I N T E R V I E W  Kathy D Miller, MD Dr Miller is the Sheila D Ward Scholar of Medicine and Associate Professor of Medicine in the Department of Hematology/Oncology at Indiana University School of Medicine in Indianapolis, Indiana.

On the question of the use of a sequential single agent over combination chemotherapy for most patients with metastatic disease, it will be no surprise to most of you that I would strongly agree.

These are the reasons why I would agree that single-agent chemotherapy in a sequential fashion should be the preferred choice for the vast majority of our patients.

Slide 1

This question has been tackled in different ways. These three separate trial designs really don’t ask the same question — they ask two different questions. These first two trial designs ask, Is a particular drug beneficial in breast cancer?

The third design, which includes a sequential strategy versus a combination strategy, asks how to best use drugs that we already know or assume are active.

Slide 2

This is a trial that is familiar to you — the docetaxel/capecitabine trial with the doses and schedules that you see here. Only about 20 percent of patients crossed over, so this is not a trial that asked, How do we best use capecitabine in our patients? It’s a trial that asked, Is capecitabine beneficial?

Slide 3

It clearly is. Those patients who received capecitabine had a much higher response rate, better time to progression and improvement in overall survival.

Slide 4

There’s a similar trial design with gemcitabine with about 15 percent crossover. The question is the same: Is gemcitabine beneficial in this disease? The question is not, How do we best use gemcitabine in our patients?

Slide 5

The results were very similar with improvements in overall survival and response rates. In both of these trials, there was increased toxicity in the combination arm and no difference in overall survival. You could argue that the slight increase in response rates is discounted, in a quality-of-life sense, by the increase in toxicities.

Slide 6

There have been many other trials that have had similar designs. These trials go back to the 1970s, and there were only three that found improvement in overall survival. What is striking about all three of them is that they coincide with the introduction of a new therapy that we have all come to accept as one that is highly active and beneficial for at least some of our patients.

There are several trials that have looked at the strategy question rather than the drug question.

Slide 7

This is a trial from our European colleagues that looked not only at single drugs but also whether sequential single agents versus sequential combinations — keeping the single agents in those combinations — are beneficial. There were no improvements in any of the endpoints with the combination therapy regimen, and treatment-related toxicity and quality of life certainly favored the single agents.

Slide 8

The largest trial and perhaps the one that has been the hallmark for this particular question is George Sledge’s E1193 trial using doxorubicin and paclitaxel as either single agents or in combination, with a planned crossover to the other agent in patients randomized to the single-agent arms.

Slide 9

On this slide, I have tried to summarize all of the results that I think are important for this trial. As initial therapy, there were slightly greater response rates and about a two-month improvement in time to progression for the combination, but there was no difference in overall survival or quality of life. In the crossover results, response rates, time to progression and overall survival at the time of second-line therapy were also essentially equivalent. With these particular agents, we couldn’t define a superior
sequence.

Slide 10

There are several other trials, going back to the 1970s, which have used this particular trial design. The Baker trial actually included five chemotherapy agents, either given all together or in sequence. The results are markedly consistent. There was no difference in overall survival in any of these efforts.

Slide 11

We have to go back to what our therapy goals for metastatic disease are. This is what my patients tell me. This is not really what I say. My patients tell me they wish to live longer and better during that time, however long or short it might be.

Our clinical trials always tell us about response rates. They may tell us about duration of response or time to progression. Occasionally, they tell us about overall survival, though these timeto- event variables in uncontrolled trials are very hard to dissect and interpret.

And they often either don’t assess quality of life, or our quality-of-life tools are fairly crude and only tell us about large differences.

We have fairly good data that tell us that those clinical trial results, other than overall survival, really don’t translate very well into extending survival and improving or maintaining quality of life.

Slide 12

That is why, in my opinion, for most patients, the sequential single-agent approach is preferred — it gives us a variety of options.

From this year’s ASCO, several other trials are looking at whether the order of those agents makes a difference. We have yet to identify any particular order that, overall, gives patients greater benefit than a different order. So there are many options that allow us to individualize both treatment options and the order of treatment for our patients, based on what our patients bring to us.

For those patients we all talk about but rarely see who have such symptomatic, rapidly progressive disease that you think — if they don’t respond to the first treatment option, you’re not going to get another chance in four, six or eight weeks — a combination is certainly appropriate. I just don’t think that happens very often, certainly not in the first-line setting. Usually, when I see patients in that situation, they’ve already had much previous chemotherapy, so my options are already limited.

Slide 13

E2100 is based on a different hypothesis. It really is not asking a chemotherapy question. It’s based on our understanding that angiogenesis is important in breast cancer. There are many proangiogenic factors that are produced by breast cancer, but the vascular endothelial growth factor (VEGF) is one of the most common and potent ones.

We do know that bevacizumab is an agent that has some activity in breast cancer but is not particularly great in refractory patients, with only about a nine percent response rate as monotherapy. This was a slight improvement in the response rate, but progression- free or overall survival did not improve in those patients.

The hypothesis that E2100 was designed to test was based on the biology that using this sort of agent earlier in the course of the disease would give greater activity.

Slide 14

This trial randomized patients to paclitaxel with or without bevacizumab. A fairly common four-week treatment cycle was used, with a schedule of weekly paclitaxel for three weeks and then one week off. Patients were eligible if they had metastatic or locally recurrent disease that was not curable with any local therapy approach.

Slide 15

Essentially, this is a trial for patients with HER2- negative disease, because the trial schema does not include trastuzumab. Patients with HER2- positive disease were only allowed into this trial if they had received previous trastuzumab therapy. Note that patients couldn’t have received previous chemotherapy for metastatic disease. We thought this could apply to someone who was in one of the adjuvant trastuzumab studies and progressed fairly soon after her trastuzumab therapy or someone who received trastuzumab monotherapy for metastatic disease and then wanted to enter this trial. In reality, there were only a handful of patients with HER2-positive disease, and they all came from South Africa, where trastuzumab is not available. Otherwise, patients had to be healthy and,
significantly, patients with CNS metastases
were excluded. We did require screening before entry.

Slide 16

We had fairly modest goals for this trial. We were hoping to find an improvement in progression-free survival from six to eight months in the paclitaxel-alone arm. That would have required 650 patients. There were planned interim analyses that used very stringent O’Brien- Fleming boundaries to release the results early, either because there was a clear benefit or a definite lack of benefit.

Slide 17

The results that we have now are from the first planned interim analysis. The trial accrued 715 eligible patients over about two and a half years. The data is now current through February 9th of this year and includes 355 events, most of which are progression. The 64 patients who were censored at death without progression had progressive disease in most of the situations — but there was not adequate imaging or documentation to meet RECIST criteria for disease progression.

Slide 18

Patients were well matched. They were the usual age for a breast cancer trial — in their mid-fifties, though there were patients enrolled who were in their mid-eighties.

Two thirds of the patients in this trial had received adjuvant chemotherapy, which is much higher than previous metastatic trials. That tells us that how we treat these patients has shifted, so you can’t easily compare results of first-line trials done today to results of first-line trials done 10 years ago.

The response data are the first encouraging bit of news, and this is my evidence that we can evaluate response in patients without measurable disease quite well.

Slide 19

E2100 did not require measurable disease because progression-free survival was the primary endpoint. As the PI, I didn’t care if you could tell me if the patient was responding, but I was fairly certain that good clinicians would know when their patients were progressing, and that was the endpoint we were interested in.

If you look at the subset of about three quarters of patients who had measurable disease, the estimates of response were very similar.

Slide 20

Progression-free survival was clearly improved by much greater than we had expected. Our estimates of a six-month progression-free survival for patients receiving paclitaxel alone was 6.1 months, with an improvement to 10.97 months or a 51 percent decrease in the risk of progression for patients receiving the combination.

Slide 21

The overall survival data, I must caution you, are early, so I cannot tell you what the median survivals are. They’re just now being reached, and that part of the curve is very unstable. What is not likely to change, however, is that there is a survival advantage in this trial, with about a 33 percent decrease in the risk of death in patients receiving the combination therapy.

Slide 22

Increase in toxicity was minimal. We’ve known about the bevacizumabassociated toxicities, and this is about what we expected. About 15 percent of patients developed hypertension and needed treatment, and there was a small proportion of patients with significant proteinuria without any other signs of renal dysfunction. In this relatively healthy breast cancer patient population, thromboembolic events and bleeding were not a problem, and they were not different in the two treatment groups.

Slide 23

There’s only one important difference in the chemotherapy-associated toxicity, which is a slight increase in the risk of Grade III neuropathy. It could be that there’s an interaction between the drugs that increases neuropathy. What’s more likely is, since paclitaxel-related neuropathy is both duration of treatment and dose related, patients were responding to therapy for a much longer time and were exposed to a greater duration of taxane therapy.

Slide 24

Overall, we were delighted to be able to finally say that the addition of bevacizumab in patients with newly diagnosed metastatic breast cancer — or at least newly requiring chemotherapy for their metastatic breast cancer — improves the progression-free survival and the objective response rate. It also appears to improve overall survival, though we’ll need further follow-up to know the magnitude of improvement in overall survival.

We are continuing to look for ways to identify those patients who are most likely to benefit from this approach. These patients were not positively selected for any molecular feature because, at this point, we don’t know how to select them.

Slide 25

We’re also looking forward to exploring bevacizumab in the adjuvant setting and will very soon be activating an adjuvant pilot trial in the Eastern Cooperative Oncology Group. This trial will hopefully then be followed, in very quick succession, with a full adjuvant trial to try and look for improvements in disease-free and overall survival.

I would agree with this second question as well. For most patients, weekly paclitaxel or capecitabine in combination with bevacizumab provides the most effective first-line chemotherapy. The trial I’ve just shown you, E2100, clearly finds significant improvements in all of those response parameters for incorporating bevacizumab with weekly paclitaxel.

It’s important to think about the eligibility criteria and which of our patients with metastatic disease would not have been eligible for this trial. We don’t yet know about the role of bevacizumab or the safety of bevacizumab in combination with trastuzumab regimens in patients who are HER2-positive. At this point, I would not add bevacizumab to those patients’ treatment. We also excluded those patients who had received adjuvant taxane therapy and progressed within 12 months. I don’t think it would be reasonable to re-treat those patients with paclitaxel, even in combination with bevacizumab. So for those patients who’ve recently had an adjuvant taxane and are progressing, we really don’t have good data.

From our previous capecitabine trials, I can tell you that those patients who receive an anthracycline and taxane and progress within 12 months do miserably, and I suspect they would do miserably no matter what we did for them. If you are interested, it’s reasonable to use bevacizumab in those patients, but we don’t have randomized trial data to support that at this point.

Slide 26

We have safety and response data with bevacizumab in combination with capecitabine in exactly that patient population. Response rates increased slightly, but most of those extra responses were fairly short lived.

Slide 27

There are also safety and response data in patients who were pretreated with vinorelbine. Either one of those would be regimens for which at least there is some evidence, although more on safety than on efficacy.

Slide 28

Finally, there is another trial that will be starting very soon called XCaliBr. This trial will look at newly diagnosed patients, essentially the same group as in the E2100 trial — needing chemotherapy, but using capecitabine in combination with bevacizumab. This trial allows but does not require patients to continue bevacizumab after initial progression, either with vinorelbine or paclitaxel, at the patients’ and investigators’ choice.

This is a fairly small Phase II trial with only 92 patients, so it certainly will not be definitive. Randomization to continuing bevacizumab or not is not included. That is an open question we need to address quickly. But the trial will, at least, give us some comparative data on response rates and time-toevent variables, with a different chemotherapy regimen. That will be particularly helpful as the taxanes are used more and more in the adjuvant setting and as we hope to move to more oral therapies and therapies that don’t include alopecia and some of those toxicities for our patients with metastatic disease.

SELECT PUBLICATIONS

* Presented at Research To Practice Breast Cancer Update CME Forum, Los Angeles, California, May 21, 2005. The enclosed graphics from the presentation are included in the PowerPoint files on CD 3. Please see page 1 for additional instructions.