Michael B. Atkins, MD: We’ve known, from our melanoma data, that immunotherapy can enhance subclinical toxicity from other therapies. And we know that from ipilimumab/dacarbazine, we saw liver toxicity with that combination that we didn’t see with either ipilimumab or dacarbazine alone. Who knew that dacarbazine caused liver toxicity? It’s not a surprise that a drug that’s not a very clean inhibitor of VEGF, such as pazopanib—which has liver toxicity—could have exacerbated liver toxicity when combined with something that enhances inflammation at the site of damage.
After seeing that initial experience with sunitinib and pazopanib being combined with anti-PD-1 inhibitors, we try to focus on more selective inhibitors of VEGF. And that was either axitinib, which is the most selective of the VEGF inhibitors, or bevacizumab, the antibody that binds VEGF. We’ve had a little bit more success combining anti-PD-1s or anti-PD-L1s with either bevacizumab or axitinib.
I led a trial that was reported out at the ESMO 2016 Congress, combining axitinib with pembrolizumab. In 52 patients, the combination was well tolerated, with very little in the way of liver toxicity. Most of the toxicity was still related to axitinib rather than pembrolizumab. The response rate was over 70%, and over 90% of patients had tumor shrinkage. The median PFS was at least 15 months at that time. That trial data led to 2 phase III trials essentially: one that’s ongoing of axitinib/pembrolizumab versus sunitinib and a subsequent trial that came out by Pfizer looking at avelumab/axitinib versus sunitinib, both of which are approved at the moment.
Another trial was a trial looking at atezolizumab, a PD-L1 antibody, together with bevacizumab. This was a small phase I trial that showed about a 40% response rate with the combination and an increase in immune infiltration in the tumor microenvironment in patients who had a biopsy before and during treatment. This led to a randomized phase II study. At the Genitourinary Cancers Symposium, David McDermott reported that what Dr. Tom Powles presented was the first randomized data comparing a combination immuno-oncology/VEGF inhibitor to a VEGF receptor TKI or to atezolizumab alone. And in that report, there was no significant difference in PFS for the intent-to-treat population for the bevacizumab/atezolizumab combination compared to sunitinib. But in the group of patients that represented about half of the patients in the trial who had PD-L1 expression that was elevated, in their immune cell infiltrate—which is the way Genentech does their PD-L1 assays—there was a doubling of PFS to 14 months for the bevacizumab/atezolizumab arm compared to 7 months or so for the sunitinib arm and a virtual doubling of the response rate.
Interestingly enough, their highest number of CRs, or complete responders, was in the atezolizumab monotherapy arm. That study had a heavy translational endpoint. At the American Association for Cancer Research Annual Meeting, Dr. McDermott presented some biomarker data from that subset, which is fairly interesting. They did RNA-sequencing on the tumor cells looking for either an angiogenesis signature, a T effector cell signature, or a myeloid signature—that were all predefined—and then divided patients based on the median expression into either high or low in those various categories and looked at how the 3 treatments behaved in those categories. They were able to identify that patients who received sunitinib and had an angiogenesis-high signature did much better than the atezolizumab-alone arm. That was not the case in the angiogenesis-alone arm. So, that’s the first time we actually have a predictive biomarker that could be potentially used for a VEGF receptor TKI.
We also saw that the bevacizumab/atezolizumab arm was the best in the T effector-signature group. It was better than the atezolizumab-alone arm even though there was some higher degree of complete responders in the atezolizumab-alone arm. But perhaps the most interesting data were in the group of patients who were T effector-high and then divided into 2 groups based on whether they had a myeloid suppressor cell signature versus not. And in the patients who were T effector-high, myeloid-less tumor infiltration with myeloid suppressor cells, there was good activity in both the atezolizumab-alone arm and the bevacizumab/atezolizumab arm. In the T effector-high, myeloid-high patients—immunosuppressive—there was very poor activity with atezolizumab alone. But that activity would be rescued by the addition of bevacizumab.
Bevacizumab/atezolizumab did very well in that group, much better than the sunitinib group, and it suggests that that’s what bevacizumab might be doing in this combination: suppressing the effects of those myeloid-derived suppressor cells, suggesting that that combination may actually be working as an immune therapy, rather than combining a VEGF inhibitor with an immunotherapy. It gives one a reason for potentially thinking about what you’re doing with those combinations in a patient population who might require that type of combination. And that’s something that will be looked at in the phase III trial of bevacizumab/atezolizumab versus sunitinib, which has been accruing and, I think, completed accrual in October of 2016.
Transcript Edited for Clarity
Brought to you in part by Eisai