ABSTRACT
PURPOSE: Classic Hodgkin lymphoma (cHL) is a B cell lymphoma that occurs primarily in young adults and, less frequently, in elderly individuals. A hallmark of cHL is the exceptional scarcity (1-5%) of the malignant Hodgkin Reed-Sternberg (HRS) cells within a network of non-malignant immune cells. Molecular determinants governing the relationship between HRS cells and their proximal microenvironment remain largely unknown. EXPERIMENTAL DESIGN: We performed spatially resolved multiplexed protein imaging and transcriptomic sequencing to characterize HRS cell states, cellular neighborhoods, and gene expression signatures of 23.6 million cells from 36 newly diagnosed Epstein-Barr virus (EBV) positive and EBV-negative cHL tumors. RESULTS: We show that MHC-I expression on HRS cells is associated with immune inflamed neighborhoods containing CD8+ T cells, MHC-II+ macrophages, and immune checkpoint expression (i.e., PD-1 and VISTA). We identified spatial clustering of HRS cells, consistent with the syncytial variant of cHL, and its association with T cell excluded neighborhoods in a subset of EBV-negative tumors. Finally, a subset of both EBV-positive and EBV-negative tumors contained regulatory T cells high neighborhoods harboring HRS cells with augmented proliferative capacity. CONCLUSIONS: Our study links HRS cell properties with distinct immunophenotypes and potential immune escape mechanisms in cHL.
ABSTRACT
Multiplexed imaging offers a powerful approach to characterize the spatial topography of tissues in both health and disease. To analyze such data, the specific combination of markers that are present in each cell must be enumerated to enable accurate phenotyping, a process that often relies on unsupervised clustering. We constructed the Pan-Multiplex (Pan-M) dataset containing 197 million distinct annotations of marker expression across 15 different cell types. We used Pan-M to create Nimbus, a deep learning model to predict marker positivity from multiplexed image data. Nimbus is a pre-trained model that uses the underlying images to classify marker expression across distinct cell types, from different tissues, acquired using different microscope platforms, without requiring any retraining. We demonstrate that Nimbus predictions capture the underlying staining patterns of the full diversity of markers present in Pan-M. We then show how Nimbus predictions can be integrated with downstream clustering algorithms to robustly identify cell subtypes in image data. We have open-sourced Nimbus and Pan-M to enable community use at https://github.com/angelolab/Nimbus-Inference.
ABSTRACT
Colitis induced by treatment with immune-checkpoint inhibitors (ICI), termed irColitis, is a substantial cause of morbidity complicating cancer treatment. We hypothesized that abnormal fecal microbiome features would be present at the time of irColitis onset and that restoring the microbiome with fecal transplant from a healthy donor would mitigate disease severity. Herein, we present fecal microbiota profiles from 18 patients with irColitis from a single center, 5 of whom were treated with healthy-donor fecal microbial transplantation (FMT). Although fecal samples collected at onset of irColitis had comparable α-diversity to that of comparator groups with gastrointestinal symptoms, irColitis was characterized by fecal microbial dysbiosis. Abundances of Proteobacteria were associated with irColitis in multivariable analyses. Five patients with irColitis refractory to steroids and biologic anti-inflammatory agents received healthy-donor FMT, with initial clinical improvement in irColitis symptoms observed in four of five patients. Two subsequently exhibited recurrence of irColitis symptoms following courses of antibiotics. Both received a second "salvage" FMT that was, again, followed by clinical improvement of irColitis. In summary, we observed distinct microbial community changes that were present at the time of irColitis onset. FMT was followed by clinical improvements in several cases of steroid- and biologic-agent-refractory irColitis. Strategies to restore or prevent microbiome dysbiosis in the context of immunotherapy toxicities should be further explored in prospective clinical trials.
Subject(s)
Biological Products , Colitis , Gastrointestinal Microbiome , Humans , Fecal Microbiota Transplantation/adverse effects , Prospective Studies , Dysbiosis/therapy , Dysbiosis/etiology , Treatment Outcome , Colitis/therapy , Colitis/complicationsABSTRACT
Programmed death-1 (PD-1) inhibitors are approved for therapy of gynecologic cancers with DNA mismatch repair deficiency (dMMR), although predictors of response remain elusive. We conducted a single-arm phase 2 study of nivolumab in 35 patients with dMMR uterine or ovarian cancers. Co-primary endpoints included objective response rate (ORR) and progression-free survival at 24 weeks (PFS24). Secondary endpoints included overall survival (OS), disease control rate (DCR), duration of response (DOR) and safety. Exploratory endpoints included biomarkers and molecular correlates of response. The ORR was 58.8% (97.5% confidence interval (CI): 40.7-100%), and the PFS24 rate was 64.7% (97.5% one-sided CI: 46.5-100%), meeting the pre-specified endpoints. The DCR was 73.5% (95% CI: 55.6-87.1%). At the median follow-up of 42.1 months (range, 8.9-59.8 months), median OS was not reached. One-year OS rate was 79% (95% CI: 60.9-89.4%). Thirty-two patients (91%) had a treatment-related adverse event (TRAE), including arthralgia (n = 10, 29%), fatigue (n = 10, 29%), pain (n = 10, 29%) and pruritis (n = 10, 29%); most were grade 1 or grade 2. Ten patients (29%) reported a grade 3 or grade 4 TRAE; no grade 5 events occurred. Exploratory analyses show that the presence of dysfunctional (CD8+PD-1+) or terminally dysfunctional (CD8+PD-1+TOX+) T cells and their interaction with programmed death ligand-1 (PD-L1)+ cells were independently associated with PFS24. PFS24 was associated with presence of MEGF8 or SETD1B somatic mutations. This trial met its co-primary endpoints (ORR and PFS24) early, and our findings highlight several genetic and tumor microenvironment parameters associated with response to PD-1 blockade in dMMR cancers, generating rationale for their validation in larger cohorts.ClinicalTrials.gov identifier: NCT03241745 .