2 Clarke Drive
Suite 100
Cranbury, NJ 08512
© 2024 MJH Life Sciences™ and OncLive - Clinical Oncology News, Cancer Expert Insights. All rights reserved.
Brian Gastman MD, discusses how findings from CheckMate76K contributed to the rationale for a biomarker analysis in that trial’s patient population, key findings from this biomarker analysis, and how these findings may inform further research regarding the benefits of immunotherapy in earlier-stage disease.
The first comprehensive biomarker analysis conducted in patients with stage IIB or IIC melanoma who received adjuvant nivolumab (Opdivo) demonstrates predictors of relapse-free survival (RFS) benefit in this population, according to Brian Gastman MD.
The phase 3 CheckMate76K trial (NCT04099251) randomly assigned patients with resected stage IIB or IIC melanoma with wide local excision and negative sentinel lymph node biopsy 2:1 to receive either nivolumab or placebo for 12 months. At a median follow-up of 15.8 months in the nivolumab arm and 15.9 months in the placebo arm, the median RFS was not reached ([NR]; 95% CI, 28.5 months to not assessed [NA]) with nivolumab (n = 526) vs NR (95% CI, 21.6 months-NA) with placebo (n = 264; HR, 0.42; 95% CI, 0.30-0.59; log-rank P < .0001).1,2
The CA209-76K study, an exploratory biomarker analysis of the CheckMate76K population showed that the RFS benefit with nivolumab was consistent across all biomarker subgroups.2 Additionally, patients in the nivolumab arm with higher interferon gamma (IFNγ) signatures, tumor mutational burden (TMB), and CD8 immunohistochemistry scores, as well as those with lower C-reactive protein levels, experienced numerically longer RFS outcomes vs those in the placebo arm, with respective hazard rations (HRs) of 0.26 (95% CI, 0.14-0.48), 0.35 (95% CI, 0.19-0.62), 0.23 (95% CI, 0.12-0.43), and 0.30 (95% CI, 0.17-0.51).
“The need for these biomarkers as prediction tools has never been more important,” Gastman said in an interview with OncLive® during the 2023 ASCO Annual Meeting.
In the interview, Gastman discussed how findings from CheckMate76K contributed to the rationale for a biomarker analysis in that trial’s patient population, key findings from this biomarker analysis, and how these findings may inform further research regarding the benefits of immunotherapy in earlier-stage disease.
Gastman is surgical director of the Melanoma & High-Risk Skin Cancer Program in the Department of Plastic Surgery at the Beachwood Family Health Center, Cleveland Clinic, Ohio.
Gastman: Immunotherapy and some other systemic therapies have revolutionized the treatment of [patients with] melanoma. Nobel Prize–winning discoveries have occurred around the subject, and patients, [such as those with metastatic, and now, high-risk stage III disease], who have never been cured in the past are now being cured or are at least [experiencing] significant improvements in long-term survival. The question becomes: Can we also help patients with lower-risk stages of disease including IIB and IIC? That’s the highest end of the lower stages.
Data presented [from] a different trial [showed] that immunotherapy given systemically improves outcomes after surgery in those patients. An alternative drug to [the 1 used in that trial] is nivolumab, which is what was used in [CheckMate76K]. The question was: Will it add efficacy in terms of RFS? It’s been presented internationally that we saw a signal.
[However], most of these patients will likely benefit from surgery, some may benefit from adding immunotherapy, and some might not benefit from either. These drugs are expensive, so there’s financial toxicity, and patients can have long-term and even permanent adverse effects. Are we overtreating patients? Are there some patients who would benefit from surgery alone or wouldn’t benefit from everything we’ve given them so far?
To better rationalize the patients we treat and why we treat them, we’re looking for some type of predictive tool. A biomarker is 1 data piece, and by itself, will likely not give us that information. However, a cluster or plethora of these biomarkers, likely down the road using machine learning because there’s so much input data, will provide a predictive tool. The beginning of that is what we presented at the plenary session [at the 2023 ASCO Annual Meeting].
This is an extension of what was presented [from the CheckMate76K trial] from an RFS perspective. It was part and parcel of the same trial. Patients who were biopsied [and were diagnosed with] melanoma went on to have wide local excision sentinel node biopsy. If the sentinel node biopsy was negative and [their disease] stage was IIB or IIC, they [were randomly assigned 2:1] to have nivolumab or placebo control. This [trial] was double blinded and randomized.
An RFS benefit was seen [with nivolumab]. It was seen in a rather short order. During the trial, standard biometric data were collected, and blood was also collected from these patients. Tumor was collected as well. Various biomarker analyses occurred. Initially, standard biomarkers were assessed, biomarkers that you would typically see in [melanoma] trials over the past few years, and most of those are the biomarkers we presented.
We saw some association [between nivolumab treatment and], for example, TMB, which is not uncommon in these trials. Another example is, some of the immune characteristics [of patients in the nivolumab arm] were pro-inflammatory, such as IFNγ signatures. These by themselves are not necessarily novel, because we might see them in a stage IV or stage III trial.
What is novel is [seeing them] in the nonmetastatic setting. In theory, although these patients have risk for metastasis, as far as we know, based on imaging as well as sentinel node biopsy, [their disease has] not left the primary tumor site. And yet, we’re seeing the results we saw clinically and similar biomarker data that we would have expected in the higher-stage setting, where there are known metastases.
They underline the complexity of [biomarker testing]. Especially when you add [findings such as] transcriptomic data, which [was not presented], it’s like drinking from a fire hose because tens of thousands of data points are being brought in about thousands of patients. To sift through all that, machine learning and artificial intelligence are needed to separate the chaff from the wheat. We have many issues in the field. We have several different drugs. We don’t know which drugs or which combinations of drugs [should be given to] which patients. In the setting where patients might be cured with surgery alone or might not be cured even with surgery and immunotherapy, being able to help these patients using biomarker correlative-based analyses is ever more important. The field keeps getting more complex.
We did not present transcriptomic data, but there may be some interesting information within the tumor itself, potentially also in the peripheral blood of these patients, at a basal layer in terms of RNA transcription, etc. That may need to be combined even with the simple clinical characteristics of these patients to ultimately come up with the best tool. We don’t understand why the thicker tumors did not spread to the lymph nodes in any of the patients in this trial. We could use this trial to figure out which patients should or shouldn’t be treated and to better understand the biology of [their disease] to best come up with the next generation of treatments for this population.
Finally, we’re talking about stage IIB and IIC. There must be [also be discussion around] stage IIA and stage I. When will we start thinking about those kinds of cancers for these kinds of treatments? That would be a lot of overtreatment if we treat all [those patients] the same with these expensive, potentially toxic therapies. However, trials like this may allow us to uncover which [of those] patients would benefit from these types of therapies.
That’s 1 of the biggest excitements of doing this trial. I’m helping [patients with stage IIB and IIC disease], and it opens the door for the patients with lower-stage disease, which make up [most patients with] melanoma. Interestingly, the lower stages, by definition, are patients who either had a sentinel node biopsy that was negative or who didn’t meet criteria for sentinel node biopsy. The data have been clear that most [melanoma] deaths, even though this is the lower-risk population, come from these patients. Most melanoma deaths come from patients with stage I and II disease, but we’re finally just starting to get a toe in the water of how to start treating them in a way to prevent those deaths in the future. These kinds of trials are that toe in the water.
In the world of adjuvant therapy, there’s also neoadjuvant therapy. We’ve only been able to test [for biomarkers] where there’s measurable, macroscopic metastases. In the nonmetastatic world, most of the primary tumor is biopsied away. In most cases, there’s no visual melanoma to do a neoadjuvant approach to evaluate pathologic response. Most of the pathology on re-resection is, at best, a little bit of cancer left, and at worst, just scar from previous surgery, even though there must still be cancer in there, otherwise we wouldn’t go back and operate in those patients. Designing a trial around patients in the neoadjuvant setting would open the doorway for the promise of neoadjuvant therapy, like what we’re seeing in [patients with] macroscopic [disease].
We want to do better than we have been doing, because some patients [develop progressive disease] despite surgery and despite adjuvant PD-1 inhibition. Figuring out a way to [administer immunotherapy] neoadjuvantly could increase [the number of patients who benefit from immunotherapy]. There’s some communication and activity on my part, my friends’ parts, and my colleagues’ parts toward that goal. If we can somehow get neoadjuvant therapy into this population, we may see a major shift in the management of melanoma.
Disclosures: Dr Gastman reports stock and other ownership interests with Castle Biosciences; consulting or advisory roles with Castle Biosciences and Quest Imaging; speakers’ bureau participation with Castle Biosciences; research funding from Alkermes, Instil Bio, Merck, NeoImmuneTech, and Quest Imaging; and travel, accommodations, and expenses from Alkermes.