DR. BIRRER: Our next speaker is Bob Bast. The way we designed this session was to have Jeff do more of a genomic biomarker review, and then Dr. Bast, because of his seminal work with CA 125 and his interest in biomarkers, would do all of the rest. With that, Bob is going to come up and give that a try.

TOP

Currently, serum biomarkers are used, as you know, in a number of different ways in the management of ovarian cancers, certainly monitoring response to treatment, distinguishing malignant from malignant from benign pelvic masses -- although this has been applied probably more in the United Kingdom than the United States -- predicting prognosis, and potentially in early detection, which has been of great interest to the SPOREs over the last several years.

TOP

If you look toward the future, I think there are possibilities for using serum biomarkers more effectively in clinical trial design: certainly as surrogate markers for response in phase II trials, in monitoring cytostatic drugs in phase II trials, and finally, perhaps, in eliminating the least effective arm in interim analysis in multi-armed randomized phase III studies.

TOP

Clearly, we are facing a crisis, in that we have an increasing number of drugs available for evaluation of ovarian cancer. Approximately the same number of patients is eligible for trials. Yet a significant fraction of women who could participate in clinical trials, perhaps as high at 50 percent who could be in phase II trials, lack measurable disease by RECIST criteria.

There is an increasing body of evidence that serum markers can accurately reflect response or lack of response to drugs.

TOP

I think there is a real opportunity to use CA 125 and perhaps other markers as surrogates in phase II trial design.

There is recent evidence from Ken Lloyd and Tim O'Brien that have really characterized CA 125 at a molecular level.

It turns out to be a mucin designated MUC-16, a large molecule of about a million daltons, heavily glycosylated with 40 or more repeats in the sequence.

This is cleaved near the membrane, and results in the shedding of a substantial amount of antigen into the serum.

TOP

CA 125 could provide a surrogate marker for response in phase II trials. Gordon Ruston has really made a career of looking at how CA 125 tracks disease in phase II settings, and a 50 or 75 percent decrease in CA 125 has correlated with response rates in some 19 phase II trials of 14 different cytotoxic drugs with approximately 1,400 patients.

Gordon, I think, last summarized this most effectively in a JCO article in 2001. If you used CA 125 at least initially in phase II trials, you possibly could double accrual.

You could discontinue trials with poor response, and this is the crux. Are companies and are investigators willing to stop working with the drug based on a progressively rising CA 125 in a critical number of patients in a phase II context.

If you need RECIST criteria for FDA approval, you could go on to gather more patients, where the CA 125 data was promising, to fulfill RECIST criteria with a larger phase II trial.

TOP

A second area where serum tumor markers are already being applied is in monitoring new cytostatic drugs. As we are all aware, many of the targeted therapies are cytostatic and stabilize disease, rather than cause partial or complete responses.

Effective drugs could arrest a rising CA 125 and recurrent disease and measure the decreased slope of the use of doubling time of the CA 125 as progression.

TOP

The GOG, as I think all the GOG members are aware, have already taken this track in looking at biochemical recurrences of ovarian carcinoma, using criteria where patients have been normal prior to treatment, and subsequently rose to exceed twice the upper limits of normal and, if that is still less than 100 units of CA 125, to confirm that with a second value within four weeks. Then patients who have greater than 100 units are entered outright.

TOP

There is currently a large phase III trial that is underway involving thalidomide versus tamoxifen in patients with biochemical recurrence.

TOP

Looking toward the future, it might be possible to have even more sophisticated analysis. Steve Skates has developed algorithms to look at rising CA 125 as an indicator of early-stage disease for screening.

You could reverse those algorithms and look at the trend of rising CA 125 being blunted or leveling off using change point analyses.

Also, you could imagine using this not only for phase III trials, but also for sequential phase II trials, not unlike we do for conventional criteria in the GOG at the present time, applying biomarker trends for phase II trials screening new drugs with historical benchmarks for what a leveling off of CA 125 would reflect.

TOP

A third application in serum biomarkers to clinical trial designs may be in phase III. We know now, from some 20 trials, that CA 125 half-life has been shown to be an important prognostic variable.

Indeed, it compares pretty favorably, although it doesn't give you exactly the same information as second look operations, in at least a couple of those papers.

However, it is not sufficiently precise to individualize therapy, but it has been highly predictive for groups of patients.

TOP

These are some early data from our Duke experience. The real half-life of CA 125 is about seven to 14 days.

If the apparent half-life reflecting residual tumor is greater than 20 days, with the yellow lines, those patients do substantially worse than those who have a shorter half life.

TOP

One possibility would be to eliminate the least effective arm at interim analysis and multi-arm randomized trial based on the rate of CA 125 decline in each arm.

One could then shift accrual to the more favorable arms, abandoning that arm based on biomarker analysis, and you could conserve patient resources, and presumably complete these trials much more rapidly.

TOP

There are probably also a number of applications for individual tissue biomarkers in the management of ovarian cancer.

Although there is probably no single biomarker that is going to work, one could imagine, instead of a whole array analysis, using a small panel of biomarkers, not unlike the analysis in breast cancer where a 21-gene panel has provided some useful information in sorting out those patients with ER positive disease, who are likely to relapse after conventional hormonal therapy, who might need chemotherapy.

You also ideally could individualize therapy with drugs in recurrent disease, and certainly stratify clinical trials for resistance markers.

TOP

I think a couple of thoughts with regard to platinum compounds, we know that, with optimal levels of platinum, either carboplatin or cisplatinum, there is about a 70 percent response rate in previously untreated ovarian cancer patients.

I think it is fair to say, in our thinking about how we apply either arrays or individual markers to the problem of prognosis, you have got to have a very high negative predictive value, arguably greater than 95 percent, in order to forego treatment.

If we are going to abandon platinum for a patient with ovarian cancer, we have got to be pretty sure that that patient is not going to respond, not unlike what we do with hormonal therapy for breast cancer where, by and large, ER and PR gives about a 95 percent predictive value.

With regard to taxanes, I think there is a different setting. From GOG-132, we know that, with an optimum or maximally tolerated dose of paclitaxel, there is only about a 50 percent response rate, maybe a little less.

Also, from both laboratory and clinical studies, it appears that taxanes and platinum compounds that are used in patients who we treat at the present time are not synergistic, arbitrarily with carboplatin and Taxol, aren't benefiting from the Taxol (getting neuropathy, immunosuppression, etc.). This is about 50% of all patients.

They are getting neuropathy, they are getting myelosuppression. Probably you could use a higher dose of carboplatin or, better yet, some other agent to which they might otherwise respond.

I think the problem, in terms of individualizing therapy, though, is that now, as these drugs have been developed and are used in combination, it is really difficult to study them as individual agents.

Hani Gabra, who is with us today, in the SCOTROC trial, I think is doing a very important study with single agent carboplatin, and very carefully collecting the tissues and serum from those individuals, so that it may well be possible to sort out biomarkers for platinum alone in a prospective trial with previously untreated patients.

TOP

Another trial that we are undertaking at M.D. Anderson with David Gershenson is to study previously untreated patients in a therapeutic window of approximately nine weeks before six cycles of conventional carboplatin and paclitaxel in patients who are suboptimally cytoreduced.

In each of these individuals, because of the cytoreduction, we have got frozen tumor tissue that can be used for array analysis or for individual marker panels.

We are looking at the response in that window, either using the rate of decline of CA 125, or the use of imaging when there is still enough disease left behind to image.

We are looking forward to correlating marker expression with single agent activity, but clearly, we are accruing only very slowly to this trial and this, I think, is one opportunity for real collaboration with a drug that is generally available, simply a different trial design.

TOP

Clearly, there has been a tremendous amount of work looking for biomarkers individually, to predict response to individual drugs in ovarian cancer, and this is a very partial list with lots of omissions.

Certainly, from the work of Fox Chase and Steve Howell and many others, p53, ERCC-1, lack of copper transporters, XIAP, have all been related, although not precisely enough yet to be useful for individualization of therapy to platinum resistance.

Biomarkers are also available for taxane resistance. Certainly Mike Seiden's group and others have been leaders in this area, looking at a number of different markers, including MDR-1 tubular mutations, HR-2 and survivin.

TOP

Looking forward, I think we are facing a number of challenges, and they are not so much technological, it is not so much the laboratory side of this. I think it is really the clinical side that is limiting our development of biomarkers.

Certainly looking at trial design, where we can get optimal information about single agents of drugs, as well as their use in combination, also to obtain large numbers of patients with tissue, body fluids and single agent response rates, not dozens of patients, but hundreds of patients for these trials, to do them right.

Also, I think it is going to be increasingly important to have the tissue banks in proximity to patients with the support for the people to pick up the tissues and freeze them promptly.

This is critical because, particularly with the GOG, some institutions have this readily available. Others don't.

There is also, I think, a real need, as Jeff mentioned before, that we develop confirmatory trials promptly, that predict response and/or that individualize treatment, and that we design these appropriately.

So, some thoughts on the use of biomarkers, either individually or in panels for several different applications.

TOP