Expert Discusses the Safety and Efficacy of Mirvetuximab Soravtansine in Ovarian Cancer Following FORWARD I Findings

Kathleen Moore, MD

Kathleen Moore, MD

Topline results from the phase III FORWARD I/GOG 3011 demonstrated that mirvetuximab soravtansine, an antibody—drug conjugate (ADC), did not meet its primary endpoint of progression-free survival (PFS) in women with folate receptor alpha (FRα)–positive platinum-resistant ovarian cancer who have had 1 to 3 prior lines of chemotherapy.¹However, an exploratory analysis showed the trial may have missed the primary endpoint due to FRα scoring methods.

In prior phase I studies of mirvetuximab soravtansine in this patient population, investigators utilized the PS2+ scoring procedure to identify levels of FRα in patients.²The phase III trial sought to compare patients with medium to high levels of FRα positivity, but investigators used the 10X scoring procedure, a method that would identify patients quicker.

After data from the phase III trial were analyzed, investigators ran an exploratory analysis on the 2 cohorts of patients with medium or high FRα levels by using the PS2+ scoring system. These data demonstrated that about one-third of the patients were not eligible for the trial because they did not meet the limitations for FRα—positivity and half of the patients categorized as FRα-high actually had medium levels of expression.

In an additional analysis of the patients identified as FRα-high on PS2+ scoring, the PFS rates andPvalues were comparable to what investigators originally expected to see prior to enrollment of FORWARD I. However, these data are exploratory and represent a smaller proportion of patients than intended when enrolling the trial. 

Investigators will enroll new patients to the MIRASOL trial to reanalyze this ADC in women with platinum-resistant ovarian cancer who have received 1 to 3 prior lines of chemotherapy and are identified as FRα-high based on the PS2+ score.

“I think based on the exploratory analysis and the phase I experience, which is all very consistent, we feel this will be a positive study to move this agent forward for our patients, but we will have to do another trial,” said Kathleen Moore, MD. “Lessons have been learned about the importance of scoring, but we will get it right.”

In an interview withTargeted Oncology, Moore, associate director of clinical research at the Stephenson Cancer Center at the University of Oklahoma, discussed the data surrounding mirvetuximab soravtansine in this patient population and the plans to reevaluate the ADC following the results from the FORWARD I/GOG 3011 clinical trial. She also highlighted other areas of investigation for women with platinum-resistant ovarian cancer.

TARGETED ONCOLOGY: What was the rationale for evaluating mirvetuximab soravtansine in these patients?

Moore:FORWARD I is a randomized phase III study evaluating mirvetuximab soravtansine, which is an ADC that targets the FRα transmembrane protein. FRα protein is almost ubiquitously expressed, if not overexpressed, in high-grade serous ovarian cancer. There are also varying levels of expression that makes it an ideal target for drug delivery because it is not expressed on normal tissues with some very rare exceptions. It allows us to focus this conjugate just to the ovarian cancer tumor cells and deliver a pay load of very potent chemotherapy, which is a very potent antimitotic agent, and we know ovarian cancer cells are very sensitive to that. It’s not paclitaxel by any imagination, but it is a microtubule toxin.

This is a drug that potentially has many applications, but in this initial registration clinical trial, it was intended to show efficacy as compared with monotherapy chemotherapy in women who had platinum-resistant ovarian cancer and 1 to 3 prior lines of chemotherapy who also overexpressed FRα protein on their tumor cells.

TARGETED ONCOLOGY: Are we seeing any advanced toxicity with this treatment?

Moore:No, and that’s the nice thing about ADCs, in general, where this is the advantage when they work. You are targeting very, very potent chemotherapy molecules that you would otherwise not be able to use, normally, systemically. You conjugate them to these ADCs, and when they are circulating in the body looking for their target, they are not active. They are not active until they are released from the conjugate, and they have to be internalized into the tumor cell itself where there is a target to be active. Even though you are using a very potent molecule of chemotherapy, you are directed it directly to the tumor cell, and you are able to use far less chemotherapy than you would otherwise need to if you were using, say, systemic paclitaxel or any of the other agents that we use where it goes throughout the body. You can use far less in the nanomolar range of cytotoxics with ADCs because you are just sending molecules directly to the tumor cell.

It would be a perfect world if you could say you have no off-target toxicities. That’s the idea, but we do have some off-target toxicities and on-target toxicities. However, they tend to be less than [with] traditional cytotoxic chemotherapy and/or differentiated, so they not overlapping with other sorts of toxicities. It fits in well with the treatment options that are otherwise available for patients with ovarian cancer.

TARGETED ONCOLOGY: What were the initial data we had seen with this treatment prior to the FORWARD I data?

Moore:The initial data we saw that has already published on where mirvetuximab soravtansine comes from a pretty extensive phase I experience. We initially, in a traditional phase I fashion, allowed treatment in very heavily pretreated patients and actually saw efficacy in that group. Then we started to see where our target was and narrow down that population of women with 1 to 3 prior lines of therapy and medium to high FRα expression. In those single-arm, early-phase studies, we saw response rates over 40%, and those response rates were quite durable for patients with platinum-resistant disease. That was the signal that led to the development of FORWARD I as a registration trial in a similar population.

We also know from that fairly large phase I experience what our expected toxicities would be, so we knew that we did not see a tremendous amount of hematologic toxicity. We knew that we had to expect some gastrointestinal toxicity, mainly diarrhea and a little bit of nausea, but those were fairly easy to mitigate in the phase I. As a class effect with many ADCs, and mirvetuximab soravtansine is certainly no exception, you do see visual disturbances, such as blurry vision and keratopathy, which are most classically associated. We did see that in the phase I experience and worked quite hard to develop mitigation strategies for that to prevent it from happening at all and/or make sure that if it did happen, it was a very low grade, 100% reversible, and we had a good plan for moving forward into a more generalized population in the phase III study.

TARGETED ONCOLOGY: Could you elaborate on the durability of this agent in the phase I experience?

Moore:In the phase I experience, the response rate for mirvetuximab soravtansine was over 40%. What you expect, from a historical standpoint, are response rates in the 15% range, maybe to 20% at best, and durability of 3 to 4 months. What we were seeing was over 40% response rate with about 7 months median PFS, and the duration of response could be even longer. We were seeing 9 months duration of response. It did seem to have a signal in that selected population of FRα medium to high of clinically relevant clinical activity in an area where we see a differentiated toxicity profile. [There’s not] no toxicity, but it is very well tolerated and different from what has been experienced with our patients with other therapies or that might be experienced in subsequent therapy, so there weren’t overlapping, cumulative toxicities of the same sort. 

TARGETED ONCOLOGY: What was the design of the trial?

Moore:FORWARD I was a planned registration study that compared mirvetuximab soravtansine with investigator’s choice of chemotherapy in women with 1 to 3 prior lines of therapy in a medium to high FRα expression. The primary endpoint was designed as PFS, in both the intent-to-treat (ITT) group and the subpopulation of FRα-high. The assumptions that went into the statistical design were that the control group would have a median PFS of about 3.5 months, which is our historical reference, and mirvetuximab soravtansine would have a median PFS of 6 months, which correlates with a hazard ratio of 0.58. We chose to use something called the Hochberg statistical analysis, which would allow us to simultaneously assess that PFS in both the ITT and FRα-high group at the same time. If both of those were less than 0.05, then it was positive. If 1 was greater than 0.05, the other would have to be a Pvalue less than 0.025 in order to be positive. That was the risk that was taken.

TARGETED ONCOLOGY: What were the findings from this trial?

Moore:What we found in this trial was, again, a very differentiated safety profile with no new safety signals, far [fewer] grade 3 treatment-emergent adverse events (AEs), far [fewer] patients having to dose reduce because of treatment-emergent AEs and very low rates in both arms of treatment discontinuation because of AEs. We saw statistically less hematologic toxicity, less neuropathy, no alopecia, and we saw what we expected for the drug, which was nausea and diarrhea of low-grade and the visual disturbances, which were all again of low-grade and mitigated with the preplanned interventions.

Unfortunately, in the primary analysis of PFS, there was no difference in the ITT group. The median in the control group was as expected at 3.39 months, but in the in the FRα-high subpopulation, it was only 4.6 months, which gave us a HR of 0.69 and aPvalue that was less than 0.05, but it was not less than 0.025, so it was not statistically significant. When we looked at primary and all the secondary analyses in the ITT group, there were really no differences, but in the FRα-high group, mirvetuximab soravtansine did numerically look superior, but none of the Pvalues were less than 0.025. This was not statistically significant, and it was a negative study.

The confusing thing was how far off the median looked from what our expectations were for the FRα-high group. It was 2 months less than what we had expected, and the other thing that struck us was the fact that of the patients who enrolled on FORWARD I, and remembering you needed to have a median or high expression of FRα, 60% were characterized as FRα high, which is far more than we expected. We wondered if something had happened with the FRα-high scoring in FORWARD I that may have compromised the study. The reason we question that is because in the phase I experience, which again was several hundred patients, we used what is called a PS2+ scoring, so the same assay was used in both but how you scored it changed. In the phase I experience, we used a scoring system that took into account the percentage of cells that were positive as well as the intensity. You combined those, and there was an algorithm for whether or not you were medium or high. That process is doable and is used as selection in other assets that require immunohistochemical scoring, but it is somewhat cumbersome, so we worked with a company to evaluate whether something of 10X scoring would be equivalent. It’s an easier and quicker way to score, and it takes into account the percentage of cells that are positive at less than 10X magnification, but not the intensity. The bridging study basically said that they were equivalent, it certainly made for a more generalizable companion diagnostic, so that is what was chosen for FORWARD I.

When we got the results of FORWARD I, we re-analyzed all of the specimens using the PS2+ by a blinded independent pathologist. This is a very exploratory analysis just trying to figure out what happened, but we found in the distribution of patients who enrolled, about one-third were ineligible for FORWARD I because their FRα was too low. Among those patients who were classified as FRα-high by the 10X scoring, half of those were actually medium. Our FRα-high subgroup, which was 1 of our primary endpoints, was actually diluted out by half by FRα-medium patients . Then we re-ran the statistical analysis just on the FRα-high [group], which was a much smaller population, so we lost power. When we looked at the PFS, it looks like the assumptions we had set forth in the study. The control is still 3.3 months because why would that have changed, but the FRα-high group is now 5.6 months, and thePvalue is 0.015. All of the secondary endpoints are either positive or trend toward positive, including overall survival where medians go from 11 to 16.5 months. That Pvalue is not statistically significant, but we lost half our power, so we can only make hypothesis based on that. However, the trend was clearly there.

We did show a well-tolerated agent with a differentiated toxicity profile and even with 10X scoring, a signal of efficacy or superiority that makes us believe still that mirvetuximab soravtansine has a place in the treatment of people with platinum-resistant ovarian cancer. The exploratory reanalysis gives us a lot of confidence that we actually have an active agent and that it confirms commitment to developing the asset and relaunching a second confirmatory trial, called MIRASOL, and that should launch at the end of 2019. It will only enroll patients with FRα-high, so not the mediums, as scored by the PS2+, the original scoring system with still 1 to 3 prior lines of chemotherapy and platinum-resistant disease. There will be a 1:1 randomization instead of 2:1 because we now have enough safety data, and PFS by investigator will be the primary endpoint. I think based on the exploratory analysis and the phase I experience, which is all very consistent, we feel this will be a positive study to move this agent forward for our patients, but we will have to do another trial. Lessons have been learned about the importance of scoring, but we will get it right. This agent will be available for patients, but it will take another few years.

TARGETED ONCOLOGY: Do you think these data will have any impact on enrollment of the next trial?

Moore:I hope not. I think that we have to be careful with the exploratory reanalysis because the intention is not to say it’s a positive study because it’s a negative study. We have to analyze the study the way it was designed, and based on the way the patients were enrolled, it is a negative study. However, we believe it is missed because of the misfire of the scoring. I think if we explain that and come back around to say we will do a trial in an appropriately selected population and the expectation is that that will be positive, then hopefully our coinvestigators will maintain their enthusiasm. 

The investigators who have used this agent know it works. I think when the initial press release came out saying it was negative, people were puzzled. There was a lot of surprise that it was negative because we have seen how well it works in an appropriately selected population. I hope enthusiasm will be maintained, but I understand when you have a negative trial, you can sometimes lose a little momentum, but usually when you have a negative trial, you don’t have the company coming back around with another trial because the trial was flawed in a way. I think that shows commitment to the asset and to our patients, so I think that people will see that.

TARGETED ONCOLOGY: In the platinum-resistant setting, are there any other agents that appear exciting?

Moore:For single agents, not really. Unfortunately, there have been a lot of promising ideas that fizzled out. The things that are exciting are still early, but there is quite a bit of work being done, not so much in platinum-resistant; that concept is probably fading out a little bit. We are looking more at patients who have homologous recombination proficiency, which makes them more likely to be platinum-resistant but also makes them resistant to PARP inhibitors and other agents that target the DNA damage response pathway. Those homologous recombination proficient patients are either because they have an acquired proficiency or they just denoted their DNA and are resistant to everything we use that targets the DNA, which is the platinums.

Trying to induce homologous recombination deficiency (HRD) in that population, there is a lot of exciting stuff going on, and it’s all combinations. It’s not just 1 agent, it’s strategies. The furthest along right now is using PARP inhibitors, or potentially other DNA damage agents, with antiangiogenic agents, such as cediranib (AZD-2171), or bevacizumab (Avastin), which was just presented, the NSGO-AVANOVA2/ENGOT-OV24 at the 2019 ASCO Meeting in a platinum-sensitive population. There are data for that synergy, we think, between antiangiogenics and PARP inhibitors. There is a lot of interest in immunotherapy and PARP inhibitors with mechanistic synergy with activation of the STING pathway. There are TOPACIO data.

There are also newer things that mechanistically are very exciting. PARP inhibitors plus BET inhibitors or bromodomain inhibitors, PARP inhibitors and PI3K inhibitors, and PARP inhibitors and MEK inhibitors. There are so many pathways that get upregulated that restore, we think, homologous recombination proficiency that if you take those out and add a PARP to it, we may be able to create new combinations for patients, but the corollary to that is we have to figure out why they are resistant and potentially individualize, so I think it is getting exciting and also complex. Combinations are where we are going to make [progress] for these therapy-resistant patients. They are coming, so I think the studies are up and running. They have a lot of translational science to try and identify the right combination for the right patient. It’s going to take time.

References:

1. Moore KN, Oza AM, Colombo N, et al. FORWARD I (GOG 301: A Phase III study of mirvetuximab soravtansine, a folate receptor alpha (FRa)-targeting antibody-drug conjugate (ADC), versus chemotherapy in patients (pts) with platinum-resistant ovarian cancer (PROC).Ann Oncol.2019;30(suppl 5; abstr 992O). doi: 10.1093/annonc/mdz250.
2. Moore KN, Matulonis UA, O’Malley DM, et al. Mirvetuximab soravtansine (IMGN853), a folate receptor alpha (FRα)-targeting antibody-drug conjugate (ADC), in platinum-resistant epithelial ovarian cancer (EOC) patients (pts): activity and safety analyses in phase I pooled expansion cohorts.J Clin Oncol.2017;35(suppl; abstr 5547). doi: 10.1200/JCO.2017.35.15_suppl.5547.