M. Oliva Bernal

3098 Background: Therapeutic augmentation of the intestinal microbiome to improve immunotherapy outcomes is an active area of investigation. Microbial Ecosystem Therapeutics (METs) are consortia of human-derived bacteria designed to be reproducible, scalable and safe alternatives to fecal transplant. MET4 is a first-in-class consortium of taxa associated with immune checkpoint inhibitor (ICI)-responsiveness. Here we describe preliminary results of MET4-IO, an interventional trial assessing the safety and ecological effects of MET4 in ICI recipients. Methods: MET4-IO is a randomized investigator-initiated trial, evaluating MET4 in solid cancer patients treated with ICI. MET4-IO involves 3 cohorts of 65 total patients: Group A, a safety cohort of 5 patients already on ICI; Group B, patients starting ICI, randomized 3:1 to receive MET4 or not; Group C, patients on ICI who experience radiological progression but not clinical deterioration, randomized 1:1 to receive MET4 or not. Stool and blood samples are collected at baseline and 4-5 additional time-points. For this interim analysis, 16S rRNA gene sequencing was performed on fecal specimens. Shannon diversity, relative abundance (RA), number and fold-change of MET4 taxa > RA 0.01 were assessed and compared to controls. Results: As of January 26, 2020, 21 patients were enrolled (A = 5,B = 12,C = 4), and 15 (71%) received MET4. The mean age was 65.9 years, 40% were females, 52% had head and neck cancer and 19% melanoma. Sixteen patients (76%) were treated with an anti-PD1 agent as monotherapy and 5 with a combination of anti-PD1 and anti-CTLA4 antibodies. G3-4 toxicities (CTCAEv5.0) attributed to ICI were observed in 13% vs. 17% of MET4 exposed and control patients, respectively. Three patients (20%) experienced toxicities attributed to MET4, all grade 1 except G2 dyspepsia in 1 patient. A greater number of MET4-associated taxa were detectable in MET4 recipients than controls (p < 0.01), with a trend towards higher cumulative RA (p = 0.10). No significant change in Shannon diversity after MET4 was observed, however controls were more likely to lose diversity overtime than MET4 recipients (p = 0.05). Colonization with MET4 varied by recipient and by taxon. Bifidobacterium, Collinsella and Enterococcus were significantly more common and abundant in MET4 recipients than controls. Conclusions: In this cohort, MET4 treatment was safe and associated with higher MET4-associated taxa in recipients than controls. Further analyses including peripheral blood immunophenotyping are ongoing. Clinical trial information: NCT03686202 .

6569 Background: Sitravatinib (receptor TKI against TYRO3, AXL, MERTK and VEGF family of receptors) is predicted to increase M1-type tumor-associated macrophages (TAMs) and decrease MDSCs in the tumor microenvironment. SNOW is a window-of-opportunity study evaluating the immunogenic and antitumor effects of preoperative sitravatinib and nivolumab in patients (pts) with OCC. Early results demonstrated the combination was safe and active (Oliva et al, SITC 2019). Biomarker analyses and updated results are presented. Methods: Pts with untreated T2-4a, N0-2 or T1>1cm-N2 OCC are eligible. All pts receive oral sitravatinib 120mg daily from day (D) 1 up to 48h pre-surgery and 1 dose of Nivolumab 240mg on D15. Surgery planned between D23-D30. Standard of care adjuvant radiotherapy given based on clinical stage. Tumor pictures, fresh tumor biopsies, blood samples taken at baseline, D15 and pre-surgery. Tumor flow cytometry and multiplex immunofluorescence staining performed on all biopsies to study changes in immune-cell populations. Tumor whole-exome sequencing (WES) performed on baseline biopsies. Results: As of Jan 31 st 2020, 10 out of 12 planned pts were enrolled. Study treatment was well-tolerated: only 1 pt had grade (G) >3 toxicity (hypertension) and 1 pt required surgery delay due to G2 thrombocytopenia. None had intraoperative complications. 1 pt had wound infection and tracheostomy bleeding 11 days post-surgery, possibly-related to study drugs. All pts had tumor reduction, 9/10 had pathological downstaging, including 1 complete response (Table). All had clear margins with no extranodal extension; none required adjuvant chemotherapy. All pts are alive with no recurrence (median follow-up= 69 weeks). Lower % of MDSCs and increased % of M1-TAMs and M1:M2 ratio trend was seen at D15 and pre-surgery, with stronger effect in major responders. Best responders (Pts S1-S2) had higher % of PD-L1+ TAMs at baseline. Tumor WES revealed an HRAS G12D mutation in pt S2 and a BLM mutation (DNA repair) in pt S6 (no downstaging). Conclusions: Pharmacodynamic analyses support the antitumor and immune effects of sitravatinib and nivolumab in OCC. Immune pathological response assessment and transcriptomics are on-going. Clinical trial information: NCT03575598 . [Table: see text]

6045 Background: The ROMA LA-OPSCC ( NCT03759730) study prospectively evaluated the oral and gut microbiota in a single-centre cohort of LA-OPSCC patients (pts) receiving chemoradiotherapy (CRT). Methods: LA-OPSCC pts treated with definitive CRT (IMRT plus single-agent cisplatin) were eligible. Oral swabs over the tumor site and stool samples were collected at baseline and end of CRT (EOT). Taxonomic profiles were generated by 16S rRNA sequencing. ANOSIM/Kruskal-Wallis tests were used to identify differences between baseline and EOT samples. Results: A total of 96 samples were collected from 24 evaluable pts (100% compliance). Baseline characteristics: median age = 61 (range, 50-71); smoking status current/former/never = 5/11/8; HPV+/- = 23/1; stage I/II/III/IVA = 7/7/9/1; use of antibiotics = 12 pts. In oral swabs, decreased Shannon diversity ( p< 0.01) and changes in abundance (adjusted p value: q< 0.05) of multiple taxa including Prevotella, Veillonella, and Streptococcuswere observed at EOT vs baseline. Stool diversity did not differ between baseline and EOT ( p= 0.42), but abundance of Ruminoccocus and Roseburia decreased ( q< 0.05). CRT-associated changes remained significant when controlled for stage, smoking, antibiotics, cisplatin dose and mucositis grade ( p< 0.01). In HPV+ pts, stage I-II baseline oral swabs had higher relative abundance of Clostridium IV ( p= 0.02) and Escherichia ( p= 0.04) than stage III, which had higher Fusobacterium ( p =0.03) and Gemella ( p< 0.01). Relative abundance of Actinobacteria (p < 0.01), Proteobacteria (p < 0.01) and Firmicutes (p = 0.03) was higher in stool from stage III pts . Akkermansia muciniphila was present in 57% of the stage I-II stool samples, and 11% in stage III ( p= 0.04). Conclusions: CRT in LA-OPSCC is associated with increases in potentially pathogenic genera in the oropharynx. HPV+ stage III disease was associated with higher Fusobacterium in the oropharynx, which has been implicated in tumor metastases, and with decreased prevalence of the immunotherapy-response-associated species Akkermansia in stool. These preliminary observations suggest an opportunity for the evaluation of IO based therapies or manipulation of the gut microbiota in this patient population. Clinical trial information: NCT03759730.

2653 Background: Pharmacodynamic biomarkers (PD) are considered fundamental for go/no-go decisions in phase 1 trials. Despite an increase in the availability of blood-based biomarker assays, the requirement of invasive non-diagnostic research tumor biopsies for trial eligibility remains common. In the immuno-oncology (IO) era, the impact of PD analysis for the confirmation of biologic activity and recommended phase 2 dose (RP2D) has not been investigated. Methods: Phase 1 studies from 01/2014 to 12/2018 were reviewed. Among 12053 abstracts screened, a total of 143 phase I-IO trials were identified. Characteristics of studies that included on-treatment PD biomarkers (tissue-derived, blood-based and radiomic) were extracted and analyzed. Outcomes from the biomarker data in terms of proof of mechanism/biologic activity and statistically significant correlation with clinical benefit (objective response or survival) were collected. Authors’ statements on the influence of PD results on RP2D were also noted. Results: Out of 143 phase 1 IO trials, 107 (75%) were monotherapy. The most frequent IO evaluated were vaccines (41%), cell therapy (16%), immunomodulators (13%) and cytokines (7%). Of the 36 combination studies, 20 (61%) included a second IO drug while 16 (39%) included molecular-targeted agents. Only 18 of 143 studies (12%) did not report any PD data. Of the remaining 125 studies, tissue-derived PD (t-PD) biomarkers alone, blood-based PD (b-PD) biomarkers alone, both t-PD and b-PD biomarkers, and imaging biomarkers were tested in 3 (2%), 97 (78%), 25 (20%), and 7 (6%), respectively. Demonstration of proof of mechanism/biologic activity only were reported in 16/28 (57%), 80/122 (66%) and 4/7 (57%) of the t-PD, b-PD and imaging biomarker studies, respectively. Significant correlation with clinical benefit was reported in 2/28 (7%), 7/122 (6%) and 0/7 (0%) of the t-PD, b-PD and imaging biomarker studies, respectively; these involved 4 vaccines (1 in combination with PD1 blockade), 1 cell therapy and 1 oncolytic virus (in combination with CTLA4 blockade). Among 35 b-PD studies with negative results, 5 also performed t-PD biomarkers, all with negative results. Notably, 3 out of 10 t-PD studies with negative results reported concurrent positive b-PD results. Based on the published reports, authors stated that biomarker results helped with RP2D determination in 16/28 (57%) of t-PD and 78/122 (64%) of b-PD studies. Conclusions: Our results suggest that in the IO era, most studies perform PD analysis, with similar proportions of t-PD and b-PD showing proof of mechanism/biologic activity. IO PD biomarkers have limited correlation with clinical benefit. Many authors considered IO PD biomarkers to be relevant in RP2D decisions, but this needs confirmation by other measures of impact. With continued technological developments utilizing circulating biomarkers, b-PD may ultimately replace many t-PD tests in future IO studies.