A quantitative approach to determining disease response during therapy using multiple biologic markers: application to carcinoma of the breast.

This analytical study was undertaken in an effort to develop a model for a quantitative approach to the evaluation of multiple biological marker levels in blood and urine as a means for determining tumor changes during treatment of patients with malignant disease. The potential biologic markers measured in patients with carcinoma of the breast consist of three urinary polyamines (putrescine, spermidine and spermine), three urinary nucleosides (pseudouridine, N2, N2-dimethylguanosine and 1-methylinosine), and plasma carcinoembryonic antigen (CEA). The distribution patterns of the seven markers measured pretreatment and five weeks after initiating therapy were examined by grouping the patients into the three categories of progression, stable, or regression based on their clinical response to treatment. In addition to the individual marker measurements, the pretreatment and posttreatment values of the ratios of the polyamine levels (spermine/putrescine, spermine/spermidine, and spermidine/putrescine) and the nucleoside levels (N2, N2-dimethylguanosine/pseudouridine, 1-methylinosine/pseudouridine, and 1-methylinosine/N2, N2-dimethylguanosine) were also evaluated. In the pretreatment measurements, CEA levels were elevated for 76% of the patients and the three nucleosides were elevated for 36% of the patients and the three nucleosides were elevated for 36% to 37% of the patients. Urinary spermidine and spermine levels were abnormal for 27% and 24%, respectively, while putrescine levels were elevated for 7% of the patients. When all 14 marker measurements and the 12 ratios of these measurements were considered, the multiple regression equation evaluated the treatment results with a multiple correlation coefficient (R = 0.891; P less than 0.100) about 2.4 times higher than with the most sensitive single marker variable, N2, N2-dimethylguanosine/pseudouridine (R = 0.377; P less than 0.05), alone. Stepwise regression analysis revealed that the minimum number of multiple marker measurements and their ratios required to achieve the maximum value of the multiple correlation coefficient (R = 0.653; p = 0.010) was fifteen. These include the pre and posttreatment measurements of CEA, spermine, N2, N2-dimethylguanosine and 1-methylinosine, as well as two ratios of the polyamines and three ratios of the nucleosides in the post-treatment of the polyamines and three ratios of the nucleosides in the post-treatment measurements. These data suggest that the utilization of regression analysis to evaluate the monitoring utility of multiple marker measurements may be of clinical value.