Abstract
Background: Higher plasma levels of insulin-like growth factor (IGF)-I and low levels of IGF binding protein (IGFBP)-3 have been associated with increased risk of incident prostate cancer (PCa), especially clinically advanced disease. However, their relationship with long-term survival after cancer diagnosis is largely unknown. Methods: Using prospectively collected blood samples during 1982-1984 in the Physicians\#8217; Health Study, we measured IGF-I, IGFBP-3, IGFBP-1, C-peptide (a marker of insulin production), and baseline PSA among men who were subsequently diagnosed with PCa during the 25 years of follow-up. We used KM-survival curves and Cox proportional hazard survival models to estimate time between cancer diagnosis to PCa -specific death, controlling for age at diagnosis, time between blood draw and PCa diagnosis, Gleason grade, and clinical stage. Analysis was further restricted to men without apparent clinically detectable PCa at baseline (PSA<4 ng/ml or, among those without available PSA, men whose PCa was diagnosed only after more than eight years of follow-up during PSA era, 1991-2007). We controlled for fasting status for all analyses related to IGFBP-1 and C-peptide. Results: Among the 769 men with available assays who were diagnosed with PCa during the follow-up, 128 (16.6%) subsequently died of the cancer. Overall, circulating levels of IGF-I, IGFBP-3, and IGFBP-1 were not significantly related to PCa -specific mortality. However, in the analysis restricted to men without clinically detectable PCa at baseline (551 men, 66 PCa deaths), both high IGF-I and low IGFBP-3 were associated with reduced PCa survival; across quartiles of blood levels, the multivariate-adjusted HRs (95% confidence interval) were 1.00 (ref), 1.06, 1.38, and 3.45 (1.39 -8.70), Ptrend=0.049 for IGF-I and 1.00 (ref), 0.60, 0.49, and 0.40 (0.16-0.98), Ptrend=0.20 for IGFBP-3, independent of age at diagnosis, clinical stage, and Gleason grade. We had previously reported that men with higher prediagnostic levels of C-peptide were at higher risk of dying from PCa. However, the association was much stronger in the analysis restricted to men without clinically detectable PCa; the multivariate-adjusted HRs were 1.00 (ref), 0.98, 1.08, and 3.46 (1.54 -7.73), Ptrend=0.0007. Baseline IGFBP-1 levels were unrelated to PCa survival. Conclusion: Among men without apparent clinically detectable PCa at baseline when blood samples were collected, higher prediagnostic levels of IGF-I, C-peptide, or lower levels of IGFBP-3 were associated with a higher risk of PCa specific mortality after diagnosis. Our long-term survival data suggest a biological role of the IGFs and insulin pathway in PCa progression independent of clinical predictors. These findings need to be further confirmed by larger studies.
Citation Information: In: Proc Am Assoc Cancer Res; 2009 Apr 18-22; Denver, CO. Philadelphia (PA): AACR; 2009. Abstract nr 3031.
100th AACR Annual Meeting-- Apr 18-22, 2009; Denver, CO