in cancer cells and seemed to localize in the cell cytoplasm as a diffuse staining as shown previously22 (Fig. 1). In cases with extensive (+++) positive MMP-9 staining in the primary tumor, 22 out of 25 patients (89%) presented with a high level (> 73 ng/ml) of serum MMP-9 immunoreactive protein. In cases with tumor staining that was weakly or moderately positive for MMP-9, the number of cases with a high circulating level of MMP-9 was 17 out of 30 cases (57%), and 6 out of 12 patients (50%) presenting with a negative MMP-9 staining in the primary tumor had a high serum level for MMP-9. This correlation was statistically significant (p = 0.028, Fisher's exact test; Table II).
|Immunohistochemicalstaining result (MMP-9)||n||SerumMMP-9 low,1n (%)||SerumMMP-9 high,2n (%)|
|Negative||12||6 (50%)||6 (50%)|
|Weak (+)||21||9 (43%)||12 (57%)|
|Moderate (++)||9||4 (44%)||5 (56%)|
|Extensive (+++)||25||3 (12%)||22 (88%)|
Circulating MMP-9 as a prognostic factor
Patients with serum MMP-9 levels ≤ 73 ng/ml survived longer than those with MMP-9 levels > 73 ng/ml. The Kaplan-Meier analysis showed that the cause-specific cumulative survival of the patients with a low level of MMP-9 was 69% after 5 years of the follow-up, whereas the survival was only 40% among the patients presenting with a high serum MMP-9 level (p = 0.027, log-rank analysis; Fig. 2a). MMP-9 serum levels also associated with relapse-free survival, but this difference did not quite reach statistical significance in the log-rank analysis (p = 0.074). The cumulative 5-year relapse-free survival rate was 66% in patients presenting with a low serum level for MMP-9, whereas in patients with a high serum MMP-9 level the rate was only 36% (Fig. 2b).
During the follow-up time, 38 patients faced a relapse (Table III). In 29 out of the 38 patients with a relapse, the pretreatment serum MMP-9 immunoreactive protein level was high (p = 0.059, Fisher's exact test). Additional analyses showed that in cases with a later hematogenic relapse, all patients (100%) had a high pretreatment serum MMP-9 level. In cases with a lymph node relapse, 75% had a high level of serum MMP-9. In patients with a local relapse, the rate of high serum level was 67% (Table III).
|Type of the relapse||n||MMP-9 low, n (%)||MMP-9 high, n (%)|
|No relapse||29||13 (45%)||16 (55%)|
|Relapse||38||9 (32%)||29 (68%)1|
|Local relapse||9||3 (33%)||6 (67%)|
|Lymph node||24||6 (25%)||18 (75%)|
|Hematogenous||5||0 (0%)||5 (100%)2|
The value of prognostic variables was also studied by using the Cox regression analysis. In this material, the most significant prognostic factor for survival was the stage of the disease. The patients with a stage IV disease had a 7.2-fold relative risk for dying compared with the patients with a stage I disease (p = 0.011). In the same analysis, high serum MMP-9 level also had a prognostic effect. Patients with a high serum value had a 2.0-fold relative risk of dying from HNSCC when compared to the patients presenting with a low serum MMP-9 (p = 0.15). We also tested the dependence between the serum MMP-9 level and the grade of the tumor. In that analysis, MMP-9 had a role as a marker predicting shortened survival, but the grade of the tumor did not have a prognostic value in this analysis. Patients with an MMP-9 serum level above 73 ng/ml had 2.6-fold relative risk of dying compared with patients with a value below 73 ng/ml (p = 0.034). Because of the limited number of patients, it was not possible to include MMP-9, stage and grade as variables in the same analysis model simultaneously.
In this study, we demonstrated that high preoperative serum MMP-9 level could be a prognostic factor for shortened survival for head and neck squamous cell carcinoma patients. To our knowledge, this is the first study that has shown such a correlation. We found that the cause-specific survival was found to be strongly dependent on the circulating pretreatment level of MMP-9 immunoreactive protein. Only 40% of the patients with a high serum level of MMP-9 (> 73 ng/ml) survived for 5 years after the diagnosis of HNSCC, whereas in patients with a low MMP-9 level (≤ 73 ng/ml) a 69% cause-specific 5-year survival rate was observed. This difference would be clinically noteworthy and it was significant in the log-rank analysis (p = 0.027). There was also a notable difference in the 5-year relapse-free survival between the patient groups presenting with high and low serum MMP-9 levels (36% and 66%, respectively), although this difference did not quite reach statistical significance probably due to the limited number of patients.
Traditionally, the MMPs have been associated with tumor cell growth and migration.4, 5, 6, 7, 8 Some immunohistochemical studies have shown that MMP-9 could have prognostic value in predicting long-term outcome in different types of tumors.14, 22, 31, 32 In our previous study, we reported that overexpression of MMP-9 immunoreactive protein in tissue section from primary tumors of head and neck carcinoma correlated with cause-specific and relapse-free survival.22 There are, however, few previous publications report that serum MMP-9 would have any prognostic role in cancer patients.15, 16, 17, 18
The prognosis of squamous cell carcinoma of head and neck is traditionally linked to the TNM stage, differentiation (grade) and performance status. However, these factors are not sufficient enough to predict outcome accurately. Therefore, there is an obvious need for new markers that could help assess patients in more consequential prognostic groups and identify high-risk patients who might benefit from different treatment modalities and follow-up programs or serve as targets for novel therapy models. A few circulating tumor markers such as endostatin, angiogenin, Tie-2 receptor and anti-p53-antibody have been investigated as prognostic indicators among HNSCC patients.33, 34, 35 None of them, however, has been accepted as a routine clinical marker.
In this work, we did not find any correlation between the MMP-9 serum immunoreactive protein level and the traditional histopathologic or clinical prognostic variables. These results are in line with previous recent reports that show no correlation between circulating MMP-9 serum protein and the traditional clinicopathologic prognostic factors among HNSCC patients.36, 37 On the other hand, Riedel et al.38 have reported that serum concentrations of MMP-9 are significantly higher in patients with an advanced stage of HNSCC.
In the present study, we show also for the first time that there is a correlation between the serum level of MMP-9 immunoreactive protein and MMP-9 immunohistochemical staining in primary tumors of HNSCC. Most (89%) of the cases presenting with an extensive (+++) positive immunostaining had a high level of circulating MMP-9 immunoreactive protein (p = 0.028). Iizasa et al.39 have investigated a possible correlation between plasma MMP-9 and tissue immunohistochemical staining for MMP-9 in non small cell lung cancer patients. Plasma concentrations of MMP-9 were significantly elevated in NSCLC patients compared to healthy controls. The positive immunostaining for MMP-9 did not, however, show any significant association with plasma concentrations of MMP-9. Tutton et al.40 have also reported that MMP-9 expression is significantly increased in colorectal cancer tissues, predominantly in the stroma of the tumor, and that plasma levels of MMP-9 are significantly elevated at all stages in colorectal cancer patients, but any correlation between tissue and circulating levels of MMP-9 was not reported. A statistically significant difference was found in this study between the mean values of serum MMP-9 immunoreactive protein in HNSCC patients and healthy controls. This is in line with previous reports.41, 42 The question still remains as to where the circulating MMP-9 originated. Our results support the theory that elevated serum MMP-9 level could at least in part be linked to the overproduction of MMP-9 by the HNSCC cells. The influence of inflammation to the MMP-9 serum levels is somewhat unclear. Kuropkat et al.42 have previously found that there were no significant correlation between serum MMP-9 level and the additional inflammatory disease status or C-reactive protein level in HNSCC patients.
In the present study, the stage of the disease was the most powerful factor in predicting survival according to the Cox regression analysis. We found that serum MMP-9 might also still have a prognostic role independent of the stage, because the risk of dying is increased 2-fold even if the stage has been included in the analysis. Due to the strength of stage as a prognostic variable and the limited number of cases in our analysis, the p-value did, however, remain nonsignificant. On the other hand, when MMP-9 and grade of the tumor were included simultaneously in the Cox analysis, MMP-9, but not tumor grade, retained its strong prognostic role. In this analysis, the patients with high serum MMP-9 had a 2.6-fold relative risk of dying when compared to the patients with a low serum MMP-9. In our previous study, we have shown that positive tissue MMP-9 immunostaining had an 11-fold relative risk of dying when compared to patients with MMP-9-negative immunostaining.22
In conclusion, we report here that high serum level of MMP-9 immunoreactive protein is associated with shortened cause-specific and relapse-free survival in HNSCC. To the best of our knowledge, this is the first study identifying circulating MMP-9 as a prognostic marker in HNSCC. Based on this result, the patient group with high pretreatment serum level of MMP-9 has an about 2-fold risk of dying when compared to patients with a low serum level of MMP-9. MMP-9 did not correlate to the tumor burden (TNM and/or stage), but it did associate to tissue MMP-9 in the corresponding primary tumor. Further studies will show whether circulating MMP-9 could serve as a marker for recurrence in follow-up.
The authors thank Mrs. Kaisu Järvenpää and Anne Bisi for skillful technical assistance, Mr. Risto Bloigu for statistical consulting and the staff of the Otorhinolaryngology Clinic at Oulu University Hospital for help in collecting the serum samples.
- 26International Union Against Cancer. TNM classification of malignant tumors. New York: Wiley-Liss, 1997. 7–49.
- 27Histological typing of tumours of the upper respiratory tract and ear, 2nd ed. Berlin: Springer-Verlag, 1991. 7–49., in collaboration with and pathologists in 8 countries.