Safety and Clinical Activity of MEDI0562, a Humanized OX40 Agonist Monoclonal Antibody, in Adult Patients with Advanced Solid Tumors

Immune checkpoint blockade using antagonistic monoclonal antibodies (mAbs) against cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4), programmed cell death (PD)-1 or its ligand (PD-L1) have demonstrated impressive long-term, protective antitumor immune responses, providing clinical benefits across multiple solid tumor types (1, 2), many of which traditionally had a poor prognosis (3–5). Unfortunately, responses appear to be limited to a small percentage of treated patients (6), and many patients are resistant to, or relapse, with these drugs in monotherapy. On the basis of the nonoverlapping mechanisms of action of these agents, it was hoped that combination regimens targeting both PD-1 and CTLA-4 would further improve outcomes (7); however, while clinical response rates can be increased, the benefits are offset by an increased frequency of adverse events (8). Thus, the focus of drug development has shifted to examine new immunomodulatory molecules which might be targeted to bypass the mechanisms responsible for primary or acquired resistance to first-generation immunotherapies (2, 7).

The costimulatory molecule OX40 is a member of the tumor necrosis factor (TNF) receptor family that is highly expressed in activated immune cells (2). Upon binding to its ligand, OX40L, it initiates an intracellular signaling cascade both in effector T cells (Teffs) to enhance immune responses and in regulatory T cells (Tregs) to overcome their suppressive activity (9–11). Engagement of OX40 by OX40L enhances activation, potentiation, proliferation, and survival of T cells, and when tumor antigens are recognized by tumor-infiltrating lymphocytes (TILs), OX40 expression increases (12). OX40 has a broad expression profile and is a target for multiple tumor settings, such as breast cancer, melanoma, B-cell lymphoma, head and neck cancer, and colon cancer (2).

MEDI0562 is an agonistic humanized IgG, subclass 1, κ light chain mAb that specifically binds to OX40 and triggers a signaling cascade, which acts to enhance T-cell survival, proliferation, and cytokine production (10). MEDI0562 was developed by humanizing a murine OX40 mAb (9B12) that was previously reported to have an acceptable toxicity profile, evidence of pharmacodynamic activity, and regression of ≥1 metastatic lesion in 12 of 30 treated patients in a phase I clinical study (10). The objectives of this first-in-human study were to evaluate the safety, tolerability, pharmacokinetics, pharmacodynamics, and antitumor activity of single doses of MEDI0562 in patients with advanced solid tumors. The study is registered with ClinicalTrials.gov as NCT02318394.

This was a first-in-human phase I study of MEDI0562, designed to assess dose escalation (3 + 3 design) and dose expansion of MEDI0562 in adult patients with advanced solid tumors. The study was conducted as a multicenter, open-label, single-arm study at sites in North America and Asia. In total, 11 sites enrolled and treated patients.

Patients were treated with one of six escalating doses of MEDI0562 every 2 weeks until confirmed disease progression or unacceptable toxicity. In the dose-escalation phase, patients received MEDI0562 through intravenous infusion (Fig. 1). Tumor assessments were performed every 8 weeks using immune-related Response Evaluation Criteria in Solid Tumors (irRECIST) (13). Enrollment was completed for the dose-escalation cohorts (0.03–10.0 mg/kg MEDI0562).

Key inclusion critieria included patients aged ≥18 years; confirmed advanced solid tumor with ≤3 prior lines of therapy for recurrent or metastatic disease; ≥1 measurable lesion and adequate organ function. Prior treatment with agonistic antibodies targeted to OX40, CD27, CD137, and CD357 was not allowed. Prior treatment with anti–PD-1/L1 and/or CTLA-4 antibodies was allowed if treatment interval was at least 60 days and prior therapy was not discontinued because of an AE. Patients with active or prior documented autoimmune disease within the previous 2 years or untreated metastatic central nervous system disease were excluded.

The study was performed in accordance with the Declaration of Helsinki and is consistent with International Conference on Harmonisation, Good Clinical Practice, and any applicable laws and requirements, and any conditions required by a regulatory authority and/or Institutional Review Board/Independent Ethics Committee that approved this study to be conducted in its territory. All patients provided written informed consent.

In the dose-escalation phase of the study, sequential cohorts of three to six patients each received one of six escalating dose levels of MEDI0562 (0.03, 0.1, 0.3, 1.0, 3.0, or 10 mg/kg) through intravenous infusion every 2 weeks, unless the maximum tolerated dose (MTD) was reached before all dose-escalation cohorts were completed. Patients remained on treatment until unacceptable toxicity, documentation of confirmed disease progression, or another reason for treatment discontinuation developed. In the event of confirmed immune-related progressive disease (irPD), a patient could continue to receive MEDI0562. Tumor assessments were performed every 8 weeks with irRECIST.

A planned dose-expansion phase did not begin due to the limited efficacy observed during the dose-escalation phase, and a strategic decision by the sponsor to pursue combination regimens in place of the monotherapy.

The primary objective of this study was to assess safety and tolerability, describe any dose-limiting toxicities (DLTs) and determine the MTD or maximum administered dose of MEDI0562 when administered as a single agent in patients with selected advanced solid tumors.

Secondary objectives included describing the antitumor activity of MEDI0562 as per irRECIST (13), as measured by objective response, disease control, duration of response, progression-free survival (PFS), and overall survival (OS). Immunogenicity was quantified by antidrug antibodies (ADAs). Pharmacokinetics were described by maximum observed concentration (C_max), area under the curve (AUC), clearance, and terminal half-life. Candidate pharmacodynamic biomarkers were also evaluated including induction of proliferation markers in various lymphocyte subsets and assessment of TILs.

The DLT-evaluation period was defined as the time period from the first dose of MEDI0562 until 28 days post dose 1. Tumor samples were evaluated pretreatment and at day 29 by quantitative digital analysis of immunohistochemistry (IHC) images and gene expression. Peripheral blood from all patients with evaluable samples was monitored using gene expression and flow cytometry, with mean percentages of T cells plotted over time by dose cohort. All flow cytometry testing was analytically validated (10) and performed by Covance Central Laboratory Services/LabCorp on BD FACSCanto instruments.

Serum samples were collected and assessed for ADAs every 2 weeks starting from day 1 (predose). Blood samples were collected for pharmacodynamic biomarkers at day 1 (predose), day 2, day 3, day 8, day 15 (predose) and every 2 weeks thereafter. Disease assessments were performed every 8 weeks (± 3 days) during the treatment period (e.g., day 57, day 113, etc) or when signs/symptoms of disease progression were reported. If irPD was observed in the absence of clinical deterioration, a confirmatory scan was performed within 4 weeks.

Categorical data were summarized by frequency distribution (number and percentage of patients falling within each category). Continuous variables were summarized by descriptive statistics including number of patients (N), mean, standard deviation (SD), median, and range (minimum and maximum). The best overall response (BOR) was summarized as the number and percentage of patients for the following categories: immune-related complete response (irCR), partial response (irPR), stable disease (irSD), disease progression (irPD), and not evaluable. The objective response rate (ORR) was defined as the proportion of patients with a BOR of confirmed irCR or irPR, and estimated with 80% and 95% confidence intervals (CIs) using the exact probability method. Efficacy analyses were based on application of irRECIST to investigator-assessed tumor measurements. Time-to-event data (including OS and PFS) were summarized using Kaplan–Meier estimates (median time and two-sided 80% and 95% CIs for median time). Unless stated otherwise, dose 1/day 1 (predose) was the baseline assessment unless data were missing, in which case, baseline was defined as the last value prior to dosing. PFS was measured from the start of treatment with investigational product until the first documentation of PD or death due to any cause, whichever occurred first. Data analyses were conducted using the SAS System (SAS Institute Inc.) Version 9.3.

Data underlying the findings described in this manuscript may be obtained in accordance with AstraZeneca's data sharing policy described at https://astrazenecagrouptrials.pharmacm.com/ST/Submission/Disclosure.

In total, 56 patients were enrolled into the study from 11 sites between March 24, 2015 and March 1, 2017, of whom 55 received ≥1 dose of MEDI0562 and were included in the as-treated population (Fig. 1). Of these patients, 49 were included in the DLT-evaluable population, and 50 patients were included in the response-evaluable population. Overall, 55 patients discontinued study treatment, with PD (n = 34, 62%) being the primary reason. Reasons for study discontinuation were death (n = 25 [45%] patients), withdrawal of consent (n = 14 [25%] patients [two patients gave AE as the reason for withdrawal of consent and 12 stated “other”]), completed protocol-defined end of study (i.e., were in study follow-up at the time of study termination; n = 13 [24%] patients), and lost to follow-up (n = 3 [5%] patients).

Baseline demographics and clinical characteristics are summarized in Table 1. The most common tumor types evaluated were squamous cell carcinoma of the head and neck (SCCHN, n = 26, 47%) and cervical cancer (n = 10, 18%).

For those patients with SCCHN, 10 were human papillomavirus (HPV) positive (38%), 12 were HPV negative (46%), and four had unknown HPV status (15%). For patients with cervical cancer, six were HPV positive (60%) and four had unknown HPV status (40%). Of the 55 patients in the as-treated population, 54 (98%) received prior anticancer treatment for recurrent or metastastic disease that included chemotherapy and mAbs. The most commonly used mAbs were cetuximab (11/55, 20%), bevacizumab (9/55, 16%), pembrolizumab (4/55, 7%), and nivolumab (3/55, 5%); the most commonly used chemotherapy agents were cisplatin (29/55, 53%), carboplatin (23/55, 42%), paclitaxel (15/55, 27%), and docetaxel (14/55, 25%). The median number of prior regimens was 2.0 (range 0–5). One patient (adenoid cystic salivary gland tumor) received radiotherapy and surgery prior to MEDI0562 treatment, but no prior systemic therapy, and all other patients received prior systemic therapy.

The median duration of patient exposure to treatment was 10 weeks (range, 2–48 weeks). AEs and treatment-related AEs (TRAEs) were reported in 96% and 67% of study patients, respectively (Table 2). The most common TRAEs were fatigue (31%) and infusion-related reaction (14%). Grade 3 TRAEs occurred in 14% of patients, and 53% of patients had TRAEs of grade 1 or 2, with no apparent dose relationship (Table 2).

No DLTs were observed in the dose-escalation portion of the study. During the expansion of dose escalation, one patient (1.0 mg/kg) had a DLT of diarrhea which was categorized as grade 3 according to symptoms (≥7 stools per day or incontinence); however, because the patient was not hospitalized, the investigator determined that the event did not meet the standard criteria of a serious AE (SAE). In this patient, treatment was discontinued and antidiarrheal agents were administered; because the event had not resolved after 8 days (possibly due to the lack of aggressive steroid treatment), the patient was subsequently withdrawn from the study. No DLTs or MEDI0562-related AEs resulted in death.

As shown in Table 3, the ORR was 4% (2/50); this included one patient with squamous cell carcinoma (SCC) of the larynx (0.03 mg/kg every 2-week group) and one with bladder cancer (3 mg/kg every 2 weeks), both of whom had irPR at the first tumor assessment (duration of response was 113 days for both patients). The patient with SCC of the larynx was PD-1 naïve at the start of the study and, following MEDI0562 treatment and nivolumab therapy (6 weeks' duration), had an OS duration of 14 months; the patient with bladder cancer had received prior pembrolizumab treatment and had an OS duration of 10 months following MEDI0562 treatment (patient received no subsequent anticancer therapy). The remaining patients had irSD (22/50, 44%), irPD (19/50, 38%), or were nonevaluable (7/50, 14%). The OS at 12 months was 47% (95% CI, 30.7–61.0) observed across all cohorts (Table 3).

Among the 55 patients in the as-treated population, six (6/55, 11%) received anti–PD-1 treatment prior to MEDI0562 (n = 3, nivolumab; n = 3, pembrolizumab). Of these six patients, four progressed following MEDI0562, one had SD, and one had a PR.

Median PFS was 3 (95% CI, 1.8–3.7) months, and PFS rate at 3 and 6 months was 53% (95% CI, 36.6–66.5) and 25% (95% CI, 12.2–40.7), respectively. Best change from baseline in tumor burden based on irRECIST is shown in Supplementary Fig. S1.

MEDI0562 exhibited linear pharmacokinetics following intravenous infusion of 0.03 to 10 mg/kg (Fig. 2). The “peak” concentration (C_max) of the treatment increased dose-proportionally from 1 to 193 μg/mL, and the area under the concentration–time curve from time 0 to the time of last quantifiable concentration (AUC_last) during the first dosing interval increased dose-proportionally from 3 to 823 μg*day/mL at 0.03 to 10 mg/kg, after the first dose. The mean serum clearance was 0.5 to 0.8 L/day and terminal half-life was 4 to 6 days.

Overall, 22 patients had biopsies taken. Of these 22 patients, samples acquired from 16 (73%) were viable to use for analyses and provided valid data.

ADAs against MEDI0562 were detected in 26 (51%) of the 51 patients who had samples postbaseline analyzed for ADAs. Variable ADA impact on pharmacokinetics was observed at doses below 3 mg/kg, while no impact was observed at doses of 3 and 10 mg/kg (Supplementary Fig. S2). Furthermore, ADAs did not appear to impact safety or pharmacodynamic responses to MEDI0562 (as measured by percentage changes from baseline in peripheral Ki67⁺ CD4⁺, Ki67⁺ CD8⁺, and OX40⁺ CD4⁺ memory T-cell levels) (Supplementary Fig. S3).

T-effector function was increased across all dose levels based on gene expression, flow cytometry, and IHC assessments in tumor/blood. A 1.5- to 3.0-fold increase in mean peaks of the percentage of peripheral Ki67⁺ CD4⁺ and Ki67⁺ CD8⁺ memory T cells across ascending dose levels was seen, with more sustained effects at higher dose levels (Fig. 3A and 3B). It was not possible to clearly define the duration of Ki67⁺ T-cell responses for each dose cohort due to differences in peak response and duration of change in each patient. However, we observed that patients receiving doses ≥0.3 mg/kg of MEDI0562 exhibited elevated Ki67⁺ T-cell responses that were sustained out to study day 84 or longer, independent of the presence of ADAs (Supplementary Fig. S3A and S3B). In addition, following treatment with MEDI0562, the percentage of OX40⁺CD4⁺ memory T cells decreased at dose levels ≥0.1 mg/kg, with more sustained effects observed at higher dose levels (Fig. 3C). These trends in change from baseline in peripheral T cells were not dose-dependent and did not reach statistical significance due, in part, to the small sample size tested at each dose and the variability of the timing associated with the maximum change from baseline across patients. In tumor, paired biopsies showed ≥2-fold increased expression of PD-L1⁺ or CD8⁺ cells in seven of 14 patients (Supplementary Fig. S4A), with a statistically significant 60% median reduction in OX40⁺ FOXP3⁺ cells at dose levels of 1 and 3 mg/kg compared with dose levels <1 mg/kg (P < 0.01; Fig. 3D). Intratumoral pharmacodynamic changes in OX40⁺ FOXP3⁺ Tregs and a gene expression signature of OX40 activation were more prominent in patients with high OX40 expression at baseline (> median) regardless of MEDI0562 dose or presence of ADAs (Supplementary Fig. S4B).

In this multicenter, open-label, single-arm, dose-escalation study, MEDI0562, a novel OX40 agonist, was safely administered at doses up to 10 mg/kg in patients with advanced solid tumors. Although no significant clinical activity was demonstrated, on-target pharmacodynamic effects were suggested by the observed increases in peripheral Ki67⁺ CD4⁺ and CD8⁺ memory T-cell proliferation and decreases in intratumoral OX40⁺ FOXP3⁺ cells. However, the small size of this study prevents us from confirming the pharmacodynamics of OX40 agonism in this setting.

Many tumors have OX40-expressing lymphocytes, and higher OX40 expression levels have been correlated with longer patient survival and better prognosis (14, 15). Therefore, OX40 has potential as an immune therapeutic target (2) and the use of mAbs to activate OX40 is a promising strategy to increase the antitumor activity of the immune system. Agonistic OX40 mAbs demonstrated promising antitumor activity in a number of preclinical models (10, 16–19), and there are several ongoing phase I/II clinical trials utilizing these new agents to treat advanced solid tumors (12). It has also been postulated that a combination of immunotherapies incorporating checkpoint inhibitors and OX40 agonists may provide enhanced efficacy over currently available treatments, providing additional benefits to patients with cancer (6).

Several phase I trials of OX40 agonists have reported preliminary safety findings (12). MOXR0916 alone or in combination with atezolizumab (anti–PD-L1) was well tolerated (20, 21); neither DLTs nor grade 3–5 AEs were observed with PF-04518600 (22, 23) at any of the doses tested. In the current analysis, no DLTs were observed with MEDI0562 at any doses tested up to 10 mg/kg and the MTD could not be determined. TRAEs were reported in two-thirds of study patients; most were grade 1 or 2, and there was no apparent dose relationship. Importantly, we found no major differences in the incidence or severity of TRAEs among the six dose groups thereby confirming that MEDI0562 is safely administered at doses up to 10 mg/kg. In addition, only two MEDI0562-related SAEs were reported, and no DLTs or MEDI0562-related AEs resulted in death.

It is notable that despite single-agent effects in preclinical models (10, 16–19), clinical activity observed with OX40 agonists alone has been minimal. In a preliminary analysis of the phase I PF-04518600 study (25 patients evaluable for response), one patient experienced irPR, and 15 experienced irSD (22). Another dose-finding study reported a best response of irSD in 11 patients treated with MOXR0916 alone for >6 months (20). In patients with locally advanced or metastatic solid tumors, MOXR0916 in combination with atezolizumab demonstrated partial response in two of 51 patients (21). Using irRECIST in the current study, we observed irPR in two patients (duration of response was 113 days for both patients) and irSD in 22 patients, four (18%) of whom had PFS >6 months.

The mean t_1/2 of serum MEDI0562 was 4 to 6 days and C_max and AUC_last increased dose-proportionally from 0.03 mg/kg every 2 weeks to 10 mg/kg. Variable ADA impact on pharmacokinetics was observed at doses <3 mg/kg, but was not evident at 3 or 10 mg/kg. The high rate (51%) of ADA detection is notable and is likely due to the nature of the antibody and its target, given that this is a humanized murine antibody and T-cell targeting immune-stimulatory agent. However, it is important to note that no clear trend was observed between the rate of ADA positivity postbaseline and dose levels. In addition, ADAs did not appear to impact the safety or the pharmacodynamic effects of MEDI0562 and thus are unlikely to have altered the efficacy of MEDI0562 in this setting; however, such conclusions can not be made in a study with such low response rates. Data with MOXR0916 also show linear pharmacokinetics and sustained peripheral blood OX40 receptor saturation at doses up to 40 mg every 2 weeks (20). PF-04518600 has exhibited increased exposure (AUC) with increasing doses during the first treatment cycle in patients with select advanced solid tumors (23).

It is widely known that the microenvironment surrounding a tumor influences the tumor biology and that tumor behavior is affected by its immunologic environment (24). The tumor microenvironment can negatively impact the immune system via immunosuppression, immunoevasion, and immunoediting (25). Increased concentrations of tumor-specific Teffs have been identified within the tumor environment and have been shown to improve prognosis in patients with different tumors (24). OX40 is expressed on both activated Teff and Treg cells, and leads to increased INFγ expression in the periphery and within the tumor microenvironment (24). The interaction of OX40 with the OX40 ligand results in enhanced CD4⁺ and CD8⁺ T-cell proliferation, stimulated cytokine production, and increased survival of antigen-specific memory T cells (10, 24, 26, 27). In addition, tumors with OX40⁺ cells have been shown to have increased CD8⁺ T-cell infiltration and higher CD8⁺ activation, suggesting that high-OX40 expression tumors may have pre-existing antitumor immunity (28). Thus, the immune-stimulating properties of OX40 agonists could overcome some of the immunosuppressive properties within the tumor environment. This hypothesis is supported by findings that a higher incidence of OX40 receptors expressed within tumor cells was correlated with increased chemosensitivity and an improved recurrence-free survival postchemotherapy in patients with ovarian cancer (24).

In our study, polychromatic flow cytometric analysis of longitudinal blood samples demonstrated transient increases in Ki67⁺ CD4⁺ and Ki67⁺ CD8⁺ memory T cells with MEDI0562 over pretreatment levels, which has been observed with other agonist OX40-targeting mAbs in preclinical mouse models and in patients with solid tumors (29–31). Of note, OX40⁺ CD4⁺ memory T cells decreased after the first dose of MEDI0562. This may be due to saturation of OX40 receptor occupancy, resulting in downregulation of the protein on the cell surface. Similar results have been reported with the OX40 agonistic mAb BMS-986178. Preclinical data suggested that maximal effects of BMS-986178 were achieved at approximately 20% receptor occupancy, with functional enhancement lost at higher drug concentrations as binding saturation occurred (32). These results were corroborated in a phase I/IIa study of BMS-986178 as monotherapy or combination therapy in patients with solid tumors; furthermore, soluble OX40 levels were also affected by the dosage levels of the combination treatment regimen (28). There is clearly a complex dose-response relationship between receptor saturation, receptor modulation, and induction of soluble receptor which requires further investigation before definitive therapeutic doses and schedules can be specified for OX40 mAbs administered with immune checkpoint inhibitors. The observation that patients treated with MEDI0562 in the current analysis exhibited increased Ki67⁺ CD4⁺ and CD8⁺ memory T-cell proliferation in the periphery and decreased intratumoral OX40⁺ FOXP3⁺ cells is consistent with the hypothesized dual mechanism of action of this agonist mAb (Supplementary Fig. S5). Furthermore, the implied enhanced pharmacodynamic effects in patients with high OX40 levels at baseline suggest that this may have the potential to be used as a selection marker to maximize the effect of MEDI0562 alone and/or in combination with other therapies.

The majority of patients in our study had advanced solid tumors and were heavily pretreated. Tumors commonly develop resistance mechanisms, including immune suppression, induction of tolerance, and systemic dysfunction of T-cell signaling as they progress (1, 2). Several immunotherapy treatments have shown clinical benefit in patients with refractory tumors, for example, anti–PD-1 in metastatic melanoma and non–small cell lung cancer (33, 34). In this study, six patients received prior treatment with a PD-1 checkpoint inhibitor (nivolumab or pembrolizumab); one of whom had a PR following MEDI0562, one had SD, and four progressed. It is hypothesized that combinations of CTLA-4 and PD-1 checkpoint inhibitors, together with OX40 agonists may provide greater benefits than in monotherapy. Preclinical data suggest that combining OX40 mAbs with checkpoint inhibitors enabled exploitation of the different mechanisms of action of these agents to refine the T-cell response and enhance tumor regression and host survival (9, 35–37). Furthermore, because the antitumor activity of immune therapies is principally a consequence of modulating the T-cell repertoire, several studies have examined T cells as biomarkers of potential activity; results indicated that combination therapy with an OX40 mAb plus a checkpoint inhibitor was able to positively influence the abundance, diversity, and functional phenotype of T-cell populations (9, 35, 38).

Several OX40 agonists have now reached early-stage clinical development as monotherapy and combination therapy, including MOXR0916, PF-04518600, BMS-986178, INCAGN01949, and GSK3174998 (6, 39). A combination trial of MEDI0562 with durvalumab (anti–PD-L1) and with tremelimumab (anti-CTLA-4) to treat advanced solid tumors is currently ongoing (NCT02705482). Bispecific antibodies targeting CTLA-4 or PD-1, plus OX40 are also entering clinical evaluation (40, 41). To date, only limited clinical efficacy results have been reported, and more mature data are required before any conclusions can be inferred regarding how these treatments may impact patients or their potential place within the current therapeutic algorithms. In addition, recent research has suggested additional questions which must be answered prior to routine use of such therapeutic combinations, including the optimal timing and sequencing of each treatment agent (42, 43), the most potent method of administration (44, 45), and the potential use of biomarkers to individualize treatment and predict efficacy outcomes (46, 47).

In conclusion, treatment with MEDI0562 was safe when administered at doses up to 10 mg/kg in a heavily pretreated patient population. In addition, on-target pharmacodynamic effects were suggested. Further evaluation with immune checkpoint inhibitors is ongoing.

B.S. Glisson reports grants from MedImmune (fees paid to institution to support this clinical trial) during the conduct of the study, as well as grants from Pfizer, ISA Pharmaceuticals, Turnstone Bio, and Cue Bio (fees to institution to support clinical research) outside the submitted work. R.S. Leidner reports other from MedImmune (institutional research support) during the conduct of the study and grants from MedImmune outside the submitted work. R.L. Ferris reports other from Aduro Biotech (consulting), Amgen (advisory board), Bain Capital Life Sciences (consulting), EMD Serono (advisory board), GlaxoSmithKline (advisory board), Iovance Biotherapeutics (consulting), Lilly (advisory board), MacroGenics (advisory board), Nanobiotix (consulting), Numab Therapeutics AG (advisory board), Oncorus (advisory board), Ono Pharmaceutical Co. (consulting), Pfizer (advisory board), PPD (advisory board), Regeneron Pharmaceuticals (advisory board), and Torque Therapeutics (consulting); grants and other from AstraZeneca/MedImmune (advisory board, clinical trial, research funding), Bristol-Myers Squibb (advisory board, clinical trial, research funding), Tesaro (advisory board, research funding), and TTMS (consulting, research funding); and grants from VentiRx Pharmaceuticals (research funding) outside the submitted work. J. Powderly reports other from AbbVie (clinical trial funding) during the conduct of the study; other from AstraZeneca (clinical trial funding, consultancy, advisory), Bristol-Myers Squibb (clinical trial funding, speakers bureau, consultancy), Merck (speakers bureau, advisory role, laboratory contract research), EMD Serono (clinical trial funding), Macrogenics (clinical trial funding), InCyte (clinical trial funding), RAPT Therapeutics (clinical trial funding), Alkermes (clinical trial funding), Tempest (clinical trial funding), Curis (clinical trial funding, consultancy, advisory role), Corvus (clinical trial funding), Top Alliance BioSciences (clinical trial funding), Precision for Medicine (clinical trial funding), MT Group (clinical trial funding), StemCell (clinical trial funding), Carolina BioOncology Institute, PLLC (founder, owner), and BioCytics, Inc. (founder, owner) outside the submitted work; additionally, Carolina BioOncology Institute PLLC and BioCytics Inc. are both developing intellectual property for personalized autologous cell therapies. N.A. Rizvi reports personal fees from AstraZeneca during the conduct of the study; personal fees from AbbVie, Boehringer Ingelheim, Calithera, Dracen, Editas, EMD Serono, G1 Therapeutics, Genentech, Gilead, GlaxoSmithKline, Merck, Novartis, Takeda; and personal fees and other from Arcus (stock options) and Gritstone (stock and stock options) outside the submitted work; and reports being listed as a coinventor on a pending patent regarding identifying determinants of cancer response to immunotherapy (PCT/US2015/062208) filed by Memorial Sloan Kettering Cancer Center and licensed to Personal Genome Diagnostics with royalties paid. Y. Zheng reports personal fees from AstraZeneca (employment and stock ownership) both during the conduct of the study and outside the submitted work. K. Streicher reports personal fees from AstraZeneca (employment and stock options) both during the conduct of the study and outside the submitted work. D.M. Townsley reports other from AstraZeneca/MedImmune (employee) during the conduct of the study. S.P. Patel reports other from AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Illumina, Rakuten, Paradigm, and Tempus (scientific advisory income), as well as research funding from Bristol-Myers Squibb, Eli Lilly, Incyte, AstraZeneca/MedImmune, Merck, Pfizer, Roche/Genentech, Xcovery, Fate Therapeutics, Genocea, and Iovance (to institution). No conflicts of interest were disclosed by the other authors.

B.S. Glisson: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. R.S. Leidner: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. R.L. Ferris: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. J. Powderly: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. N.A. Rizvi: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. B. Keam: Conceptualization, formal analysis, supervision, methodology, writing- original draft, writing-review and editing. R. Schneider: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. S. Goel: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. J.P. Ohr: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. J. Burton: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. Y. Zheng: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. S. Eck: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. M. Gribbin: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. K. Streicher: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing- review and editing. D.M. Townsley: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing. S.P. Patel: Conceptualization, formal analysis, supervision, methodology, writing-original draft, writing-review and editing.

The authors would like to thank the patients and their families for their participation in this study.

The study (NCT02318394) was funded by MedImmune, the manufacturer of MEDI0562.

The authors would like to acknowledge Rachel Cicchelli, PhD, of iMed Comms, Macclesfield, UK, an Ashfield Company, part of UDG Healthcare plc for medical writing support that was funded by MedImmune in accordance with Good Publications Practice (GPP3) guidelines (http://www.ismpp.org/gpp3).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.