Hepatocyte growth factor (HGF) induces tumor and endothelial cell migration, proliferation and angiogenesis through the receptor tyrosine kinase c-Met. c-Met is frequently overexpressed in human gliomas and expression level correlates with malignancy grade and prognosis. c-Met undergoes ectodomain shedding and soluble c-Met measurements in the plasma and urine of mice harboring human xenografts increase directly with tumor burden. We therefore examined the relationship between urinary c-Met levels in glioblastoma patients and prognosis.
Materials/Methods
c-Met protein was measured by electrochemiluminescent immunoassay in the urine of 23 patients with glioblastoma who had undergone radiotherapy. Urine was collected before, during, and at the completion of radiotherapy and at 1 month follow-up (80 specimens total). Baseline levels and changes in c-Met values were measured to determine whether they could predict disease progression. The change in c-Met values for each patient was determined by linear regression analysis using all available data points. Kaplan-Meier analysis was used to estimate time to disease progression in patients with high and low baseline c-Met values (relative to median), and in patients with increasing or decreasing c-Met values. The significance of the survival distributions was assessed by log-rank test. Cox proportional hazard models were used to evaluate the effect of RTOG RPA class and c-Met on time to disease progression.
Results
The median time to disease progression of all failed patients was 30.7 weeks (range, 5.3 - 118.1). Baseline c-Met levels were not found to be predictive of disease progression (p = 0.751), however, changes in c-Met levels were. Median time to progression in patients with increased and decreased c-Met values was 26.9 and 53.6 weeks, respectively. The log-rank test applied to the Kaplan-Meier curves resulted in a statistically significant p value of 0.0387. When adjusting for RTOG RPA class, patients in class IV had a hazard ratio of 1.73 and patients in class V had a hazard ratio of 2.56, indicating that in class V patients, increases in c-Met levels results in an increased risk by a factor of 2.56 compared to 1.73 in class IV patients. Although the coefficients for each of the covariates were not statistically significant, likely due to a small sample size, the adjusted log-rank p value for the survival curves was .08 suggesting that RTOG RPA class is probably an important explanatory variable in the outcome of these GBM patients.
Conclusions
Changes in urinary c-Met levels, but not baseline values, predict disease progression in patients with glioblastoma. We plan to extend these preliminary results to a larger sample cohort.
National Cancer Institute, NIH, Bethesda, MD
Author Disclosure: U.T. Shankavaram, None; A. Baschnagel, None; A. Russo, None; K. Jedlicka, None; M. Wernick, None; S. Smith, None; T. Cooley-Zglea, None; D. Citrin, None; D.P. Bottaro, None; K. Camphausen, None.
FULL TEXT