TIMP-1 and CEA as biomarkers in third-line treatment with irinotecan and cetuximab for metastatic colorectal cancer

Abstract

KRAS wild-type (wt) status determines indication for treatment with combination therapy, including epidermal growth factor receptor (EGFR) inhibitors, but still, the overall response rate in KRAS wt patients is less than 40 %. Consequently, the majority of patients will suffer from substantial side effects and no apparent benefit. Tissue inhibitor of metalloproteinases-1 is a glycoprotein, which regulates metalloproteinases and may consequently imply a central role in tumour progression. Furthermore, it is closely related to the EGFR regulation and has shown promising potential as a biomarker in colorectal cancer (CRC). The aim of the present study was to investigate the clinical value of TIMP-1 in patients with metastatic colorectal cancer (mCRC) treated with cetuximab and irinotecan. Patients with chemotherapy-resistant mCRC referred to third-line treatment with cetuximab (initial 400 mg/m2 followed by weekly 250 mg/m2)/irinotecan (350 mg/m2 q3w) were prospectively included in the biomarker study, as previously published. Pre-treatment blood samples were collected, and plasma TIMP-1 was measured by a validated in-house ELISA assay. In addition, carcinoembryonic antigen (CEA) measurement was performed with a standardised method. A total of 107 patients were included in the biomarker study. The median baseline plasma TIMP-1 level was 271.1 ng/ml (range 65.9–1432 ng/ml) with no significant associations with baseline clinical characteristics. Median baseline plasma TIMP-1 levels were significantly higher in patients with early progression compared to patients who achieved disease control, 349 ng/ml (233–398 95 % confidence interval (CI)) and 215 ng/ml (155–289 95 % CI), respectively, p = 0.03, suggesting some association with treatment efficacy. When dividing patients according to TIMP-1 tertiles, the median progression-free survival (PFS) in patients with a high level of TIMP-1 was 2.4 months (95 % CI 2.1–4.1) compared to 3.3 months (95 % CI 2.1–6.2) and 4.7 months (95 % 3.2–7.6) in patients with intermediate or low levels, respectively. Analysis of TIMP-1 as a continuous variable revealed a shorter PFS associated with increasing levels of TIMP-1 (hazard ratio (HR) 1.36). These results translated into a significantly lower overall survival (OS) in patients with a high baseline TIMP-1 level (4.5 months (95 % CI 3.4–5.4)), compared to those with intermediate or low TIMP-1 levels (7.8 months (95 % CI 4.4–13.7) and 12.0 months (95 % CI 10.1–14.3), respectively, p < 0.0001). An 83 % higher hazard for death was revealed (HR = 1.83) with each twofold increase in the TIMP-1 level. Pre-treatment levels of CEA were not associated with any of the baseline characteristics (except primary tumour localisation) or to differences in PFS or OS. The rank correlation between CEA and TIMP-1 was r = 0.50, and a test for interaction between TIMP-1 and CEA (dichotomised at 5 ng/ml) in survival analysis was not significant (p = 0.18). A multivariate analysis for PFS and OS resulted in a model with significant contributions from TIMP-1, KRAS, and the number of metastatic sites. We have confirmed the potential prognostic value of TIMP-1 measurement prior to palliative chemotherapy for mCRC. However, validation in randomised trials will be essential with the perspective of establishing a potential predictive role of plasma TIMP-1 in this setting.