Premium
The predictive value of a combination of tumor markers in monitoring patients with ovarian cancer
Author(s) -
Lahousen Manfred,
Stettner Haro,
Pickel Hellmuth,
Urdl Wolfgang,
Pürstner Peter
Publication year - 1987
Publication title -
cancer
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.052
H-Index - 304
eISSN - 1097-0142
pISSN - 0008-543X
DOI - 10.1002/1097-0142(19871101)60:9<2228::aid-cncr2820600921>3.0.co;2-n
Subject(s) - medicine , carcinoembryonic antigen , ovarian cancer , tumor marker , discriminant function analysis , cancer , cancer antigen , oncology , ferritin , chemotherapy , receiver operating characteristic , stage (stratigraphy) , laparotomy , gastroenterology , surgery , machine learning , computer science , paleontology , biology
To evaluate the predictive value of the serial determination of various tumor markers, we measured carcinoembryonic antigen, ferritin, cancer antigen 125, and tissue polypeptide antigen in 109 patients with ovarian cancer before surgery, during postoperative chemotherapy, and follow‐up. From these patients two groups were randomly selected. Group 1 (30 patients) had a favorable course, and Group 2 (30 patients) had an unfavorable course. Using the discriminant analysis we calculated a linear discriminant function and a cut‐off score. The two groups were thereby separated according to their scores (characteristic values) from their marker values. The scores accurately reflected the clinical course in 55 of the 60 patients (91.7%). This discriminant function was then used to make a prognosis in 49 patients. In 21 patients an elevated characteristic value (⩽ cut‐off score) indicated disease progression 5 months before clinical confirmation was possible. The remaining 28 patients scored below the cut‐off point. From six to 65 months (mean, 26.2) after surgery all are free of recurrence. It is concluded that invasive procedures, second‐look laparotomy, for instance, may not be necessary in following up ovarian cancer patients with normal tumor marker profiles.