Genomic, Immune, Epigenetic Features Drive Treatment Decision-Making in High-Risk NSCLC Subtypes

Aggressive molecular subsets of non–small cell lung cancer (NSCLC), including STK11-, KEAP1-, and SMARCA4-mutated disease, as well as NUT carcinoma, continue to present substantial therapeutic challenges despite advances in precision oncology, according to Benjamin Herzberg, MD.

During an interview with OncLive at the 20th Annual New York Lung Cancers Symposium®, Herzberg provided a detailed overview of the biologic features that define these high-risk tumors, the limitations of current treatment approaches, and the emerging strategies aimed at addressing profound unmet needs for patients with these more aggressive subtypes. He also underscored the importance of recognizing these distinct subtypes early in the diagnostic process.

“If you have a patient who has an aggressive mutation and does poorly with chemo-immunotherapy [chemo-IO] and you don’t have that much to offer next, knowing they have that mutation can open up a world of trials for that patient,” Herzberg explained during the interview. “This really underlines the importance of comprehensive genomic testing in all of our patients.”

Herzberg is a medical oncologist at Herbert Irving Comprehensive Cancer Center and assistant professor of medicine at Columbia University Medical Center in New York,

OncLive: How does the presence of STK11-, KEAP1-, and SMARCA4 mutations influence treatment strategies for NUT carcinoma and other aggressive lung cancer subtypes?

Herzberg: [We know] that STK11, KEAP1, SMARCA4 confer particularly poor prognoses when given chemotherapy, immunotherapy, and radiation therapy, especially when they’re found in association with a KRAS mutation. The most important thing to know is that when I see these mutation patterns, I already have a sense that I’m worried about these patients with our conventional therapies. That’s one important [consideration].

NUT carcinoma and SMARCA4-undifferentiated tumors are tumors that we should suspect when a pathologist says, ‘This looks like lung cancer, but it looks [poorly] differentiated. It looks maybe a little bit like squamous, but it doesn’t express all the squamous markers. We’re not sure.’ That’s something like a NUT carcinoma, [which is] also found a lot in younger patients or never-smokers. It’s something that we should be on the lookout for, because the treatment [of patients with this subtype] is quite different. Their prognosis also tends to be very poor, so we have to identify these patients, because there are some trials for them.

SMARCA4-undifferentiated tumors probably evolved from an NSCLC to be even [clinically] worse. There’s been a mishmash of strategies [to manage this subtype] and it’s been hard to tell exactly what to do with these patients. Immunotherapy sometimes seems to help, [but] sometimes it doesn’t. There’s a lot of research going on in this area.

What do these lung cancer subtypes have in common? How do we currently approach the What are some of the current treatment modalities for these aggressive lung cancer subtypes, and what are their limitations?

The biggest underlying commonality of all these tumors is that they seem to be what is sometimes referred to metaphorically as immune cold. They don’t really seem to activate immune responses in patients. The treatments that we have as of today to generate more immune responses in patients, like immune checkpoint inhibitors— [which] are classic treatments for lung cancer—work less well in this patient population. Therefore, the biggest current strategy that’s been explored is to try to give these patients a CTLA-4 inhibitor in conjunction with a PD-1 inhibitor.

We’ve had many trials in lung cancer evaluating CTLA-4 plus PD-1 inhibition plus chemotherapy for all patients, and I would say that those trials are positive trials when you compare [the chemo-IO regimen] against chemotherapy. These strategies are approved — they’re in FDA [labels], [and] they’re in the National Comprehensive Cancer Center guidelines. However, [a] common issue we face is who to choose for these augmented strategies. The adverse effects are greater [and] it’s not 100% clear that patients really do better long term.

Several groups, most notably Ferdinandos Skoulidis, MD, PhD, MRCP of the University of Texas MD Anderson Cancer Center in a large paper that involved a large portion of [the] lung cancer community, showed that there’s reason to believe that in these poor-prognostic subtypes with STK11, KRAS, andKEAP1 [mutations], giving a CTLA-4 inhibitor may confer particular benefit. What I do today is I don’t say I have to give augmented immunotherapy for these patients. It sort of puts the thumb on the scale when I was already thinking about giving a CTLA-4 inhibitor, [making me] be more proactive when I see one of these poor-prognostic mutations.

We are testing this hypothesis prospectively in a phase 3b study called the TRITON study [NCT06008093], which randomly assigned patients to PD-1 inhibition alone vs PD-1/CTLA-4 inhibition if they have one of these poor-prognostic mutations.2 We’ll see what the results are there, [but right now], it’s a soft factor in my clinical decision-making, where I think to augment the immunotherapy that I give my patients.

What are some upcoming clinical studies evaluating novel therapies for STK11-, KEAP1-, and SMARCA4-mutated lung cancers and NUT carcinoma?

Other than CTLA-4 plus PD-1 [inhibition], which has some evidence but not tremendous evidence, everything else we’re talking about for these tumor types is really a new therapeutic strategy. That’s daunting to some extent, because none of these drugs have been approved, but exciting.

STK11 is [a] mutation that lowers something called cGAS–STING signaling in the tumor. This is an immune pathway that is supposed to respond when your cells detect viral DNA around, but it is part of the anticancer immune response as well. STK11-mutated tumors can’t mount this response. There’s exciting work on STING agonists to try to reactivate this pathway. There’s been other work with other epigenetic therapies to try to make these cold tumors hot.

KEAP1 really causes a tolerance in tumors to stress of all kinds. KEAP1 tumors don’t care about radiation, they don’t care about chemotherapy, they don’t care about immunotherapy. They do this by shrugging off oxidative damage. A lot of the strategies that we’re seeing now for KEAP1-mutated disease are attempting to take advantage of this fact. Glutamine, for example, is important for soaking up reactive oxygen species, so there’s an investigator-initiated trial at NYU Langone Cancer Center being conducted by my colleague Salman R. Punakar, MD, [evaluating] glutamine-directed [therapy]. We’ve seen other strategies, like activators of KEAP1, [in development] to address this.

For SMARCA4-deficient tumors, it turns out that there is one other protein in the cell that can compensate for the loss of SMARCA4. That’s a protein [produced from the] SMARCA2 gene. [The] therapeutic strategy [for this tumor subtype] has been to knock out SMARCA2 at the same time. This strategy has been tested and shown responses in patients. Approximately 25% of patients with these mutations will experience major tumor shrinkage. Therefore, SMARCA4-directed therapy might be another synthetic lethal therapy that we add to our toolkit going forward.

Is there anything else you’d like to highlight about unmet needs, especially in NUT carcinoma?

[Treatment for] NUT carcinoma is a huge unmet need. My colleague at Dana-Farber Cancer Institute, Jia Luo, MD, along with several colleagues across the country, [is] working with BET inhibitors, a type of therapy that directly addresses the fusion protein in that cancer type. That’s a big unmet need.

To a certain extent, all of these are unmet needs, because we know these patients have worse outcomes with our standard-of-care therapies. However, if we administer some targeted therapies instead of chemo-IO, we see some benefits with that. What’s really going to break the bank here is novel strategies.

Are there populations in whom these mutations and their epigenetic features are more common?

For NUT carcinoma, these tend to occur in both smokers and never smokers. I think about them when I see a squamous-like carcinoma happening in someone who maybe doesn’t have a long smoking history or [in whom the cancer] is behaving extremely aggressively. That’s something that makes me nervous. They’re often confused for head and neck cancers. They’re often confused for lung cancers. We have to have a high index of suspicion to ask a pathologist to test for a NUT fusion protein in this cancer, or if there is anything that makes them suspect that this is not a normal squamous cancer.

For these other tumor types, [I would emphasize the] importance of getting comprehensive genomic sequencing for all patients with lung cancer. Even if only a small slice of the [mutations] we find are targetable today, more and more will be targetable in the future.

References

1. Herzberg B. The Bad Boys of Lung Cancer: NUT Carcinoma, SMARCA4-Deficient, STK11/KEAP1 Mutated. Presented at: 20th Annual New York Lung Cancers Symposium; November 15, 2025; New York, NY.
2. A study to investigate the efficacy of durvalumab plus tremelimumab in combination with chemotherapy compared with pembrolizumab in combination with chemotherapy in metastatic NSCLC patients with non-squamous histology who have mutations and/or co-mutations in STK11, KEAP1, or KRAS (TRITON). Clinicaltrials.gov. Updated October 27, 2025. Accessed November 18, 2025. https://www.clinicaltrials.gov/study/NCT06008093