Robert L. Coleman, MD: The ARIEL3 trial was done in a population that was similar to those in the NOVA trial. Patients who were platinum-sensitive responders were then randomized to rucaparib, 2:1 randomization, as the other ones were, versus placebo. This trial was ultimately presented in September of 2017 at the European Society of Medical Oncology meeting, and it showed a very strong effect on delay of progression, just like the other 2 trials had done.
As aligned with SOLO-2, this was an investigator-assessed progression-free survival endpoint. They shared that similarity, but it was in a much broader population and was slightly different from the way NOVA allocated their patients. In NOVA, there was a germline patient population, and then there was everybody else. In ARIEL3, it was just everybody else. But they did a step-down procedure, from a statistical standpoint. They looked at the patients who were germline or somatic: BRCA mutation carriers or those with a BRCA mutation in the tumor. That was one cohort. If there was a positive result, then they expanded it to the HRD population. So, that included the germline and somatic patients, and also, now, the HRD patients. And, in this particular trial, the HRD assay was assessed by a different methodology. It was using loss of heterozygosity as the global review in how damaged the DNA repair mechanism was. It was done by the Foundation Medicine test as opposed to the Myriad Genetics test, which was done in NOVA.
If the second population that stepped down into the larger patient population was positive, then it extended the analysis to all 3. So, it started with the BRCA mutation. Then, the HRD plus BRCA mutation. And then, all-comers. That step-down procedure allowed them to statistically assess all 3 patient populations. And they did this not only with the progression-free survival outcomes but, also, from the quality of life aspects. They assessed quality of life, and they expanded it down the cohorts. The point was, if any one of those analyses were negative, they stopped.
As I mentioned, as it showed in previous trials, the addition of a PARP inhibitor in this step-down procedure improved the progression-free survival to a dramatic degree. Again, the hazard ratios were all in the low 0.2 range. With each of the larger cohorts, the hazard ratios rose. But, they were still significantly positive.
One caveat that was different in ARIEL3, from the other trials, was that they actually looked at what the likelihood of response was in the patients that were randomized. They said, “Wait a minute, these patients were treated with chemotherapy.” Yes, that’s true. But all 3 of the trials included patients who only had partial responses to treatment. So, it was about 50/50, 60/40, and ARIEL3 was actually about 30/70. There was a significant number of patients who had a partial response to their chemotherapy before they got randomized to placebo, versus their active compound.
In ARIEL3, we looked at what the response rate to the drug was in the patients that we could measure response in—that happened to be a significant proportion of the patients that had persistent disease or had partial responses to the platinum-based regimen. In those cohorts, it showed that patients who carried a germline or somatic mutation in BRCA had about a 40% response rate when they were given the drug. So, they had just responded to the platinum-based regimen. Then, we gave them the PARP inhibitor, and 40% of those patients went on to respond again. It carried out to each of the additional sub-cohorts, but the response rates, as you would expect, were reduced as they lost their homologous recombination deficiency signal.
When we looked at the adverse effects in that particular trial, we saw some very commonalities, gastrointestinal side effects, as I mentioned, and nausea, vomiting, taste changes, and weight loss. Those were very similar across the board. Some nuances in the ARIEL3 patient population, that were well documented, were changes in the transaminase enzymes. We saw a significant increase in the transaminases, on the average of below 2times the upper limit of normal, which decreased as the trial went on. Another effect, that was similar across the board with all PARP inhibitors, was a change in the creatinine. These were well documented and reported. But as I mentioned in the beginning, the side effect that we were concerned about, which we were very happy to report that it did not occur in all of the patients who were randomized to a higher rate of placebo versus drug, was myelodysplastic syndrome or AML.
The exclusion criteria for the 3 trials was also slightly different. In NOVA, there was a requirement that in patients who had had a response to the platinum-based regimen, the size of the individual tumor nodule needed to be less than 2 cm. And, the CA-125 values needed to be in the normal range. In the ARIEL3 trial, there was no restriction on the size of the tumor. But the CA-125 values had to be normal. People were a little confused by that, but the intent was to try to identify a patient cohort population that was responding well to platinum-based therapy. It was thought that patients that had normalized CA-125 but still had disease that was visible by CAT scan may, in fact, be on their way to responding. It’s interesting. As I mentioned in the partial response outcomes, patients in the placebo group did respond. And so, that might serve as some proof of a concept that these patients hadn’t yet reached their maximal or optimal response time point. We had stopped the evaluation, or stopped the treatment with the chemotherapy a bit early. But they had already normalized their CA-125. This was maybe a precursor of what their best response, ultimately, was going to be.