Preliminary Analysis Demonstrates Durvalumab (Anti-PD-L1) & Lenalidomide Is Superior to Single-Agent Durvalumab (anti-PD-L1) in Refractory/Advanced Cutaneous T Cell Lymphoma in a Randomized Phase 2 Trial

Introduction: Mycosis fungoides (MF) and the leukemic variant Sézary syndrome (SS), commonly referred to as cutaneous T cell lymphoma (CTCL), are characterized by clonal expansion of malignant T cells in the skin within a background of chronic inflammation. Advanced stages have an unfavorable prognosis. The skin microenvironment plays an important role in the development and progression of MF and SS. We have shown that the malignant T cells escape immune surveillance via immune checkpoint signaling such as the PD-1/PD-L1 axis. In our phase I dose-escalating study anti-PD-L1 (durvalumab) & lenalidomide demonstrated a tolerable safety profile with significant clinical activity in refractory/advanced CTCL. We initiated the randomized Phase 2 portion to compare single agent durvalumab to durvalumab combined with lenalidomide in relapsed/advanced CTCL (NCT03011814). The primary end point was objective response rate (ORR) using the global composite response (based on skin, blood, nodes, and viscera) according to consensus guidelines. Secondary end points included duration of response, progression-free survival, and toxicity. Relationships between gene expression profile, tumor-microenvironment, and antitumor activity were exploratory end points.

Methods: Adult patients with histologically confirmed MF or SS, who had failed ≥2 systemic therapies were enrolled and randomized 1:1 to single agent durvalumab (1500 mg (day 1 of 28-day cycle) or durvalumab (same dose) & lenalidomide (10 mg for cycle 1, 15 mg for cycle 2, then 20 mg for subsequent cycles daily for 21 days of each 28-day cycle). The study used a “pick a winner” design based on ORR. Serial skin and blood samples were collected to assess the impact on the tumor microenvironment (TME) and anti-tumor activity.

Results: A total of 23 patients were randomized and had the following characteristics (11 durvalumab vs 12 durvalumab/lenalidomide): median age 57 yrs (35-79) vs 63 yrs (32-88); stage IB, 2 (18%) vs 4 (33%); stage IIB, 4 (36%) vs 3 (25%); stage III/IV, 5 (45%) vs 5 (42%); MF, 8 (73%) vs 8 (67%); e-MF/SS, 4 (36%) vs 2 (17%); large cell transformation 3 (27%) vs 5 (42%). The median number of prior systemic treatments for both single and combo arm was 3. Global best ORR was 58% for durvalumab/ lenalidomide vs 36% for durvalumab alone. Median follow up time was 7.9 (range, 0.9-27.6) months. Analysis is ongoing. One patient died one month after discontinuation from study due to PD. No serious AEs were observed. The most common treatment-emergent adverse events (shown below for combo arm) were more frequent in the durvalumab/lenalidomide arm vs durvalumab arm and included fatigue (n=9), diarrhea (n=4), decreased platelets (n=4), leg edema (n=3), constipation (n=3), hyperglycemia (n=3), anemia (n=3), and leukopenia & neutropenia (n=3). The majority of AEs with both treatment arms were mild to moderate in severity (grade I/II, 91%; grade III, 8 %). No grade IV event except 1 neutropenia on combo arm was observed. Median cycles of treatment were 4 (range, 1-12) and 5 (range 1-7) for single and combo arms.) Five pts remain on treatment on single agent durvalumab vs 3 pts on combo arm. Molecular profiling using CIBERSORT and single cell analysis revealed distinct changes of immune cell signatures and cellular signaling interactions within the TME that occurred in each treatment arm when compared to baseline.

Conclusions: This randomized phase 2 trial of durvalumab (anti-PD-L1) +/- lenalidomide evaluating anti-tumor activity demonstrated superior clinical activity of combinatorial durvalumab/lenalidomide vs single-agent durvalumab in refractory/advanced CTCL. Responses were durable and ongoing, and treatment was well tolerated. Our correlative results from sequential skin biopsies demonstrated immune signatures for enhanced anti-tumor responses on skin biopsies at baseline may be predictive of response to checkpoint blockade and yield insights into mechanisms of therapeutic resistance.