|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2003 American Society for Clinical Oncology Use of CA-125 to Assess Response to New Agents in Ovarian Cancer Trials
From the Department of Medical Oncology, Mount Vernon Cancer Centre, Northwood, Middlesex, United Kingdom. Address reprint requests to Gordon J.S. Rustin, MD, Department of Medical Oncology, Mount Vernon Cancer Centre, Northwood, Middlesex HA6 2RN, United Kingdom; email: rustin{at}mtvern.co.uk.
The purpose of this article is to demonstrate how CA-125 could be used in clinical trials to ascertain the efficacy of new drugs for ovarian cancer. Studies that have investigated the use of CA-125 in assessing response and progression of ovarian cancer are reviewed. A precise CA-125 response definition that requires either a 50% or 75% decrease in CA-125 levels has been shown in trials of both initial chemotherapy and in phase II trials to predict accurately the response in comparison with standard response criteria. A simpler response definition that is based on just a 50% decrease in CA-125 levels has been proposed by the Gynaecological Cancer Intergroup (GCIG) but requires further validation. Definitions for progression have also been proposed by the GCIG on the basis of a confirmed doubling of CA-125 levels from either the upper limit of normal or the nadir CA-125 level. These CA-125 definitions for progression falsely predict progression in fewer than 2% of patients and can be used to define the date of progression. Precise definitions for response and progression according to CA-125 should be incorporated into ovarian cancer clinical trial protocols. These definitions already have been shown to be valuable in assessing efficacy of new agents but require further prospective evaluation.
ALMOST THREE fourths of women diagnosed with ovarian cancer are given chemotherapy, yet die from their disease. Better drugs are required to treat this disease. Two major end points used in clinical trials of new anticancer drugs are response rates and progression-free survival. A substantial proportion of ovarian cancer patients have disease that is poorly visualized on scans, so these patients cannot be evaluated for response. Serum levels of CA-125 are elevated in more than 90% of patients with advanced epithelial ovarian cancer.1,2 CA-125 is used routinely in clinical practice to determine whether patients are responding to therapy or have progression of their ovarian cancer. A serial decrease usually leads to continuation of the current therapy, whereas a serial increase is likely to lead to a change in patient management. There is, therefore, the potential to use CA-125 to define response in these patients.3 For use in clinical trials, precise definitions have to be produced from one data set and then validated in several different data sets. This review describes the development of definitions for response on the basis of CA-125. Paradoxically, the widespread use of CA-125 during initial chemotherapy and follow-up frequently leads to a change or reintroduction of chemotherapy. If second-line therapy is started before progression has been confirmed by standard criteria, it becomes impossible to define the date of progression. This review therefore also addresses the problem of how the date of progression can again be incorporated as a reliable end point by incorporating CA-125 levels into the definition of the date of progression.
Therasse et al4 have recently published new response evaluation criteria in solid tumors (Response Evaluation Criteria in Solid Tumors Group [RECIST]) that are the result of a large international collaboration between the European Organization for Research and Treatment of Cancer, the National Cancer Institute of the United States, and the National Cancer Institute of Canada Clinical Trials Group. These guidelines supersede the previously used World Health Organization standard criteria5,6 and use unidimensional measurements of target lesions to calculate response instead of the previous bidimensional approach. These guidelines are now being incorporated into the end points of all new clinical trials. The regulatory authorities accept the definitions in these guidelines for licensing new drugs, and any definitions incorporating new methods of measuring tumors (such as tumor markers) must be compared with these standard criteria.
Initial Chemotherapy Several definitions of CA-125 response have been proposed,3,7–9 and Table 1  shows how the response rate can vary from 10% to 62% in the same trial if different response criteria are used. Only one CA-125 response definition has been retrospectively validated and is discussed further.9 The initial CA-125 response criteria were derived from 277 patients within the North Thames Ovary Trial, which compared maintenance radiotherapy with carboplatin. The sensitivity and specificity of different criteria were examined in this patient group. To reduce false-positive prediction of response to less than 2% and to maintain a sensitivity of at least 65%, two definitions were finally accepted. Patients could be classified as responding according to CA-125 if there was either a 50% decrease over four samples or a 75% decrease over three samples (Table 2). Once these 50% and 75% CA-125 response criteria had been established, they were then tested in two other patient groups: 254 patients in the North Thames Ovary Trial of five versus eight courses of chemotherapy and 458 patients in the Gynecologic Oncology Group (GOG) Protocol 97, which both investigated dose-intense versus standard chemotherapy. Of all 620 patients assessable by CA-125 criteria in the GOG trial, only two patients (0.3%) showed a CA-125 response at the time of clinical progression. CA-125 response rate was 66% in all 317 patients assessable by CA-125 criteria and 67% in 221 patients nonmeasurable according to GOG standard criteria. This compares to a GOG-defined response rate of 62%.
To take account of natural variations in CA-125, missing samples, extra samples, and changing upper limits of normal, the definition had to be precise, so it was written using mathematic logic (Table 3 ). For ease of use and to prevent errors, it was therefore incorporated into a computer program called Tumor Marker Evaluation, which was written using Microsoft Visual FoxPro (Redmond, WA) for Windows 95. Many trial groups and pharmaceutical companies have since analyzed their CA-125 data using this program, which leads to additional validation of the definitions. Inquiries should be directed to the medical oncology data manager at rustin{at}mtvern.co.uk.
Relapse Therapy A difficulty in correlating CA-125 response with clinical response is that some patients are only assessable by scans and some are only assessable by CA-125. Table 1  demonstrates how the use of a correlation coefficient can be misleading if one is in effect comparing apples with oranges. To overcome this problem, a recent analysis examined how accurate the response rates obtained using the 50% and 75% response definitions (Tables 2 and 3) were at predicting the activity of drugs in phase II trials for ovarian cancer, compared with the response rates obtained by standard criteria.11 In this analysis, data were obtained from 25 treatment groups within 19 phase II clinical trials that studied 14 different cytotoxic drugs for relapsed ovarian cancer. Response rates were estimated in 1,457 assessable patients according to standard criteria and in 1,092 assessable patients according to CA-125. For each trial, the observed response rates acted as evaluation of how the two criteria would perform in a hypothetical Gehan two-stage phase II trial, which accepts a target drug efficacy rate of 20% and a rejection error of 5%. The results from this analysis showed that there was no statistical difference between response rate obtained by standard or CA-125 criteria and, therefore, no difference in accurately predicting whether a phase II drug is active and worth pursuing in further clinical trials. Examples of some of the trials are shown in Table 4. The overall results showed that CA-125 and clinical response criteria were concordant in 20 of 25 groups, with less than 5% chance of rejecting the drug in nine groups and greater than 5% chance of rejecting the drug in 11 groups. In four groups, the drug had less than 5% chance of being rejected by CA-125 but greater than 5% chance of being rejected by standard criteria, whereas there was only one group in which the drug was rejected by CA-125 but not by standard criteria.
Concern has been raised by several authors that CA-125 response may be altered by certain drugs especially paclitaxel.12,13 Bridgewater et al,14 however, reanalyzed the data of four trials of paclitaxel as first- or second-line therapy using the 50% and 75% response criteria as stated above. In 144 patients treated, the response rate according to standard criteria and CA-125 criteria were equivalent (30.7% v 31.7%). Furthermore, the false-positive rate was 2.9% with paclitaxel and 2.2% with cisplatin, indicating that precise CA-125 response criteria behave similarly with the two drugs. The use of weekly paclitaxel and therefore weekly measurements of CA-125 could be a possible explanation for why some investigators have found CA-125 to be less reliable when paclitaxel therapy is monitored. The 50% and 75% response definitions were derived and tested in patients receiving therapy once every 3 to 4 weeks, so any interference from tumor lysis by the therapy would be lessened. The above data clearly identify CA-125 as an accurate, reliable, and inexpensive means of assessing response. The data also demonstrate how in the majority of circumstances, CA-125 assessment can replace the need for expensive and time-consuming radiologic assessment. Examples of the use of these definitions can be found in the publications on trials of altretamine, weekly cisplatin plus oral etoposide, and platinum plus tamoxifen.15–17
In the guidelines outlining the new RECIST criteria, Therasse4 discussed the role combining standardized tumor marker response definitions with the RECIST response criteria in future clinical trials. For ovarian cancer trials this has the advantage of evaluating more patients according to CA-125 and RECIST or by both criteria. However, the CA-125 response and progression criteria were derived from known data sets and have been validated retrospectively. Therefore, at present, the use of CA-125 response criteria as a secondary end point in clinical trial design should be discussed with the regulatory authorities before being incorporated in a trial protocol. The initial aim of any phase II trial is to determine whether the activity of a drug is sufficient to justify additional investigation. If response rates according to CA-125 are lower than a predetermined threshold efficacy, the drug should be rejected and additional studies could be terminated. However, if response rates according to CA-125 are satisfactory, the patient numbers within the trial should be expanded to allow sufficient patients to be evaluated by both CA-125 and RECIST criteria.
Once it has been decided to incorporate CA-125 criteria into future trials, a standardized CA-125 definition must be used. Many clinicians have been confused by the complexity of the Rustin CA-125 response criteria and are not clear that a patient is classified as a responder if they have either a 50% or 75% response, the former requiring four samples but the latter requiring only three samples. A major reason for requiring two definitions was that from retrospective analysis of trial data not originally designed to use CA-125 as a response indicator, most patients did not have two pretreatment CA-125 samples and it was not clear whether the single pretreatment level was truly reliable. The Rustin criteria therefore accepted patients as assessable according to CA-125 providing they had one sample between 9 days before and 35 days after the start of chemotherapy. Thus, it was possible to assess response even in patients whose first CA-125 sample was after the start of therapy. The Gynaecological Cancer Intergroup (GCIG) recently proposed a simplified CA-125 definition that uses just the 50% response criteria but requires two pretreatment CA-125 levels at least twice the upper limit of normal, with samples taken at least 1 week and not than more than 3 months apart. This definition, which is outlined in Table 5
The GCIG is eager for current and future clinical trials in ovarian cancer incorporate this definition into their protocol and make the data available either to Monica Bacon (mbacon{at}ctg.queensu.ca) or Gordon Rustin (see contact details) so that the definition can be validated. A French group has recently evaluated this definition in 595 patients from six consecutive phase II trials.18 Response rates were assessed independently according to RECIST in 338 patients with measurable disease and according to the 50% CA-125 response definition in 490 patients. They showed that the CA-125 definition both predicted progression-free survival and correlated with RECIST response. Patients who have early disease progression or severe drug toxicity and have not had four samples taken for CA-125 before discontinuing the study are not assessable according to CA-125. If there is a high proportion of such patients (eg, in a trial of a toxic inactive drug), the denominator of CA-125–assessable patients could be low enough to produce an erroneously high response rate. This potential bias is overcome by performing an intention-to-treat analysis in which the denominator of eligible patients is all patients with two elevated pretreatment CA-125 levels.
Complete or Partial Response or Stable Disease According to CA-125 Many drugs now entering clinical trials are expected to produce disease stabilization rather than partial response. A particularly good group of patients in whom to test these drugs are those with an asymptomatic increase in CA-125 levels who qualify as having progression according to CA-125. It is important that if an elevated CA-125 level is used as an entry requirement that it complies with the definition of progression according to CA-125 described below. There are then several options for how response could be ascertained according to CA-125. The simplest is to use the response criteria discussed above. If only disease stabilization is expected, one could look for decreases of less than 50% that are maintained for a specified period, as performed in a trial of isotretinoin and calcitriol.19 Another option would be to consider the CA-125 level at the start of the new trial therapy as the nadir CA-125 level. A confirmed doubling from that level would then be considered progression. The time before progression could be considered the period equivalent to stable disease. Any unrecognized CA-125 definition such as stable disease would have to be precisely defined. It would then require testing in different trials and validation against standard criteria. If just a doubling of CA-125 levels is used as an entry requirement without the need for the increase to be to above a certain level, it is likely that the CA-125 level will not continue to increase at the same rate. Thus, a slower increase of CA-125 after starting the test therapy might be mistakenly assumed to indicate activity. A cautious approach is therefore necessary when introducing any new definitions that are based on CA-125.
An increasing CA-125 has been shown to predate clinical relapse in approximately 70% of patients with a median lead time of 4 months.20,21 It is therefore used as an early indicator of disease relapse. In many clinical trials of first- and second-line treatment for ovarian cancer in which progression-free survival is a major end point, it is well known that many investigators will instigate second-line treatment because of an increase in serum CA-125 levels before clinical or radiologic signs of progression. This then causes great confusion when the date of disease progression is determined; wide variability exists among different trial groups about what to do with the data. Some groups will include these patients in the progressive disease population, other groups will censor these patients, and others will ignore treatment before the documented date of clinical progression. The GCIG therefore proposed that a precise definition of progression be used as a secondary end point in clinical trials.22
Several definitions of progression according to CA-125 have been proposed. An increase of 50%, 100%, or just to above the normal range have all be shown to be predictive of relapse.1,3,21,23 Rustin et al24 produced and validated a definition that was based on a serial increase of 25% over four samples, 50% over three samples, or levels persistently elevated above 100 U/mL, which required a computer program to maintain accuracy. Two simpler definitions have now been produced. The first was developed after studying 255 patients in the North Thames Ovary Trial of five versus eight courses of chemotherapy.20 It was found that a confirmed increase of serum CA-125 to more than twice the upper limit of normal after first-line chemotherapy predicted tumor relapse with a sensitivity of 84% and a false-positive rate of less than 2%. The second definition was developed from studying 88 patients whose CA-125 levels remained persistently elevated during and/or after first-line chemotherapy. In this group a confirmed doubling of CA-125 from its nadir level predicted progression with a sensitivity of 94% and almost 100% specificity.25 The GCIG has produced a definition on the basis of these last two criteria (Table 6
In conclusion, this review describes in detail the currently available definitions of response and progression according to CA-125. Their routine use in clinical trials will lead to a greater proportion of patients becoming eligible because the great majority will be assessable by standard criteria, CA-125 criteria, or both. The use of the CA-125 criteria should enable positive or negative decisions on further development of new drugs to be made more quickly, leading to both cost savings and reduction in exposure of patients to inactive drugs. The CA-125 definitions should also lead to a reduction in expensive, time-consuming, and often uncomfortable radiologic investigations. The regulatory authorities should be consulted before these CA-125 definitions are used as major end points in clinical trials. There are various quality control issues that need to be addressed when tumor markers are used in clinical trials. The laboratories performing the assays should be participating in a quality control scheme. All samples from an individual patient should be assayed by the same type of assay because there can be considerable differences in the upper limit of normal among assays. Investigators need to be aware of exogenous factors that might affect CA-125 levels. These include recent surgery, peritoneal or pleural fluid drainage, and administration of mouse antibodies. Additional prospective validation of the CA-125 response and progression definitions is required so that in conjunction with RECIST, they can form the basis of evaluating efficacy in future ovarian cancer trials.
DR. CANNISTRA: Is there any concern regarding the use of CA-125 as a surrogate marker of response to biological agents? Is it conceivable that perhaps we could miss biological or clinical activity of an agent that inhibits signal transduction or an antiangiogenic agent if we simply look at CA-125 response, for example, in a minimal disease setting where you can’t use RECIST criteria [J Natl Cancer Inst 92:205–216, 2000]? DR. ROWINSKY: With many of these new therapeutics, there’s going to be a range of responses with regard to the tumor as well as to CA-125. So CA-125 in some individuals will stabilize in a cytostatic sense. I’d be worried about whether or not we would miss a drug. We have to validate what the various types and rates of responses mean. I’m worried about the fact that in a population, a drug may be valuable as far as decreasing time to progression and having low toxicity, CA-125 could be increased, but the rates of CA-125 could be declining. We’re going to have to validate what a rate decrement really means in terms of the validity of new agents. DR. CANNISTRA: Is there some merit to considering whether or not from a serologic point of view you should define what stable disease is? DR. BAST: You ought to be able to reverse the methodology. DR. SKATES: Yes, with enough data. DR. BAST: The other confounding possibility is that some of these signal transduction inhibitors might affect the secretion or processing of CA-125. Now that CA-125 has been cloned, our lab has begun to look at what modifies and modulates CA-125 expression, at least in tissue culture. We have tested inhibitors of several pathways and have seen little effect on CA-125 expression or secretion. DR. CANNISTRA: That’s encouraging, since it means that CA-125 may still prove to be a useful marker of response in patients treated with signal transduction inhibitors, although this would have to be formally demonstrated. DR. BAST: For clinical monitoring, this is an advantage, but it does not help us understand the regulation of CA-125. However, it is probably a useful principle that if you are going to use any marker to monitor patients who are being treated with signal transduction inhibitors, one should be certain that the drug does not affect processing of that particular marker. DR. CANNISTRA: How many of us are using CA-125 as an endpoint in their clinical studies? DR. VASEY: We’ve just completed a trial of Capecitabine (Xeloda) using CA-125 as a primary endpoint. One of the things we found is that if you’re using a resistant patient population, you are losing more eligible and evaluable patients because of the need to exclude people who have ascites. There are patients in whom the number of ascitic drains you have to do actually decreases as the patient responds. For example, with topotecans, sometimes four cycles are needed before you get a response. This has actually cut down on the ability to use CA-125 as your endpoint. DR. RUSTIN: Also, what you do about patients who come off a drug because it’s too toxic or not of any use? You could say they’re not CA-125 evaluable. The way of getting around that is to make sure that you do an intention-to-treat analysis. DR. CANNISTRA: How often is CA-125 being used as a marker of response in GOG trials? DR. BOOKMAN: Not at all. It’s used as a marker of progression on the current phase 3 randomized trial. It’s not incorporated as a measure of response on any of the phase 2 trials. DR. CANNISTRA: Are there plans to do so? DR. BOOKMAN: It’s more complicated than that. When the phase 2 trials are divided based on the treatment-free interval, ideally, you’d want to think of that more as a progression-free interval, because when you start treatment on a patient, it can be a somewhat arbitrary decision. Let’s say somebody’s CA-125 goes up and then 3 months later their CT scan shows an abnormality and then 3 months after that they get chemotherapy. Which one do you use as the endpoint to define your treatment-free interval? We’ve taken a conservative approach and said that it’s the clinical evidence of disease, meaning not just CA-125, but something in addition to that. However, one could have just as easily argued to take an endpoint based on CA-125. So it’s not currently used for either of the two.
Supported by grants from The Cancer Treatment and Research Trust.
1. Bast RC, Klug TL, John ES et al: A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med, 309:883–887, 1983[Abstract] 2. Tuxen MK, Soletormos G, Dombernowsky P: Tumor markers in the management of patients with ovarian cancer. Cancer Treat Rev 21:215–245, 1995[CrossRef][Medline]
3. Rustin GJS, Nelstrop AE, Bentzen SM, et al: Use of tumour markers in monitoring the course of ovarian cancer. Ann Oncol 10:21–27, 1999 (suppl 1)
4. Therasse P, Arbuck SG, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumours. J Natl Cancer Inst 92:205–216, 2000 5. WHO Handbook for Reporting Results of Cancer Treatment. World Health Organization, Geneva, offset publication 48, 1979 6. Miller AB, Hogestraeten B, Staquet M, et al: Reporting results of cancer treatment. Cancer 47:207–214, 1981[CrossRef][Medline] 7. Markman M: A proposal to use CA125 to evaluate activity of new neoplastic agents in ovarian cancer. Gynecol Oncol 51:297–298, 1993[CrossRef][Medline] 8. Ng LW, Homesley HD, Barrett RJ, et al: CA125 values predictive of clinical response during second-line chemotherapy for epithelial ovarian cancer. Am J Clin Oncol 12:106–109, 1989[Medline]
9. Rustin GJ, Nelstrop AE, McClean P, et al: Defining response of ovarian carcinoma to initial chemotherapy according to serum CA 125. J Clin Oncol 14:1545–1551, 1996
10. Eisenhauer EA, ten Bokkel Huinink WM, Swenerton KD: European-Canadian randomised trial of paclitaxel in relapsed ovarian cancer: High dose versus low dose and long versus short infusion. J Clin Oncol 12:2654–2666, 1994
11. Rustin GJS, Nelstrop AE, Bentzen SM, et al: Selection of active drugs for ovarian cancer based on CA 125 and standard response rates to phase II trials. J Clin Oncol 18:1733–1739, 2000 12. Davelaar EM, Bonfer JMG, Verstraeten RA, et al: A valid marker in ovarian carcinoma patients treated with paclitaxel. Cancer 78:118–127, 1996[CrossRef][Medline] 13. Pearl ML, Yashar CM, Johnstone CM, et al: Exponential regression of CA125 during salvage treatment for ovarian cancer with Taxol. Gynecol Oncol 53:339–343, 1994[CrossRef][Medline]
14. Bridgewater JA, Nelstrop AE, Rustin GJS, et al: Comparison of standard and CA 125 response criteria in patients with epithelial ovarian cancer treated with platinum or paclitaxel. J Clin Oncol 17:501–508, 1999
15. Rustin GJS, Nelstrop AE, Crawford M, et al: Phase II trial of oral altretamine for relapsed ovarian carcinoma: Evaluation of defining response by serum CA125. J Clin Oncol 15:172–176, 1997
16. Meyer T, Nelstrop AE, Mahmoudi M, et al: Weekly cisplatin and oral etoposide as treatment for relapsed epithelial ovarian cancer. Ann Oncol 12:1705–1709, 2001 17. Benedetti Panici P, Greggi S, Amoroso M, et al: A combination of platinum and tamoxifen in advanced ovarian cancer failing platinum-based chemotherapy: Results of a phase II study. Int J Gynecol Cancer 11:438–444, 2001[CrossRef][Medline] 18. Guastella JP, Vincent P, Rol A, et al: CA 125 evaluation of chemotherapy response in patients with recurrent ovarian cancer: Rustin criteria revisited. Proc Am Soc Clin Oncol 21:204a, 2002 (abstr 815) 19. Rustin GJS, Quinnell TG, Johnson J, et al: Trial of isotretinoin and calcitriol monitored by CA 125 in patients with ovarian cancer. Br J Cancer 74:1479–1481, 1996[Medline]
20. Rustin GJS, Nelstrop AE, Tuxen MK, et al: Defining progression of ovarian carcinoma during follow-up according to CA 125: A North Thames Ovary Group Study. Ann Oncol 7:361–364, 1996
21. Van der Burg ME, Lammes FB, Verweij J: The role of CA 125 in the early diagnosis of progressive disease in ovarian cancer. Ann Oncol 1:301–302, 1990
22. Vergote I, Rustin GJS, Eisenhauer EA, et al: New guidelines to evaluate the response to treatment in solid tumours (ovarian cancer). J Natl Cancer Inst 92:1534–1535, 2000
23. Krebs HB, Goplerud DR, Kilpatrick SJ, et al: The role of CA 125 as a tumour marker in ovarian cancer. Obstet Gynecol 67:473–477, 1986 24. Rustin GJ, Nelstrop A, Stilwell J, et al: Savings obtained by CA-125 measurements during therapy for ovarian carcinoma: The North Thames Ovary Group. Eur J Cancer 28:79–82, 1992
25. Rustin GJS, Marples M, Nelstrop AE, et al: Use of CA 125 to define progression of ovarian cancer in patients with persistently elevated levels. J Clin Oncol 19:4054–4057, 2001 Submitted January 29, 2003; accepted March 14, 2003. This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2003 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
