- Email: [email protected]
Serum TIMP-1 Predicts Survival Outcomes of Invasive Breast Carcinoma Patients: A Meta-analysis
Archives of Medical Research 42 (2011) 463e468
ORIGINAL ARTICLE
Serum TIMP-1 Predicts Survival Outcomes of Invasive Breast Carcinoma Patients: A Meta-analysis Ju-Han Lee,* Jung-Woo Choi,* and Young-Sik Kim Department of Pathology, Korea University Ansan Hospital, Ansan, Korea Received for publication March 4, 2011; accepted September 6, 2011 (ARCMED-D-11-00113).
Background and Aims. Tissue inhibitor of metalloproteinase-1 (TIMP-1) is a small secretory glycoprotein with multifunctional activity including anti-apoptosis and the inhibition of matrix metalloproteinase in invasive breast carcinomas. There have been contradictory results as to whether TIMP-1 is a poor or good prognostic factor in breast cancer patients. To address this controversy, we conducted a meta-analysis for the relationship between TIMP-1 levels and prognostic parameters in the breast cancer. Methods. The relevant published studies were pooled according to the defined selection criteria. The effect sizes of overall survival and prognostic parameters were calculated by a hazard ratio (HR) or an odds ratio (OR). HRs or ORs were combined using a randomeffects model. Results. Survival outcomes between high or elevated and low or normal serum TIMP-1 levels were compared by uni- and multivariate analyses involving 886 and 844 breast cancer patients, respectively. Patients with high or elevated serum TIMP-1 levels had unfavorable survival outcomes compared to patients with low or normal serum TIMP-1 levels in the uniand multivariate analyses (HR, 1.7 and 2.4; p !0.001 and p 5 0.033, respectively). However, no survival difference was evident in the data from tissue TIMP-1 levels by enzyme-linked immunosorbent assay and the expression of tissue TIMP-1 mRNA. The high or positive immunohistochemical expression of tissue TIMP-1 protein was not related to adjusted and unadjusted HRs, lymph node metastasis, and clinical stages. Conclusions. This meta-analysis indicates that serum TIMP-1 levels may be useful for predicting survival outcomes of invasive breast cancer patients. Ó 2011 IMSS. Published by Elsevier Inc. Key Words: TIMP-1, Breast cancer, Survival, Meta-analysis.
Introduction Breast cancer is the most common type of cancer among women in developed countries. An estimated 194,280 new breast cancer patients were diagnosed and 40,610 patients died in the United States in 2009 (1). Currently, the prognosis of breast cancer patients is determined by a number of clinical and biological parameters such as TNM stage, histological grade, steroid hormone receptor
*
These authors contributed equally to this work. Address reprint requests to: Young-Sik Kim, MD, PhD, Department of Pathology, Korea University Ansan Hospital, 516, Gojan-1 Dong, Danwon-Gu, Ansan-Si, Gyeonggi-Do, 425-707, Republic of Korea; Phone: 82-31-412-5322; FAX: 82-31-412-5324; E-mail: [email protected]
status, and HER-2/neu gene expression. However, for risk stratification, these factors are relatively inadequate or insufficient to precisely evaluate the prognosis of an individual breast cancer patient, leading to unnecessary adjuvant chemotherapy or unexpected cancer recurrences (2). Among the promising new biomarkers of breast cancers, tissue inhibitor of metalloproteinase-1 (TIMP-1) is a cellular survival factor with multiple functions that involve antiapoptosis, cell growth and differentiation, as well as matrix metalloproteinase (MMP) inhibition (3e9). Recent studies have shown that high levels of tissue or serum TIMP-1 are associated with worse outcomes of breast cancer patients (10e25). However, the prognostic value of TIMP-1 in breast cancer has been challenged by conflicting results according to the analytical method or sample type used. In an attempt
0188-4409/$ - see front matter. Copyright Ó 2011 IMSS. Published by Elsevier Inc. doi: 10.1016/j.arcmed.2011.09.006
464
Lee et al./ Archives of Medical Research 42 (2011) 463e468
to address these controversies, we performed a meta-analysis to clarify the relationship between TIMP-1 and the prognosis of breast cancer, especially focusing on the analytical method and clinical material chosen. Materials and Methods Data Collection and Selection Criteria for Meta-analysis We extensively searched the following online databases using the keywords ‘‘TIMP and breast cancer’’; 1) Medline using PubMed (http://www.ncbi.nlm.nih.gov/pubmed), 2) ISI Science Citation Index using the ISI Web of Science search interface (http://apps.isiknowledge.com), and 3) Conference Papers Index using the CSA Illumina search interface (http://www.csa.com/csaillumina). We also manually searched the reference lists of the found articles. Overlapping articles or duplicate data were excluded by examining the authors’ names and affiliations for each publication. The selection process of the papers is shown in Figure 1. The following types of articles were included: 1) original articles demonstrating that TIMP-1 was assessed only in primary breast cancer tissue or patients; articles that dealt with cell lines or animals were excluded; 2) articles published before November 2010 in English; 3) the most informative article when multiple articles were published by the same authors or groups. The following articles were excluded: 1) articles containing data inappropriate for meta-analysis; 2) review articles without original data; and 3) case reports. Data Pooling and Statistics Hazard ratios (HRs) or odds ratios (ORs) with 95% confidence intervals (CIs) were calculated by means of metaanalysis (26e29). For studies without HRs of survival, 370 papers were retrieved for key words and their abstracts were reviewed
288 articles were excluded, due to: (1) 22 papers were review or comment (2) 262 papers were obviously irrelevant or not associated with TIMP in primary breast cancer tissue or patients (3) 4 papers were not published in English
82 full texts were reviewed
66 articles were excluded, due to: (1) 46 papers did not provide sufficient data (2) 20 papers were published by the same authors or groups 16 studies were included in this meta-analysis
Figure 1. Paper selection flowchart for meta-analysis.
we assessed HRs and CIs using a published approximation method (27). HRs or ORs were combined using a randomeffects model (DerSimonian-Laird method). Q statistics were used to test interstudy heterogeneity, which is an adaptation of the c2 goodness-of-fit test. It was considered statistically significant with p !0.10. Sensitivity analysis was performed to examine the influence of each study on the pooled OR by serially omitting an individual study and pooling the remaining studies. Possible publication bias was investigated by visual assessment of a funnel plot and Egger’s test for the degree of asymmetry ( p !0.05 was considered a potential publication bias). Statistical analysis was performed using Comprehensive Meta-analysis Software v.2.0 (Biostat, Englewood, NJ); p value !0.05 was considered statistically significant. Results Serum TIMP-1 Levels by Enzyme-linked Immunosorbent Assay (ELISA) Four studies reported serum TIMP-1 levels on the survival outcomes of breast cancer patients (10e13). These studies, which included 310 breast cancer patients with high or elevated serum TIMP-1 levels and 576 with low or normal serum TIMP-1, presented unadjusted HRs and CIs on overall survival. The estimated unadjusted HRs ranged from 1.37e3.04. High or elevated serum TIMP-1 levels were significantly associated with poor overall survival rates compared with low or normal serum TIMP-1. The pooled HR for these studies was 1.662 (95% CI: 1.350e2.046; p !0.001) (Figure 2). No statistical heterogeneity was found among the studies (Q 5 1.208, df 5 3, p 5 0.751). Three studies addressed multivariate adjusted HRs and CIs on overall survival. Breast cancer patients consisted of 282 patients with high or elevated serum TIMP-1 levels and 562 patients with low or normal serum TIMP-1 (10e12). The prognostic variables used in the multivariate survival studies were patient age, histological grade, tumor size, lymph node metastasis, TNM stage, menopausal status, hormonal receptor status, c-erbB2, and MMP expression (Table 1). In our analysis, the estimated adjusted HRs ranged from 1.44e4.76. Pooled HRs for these studies was 2.371 (95% CI: 1.073e5.237; p 5 0.033) (Figure 3). There was statistical heterogeneity among the studies (Q 5 5.269, df 5 2, p 5 0.072). Tissue TIMP-1 Levels by ELISA Two studies including 1783 breast cancer patients with high tissue TIMP-1 levels and 1338 with low tissue TIMP-1 presented unadjusted HRs and CIs on overall survival (14,15). There was statistical heterogeneity among the studies (Q 5 11.052, df 5 1, p 5 0.001). There was no association between tissue TIMP-1 levels and overall survival rates (HR 5 2.656, 95% CI: 0.595e11.856; p 5 0.201).
TIMP-1 in Breast Cancer
Study name
Statistics for each study
465
Hazard ratio and 95% CI
Hazard Lower Upper ratio limit limit Z-Value p-Value Talvensaari-Mattila A Wurtz SO Lipton A Wu ZS
3.037 1.404 1.699 1.372 1.662
0.737 12.514 0.804 2.452 1.347 2.144 0.475 3.964 1.350 2.046
1.538 1.194 4.469 0.585 4.785
0.124 0.232 0.000 0.559 0.000 0.01
0.1
low TIMP-1
1
10
100
high TIMP-1
Figure 2. Hazard ratios with corresponding 95% CIs of individual studies and pooled data for overall survival between high or elevated serum TIMP-1 levels and normal or low serum TIMP-1 in univariate analysis.
Expression of Tissue TIMP-1 mRNA Two studies that included 1418 breast cancer patients described multivariate adjusted HRs and CIs on overall survival (16,17). Statistical heterogeneity was found between studies (Q 5 4.063, df 5 1, p 5 0.044). There was no association between tissue mRNA expression of TIMP-1 and overall survival rates (HR 5 1.171, 95% CI: 0.764e1.795; p 5 0.468). Expression of Tissue TIMP-1 Protein in Tumor Cells Articles that dealt only with TIMP-1 expression in cancer cells were included because TIMP-1 is expressed in stromal cells as well as cancer cells. Five studies using immunohistochemistry presented unadjusted HRs and CIs of TIMP-1 expression in breast cancer cells on overall survival. These studies included 810 breast cancer patients with high or positive TIMP-1 protein expression and 434 patients with low or negative TIMP-1 protein expression (18e22). There was statistical heterogeneity among the studies (Q 5 14.244, df 5 4, p 5 0.007). The estimated unadjusted HRs ranged from 0.4e2.0. The high or positive TIMP-1 expression was
not associated with overall survival rates (HR 5 1.104, 95% CI: 0.789e1.543; p 5 0.564). Four studies presented multivariate adjusted HRs and CIs on overall survival, which included 715 breast cancer patients with high or positive TIMP-1 expression in tumor cells and 375 patients with low or negative TIMP-1 expression (18e21). Statistical heterogeneity was found among the studies (Q 5 7.125, df 5 3, p 5 0.068). The high or positive TIMP-1 expression was not related to overall survival rates (HR 5 1.282, 95% CI: 0.970e1.693; p 5 0.081). Six studies investigated the relationship of TIMP-1 expression and lymph node metastasis (13,18,22e25). The high or positive expression of TIMP-1 protein was observed in 443 (70%) of 637 cases with lymph node metastasis and in 292 (65%) of 446 cases without lymph node metastasis. Statistical heterogeneity was detected among the studies (Q 5 20.937, df 5 5, p 5 0.001). The high or positive expression of TIMP-1 protein was not associated with lymph node metastasis (HR 5 0.952, 95% CI: 0.481e1.882; p 5 0.887). Four studies showed the relationship of TIMP-1 expression and clinical stage (13,21,22,25). The high or positive
Table 1. Covariates used in the multivariate survival models Study Serum TIMP-1 Talvensaari-Mattila et al. 2005 W€ urtz et al. 2008 Lipton et al. 2008 Tissue TIMP-1 mRNA Sieuwerts et al. 2007 Nakopoulou et al. 2002 Tissue TIMP-1 protein Willemoe et al. 2009 Zhang et al. 2008 Vizoso et al. 2007 Kuvaja et al. 2005
Variables used in Cox proportional hazard analysis
TIMP-1,* T, age TIMP-1,* size, grade, age, menopausal status, hormonal receptor status TIMP-1,* serum c-erbB2,* soft tissue metastasis,* liver metastasis,* KPS* TIMP-1, size,* N,* grade,* age,* menopausal status, ER,* PR* TIMP-1,* size, histological type, N, grade, stage, c-erbB2, Bcl-2, ER,* PR TIMP-1, size,a histological type,a N,a grade,a ER,a chemotherapy regimen* TIMP-1,* size,* N,* grade,* c-erbB2,* MMP-13* TIMP-1,* TIMP-2,* stage,* grade, age, ER, PR,* MMP-9,* MMP-11* TIMP-1,* size,* stage,* grade, menopausal status
T, tumor size; KPS, Karnofsky performance score; N, regional lymph node involvement; ER, estrogen receptor status; PR, progesterone receptor status. a Unmentioned statistical value. *p !0.05. Independent prognostic variable.
466
Lee et al./ Archives of Medical Research 42 (2011) 463e468
Study name
Statistics for each study
Hazard ratio and 95% CI
Hazard Lower Upper ratio limit limit Z-Value p-Value Talvensaari-Mattila A 3.400 1.197 9.661 Wurtz So 4.757 1.287 17.581 Lipton A 1.440 1.104 1.878
2.297 2.338 2.694 2.371 1.073 5.237 2.135
0.022 0.019 0.007 0.033 0.01
0.1
1
10
100
low TIMP-1 high TIMP-1 Figure 3. Hazard ratios and pooled data for overall survival between high or elevated serum TIMP-1 levels and normal or low serum TIMP-1 in multivariate analysis.
expression of TIMP-1 protein was found in 45 (63%) of 71 cases with stage III or IV and in 225 (66%) of 342 cases with stage I or II. The high or positive expression of TIMP-1 protein was not associated with clinical stage (HR 5 0.897, 95% CI: 0.357e2.252; p 5 0.816). There was no statistical heterogeneity among the studies (Q 5 5.842, df 5 3, p 5 0.120). Sensitivity Analysis and Publication Bias Sensitivity analyses revealed that Lipton’s study (12) slightly influenced the adjusted and unadjusted HRs (Figure 4). None of the other studies significantly affected the pooled HRs or ORs with CIs. In the funnel plots and the Egger’s regression tests, there was no evidence of publication bias for unadjusted survival analysis according to the serum TIMP-1 levels (Figure 5). In addition, publication bias was not observed in the analyses of survival curves and lymph node metastasis according to TIMP-1 protein expression. However, evidence of publication bias was suggested for adjusted survival analysis according to the serum TIMP-1 levels and for investigation of clinical stage according to TIMP-1 protein expression.
Discussion This pooled analysis using both uni- and multivariate survival data from 886 and 844 breast cancer patients, Study name
respectively, found that high levels of serum TIMP-1 are significantly associated with poor survival rates of breast cancer patients. TIMP-1 is a small secretory glycoprotein with a molecular size of 28.5 kDa that inhibits the proteolytic activity of extracellular matrix by MMPs (30,31). Given that MMPs play a central role in cancer cell invasion and dissemination, one would expect high levels of TIMP-1 to be related with a favorable cancer prognosis. Nakopoulou et al. (22) and Kuskunovic et al. (23) reported that overexpression of TIMP-1 protein was closely associated with favorable prognosis of breast cancer patients. However, this concept has been challenged by the evidence that TIMP-1 inhibits apoptosis of cancer cells (5,7e9). In fact, some investigations have shown that overexpression of TIMP-1 at mRNA and protein levels are observed in different carcinomas, including breast carcinoma. In addition, overexpression of TIMP-1 is highly associated with poor prognosis of patients with these carcinomas (10,15,17,19e21,32,33). The mechanism by which TIMP-1 suppresses apoptosis of breast cancer cells has been proposed (34,35). TIMP-1 binds to CD63 at the cell surface, leading to activation of the CD63/integrin b1 complex and its downstream signaling molecules including focal adhesion kinase, phosphoinositide 3-kinase, Akt, Bad and Bcl-2/ Bcl-XL. This meta-analysis reveals that serum TIMP-1 levels by ELISA have a clinical implication, which is significantly associated with a poor survival in breast cancer patients.
Statistics with study removed
Hazard ratio (95% CI) with given named study removed
Lower Upper Point limit limit Z-Value p-Value Talvensaari-Mattila A 1.640 Wurtz SO 1.708 Lipton A 1.520 Wu ZS 1.675
1.329 1.365 0.954 1.355 1.662 1.350
2.024 2.137 2.422 2.071 2.046
4.609 4.677 1.761 4.763 4.785
0.000 0.000 0.078 0.000 0.000 0.01
0.1
low TIMP-1
1
10
100
high TIMP-1
Figure 4. Sensitivity analysis of meta-analysis for unadjusted HR of overall survival according to serum TIMP-1 levels.
TIMP-1 in Breast Cancer
467
Conflict of Interest
Funnel Plot of Standard Error by log hazard ratio 0.0
The authors declare no conflict of interest.
Standard Error
0.2
Acknowledgment This study was supported by a Korea University Grant. 0.4
References
0.6
0.8 -2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
log hazard ratio
Figure 5. Funnel plot of meta-analysis for unadjusted HR of overall survival according to serum TIMP-1 levels.
The HR of high serum TIMP-1 levels by ELISA was approximately 2-fold. In contrast, patients with high levels of tissue TIMP-1 protein or mRNA failed to demonstrate any survival differences from patients with low levels of tissue TIMP-1. This suggests that serum TIMP-1 is more effective and clinically useful to predict the prognosis of breast cancer patients than tissue levels of TIMP-1 expression. It remains unclear that immunohistochemical scoring for TIMP-1 is an indicator for a poor prognosis of breast cancer patients. In this study, high or positive expression of TIMP1 protein was not associated with adjusted and unadjusted HRs, lymph node metastasis, and clinical stages. The immunohistochemical results of TIMP-1 have been a matter of debate because of different antibodies used, variations in tissue fixation time, no or less optimal epitope retrieval, and discordant data interpretation (36). Therefore, immunohistochemical quantification of tissue TIMP-1 expression as prognostic stratification in breast cancer needs to be further validated. There are some limitations to our study. First, studies used different analytical methods and clinical sample types. Serum TIMP-1 levels could be compared with overall survival, whereas serum TIMP-1 levels could not be compared with several parameters such as clinical stage, lymph node metastasis, and tumor size due to insufficient data for meta-analysis. Second, we calculated the loghazard ratio and its variance from Kaplan-Meier survival curves using the indirect method (27). However, the indirect approach for extracting summary statistics is sufficiently correct to make a useful graphic summary of two survival curves (27). In summary, our study indicates that serum measurement of TIMP-1 levels by ELISA from patients with invasive breast cancer may be useful in predicting their survival outcomes although additional large-scale investigations are needed.
1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CA Cancer J Clin 2009;59:225e249. 2. Thomssen C, Janicke F. Do we need better prognostic factors in nodenegative breast cancer? Pro Eur J Cancer 2000;36:293e298. 3. Hayakawa T, Yamashita K, Tanzawa K, et al. Growth-promoting activity of tissue inhibitor of metalloproteinases-1 (TIMP-1) for a wide range of cells. A possible new growth factor in serum. FEBS Lett 1992;298:29e32. 4. Khokha R. Suppression of the tumorigenic and metastatic abilities of murine B16-F10 melanoma cells in vivo by the overexpression of the tissue inhibitor of the metalloproteinases-1. J Natl Cancer Inst 1994; 86:299e304. 5. Guedez L, Courtemanch L, Stetler-Stevenson M. Tissue inhibitor of metalloproteinase (TIMP)-1 induces differentiation and an antiapoptotic phenotype in germinal center B cells. Blood 1998;92: 1342e1349. 6. Kim YS, Seo DW, Kong SK, et al. TIMP1 induces CD44 expression and the activation and nuclear translocation of SHP1 during the late centrocyte/post-germinal center B cell differentiation. Cancer Lett 2008;269:37e45. 7. Lee SJ, Yoo HJ, Bae YS, et al. TIMP-1 inhibits apoptosis in breast carcinoma cells via a pathway involving pertussis toxin-sensitive G protein and c-Src. Biochem Biophys Res Commun 2003;312: 1196e1201. 8. Liu XW, Bernardo MM, Fridman R, et al. Tissue inhibitor of metalloproteinase-1 protects human breast epithelial cells against intrinsic apoptotic cell death via the focal adhesion kinase/phosphatidylinositol 3-kinase and MAPK signaling pathway. J Biol Chem 2003; 278:40364e40372. 9. Liu XW, Taube ME, Jung KK, et al. Tissue inhibitor of metalloproteinase-1 protects human breast epithelial cells from extrinsic cell death: a potential oncogenic activity of tissue inhibitor of metalloproteinase-1. Cancer Res 2005;65:898e906. 10. Talvensaari-Mattila A, Turpeenniemi-Hujanen T. High preoperative serum TIMP-1 is a prognostic indicator for survival in breast carcinoma. Breast Cancer Res Treat 2005;89:29e34. 11. Wurtz SO, Moller S, Mouridsen H, et al. Plasma and serum levels of tissue inhibitor of metalloproteinases-1 are associated with prognosis in node-negative breast cancer: a prospective study. Mol Cell Proteomics 2008;7:424e430. 12. Lipton A, Leitzel K, Chaudri-Ross HA, et al. Serum TIMP-1 and response to the aromatase inhibitor letrozole versus tamoxifen in metastatic breast cancer. J Clin Oncol 2008;26:2653e2658. 13. Wu ZS, Wu Q, Yang JH, et al. Prognostic significance of MMP-9 and TIMP-1 serum and tissue expression in breast cancer. Int J Cancer 2008;122:2050e2056. 14. Schrohl AS, Holten-Andersen MN, Peters HA, et al. Tumor tissue levels of tissue inhibitor of metalloproteinase-1 as a prognostic marker in primary breast cancer. Clin Cancer Res 2004;10:2289e2298. 15. McCarthy K, Maguire T, McGreal G, et al. High levels of tissue inhibitor of metalloproteinase-1 predict poor outcome in patients with breast cancer. Int J Cancer 1999;84:44e48. 16. Sieuwerts AM, Usher PA, Meijer-van Gelder ME, et al. Concentrations of TIMP1 mRNA splice variants and TIMP-1 protein are
468
17.
18.
19.
20. 21.
22.
23.
24.
25.
Lee et al./ Archives of Medical Research 42 (2011) 463e468 differentially associated with prognosis in primary breast cancer. Clin Chem 2007;53:1280e1288. Nakopoulou L, Giannopoulou I, Stefanaki K, et al. Enhanced mRNA expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) in breast carcinomas is correlated with adverse prognosis. J Pathol 2002;197:307e313. Willemoe GL, Hertel PB, Bartels A, et al. Lack of TIMP-1 tumour cell immunoreactivity predicts effect of adjuvant anthracycline-based chemotherapy in patients (n 5 647) with primary breast cancer. A Danish Breast Cancer Cooperative Group Study. Eur J Cancer 2009; 45:2528e2536. Zhang B, Cao X, Liu Y, et al. Tumor-derived matrix metalloproteinase-13 (MMP-13) correlates with poor prognoses of invasive breast cancer. BMC Cancer 2008;8:83. Vizoso FJ, Gonzalez LO, Corte MD, et al. Study of matrix metalloproteinases and their inhibitors in breast cancer. Br J Cancer 2007;96:903e911. Kuvaja P, Talvensaari-Mattila A, Paakko P, et al. The absence of immunoreactivity for tissue inhibitor of metalloproteinase-1 (TIMP1), but not for TIMP-2, protein is associated with a favorable prognosis in aggressive breast carcinoma. Oncology 2005;68:196e203. Nakopoulou L, Giannopoulou I, Lazaris A, et al. The favorable prognostic impact of tissue inhibitor of matrix metalloproteinases-1 protein overexpression in breast cancer cells. APMIS 2003;111:1027e1036. Kuskunovic S, Radovic S, Doric M, et al. Immunohistochemical expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) in invasive breast carcinoma. Bosn J Basic Med Sci 2009;9:125e130. Del Casar JM, Gonzalez LO, Alvarez E, et al. Comparative analysis and clinical value of the expression of metalloproteases and their inhibitors by intratumor stromal fibroblasts and those at the invasive front of breast carcinomas. Breast Cancer Res Treat 2009;116:39e52. Zhang YG, Du J, Tian XX, et al. Expression of E-cadherin, betacatenin, cathepsin D, gelatinases and their inhibitors in invasive ductal breast carcinomas. Chin Med J 2007;120:1597e1605.
26. Normand SL. Meta-analysis: formulating, evaluating, combining, and reporting. Stat Med 1999;18:321e359. 27. Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med 1998;17:2815e2834. 28. Lee JH, Lee ES, Kim YS. Clinicopathologic significance of BRAF V600E mutation in papillary carcinomas of the thyroid: a meta-analysis. Cancer 2007;110:38e46. 29. Lee JH, Kim SH, Lee ES, et al. CD24 overexpression in cancer development and progression: a meta-analysis. Oncol Rep 2009;22: 1149e1156. 30. Brew K, Dinakarpandian D, Nagase H. Tissue inhibitors of metalloproteinases: evolution, structure and function. Biochim Biophys Acta 2000;1477:267e283. 31. Lambert E, Dasse E, Haye B, et al. TIMPs as multifacial proteins. Crit Rev Oncol Hematol 2004;49:187e198. 32. Holten-Andersen MN, Stephens RW, Nielsen HJ, et al. High preoperative plasma tissue inhibitor of metalloproteinase-1 levels are associated with short survival of patients with colorectal cancer. Clin Cancer Res 2000;6:4292e4299. 33. Yoshikawa T, Saitoh M, Tsuburaya A, et al. Tissue inhibitor of matrix metalloproteinase-1 in the plasma of patients with gastric carcinoma. A possible marker for serosal invasion and metastasis. Cancer 1999; 86:1929e1935. 34. Jung KK, Liu XW, Chirco R, et al. Identification of CD63 as a tissue inhibitor of metalloproteinase-1 interacting cell surface protein. EMBO J 2006;25:3934e3942. 35. Wurtz SO, Schrohl AS, Mouridsen H, et al. TIMP-1 as a tumor marker in breast cancerean update. Acta Oncol 2008;47:580e590. 36. Sorensen IV, Fenger C, Winther H, et al. Characterization of antiTIMP-1 monoclonal antibodies for immunohistochemical localization in formalin-fixed, paraffin-embedded tissue. J Histochem Cytochem 2006;54:1075e1086.
ORIGINAL ARTICLE
Serum TIMP-1 Predicts Survival Outcomes of Invasive Breast Carcinoma Patients: A Meta-analysis Ju-Han Lee,* Jung-Woo Choi,* and Young-Sik Kim Department of Pathology, Korea University Ansan Hospital, Ansan, Korea Received for publication March 4, 2011; accepted September 6, 2011 (ARCMED-D-11-00113).
Background and Aims. Tissue inhibitor of metalloproteinase-1 (TIMP-1) is a small secretory glycoprotein with multifunctional activity including anti-apoptosis and the inhibition of matrix metalloproteinase in invasive breast carcinomas. There have been contradictory results as to whether TIMP-1 is a poor or good prognostic factor in breast cancer patients. To address this controversy, we conducted a meta-analysis for the relationship between TIMP-1 levels and prognostic parameters in the breast cancer. Methods. The relevant published studies were pooled according to the defined selection criteria. The effect sizes of overall survival and prognostic parameters were calculated by a hazard ratio (HR) or an odds ratio (OR). HRs or ORs were combined using a randomeffects model. Results. Survival outcomes between high or elevated and low or normal serum TIMP-1 levels were compared by uni- and multivariate analyses involving 886 and 844 breast cancer patients, respectively. Patients with high or elevated serum TIMP-1 levels had unfavorable survival outcomes compared to patients with low or normal serum TIMP-1 levels in the uniand multivariate analyses (HR, 1.7 and 2.4; p !0.001 and p 5 0.033, respectively). However, no survival difference was evident in the data from tissue TIMP-1 levels by enzyme-linked immunosorbent assay and the expression of tissue TIMP-1 mRNA. The high or positive immunohistochemical expression of tissue TIMP-1 protein was not related to adjusted and unadjusted HRs, lymph node metastasis, and clinical stages. Conclusions. This meta-analysis indicates that serum TIMP-1 levels may be useful for predicting survival outcomes of invasive breast cancer patients. Ó 2011 IMSS. Published by Elsevier Inc. Key Words: TIMP-1, Breast cancer, Survival, Meta-analysis.
Introduction Breast cancer is the most common type of cancer among women in developed countries. An estimated 194,280 new breast cancer patients were diagnosed and 40,610 patients died in the United States in 2009 (1). Currently, the prognosis of breast cancer patients is determined by a number of clinical and biological parameters such as TNM stage, histological grade, steroid hormone receptor
*
These authors contributed equally to this work. Address reprint requests to: Young-Sik Kim, MD, PhD, Department of Pathology, Korea University Ansan Hospital, 516, Gojan-1 Dong, Danwon-Gu, Ansan-Si, Gyeonggi-Do, 425-707, Republic of Korea; Phone: 82-31-412-5322; FAX: 82-31-412-5324; E-mail: [email protected]
status, and HER-2/neu gene expression. However, for risk stratification, these factors are relatively inadequate or insufficient to precisely evaluate the prognosis of an individual breast cancer patient, leading to unnecessary adjuvant chemotherapy or unexpected cancer recurrences (2). Among the promising new biomarkers of breast cancers, tissue inhibitor of metalloproteinase-1 (TIMP-1) is a cellular survival factor with multiple functions that involve antiapoptosis, cell growth and differentiation, as well as matrix metalloproteinase (MMP) inhibition (3e9). Recent studies have shown that high levels of tissue or serum TIMP-1 are associated with worse outcomes of breast cancer patients (10e25). However, the prognostic value of TIMP-1 in breast cancer has been challenged by conflicting results according to the analytical method or sample type used. In an attempt
0188-4409/$ - see front matter. Copyright Ó 2011 IMSS. Published by Elsevier Inc. doi: 10.1016/j.arcmed.2011.09.006
464
Lee et al./ Archives of Medical Research 42 (2011) 463e468
to address these controversies, we performed a meta-analysis to clarify the relationship between TIMP-1 and the prognosis of breast cancer, especially focusing on the analytical method and clinical material chosen. Materials and Methods Data Collection and Selection Criteria for Meta-analysis We extensively searched the following online databases using the keywords ‘‘TIMP and breast cancer’’; 1) Medline using PubMed (http://www.ncbi.nlm.nih.gov/pubmed), 2) ISI Science Citation Index using the ISI Web of Science search interface (http://apps.isiknowledge.com), and 3) Conference Papers Index using the CSA Illumina search interface (http://www.csa.com/csaillumina). We also manually searched the reference lists of the found articles. Overlapping articles or duplicate data were excluded by examining the authors’ names and affiliations for each publication. The selection process of the papers is shown in Figure 1. The following types of articles were included: 1) original articles demonstrating that TIMP-1 was assessed only in primary breast cancer tissue or patients; articles that dealt with cell lines or animals were excluded; 2) articles published before November 2010 in English; 3) the most informative article when multiple articles were published by the same authors or groups. The following articles were excluded: 1) articles containing data inappropriate for meta-analysis; 2) review articles without original data; and 3) case reports. Data Pooling and Statistics Hazard ratios (HRs) or odds ratios (ORs) with 95% confidence intervals (CIs) were calculated by means of metaanalysis (26e29). For studies without HRs of survival, 370 papers were retrieved for key words and their abstracts were reviewed
288 articles were excluded, due to: (1) 22 papers were review or comment (2) 262 papers were obviously irrelevant or not associated with TIMP in primary breast cancer tissue or patients (3) 4 papers were not published in English
82 full texts were reviewed
66 articles were excluded, due to: (1) 46 papers did not provide sufficient data (2) 20 papers were published by the same authors or groups 16 studies were included in this meta-analysis
Figure 1. Paper selection flowchart for meta-analysis.
we assessed HRs and CIs using a published approximation method (27). HRs or ORs were combined using a randomeffects model (DerSimonian-Laird method). Q statistics were used to test interstudy heterogeneity, which is an adaptation of the c2 goodness-of-fit test. It was considered statistically significant with p !0.10. Sensitivity analysis was performed to examine the influence of each study on the pooled OR by serially omitting an individual study and pooling the remaining studies. Possible publication bias was investigated by visual assessment of a funnel plot and Egger’s test for the degree of asymmetry ( p !0.05 was considered a potential publication bias). Statistical analysis was performed using Comprehensive Meta-analysis Software v.2.0 (Biostat, Englewood, NJ); p value !0.05 was considered statistically significant. Results Serum TIMP-1 Levels by Enzyme-linked Immunosorbent Assay (ELISA) Four studies reported serum TIMP-1 levels on the survival outcomes of breast cancer patients (10e13). These studies, which included 310 breast cancer patients with high or elevated serum TIMP-1 levels and 576 with low or normal serum TIMP-1, presented unadjusted HRs and CIs on overall survival. The estimated unadjusted HRs ranged from 1.37e3.04. High or elevated serum TIMP-1 levels were significantly associated with poor overall survival rates compared with low or normal serum TIMP-1. The pooled HR for these studies was 1.662 (95% CI: 1.350e2.046; p !0.001) (Figure 2). No statistical heterogeneity was found among the studies (Q 5 1.208, df 5 3, p 5 0.751). Three studies addressed multivariate adjusted HRs and CIs on overall survival. Breast cancer patients consisted of 282 patients with high or elevated serum TIMP-1 levels and 562 patients with low or normal serum TIMP-1 (10e12). The prognostic variables used in the multivariate survival studies were patient age, histological grade, tumor size, lymph node metastasis, TNM stage, menopausal status, hormonal receptor status, c-erbB2, and MMP expression (Table 1). In our analysis, the estimated adjusted HRs ranged from 1.44e4.76. Pooled HRs for these studies was 2.371 (95% CI: 1.073e5.237; p 5 0.033) (Figure 3). There was statistical heterogeneity among the studies (Q 5 5.269, df 5 2, p 5 0.072). Tissue TIMP-1 Levels by ELISA Two studies including 1783 breast cancer patients with high tissue TIMP-1 levels and 1338 with low tissue TIMP-1 presented unadjusted HRs and CIs on overall survival (14,15). There was statistical heterogeneity among the studies (Q 5 11.052, df 5 1, p 5 0.001). There was no association between tissue TIMP-1 levels and overall survival rates (HR 5 2.656, 95% CI: 0.595e11.856; p 5 0.201).
TIMP-1 in Breast Cancer
Study name
Statistics for each study
465
Hazard ratio and 95% CI
Hazard Lower Upper ratio limit limit Z-Value p-Value Talvensaari-Mattila A Wurtz SO Lipton A Wu ZS
3.037 1.404 1.699 1.372 1.662
0.737 12.514 0.804 2.452 1.347 2.144 0.475 3.964 1.350 2.046
1.538 1.194 4.469 0.585 4.785
0.124 0.232 0.000 0.559 0.000 0.01
0.1
low TIMP-1
1
10
100
high TIMP-1
Figure 2. Hazard ratios with corresponding 95% CIs of individual studies and pooled data for overall survival between high or elevated serum TIMP-1 levels and normal or low serum TIMP-1 in univariate analysis.
Expression of Tissue TIMP-1 mRNA Two studies that included 1418 breast cancer patients described multivariate adjusted HRs and CIs on overall survival (16,17). Statistical heterogeneity was found between studies (Q 5 4.063, df 5 1, p 5 0.044). There was no association between tissue mRNA expression of TIMP-1 and overall survival rates (HR 5 1.171, 95% CI: 0.764e1.795; p 5 0.468). Expression of Tissue TIMP-1 Protein in Tumor Cells Articles that dealt only with TIMP-1 expression in cancer cells were included because TIMP-1 is expressed in stromal cells as well as cancer cells. Five studies using immunohistochemistry presented unadjusted HRs and CIs of TIMP-1 expression in breast cancer cells on overall survival. These studies included 810 breast cancer patients with high or positive TIMP-1 protein expression and 434 patients with low or negative TIMP-1 protein expression (18e22). There was statistical heterogeneity among the studies (Q 5 14.244, df 5 4, p 5 0.007). The estimated unadjusted HRs ranged from 0.4e2.0. The high or positive TIMP-1 expression was
not associated with overall survival rates (HR 5 1.104, 95% CI: 0.789e1.543; p 5 0.564). Four studies presented multivariate adjusted HRs and CIs on overall survival, which included 715 breast cancer patients with high or positive TIMP-1 expression in tumor cells and 375 patients with low or negative TIMP-1 expression (18e21). Statistical heterogeneity was found among the studies (Q 5 7.125, df 5 3, p 5 0.068). The high or positive TIMP-1 expression was not related to overall survival rates (HR 5 1.282, 95% CI: 0.970e1.693; p 5 0.081). Six studies investigated the relationship of TIMP-1 expression and lymph node metastasis (13,18,22e25). The high or positive expression of TIMP-1 protein was observed in 443 (70%) of 637 cases with lymph node metastasis and in 292 (65%) of 446 cases without lymph node metastasis. Statistical heterogeneity was detected among the studies (Q 5 20.937, df 5 5, p 5 0.001). The high or positive expression of TIMP-1 protein was not associated with lymph node metastasis (HR 5 0.952, 95% CI: 0.481e1.882; p 5 0.887). Four studies showed the relationship of TIMP-1 expression and clinical stage (13,21,22,25). The high or positive
Table 1. Covariates used in the multivariate survival models Study Serum TIMP-1 Talvensaari-Mattila et al. 2005 W€ urtz et al. 2008 Lipton et al. 2008 Tissue TIMP-1 mRNA Sieuwerts et al. 2007 Nakopoulou et al. 2002 Tissue TIMP-1 protein Willemoe et al. 2009 Zhang et al. 2008 Vizoso et al. 2007 Kuvaja et al. 2005
Variables used in Cox proportional hazard analysis
TIMP-1,* T, age TIMP-1,* size, grade, age, menopausal status, hormonal receptor status TIMP-1,* serum c-erbB2,* soft tissue metastasis,* liver metastasis,* KPS* TIMP-1, size,* N,* grade,* age,* menopausal status, ER,* PR* TIMP-1,* size, histological type, N, grade, stage, c-erbB2, Bcl-2, ER,* PR TIMP-1, size,a histological type,a N,a grade,a ER,a chemotherapy regimen* TIMP-1,* size,* N,* grade,* c-erbB2,* MMP-13* TIMP-1,* TIMP-2,* stage,* grade, age, ER, PR,* MMP-9,* MMP-11* TIMP-1,* size,* stage,* grade, menopausal status
T, tumor size; KPS, Karnofsky performance score; N, regional lymph node involvement; ER, estrogen receptor status; PR, progesterone receptor status. a Unmentioned statistical value. *p !0.05. Independent prognostic variable.
466
Lee et al./ Archives of Medical Research 42 (2011) 463e468
Study name
Statistics for each study
Hazard ratio and 95% CI
Hazard Lower Upper ratio limit limit Z-Value p-Value Talvensaari-Mattila A 3.400 1.197 9.661 Wurtz So 4.757 1.287 17.581 Lipton A 1.440 1.104 1.878
2.297 2.338 2.694 2.371 1.073 5.237 2.135
0.022 0.019 0.007 0.033 0.01
0.1
1
10
100
low TIMP-1 high TIMP-1 Figure 3. Hazard ratios and pooled data for overall survival between high or elevated serum TIMP-1 levels and normal or low serum TIMP-1 in multivariate analysis.
expression of TIMP-1 protein was found in 45 (63%) of 71 cases with stage III or IV and in 225 (66%) of 342 cases with stage I or II. The high or positive expression of TIMP-1 protein was not associated with clinical stage (HR 5 0.897, 95% CI: 0.357e2.252; p 5 0.816). There was no statistical heterogeneity among the studies (Q 5 5.842, df 5 3, p 5 0.120). Sensitivity Analysis and Publication Bias Sensitivity analyses revealed that Lipton’s study (12) slightly influenced the adjusted and unadjusted HRs (Figure 4). None of the other studies significantly affected the pooled HRs or ORs with CIs. In the funnel plots and the Egger’s regression tests, there was no evidence of publication bias for unadjusted survival analysis according to the serum TIMP-1 levels (Figure 5). In addition, publication bias was not observed in the analyses of survival curves and lymph node metastasis according to TIMP-1 protein expression. However, evidence of publication bias was suggested for adjusted survival analysis according to the serum TIMP-1 levels and for investigation of clinical stage according to TIMP-1 protein expression.
Discussion This pooled analysis using both uni- and multivariate survival data from 886 and 844 breast cancer patients, Study name
respectively, found that high levels of serum TIMP-1 are significantly associated with poor survival rates of breast cancer patients. TIMP-1 is a small secretory glycoprotein with a molecular size of 28.5 kDa that inhibits the proteolytic activity of extracellular matrix by MMPs (30,31). Given that MMPs play a central role in cancer cell invasion and dissemination, one would expect high levels of TIMP-1 to be related with a favorable cancer prognosis. Nakopoulou et al. (22) and Kuskunovic et al. (23) reported that overexpression of TIMP-1 protein was closely associated with favorable prognosis of breast cancer patients. However, this concept has been challenged by the evidence that TIMP-1 inhibits apoptosis of cancer cells (5,7e9). In fact, some investigations have shown that overexpression of TIMP-1 at mRNA and protein levels are observed in different carcinomas, including breast carcinoma. In addition, overexpression of TIMP-1 is highly associated with poor prognosis of patients with these carcinomas (10,15,17,19e21,32,33). The mechanism by which TIMP-1 suppresses apoptosis of breast cancer cells has been proposed (34,35). TIMP-1 binds to CD63 at the cell surface, leading to activation of the CD63/integrin b1 complex and its downstream signaling molecules including focal adhesion kinase, phosphoinositide 3-kinase, Akt, Bad and Bcl-2/ Bcl-XL. This meta-analysis reveals that serum TIMP-1 levels by ELISA have a clinical implication, which is significantly associated with a poor survival in breast cancer patients.
Statistics with study removed
Hazard ratio (95% CI) with given named study removed
Lower Upper Point limit limit Z-Value p-Value Talvensaari-Mattila A 1.640 Wurtz SO 1.708 Lipton A 1.520 Wu ZS 1.675
1.329 1.365 0.954 1.355 1.662 1.350
2.024 2.137 2.422 2.071 2.046
4.609 4.677 1.761 4.763 4.785
0.000 0.000 0.078 0.000 0.000 0.01
0.1
low TIMP-1
1
10
100
high TIMP-1
Figure 4. Sensitivity analysis of meta-analysis for unadjusted HR of overall survival according to serum TIMP-1 levels.
TIMP-1 in Breast Cancer
467
Conflict of Interest
Funnel Plot of Standard Error by log hazard ratio 0.0
The authors declare no conflict of interest.
Standard Error
0.2
Acknowledgment This study was supported by a Korea University Grant. 0.4
References
0.6
0.8 -2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
log hazard ratio
Figure 5. Funnel plot of meta-analysis for unadjusted HR of overall survival according to serum TIMP-1 levels.
The HR of high serum TIMP-1 levels by ELISA was approximately 2-fold. In contrast, patients with high levels of tissue TIMP-1 protein or mRNA failed to demonstrate any survival differences from patients with low levels of tissue TIMP-1. This suggests that serum TIMP-1 is more effective and clinically useful to predict the prognosis of breast cancer patients than tissue levels of TIMP-1 expression. It remains unclear that immunohistochemical scoring for TIMP-1 is an indicator for a poor prognosis of breast cancer patients. In this study, high or positive expression of TIMP1 protein was not associated with adjusted and unadjusted HRs, lymph node metastasis, and clinical stages. The immunohistochemical results of TIMP-1 have been a matter of debate because of different antibodies used, variations in tissue fixation time, no or less optimal epitope retrieval, and discordant data interpretation (36). Therefore, immunohistochemical quantification of tissue TIMP-1 expression as prognostic stratification in breast cancer needs to be further validated. There are some limitations to our study. First, studies used different analytical methods and clinical sample types. Serum TIMP-1 levels could be compared with overall survival, whereas serum TIMP-1 levels could not be compared with several parameters such as clinical stage, lymph node metastasis, and tumor size due to insufficient data for meta-analysis. Second, we calculated the loghazard ratio and its variance from Kaplan-Meier survival curves using the indirect method (27). However, the indirect approach for extracting summary statistics is sufficiently correct to make a useful graphic summary of two survival curves (27). In summary, our study indicates that serum measurement of TIMP-1 levels by ELISA from patients with invasive breast cancer may be useful in predicting their survival outcomes although additional large-scale investigations are needed.
1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2009. CA Cancer J Clin 2009;59:225e249. 2. Thomssen C, Janicke F. Do we need better prognostic factors in nodenegative breast cancer? Pro Eur J Cancer 2000;36:293e298. 3. Hayakawa T, Yamashita K, Tanzawa K, et al. Growth-promoting activity of tissue inhibitor of metalloproteinases-1 (TIMP-1) for a wide range of cells. A possible new growth factor in serum. FEBS Lett 1992;298:29e32. 4. Khokha R. Suppression of the tumorigenic and metastatic abilities of murine B16-F10 melanoma cells in vivo by the overexpression of the tissue inhibitor of the metalloproteinases-1. J Natl Cancer Inst 1994; 86:299e304. 5. Guedez L, Courtemanch L, Stetler-Stevenson M. Tissue inhibitor of metalloproteinase (TIMP)-1 induces differentiation and an antiapoptotic phenotype in germinal center B cells. Blood 1998;92: 1342e1349. 6. Kim YS, Seo DW, Kong SK, et al. TIMP1 induces CD44 expression and the activation and nuclear translocation of SHP1 during the late centrocyte/post-germinal center B cell differentiation. Cancer Lett 2008;269:37e45. 7. Lee SJ, Yoo HJ, Bae YS, et al. TIMP-1 inhibits apoptosis in breast carcinoma cells via a pathway involving pertussis toxin-sensitive G protein and c-Src. Biochem Biophys Res Commun 2003;312: 1196e1201. 8. Liu XW, Bernardo MM, Fridman R, et al. Tissue inhibitor of metalloproteinase-1 protects human breast epithelial cells against intrinsic apoptotic cell death via the focal adhesion kinase/phosphatidylinositol 3-kinase and MAPK signaling pathway. J Biol Chem 2003; 278:40364e40372. 9. Liu XW, Taube ME, Jung KK, et al. Tissue inhibitor of metalloproteinase-1 protects human breast epithelial cells from extrinsic cell death: a potential oncogenic activity of tissue inhibitor of metalloproteinase-1. Cancer Res 2005;65:898e906. 10. Talvensaari-Mattila A, Turpeenniemi-Hujanen T. High preoperative serum TIMP-1 is a prognostic indicator for survival in breast carcinoma. Breast Cancer Res Treat 2005;89:29e34. 11. Wurtz SO, Moller S, Mouridsen H, et al. Plasma and serum levels of tissue inhibitor of metalloproteinases-1 are associated with prognosis in node-negative breast cancer: a prospective study. Mol Cell Proteomics 2008;7:424e430. 12. Lipton A, Leitzel K, Chaudri-Ross HA, et al. Serum TIMP-1 and response to the aromatase inhibitor letrozole versus tamoxifen in metastatic breast cancer. J Clin Oncol 2008;26:2653e2658. 13. Wu ZS, Wu Q, Yang JH, et al. Prognostic significance of MMP-9 and TIMP-1 serum and tissue expression in breast cancer. Int J Cancer 2008;122:2050e2056. 14. Schrohl AS, Holten-Andersen MN, Peters HA, et al. Tumor tissue levels of tissue inhibitor of metalloproteinase-1 as a prognostic marker in primary breast cancer. Clin Cancer Res 2004;10:2289e2298. 15. McCarthy K, Maguire T, McGreal G, et al. High levels of tissue inhibitor of metalloproteinase-1 predict poor outcome in patients with breast cancer. Int J Cancer 1999;84:44e48. 16. Sieuwerts AM, Usher PA, Meijer-van Gelder ME, et al. Concentrations of TIMP1 mRNA splice variants and TIMP-1 protein are
468
17.
18.
19.
20. 21.
22.
23.
24.
25.
Lee et al./ Archives of Medical Research 42 (2011) 463e468 differentially associated with prognosis in primary breast cancer. Clin Chem 2007;53:1280e1288. Nakopoulou L, Giannopoulou I, Stefanaki K, et al. Enhanced mRNA expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) in breast carcinomas is correlated with adverse prognosis. J Pathol 2002;197:307e313. Willemoe GL, Hertel PB, Bartels A, et al. Lack of TIMP-1 tumour cell immunoreactivity predicts effect of adjuvant anthracycline-based chemotherapy in patients (n 5 647) with primary breast cancer. A Danish Breast Cancer Cooperative Group Study. Eur J Cancer 2009; 45:2528e2536. Zhang B, Cao X, Liu Y, et al. Tumor-derived matrix metalloproteinase-13 (MMP-13) correlates with poor prognoses of invasive breast cancer. BMC Cancer 2008;8:83. Vizoso FJ, Gonzalez LO, Corte MD, et al. Study of matrix metalloproteinases and their inhibitors in breast cancer. Br J Cancer 2007;96:903e911. Kuvaja P, Talvensaari-Mattila A, Paakko P, et al. The absence of immunoreactivity for tissue inhibitor of metalloproteinase-1 (TIMP1), but not for TIMP-2, protein is associated with a favorable prognosis in aggressive breast carcinoma. Oncology 2005;68:196e203. Nakopoulou L, Giannopoulou I, Lazaris A, et al. The favorable prognostic impact of tissue inhibitor of matrix metalloproteinases-1 protein overexpression in breast cancer cells. APMIS 2003;111:1027e1036. Kuskunovic S, Radovic S, Doric M, et al. Immunohistochemical expression of tissue inhibitor of metalloproteinase-1 (TIMP-1) in invasive breast carcinoma. Bosn J Basic Med Sci 2009;9:125e130. Del Casar JM, Gonzalez LO, Alvarez E, et al. Comparative analysis and clinical value of the expression of metalloproteases and their inhibitors by intratumor stromal fibroblasts and those at the invasive front of breast carcinomas. Breast Cancer Res Treat 2009;116:39e52. Zhang YG, Du J, Tian XX, et al. Expression of E-cadherin, betacatenin, cathepsin D, gelatinases and their inhibitors in invasive ductal breast carcinomas. Chin Med J 2007;120:1597e1605.
26. Normand SL. Meta-analysis: formulating, evaluating, combining, and reporting. Stat Med 1999;18:321e359. 27. Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med 1998;17:2815e2834. 28. Lee JH, Lee ES, Kim YS. Clinicopathologic significance of BRAF V600E mutation in papillary carcinomas of the thyroid: a meta-analysis. Cancer 2007;110:38e46. 29. Lee JH, Kim SH, Lee ES, et al. CD24 overexpression in cancer development and progression: a meta-analysis. Oncol Rep 2009;22: 1149e1156. 30. Brew K, Dinakarpandian D, Nagase H. Tissue inhibitors of metalloproteinases: evolution, structure and function. Biochim Biophys Acta 2000;1477:267e283. 31. Lambert E, Dasse E, Haye B, et al. TIMPs as multifacial proteins. Crit Rev Oncol Hematol 2004;49:187e198. 32. Holten-Andersen MN, Stephens RW, Nielsen HJ, et al. High preoperative plasma tissue inhibitor of metalloproteinase-1 levels are associated with short survival of patients with colorectal cancer. Clin Cancer Res 2000;6:4292e4299. 33. Yoshikawa T, Saitoh M, Tsuburaya A, et al. Tissue inhibitor of matrix metalloproteinase-1 in the plasma of patients with gastric carcinoma. A possible marker for serosal invasion and metastasis. Cancer 1999; 86:1929e1935. 34. Jung KK, Liu XW, Chirco R, et al. Identification of CD63 as a tissue inhibitor of metalloproteinase-1 interacting cell surface protein. EMBO J 2006;25:3934e3942. 35. Wurtz SO, Schrohl AS, Mouridsen H, et al. TIMP-1 as a tumor marker in breast cancerean update. Acta Oncol 2008;47:580e590. 36. Sorensen IV, Fenger C, Winther H, et al. Characterization of antiTIMP-1 monoclonal antibodies for immunohistochemical localization in formalin-fixed, paraffin-embedded tissue. J Histochem Cytochem 2006;54:1075e1086.