REFINE-Lung puts pembrolizumab overtreatment under a randomized lens
Some patients with advanced cancer may be receiving more immunotherapy than they need. In advanced non-small-cell lung cancer, that concern now sits at the center of REFINE-Lung, a UK multicentre phase 3 study designed to test whether pembrolizumab can be given less frequently without sacrificing therapeutic benefit.
The trial arrives at a practical problem oncology has struggled to solve with standard trial logic. Conventional two-arm non-inferiority studies are slow and resource-heavy when the question is not whether a drug works, but how much of it is enough. For anti-PD-1 therapy, the issue is especially sharp because the treatment carries substantial cost, can affect quality of life, and may expose patients to avoidable toxicity when the dosing schedule is more intensive than necessary.
A single drug can be clinically effective and still be administered in a way that is harder on patients than it needs to be.
REFINE-Lung is built to answer that question directly. The study evaluates reduced-frequency pembrolizumab in advanced non-small-cell lung cancer and uses a multi-arm multi-stage response over continuous interventions design, or MAMS-ROCI, to identify the optimal dose frequency. Rather than forcing investigators to test one alternative schedule against standard care in a narrow, sequential fashion, the design is intended to compare multiple dosing approaches more efficiently and to determine whether a less intensive regimen can hold up in the clinic.
The underlying clinical logic is straightforward. Anti-PD-1 therapy does not need to be biologically maximal at every interval to remain effective, and the real-world burden of treatment extends beyond tumor control alone. Every infusion schedule carries consequences for time spent in care, treatment-related toxicity, and financial pressure on health systems and patients. If a lower-frequency schedule can maintain efficacy, the payoff would be measured not only in resource use but in the daily experience of people living with advanced cancer.
Pembrolizumab sits at the center of that debate because it is already widely used and embedded in treatment pathways for advanced disease. As with many checkpoint inhibitors, the key question is not whether the drug has activity, but whether its administration can be refined so that treatment intensity better matches clinical need. MAMS-ROCI is meant to make that refinement testable rather than speculative.
Why standard non-inferiority trials struggle with dose reduction
Traditional non-inferiority trials work well when the clinical question is cleanly framed around one alternative treatment versus a standard comparator. Dose-optimization questions are messier. Different schedules may all be plausible, and the goal is not necessarily to show superiority, but to identify the least burdensome regimen that still preserves acceptable outcomes. Testing those possibilities one by one can consume large numbers of patients and extend the time needed to reach a usable answer.
The REFINE-Lung design was created to address exactly that inefficiency. By using a multi-arm, multi-stage approach, it can evaluate reduced-frequency pembrolizumab more flexibly than a conventional two-arm study. The continuous-intervention framework is particularly relevant for a question like dosing frequency, where the biologic and clinical trade-offs do not sit at a single neat threshold.
This matters because dose optimization in immuno-oncology is not a side issue. If clinicians can safely reduce the frequency of anti-PD-1 therapy, the change would ripple through infusion centers, pharmacy budgets, and patient schedules. It would also fit a broader movement in oncology toward matching treatment intensity to what is actually necessary for disease control rather than defaulting to historical schedules that may be more conservative than evidence demands.
A template for immunotherapy optimization across tumor types
REFINE-Lung is not being developed in isolation. A similarly designed basket study in renal cancer and melanoma is underway alongside it, extending the same concept across disease settings where immunotherapy is central to care. Together, these studies are meant to test whether a shared trial architecture can answer dose-optimization questions across multiple cancers, not just one.
The appeal of that approach is obvious in principle. Many new or existing agents are used for long periods, and each long-term treatment course raises questions about how much drug exposure is required for benefit. A design that can be adapted to dose, frequency, or duration may save time while still generating clinically useful evidence. In a field where new therapies are often evaluated for efficacy but not fully recalibrated for minimal effective use, that shift is overdue.
The biology also supports the question. Checkpoint blockade works by releasing immune restraint, not by directly killing tumor cells in the way cytotoxic chemotherapy does. Once the immune response is activated, the marginal value of continuing treatment at the same cadence may not scale linearly with more frequent dosing. That does not prove a reduced schedule is sufficient, but it makes the hypothesis credible enough to test rigorously rather than assume away.
For patients, the implications are immediate. Reduced-frequency pembrolizumab could mean fewer treatment visits, less disruption to daily life, and potentially less toxicity exposure if the schedule proves non-inferior in the relevant outcomes. For health systems, it could mean lower drug use and more efficient deployment of infusion capacity. The unanswered piece is not whether dose optimization is worth pursuing; it is whether the trial architecture can identify the right balance with enough precision to change practice.
The real test is whether less frequent treatment still holds the line
REFINE-Lung is a study about restraint as much as it is about efficacy. By challenging the assumption that every approved immunotherapy schedule is already optimized, the trial turns a familiar treatment into a question of calibration.
If the reduced-frequency approach performs well, the result would do more than adjust pembrolizumab scheduling in advanced non-small-cell lung cancer. It would validate a trial framework built for de-escalation questions that conventional non-inferiority studies handle poorly. If it does not, the field still gains a clearer boundary around where treatment can no longer be safely thinned out.
The study’s importance rests on that boundary setting. Dose reduction in immunotherapy has been difficult to study because the relevant outcomes are tied to long follow-up, practical feasibility, and the risk of under-treating patients who still need full exposure. REFINE-Lung does not eliminate those challenges. It gives them a more efficient structure and asks whether the standard rhythm of pembrolizumab is truly necessary for every patient enrolled in the advanced disease setting.
That answer will matter well beyond one tumor type, because the same question now runs through modern oncology: when a drug is effective, how often does it actually need to be given?