Vol. 47 No. 1/2000 149–156 QUARTERLY

A pilot study on relationships of selected molecular factors (c-myc oncogene average gene copy numbers (AGCN); serum CEA and CA 15.3 antigen levels; tumor cells' DNA values), to the ex vivo chemosensitivity of primary female human breast cancer in a modified adenosine triphosphate cell viability chemosensitivity assay (ATP-CVA), was performed. Four drug combinations were tested. A group of 75 cases of female primary breast cancer was assessed. Numerous correlations were found among molecular factors tested but none, with the exception of tumor grading, of these reflected ex vivo chemosensitivity of tumors tested. The results suggest that the parameters tested may not be important factors related to adjuvant chemoresponsiveness of primary human breast cancer to tested drug combinations.

was performed.Four drug combinations were tested.A group of 75 cases of female primary breast cancer was assessed.Numerous correlations were found among molecular factors tested but none, with the exception of tumor grading, of these reflected ex vivo chemosensitivity of tumors tested.The results suggest that the parameters tested may not be important factors related to adjuvant chemoresponsiveness of primary human breast cancer to tested drug combinations.
Breast cancer is still a leading cause of death among women concerning the neoplasmic disease around the world.Patients with operable breast cancer are primarily treated by surgery, and then by postoperative hormonal therapy, radiotherapy and/or chemotherapy, with the exception of patients with early-stage disease.The necessity of choosing a good adjuvant treatment for breast cancer patients, e.g. an effective chemotherapy regimen, is one of the key problems in the therapy.One of the ways of selecting the proper treatment is the analysis of clinicopathological data, as well as the status of some molecular factors known as having a predictive value in breast cancer, e.g.status of erbB-2 oncogene and/or its product (recently discussed by Clark, 1998).The other factors studied in the case of breast cancer (without established predictive value, however), are other oncogenes [e.g.myc family (Ryan & Birnie, 1996)] and serum markers, e.g.CEA and CA-15.3 antigens which, used in combination, correlate to breast cancer dissemination and may be useful markers of the effectiveness of therapy used, as well as markers of disease relapses (Molina & Gion, 1998;Robertson et al., 1999).
It is currently accepted that cancer patients' clinical and histopathological evaluation does not provide sufficient information as regards the response to chemotherapy.Drug selection for individual patients is usually based on reported response rates for clinical trials (so-called empiric treatment), but not on ex vivo individualized chemosensitivity assays.The response rates are usually better after treatment with the ex vivo selected chemotherapy regimens than with empiric ones, however, only one third of patients entered in prospective trials of ex vivo chemosensitivity test-ing were treated with an ex vivo best regimen, as reviewed by Cortazar & Johnson (1999).
In, at least, some cancer patients, more effective clinical response to chemotherapy may be achieved by selecting the most active drugs on the basis of ex vivo chemosensitivity test results for individual patients (Zabel et al., 1992;1997;Cree & Kurbacher, 1997;Cortazar & Johnson, 1999;Kurbacher et al., 1999).One of the ex vivo assays widely used in prediction of chemotherapy response, or in vitro drug testing in a variety of cancers, is the adenosine triphosphate cell viability chemosensitivity assay (ATP-CVA) and its modifications (Cree & Kurbacher, 1997;Falkiewicz & Vogt, 1999).It has a high evaluability rate and its results are independent of presence of non-cancerous cells in specimens evaluated (Cree & Kurbacher, 1997;Falkiewicz & Vogt, 1999).ATP-CVA is individually able to detect the majority of patients with in vivo poor or good response to chemotherapy; its sensitivity appears to be more than 85% and its specificity not less than 80%, with the true positive predictive value of more than 92% and true negative predictive value of more than 70% (Cree & Kurbacher, 1997;De Vita, 1997;Falkiewicz & Vogt, 1999).
As we hypothesize, even better results may be obtained when the ex vivo chemosensitivity assay will be combined with assessment of molecular factors with established or probable predictive value in individual patients with cancer.Therefore, we performed a preliminary study on the correlation of selected molecular factors [c-myc oncogene average gene copy numbers (AGCN); serum CEA and CA-15.3 antigen levels; tumor cells' DNA values] with the ex vivo chemosensitivity of ovarian cancer in the modified ATP-CVA.
Statistical analyses were performed using Statistica for Windows 4.3 software (Statsoft Inc., OK, USA).

c-myc Oncogene average gene copy numbers (AGCN).
The AGCNs were determined using double differential polymerase chain reaction (ddPCR).The general ddPCR reaction conditions, reproducibility and clinical usefulness have been described elsewhere (Brandt et al., 1994;1995a;1995b), and we used the ddPCR and electrophoresis conditions exactly as previously described (Vogt et al., 1998a).Primers for SOD2 and HBB (Brandt et al., 1995a), and for c-myc (Beckmann et al., 1999) fragments amplification have been described previously.

Serum carcinoembryonic antigen (CEA) and carbohydrate antigen 15.3 (CA-15.3).
The protein concentrations in serum were determined using commercially available CEA and CA-15.3 enzyme-linked immunosorbent assays (Abbott), in accordance with the manufacturer's instructions.DNA values.Tumor tissue was cut up with a scalpel to 1 ml of Partec HR-A solution (Lysing solution, Partec, Germany), then in-cubated for 10 min on a shaker board at room temperature.The cell suspension obtained was filtered through a 50 mm nylon mesh, 3 ml of Partec HR-B solution (Partec, Germany) was added, and the solution was incubated for 10 min at room temperature (DAPI-Staining).Analysis was performed directly with the Partec PAS Flow-Cytometry Analyzer using the HBO-Lamp excitation and Modfit or DPAC-Software for Cell Cycle Analysis and Ploidy Determination.The diploid region was defined as a DNA index from 0.95 to 1.05 and the tetraploid region as being from 1.9 to 2.1.

RESULTS AND DISCUSSION
The descriptive results of the determination of selected molecular factors in the patients studied are presented in Table 1.Some interesting statistically significant relationships among the patients' and tumor parameters were found and summarized in Table 2.
The c-myc gene amplification determined by ddPCR was found in 24 cases (32.0%) (according to the published cut-off value of AGCN > 2.0) (Berns et al., 1995)  What one may suspect is that the fraction of cells in G 0 G 1 phase significantly negatively correlates with their fraction in S phase or in G 2 M phase, as well as with DNA index; DNA index correlates positively with the fraction of cells in S phase.
In ATP-CVA, all 75 tumors tested gave the in vitro cultures (100%).Chemoresponse ex vivo of the tumors assessed was not affected by the molecular parameters tested.However, tumor grading is the single factor found to significantly influence the ex vivo chemosensitivity of cancerous cells, but not in every drug regimen.The results of statistical analysis of cor-relation between areas under chemoresponse curves (AUC) obtained in ATP-CVA and other factors tested are shown in Table 3.
The most effective drug regimen in the study was the PMi combination, giving the mean AUC of 17091, slightly less effective was the PE combination, giving the mean AUC of 16332, a worse regimen was the FEC combination, giving the mean AUC of 14528.The least effective regimen was the CMF combination, giving the mean AUC of 10735.In the study, all the ex vivo chemosensitivity results (expressed as AUCs) correlate strictly with each other (Table 4), despite the reported great variability in the effectivity of single drugs and drug combinations in breast cancer  specimens tested ex vivo (Kurbacher et al., 1996b).Various modifications of ATP-CVA have been applied in some studies for the ex vivo prediction of breast cancer patients' response to chemotherapy (Hunter et al., 1993;Petty et al., 1994;Köchli et al., 1993;1995;Cree et al., 1996;Kurbacher et al., 1996a;1996b;1999), or for chemosensitivity testing in breast cancer cell lines (several examples).We recently tested the combinations of ATP-CVA with cellular DNA parameters, oncogenes' gene dosage abberations and/or overexpression in a primary non-small cell lung cancer (Vogt et al., 1998b) and ovarian cancer (Vogt et al., 1999b).To the best of our knowledge, prior to our recent studies (Falkiewicz et al., 1999;Schlotter et al., 1999), ATP-CVA results in breast cancer have only once been correlated with molecular factors known as having a prognostic or predictive value in breast cancer, namely with p53 gene product expression, in a study which demonstrated a correlation between p53 protein expression and chemosensitivity of human breast cancer specimens (Petty et al., 1994).
In conclusion, the lack of correlation of c-myc AGCN, c-myc amplification status, serum CEA and CA-15.3 antigen levels, or proportions of cells in various cell cycle phases with chemosensitivity to all tested drug combinations, found in the present study, may suggest that the parameters tested may not be important factors related to adjuvant chemoresponsiveness of primary human breast cancer to tested drug combinations.In order to obtain useful and valuable results of combined testing we recently analysed the combination of other molecular factors (namely erbB oncogene family gene dosages and erbB-2 expression) with ATP-CVA in the breast cancer (Schlotter et al., 1999).The results of the study are promising and suggest that erbB-2 expression (but not erbB family gene aberrations), may be effectively combined in ex vivo chemoresponsiveness prediction of primary human breast cancer, but this preliminary suggestion awaits further clinical confirmation.

Vol. 47 Chemosensitivity of breast cancer ex vivo 153 Table 3 .
Relationships (p values) of AUC for four drug regimens tested to c-myc oncogene average gene copy numbers (AGCN), serum CEA and CA-15.3 antigen levels, and tumor cells' DNA values

Table 4 . Relationships (p values) between AUC and four drug regimens tested
a <0.000000 a -<0.000000 a PMi 0.001 a 0.000001 a <0.000000 a a Spearman rank order correlation