Targeting the Inducible T-cell Costimulator (ICOS) in Patients with Relapsed/Refractory T-follicular Helper Phenotype Peripheral T-cell and Angioimmunoblastic T-cell Lymphoma

Peripheral T-cell lymphoma (PTCL) is a rare and heterogeneous group of lymphomas involving T lymphocytes which comprises about 5% to 10% of all lymphomas, with the most common subtypes being PTCL-not otherwise specified (PTCL-NOS), anaplastic large cell lymphoma, and angioimmunoblastic T-cell lymphoma (AITL; ref. 1). The first-line treatment for T-cell lymphomas has not been well defined but patients are frequently treated with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) or CHOP-like regimens with or without autologous stem cell transplantation (2). Despite aggressive treatment, a substantial proportion of patients develop recurrence and 5-year survival rates in relapsed/refractory PTCL range from 10% to 30% (3). Several molecularly targeted and biological based agents have been investigated in the treatment of relapsed or refractory PTCL, including histone deacetylase inhibitors (e.g., romidepsin, belinostat), folate analogue metabolic inhibitors (e.g., pralatrexate), and CD-30 directed antibody–drug conjugates (e.g., brentuximab vedotin; refs. 4–6). Although these agents are approved/available in the United States for the treatment of relapsed or refractory PTCL, the duration of response is often short and the disease remains incurable in most cases. The identification of novel therapeutic interventions targeting relevant molecular and immunological pathways in PTCL remains an unmet need.

T follicular helper (T_FH) cells are a subset of CD4⁺ T cells that reside in the light zone of germinal centers and are considered the cell of origin of some subtypes of PTCL, specifically AITL and T-follicular helper phenotype (TFH) PTCL. Gene expression profiling studies have demonstrated that T_FH cells are characterized by the expression of CD10, B-cell lymphoma 6 (BCL-6), programmed cell death-1 (PD1), CXC chemokine ligand 13 (CXCL13), CXC chemokine receptor 5 (CXCR5), signaling lymphocytic activation molecular-associated protein (SAP), and inducible T-cell costimulator (ICOS; refs. 7, 8). Per the latest WHO 2016 classification (and later confirmed by the latest WHO 2022), two of these antigens are needed to make the diagnosis of PTCL-T_FH (9). ICOS (CD278) is a member of the CD28 (B7) family and through interactions with its ligand (ICOS-L), plays an essential role in the regulation of the T-cell dependent B-cell response in germinal centers (10). These lymphomas are currently under the category of nodal TFH cell lymphomas (11). Marafioti and colleagues evaluated the expression pattern of ICOS using immunohistochemistry in 633 various types of lymphomas, with ICOS positivity confined to AITL (85/86 cases, 99%), TFH-PTCL (18/18 cases, 100%), and a proportion of PTCL-NOS (24/56 cases, 43%) that also expressed other T_FH-associated molecules (8). ICOS expression is also seen in about 70% to 75% of mycosis fungoides and Sezary syndrome cases (12).

MEDI-570 is a human afucosylated, immunoglobulin G1 kappa (IgG1k) monoclonal antibody directed against ICOS. This agent was generated from human anti-ICOS IgG2k monoclonal antibody JTA-009 in a fucosyltransferase-deficient Chinese hamster ovary producer cell line, engineered for enhanced antibody-dependent cellular toxicity effector function (13, 14). The first-in-human study of MEDI-570 was conducted in patients with moderate to severe systemic lupus erythematosus (SLE; NCT01127321) based on the observation that CD4⁺ICOS⁺ T cells are increased in this patient population. A total of 17 subjects were enrolled with MEDI-570 given as a single subcutaneous injection at flat doses ranging from 0.03 mg to 1 mg. This study was terminated early despite not reaching dose-limiting toxicity, due to the occurrence of adverse events deemed to be unacceptable in the context of chronic SLE therapy (15). However, a single dose of MEDI-570 across all treatment groups reduced the mean percentage of memory CD4⁺ICOS⁺ T cells by approximately 70% compared with baseline as early as day 3 (MEDI-570 Investigator's Brochure).

The observation that MEDI-570 depleted ICOS+ T cells prompted a phase I study of MEDI-570 in ICOS+ lymphomas by the Cancer Therapy Evaluation Program of the National Cancer Institute through its NExT program. The current study NCI-9930 was conducted to evaluate the safety and preliminary efficacy of MEDI-570 in T-cell NHL, particularly AITL and TFH-PTCL, as they are known to have high ICOS expression. A starting intravenous dose of 0.01 mg/kg was selected given that this is close to the highest dose (1 mg) administered subcutaneously in the SLE trial without dose-limiting toxicity. Because this agent has not been administered intravenously in repeated cycles, a dose-finding phase I study was performed in patients with pretreated, relapsed, or refractory T-cell NHL.

NCI-9930 was a phase I, dose-finding, nonrandomized, open-label, multicenter clinical study with the primary objective to determine the safety, maximum tolerated dose (MTD) and recommended phase II dose (RP2D) of MEDI-570 in patients with TFH-PTCL, AITL, and other T-cell NHL known to express ICOS (NCT02520791). Secondary objectives were to evaluate the pharmacokinetic (PK) profile of MEDI-570, its efficacy endpoints including objective response rate (ORR) and progression-free survival (PFS), and its long-term effects on the immune system and T-cell lymphocyte subsets. Exploratory objectives were to investigate biomarkers of response and resistance to MEDI-570 in the study population. This study was conducted under the auspices of the US National Cancer Institute (NCI) Experimental Therapeutic Clinical Trials Network (ETCTN) and Early Drug Development Opportunity Program (EDDOP), and recruited patients from 10 sites between September 13, 2016, and March 4, 2021. The protocol was approved by the independent review boards of all participating ETCTN sites, patients provided informed written consent and the study was conducted in compliance with the Declaration of Helsinki and Good Clinical Practice Guidelines of the International Conference on Harmonization.

In the dose escalation part of NCI-9930, patients must have a confirmed diagnosis of PTCL or AITL that was refractory or relapsed to at least one line of therapy. Advanced cutaneous T-cell lymphoma; small/medium T-cell lymphoma; mycosis fungoides stage IB, IIA, IIB, III, and IV that have relapsed after at least one specific prior therapy; and follicular lymphoma (grade 1, 2, or 3A) that has relapsed or refractory to at least two lines of therapy and after autologous stem cell transplantation, were also eligible. In the dose expansion part of the study, two separate cohorts were established: (i) TFH-PTCL or AITL that has relapsed or refractory to at least one line of therapy; (ii) AITL that has not been treated with any therapy or deemed by the investigator as being not suitable for chemotherapy. All patients must be greater than or equal to 18 years of age and have adequate bone marrow, renal, and hepatic functions. Exclusion criteria included any history or evidence of opportunistic infection within 6 months of screening, prior allogeneic stem cell transplantation, and autologous stem transplantation within 3 months of study entry and a CD4⁺ less than 100. Full eligibility criteria are presented in the Supplementary Methods.

The dose escalation part of this study utilized the accelerated titration dose escalation design with single patient cohorts until a grade 2 or higher toxicity that was at least possibly related to MEDI-570 was encountered, at which point dose escalation would revert to a 3 + 3 design (16). During dose escalation, up to 6 additional patients with AITL may be enrolled as backfill into the highest dose level that has previously cleared the dose-limiting toxicity (DLT) assessment period in at least three evaluable patients. The dose expansion cohort has a target sample size of 10 TFH-PTCL or AITL patients with prior therapy. This sample size was chosen/selected based on the assumption of an objective response rate at P0 of 15% and P1 of 40%. Patients enrolled in the dose expansion cohort received MEDI-570 at the RP2D. No intrapatient dose escalation was allowed in any part of the study.

The following five dose levels of MEDI-570 were evaluated: 0.01, 0.1, 0.3, 1, and 3 mg/kg. MEDI-570 is administered intravenously in the outpatient setting over at least 1 hour, in 21-day cycles for up to 12 cycles. MEDI-570 was stable for up to 4 hours and a new preparation may have been needed for completion of any skipped dose, or if extended infusion was needed (e.g., due to infusion reactions). Initially, no premedications were required but as a result of a grade 3 infusion-related reaction encountered at dose level 2 (0.1 mg/kg), the protocol was amended to mandate pre-medications consisting of diphenhydramine 50 mg orally or IV (or equivalent dose of antihistamine drug) and acetaminophen 500 to 1,000 mg orally (or equivalent dose of antipyretic) to be given 1.5 hours prior to infusion of MEDI-570. Because of potential T-cell depletion from MEDI-570, patients were recommended antiviral prophylaxis with acyclovir and anti-Pneumocystis jirovecii pneumonia (PJP) prophylaxis with trimethoprim-sulfamethoxazole or equivalent (if history of sulfa allergy) with dosing according to institutional practices. If CD4⁺ T cells were less than 50 cells/mm³, prophylaxis for mycobacterium avium complex (MAC) was recommended.

Baseline evaluations including medical history; physical examination; assessment of B symptoms; bloodwork for hematology, absolute CD4 count, serum chemistry and pharmacodynamics biomarkers; urinalysis and tuberculosis skin test were performed within 7 days prior to start of protocol therapy. Bloodwork was repeated weekly during cycle 1 and then at the start of every subsequent cycle. Patients were monitored for cytomegalovirus (CMV) reactivation by polymerase chain reaction (PCR) in peripheral blood. Computerized tomography (CT) scans and 18F-fluorodeoxyglucose-positron emission tomography (FDG-PET) scans (only for PTCL, AITL, and follicular lymphoma patients) were done within 4 weeks prior to start of study therapy. CT scans were repeated every 2 cycles, whereas follow-up FDG-PET scans were only needed at end of treatment or at disease progression. Collection of archival tumor tissue was mandatory for all patients, whereas fresh tumor biopsy at baseline and at cycle 1 week 3 were considered optional.

Toxicities were assessed using the NCI Common Terminology Criteria for Adverse Events version 5.0. The following adverse events at least possibly related to MEDI-570 were considered as DLT: (i) Any grade 3 or higher non-hematologic toxicity, except for: alopecia; tumor lysis syndrome not requiring dialysis, hospital admission for less than 72 hours, or admission to an intensive care unit; grade 3 infusion-related reactions that responded to medical treatment; grade 3 nausea, vomiting or diarrhea that resolved to less than or equal to grade 1 by appropriate medical intervention; electrolyte toxicities that can be corrected to less than grade 1 or baseline within 48 hours; grade 3 rash that improves to a lower grade within 72 hours; (ii) Severe opportunistic infection or any grade 3 infection; (iii) Delay in starting cycle 2 by 14 days or longer due to toxicity; (iv) Other toxicity of concern to investigators that are considered to be related to study drug; (v) Hematologic toxicity including grade 4 neutropenia lasting more than 7 days; grade 4 thrombocytopenia; grade 3 thrombocytopenia with bleeding; grade 3 or 4 febrile neutropenia; and other hematologic toxicities of concern (see Supplementary Methods for full DLT criteria).

Antitumor activity was evaluated every 2 cycles, and response assessment was made using the new Revised Criteria based on the Lugano Classification for patients with PTCL, AITL, and follicular lymphoma (17). Patients with mycosis fungoides were evaluated as per the Consensus Statement of the International Society for Cutaneous Lymphomas and the European Organisation for Research and Treatment of Cancer (18). Patients were evaluable for toxicity if they at least received 1 dose of study drug. Patients were evaluable for efficacy if at least one cycle of study drug was completed.

Samples for pharmacokinetics analysis of MEDI-570 were collected in peripheral blood prior to dose on day 1, immediately after end of infusion, and then at 6, 24, 72, and 168 hours after dose of cycle 1 and cycle 2, on day 1 pre-dose of every subsequent cycle starting at cycle 3 and onwards, and at the end of study. Serum concentrations of MEDI-570 were measured by a quantitative Electrochemiluminescent Assay (ECLA, Meso Scale Discovery), using biotinylated mouse anti-MEDI-570 antibody for capture and a sulfo-TAG labeled mouse anti–MEDI-570 monoclonal antibody for capture. The assay was bioanalytically validated with a sensitivity of 5 ng/mL.

Blood samples for anti–MEDI-570 antibodies were collected prior to dose on cycle 1 day 1, on day 1 of subsequent cycles starting at cycle 2, and at the end of the study. An ECLA was used to detect the presence of anti-MEDI-570 antibodies in human serum. The method included procedures to detect (screen), confirm, and titer anti-drug antibodies against MEDI-570 in the serum of human subjects.

Formalin-fixed, paraffin-embedded slides from representative sections of archived tumor specimens were stained using rabbit anti-ICOS monoclonal antibody against ICOS clone SP98 (Spring Bioscience). Staining was conducted on the Discovery Ultra platform (Ventana) and expression of ICOS was scored semiquantitatively by a board-certified pathologist (L. Cheng).

Absolute quantities of circulating T, B, or natural killer (NK)-cell populations and ICOS-expressing T cells were monitored in fresh whole blood specimens using bioanalytically validated immunophenotyping assays. In brief, whole blood specimens were collected in optimum collection tubes (CytoChex-BCT or ACD-B for T, B, and NK or ICOS-expressing T-cell assays, respectively) and incubated with optimized quantities of fluorochrome-labelled antibodies as reported previously. Data were plotted using Tibco Spotfire version 10.3 (19).

Peripheral blood mononuclear cells (PBMC) were isolated and viably frozen at multiple time points for each patient: at screening, prior to dose on cycle 1 day 1, weekly during cycle 1, on day 1 of subsequent cycles starting at cycle 2, and at the end of the study. Multi-parameter flow cytometry analyses were performed on viably frozen PBMCs in batches (see Supplementary Table S1 for panels and antibodies, and Supplementary Table S2 for timepoints). Flow cytometry data were acquired using a 5-laser BD LSRFortessa X-20 (BD). Data were analyzed using FlowJo Software (BD), and computational cell clustering was performed on normalized-arcsin transformed data using Rphenograph R package (V0.99.1; ref. 20). Uniform manifold approximation and projection (UMAP), using the umap R package (V0.2.4.1) was performed to visualize the cell clusters (21).

The Kaplan–Meier method was carried out for progression-free survival (PFS) and overall survival (OS) analyses. Descriptive statistics were used to report patients’ clinical characteristics, response rate, adverse events, and all other outcome measures.

Detailed data cannot be shared publicly to protect the privacy of individual clinical trial participants. Information to support the findings of these analyses are available by contacting the corresponding author, who upon reasonable request and understanding of the intended use of the data will work with the Clinical Trials Evaluation Program to provide the requested information in the manner which continues to protect individual patient information. Data generated by these analyses are provided in the article.

Between September 13, 2016, and March 4, 2021, this study enrolled 23 patients including one screen failure. Median follow-up was 9.0 (range, 1.8–44.2) months. Of these 23 patients, 18 were enrolled in the dose escalation part (including the screen failure case), 5 in dose expansion cohort (a) for relapsed/refractory TFH-PTCL or AITL, and no patients were enrolled in dose expansion cohort (b) for untreated AITL. Patient and disease characteristics are summarized in Table 1. The median age was 67 years (range, 29–86); patients were heavily pretreated with a median of 3 prior lines of therapies (range, 1–16). For PTCL/AITL patients (n = 21), the median number of prior lines were 2 (range, 1–8).

During dose escalation, MEDI-570 was administered to 17 patients without any DLT. The first patient enrolled at dose level 1 with MEDI-570 given at 0.01 mg/kg dose experienced grade 2 headache (baseline grade 1) and grade 2 vomiting that were deemed by the investigator to be possibly related to study drug. Hence, dose escalation reverted from accelerated titration design to 3 + 3 design, with three patients enrolled in each of dose levels 1 to 4, without DLT being observed in any of these patients. In dose level 5 (3 mg/kg), which was the maximum planned dose, five patients were enrolled without DLT, and thus, a sixth patient was not enrolled because it would not alter the decision for RP2D determination. As no DLT was encountered in all five dose levels during dose escalation, the highest dose of MEDI-570 administered, 3 mg/kg intravenously every 3 weeks, was declared as the RP2D. Five patients were enrolled into the expansion cohort; none of the patients treated at the RP2D experienced any DLT-qualifying adverse events.

Table 2 depicts treatment emergent adverse events (TEAE) with frequencies of at least 15% of all patients. The most common grade 3–4 TEAEs were decreased CD4⁺ count (less than 200 cells/mm³) seen in 13 (57%) patients, lymphopenia (less than 500 cells/mm³) in 6 (22%) patients, anemia in 3 (13%) patients, hypertension in 3 (13%) patients, infusion related reaction in 2 (9%) patients and hypophosphatemia in 2 (9%) patients. The most common grade 1–2 TEAEs were infusion related reactions (IRR) in 11 (48%) patients, fatigue in 8 (35%) patients, nausea in 7 (30%) patients, and constipation in 6 (26%) patients. IRRs were limited to C1 and, after introduction of premedications, no grade >2 IRR were reported. CD4 counts were collected from the end of study treatment in 13 patients and among those with serial collections, recovery was observed over time (Supplementary Fig. S1).

There were five serious adverse events (SAE) which occurred in four patients, deemed as at least possibly related to MEDI-570. One patient treated at dose level 2 with MEDI-570 at 0.1 mg/kg experienced a grade 3 IRR related to study drug. This SAE triggered a protocol amendment for all subsequent patients to be premedicated with diphenhydramine and acetaminophen. A second patient treated at dose level 2 with MEDI-570 at 0.1 mg/kg after developed 2 SAEs after receiving an allogeneic stem cell transplant (SCT) using a matched unrelated donor with an investigational conditioning regimen (romidepsin in combination with myeloablative doses of fludarabine and busulfan). The allogeneic SCT occurred after achieving a partial response post cycle 4 of MEDI-570. The patient had grade 3 BK virus cystitis/nephritis, grade 3 adenovirus infection and grade 2 CMV reactivation, which were all deemed possibly related to MEDI-570, possibly related to underlying PTCL, and probably related to the allogeneic SCT and its conditioning regimen in the context of a clinical trial. This patient ultimately had a grade 5 upper gastrointestinal hemorrhage event deemed possibly related to MEDI-570, in the setting of adenovirus infection, and graft versus host disease. A third patient treated at dose level 3 with MEDI-570 at 0.3 mg/kg developed grade 3 pericarditis deemed possibly related to study drug, possibly related to disease and probably related to viral etiology. A fourth patient with AITL treated at dose level 4 with MEDI-570 at 1 mg/kg developed a diagnosis of diffuse large B cell lymphoma in the setting of EBV reactivation.

Patients discontinued MEDI-570 mainly due to progressive disease (12/23 patients = 52%), completion of therapy or to undergo consolidation with allogeneic stem cell transplantation. No patients discontinued MEDI-570 due to AEs or SAEs.

The ORR for all (evaluable) patients was 7 of 23 patients or 30%. There were 2 patients who attained CR and 5 patients who attained PR as their best objective response on MEDI-570, all with AITL (Table 3), yielding an overall response rate (ORR) of 44% (7/16 patients, 95% CI: 20%–70%) in this subtype. These patients were treated at dose levels 1, 2, 5, and dose expansion. Median duration of response was 6.1 (range 0.7–36.8) months (including a patient that underwent allogeneic SCT as consolidation). For AITL patients, the median PFS was 2.9 (95% CI, 1.9–6.7) months, and median OS was 17.1 (95% CI, 6.7–not reached) months. No responses were seen in PTCL and CTCL patients. The waterfall plots and spider plots of all patients by dose level and by histology are shown in Fig. 1.

MEDI-570 exposure increased in a dose dependent manner at doses greater than or equal to 0.1 mg/kg. Non-linearity in PK was noted at the lower dose level of 0.01 mg/kg, as indicated by the greater than proportional increase in exposures at higher doses (Fig. 2). These results are consistent with the longer half-life (t_1/2) and slower clearance observed at the higher doses (≥0.1 mg/kg), relative to that at 0.01 mg/kg (Supplementary Table S3). Generally, moderate PK variability (<∼40% CV) was observed across dose levels.

All samples, evaluated across all subjects, were observed to be negative for the presence of anti–MEDI-570 antibodies.

Immunophenotypic analysis of fresh whole blood revealed that MEDI-570 resulted in reductions in circulating quantities of CD4⁺ ICOS⁺ T cells in a dose-dependent manner with a maximum effect observed at doses of 0.3 mg/kg and higher (Fig. 3A). Similar results were observed with total CD4⁺ T cells demonstrating that these reductions were not merely the result of MEDI-570 interference with detection of ICOS expression but represent a bona-fide loss of target cells (Fig. 3B). Assessment of T-cell subsets revealed that MEDI-570 induced preferential reductions in circulating quantities of CD4⁺ CD45RA⁻ memory T cells and follicular helper T cells, both of which comprise the majority of ICOS-expressing T cells (Fig. 3C and D). Analysis of additional circulating lymphocyte and leukocyte populations revealed MEDI-570–induced reductions in CD8⁺ T cells, CD14⁺ monocytes, and CD16⁺/56⁺ natural killer cells (Supplementary Fig. S2A–S2C). The effects observed on CD19⁺ B cells were inconsistent following MEDI-570 administration with some patients exhibiting reductions and others demonstrating expansions (Supplementary Fig. S1A–S1D).

Immune analysis of viably frozen PBMCs showed that independent of dose, MEDI-570 treatment significantly reduced CD3⁺ T cells and CD4⁺ T cells between baseline/cycle 1 week 1, and cycle 1 week 3/cycle 2 week 1 (Fig. 4A). The number of CD8⁺ T cells was not significantly affected when grouped in the same way. High-dimensional clustering using phenograph was performed as an additional exploratory analysis of the circulating immune profile in patients (Fig. 4B and C). For each flow cytometry panel, unsupervised phenograph clustering was performed on all live cells from each patient and time point, resulting in 23 unique immune cell clusters for panel 1 and 27 clusters for panel 2 (Supplementary Table S1), including in both panels, cell clusters representing lymphoma cells (CD3^neg/low, CD4⁺ cells). When comparing the baseline peripheral immune profiles of patients grouped by diagnosis (Fig. 4D and E) and best response (Fig. 4F and G; AITL only), we observed differences (not statistically significant) in several immune cell clusters. The abundance of γδ T cells (Fig. 4D, cluster 6), CD4⁺ T cells (Fig. 4D, cluster 8; Fig. 4E, clusters 1, 5, 16, 20, and 22) including regulatory T cells (Fig. 4E, clusters 4 and 25), CD8⁺ T cells (Fig. 4E, clusters 2, 11, 19, and 21) and potential lymphoma clusters (Fig. 4D, clusters 15, 18, and 20; Fig. 4E, clusters 8, 9, 14, 18, and 27) differed by diagnosis. We observed fewer CD28⁺TIGIT⁺PD-1⁺CD4⁺ T cells (Fig 4F, cluster 8) and more CD127⁺FOXP3⁺PD-1⁺ICOS^lowCD4⁺ T cells (Fig. 4G, clusters 5 and 20) in AITL patients who responded (PR) to MEDI-570 in comparison to progressors (PD).

Immunohistochemical staining for ICOS was conducted on 14 formalin-fixed, paraffin-embedded tumor specimens from 9 patients including primarily archival (5/14 confirmed) and/or baseline and cycle 1 week 3 (6/14 confirmed), if provided. All evaluated biopsy samples for AITL and PTCL-NOS or TFH-PTCL demonstrated positive, heterogenous ICOS staining ranging from scattered weak staining to strong membranous expression (Supplementary Fig. S3). The low numbers of paired time point samples and the quality of the biopsy sampling precluded any further correlative assessments based on IHC expression.

This study closed to accrual and treatment on March 4, 2021. At the time of data cut-off on January 16, 2022, all patients have come off study treatment, and 14 are alive.

In this phase I, first in human, open-label and dose-finding study NCI-9930, the safety, PK, PD, MTD, and preliminary efficacy of MEDI-570 was evaluated in patients with previously and heavily pre-treated various T-cell malignancies. No DLTs were reported in the dose escalation phase; therefore, the highest dose of MEDI-570 at 3 mg/kg was declared as the RP2D. The study was terminated early due to drug product supply limitations.

In the current study, MEDI-570 was successfully administered through the intravenous route. Its serum concentrations increased in a dose-dependent manner. A longer MEDI-570 half-life was observed at higher doses. Clinical responses were seen across the range of doses tested. Because of the occurrence of IRR at dose level 2, a study amendment was proposed with the use of diphenhydramine and acetaminophen as premedications prior to MEDI-570 infusion. While grade 3/4 decreased CD4⁺ counts were seen, these did not translate to increased risk of opportunistic infections and recovery was seen after end of treatment. There were no discontinuations due to AEs or SAEs. There was one grade 5 SAE possibly related to MEDI-570 in a patient who was post allogeneic SCT several months after MEDI-570 was completed. Despite the effect of MEDI-570 in suppressing CD4⁺ T cells, there was no evidence of recurrent opportunistic infections in this study.

MEDI-570 showed encouraging single agent activity in relapsed or refractory AITL with an objective response rate of 44% and a median duration of response of 6.1 months in 16 patients with such histology treated on this study. This is particularly important when considering the median number of prior therapies was 2 (range 1–8) for AITL/PTCL patients, suggesting this is a highly active agent in these lymphomas. There are other available agents for R/R PTCL and AITL that elicit objective responses that include folate analogue metabolic inhibitor pralatrexate (29%, AITL 8%), histone deacetylase inhibitors romidepsin (25%, AITL 30%), and belinostat (23%, AITL 46%), Aurora A kinase inhibitor alisertib (24%, AITL 33%), CD30-directed antibody–drug conjugate brentuximab vedotin (42%, AITL 54%), and PI3 kinase inhibitor duvelisib (40%, AITL 50%; refs. 5, 22–26). In context of this therapeutic landscape, MEDI-570 demonstrates a promising activity and suggests that ICOS inhibition is a relevant strategy for the treatment of AITL with tolerable toxicity profile in population of heavily pretreated patients. Given its toxicity profile and fixed duration option, targeting ICOS could be further explored in combination with frontline regimens for T-cell lymphomas.

While there are other molecules in clinical development targeting the costimulatory molecule ICOS in solid tumors with variable efficacy (27, 28), these agents are mechanistically different from MEDI-570. MEDI-570 binds to ICOS and mediates enhanced antibody-dependent cellular cytotoxicity based on its increased binding to Fc receptors, in contrast to costimulatory molecules that propose to enhance adaptive immune response. Results of T-cell subset analyses of whole blood and PMBCs both demonstrated that MEDI-570 caused preferential depletion of CD4⁺ T cells, particularly CD4⁺ ICOS⁺ T cells, indicating its potential for the targeting of T_FH cells, which are essential as cell of origin for the development of T_FH-subtype PTCL, especially AITL. The sample size of the current study was too limited to demonstrate an association between clinical response to MEDI-570 and decreased CD4⁺ ICOS⁺ T cells, but a trend was observed.

It is noteworthy to emphasize the preferential activity of MEDI-570 in AITL underscores the role of ICOS as a potential and novel biomarker-driven approach for T-cell NHL. The safety profile of MEDI-570 renders this agent a good candidate for combinatorial approaches to improve clinical responses. Romidepsin and 5-azacitidine were safely combined and elicited high response rates with an objective response rate and CR rate of 80% and 60%, respectively (29). Similarly, the combination of romidepsin and duvelisib yielded higher responses (CR rate of 34%) but with higher grade toxicities such as grade 3 neutropenia (36%), diarrhea (15%), and transaminitis (14%; ref. 30).

The main limitation of this study was its small sample size (as expected for a rare disease). Another limitation was that the current study could not be completed due to drug expiration and subsequent lack of drug supply to enroll more patients in the expansion and backfill cohorts. Another challenge was the slow pace of accrual (duration of 5 years approximately) owing to the phase I nature of the study in a rare lymphoma subtype.

In summary, the targeting of ICOS by MEDI-570 represents a novel approach for T-cell NHL, specifically for T_FH-subtypes such as AITL. MEDI-570 showed encouraging activity and safety profile in heavily pretreated AITL patients. Further studies should focus on confirmatory studies especially in AITL, with larger sample size and combinatorial approaches. There is a strong scientific rationale to continue the investigation of the role of ICOS in AITL as it may represent a novel precision-based immunotherapeutic approach in this T_FH-cell driven malignancy.

J. Chavez reports personal fees from AstraZeneca during the conduct of the study; personal fees from GenMab, ADC Therapeutics, BMS, AdiCet, Novartis, and Kite/Gilead, and grants from Merck, Adaptive, and BeiGene outside the submitted work. B. William reports other support from Guidepoint Global, Morphosys, and ADC Therapeutics outside the submitted work. S. Smith reports other support from Ono Pharmaceuticals, Gilead, BMS, Morphosys, Janssen, and Karyopharm outside the submitted work. A. Prica reports personal fees from AstraZeneca and Kite-Gilead outside the submitted work. J. Zain reports grants from Seattle genetics, Myeloid, and CRSPR, personal fees from Kyowa Kirin, grants and personal fees from Secura Bio, and grants from AstraZeneca outside the submitted work. H. Shah reports nonfinancial support from AstraZeneca, Seattle Genetics, Beigne, ADCT, and LOXO Oncology outside the submitted work. N. Mehta-Shah reports personal fees from Secura Bio, Kyowa Hakka Kirin, AstraZeneca, Genentech, Daiichi Sankyo, Ono, C4 Therapeutics, and Karyopharma outside the submitted work. P. Ramakrishnan Geethakumari reports other support from BMS, Kite pharma, Ono pharma, Cellectar, ADCT, Rafael Pharma, and Pharmacyclics LLC outside the submitted work. B.X. Wang reports personal fees from AstraZeneca, Tessa Therapeutics, and Providence Therapeutics outside the submitted work. S. Zantinge reports grants from NIH during the conduct of the study. N. Standifer reports other support from AstraZeneca outside the submitted work. S. Sharma reports personal fees and other support from AstraZeneca and personal fees and other support from Gilead Sciences outside the submitted work. G. Carlesso reports personal fees from AstraZeneca outside the submitted work as well as a pending patent (PCT/US2021/058110, which is directed to anti-ICOS antibodies for the treatment of lymphomas and is related to the study). L.L. Siu reports personal fees from AstraZeneca and grants from Astra Zeneca during the conduct of the study, personal fees from Merck, Pfizer, Roche/Genentech, GlaxoSmithKline, Voronoi, Arvinas, Tessa, Navire, Relay, Coherus, Amgen, Marengo, Daiichi Sankyo, InteRNA, Medicenna, Tubulis, and LTZ Therapeutics, grants from Novartis, Bristol-Myers Squibb, Pfizer, Boehringer-Ingelheim, GlaxoSmithKline, Roche/Genentech, Merck, Bayer, Abbvie, Amgen, Symphogen, Intensity Therapeutics, Mirati Therapeutics, Shattucks Laboratory, BioNTech, 23Me, and EMD Serono, and personal fees from Agios and Treadwell Therapeutics outside the submitted work. No disclosures were reported by the other authors.

J.C. Chavez: Conceptualization, resources, data curation, supervision, validation, investigation, methodology, writing–original draft. F.M. Foss: Data curation, investigation, writing–review and editing. B.M. William: Data curation, investigation, writing–review and editing. J.E. Brammer: Data curation, investigation, writing–review and editing. S.M. Smith: Data curation, investigation, writing–review and editing. A. Prica: Data curation, investigation, writing–review and editing. J.M. Zain: Data curation, investigation, writing–review and editing. J.M. Tuscano: Data curation, investigation, writing–review and editing. H. Shah: Data curation, investigation, writing–review and editing. N. Mehta-Shah: Data curation, investigation, writing–review and editing. P.R. Ramakrishnan Geethakumari: Data curation, validation, writing–review and editing. B.X. Wang: Data curation, formal analysis, visualization, writing–review and editing. S. Zantinge: Resources, data curation, supervision, investigation, writing–review and editing. L. Wang: Resources, data curation, validation, writing–review and editing. L. Zhang: Resources, validation, writing–review and editing. A. Boutrin: Resources, methodology, writing–review and editing. W. Zhao: Resources, methodology, writing–review and editing. L. Cheng: Resources, methodology, writing–review and editing. N. Standifer: Resources, data curation, supervision, methodology, writing–review and editing. L. Hewitt: Resources, methodology, writing–review and editing. E. Enowtambong: Resources, methodology, writing–review and editing. W. Shao: Resources, methodology, writing–review and editing. S. Sharma: Resources, writing–review and editing. G. Carlesso: Resources, supervision, methodology, writing–review and editing. J.A. Moscow: Conceptualization, writing–original draft, project administration, writing–review and editing. L.L. Siu: Conceptualization, resources, data curation, methodology, writing–original draft.

This study was supported by NCI UM1 Grant # CA186644, held by the North American Star Consortium