At the time that OncTalk (the predecessor to GRACE) was just getting off the ground in the fall of 2006 (wow, three years have gone quickly!), Avastin (bevacizumab) was just getting FDA approval in the first line treatment of advanced NSCLC. The main focus was on the randomized trial that showed a survival benefit and led to its approval, and in the rush to generate summaries of the treatment highlights for various treatment settings in lung cancer, the story of Avastin skipped over its development in lung cancer. Many people may know that Avastin is given only in patients with non-squamous cancers, but it’s worth backtracking to understand why.

The first significant trial of Avastin in NSCLC was a randomized phase II trial done at Vanderbilt University and that enrolled about 100 patients with advanced NSCLC who received either chemotherapy with carbo/taxol (paclitaxel) alone every three weeks., or the same chemo with Avastin at either 7.5 or 15 mg/kg IV every three weeks. This trial was open to patients with any NSCLC histology. Importantly, patients who were enrolled on the chemo only arm as first line therapy were allowed to cross over to the Avastin alone (at the higher dose) after progression on carbo/taxol.

(Click on image to enlarge)

The results of the trial were overall favorable for Avastin, at least at the higher dose, though the results were somewhat puzzling. The carbo/taxol/Avastin high dose arm had a significant improvement in median progression-free survival (PFS) of over 7 months compared with the very typical 4 month range in the other two arms. The high dose Avastin arm also had a median overall survival (OS) of nearly 18 months, which is terrific (but also was far better than the 12 months seen in the phase III trial with this regimen, illustrating how we should never presume that good results in smaller phase II trials will be replicated in larger, multicenter phase III trials). But what was confusing was that the patients on the chemo alone arm had a nearly unprecedented median OS of nearly 15 months, which was better than what they saw in the chemo/low dose Avastin arm, and also about double what you see in at least larger trials with this chemo regimen.

Some of this may be that the patients at a major academic center like Vanderbilt have more favorable features than a broader population because only those motivated and healthy enough to get there were included (so-called selection bias), but there were two other reasons that we might identify as important. First, it could be important that patients assigned to chemo crossed over to the Avastin, since 19 of the 32 patients in that arm received it, for a median duration of 12 weeks. Although there were no patients who showed significant tumor shrinkage that could be classified as a response, five remained on Avastin without progression for 6 months or longer. In the figure below, the horizontal lines represent the time that patients remained on Avastin without progressing:

This is probably the best evidence we have of the potential value of single agent Avastin as a maintenance therapy after first line chemo (which is part of our standard approach for Avastin now), though this group of patients were receiving Avastin for the first time rather than continuing after several cycles of chemo and Avastin together.

But the other potentially very important issue is in the bottom of the table above. There were six patients with life threatening or fatal hemoptysis (coughing up blood), comprising 9% of the patients who received Avastin on the trial. All 6 had large central tumors, and unfortunately 5 of them happened to be on the low dose Avastin arm, so this may have really affected the median OS greatly (particularly in a small group, like the 32 patients on this arm). Not surprisingly, these situations caused a lot of concern, and they were studied and found to be very disproportionately likely to have a squamous cell carcinoma (4/13 patients with squamous cell, or 31%, compared with 2/53 patients with adenocarcinoma, or 4%). When the results of the trial were re-analyzed without the patients without squamous NSCLC, the results looked better overall and no longer showed the puzzling results of the patients on the low dose Avastin arm doing worse than the chemo alone arm:

Of course, the story went on, and we’ve gained a lot more experience with Avastin since then, and nearly all of it in patients without squamous cell NSCLC. The larger ECOG 4599 trial was designed to exclude patients with squamous NSCLC based on the concern around safety, and bleeding complications weren’t entirely eliminated but were down in the 2% range. Though we can’t know for sure, it’s possible that if the investigators hadn’t reviewed these bleeding cases carefully, people with a higher risk of bleeding complications could have been enrolled in large numbers on the phase III Avastin studies, leading to bleeding complications that the trials might have been stopped early, with an alarmingly higher risk of bleeding now than we see with it, and leaving many patients who now benefit from it from having that opportunity.

Meanwhile, several trials with other anti-angiogenic agents like sutent (sunitinib) and nexavar (sorafenib) have raised concerns poor outcomes in the trials that have enrolled patients with squamous cell carcinomas. Overall, I’d say the limited evidence should continue to make us very cautious about giving any anti-angiogenic agents to patients with squamous NSCLC.