Objective: To determine whether urinary matrix metalloproteinases (MMPs) predict the presence of ovarian cancer in patients with CA125 levels below the normal threshold of 35U/mL, a critical group of patients for whom no ovarian cancer biomarker is currently available. To determine whether these noninvasive biomarkers provide clinically useful information in the general ovarian cancer patient population as well.
Methods: ELISA analyses and substrate gel electrophoresis detected the levels and activity of urinary MMP-2, MMP-9, MMP-9/neutrophil gelatinase-associated lipocalin (NGAL) complex, and MMP-9 dimer in all ovarian cancer patients (n=97), those with CA125 <35U/mL (n=26) and controls (n=81).
Results: In patients with CA125 <35U/mL, receiver-operating characteristic (ROC) area under curve (AUC) analysis demonstrated that either urinary MMP-2 or MMP-9 or NGAL significantly discriminated between controls and ovarian cancer patients with normal CA125. Multivariate logistic regression revealed that the combination of urinary MMP-2 and MMP-9 provided the best diagnostic accuracy when multiplexed. When further multiplexed with age, the diagnostic accuracy of these biomarkers increased to a significant AUC of 0.820. These findings were consistent among the general ovarian cancer population studied as well, where the combination of urinary MMP-2 and MMP-9 multiplexed with age resulted in a highly significant AUC of 0.881. Pearson chi-square analysis revealed that higher urinary levels of either MMP-2 or MMP-9 were strongly associated with the increasing percentage of women with ovarian cancer independent of CA125 levels.
Conclusion: This study demonstrates the potential utility of urinary MMP-2 and MMP-9 to differentiate between ovarian cancer patients with normal CA125 levels and controls and suggests that urinary MMP-2 and MMP-9 may be a clinically useful aid in the diagnosis of advanced or recurrent ovarian cancer.
Copyright © 2011 Elsevier Inc. All rights reserved.