2 January 2024 in Nature Biotechnology
Adoption of high-content omic technologies in clinical studies, coupled with computational methods, has yielded an abundance of candidate biomarkers. However, translating such findings into bona fide clinical biomarkers remains challenging. To facilitate this process, we introduce Stabl, a general machine learning method that identifies a sparse, reliable set of biomarkers by integrating noise injection and a data-driven signal-to-noise threshold into multivariable predictive modeling. Evaluation of Stabl on synthetic datasets and five independent clinical studies demonstrates improved biomarker sparsity and reliability compared to commonly used sparsity-promoting regularization methods while maintaining predictive performance; it distills datasets containing 1,400–35,000 features down to 4–34 candidate biomarkers. Stabl extends to multi-omic integration tasks, enabling biological interpretation of complex predictive models, as it hones in on a shortlist of proteomic, metabolomic and cytometric events predicting labor onset, microbial biomarkers of pre-term birth and a pre-operative immune signature of post-surgical infections. Stabl is available at https://github.com/gregbellan/Stabl.
28 February 2023 in Research Square
High-content omic technologies coupled with sparsity-promoting regularization methods (SRM) have transformed the biomarker discovery process. However, the translation of computational results into a clinical use-case scenario remains challenging. A rate-limiting step is the rigorous selection of reliable biomarker candidates among a host of biological features included in multivariate models. We propose Stabl, a machine learning framework that unifies the biomarker discovery process with multivariate predictive modeling of clinical outcomes by selecting a sparse and reliable set of biomarkers. Evaluation of Stabl on synthetic datasets and four independent clinical studies demonstrates improved biomarker sparsity and reliability compared to commonly used SRMs at similar predictive performance. Stabl readily extends to double- and triple-omics integration tasks and identifies a sparser and more reliable set of biomarkers than those selected by state-of-the-art early- and late-fusion SRMs, thereby facilitating the biological interpretation and clinical translation of complex multi-omic predictive models. The complete package for Stabl is available online at https://github.com/gregbellan/Stabl.
8 July 2025 in Cancer Immunology Research
Vγ9Vδ2 (TCRVγ9+ TCRVδ2+) T cells are promising immunotherapeutic targets with effective antitumor properties in both in vitro and preclinical models of triple-negative breast cancer (TNBC). However, no information about their potential role in the context of human TNBC progression and response to immunotherapy has been reported. One key reason for this is the scarcity of Vγ9Vδ2 T-cell infiltrates relative to their Vδ1 (TCRVδ1+) and αβCD8 (TCRαβ+ CD8αβ+) T-cell counterparts. We provide comprehensive single-cell profiling of Vγ9Vδ2 T cells from patients with TNBC, prior to and following PD-(L)1 blockade therapy. We report that baseline Vγ9Vδ2 T-cell infiltrate expressing a unique cytotoxic type I phenotype could be associated with improved survival in patients with TNBC. Vγ9Vδ2 T cells harboring characteristics of enhanced antitumor activity (KLRC1+) were further associated with improved response to PD-(L)1 blockade therapy in patients with TNBC. Vγ9Vδ2 T cells had low expression levels of T-cell exhaustion (PD-1LowTOXLow) and T-cell receptor signaling hallmarks compared with Vδ1 and αβCD8 T cells, along with skewed differentiation profiles toward early effector memory phenotypes, both before and after anti–PD-1 therapy in TNBC tumors. Consistently, we observed limited activity of anti–PD-1 on tumor-infiltrating Vγ9Vδ2 T cells. In vitro, the use of anti–butyrophilin-3A antibodies in addition to anti–PD-1 reinvigorated the cytotoxic type I functions of peripheral Vγ9Vδ2 T cells from patients with breast cancer. Together, these data provide a rationale for Vγ9Vδ2 T cell–based combination therapy in patients with TNBC.
11 February 2026 in Communications Medicine
Background Postoperative pancreatic fistula (POPF) is the major driver of postoperative morbidity after pancreatoduodenectomy (PD) and a healthcare issue. In patients with pancreatic tumors the occurrence of POPF could lead to a complete failure of the oncologic strategy by delaying or annihilating the delivery of the indicated adjuvant chemotherapy. However, current preoperative prediction models lack precision. This study aimed to determine the ability of a high dimensional analysis of the patient’s peripheral immune system before PD to predict POPF. Methods Twenty-two patients in the prospective IMMUNOPANC trial (NCT03978702) underwent PD. Blood samples collected preoperatively were analyzed by combining single-cell mass cytometry and a sparse machine-learning pipeline, Stabl, to identify the most relevant POPF-predictive features. The logistic regression model output was evaluated using a five-fold cross-validation procedure. Results Eight (36%) patients experience POPF (grade B, n = 7; grade C, n = 1). The multivariable predictive model includes 11 features—six natural killer, three CD8+ T, and two CD4+ T lymphocyte cell clusters—revealing a preoperative POPF lymphocyte signature (Pancreatic Fistula Lymphocyte Signature, PFLS). The Stabl algorithm identifies a predictive model classifying POPF patients with high performance (area under the receiver operating characteristic curve=0.81, P = 2.04e-02). Conclusions In summary, preoperative circulating immune-cell composition can predict POPF in patients undergoing PD. The clinical application of the PFLS may enable the early identification of patients at high risk before pancreatic surgery, giving clinicians the opportunity to anticipate and mitigate POPF risk through tailored strategies in pre-, intra-, and post-operative settings.