Abstract
Fully active MMP-2 is expressed at such low levels in human tissues that studies often fail to confirm its value as a cancer marker despite strong associations with malignancy. Our study utilized careful extraction, accurate activity measurements, standardization to purified controls and a new statistical metric to determine whether active MMP-2 is an effective indicator of colorectal cancer compared to pro-MMP-2 or pro-MMP-9. MMP-2 and MMP-9 activities were analyzed in matched normal and cancer samples from 269 patients by gelatin zymography, computer-assisted image analysis, serial dilutions of strong samples and standardization to controls. An index of effect size was designed for comparative evaluation of active MMP-2, pro-MMP-2 and pro-MMP-9 activities. For each gelatinase, mean activity and protein levels/mg soluble protein in normal mucosa and colorectal cancer were calculated for the first time with respect to commercial standards. Active MMP-2 activity, detected in 99% of colorectal cancers, was higher in 95% of cancers (on average 10-fold) than in normal mucosa. Levels of pro-MMP-2 and pro-MMP-9, but not active MMP-9, activities were also significantly higher in cancers versus normal. However, active MMP-2 activity provided the most effective test for the presence of cancer (p<0.0.0001) with an effect size statistically significantly larger than for either pro-MMP-2 or pro-MMP-9. Receiver operating characteristic (ROC) curves demonstrated that a cut-off for active MMP-2 of >44 SDU activity/mg soluble protein (>180 pg/mg), which is three times mean normal levels, would permit detection of colorectal cancer with an estimated sensitivity of 84% and estimated specificity of 93%.