Precision Medicine with Duvelisib and the Evolving Landscape of Lymphoma

The treatment landscape for peripheral T-cell lymphoma (PTCL) is currently defined by a critical paradox: while our understanding of the disease’s biological heterogeneity is expanding rapidly, the clinical toolkit remains frustratingly limited. For years, a "one-size-fits-all" chemotherapy approach has served as the backbone of care, yet many patients do not experience durable remissions. As the oncology community moves toward more refined, biologically matched therapies, duvelisib (Copiktra), a dual PI3K inhibitor, has emerged as a compelling candidate for addressing these unmet needs, based on findings from the expansion phase of the phase 2 PRIMO trial (NCT03372057)published in the Journal of Clinical Oncology.¹

In a recent conversation with Steven M. Horwitz, MD, the rationale behind duvelisib’s innovative dosing strategy and its efficacy in specific subtypes like angioimmunoblastic T-cell lymphoma (AITL) took center stage. A pivotal component of its clinical application is the transition from a high-intensity lead-in dose of 75 mg twice daily to a 25 mg maintenance dose after 2 cycles. This strategy was born from clinical logic: the higher initial dose is essential for rapid target engagement and early disease control in aggressive lymphomas, while the step-down to a lower dose aims to mitigate long-term adverse effects (AEs), such as infections, which often limit a patient’s ability to remain on therapy.

Beyond dosing, the biological interplay between duvelisib and the tumormicroenvironment suggests a new frontier in T-cell malignancy management. Horwitz highlighted the marked sensitivity of AITL to PI3K inhibition, suggesting that duvelisib’s dual inhibition of PI3K-δ and PI3K-γ may drive both direct antitumor activity and broader remodeling of the tumor microenvironment through effects on malignant T cells, macrophage repolarization, and immune signaling pathways, potentially contributing to the notably high complete response rates observed in this subtype.

CancerNetwork: What clinical unmet need in PTCL is duvelisib attempting to address?

Horwitz: The main, or the overriding clinical need in PTCL or T-cell lymphomas is simply more effective therapies. We have a handful of therapies that work okay or only very well in certain subtypes. We have chemotherapies that work okay, but there is a significant proportion of patients who are not cured or not maintained with disease control on the available therapies. If anything, for FDA-approved agents, we’ve lost one in the last few years rather than gained. We need more tools to treat patients. That is the overriding need.

The other area where T-cell lymphoma is heading, as we try to develop better therapies, is moving away from the historic "one-size-fits-all" approach. Chemotherapy has been the main tool, and that has been applied broadly across what we increasingly understand are very biologically heterogeneous diseases. We need new therapies, but equally important, we need therapies whose mechanisms we truly understand—where they work, where they may fall short, and ultimately why. This will help us match the therapies better to patients and eventually build better platforms, capitalizing on that knowledge as opposed to treating everyone the same and hoping for good results here and there.

In the treatment course, a high-intensity lead-in dose of 75 mg twice daily for 2 cycles was utilized, followed by a 25 mg twice-daily maintenance dose. How does this specific transition address the historical challenge of balancing rapid initial disease control with the management of late-onset adverse effects compared with continuous low-dose strategies?

When we did the phase 1 study of duvelisib, it was a dose-escalation study, and it treated all lymphomas.² What happened is we were seeing efficacy at 15 mg and 25 mg. There was a waitlist in patients with indolent lymphomas, primarily CLL and follicular lymphoma, who could wait and take those spots, and the patients with more aggressive disease, which included the patients with PTCL, could not get on study.

There was a decision made in the phase 1 to branch off patients with low-grade more indolent lymphoma at 25 mg twice a day. That ended up being the approved dose for that. The dose-escalation phase primarily enrolled patients with aggressive lymphomas. In the phase 1 experience, nearly all patients with T-cell lymphoma received higher dose levels—most commonly 75 mg twice daily, corresponding to the maximum tolerated dose or above—with only a small number treated at intermediate doses such as 50 or 60 mg twice daily, forming the basis of the initial efficacy dataset.

We had this phase 1 with [an approximately 50%] response rate, but no data at lower doses. We had subsequently done a combination study with romidepsin [Istodax], which is separately published as an investigator-initiated study where we looked at some combinations, but we also did some single-cycle lead-ins to get some biopsies and do some corelative science. We had some patients at 75 mg and 25 mg, and in that small cohort, we didn’t see a difference. Going forward, we predicted that it wouldn’t matter, 25 mg vs 75 mg.

Higher doses, such as 75 mg twice daily, were thought to provide more sustained dual inhibition of both PI3K-δ and PI3K-γ, whereas PI3K-δ inhibition can be achieved at lower dose levels. Importantly, much of the prior clinical experience in lymphoma had been with predominantly PI3K-δ–selective inhibitors, making duvelisib’s combined γ/δ inhibition potentially distinct from earlier agents.. We didn’t know that it mattered. As part of this, there was a pilot to try to dose optimize and try to understand. What it turned out from that dose optimization cohort is that there were more patients who [experienced progression] early at 25 mg. This is why, if you look at that dose optimization cohort, more patients came off early at 25 mg.

Pharmacokinetic and pharmacodynamic analyses suggested that lower dose levels may not have achieved sufficiently rapid or sustained target engagement, particularly for downstream immune markers such as interferon-γ. In contrast, the 75 mg twice-daily dose appeared to provide more robust biologic activity, potentially contributing to the clinical responses observed early in development.

Circle back: in the phase 1 study, we also did an exploratory PET scan at week 3 or 4 in the first cycle. We were worried that if patients with aggressive disease were progressing, we wanted to recognize that early. It turned out, in that small experience, almost everyone who responded [did so] by that first PET scan. The response readout was early, and that was primarily at 75 mg.

Then it was a clinical logic decision, not a data-driven decision, to say: "Okay, if we need 75 mg to get early target engagement and early disease control, but we know there are already long-term toxicities with this, we think those are going to be worse with 75 mg." The decision was made that once we had good disease control, we would reduce to 25 mg to try to minimize some of the long-term toxicity, which would limit patients' ability to stay on treatment. That ended up working out. The initial decision about the initial dosing was data-driven, and then the decision of how to optimize that for long-term safety was a clinical logic decision, and that is how that strategy played out.

The trial demonstrated a complete response rate of approximately 51.4% in patients with angioimmunoblastic T-cell lymphoma (AITL), a biologically distinct subtype within the T-follicular helper (TFH) cell lymphoma spectrum. What biological or microenvironmental characteristics of AITL do you believe contribute to this heightened sensitivity to dual PI3K inhibition?

We don’t know the answer to that in terms of exactly what the cause is, but as you alluded to, AITL or follicular helper T-cell lymphomas certainly are more microenvironment-rich than malignancies like anaplastic large cell lymphoma.

We know that PI3K gamma inhibition leads to some changes in, for example, macrophage repolarization, and in some early preclinical studies that we conducted, there were cell lines that responded and cell lines that did not respond. Therefore, we think there is a combination of cell-autonomous and non-cell-autonomous effects. One potential explanation is that AITL, which is characterized by a highly complex and immune-rich tumor microenvironment, may be particularly sensitive to therapies that modulate microenvironmental signaling. The notably high response rates observed in this subtype raise the possibility that PI3K-γ inhibition, in addition to PI3K-δ blockade, contributes meaningfully to duvelisib’s activity in this disease.

We are likely getting some "on-tumor" effects and some microenvironment effects. That is a hypothesis based on what we know about that disease and the clinical data, but we do not know for certain that there is a specific target that we are engaging that makes a difference in AITL as opposed to PTCL-NOS or other subtypes.

Reference

1. Mehta-Shah N, Zinzani PL, Jacobsen ED, et al. Duvelisib induces deep responses in PTCL: final results of the phase 2 PRIMO trial of duvelisib in relapsed/refractory peripheral T-cell lymphoma. J Clin Oncol. Published April 22, 2026. doi:10.1200/JCO-25-03120
2. Horwitz SM, Koch R, Porcu P, et al. Activity of the PI3K-δ,γ inhibitor duvelisib in a phase 1 trial and preclinical models of T-cell lymphoma. Blood. 2017;131(8):888-898. doi:10.1182/blood-2017-08-802470