BOND-2 study: Cetuximab/bevacizumab plus or minus irinotecan in patients with metastatic colorectal cancer

Rationale and design

In the BOND-1 trial, patients with irinotecan-refractory metastatic colorectal cancer were treated with cetuximab alone or in combination with irinotecan (Cunningham 2003). In the BOND-2 trial, we added bevacizumab to both arms (Saltz 2005). We wanted to know whether adding bevacizumab improved efficacy and whether it was safe and tolerable to give both monoclonal antibodies at the same time.

We had planned to accrue 75 patients in each arm; however, shortly after opening the study in December 2003, both agents became commercially available and within a month or so, the rapid rate of accrual slowed considerably. By the end of 2004, the number of eligible patients — patients who were bevacizumab naïve — was virtually nil, and we revised the statistical goals. The current data set is 39 patients who received irinotecan, cetuximab and bevacizumab and 35 patients who received cetuximab and bevacizumab. We anticipate that will be close to the final accrual total.

Efficacy data

The combination of cetuximab, bevacizumab and irinotecan resulted in a response rate of 38 percent, and the time to progression was 8.5 months. The historical reference points in two previous trials of cetuximab and irinotecan without bevacizumab was a response rate of 23 percent and a time to progression of approximately four months (Saltz 2001, Cunningham 2004; [1.1]).

What I found even more interesting is the response rate with the two antibodies. In the BOND-2 trial, the response rate was 23 percent, and in the three historical references for cetuximab alone, the response rates ranged from nine to 12 percent (Saltz 2002, Cunningham 2004, Lenz 2004). In the prior studies, the time to tumor progression for single-agent cetuximab averaged 1.5 months, whereas in the BOND-2 trial the median time to tumor progression with the two antibodies at this analysis was 6.9 months.

This was a randomized Phase II study, and the comparison of the two arms was not the primary statistical hypothesis. We have to be very careful about interpreting a small study with a historical control; however, it’s intriguing that the three-drug combination seems to show a significant increase in both response rate and time to tumor progression.

Toxicity data

We saw no clear evidence of synergistic toxicity in the BOND-2 trial. The toxicities seen were essentially those of single-agent cetuximab or bevacizumab, such as skin rash and hypertension. We did see some instances in which it was difficult to discern whether we were seeing side effects or simply advancing cancer.

The incidence of these events was consistent with what has been reported in previous trials, so I don’t believe they indicate a synergism of toxicities. However, we have less than 80 patients, and, as we gain more experience with the combination, we may begin to see some emerging toxicities.

Implications of the BOND-2 study in clinical practice and future research

Data from the BOND-2 trial raise some interesting questions: What do we do with an interesting study that treats a population that no longer exists, and how do we extrapolate the data to clinical practice today?

This trial doesn’t tells us whether to add bevacizumab to cetuximab-based therapy in patients who have already received bevacizumab, but our next study will basically repeat this trial in patients who have failed bevacizumab.

Until we have that data, I do not advocate routinely adding bevacizumab to cetuximab in patients who have previously received bevacizumab. However, if a patient is bevacizumab naïve, I believe these data support adding bevacizumab to cetuximab in a salvage setting.

While very few circumstances exist where the BOND-2 trial should change routine practice, it does provide important safety and pilot data for moving these two antibodies to front-line trials.

In a proposed design for a new Intergroup study, patients will be allowed to receive either FOLFOX or FOLFIRI at the physician’s discretion. They will then be randomly assigned to receive bevacizumab, cetuximab or both in addition to the chemotherapy. In addition, we are considering several other constructs to evaluate the double-antibody approach as front-line therapy.

EGFR staining and response to cetuximab

We just published an article in the Journal of Clinical Oncology that reports activity with cetuximab in colorectal cancer in tumors that do not express the EGFR by immunohistochemistry (IHC; [Chung 2005]). We reviewed charts of patients treated for colorectal cancer with cetuximab at Memorial Sloan-Kettering in a nonprotocol setting and found 16 patients with documented EGFR-negative tumors.

A reference pathologist confirmed that these tumors were negative, and then a reference radiologist reviewed the patients’ scans prior to and during cetuximab treatment. It was confirmed that four patients had major objective responses and two had minor regressions on cetuximab, so a fair amount of antitumor activity was confirmed in these EGFR-negative tumors.

These are very compelling data. We all wanted to believe that EGFR would be an important prognostic indicator, but our technology for assessing EGFR expression is flawed. We generally use the primary tumor as the basis for the EGFR status of the metastasis, but that appears to be inaccurate. Data shows that EGFR degrades over time.

In addition, EGFR staining is very sensitive to the type of fixative used on the tissue. It appears that EGFR exists in two conformations — tethered and untethered. The antibodies we use do not discriminate between these conformations, but only the untethered has biologic activity in terms of signal transduction, and it represents a very small percentage of the total.

We know that blocking the EGFR alters regulation of VEGF expression, so scientific reasons exist to combine bevacizumab and cetuximab. However, EGFR staining has no prognostic significance, so the BOND-2 study did not require staining for entry. When we examined the original ImClone and BOND study, we saw exactly the same activity level regardless of whether the staining was very weak or very strong.

Putting all this together, I believe EGFR staining should not be permitted in standard practice. I find it a waste of money, and it’s worrisome that physicians might rebiopsy a patient just to obtain this material. At this time, no clinical decision should be made on the basis of EGFR staining. Specifically, no patient should be excluded from a therapy — cetuximab or otherwise — simply because their IHC staining for EGFR is negative and, just as importantly, no patient should be treated with these agents simply because the tumor is strongly EGFR positive.

This doesn’t mean cetuximab isn’t an EGFR-specific targeted agent — it almost certainly is. The fact is that cetuximab does bind to the epidermal growth factor receptor. The question is whether quantitating EGFR by IHC can give us any handle on cetuximab activity, and the answer is no.

These EGFR-negative data are very fresh and have enormous implications. They confirm what virtually every GI academic oncologist has known for a long time, which is that EGFR staining is a sham. It is wrong to use it for decision-making, and it is unethical to exclude patients from treatment on the basis of it.

Impact of BOND-2 and ECOG-E3200 data on clinical practice

When Hurwitz presented the data on front-line IFL plus bevacizumab, I chose a fairly broad interpretation, as did many other oncologists, and decided it indicated that bevacizumab contributed to the activity of front-line chemotherapy (Hurwitz 2004). I extrapolated that to second-line therapy in my practice in bevacizumab naïve patients. Indeed, E3200 data confirmed that adding bevacizumab to FOLFOX in this patient population was beneficial (Mitchell 2005). The BOND-2 data builds on that and showed that adding bevacizumab to cetuximab-based therapy improves activity.

The E3200 study and BOND-2 study both show the utility of adding bevacizumab to a standard therapy and, at the same time, they’re both treating a population that no longer exists — bevacizumab naïve patients. We need to investigate what will happen in patients who have been exposed to one of the drugs in advance, because one of the major questions regarding bevacizumab is whether it should be continued forever. Some very intelligent thought leaders feel that bevacizumab should be continued with sequential regimens in colorectal cancer. While I believe that’s an interesting hypothesis, until it’s tested it remains a hypothesis.

I am concerned that physicians will extrapolate the E3200 data to justify continuing bevacizumab after progression. They may think, “I used bevacizumab at 5 mg/kg front line, so now I’ll continue it and maybe double the dose to 10 mg/kg as second-line therapy because that’s what they did in E3200.” That may be the right thing to do, but we don’t have the data. Bevacizumab is expensive and while the subjective toxicity is minimal, it has some very rare but very serious potential toxicities.

Adjuvant therapy for patients with Stage III disease

I’m pretty comfortable with the MOSAIC data (de Gramont 2005), so I generally use FOLFOX in the adjuvant setting for patients with Stage III disease. When I have a patient who is particularly dependent on their fine-motor skills, I discuss with them whether we want to include oxaliplatin in their treatment because the neurotoxicity might compromise their quality of life. If I’m concerned about a patient’s ability to tolerate combination chemotherapy, I might consider using one of several schedules of 5-FU/leucovorin or capecitabine.

We don’t know the efficacy of FOLFIRI in the adjuvant setting, but if the PETACC-3 and ACCORD-2 studies are positive, then we would have an interesting alternative to FOLFOX for combination therapy without long-term neurotoxicity. We do know that IFL was not effective in the adjuvant setting, so that’s not an option that should be considered.

Capecitabine as adjuvant therapy: The X-ACT trial

In a reliable patient, capecitabine is a reasonable alternative when we don’t want to use oxaliplatin, bearing in mind that the capecitabine data were generated in Europe (Cassidy 2004) and, for reasons that are not completely clear, European patients tolerate capecitabine better than American patients. In the X-ACT trial, the European adjuvant trial comparing capecitabine to Mayo Clinic 5-FU/leucovorin, the results for patients with Stage III disease who received capecitabine looked remarkably good. The study was designed as a non-inferiority study, but in a number of parameters capecitabine actually appears to be modestly superior.

Select publications

Dr Saltz is a Professor of Medicine at Weill College of Cornell University, an Attending Physician and the Colorectal Disease Management Team Leader at Memorial Sloan-Kettering Cancer Center in New York, New York.