Evolving Technology Behind ADCs Could Improve Benefit in Advanced Ovarian Cancer

Michael Birrer, MD, PhD

Antibody-drug conjugates (ADCs) have been incorporated into the treatment of patients with solid tumors, such as ovarian cancer. However, the technology behind older ADCs can limit their effectiveness, according to Michael Birrer, MD, PhD.

However, new technology designed to deliver payloads more accurately to tumor cells represents an important change in ADCs, he noted. Additionally, increased knowledge behind biomarkers, such as NaPi2B, could aid the effectiveness of ADCs for patients with advanced ovarian cancer.

“In the last 10 years, [ADCs] have now evolved into solid tumors,” Birrer said. “In breast cancer, there is a substantial track record [for them], and in ovarian cancer, there has now been a fair amount of work. ADCs are going to have a huge role in the future.”

In an interview with OncLive^®, Birrer, vice chancellor at University of Arkansas for Medical Sciences, director at Winthrop P. Rockefeller Cancer Institute, and director of the Cancer Service Line at UAMS, further discussed how new technology has transformed ADCs, how upifitamab rilsodotin (XMT-1536, UpRi) could have a role in treating patients with ovarian cancer, and why the NaPi2B biomarker could be a vital target.

OncLive^®: What are the unmet needs for patients with platinum-resistant and platinum-sensitive ovarian cancer? Who progresses after PARP inhibitors in the maintenance setting?

Birrer: Platinum-resistant ovarian cancer continues to be the major unmet need within epithelial ovarian cancer and its treatment. I would like to say we have made a lot of progress on that, but not really. If you look at the big picture, there was the phase 3 SOLO-1 trial [NCT01844986] where PARP inhibitors were used in the maintenance of [patients harboring] BRCA1 and BRCA2 [mutations]. There may be a subset of patients we are now curing that we did not before. However, that is still questionable.

The overall numbers in ovarian cancer are that 75% of patients with advanced-stage [disease] will eventually recur. Once they recur, most [experts] still think the disease is incurable, and all those patients will develop platinum-resistant disease, [becoming] an unmet need. The overall survival for platinum-resistant disease is about 11 to 12 months. The progression-free survival [PFS] is around 4 to 5 months.

The glimmer of hope was the phase 3 AURELIA study [NCT00976911], where bevacizumab [Avastin] was added to chemotherapy, and there was a statistically significant prolongation of PFS. If you look at the [paclitaxel and] bevacizumab combination, it was striking in that the PFS was out to 11 months and the overall survival [OS] was out to 20 months. [It is important to note] that that was a subset analysis. It was not preordained, so you cannot really believe it. But, that combination has become our standard.

[Platinum-resistant ovarian cancer] is the testing ground for most drugs. Because of that unmet need, the bar is low. We know the milestones that we can aim for, and it has been a debate as to whether that should be the case, because those tumors are badly beat up with all the chemotherapy. That’s why when you see a lot of trials in platinum-resistant [ovarian cancer], you will see a limitation on the prior lines of therapy—usually between 1 and 3.

What are the options for patients with platinum-sensitive ovarian cancer who progress on a PARP inhibitor in the maintenance setting?

You will see less trials available for platinum-sensitive [ovarian cancer] because the readout takes longer for companies to see whether their drugs are effective. Again, they shift toward [patients with] platinum-resistant disease, but platinum-sensitive recurrence is still an important issue. Of all the [patients with] platinum-resistant disease, most will go through a platinum-sensitive portion. If they have been on PARP inhibitors for maintenance, then the issue becomes, if they progress, are they still sensitive to a PARP inhibitor? That is the huge question for [patients with] platinum-sensitive recurrence who have been on maintenance PARP inhibitor.

What is the mechanism of resistance to the PARP inhibitor? My guess is that it is heterogeneous. Some are going to have reversion mutations, some are going to have [replication] fork stabilization, some are going to have P-glycoprotein activity, and [these patients] will be dependent upon the mechanism of PARP inhibitors resistance, which will determine whether retreatment with PARP makes sense. For instance, if there is a reversion mutation, those patients are not going to respond to a PARP inhibitor. That is an area of evolving biology that will [expand] in the future.

Could you elaborate on some of the ADCs under development in advanced ovarian cancer? How do they compare regarding mechanism of action?

ADCs are an area of rapid evolution. First, they are not new. They have been around for about 30 years. Where they made a claim to fame has been in liquid tumors, and part of that is because liquid tumors are genomically less complex. There is less heterogeneity, and you get better responses.

The components of an ADC are threefold in terms of how they work. One is the target, meaning what the antibody binds to. The second is the linker. There had been a lot of biochemistry by companies involved in trying to design that the perfect linker. Finally, [there is] the payload. The payload is well defined at this point. The payload was a problem 30 years ago. The original ADCs failed mostly because the payloads were known chemotherapeutic agents, which were nowhere near powerful enough to function as an ADC.

Now we know you cannot have an effective ADC unless it has an agent that is way too toxic to give as a free drug. With an ADC, it is active enough that when it kills the cell, then there is enough drug still released for a bystander effect. Effective ADCs work because they have a robust bystander effect.

For [ovarian cancer], most work has been on the folate receptor α [FRα], which is a validated and outstanding target. It is differentially overexpressed in epithelial ovarian cancer. In high-grade serous [disease], roughly 50% to 60% have high expression. It is not expressed in high levels in normal tissues. Type II alveolar cells express the FRα, which is why there was concern that there might be pneumonitis. [However], it never panned out.

[Moreover], there is some expression in renal tubules, but there has been no renal toxicity from ADCs targeting that. Another target that is up and coming is NaPi2B, which is a phosphate transporter. The purpose of this protein is unknown, but [it is] expressed at fairly high levels in about 60% of ovarian cancers. It is not expressed at very high levels in normal tissue, making the ideal target for an ADC.

The other components to mention are the FRα ADCs are the oldest, meaning that the 1 agent that is the furthest along in development uses fairly dated technology terms of attaching the payload to the antibody. But that has worked its way through several important clinical trials. The NaPi2B has been around for a while, and there was a series of ADCs that were tested, which did not do well. [However], the newer companies have newer technology, which allows them to attach more drug firmly to the antibody, so you get a bigger payload and a bigger bystander effect.

Expanding on NaPi2B, what is known so far about this marker? What work is being done to target it?

NaPi2B is a phosphate transporter. It is not clear that it is critical to the biology of the tumor. There have been several preclinical projects and experiments exploring that. There is certainly no evidence that its overexpression is a driver of the tumor.

However, for an ADC, that is OK. When you have HER2/neu in breast [cancer] and you are using trastuzumab [Herceptin] as an antibody, not an ADC, you assume that the target is involved in the biology. Inhibiting the target, like trastuzumab does [with HER2/neu in breast cancer], will obviously inhibit and kill breast cancer cells. However, this is an ADC. [The goal is for] the antibody to drag the payload to the tumor cell. It does not matter about the biology.

NaPi2B is fine in that regard. The other requirement is ensuring that it is not expressed on critical normal tissue, because you will be killing [healthy] heart, kidney, or liver cells. In this case, NaPi2B is very restricted in its expression. For ovarian cancer, we would put NaPi2B up there with a FRα as a validated overexpressed target.

Not only do you need to know that the target is expressed, and it is differentially expressed for normal tissues, but [you need to know] if it is heterogeneously expressed, if it is expressed in some of the tumor cells but not any other, or that there are anatomic locations within the abdomen for advanced-stage ovarian cancer. Then you could imagine how that would affect the efficacy of the ADC, because you are going to kill off all of those high expressers and be left with the non-expressers.

If [non-expressers comprise] 30% of the tumor in a patient, then, at most, you are going to get a partial response and develop resistant tumors rapidly. For FRα and NaPi2B, we have good data that there may be slight heterogeneity, but not enough to be overly concerned by it.

If you are using archival material, meaning when the patient is diagnosed that you use the tumor and you stain it for the target, but you are treating a patient with platinum-resistant disease 3 years later, you would like to know that this target does not change a lot. For FRα and NaPi2B, good data suggest there is not a lot of fluctuation. Although, being a biomarker [proponent], I would like to have more data, but we certainly have enough data to be comfortable about it.

Please discuss the unique drug to antibody ratio with upifitamab rilsodotin (UpRi).

[This technology] is the next generation for ADCs. Historically, the way ADCs have worked [is taking] the payload and attaching it directly to the antibody. Two things happen in that process. One is that most of the time, the conjugation process going through a proprietary linker is essentially described as a bell curve [with a Poisson distribution]. You have some antibodies in that solution that have 1 molecule of payload, and you have some that have 15 molecules of payload. Then the bell curve may average 3 or 4, and that is the solution that you are going to give the patient.

You can imagine what the outcome on that is. If the antibody with 1 payload binds to the tumor cell, there may not be enough payload to have any bystander effect. That area where that antibody binds, you will see the 1 cell get killed, but nothing else.

The flip side is if you have 15 molecules hanging off that antibody, 1 of those molecules might fall off. Now you have free drug. When using this random process of attaching payload to antibody, which is the older technology, you are going to get some issues on less efficacy, and you probably have a little more toxicity because you have free drug floating around. If you look, at the FRα, there is some ocular toxicity and myelosuppression, which is likely from circulating free drug.

NaPi2B is a newer target [utilizing what] they call a polymer technology. This polymer coats the antibody and it allows for more payload to be attached to the antibody. Again, it's a polymer, not a proprietary linker, where the payload may leak off. The polymer is secure, and once the antibody gets into the cell, the polymer dissolves, dissociates the payload, the cell is killed, and that extra drug is available for bystander effect. That technology allows you to deliver more drug per tumor cell and allows you to have more bystander effect.

The more bystander effect you have, the more efficacious the ADC. More importantly, it means that the quantification of the biomarker becomes less important. For the older technology, [it is] critical to try to get tumor cells to have high expression. There are lots of ways of quantifying that. However, it involves work and pathologies to look at it. The latter technology, because there is so much more of a bystander effect, you may just need to say yes or no, whether the tumor cells express. Even if they express low, it is sufficient to get a response. That is a huge step forward, not just from efficacy, but from a biomarker characterization.

The newer technologies coming down the pike from the same company have a newer process, which is coming online. We are beginning to see some data where, not only can they get more drug on the polymer, they can determine exactly the number of payloads they put on. Every antibody in that solution will have 6 payloads. You could imagine a trial where we are going to give these 40 patients 6 molecules [of payload], these 40 patients 12 molecules, and maybe another set [of patients] 18 molecules, and we are going to ask: “Who gets the best response-to-toxicity ratio?” This is becoming much more defined.

What has been observed in UpRi so far in terms of safety and efficacy?

The response rates are between 30% and 35%, which is exactly what you want to see with an agent. The toxicity profile is not only acceptable, but also better than what I have seen with other ADCs.

There is a diagnostic assay that is also in development. What is the optimal utility of this?

On the mirvetuximab soravtansine studies, we needed to look at not only the number of cells expressing, but the intensity of that expression. It is doable, but the assay needs to be locked in and fixed.

The phase 3 FORWARD I trial [NCT02631876] was a randomized [study] for mirvetuximab soravtansine, and it was ultimately a negative study, which was disappointing. But it really was not a negative study in the sense that it was a biomarker problem. If you look at the [histology score] and correlate it with response, those patients who had 2+ or 3+ staining tumors had the response rate we expected. The new study following that has now been positive, so I have a feeling that mirvetuximab soravtansine is slowly going to get there.

Ultimately, mass spectrometry quantification for either FRα or NaPi2B would be easy. You would get an actual quantification number.

The assay for NaPi2B is easier because of that technology, meaning that while we could stain and quantify 3+, 2+, 1+ for NaPi2B, it is probably not necessary. It is a binary readout. That is how the technology of how an ADC works and the evolving technology makes and informs the biomarker assay.

Are there any other emerging markers in ovarian cancer for these ADCs?

There are tremendous genomic data out now with The Cancer Genome Atlas and others. We have been trying to identify differentially expressed proteins on ovarian cancer. None have gotten close to what NaPi2B or FRα have. TROP-2 is out there, although it [may not be] better than [NaPi2B or FRα].

The other one would be tissue factor. [Tisotumab vedotin (Tivdak)] has recently gotten a lot of excitement in cervical cancer, which was the low-hanging fruit because those patients do so poorly. However, tissue factor is overexpressed in ovarian cancer. That may be another target for which there is already an ADC, and we can move forward on it.

Where do you see ADCs factoring into the future of treatment for ovarian cancer?

ADCs are here to stay. The first direction will be new targets, which we just discussed. We are not finished with the targets. It's certainly conceivable that for patients of recurrent ovarian cancer, you could use a FRα ADC. Some of the patients I have treated could get a year and a half to 2 years out of it. For platinum-resistant patients, they then progress and flip to NaPi2B, because they are not exclusive. They are overlapping. Since the payloads are different, there should not be any drug resistance.

You could theoretically think about having a series of ADCs, where patients go from 1 to the other. You are not curing them, but you are giving them a survival advantage and great quality of life.

The second direction you are going to see is this continual improvement in technology. Other than the target, you are going to see better improvement in terms of attaching the payload in a defined way to the antibody. As described before, we know what we are delivering as opposed to this Poisson distribution of antibodies that have [varying] payload on them. You will also see different payloads and some more linker development, not just in terms of defining the payload, but also releasing it more effectively within the tumor cell.

[Finally], you are going to see ADCs, once they get FDA approval in platinum resistance, get rushed to the [frontline setting]. We already talked about using an ADC for newly diagnosed patients with ovarian cancer and getting rid of [paclitaxel]. Only [42%] of patients respond to [paclitaxel]. [Almost] 60% of the patients were not benefiting from it, and it was having an impact on their quality of life.

If you had an ADC that did the same thing, [with] maybe an even a higher response rate, you combine it with carboplatin, and you are going to get the same response rate, if not better—and patients are going to have a better quality of life.