Phase I Study of Stereotactic Body Radiotherapy plus Nivolumab and Urelumab or Cabiralizumab in Advanced Solid Tumors

This article is featured in Highlights of This Issue, p. 5441

While anti-programmed death-1 (PD1) immunotherapy has improved outcomes across several tumor types, most patients with cancer do not benefit, and acquired resistance remains a major barrier to antitumor activity. Stereotactic body radiotherapy (SBRT) is a mechanistically rational combination partner based on its ability to ablate tumors with genomic resistance to immunotherapy such as human leukocyte antigen loss or JAK mutation (1, 2) and its capacity to induce antitumor immunity (3). We previously performed a phase I study of SBRT followed by pembrolizumab demonstrating feasibility and promising local control for irradiated lesions (3, 4). Furthermore, we observed an association between irradiated-lesion response and induction of IFN-associated gene expression and suppression of DNA damage response (3).

Murine modeling of SBRT and anti-PD1 therapy suggests that suboptimal T-cell activation and immunosuppressive tumor-associated macrophages (TAM) limit antitumor immunity. By inducing T-cell proliferation and cytokine secretion, CD137 agonism can enhance natural killer–cell antibody-dependent, cell-mediated cytotoxicity and stimulate antigen-presenting cells to produce larger responses in tumors with baseline susceptibility to immunotherapy (5). Moreover, preclinical studies of SBRT + anti-PD1 + CD137 agonism have reported abscopal responses mediated by CD8 T cells and cross-priming (6). In addition, colony-stimulating factor 1 receptor (CSF1R) is canonically expressed on macrophages, and anti-CSF1R agents can favorably alter the ratio of M1:M2-polarized TAMs leading to greater reductions in tumor volume when combined with radiation (7).

Building upon our prior investigation of multi-site SBRT with anti-PD1 immunotherapy, we designed a phase I protocol to investigate the safety of SBRT to 1–4 metastases given with concurrent and post-SBRT nivolumab+urelumab (CD137 agonist) or nivolumab+cabiralizumab (CSF-1R antagonist). Herein, we present organ-site specific SBRT dose determination from these first-in-human combinations and identify a potential association between baseline serum IL8 levels and resistance to SBRT + anti-PD1 + anti-CD137/anti-CSF1R.

This single-institution, phase I protocol had a primary objective of determining the recommended SBRT doses associated with a DLT rate ≤33% for radiated metastases in prespecified anatomic cohorts. DLT was defined as any possibly treatment-related Common Terminology Criteria for Adverse Events version 4.0 grade ≥3 toxicity within 3 months of day 1 of SBRT with the exception of DLTs affecting skin, mucosa, or any organ in the radiation field expected to resolve to grade 0–1 within 3 weeks of treatment without systemic steroids. DLTs were attributed to the combination of SBRT and immunotherapy on a weekly basis. Secondary endpoints included the incidence of any grade ≥3 toxicity, overall response rate (ORR) as assessed by the RECIST version 1.1, progression-free survival (PFS), and overall survival (OS) with exploratory endpoints including molecular correlative studies.

Enrolled patients were nonrandomly assigned to urelumab or cabiralizumab depending upon study drug availability and physician preference. Urelumab was preferred for tumor types “responsive” to immune checkpoint inhibition (e.g., histologies for which anti-PD/L1 therapy had been FDA approved or any histology with prior response to immunotherapy) whereas cabiralizumab was preferred in other tumor types (e.g., immunotherapy-“refractory” histologies such as microsatellite-stable colorectal or pancreatic carcinomas). Patients were allocated in this manner in an attempt to maximize potential clinical benefit based upon the mechanism of action of each study drug. For example, tumors more generally associated with a T cell–inflamed status and a microenvironment more likely to harbor T cells amenable to CD137 were allocated to the urelumab cohort. Meanwhile, patients harboring metastases from cancer types more generally associated with non-T cell–inflamed tumors were allocated to the cabiralizumab arm because this agent could facilitate greater activity of SBRT + anti-PD1 via depletion of immunosuppressive and putative M2-like macrophages. Triple immunotherapy with nivolumab, urelumab, and cabiralizumab was not investigated because of a lack of preclinical toxicity data to suggest that such an approach would be clinically tolerable and lack of interest in such an approach from the drug sponsor providing the study agents for this investigator-initiated clinical trial.

Each patient was assigned to one of five anatomic SBRT cohorts including (i) peripheral lung, (ii) central lung/mediastinal/thoracic lymph nodes, (iii) liver, (iv) spinal/paraspinal/osseous, and (v) abdominal/pelvic, although lesions in multiple anatomic cohorts could be radiated in each patient. Six participants were assigned to each of the five anatomic cohorts and monitored for ≥3 months for DLTs. If ≤1 patient in the cohort experienced DLT, the starting SBRT dose would be deemed acceptable. The protocol included prespecified de-escalation SBRT doses for all anatomic cohorts should >1 patient experience a DLT at the starting dose level. The study was approved by the institution review board at the University of Chicago (Chicago, IL; NCT03431948).

Key eligibility criteria for adult patients ≥18 years of age were: (i) a histologically confirmed advanced solid tumor with no curative treatment options that had progressed on standard therapy, (ii) Eastern Cooperative Oncology Group (ECOG) performance status 0–1, (iii) adequate organ/bone marrow function, and (4) measurable disease targetable for SBRT. Pertinent exclusion criteria included: (i) no investigational therapy or anti-cancer mAb treatment within 4 weeks of protocol-directed therapy, (ii) no anti-cancer therapy within 2 weeks of protocol-directed therapy, (iii) no active central nervous system metastases, (iv) no prior radiation to lesions targeted for SBRT, and (v) no comorbidities affecting the ability to receive study drugs. Prior receipt of immunotherapy was allowed. All participants were required to provide signed written informed consent, and the study was conducted in accordance with the principles of the Declaration of Helsinki.

SBRT was delivered to 1–4 metastases with targeted lesions selected at the treating radiation oncologist's discretion after considering causation of symptoms, size, and surrounding organs at risk. Lesions receiving >10% of the prescription dose from a previous radiation course were not eligible for SBRT. Starting doses and (proposed de-escalated doses) for each anatomic site were 45 (42) Gy in 3 fractions for peripheral lung lesions, 50 (47.5) Gy in 5 fractions for central lung lesions/mediastinal/thoracic lymph nodes, 45 (42) Gy in 3 fractions for liver lesions, 30 (27) Gy in 3 fractions for spinal/paraspinal/osseous lesions, and 45 (42) Gy in 3 fractions for abdominal/pelvic lesions. Linear accelerators were used to deliver photon-based radiation once daily every other day over a maximum period of 14 days. Dose constraints for organs at risk and de-escalation doses for each anatomic cohort were prespecified in the study protocol. The maximal planning target volume (PTV) for any single target was 65 mL with lesions larger than this partially radiated to a PTV ≤65 mL. Although uniform and symmetric contraction was preferred when generating a reduced-volume PTV, asymmetric contractions were permitted for minimization of dose to surrounding organs at risk.

Immunotherapy was started within 7 days of day 1 of SBRT. Nivolumab (10 mg/mL) was initially given at a dose of 240 mg i.v. every 2 weeks but later amended to 480 mg i.v. every 4 weeks based on a change in on-label use of the drug. Urelumab (5 mg/mL) was given as an 8 mg i.v. infusion every 4 weeks, and cabiralizumab (20 mg/mL) was given as a 4 mg/kg i.v. infusion every 2 weeks. The urelumab dose of 8 mg i.v. every 4 weeks was chosen to limit the known potential for hepatotoxicity and transaminitis with this agent when given as monotherapy and, while this dose is pharmacodynamically active, it may result in suboptimal agonism of the CD137 target (8). Dual-agent immunotherapy was continued until participants experienced disease progression, unacceptable toxicity, withdrawal of consent, or death.

Endpoints were assessed from day 1 of SBRT. Patients were clinically evaluated while receiving SBRT and with each cycle of immunotherapy. Radiographic tumor response evaluation was performed using CT obtained after nivolumab cycles 3 and 6 and every 12 weeks until disease progression. Tumor response was determined on the basis of RECIST version 1.1 (9) principles using change in aggregate sum of largest axial diameter of target lesions. Because RECIST target lesions were chosen without consideration of their radiation status, target lesions could include a mix of irradiated/nonirradiated tumors for patients with ≥1 target lesion(s) not receiving SBRT.

Pre-treatment serum samples were collected from consenting patients for evaluation of serum IL8 (n = 55/60 patients) based on the known pro-tumoral and immunosuppressive aspects of the molecule (10, 11). Tumor PD-L1 expression (i.e., tumor proportion score) was evaluated via IHC on pre-SBRT biopsies or archival tissue for n = 41/60 patients using antibody clone E1L3N which has been validated for PD-L1 scoring and has shown high concordance with commercially available PD-L1 antibodies (12). Of note, research biopsies were not mandatory due to a desire to spare patients the potential risks associated with this procedure, and a minority of patients ultimately underwent research biopsies.

Patient characteristics and DLTs were reported using descriptive statistics. The ORR was calculated using RECIST target lesions including irradiated and nonirradiated metastases with the response rate and mean percent change in the longest diameter sum of irradiated and unirradiated RECIST target lesions also calculated separately. A measure of clinical benefit from protocol-directed therapy was determined by calculating the proportion of patients with at least stable disease for ≥6 months from day 1 of SBRT. The Kaplan–Meier method was used to estimate PFS and OS. PFS was calculated until clinical or radiographic progression, death, or censoring at the time of last follow-up. OS was calculated until date of death or last known contact.

The association between pre-SBRT elevated baseline serum IL8 levels (reference range <57.8 pg/mL) and ORR [subdivided by (1) complete/partial response, (2) stable disease, (3) progressive disease/early progression or death/not evaluable] was performed using univariate ordinal logistic regression. The association between any tumor PD-L1 expression and the maximum percent change in aggregate diameter of irradiated target lesions as a continuous variable was analyzed using a two-sample t test. All statistical analyses were performed using JMP software, version 14.0 (SAS Institute).

As displayed in Fig. 1, 60 patients were enrolled with 23 (38%) assigned to the urelumab arm and 37 (62%) assigned to the cabiralizumab arm. Of these, all were evaluable for toxicity with 54 also clinically/radiographically evaluable for best overall response and 41 radiographically evaluable for best overall response at 3 months after receiving 3 months of protocol-directed therapy. Patient and tumor characteristics are listed in Table 1. The majority of patients were female (n = 40, 66%) with ECOG performance status 0 (n = 33, 55%) and a mean age of 58 years. This was a heavily pre-treated cohort with the median number of prior oncologic therapies being 5 [interquartile range (IQR): 3–7]. Eleven patients (18%) had received prior PD-(L)1 blockade and 24 of 41 (59%) patients with evaluable pre-SBRT tumor specimens had tumor PD-L1 IHC expression ≥1%. Supplementary Table S1 lists the tumor histologies for all patients with the most common types being colorectal (n = 13, 22%), breast (n = 8, 13%), pancreatic/ampullary (n = 7, 12%), endometrial (n = 6, 10%), and ovarian/fallopian (n = 6, 10%). Of patients with colorectal/endometrial cancer evaluable for microsatellite instability (n = 5, 4 colorectal, 1 endometrial), all were microsatellite stable.

Twelve patients were assigned to the peripheral lung anatomic cohort (20%), 14 to the central lung/cervical/mediastinal lymph node cohort (23%), 13 to the liver cohort (22%), 4 to the spinal/paraspinal/osseous cohort (7%), and 17 to the abdominopelvic cohort (28%). The majority of patients received SBRT to ≥1 metastases all within a single anatomic cohort location (n = 39, 65%), and most received SBRT to ≥2 lesions (n = 48, 80%). In total, 142 metastases received SBRT (n = 55 in urelumab arm, n = 87 in cabiralizumab arm) with 7 (12%) of these treated with partial SBRT to a PTV ≤65 mL. All patients received at least one infusion of immunotherapy within 7 days of day 1 of SBRT with immunotherapy continued for a median of 9.7 weeks (IQR: 4.4–16.8) among patients no longer receiving study drug (n = 57/60, 95%). Reasons for study drug discontinuation included disease progression (n = 33/57, 58%), unacceptable toxicity (n = 11/57, 19%), completion of 1 year of therapy (n = 5/57, 9%), death (n = 5/57, 9%), comorbidity (n = 2/57, 4%), and consent withdrawal (n = 1/57, 2%).

Median follow-up among living patients was 13.8 months (IQR: 8.8–18.4). DLT data for all patients and patients receiving ≥3 months of therapy or DLT within 3 months of starting therapy (n = 44/60, 73%) are displayed in Table 2. Overall, there were 7 DLTs (12%) among 60 patients with 4 of these occurring in 44 patients (16%) receiving ≥3 months of therapy or taken off study for DLT within 3 months. DLTs occurred in 1 of 12 patients in the peripheral lung cohort (8%), 2 of 14 patients in the central lung/cervical/mediastinal lymph node cohort (14%), 1 of 13 patients in the liver cohort (8%), 0 of 4 patients in the spinal/paraspinal/osseous cohort (0%), and 3 of 17 patients in the abdominopelvic cohort (21%). Common DLTs were grade ≥3 creatine phosphokinase (CPK) elevation (n = 3/7, 43%) which is a known, clinically irrelevant on-target laboratory finding associated with CSF1R blockade (13), and all DLTs occurred in patients in the cabiralizumab arm. Because the DLT rate for all SBRT anatomic cohorts was ≤33%, no prespecified SBRT dose de-escalation was recommended, and the initial SBRT doses for each anatomic site were deemed safe. Complete toxicity data including treatment-related grade ≥2 toxicities by anatomic cohort and all adverse events for enrolled patients are displayed in Supplementary Tables S2 and S3.

Nineteen patients (27%) received <3 months of therapy and/or were not imaged at 3 months after starting protocol-directed therapy. Of these, 10 experienced early clinical progression, 3 died, 2 experienced unacceptable comorbidity, 1 withdrew consent, and 3 experienced DLT prior to protocol-directed imaging. Interestingly, this cohort was seemingly enriched for patients with pancreatic/ampullary carcinoma (n = 6, 32%) or colorectal cancer (n = 5, 26%) with a significant association between assignment to the liver SBRT anatomic cohort and removal from the study before radiographic response assessment (P = 0.03). Among 54 patients evaluable for response either by imaging at 3 months (n = 41) or early clinical progression/death (n = 13), responses were complete in 2 patients (3%, n = 1 endometrial adenocarcinoma and n = 1 squamous cell carcinoma of unknown primary), partial in 7 patients (12%, n = 1 adenoid cystic carcinoma, n = 2 breast carcinoma, n = 1 endometrial adenocarcinoma, n = 1 prostate adenocarcinoma, n = 2 urothelial carcinoma), stable at 3 months in 12 patients (20%), and early progression or death in 31 patients (52%) with an associated ORR of 17%. A swimmer plot displaying duration of response by study drug and best overall RECIST response is displayed in Fig. 2. The median duration of response was 24 weeks (range: 11–61) with 2 responding patients developing subsequent progression and the remaining patients (n = 7) experiencing continued response at last follow-up.

Within the subset of 41 patients with 78 target lesions [n = 55/78 (71%) irradiated and n = 23/78 (29%) unirradiated] imaged at 3 months after receiving ≥3 months of protocol-directed therapy, responses were complete in 2 patients (5%), partial in 7 patients (17%), stable at 3 months in 12 patients (29%), and progression in 20 patients (49%) resulting in an ORR of 22%. Among the 9 responding patients, 7 were immunotherapy naïve, and there was no significant association between anatomic cohort categorization or number of lesions or distinct anatomic sites receiving SBRT and response. Both patients experiencing a complete response and 5 of 7 patients experiencing a partial response had all RECIST target lesions irradiated. The remaining 2 patients experiencing a partial response had SBRT to n = 0/1 and n = 1/2 RECIST target lesions, respectively. The number and proportion of patients receiving SBRT to all RECIST target lesions and/or all known metastases according to response is further detailed in Supplementary Table S4. The ORR among all participants (n = 60) was 15% if all nonevaluable patients are counted as nonresponders.

Waterfall plots for patients receiving ≥3 months of protocol therapy and imaging at 3 months are displayed in Fig. 2 for all target lesions as well as irradiated and unirradiated target lesions. The mean percent change in longest tumor diameter sum for irradiated target metastases was a decrease of 31% (SD: 31%) while the mean percent change in longest tumor diameter sum for unirradiated target metastases was an increase of 27% (SD: 77%). The response rate for imaged irradiated target lesions was 44% (n = 14/32), and the response rate for imaged unirradiated target lesions was 11% (n = 2/19). Of note, 32% of patients treated on protocol with imaging at 3 months (n = 13/41, 4 cabiralizumab/9 urelumab) experienced at least stable disease for ≥6 months, and the proportion of patients with disease stability for ≥6 months did not significantly differ between those receiving radiotherapy to ≥1 partially radiated metastasis >65 mL versus those receiving radiation to the entire PTV(s) for irradiated metastases ≤65 mL [n = 3/6 (50%) vs. n = 10/35 (29%), respectively, P = 0.36], although statistical inference is limited with this low sample size. Kaplan–Meier curves displaying estimated PFS and OS are presented in Fig. 3 with a median PFS of 3.0 months [95% confidence interval (CI): 2.9–4.8] and a median OS of 17.0 months (95% CI: 6.8–undetermined) for all enrolled patients.

The association between pre-SBRT serum IL8 level and ORR was statistically significant (P = 0.006) with all values above the reference range recorded in patients experiencing disease progression or early death or not evaluable for response as depicted in Fig. 4. Tumor PD-L1 expression was associated with the percent change in imaged irradiated target lesion longest diameter sum (P = 0.02) with the mean percent change being −8.4% and −45.3% among patients with tumor PD-L1 expression of 0% and ≥1%, respectively.

We performed a phase I study of SBRT with nivolumab + urelumab or nivolumab + cabiralizumab in patients with advanced solid tumors. Despite limitations including the heterogeneous patient population under study and nonrandom assignment of study drug, our results suggest acceptable toxicity and modest antitumor activity for these novel combinations. The overall DLT rate of 12% and the DLT rate of 16% for patients receiving ≥3 months of therapy or taken off study for DLT within 3 months suggest that the initial starting SBRT doses for each anatomic site are associated with acceptable toxicity and are appropriate for further prospective investigation. The ORR of 15% among 54 patients clinically or radiographically evaluable for response is similar to the 13% ORR reported in a phase I study of SBRT followed by pembrolizumab in a similar patient demographic with mixed histology, heavily pretreated tumors (4). Responses in unirradiated RECIST target lesions were observed in 11% of patients and more robust responses in irradiated RECIST target lesions were observed in tumors expressing PD-L1. Perhaps most interestingly, pre-SBRT serum IL8 levels were strongly associated with ORR suggesting a potential targetable mechanism of treatment resistance.

This study adds to a growing body of literature investigating combined radiation and immunotherapy as anti-cancer therapy in the metastatic setting with widely ranging ORRs from 13% to 56% (4, 14–17). One potential explanation for this marked variability in reported ORRs is the heterogeneous groups of patients under study. However, a more nuanced explanation may relate to differences in technical aspects of treatment such as SBRT dosing, immunotherapy combinations, and treatment sequencing. For example, Theelen and colleagues reported a modest ORR of 36% failing to meet the prespecified endpoint for clinical benefit for patients with metastatic non–small cell lung cancer receiving sequential low-dose SBRT (24 Gy in 3 fractions) and pembrolizumab (14) while Hammers and colleagues reported an impressive ORR of 56% among unirradiated lesions in patients with metastatic renal cell carcinoma receiving PD1-CTLA4 checkpoint blockade and higher dose, concurrent SBRT (50 Gy in 5 fractions; ref. 15). Although the optimal radiation dosing, fractionation, and sequencing to best synergize with immunotherapy remain unknown (18, 19), preclinical studies suggest that preexisting intratumoral T cells can survive even 20 Gy in a single fraction with preserved ability to mediate antitumor responses (20). Taken together with data suggesting a negative association between immunotherapy response and tumor burden (21) which may be optimally reduced following aggressive SBRT regimens, combining higher doses of SBRT with immunotherapy is a logical strategy.

Furthermore, partially radiating lesions >65 mL is a rather unique aspect of our study with results suggesting that a heterogeneous dose distribution within large metastases may be sufficient for immune priming and local tumor control. Responsible mechanisms include biological mediators beyond radiation-induced cell killing such as enhanced effector T-cell trafficking through portions of the metastasis receiving less than prescription dose (4). Additional support for this hypothesis includes our previous report demonstrating no significant association between the proportion of partially radiated large-volume tumors receiving ≥8–12 Gy/fraction and response to multi-site SBRT and sequential pembrolizumab as well as no significant difference in control for metastases receiving full- or partial-volume SBRT followed by sequential pembrolizumab (3). Related gene expression analyses suggest a molecular basis for these observations including SBRT-induced expression of genetic pathways governing B-cell development, antigen presentation, and dendritic cell maturation (22, 23) as well as SBRT-induced suppression of pathways intertwined with the cell cycle, histone modification, and DNA repair. Overall, these findings suggest that SBRT at doses used in the presently reported protocol and delivered to the entirety or portions of metastatic lesion(s) is capable of inducing type I/II IFN signaling and downregulating mitosis and homologous repair (3).

The theoretical advantages to combined SBRT and dual-agent immunotherapy have a biological basis including the previously discussed genetic changes (3) as well as SBRT-induced modulation of DNASE1 and TREX1 (24) and downregulation of genes governing immune escape (25). Despite these mediators of favorable biological interaction, our observed ORR of 17% among clinically and radiographically evaluable patients and the relative lack of responses in unirradiated lesions suggest that further study is necessary to clinically manifest this potential. It was hoped that CSF1R blockade could shift the balance of M1:M2 TAMs to limit macrophage-associated immunosuppression in lesions predisposed to poor anti-PD1/L1 response (7) and CD137 agonism could synergize with SBRT immune priming to boost and deepen antitumor activity in lesions with a baseline predilection for anti-PD1/L1 response (5, 6). One explanation for the lower than anticipated ORR is the heavily pretreated patient population with a high rate of early death or clinical progression suggesting enrichment for multifaceted immunorefractory phenotypes with escape mechanisms bypassing the PD1/L1 and CD137 axes as well as TAM-mediated immunosuppression. Biomarkers may identify individuals most likely to benefit from combined radioimmunotherapy as evidenced by a model including Epstein-Barr virus/human papillomavirus status, tumor mutation burden, and PD-L1 expression that significantly predicts ORR for patients with metastatic head and neck cancer receiving concurrent SBRT and nivolumab (26). The importance of PD-L1 expression as a potential biomarker for response to SBRT and dual-agent immunotherapy is bolstered by our own data demonstrating a significant association between the depth of response among irradiated RECIST target lesions and tumoral PD-L1 positivity.

Otherwise, targeting molecular pathways governing resistance to SBRT + immunotherapy represents a promising avenue of future investigation. The significant association between elevated pre-SBRT serum IL8 levels and poor responses in this study suggests one biologically rational candidate for prospective evaluation given the known role IL8 plays in mediating pro-tumorigenic restructuring of the tumor microenvironment (10, 27–28). The promise of this strategy is further highlighted by reports documenting an association between elevated IL8 levels and reduced responses to PD1/L1 blockade (10, 11) as well as an association between dynamic changes in serum IL8 and clinical efficacy of single- or dual-agent immune checkpoint inhibition (29). Because this may be mechanistically related to IL8's role in generating and maintaining a tumor microenvironment rich in neutrophils and monocytes while relatively deficient in cells mediating adaptive immune responses, combining anti-IL8 antibodies such as BMS-986253 (30) with SBRT plus anti-PD1/L1 may be a particularly promising approach to enhance antitumor responses. We have initiated a phase I/II trial to explore this possibility (NCT04572451) by investigating the safety and preliminary efficacy of SBRT + anti-PD1 + anti-IL8 therapy in patients with advanced solid tumors and detectable serum IL8.

In conclusion, our findings suggest that high-dose SBRT to 1–4 metastases may be safely combined with anti-PD1 + anti-CD137 or anti-CSF1R systemic agents and, to our knowledge, is the first report to investigate SBRT combined with novel checkpoint blockade beyond CTLA4 and/or PD1/L1. Tumor PD-L1 expression represents a promising biomarker that was associated with depth of irradiated RECIST target lesion response. Furthermore, targeting potential mediators of resistance such as IL8 may further enhance responses and warrants prospective investigation.

G.F. Fleming reports other support from Merck, Roche, AstraZeneca, Molecular Templates, Corcept, Compugen, Celldex, Sermonix, Abbvie, and CytomX; personal fees and other support from GSK outside the submitted work. T.G. Karrison reports grants from NCI during the conduct of the study. C.-Y. Liao reports personal fees from Eisai, Exelixis, Ipsen, Incyte, QED, Genentech, and Blueprint Medicine outside the submitted work. A.V. Desai reports other support from Eli Lilly, Ymabs Therapeutics Inc, Jubilant DraxImage, GlaxoSmithKline, Roche, and Merck outside the submitted work; and travel, accommodations, expenses support from GlaxoSmithKline; consulting/advisory role with Merck; Ology Medical Education; stock ownership with Pfizer; Viatris. M.J. Ratain reports grants from Bristol-Myers Squibb during the conduct of the study; grants from AbbVie, Bristol-Myers Squibb, Xencor, Boston Biomedical, and Corvus; grants and personal fees from Genentech; personal fees from multiple generic pharmaceutical companies, Arvinas, Ayala, Oncovalent, Pneuma Respiratory, Credit Suisse, EQRx, bluebird bio, and Cyclacel outside the submitted work; in addition, M.J. Ratain has a patent for royalties related to UGT1A1 genotyping licensed and with royalties paid from Mayo Medical and a patent for low dose tocilizumab for COVID-19 pending; and is director of Value in Cancer Care Consortium. R. Nanda reports personal fees from Fujifilm, Immunomedics, Gilead, Macrogenics, Merck, OncoSec, SeaGen, and G1 Therapeutics; grants from Arvinas, AstraZeneca, Celgene, Corcept Therapeutics, Genentech, Roche, Immunomedics, Gilead, Merck, OBI Pharm, Inc., Odonate Therapeutics, OncoSec, Pfizer, Taiho, and SeaGen outside the submitted work. O.M. Hahn reports personal fees from Elvesier - pathways chair outside the submitted work. P.H. O'Donnell reports other support from Bristol Myers Squibb during the conduct of the study; personal fees from Genentech, Roche, Merck, Astellas Pharma, SeaGen, Janssen, Pfizer, Nektar, and Dragonfly Therapeutics; grants from Boehringer Ingelheim, Merck, Genentech/Roche, AstraZeneca/Medimmune, Acerta Pharma, Janssen, SeaGen, Astellas Pharma, and Bristol Myers Squibb outside the submitted work. E.E. Vokes reports personal fees from AbbVie, AstraZeneca, Beigene, BioNTech, Eli Lilly, EMD Serono, Genentech, GlaxoSmithKline, Merck, and Novartis outside the submitted work. H.L. Kindler reports other support from BMS during the conduct of the study; personal fees, non-financial support, and other support from AstraZeneca, Merck; personal fees from Aldeyra, Kyowa, and Novocure, personal fees and other support from Bayer, BMS, and Deciphera; personal fees and non-financial support from Boehringer Ingelheim, Paredox, and Inventiva; other support from Aduro, Harpoon, GSK, Lilly, Inbrx, Polaris, Macrogenics, Verastem, and Blueprint outside the submitted work. R.R. Weichselbaum reports stock and other ownership interests with Boost Therapeutics, Immvira LLC, Reflexion Pharmaceuticals, Coordination Pharmaceuticals Inc., Magi Therapeutics, and Oncosenescence; has served in a consulting or advisory role for Aettis Inc., AstraZeneca, Coordination Pharmaceuticals, Genus, Merck Serono S.A., Nano proteagen, NKMax America Inc, Shuttle Pharmaceuticals, and Highlight Therapeutics SL; reports research grants with Varian and Regeneron; and has received compensation including travel, accommodations, or expense reimbursement from AstraZeneca, Boehringer Ingelheim LTD, and Merck Serono S.A. S.J. Chmura reports other support from Astellas Pharma, Genentech, and AstraZeneca; grants from BMS and Merck, outside the submitted work. J.J. Luke reports grants from Bristol-Myers Squibb outside the submitted work; and scientific advisory board membership with (no stock) 7 Hills, Fstar, RefleXion, Xilio (stock) Actym, Alphamab Oncology, Arch Oncology, Kanaph, Mavu, Onc.AI, Pyxis, and Tempest; reports consultancy with compensation with Abbvie, Alnylam, Bayer, Bristol-Myers Squibb, Checkmate, Crown, Cstone, Eisai, EMD Serono, Flame, Genentech, Gilead, Kadmon, KSQ, Janssen, Immunocore, Inzen, Macrogenics, Merck, Mersana, Nektar, Novartis, Pfizer, Regeneron, Ribon, Rubius, Silicon, Synlogic, TRex, Werewolf, and Xencor; reports research support from (all to institution for clinical trials unless noted) AbbVie, Agios (IIT), Array (IIT), Astellas, Bristol-Myers Squibb (IIT & industry), Corvus, EMD Serono, Fstar, Genmab, Ikena, Immatics, Incyte, Kadmon, KAHR, Macrogenics, Merck, Moderna, Nektar, Numab, Replimmune, Rubius, Spring bank, Synlogic, Takeda, Trishula, Tizona, and Xencor; has patents: (both provisional) Serial #15/612,657 (Cancer Immunotherapy), PCT/US18/36052 (Microbiome Biomarkers for Anti-PD-1/PD-L1 Responsiveness: Diagnostic, Prognostic and Therapeutic Uses Thereof). No disclosures were reported by the other authors.

C.C. Foster: Data curation, formal analysis, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. G.F. Fleming: Investigation, writing–review and editing. T.G. Karrison: Data curation, formal analysis, writing–review and editing. C.-Y. Liao: Investigation, writing–review and editing. A.V. Desai: Investigation, writing–review and editing. J.W. Moroney: Investigation, writing–review and editing. M.J. Ratain: Investigation, writing–review and editing. R. Nanda: Investigation, writing–review and editing. B.N. Polite: Investigation, writing–review and editing. O.M. Hahn: Investigation, writing–review and editing. P.H. O'Donnell: Investigation, writing–review and editing. E.E. Vokes: Investigation, writing–review and editing. H.L. Kindler: Investigation, writing–review and editing. R. Hseu: Resources, data curation, project administration. L.A. Janisch: Data curation, investigation, project administration. J. Dai: Investigation, writing–review and editing. M.D. Hoffman: Investigation, writing–review and editing. R.R. Weichselbaum: Investigation, writing–review and editing. S.P. Pitroda: Investigation, writing–review and editing. S.J. Chmura: Conceptualization, resources, data curation, supervision, investigation, methodology, project administration, writing–review and editing. J.J. Luke: Conceptualization, resources, data curation, supervision, investigation, methodology, project administration, writing–review and editing.

Study drug but not funding was provided by Bristol-Myers Squibb as an investigator-sponsored trial.

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.