Skip to main content
. Author manuscript; available in PMC: 2012 May 1.
Published in final edited form as: Hum Pathol. 2011 Jan 15;42(5):688–701. doi: 10.1016/j.humpath.2010.08.021

Table 2.

Parameter estimates (robust standard errors) and approximate p-values from two fitted multinomial logit models that discriminate among colorectal tissue types using active MMP-2, pro-MMP-2 and pro-MMP-9 activity levels

Fitted Multinomial Logit Model
M1: Main effect of Active MMP-2 M2: Main effect of Active MMP-2, controlling for the main effects of Pro-MMP-9 & Pro-MMP-2
Adenoma (A) vs. Normal (N):
 Intercept −1.315*** (0.329) −1.149** (0.504)
 Active MMP-2 0.037** (0.012) 0.050** (0.017)
 Pro-MMP-2 −0.005 (0.003)
 Pro-MMP-9 0.001 (0.001)
High Grade Dysplasia(HGD) vs. Normal (N):
 Intercept −2.798*** (0.628) −2.046** (0.716)
 Active MMP-2 0.066*** (0.017) 0.095*** (0.030)
 Pro-MMP-2 −0.013 (0.008)
 Pro-MMP-9 0.001 (0.001)
Cancer (C) vs. Normal (N):
 Intercept −4.494*** (1.196) −3.606** (1.316)
 Active MMP-2 0.084*** (0.021) 0.114*** (0.033)
 Pro-MMP-2 −0.014 (0.008)
 Pro-MMP-9 0.001 (0.001)
−2LL 154.18 144.70
Pseudo-R2 Statistic 0.269 0.314
*

p<.05;

**

p<.01;

***

p<.001;

tissue samples, n=79, patients, n=30