Purpose: The CROSS trial established neoadjuvant chemoradiotherapy (nCRT) for patients with resectable esophageal adenocarcinoma (rEAC). In the PERFECT trial, we investigated the feasibility and efficacy of nCRT combined with programmed-death ligand-1 (PD-L1) inhibition for rEAC. Patients and Methods: Patients with rEAC received nCRT according to the CROSS regimen combined with five cycles of atezolizumab (1,200 mg). The primary endpoint was the feasibility of administering five cycles of atezolizumab in ≥75% patients. A propensity score–matched nCRT cohort was used to compare pathologic response, overall survival, and progression-free survival. Exploratory biomarker analysis was performed on repeated tumor biopsies. Results: We enrolled 40 patients of whom 85% received all cycles of atezolizumab. Immune-related adverse events of any grade were observed in 6 patients. In total, 83% proceeded to surgery. Reasons for not undergoing surgery were progression (n = 4), patient choice (n = 2), and death (n = 1). The pathologic complete response rate was 25% (10/40). No statistically significant difference in response or survival was found between the PERFECT and the nCRT cohort. Baseline expression of an established IFNγ signature was higher in responders compared with nonresponders (P = 0.043). On-treatment nonresponders showed either a high number of cytotoxic lymphocytes (CTL) with a transcriptional signature consistent with expression of immune checkpoints, or a low number of CTLs. Conclusions: Combining nCRT with atezolizumab is feasible in patients with rEAC. On the basis of our exploratory biomarker study, future studies are necessary to elucidate the potential of neoadjuvant immunotherapy in patient subgroups. See related commentary by Catenacci, p. 3269

Immunotherapy is a new anti-cancer treatment option, showing promising results in clinical trials. To investigate potential immune biomarkers in esophageal adenocarcinoma (EAC), we explored immune landscape patterns in the tumor microenvironment before and after neoadjuvant chemoradiation (nCRT). Sections from matched pretreatment biopsies and post-nCRT resection specimens (n = 188) were stained for (1) programmed death-ligand 1 (PD-L1, CD274); (2) programmed cell death protein 1 (PD-1, CD279), forkhead box P3 (FOXP3), CD8, pancytokeratin multiplex; and (3) an MHC class I, II duplex. The densities of tumor-associated immune cells (TAICs) were calculated using digital image analyses and correlated to histopathological nCRT response [tumor regression grade (TRG)], survival, and post-nCRT immune patterns. PD-L1 positivity defined by a combined positive score of >1 was associated with a better response post-nCRT (TRG 1-3 versus 4, 5, p = 0.010). In addition, high combined mean densities of CD8 + , FOXP3 + , and PD-1 + TAICs in the tumor epithelium and stroma of biopsies were associated with a better response (TRG 1-3 versus 4, 5, p = 0.025 and p = 0.044, respectively). Heterogeneous TAIC density patterns were observed post-nCRT, with significantly higher CD8 + and PD-1 + TAIC mean densities compared with biopsies (both p = 0.000). Three immune landscape patterns were defined post-nCRT: 'inflamed', 'invasive margin', and 'desert', of which 'inflamed' was the most frequent (57%). Compared with matched biopsies, resection specimens with 'inflamed' tumors showed a significantly higher increase in CD8 + density compared with non-inflamed tumors post-nCRT (p = 0.000). In this cohort of EAC patients, higher TAIC densities in pretreatment biopsies were associated with response to nCRT. This warrants future research into the potential of the tumor-immune landscape for patient stratification and novel (immune) therapeutic strategies.

Deletion of endoplasmic reticulum resident chaperone Grp78 results in activation of the unfolded protein response and causes rapid depletion of the entire intestinal epithelium. Whether modest reduction of Grp78 may affect stem cell fate without compromising intestinal integrity remains unknown. Here, we employ a model of epithelial-specific, heterozygous deletion by use of mice and organoids. We examine models of irradiation and tumorigenesis, both and Although we observed no phenotypic changes in heterozygous mice, heterozygous organoid growth was markedly reduced. Irradiation of heterozygous mice resulted in less frequent regeneration of crypts compared with nonrecombined (wild-type) mice, exposing reduced capacity for self-renewal upon genotoxic insult. We crossed mice to-mutant animals for adenoma studies and found that adenomagenesis in heterozygous- heterozygous mice was reduced compared with heterozygous controls (1.43 vs. 3.33; < 0.01). In conclusion, epithelium-specific heterozygosity compromises epithelial fitness under conditions requiring expansive growth such as adenomagenesis or regeneration after γ-irradiation. These results suggest that Grp78 may be a therapeutic target in prevention of intestinal neoplasms without affecting normal tissue. Heterozygous disruption of chaperone protein Grp78 reduces tissue regeneration and expansive growth and protects from tumor formation without affecting intestinal homeostasis. .

4045 Background: The CROSS study demonstrated the superiority of neoadjuvant chemoradiotherapy (nCRT) over surgery alone (van Hagen et al. NEJM. 2012). However, for resectable esophageal adenocarcinoma (rEAC) 5y survival is only 43%. PD1/PDL1 checkpoint inhibitors have shown promising efficacy for several cancer types, including esophageal cancer. To further improve outcomes in rEAC, we performed a phase II trial of nCRT combined with atezolizumab, a PD-L1 inhibitor. Methods: Pts with rEAC received standard dose CROSS regimen (5 cycles of IV: carboplatin AUC2, paclitaxel 50 mg/m2 and concurrent 23 fractions of 1.8 Gy on weekdays) with atezolizumab (5 cycles: 1200 mg IV, 3 weekly). Primary endpoint was the percentage of pts completing treatment with atezolizumab. Secondary endpoints included: toxicity, post-operative complications (Clavien-Dindo), Mandard score, R0 resection rate, PFS and OS. In total 40 pts will be enrolled. Results: Since July 2017, 39 pts have been enrolled (87% males, median age 63). Neoadjuvant treatment was completed by 31 pts and is ongoing in 8 pts. All cycles/fractions of nCRT were administered in 29/31 pts; 26 pts completed all cycles of atezolizumab, 24 pts finished complete neoadjuvant treatment. Reasons for missing any cycle of chemotherapy/atezolizumab included: toxicity (6 pts, in 3/6 pts immune-related adverse events (irAE)) and progression (1 pt). Grade 3-4 toxicity was observed in 15/31 pts (6/31 irAEs of any grade) which did not delay surgery. Thus far 23/31 pts were resected, 3 pts are planned for surgery, 3 pts had interval metastases preoperatively, 1 pt died during treatment (pulmonary embolism), and 1 pt declined surgery. Clavien-Dindo grade 3-4 complications were seen in 11/23 pts with no surgery related mortality. A pathological complete response (pCR), Mandard 1 was seen in 9/23 (39%) pts. All patients underwent an R0 resection. Updated results will be presented at the meeting. Conclusions: Based on data thus far, atezolizumab added to nCRT is feasible. A pCR was observed in 39% of patients, which is promising compared to 23% in the CROSS study. Treatment is associated with irAE which are manageable. Biomarker research will be performed on blood (circulating tumor DNA), tissue (immune microenvironment) and feces (microbiome). Clinical trial information: NCT03087864.