Abstract Abstract #6057 *Würtz SØ and Klintman M contributed equally, Malström P and Brünner N shares the senior authorship.
 Background. Predictive markers are scarcely represented in breast cancer. TIMP-1 has in a previous study (n = 173) performed in our laboratory been shown to carry predictive information for the response to chemotherapy in metastatic breast cancer as high tumor tissue levels of TIMP-1 were significantly associated with a low objective response rate to treatment with the most frequently used chemotherapeutic drugs (CMF or anthracyclines). The purpose of the present study was to validate these previous results with a new independent patient population.
 Methods. The TIMP-1 level was measured using a validated ELISA in 168 primary tumor extracts from patients with metastatic breast cancer and levels were associated with the objective response to CMF and anthracyclines. Patients were included in the study provided that they had received chemotherapy for their metastatic disease and that frozen tumor tissue as well as data on their objective response to chemotherapy was available.
 Results. The median TIMP-1 level in responders was 17.3 (2.9 – 75.8) ng TIMP-1/mg protein and in non-responders it was 19.6 (0-190.3) ng TIMP-1/mg protein. When analysed as a continuous log-transformed variable, increasing tumor levels of TIMP-1 were associated with a decreasing probability of objective response to CMF or anthracyclines (OR = 1.59, 95% CI: 0.97 – 2.62, p = 0.07). This is very similar to the original study and thus supporting previous data. Next, we used a more clinically relevant approach for analysing the data from the validation study. For this analysis, objective response was scored as complete or partial response versus stable disease lasting at least six months (clinical benefit) versus progressive disease (non-response). This analysis showed that increasing levels of TIMP-1 were associated with a poor clinical benefit rate from chemotherapy (OR = 1.56, 95% CI: 0.98 - 2.51, p=0.06). The original study has not previously been analysed using this endpoint. When performing this analysis, similar results were found (OR = 1.62, 95% CI: 1.06 - 2.48, p = 0.02). Combining the new population and the original population in a pooled analysis (n = 341) using a random effects model showed that high levels of TIMP-1 were highly significantly associated with a poor clinical benefit rate from chemotherapy compared with patients with low levels (OR = 1.59, 95% CI: 1.16 - 2.18, p=0.003).
 Conclusion. The present validation study supports previous findings that primary tumor tissue levels of TIMP-1 carries predictive information in metastatic breast cancer treated with chemotherapy. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 6057.