| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
Departments of
1
Laboratory Medicine and
3 Gynecology, Masaryk Memorial Cancer Institute, Zluty kopec 7, 656 65 Brno, The Czech Republic;
2
Institute of Computer Science, Pod Vodarenskou vezi 2, Prague, The Czech Republic;
4
Department of Nuclear Medicine, Charles University, E. Benese 13, 30599 Pilsen, The Czech Republic;
aauthor for correspondence: fax 420-5-4321-1169, e-mail valik{at}mou.cz
Recent randomized studies have reported significantly improved quality of life and survival in cancer patients who underwent tumor marker-directed treatment compared with patients treated according to Union Internationale Contre le Cancer (UICC) criteria [see Ref. (1) and references therein]. According to these studies, systematic analysis of tumor markers may document whether therapy is effective and should be continued in spite of its ultimate adverse toxic effects or whether it should be terminated because of ineffectiveness. The advantages of biochemical assessment are objectivity and reproducibility as well as the fact that tumor marker assays are subjected to longitudinal external and internal quality control. We aimed to determine whether changes in serum concentrations of the tumor marker CA125 evaluated by a newly developed algorithm predict response to chemotherapy in paclitaxel-treated ovarian cancer patients.
Ovarian cancer patients with advanced disease (stage III at the time of diagnosis; n = 101; median age, 56 years; age range, 33–69 years) were enrolled in a prospective study monitoring changes in serum CA125 and the effect of therapy. Patients were treated with paclitaxel (Taxol; Bristol-Myers-Squibb) as part of a first-line combination chemotherapy. They received a median of six courses (range, three to nine) of chemotherapy. During a total of 689 chemotherapy cycles, serum CA125 concentrations were evaluated (AxSYM; Abbott Laboratories) and externally controlled through the CAPTMX tumor marker survey. Serum samples were obtained before starting therapy (specimen c0) and before each chemotherapy cycle (specimens c1–cn). The univariate Cox regression model and the stepwise Cox regression model were used for selecting the optimal variables that most accurately predicted disease-free interval. The statistical package SAS, release 8.22 (SAS, Inc.), was used. The time-to-progression (TTP) in different patient groups was compared using Wilcoxon and log-rank tests. CA125 concentrations between groups of responders and nonresponders to therapy were compared using the Wilcoxon test for unpaired data, the median test, and analysis of variance.
When we used 30 units/mL as a cutoff value for CA125, it indicated disease recurrence with the following parameters, based on our patient database (n = 1378 ovarian cancers; clinical remission in 972 cases, progression in 406 cases): specificity, 94.2% [95% confidence interval (CI), 92.9–95.5%]; sensitivity, 76.9% (95% CI, 71.0–82.7%); positive predictive value, 68.0% (95% CI, 61.9–74.1%); negative predictive value, 96.2% (95% CI, 95.1–97.3%). Therapy was considered effective (responders) if complete or partial remission was achieved according to UICC criteria. Response was achieved in 76 patients. The response rate was 75.2% (95% CI, 66.8–83.7%), whereas progression occurred in 25 patients, or 24.8% (95% CI, 16.3–33.2%) of the cases (nonresponders). TTP varied from 4 to 38 months and is still being assessed in 17 patients (censored patients). We measured tumor marker concentrations before and after initiation of treatment (c0 ... cn), relative changes in tumor markers after the first and second cycles of chemotherapy (rc1 = c0/c1; rc2 = c1/c2, ... rcn = cn - 1/cn), and the ratios of these changes (rc1/rc2 ... rcn - 1/rcn) to compute a marker trend. We selected variables from sets c0, c1, c2, ... cn and their ratios, using the stepwise Cox model (Table 1
). In a multivariate stepwise Cox model, variables rc1, rc2, and rc1/rc2 were the most significant. From evident decreases in serum CA125 (decrease of 
50% from initial concentration after two cycles of chemotherapy: calculation = 1/1.6/1.26 = 0.4960 x 100% = 49.6%), we could distinguish a group of nonresponders and responders based on changes in CA125 concentrations (P <0.0001). Interestingly, in the group of nonresponders, a decrease of 
12% (median, 9%; interquartile range, 26%; limits, -4% and 22%) of CA125 was also observed after two chemotherapy cycles. Regression analysis of CA125 concentrations categorized responders into two subgroups after the second chemotherapy cycle: (a) good responders, who fulfilled the criteria rc1/rc2 >1.31 AND rc1 >1.6 AND rc2 >1.26 (see Table 1
); and (b) less favorable responders, who did not fulfill any of the above conditions. Prediction of duration of TTP was P <0.0002 with the Wilcoxon test and P <0.0036 with the log-rank test. TTP probabilities according to Kaplan–Meier are shown in Fig. 1
. Forty-three patients (56.6%; 95% CI, 45.4–67.7%) segregated into the group with a good prognosis, and 33 patients (43.4%; 95% CI, 32.3–54.6%) segregated into the group with a less favorable prognosis. The true TTP is shown in Fig. 1
, which includes 17 censored patients still undergoing chemotherapy (11 patients in the prognostically favorable subgroup, 6 patients in the unfavorable subgroup).
|
|
Significant health outcomes have been identified for use in making practice guidelines in clinical oncology based on overall survival, including disease-free survival; quality of life; reduced toxicity; and cost-effectiveness (2). The therapeutic efficacy evaluated in individual cancer cases according UICC criteria has several weaknesses: a lack of sensitivity in measuring the extent of metastatic disease and a lack of reproducibility between clinicians in interpreting changes seen from imaging methods (3). A recent American Society of Clinical Oncology statement (4) stipulated that tumor markers are the preferred method for monitoring disease progression in patients on systemic therapy when the disease is not easily assessable by imaging techniques (i.e., blastic bone metastases). The important advantage of marker measurement compared with imaging methods is that increases in tumor markers may draw attention to disease progression in distant organ(s) that may not have been studied by imaging methods. Some randomized studies have reported important improvements in quality of life and survival (1) for patients with marker-directed treatment compared with patients treated according to UICC criteria (5). Monitoring of therapy efficacy is emerging as a valuable clinical application of serum tumor markers, and positive correlations have been reported between changes in concentrations of some markers and the response to systemic therapy in patients with cancer (6). A decrease in tumor marker concentrations to normal was reported to indicate remission of disease (7)(8) or to predict response to therapy (9).
Taxanes are, at present, established compounds considered effective in the treatment of ovarian cancer (10). Evaluation of CA125, but not carcinoembryonic antigen, was reported to be useful according to some studies (11)(12). Normalization of CA125 after the first course of treatment corresponded to a 53% survival rate, and the rate of decrease in CA125 concentrations was a sensitive predictor of the likelihood of achieving complete remission and prolonged survival (13). A decrease in CA125 was an independent prognostic factor for survival of women with advanced ovarian cancer and allowed identification of a high-risk population among patients with advanced ovarian cancer (providing the possibility of considering second-line chemotherapy earlier). The discrimination power of serial CA125 determinations for long-term survival seemed to be temporary, however, and prediction of the outcome for individual patients was much less precise (14). Several definitions for trends in serum CA 125 have been proposed (15)(16)(17), but the only definition that has been validated prospectively is that of Rustin et al. (18).
A recent analysis of 1396 patients involved in the testing of 14 investigational drugs for relapsed ovarian cancer in phase II clinical trials found that CA125 response rates were slightly higher than standard response rates, by a factor of 1.11, and could therefore provide a reliable, less expensive, and more available means of identifying active drugs worthy of further study (19). Some authors have raised concerns that specific drugs, such as paclitaxel, may render the CA125 concentration unreliable for indicating response (20). This question has been reexamined recently in 144 patients treated with paclitaxel in four different trials using the 50% and 75% response criteria defined above (10). Progression-free survival for responders compared with nonresponders was equivalent regardless of whether CA125-based criteria or standard response criteria were used. The false-positive rate for CA125 response was <3%, suggesting that if a decrease in the CA125 concentration indicates a response, then there is a response in 97% of cases and radiologic reassessment is probably not warranted. False-negative rates, however, were higher, reaching 21%, and stopping treatment based on the absence of a response according to CA125 alone would therefore pose the risk of undertreating patients. Such undertreatment could be avoided if therapy was continued until there was evidence of progression by clinical, radiologic, and/or CA125 criteria.
In summary, we were interested whether sequential changes in tumor marker concentrations computed by a newly developed algorithm (21)(22) reflected a response to chemotherapy in paclitaxel-treated advanced ovarian cancer patients to predict further courses of disease. Our results support, in principle, those presented by Rustin and coworkers (18)(23) and Peters-Engl et al. (14). In addition, we think that assessing therapy efficacy based on serum CA125 after the second cycle (and/or after the third cycle and radiologic confirmation of progressive disease) opens an option for clinicians to consider appropriate modification of chemotherapy. Such an approach may contribute to individualization of treatment of ovarian cancer patients.
Acknowledgments
This work was supported in part by Grant IGA MZCR 6405-3.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
), who are still undergoing chemotherapy.