Y-90 Radioembolization Plus Pembrolizumab Generates Early Efficacy Signals in HCC

Matthew S. Johnson, MD

Pembrolizumab (Keytruda) in combination with TheraSphere™ Yttrium-90 (Y-90) radioembolization represents a potential combination treatment avenue for patients with poor-prognosis, advanced hepatocellular carcinoma (HCC), according to Matthew S. Johnson, MD.

The open-label, single-arm, multicenter, early phase 1 pilot HCRN GI15-225 study (NCT03099564) assessed the combination of pembrolizumab and Y-90 radioembolization in 27 patients with HCC with poor prognoses. In this study, the 6-month progression-free survival (PFS) rate was 55.6% (95% CI, 35.3%-74.5%), the median PFS was 9.95 months (95% CI, 4.14-15.24), and the median overall survival (OS) was 27.30 months (95% CI, 10.15-39.52). Furthermore, the overall response rate (ORR) among response-evaluable patients (n = 26) was 30.8% (95% CI, 14.3%-51.8%), and the disease control rate was 84.6% (95% CI, 65.1%-95.6%).1

“These are good results in this population that would otherwise not have a lot of hope,” Johnson said in an interview with OncLive®.

In the interview, Johnson discussed the unique features of the patient population that this pilot study enrolled; key efficacy and safety findings from this research; and how the design of this study demonstrates the importance of standardizing end points, patient populations, and language across HCC clinical trials.

Johnson is the Gary J. Becker Professor of Radiology Research, a professor of radiology and imaging sciences, and a professor of surgery at the Indiana University (IU) School of Medicine in Indianapolis.

OncLive: What was the rationale for the pilot study of pembrolizumab plus Y-90 radioembolization in patients with HCC?

Johnson: Radioembolization is effective in treating [patients with] HCC but less effective in patients with advanced disease, which has often been quantified through the Barcelona Clinic Liver Cancer [BCLC] staging and treatment algorithm. That advanced disease state includes portal vein involvement by tumor; ECOG performance status [PS] of 1 or greater; lymph nodes in the porta hepatis; and widespread metastases outside the liver. The most important of those [factors] that we’ve seen are extrahepatic metastases and portal vein involvement.

Our study was done in patients with advanced disease, which was defined as portal vein involvement, and/or multifocal disease, and/or diffuse disease or ill-defined disease in patients who did not have extrahepatic metastases. We didn’t want to include [patients with] widespread metastases because we didn’t think radioembolization would have an appreciable effect on the quality of life or OS of those patients. However, in patients who were likely to have a response to radioembolization, we had seen anecdotally and in previous retrospective studies that those patients appeared to have done better when we included an immuno-oncologic drug, [such as] pembrolizumab.

What patient population did this study observe?

This was a prospective, non-randomized, multicenter study [conducted] at 3 centers, [including IU. Patients [were] presented to our tumor board—we have an HCC conference every week, and we discuss all the patients who come in to our multidisciplinary treatment team. [IU], and I’m sure the other sites, evaluated these patients and determined whether they would be appropriate candidates [for the study].

Appropriate candidates for this study were patients with advanced disease as I’ve described. Importantly, they [could only have] up to Child-Pugh 7 [disease], [meaning patients with] either Child-Pugh 5 or 6 [disease], which would be Child-Pugh A, or Child-Pugh 7 [disease], which would be an early Child-Pugh B. If a patient had an ECOG PS of 0 or 1; Child-Pugh A status or Child-Pugh 7 score; had advanced disease; and had not been previously treated with radioembolization, they could be considered for our study.

What was the value of using 6-month PFS rate as the primary end point of this study?

I did not design the trial protocol, but 6-month PFS rate is an excellent end point. I agreed with it because [6 months was] a time [at which] we expected the patients to be alive. You have to choose a time to evaluate, and 6 months was an excellent time for this study because of how long we expected patients to survive, and [it was a time point where] we could [potentially] demonstrate a difference from previous studies. For example, in a study of patients with advanced HCC treated only with radioembolization, if patients had Child-Pugh B [disease], they wouldn’t have been expected to live much [longer than] 6 months. [Patients with] Child-Pugh A [disease] lived for approximately 13.5 months.

We expected our patients to be alive at [6 months, at which point] we could evaluate the effects of treatment after radioembolization. [By 6 months], the effects of that treatment modality had stabilized. It’s much more difficult to evaluate patients for radioembolization earlier than 3 months because of the continuing effect [of the therapy] upon the tumors at that point. [Six months] is a good time to evaluate because of [patients’] expected survival, their disease state, and the radioembolization effects, as well as because it’s [a standard time for investigators to follow]. All those aspects matter, [including] patient population and [treatment] expectations.

What is the importance of standardizing end points, patient populations, and treatment modalities during clinical research?

It’s important to agree on standard terminology, populations, treatments, and outcomes because otherwise, it’s almost impossible to evaluate the effects of a therapy. We’re early on in the [research regarding] combination therapy [involving radioembolization] for HCC. Radioembolization has been studied for many years; the immuno-oncologic agents have not been around for nearly as long. They are an impressive improvement on [historic therapies], but [they are] new.

We are getting better at understanding the differences [between patients with HCC]. The BCLC staging system [was impactful] when it came out, because it puts patients into readily defined populations. There’s wide variability in each of those populations, so we have to be even more detailed than what is allowed by that staging system, but in my opinion, it is the best we have for Western HCC. We have a staging system where we can [categorize patients]. We can go into subsets of that staging system, and then we can try to standardize therapies and all talk the same language.

Interventional radiologists approach HCC treatment differently vs medical oncologists, surgeons, and radiation therapists. [We all see] the same patients, but we all approach them from different points of view. When I talk about a patient with HCC, it’s important that when I’m talking to a medical oncologist, we both agree on the terms that we use. For example, people use portal vein thrombosis interchangeably with portal vein involvement. However, portal vein thrombosis occurs in a lot of different populations, including patients without cirrhosis, or frequently, patients with cirrhosis. [However, portal vein thrombosis] does not affect outcomes with treatments for HCC like it would if it were portal vein involvement and there was cancer within the vein. We need to agree upon how portal vein thrombosis or portal vein involvement [are defined].

In this study, we [evaluated patients with] portal vein involvement. That means the tumor has invaded the portal vein. That’s a much worse prognostic indicator than a patient with bland thrombus. You can analyze every aspect of the population definitions and look at their massive impact upon what we expect. For example, moving from Child-Pugh 5 to Child-Pugh 6 [status] likely has an impact, as does moving from Child-Pugh 6 to Child-Pugh 7 [status], which is why most medical oncologic studies exclude patients with Child-Pugh scores of 8 and above. The Child-Pugh status of 8 and above means a patient’s cirrhosis has a nonmalignant effect that is often worse than the malignancy. Ascites, encephalopathy, and liver failure have a massive impact on OS. You have to be able to evaluate separately the HCC and the cirrhosis, or the cause of the HCC.

Patients with hepatitis B frequently have cancer without cirrhosis. That’s another [disease feature] that affects therapies and is why perhaps [HCC treatments are] different in the West than the treatment algorithms used in the East, where hepatitis B is more prevalent. All those aspects come into play, and you have to understand them. It’s frequently difficult to discuss trials like this when it’s difficult to define the patient population and try to compare [it with those observed in other trials].

Whose responsibility is it to establish this standardization within clinical research in HCC?

I’m a big proponent of multidisciplinary discussions, and I’ve worked with a lot of medical oncologists, transplant surgeons, and radiation oncologists. I’m sure each of those specialties would agree with me and with other interventional radiologists that we need this [standardization]. The probable responsible parties would create a multidisciplinary task force to look at those things.

Another example of the importance of terms is that our study used RECIST v1.1 criteria as our method of evaluating outcomes. That’s a reading of images that considers the size of the tumors, but not their vascularity. Modified RECIST [mRECIST] v1.1 criteria accounts for vascularity. In interventional radiology, where what we do affects vascularity, in our studies, we prominently use mRECIST v1.1 criteria to evaluate tumors. A tumor that is the same size, but is dead, would have a complete response per mRECIST v1.1 criteria, whereas per regular RECIST v1.1 criteria, a tumor that used to be fully vascularized and now is dead but is the same size [as when it was vascularized] only achieves stable disease.

When you’re trying to use imaging outcomes, for example, ORR or time to progression, you need to know which of those modalities you are using to evaluate outcomes because they’re going to give different results. Then, when [a study shows an] excellent ORR and they’re using mRECIST v1.1 criteria, [that outcome will] likely be better than the ORR [observed with standard] RECIST v1.1 criteria.

We all should be using the same terminology. In this study, we used RECIST v1.1 criteria for the primary outcomes and mRECIST v1.1 criteria for the secondary outcomes. That’s the most appropriate. Use them both, and then be able to talk about them both and what they might mean.

What were the key efficacy and safety findings from the pilot study?

The key finding for efficacy was OS. The median OS was 27.30 months. That’s outstanding in patients with advanced disease. We don’t have another trial that was done with [the same] population [to compare outcomes]. However, we have other—predominantly retrospective,—studies that use different methodologies and terminologies where you get a gestalt that the OS wasn’t [as long as it was in this pilot study].

[A median OS of] 27.30 months in patients with multifocal cancer, which is maybe not as bad as those with ill-defined, invasive tumors, or those with portal vein involvement, is still excellent. We have an excellent outcome in the 27 patients who continued all the way through the trial. One patient died before the outcomes assessment, and that’s included in the OS.

[The combination] was safe, and 48.1% of patients had grade 3 or higher adverse effects [AEs]. That’s not awful, especially in the world of oncology, but [the rate of grade 3 or higher AEs] was substantially more than you would expect from radioembolization alone.

It appears we have demonstrated both safety and efficacy in this small trial, but we need to do more. However, like always, there will be a cost to doing combination therapy. The next stage of a trial [being conducted] at IU is investigating immuno-oncologic agents. However, when I see patients with portal vein involvement, diffuse disease that’s ill-defined, or multifocal tumors, it’s tough for me not to want to give them an immuno-oncologic agent at the same time I’m administering Y-90 radioembolization. I gained [enough knowledge] from this study that I’m fairly comfortable [with this regimen] and it looks attractive and effective, and it is safe, to the extent that it can be safe.

What AE management strategies are important to consider when treating patients with radioembolization?

If you do lobar radioembolization, [like many of the patients on this trial], of the entire right or left lobe with the tumor in it, [approximately] 1 out of 7 of those patients may have some degree of abdominal pain, nausea, or vomiting that might last a day or 2 after the procedure, sometimes longer. However, that’s it, it’s mild, and 1 out of 7 patients may be an overestimation. Approximately 50% to 70% of patients have some degree of fatigue that lasts a few days. That’s pretty much it for complications that occur regularly [with radioembolization].

The AE we’re all afraid of is nontarget radioembolization, where you put this high-activity radioembolic [agent] into a patient and it goes somewhere else. In the early days of radioembolization, that was not uncommon. However, [physicians previously] were putting the [radioembolic agents] inside a celiac artery at the beginning, and it would go everywhere, including the pancreas, stomach, etc.

Nowadays, with Y-90, radioembolization is an entire treatment plan. You evaluate the patients, you plan, you do the treatment, and then you follow them like in a medical oncologic [treatment] course. Because [the glass microspheres of Y-90] are so small, the chances of them refluxing back into the arteries you are putting them in are extremely small. For example, I’ve done approximately 900 Y-90 administrations, and 1 patient had nontarget radioembolization with that device. She had grossly abnormal anatomy, and I told her at the time that I might end up hitting her stomach, and I did. She needed medical therapy for an ulcer that I caused. Nontarget radioembolization is extremely rare [with Y-90].

The complication profile from radioembolization is not [large]. Regarding lab values that change, the most frequent is decreased white blood cell count, which is generally almost never clinically significant, but it does occur. We’re used immuno-oncologic [agents in this trial], so the possibility [of this AE] we were concerned about was clinically significant.

We were also concerned about the cumulative effect of the complications [with the combination therapy]. Increased bilirubin levels have also been reported with radioembolization. Bilirubin toxicity is common; any patient dying from cirrhosis and cancer may eventually have bilirubin elevation, so it’s tough to say whether the radioembolization or the progression of the disease causes that problem. It’s probably both, because radioembolization affects normal hepatic parenchyma. Hypervascular tumors suck out more of the radioembolic agent, so there’s a differential in how much each takes. At IU, we’ve done some studies in dosimetry, and it looks like patients [with HCC who receive lobar treatment] get approximately 2.5 to 3 times as much of the radioembolic agent into the tumors as they would the normal parenchyma. We don’t yet know where hepatic toxicity exists for a normal hepatocyte.

We also don’t know how much cirrhosis affects susceptibility to radiation injury. [Regarding] the combination, we were concerned about bilirubin toxicity, leukocytopenia, and any other combination of AEs that might occur. During therapy, 48.1% of patients developed a grade 3 or higher toxicity, and that was partially bilirubin toxicity. How much of that was due to the combination therapy, and how much of that was due to disease progression? Those are questions future studies should evaluate.

[The AEs observed in this study] didn’t look like they would prevent us from using this combination. The patients may have done worse without combination therapy. If I could, I’d recommend combination therapy to patients in this population now. However, it would be better to study [this combination in] bigger, more general populations and see the effects of the therapy.

How has this study moved the needle forward for the use of Y-90 radioembolization as a part of combination therapy in HCC?

[This study] didn’t stall [progress with Y-90 research] because it was prospective. Precious few prospective studies have evaluated combination therapy to this extent. We have other data demonstrating the benefit of radioembolization outside anecdotal and small studies. However, few studies include patient populations with ill-defined disease because they’re difficult to evaluate. If you cannot measure a tumor, how do you evaluate outcomes other than OS? We included those patients [in this study]. That’s an important point: Include patients who might otherwise [be excluded from] studies that can’t evaluate them.

[Additionally, the combination] was safe. Grade 3 and 4 AEs occurred, but only 1 patient died before the 6-month evaluation, and that’s in a patient population with advanced cancer. This is a safe combination, and it appears to be effective. I agree that [this study] moved the needle forward, and I believe that [effective] combination therapies are the next major [regimens to evaluate].

Doctors often want to protect their turf. They say, ‘We have this therapy, and everyone should get this therapy because this is what I do.’ That is terrible for patients, and it is the wrong way to move. The way patients get the best therapy is when everybody works together, brings their expertise to the table, and respects each other. If you do that, like we did here, patients live longer, and they live better. This study was another demonstration that multidisciplinary approaches [for treating] patients with cancer are essential.

Reference

Yu S, Yu M, Keane B, et al. A pilot study of pembrolizumab in combination with Y90 radioembolization in subjects with poor prognosis hepatocellular carcinoma. Oncologist. 2024;29(3):270-e413. doi:10.1093/oncolo/oyad331