ECOG-E2100: Phase III randomized trial of paclitaxel with or without bevacizumab as first-line chemotherapy for metastatic disease

Background

ECOG-E2100 (E2100) was based on preclinical work from our laboratory and a number of others. In our group, Dr Chris Sweeney conducted a study in which he evaluated the ability of the taxanes to interact with bevacizumab against endothelial cells and found that the taxanes are good anti-angiogenic agents themselves.

We were able to show synergistic activity between bevacizumab and taxanes against endothelial cells, and we reasoned that the best way to use those data in the clinic was to administer a weekly taxane along with bevacizumab. We used paclitaxel because our experience — and I think that of many — is that weekly paclitaxel is better tolerated than weekly docetaxel.

Trial design

E2100 was a randomized trial of weekly paclitaxel with or without every two-week bevacizumab (Miller 2005a; [1.1]). Patients could receive up to 18 courses of therapy. Each course consisted of a four-week cycle of therapy in which patients received three weeks of paclitaxel followed by a one-week rest. If the patients became tired of chemotherapy, they could discontinue paclitaxel and continue bevacizumab.

The primary endpoint was progression-free survival, and secondary endpoints included overall survival and the usual toxicity parameters. As a result of the previous toxicity seen in the lung cancer trial, we had very stringent criteria for discontinuing this trial if we saw an excess number of patients developing Grade IV hypertension or bleeding (Miller 2005a). When the trial was initiated, the National Cancer Institute had significant concerns about patient safety as a result of the initial experience with bevacizumab in lung cancer. Fortunately, early analyses demonstrated that was not an issue.

Safety

The side effects were relatively minimal. Predominantly, we saw mild to moderate increases in blood pressure, which is readily handled from a clinical standpoint. Of course, we’ll have to be careful with the hypertension as we move bevacizumab into the adjuvant setting. We also saw a low incidence of serious bleeding. Overall, bevacizumab was a nontoxic addition to chemotherapy (Miller 2005a; [1.2]).

To a certain extent, I wonder whether many of the toxicities observed with bevacizumab are, in fact, relatively tumor specific. For instance, the bowel perforations in patients with colorectal cancer and the bleeding problems in patients with lung cancer have not been an issue in patients with breast cancer.

Evaluating the literature, one could say that bevacizumab increases thrombotic events. In E2100, we saw a low level of thrombotic events (Miller 2005a; [1.2]). I suspect you can’t do a lot to blood vessels without altering the risk of thrombotic episodes. Going along with that is the increase in migraines seen in almost every trial conducted with bevacizumab.

These are classic migraines, which in my experience tend to occur more commonly in patients who have a prior history of migraines. In the initial Phase I/II breast cancer trial in which we dose escalated, migraines were the doselimiting toxicity once we reached a dose of 20 mg/kg (Cobleigh 2003).

In E2100, we haven’t analyzed whether age or pre-existing risk factors influence the incidence of thrombosis. Patients who enroll in large cooperative group trials tend to be younger than the general population of patients with breast cancer. For instance, in E2100, the median age was approximately 55 (Miller 2005a), which is close to a decade younger than the age of the average patient with breast cancer in the United States. So it’s entirely possible that in a more elderly group, we’ll see more toxicity.

Efficacy

Progression-free survival went from just over six months for paclitaxel alone to almost 11 months for paclitaxel plus bevacizumab — a 4.5- to five-month improvement in progression-free survival (Miller 2005a; [1.3]). If we were to evaluate the randomized trials in metastatic breast cancer conducted over the past 20 years outside of HER2-positive disease, I would say that I cannot remember any trial showing this significant of an improvement in progression-free survival.

Effects on survival

It’s still too early to evaluate overall survival. The p-value is statistically significant for overall survival, but I see a certain amount of choppiness in the overall survival curves (Miller 2005a). We don’t have enough events and haven’t followed patients long enough. In six to 12 months, we’ll have a lot more events and the curves will be more believable. We’re encouraged because the early data suggest an overall survival advantage.

Implications of the results from E2100

Bevacizumab ought to be considered for use along with taxane-based therapy as front-line therapy for patients with metastatic breast cancer. I certainly would not argue with those who suggest we need safety data for bevacizumab in combination with docetaxel or nanoparticle albumin-bound (nab) paclitaxel. However, much of our preclinical testing was with docetaxel, and I would expect docetaxel to work well with bevacizumab.

I would not be surprised if nanoparticle taxane therapy would also work well. In fact, the nanoparticle taxanes have — as a possible mechanism of action — an effect on endothelial cells. We might see some synergistic activity there also. I’m not aware of any safety data for nab paclitaxel in combination with bevacizumab, but I suspect it would be safe. I would not have a problem with someone using the combination, and I would not expect any unusual toxicity.

Dosing of bevacizumab

The dose we choose is based on our Phase I/II dosing in patients with breast cancer (Cobleigh 2003). Colon and lung cancer had different paths based on randomized Phase II trials conducted in those diseases. I don’t know what represents the right dose of bevacizumab. We know 20 mg/kg is too much because of the dose-limiting toxicity of migraines (Cobleigh 2003).

In the Phase I trials, once we got past approximately one mg/kg, all circulating, free VEGF was bound. So somewhere in between one and 20 mg/kg is the right dose. We used 10 mg/kg in E2100. In the Phase I/II breast cancer trial, we looked at doses of three mg/kg, 10 mg/kg and 20 mg/kg. Even though the numbers were small, we had a sense that when we went from three to 10 mg/kg, the responses were more brisk, and we saw relatively more patient benefit (Cobleigh 2003).

Trials combining bevacizumab and trastuzumab

When we launched E2100, literally no data existed concerning the use of combination monoclonal antibody therapy in patients with breast cancer for any antibody, let alone bevacizumab and trastuzumab. Since that time, Mark Pegram and his colleagues at UCLA, based on some wonderful preclinical work, combined bevacizumab and trastuzumab in a Phase I trial that was recently reported (Pegram 2004), and they now have an ongoing Phase II trial.

The Phase I trial enrolled a total of nine patients because the combination was fairly nontoxic, and the trial zipped through the dose levels. Five of the nine patients receiving bevacizumab and trastuzumab had an objective response, and other patients had prolonged stabilization of their disease (Pegram 2004). You don’t want to make too much of a Phase I trial or a trial with nine patients, but for a treatment with no chemotherapy involved, this is a respectable response rate. If bevacizumab plus trastuzumab plays out in the larger Phase II trial, it will be a felicitous combination.

Combining bevacizumab with a taxane and capecitabine

It would be reasonable to evaluate a combination of a taxane, capecitabine and bevacizumab despite the negative second-line trial of capecitabine and bevacizumab (Miller 2005b), which was conducted in a population of patients with very advanced disease.

I am the principal investigator for a new trial called XCaliBr, which will evaluate the combination of bevacizumab and capecitabine as front-line therapy for patients with metastatic breast cancer who have relapsed after an adjuvant anthracycline/taxane combination. We don’t have data about the use of capecitabine plus bevacizumab as first-line therapy, and we have negative second-line data (Miller 2005b). It’s incumbent on us to generate some positive front-line data before we say it’s the right thing to do.

Adjuvant bevacizumab trials

An ECOG pilot trial of adjuvant bevacizumab, which will be primarily evaluating safety issues, will involve over 200 patients and will open within the next few months. Our belief is that given adequate safety data in the adjuvant setting — which we hope to have within 12 to 18 months — we’ll be able to go directly to a large Phase III trial comparing chemotherapy to chemotherapy plus bevacizumab.

Of course, many questions can be asked in the adjuvant setting with bevacizumab (eg, which combination chemotherapy or what duration of therapy), which may require more than one trial. We will also need more than one trial because we’ll have to evaluate both HER2-negative and HER2-positive disease.

Clinical impact of adjuvant trastuzumab trial data

As a result of the data presented at ASCO in 2005, trastuzumab became a standard of care in the adjuvant setting for HER2-positive breast cancer. We saw a stunning validation of the biology of HER2 and the concept that we could diminish the likelihood of recurrence and improve overall survival through the use of targeted therapy. This validates 15 years of preclinical and clinical research, from Slamon’s initial observation in the late 1980s that HER2 was a bad actor in breast cancer (Slamon 1987) to the pivotal metastatic trial led by Slamon (Slamon 2001) and now the adjuvant trial data (Piccart-Gebhart 2005; Romond 2005). We have consistently seen that when HER2 is overexpressed or amplified, it markedly increases a patient’s risk of early relapse.

In the HERA trial, we saw that by two years after randomization, one quarter of the patients in the control arm had relapsed. In the joint analysis of NCCTGN9831 and NSABP-B-31, around 25 percent had relapsed by approximately three years (Piccart-Gebhart 2005; Romond 2005). This is a bad disease, and partly because of that, we see a high event rate early in these trials. A striking benefit was seen with trastuzumab, including survival with a median follow-up of just two years. That is unprecedented in any adjuvant trial.

It’s interesting to imagine what the impact of the estrogen receptor trials would have been if we had enrolled 3,000 patients on those studies two or three decades ago. The data probably would have been similar to the adjuvant trastuzumab trial data. The message is that if we understand biology and target it appropriately, we obtain a great result, whereas when we use relatively nonspecific therapies, we can tweak them — changing dose duration, dose density and dose intensity — and obtain slightly better results, but we’ll never achieve the revolutionary results that we saw in the adjuvant trastuzumab trials.

Distant disease recurrence reduction with adjuvant trastuzumab

In the joint analysis of NCCTG-N9831 and NSABP-B-31, the hazard rates for distant disease recurrence in patients who received trastuzumab appeared to improve with time. It’s still early to analyze these data because few patients in either trial are four years out; however, the distant disease-free survival curve appears to plateau in the trastuzumab arm. If that’s the case, it’s astonishing. We’ve never seen a true plateau in any adjuvant trial. When we examine diseasefree survival curves like this, we need to ignore a fair amount of the right side of the curve because there are so few numbers, but if that is maintained it will be an exciting finding.

Concurrent or sequential trastuzumab administration with chemotherapy

In the HERA trial, all the patients received trastuzumab after rather than concurrent with chemotherapy, and those data were positive with an impressive 45 percent reduction in hazard rate (Piccart-Gebhart 2005). On the other hand, in the Intergroup trial, it appears that concurrent therapy is superior to the sequential schedule (Perez 2005; [1.4]). These are different data sets, and both trials have a short median follow-up and a relatively small number of events, so we shouldn’t make too much of this yet.

Concurrent therapy after the anthracycline is probably better. I base that belief on the results of the pivotal trials and on a large body of preclinical data that suggest trastuzumab is a good amplifier of chemotherapy-induced apoptosis. Also, considering how rapidly the efficacy curves separate in the joint analysis data, it almost makes one want to start trastuzumab about 10 seconds after a core biopsy is obtained.

Assessment of HER2 status

I believe every patient with primary breast cancer should be tested for HER2 by FISH, although not everyone agrees with me. The NCCTG-N9831 trial required tumors to be either FISH-positive or IHC 3+ with central review; however, we know that even with central review of IHC 3+ results, a certain number of tumors were found to be HER2-negative when a FISH assay was performed. If we analyzed the data for only the FISH-positive population in this study, the results might be even more impressive. The BCIRG 006 trial accepted only FISH-positive cases, so it’s possible that the results of that trial will be even more positive.

Clearly, many hospitals report inaccurate FISH and IHC results. We know this because of the analyses done by the NSABP and NCCTG, in which comparisons of central lab testing with local hospital testing demonstrated a shocking degree of difference in some cases (Perez 2005; [1.5]). It’s frightening to think that some patients will lose a chance of being cured of breast cancer because the laboratory results were wrong. We need to develop strong national or international standards for testing HER2. We don’t have standardization for testing estrogen receptor either.

The FISH positivity rate in patients whose tumors are reported as IHC 0 and 1+ is low, but it’s real. In the data generated by Genentech as part of the initial suite of trials, the rates for both 0 and 1+ tumors being FISH-positive were under 10 percent — I believe it was seven percent for 1+ and two percent for 0. That sounds like a low number, but given that most tumors are IHC 0 or 1+ in a general population, seven percent represents a fair number of untreated patients. We need to revisit the issue of whether to retest IHC scores of 0 and 1+ now that it appears adjuvant treatment may lead to a cure in patients with HER2-positive disease.

Combining adjuvant trastuzumab with chemotherapy regimens

The carboplatin/docetaxel/trastuzumab regimen will be an important issue in the future. The first planned analysis of the BCIRG 006 trial could take place within the next few months. This is a crucial trial because the third arm — carboplatin/docetaxel plus trastuzumab — is compared to two more or less standard arms seen in the joint analysis. The trial will provide important data regarding cardiotoxicity versus efficacy with these regimens.

Dr Slamon shared some interesting cardiac data from an analysis of approximately 3,000 patients on the BCIRG 006 trial. A lower incidence of congestive heart failure (CHF) and fewer declines in left ventricular ejection fraction (LVEF) were seen when trastuzumab was given with carboplatin/docetaxel versus in proximity to doxorubicin (1.6). This raises the question of whether we need an anthracycline at all with trastuzumab-based therapy.

If carboplatin/docetaxel/trastuzumab has similar efficacy to AC followed by docetaxel/trastuzumab, it would probably become the de facto standard in a short period of time. Also, Robert’s study comparing paclitaxel/trastuzumab with or without carboplatin as front-line therapy for metastatic breast cancer resulted in a doubling in time to progression for patients who received all three agents in a true HER2-positive population (Robert 2002). That was impressive, and if we see anything like it in the adjuvant setting, it’s also likely to be a good combination.

Role of delayed adjuvant trastuzumab

The HERA trial suggests administering trastuzumab after chemotherapy may be beneficial, so the question becomes, how long after chemotherapy will it be beneficial? In the case of estrogen receptors, we have two European randomized trials that evaluated the late use of tamoxifen in patients with estrogen receptorpositive breast cancer, and both were positive. Will we see a similar benefit with delayed adjuvant trastuzumab? It’s a reasonable and important question, particularly for those patients in the control arms of N9831 and B-31 who are 18 months out from treatment. I’m not going to be dogmatic about this, but I do believe it’s reasonable to discuss the option of trastuzumab with such patients.

Select publications

Dr Sledge is a Professor of Medicine and Ballve-Lantero Professor of Oncology at the Indiana University Medical Center in Indianapolis, Indiana.