Robert L. Coleman, MD: For patients who unfortunately experience a recurrence, we initiate a conversation about what to do next. Fortunately, as I mentioned, we have a lot of options. For the patients that fall into that platinum-sensitive category, or chemotherapy-sensitive category, our best treatment options, at least at this time, seem to be with a platinum-based therapy. Right now, our consideration is that these should be platinum-based combinations. There have been a series of trials in platinum-sensitive patients that have tried to determine which of these is best. The first in this line of investigation looked to establish whether a combination was better than a single-agent platinum therapy. In that specific example, carboplatin was compared to gemcitabine and carboplatin. It was shown that the gemcitabine and carboplatin combination was better.
Then, we moved to the addition of bevacizumab. The OCEANS trial added bevacizumab to gemcitabine and carboplatin. This was a really interesting trial because the progression-free survival in the placebo arm, in that trial, was the same as the experimental arm in the prior trial. They overlapped, identically. But interestingly, the overall survival of the patients who went on to OCEANS in both cohorts was double what we saw in the first trial. So, again, to my earlier comments about the prevalence of the disease increasing, this was a clear example where patients were actually living longer, at that point in time.
Then, we did another trial called GOG-213, which looked at the addition of bevacizumab to paclitaxel and carboplatin. Again, another doublet. That particular trial, which actually has a secondary objective for secondary surgery, is still yet to report. But we did report on the benefit of bevacizumab to paclitaxel and carboplatin versus paclitaxel/carboplatin, alone.
This is the only trial that is adequately powered to address overall survival, and it demonstrated that effect. Bevacizumab was approved in a supplemental indication to paclitaxel and carboplatin, and gemcitabine and carboplatin in December of last year. It was a big win.
But it does raise a conceptual and a contextual issue about, what do you do with a patient who also might be amenable to a PARP inhibitor? GOG-213 was a study that randomized patients at the point of diagnosis. So, at that time, you have to raise the issue or discussion with the patient regarding the use of a maintenance PARP because, in the trials that all evaluated maintenance therapy with a PARP inhibitor, they were not given bevacizumab. They were all given as a platinum-based doublet. If they got a response, then they were randomized to placebo versus treatment with a PARP inhibitor. So, my discussion, at the time of diagnosis, is, “How do I pick who’s going to go on to bevacizumab? That trial showed an overall survival advantage?” Or, “Who’s going to go on to PARP? Those 3 trials showed a benefit in progression-free survival?”
We have yet to see what the overall survival is. It’s a tough decision. So, it’s multifactorial and the patient and the provider need to be intimately involved in this decision-making process. There are subtleties that are not often realized just on paper. For instance, when you use a drug like paclitaxel, you get hair loss. That’s a big issue for some patients. For others, it may not be a big issue. They may look at that data for overall survival as the be-all and end-all, knowing that they’ll lose their hair, that they have a risk for hypersensitivity reactions to the carboplatin, and that they may have issues with neuropathy. It’s not easy. So, those kinds of features have to be made clear as part of the discussion.
Not everybody does this, but I have been looking at the patients that are now known to have the BRCA mutation, and we germline test everybody with a new diagnosis of ovarian cancer as part of our policy. And so, we know who these patients are ahead of time. In those groups of patients, we happen to see the strongest benefit across all of the trials. NOVA, SOLO-2, and ARIEL3 all show that the patients that carry or have a germline somatic mutation did the very, very best. Those patients may, ultimately, have an overall survival advantage, as well. Those patients strongly fall into that side of potentially giving them a PARP inhibitor. And so, I have that discussion with them.
For all of the others, the wild-type patients, I will have this discussion and offer an intermediary ground. We can test the tumor for homologous recombination deficiency and then determine whether or not they should go on one of the other arms. As I mentioned, homologous recombination deficiency was only individually assessed in the NOVA trial. And, even then, there were some patients included who had somatic BRCA mutations. But in that group that had HRD and were not wild type and were not germline, they showed a strong effect for therapy.
In the ARIEL3 patient population, the HRD subpopulation also seemed to show an effect in that sub-cohort. And, again, these are secondary exploratory analyses. But it does seem to say that damaged machinery is a potential cohort in patients who respond, even in the maintenance setting. And, most of that’s probably because switch maintenance and treatment are probably the same thing. In other words, I think we’re kidding ourselves in looking at a patient who’s had a complete response following platinum-based therapy as, actually, disease-free. Many of those patients have small volume disease, that we can’t detect by imaging. Essentially, what we’re doing is we’re continuing treatment. We know, from ARIEL3, that we’re continuing active treatment because even in the ones that had tumors that we can’t see, 40% of those patients actually responded.
Treatment and switch maintenance—it’s kind of a nuance, but it does bring up an important other caveat which I want to address. What do you do with a patient who actually has a complete response? You’ve treated them with platinum-based therapy, they’re happy, their CA-125 is normal, and their CAT scans are negative. Now what? They know that their likelihood of the cancer coming back is extremely high. It’s higher than it is in the frontline setting. Should we treat them with a PARP inhibitor, or not?
That’s another category of patients in which HRD may be of some value. So, if we check and we find that they actually have HRD, then I think our thoughts are that PARP inhibitors might work quite well, even in the patients who have a complete response. Until the data are sorted out, clearly, across all 3 trials in those patients, this will help us to better define it. But ultimately, I think when the overall survival data mature and we can put that in context with the bevacizumab story, that I mentioned before, and the toxicities that are different between the 2 modalities of treatment, that will ultimately help us define what the optimal way to treat these patients will be.