CD24 Expression Confers Resistance to Anti–PD-1 Therapy in NSCLC, RET-TKI Dose Reductions Do Not Compromise Outcomes in RET+ Disease

CD24 has emerged as a target of interest for combination therapies with immunotherapy in non–small cell lung cancer (NSCLC) and dose reductions of RET TKIs may be able to improve patient tolerability to these agents without compromising efficacy, according to findings from two studies presented by Daniel Boiarsky, MD, and William Phillips, MD, respectively, during the OncLive National Fellows Forum: Lung Cancer.^1,2

“The Fellows Forum was great,” Phillips, a fellow in the Thoracic Oncology Senior Fellowship Program at the University of Colorado Anschutz in Aurora, said in an interview with OncLive®. “Coming a day early to [the 2026 ASCO Annual Meeting] was a nice way to meet a lot of peers who have similar interests and have some fun. It’s also a nice way to meet some of the more senior faculty in the field who have more experience and a lot of knowledge to share about research. I really enjoyed getting to meet people, getting to hear some interesting science, and starting off the meeting off in a really positive way.”

Could CD24 represent a resistance marker and candidate target in NSCLC?

Boiarsky and coauthors set out to identify novel biomarkers and therapeutic targets capable of extending the benefit of anti–PD-1 therapy beyond the minority of patients who currently respond.¹

"Although immune therapies can have substantial benefit in a small proportion of patients, most patients don’t benefit, and unfortunately we haven’t really been able to make too much progress on the initial successes in the field," Boiarsky said in an interview with OncLive®. "The motivation for our study was to really try to identify novel biomarkers and therapeutic targets that can augment responses to anti–PD-1 therapy in metastatic non–small cell lung cancer."

The team used exceptional responders to anti–PD-1 therapy as a discovery engine. "The idea behind the study was to take exceptional responders to anti–PD-1 therapy in lung cancer and use them as a discovery engine to identify novel targets and biomarkers," Boiarsky explained.

The discovery cohort pooled 367 patients with metastatic NSCLC treated with PD-1 agents with/without CTLA-4 blockade across cohorts gathered from Yale, Stand Up To Cancer, and Samsung Medical Center, with pretreatment biopsies profiled by bulk RNA sequencing. Patients were stratified into long-term responders (LTR; progression-free survival [PFS] ≥ 36 months; n = 27), short-term responders (STR; PFS 6-36 months; n = 122), and primary progressors (PP; PFS < 3 months; n = 218).

Differential expression analysis showed that a number of tumor-associated macrophage (TAM) genes, including CD206 and CD163, were upregulated in LTRs. "Somewhat surprisingly, we saw that a number of TAM genes were upregulated in long-term responders, including genes which are typically thought to mark immunosuppressive macrophages, like CD206 and CD163," Boiarsky said. Conversely, CD24 was upregulated in the PP group.

Findings were validated in a Yale cohort (n = 57) profiled with multiplex immunofluorescence and spatial transcriptomics, plus 2 additional validation cohorts for the CD24 signal (Greece, n = 63; University of Queensland, n = 26). In the Yale validation cohort, median PFS was 2.3 months (95% CI, 2.1-3.0) in the PP group (n = 24), 11.1 months (95% CI, 9.9-13.7) in the STR group (n = 23), and not reached (NR; 95% CI, 51.8-NR) in the LTR group (n = 10); best overall response of a complete or partial response was achieved in 4.2%, 47.8%, and 70.0% of these groups, respectively.

Density of CD14-positive TAMs was significantly higher in LTRs than in PP or STR patients, both overall and within the tumor compartment specifically. "We saw that these long-term responders...had increased density of CD14-positive macrophages, specifically close to their tumor cells," Boiarsky said. Gene set enrichment analysis further showed that inflammatory and antigen-presentation programs, including interferon-γ response, TNF-α signaling via NF-κB, and antigen processing and presentation, were enriched in LTRs across tumor, immune, and macrophage compartments.

On multivariable analysis adjusting for sex, histology, line of therapy, and chemotherapy exposure, cancer-cell CD24 expression was independently associated with worse outcomes in the TMA471 (HR, 1.3; 95% CI, 1.0-1.8; P = .04) and TMA4301 (HR, 1.9; 95% CI, 1.1-3.2; P = .015) validation cohorts. CD24 expression was also higher in tumor tissue than in matched normal tissue (P < .0001), and a high tumor-to-normal CD24 ratio was associated with worse survival after surgical resection (HR, 1.84; 95% CI, 1.16-2.94; P = .010; n = 55). By contrast, tumor expression of CD47 did not associate with resistance to anti–PD-1 therapy in either cohort.

"To our knowledge, this is really the first gene or biomarker that’s been identified in lung cancer independently across multiple cohorts, whose overexpression is associated with resistance to anti–PD-1 therapy," Boiarsky said, noting that CD24 is already being pursued therapeutically in other tumor types. Investigators are now working to validate the CD24 association at the protein level and to determine whether CD24 modulates adaptive immunity independent of its effects on macrophages.

Looking ahead, Boiarsky said the field’s inability to move beyond early anti–PD-1 successes stems partly from unselected trial designs. "We think future trials really need to have a biomarker-selected approach," he said. "I think this is one of the reasons why we haven’t been able to move the field forward to date."

How do dose reductions in RET-Positive NSCLC affect efficacy?

In a separate presentation, Phillips and his coauthors examined whether dose reductions of RET-directed TKIs such as selpercatinib (Retevmo) and pralsetinib (Gavreto) affect real-world outcomes in advanced RET-positive NSCLC.² "This study was formed in collaboration with [a patient advocacy group], and they basically said, ‘Why don’t we look at how dose reductions have impacted people in the real world?’" Phillips explained.

The multicenter retrospective cohort study, conducted across the University of Colorado and Mass General Brigham Cancer Institute, included 46 adult patients with advanced RET-positive NSCLC and no prior RET inhibitor exposure who were treated with selpercatinib (76.1%) or pralsetinib (23.9%). Dose reduction, which was defined as a reduction to 75% or less of the FDA-approved dose, occurred in 19 patients (41.3%).

Line of therapy was the only baseline characteristic that differed significantly between groups: patients who underwent dose reduction were more likely to have received RET-TKI therapy in a later line compared with in the first-line setting, at 78.9% vs 21.1%, respectively, a factor that was adjusted for in the time-to-event analyses. The most common reason for dose reduction was a laboratory abnormality (58%), including myelosuppression, transaminitis, hyperkalemia, and lymphopenia, more often than a chronic subjective toxicity such as fatigue or arthralgia.

At a median follow-up of 38.7 months, dose reduction was not associated with inferior outcomes: the HR for PFS was 0.72 (P = .46) and the HR for time to next systemic therapy was 0.62 (P = .26). Rates of CNS (central nervous system) progression were also similar between groups regardless of dose-reduction status (P = .445).

"Dose reductions of RET-targeted therapy were common; [approximately] 40% of patients...required a dose reduction," Phillips said. "Patients who had those dose reductions did not seem to have worse outcomes, and I think this provides a lot of reassurance for patients who may be having side effects that are cumulative, like tiredness, GI upset, [or] swelling of the legs...that if I need to go down on my dose in order to stay on my therapy with a better quality of life, we’re able to do that with some data saying that it’s safe and you’re not compromising those outcomes from a cancer control perspective."

Investigators noted that current trial designs and adverse effect reporting frameworks often fail to capture chronic, low-grade toxicity burden or real-world dosing patterns, and pointed to the FDA’s Project Optimus initiative, which mandates evaluation of multiple dose levels prior to phase 3 development, as a mechanism to shift the field from a maximum-tolerated-dose paradigm toward an optimal biological dose.

Phillips underscored that real-world data of this kind serves a complementary role alongside clinical trials. "Real-world evidence gives you the ability to include a more diverse patient population who reflects who we’re seeing in clinic...it lets you ask more creative questions that are really patient facing," he said. He also pointed to the study’s origins as a lesson in early patient engagement: "This was a really good learning opportunity for the importance of getting patients involved early on in your study design, because they’re the ones on the receiving end of treatments and side effects."

References

1. Boiarsky D. Tumor-associated macrophages and tumor cell CD24 expression define opposing axes of long-term response and resistance to PD-1 blockade in metastatic non-small cell lung cancer (mNSCLC). Presented at: OncLive National Fellows Forum: Lung Cancer; May 28, 2026; Chicago, IL.
2. Phillips WJ. RET-y or not? Clinical impact of targeted therapy dose reductions in advanced RET+ NSCLC. Presented at: OncLive National Fellows Forum: Lung Cancer; May 28, 2026; Chicago, IL.