A joint hazard and time scaling model to compare survival curves.

To provide a more general method for comparing survival experience, we propose a model that independently scales both hazard and time dimensions. To test the curve shape similarity of two time-dependent hazards, h1(t) and h2(t), we apply the proposed hazard relationship, h12(tKt)/ h1(t) = Kh, to h1. This relationship doubly scales h1 by the constant hazard and time scale factors, Kh and Kt, producing a transformed hazard, h12, with the same underlying curve shape as h1. We optimize the match of h12 to h2 by adjusting Kh and Kt. The corresponding survival relationship S12(tKt) = [S1(t)]KtKh transforms S1 into a new curve S12 of the same underlying shape that can be matched to the original S2. We apply this model to the curves for regional and local breast cancer contained in the National Cancer Institute's End Results Registry (1950-1973). Scaling the original regional curves, h1 and S1 with Kt = 1.769 and Kh = 0.263 produces transformed curves h12 and S12 that display congruence with the respective local curves, h2 and S2. This similarity of curve shapes suggests the application of the more complete curve shapes for regional disease as templates to predict the long-term survival pattern for local disease. By extension, this similarity raises the possibility of scaling early data for clinical trial curves according to templates of registry or previous trial curves, projecting long-term outcomes and reducing costs. The proposed model includes as special cases the widely used proportional hazards (Kt = 1) and accelerated life (KtKh = 1) models.

A tyrosine kinase profile of prostate carcinoma.

Tyrosine kinases play central roles in the growth and differentiation of normal and tumor cells. In this study, we have analyzed the general tyrosine kinase expression profile of a prostate carcinoma (PCA) xenograft, CWR22. We describe here an improved reverse transcriptase-PCR approach that permits identification of nearly 40 different kinases in a single screening; several of these kinases are newly cloned kinases and some are novel. According to this, there are 11 receptor kinases, 9 nonreceptor kinases, and at least 7 dual kinases expressed in the xenograft tissue. The receptor kinases include erbB2, erbB3, Ret, platelet-derived growth factor receptor, sky, nyk, eph, htk, sek (eph), ddr, and tkt. The nonreceptor kinases are lck, yes, abl, arg, JakI, tyk2, and etk/bmx. Most of the dual kinases are in the mitogen-activating protein (MAP) kinase-kinase (MKK) family, which includes MKK3, MKK4, MEK5, and a novel one. As a complementary approach, we also analyzed by specific reverse transcriptase-PCR primers the expression profile of erbB/epidermal growth factor receptor family receptors in a variety of PCA specimens, cell lines, and benign prostatic hyperplasia. We found that erbB1, -2, and -3 are often coexpressed in prostate tissues, but not in erbB4. The information established here should provide a base line to study the possible growth and oncogenic signals of PCA.