Anti-CD19 chimeric antigen receptor T cell (CAR-T) therapy has transformed the care of patients with relapsed/refractory large B cell lymphoma (LBCL). However, approximately 60% of CAR-T recipients ...ultimately will experience disease recurrence or progression. Salvage therapies after CAR-T treatment failures are of limited efficacy and have a short duration of response. The objective of the present study was to evaluate the role of allogeneic hematopoietic cell transplantation (allo-HCT) after CAR-T therapy in LBCL patients. This was a multicenter observational study reporting the outcome of 39 adult LBCL patients who underwent allo-HCT following anti-CD19 CAR-T therapy. The median patient age was 47 years (range, 20 to 68 years). HLA-matched sibling, HLA-matched unrelated, and alternative donors were used in 36%, 36%, and 28% of transplantations, respectively. Conditioning regimens were primarily of low or intermediate intensity. Disease status at allo-HCT was complete response in 41%, partial response in 38%, and progressive disease in 21%. Allo-HCT was performed at a median of 127 days (range, 82 to 206 days) after CAR-T therapy. A high incidence of hepatic toxicity (28%), including sinusoidal obstruction syndrome (15.4%; 95% confidence interval; CI, 6.2% to 28.5%), was observed. The 1-year cumulative incidence of grade II-IV and grade III-IV acute graft-versus-host disease (GVHD) was 38.5% (95% CI, 23.2% to 53.6%) and 15.4% (95% CI, 6.1% to 28.5%), respectively. The 2-year cumulative incidence of moderate-severe chronic GVHD was 11.1% (95% CI, 3.3% to 24.3%). Overall, 2-year nonrelapse mortality and relapse/progression incidence were 26% (95% CI, 13% to 41%) and 43% (95% CI, 27% to 59%), respectively. With a median follow-up of 32 months, the 2-year overall survival (OS) and progression-free survival (PFS) were 45% (95% CI, 31% to 66%) and 31% (95% CI, 19% to 50%), respectively. In multivariable analyses, pre-HCT elevated lactate dehydrogenase level and transformed lymphoma were predictive of OS and PFS, respectively. Our data suggest that allo-HCT after anti-CD19 CAR-T treatment failure is feasible with a relatively promising efficacy but possibly high toxicity rate.
Oral mucositis (OM) is a debilitating multifactorial complication of hematopoietic stem cell transplantation (HSCT). We hypothesized that the oral microbiome is disturbed during transplantation and ...may partially account for the variability in OM severity across patients.
In this single-center study, we prospectively collected saliva samples weekly in a cohort of 184 adults undergoing allogeneic HSCT. A total of 625 samples were collected starting from 7 days before HSCT to 34 days post-transplant. Samples underwent 16S rRNA gene sequencing. Sixty pre- and post-HSCT samples underwent mass spectrometry-based metabolomic profiling (Metabolon, Durham, NC). OM was graded according to the Common Terminology Criteria for Adverse Events (CTCAE) v4.0.
Myeloablative conditioning regimens were used in 70% of patients. The median time to OM development was 7 days; 58% and 43% of patients developed grade 2-4 and grade 3-4 OM, respectively. Pre-transplantation oral α-diversity, as measured by the Shannon Index, was similar to that of healthy controls (n=20) (p=0.460), but later decreased, reaching a nadir on day 12 (Fig. A). Among patients with grade 2-4 OM, the reduction in α-diversity over time was more pronounced compared to transplant-recipients having no or only mild OM (grade 0-1) (Fig. B).
In the group treated with MTX (n=135), patients harboring Rothia, Kingella, and Atopobium in the saliva before stem cells infusion were more likely to develop grade 3-4 OM than their counterparts. Furthermore, in samples collected between days 7 to 13, there was a higher abundance of Methylobacterium among patients with grade 3-4 OM, while Treponema and the genus TG5 (Fig. 3) were more common in patients with grade 0-1 OM (Fig. C).
Finally, metabolic profiling of saliva collected pre and post HSCT showed marked changes before and after transplantation among patients who developed mucositis. Microbiome-associated metabolites enriched in patients with mucositis included histidine, phenylalanine, tyrosine, and tryptophan.
In this largest analysis of the oral microbiome in HSCT recipients, we demonstrate that the oral microbiota is disrupted during allogeneic HSCT. Microbiota injury is more profound in patients who developed oral mucositis. Our findings indicate that there may be a distinct microbiological and metabolic signature in patients with mucositis, with some changes preceding the clinical phenotype. These may serve as potential intervention points to treat and prevent mucositis.
Translating spoken languages into Sign languages is necessary for open communication between the hearing and hearing-impaired communities. To achieve this goal, we propose the first method for ...animating a text written in HamNoSys, a lexical Sign language notation, into signed pose sequences. As HamNoSys is universal by design, our proposed method offers a generic solution invariant to the target Sign language. Our method gradually generates pose predictions using transformer encoders that create meaningful representations of the text and poses while considering their spatial and temporal information. We use weak supervision for the training process and show that our method succeeds in learning from partial and inaccurate data. Additionally, we offer a new distance measurement that considers missing keypoints, to measure the distance between pose sequences using DTW-MJE. We validate its correctness using AUTSL, a large-scale Sign language dataset, show that it measures the distance between pose sequences more accurately than existing measurements, and use it to assess the quality of our generated pose sequences. Code for the data pre-processing, the model, and the distance measurement is publicly released for future research.
Translating spoken languages into Sign languages is necessary for open communication between the hearing and hearing-impaired communities. To achieve this goal, we propose the first method for ...animating a text written in HamNoSys, a lexical Sign language notation, into signed pose sequences. As HamNoSys is universal by design, our proposed method offers a generic solution invariant to the target Sign language. Our method gradually generates pose predictions using transformer encoders that create meaningful representations of the text and poses while considering their spatial and temporal information. We use weak supervision for the training process and show that our method succeeds in learning from partial and inaccurate data. Additionally, we offer a new distance measurement that considers missing keypoints, to measure the distance between pose sequences using DTW-MJE. We validate its correctness using AUTSL, a large-scale Sign language dataset, show that it measures the distance between pose sequences more accurately than existing measurements, and use it to assess the quality of our generated pose sequences. Code for the data pre-processing, the model, and the distance measurement is publicly released for future research.
Diffusion-based generative models have recently shown remarkable image and video editing capabilities. However, local video editing, particularly removal of small attributes like glasses, remains a ...challenge. Existing methods either alter the videos excessively, generate unrealistic artifacts, or fail to perform the requested edit consistently throughout the video. In this work, we focus on consistent and identity-preserving removal of glasses in videos, using it as a case study for consistent local attribute removal in videos. Due to the lack of paired data, we adopt a weakly supervised approach and generate synthetic imperfect data, using an adjusted pretrained diffusion model. We show that despite data imperfection, by learning from our generated data and leveraging the prior of pretrained diffusion models, our model is able to perform the desired edit consistently while preserving the original video content. Furthermore, we exemplify the generalization ability of our method to other local video editing tasks by applying it successfully to facial sticker-removal. Our approach demonstrates significant improvement over existing methods, showcasing the potential of leveraging synthetic data and strong video priors for local video editing tasks.
Given the remarkable results of motion synthesis with diffusion models, a natural question arises: how can we effectively leverage these models for motion editing? Existing diffusion-based motion ...editing methods overlook the profound potential of the prior embedded within the weights of pre-trained models, which enables manipulating the latent feature space; hence, they primarily center on handling the motion space. In this work, we explore the attention mechanism of pre-trained motion diffusion models. We uncover the roles and interactions of attention elements in capturing and representing intricate human motion patterns, and carefully integrate these elements to transfer a leader motion to a follower one while maintaining the nuanced characteristics of the follower, resulting in zero-shot motion transfer. Editing features associated with selected motions allows us to confront a challenge observed in prior motion diffusion approaches, which use general directives (e.g., text, music) for editing, ultimately failing to convey subtle nuances effectively. Our work is inspired by how a monkey closely imitates what it sees while maintaining its unique motion patterns; hence we call it Monkey See, Monkey Do, and dub it MoMo. Employing our technique enables accomplishing tasks such as synthesizing out-of-distribution motions, style transfer, and spatial editing. Furthermore, diffusion inversion is seldom employed for motions; as a result, editing efforts focus on generated motions, limiting the editability of real ones. MoMo harnesses motion inversion, extending its application to both real and generated motions. Experimental results show the advantage of our approach over the current art. In particular, unlike methods tailored for specific applications through training, our approach is applied at inference time, requiring no training. Our webpage is at https://monkeyseedocg.github.io.