Machine learning-based scoring models to predict hematopoietic stem cell mobilization in allogeneic donors

Predictive model of the efficiency of hematopoietic stem cell collection in patients with multiple myeloma and lymphoma based on multiple peripheral blood markers

INTRODUCTION

Autologous hematopoietic stem cell transplantation (ASCT) has gained extensive application in the treatment of lymphoma and multiple myeloma (MM). Plenty of studies demonstrate that peripheral blood indicators could be considered potential predictive biomarkers for hematopoietic stem cells (HSCs) collection efficiency, including white blood cell count (WBC), monocyte count (Mono), platelet count (PLT), hematocrit, and hemoglobin levels. Currently, clinically practical predictive models based on these peripheral detection indicators to quickly, conveniently, and accurately predict collection efficiency are lacking.

METHODS

In total, 139 patients with MM and lymphoma undergoing mobilization and collection of ASCT were retrospectively studied. The study endpoint was successful collection of autologous HSCs. We analyzed the effects of clinical characteristics and peripheral blood markers on collection success, and screened variables to establish a prediction model. We determined the optimal cutoff value of peripheral blood markers for predicting successful stem cell collection and the clinical value of a multi-marker prediction approach. We also established a prediction model for collection efficacy.

RESULTS

Univariate and multivariate logistic regression analyses showed that the mobilization regimen, Mono, PLT, mononuclear cell count (MNC), and peripheral blood CD34+ cell count (PB CD34+ counts) were significant predictors of successful collection of peripheral blood stem cells (PBSC). Two predictive models were constructed based on the results of multivariate logistic analyses. Model 1 included the mobilization regimen, Mono, PLT, and MNC, whereas Model 2 included the mobilization regimen, Mono, PLT, MNC, and PB CD34+ counts. Receiver operating characteristic (ROC) curve analysis showed that the PB CD34+ counts, Model 1, and Model 2 could predict successful HSCs collection, with cutoff values of 26.92 × 106/L, 0.548, and 0.355, respectively. Model 1 could predict successful HSCs collection with a sensitivity of 84.62%, specificity of 75.73%, and area under the curve (AUC) of 0.863. Model 2 could predict successful HSCs collection with a sensitivity of 83.52%, specificity of 94.17%, and AUC of 0.946; thus, it was superior to the PB CD34+ counts alone.

CONCLUSION

Our findings suggest that the combination of the mobilization regimen, Mono, PLT, MNC, and PB CD34+ counts before collection has predictive value for the efficacy of autologous HSCs collection in patients with MM and lymphoma. Using models based on these predictive markers may help to avoid over-collection and improve patient outcomes.

Scoping Review of Factors Associated with Stem Cell Mobilization and Collection in Allogeneic Stem Cell Donors

There is wide interindividual variation in the efficacy of CD34+ cell mobilization and collection in healthy allogenic hematopoietic stem cell donors. Donor characteristics, blood cell counts, and various factors related to mobilization and collection have been associated with blood CD34+ cell count and CD34+ cell yield after granulocyte colony-stimulating factor (G-CSF) mobilization and collection. Given the heterogenous nature of the literature reporting these associations, in this scoping review we clarify the determinants of CD34+ count and yield. Studies published between 2000 and 2023 reporting allogeneic donors undergoing G-CSF mobilization and peripheral blood stem cell (PBSC) collection were evaluated. Eligible studies were those that assessed blood CD34+ cell count or CD34+ cell yield in the first PBSC collection after mobilization with 4 or 5 days of G-CSF treatment. Associations were recorded between these outcomes and donor factors (age, sex, weight, ethnicity), mobilization factors (G-CSF scheduling or dose), collection factors (venous access, processed blood volume [PBV]) or laboratory factors (blood cell counts at baseline or after mobilization). The 52 studies evaluated between 15 and 20,884 donors. Forty-three studies were retrospective, 33 assessed blood CD34+ cell counts, and 39 assessed CD34+ cell yield from PBSCs. Blood CD34+ cell counts consistently predicted CD34+ cell yield. Younger donors usually had higher blood CD34+ cell counts and CD34+ cell yield. Most studies that investigated the effect of donor ancestry found that donors of non-European ancestry had higher blood CD34+ cell counts after mobilization and higher CD34+ cell yields from collection. The poor consensus about the best predictors of blood CD34+ cell count and yield necessitates further prospective studies, particularly of the role of donor ancestry. The current focus on donor sex as a major predictor requires re-evaluation.

Hematopoietic stem cell collection for sickle cell disease gene therapy

PURPOSE OF REVIEW

Gene therapy for sickle cell disease (SCD) is advancing rapidly, with two transformative products recently approved by the US Food and Drug Administration and numerous others under study. All current gene therapy protocols require ex vivo modification of autologous hematopoietic stem cells (HSCs). However, several SCD-related problems impair HSC collection, including a stressed and damaged bone marrow, potential cytotoxicity by the major therapeutic drug hydroxyurea, and inability to use granulocyte colony stimulating factor, which can precipitate severe vaso-occlusive events.

RECENT FINDINGS

Peripheral blood mobilization of HSCs using the CXCR4 antagonist plerixafor followed by apheresis collection was recently shown to be safe and effective for most SCD patients and is the current strategy for mobilizing HSCs. However, exceptionally large numbers of HSCs are required to manufacture an adequate cellular product, responses to plerixafor are variable, and most patients require multiple mobilization cycles, increasing the risk for adverse events. For some, gene therapy is prohibited by the failure to obtain adequate numbers of HSCs.

SUMMARY

Here we review the current knowledge on HSC collection from individuals with SCD and potential improvements that may enhance the safety, efficacy, and availability of gene therapy for this disorder.

Overview on novel strategies and current guidelines for hematopoietic stem cell mobilisation and collection

The success of the autologous stem cell transplantation is strictly related to an adequate hematopoietic stem cell mobilization and collection. The minimum threshold for a successful mobilization is currently defined as 2 × 106/kg CD34+ cells. However, the optimal stem cell mobilization strategy is still controversial. The availability of plerixafor, a selective and reversible CXCR4 inhibitor, has been associated with an higher use of chemo-free protocols by many centres. In the near future, it is conceivable that artificial intelligence may became more accurate and comprehensive, possibly guiding clinicians in choosing the optimal mobilisation treatment for the various patients undergoing hematopoietic stem cell transplantation. Machine learning-based scoring models may be the basis for the development of "intelligent" mobilisation algorithms.

Updates on the mobilization pipeline for hematopoietic stem cell collection

Hematopoetic Stem Cell Transplantation is a life saving procedure which requires mobilization of stem cells for apheresis procedure. In this review we aimed to examine mobilizing agents that are in use and under investigation. Apheresis practitioners who oversee stem cell collections should be familiar with the recent advances in mobilization agents to utilize most up-to-date information for better patient outcomes.

Innovations in hematopoietic stem-cell mobilization: a review of the novel CXCR4 inhibitor motixafortide

Hematopoietic stem-cell transplantation (HCT) and stem-cell-based gene therapies rely on the ability to collect sufficient CD34+ hematopoietic stem and progenitor cells (HSPCs), typically via peripheral blood mobilization. Commonly used HSPC mobilization regimens include single-agent granulocyte colony-stimulating factor (G-CSF), plerixafor, chemotherapy, or a combination of these agents. These regimens, however, frequently require multiple days of injections and leukapheresis procedures to collect adequate HSPCs for HCT (minimum = >2 × 106 CD34+ cells/kg; optimal = 5-6 × 106 CD34+ cells/kg). In addition, these regimens frequently yield suboptimal CD34+ HSPC numbers for HSPC-based gene-edited therapies, given the significantly higher HSPC number needed for successful gene-editing and manufacturing. Meanwhile, G-CSF is associated with common adverse events such as bone pain as well as an increased risk of rare but potentially life-threatening splenic rupture. Moreover, G-CSF is unsafe in patients with sickle-cell disease, a key patient population that may benefit from autologous HSPC-based gene-edited therapies, where it has been associated with unacceptable rates of serious vaso-occlusive and thrombotic events. Motixafortide is a novel CXCR4 inhibitor with extended in vivo activity (>48 h) that has been shown in preclinical and clinical trials to rapidly mobilize robust numbers of HSPCs in preparation for HCT, while preferentially mobilizing increased numbers of more primitive HSPCs by immunophenotyping and single-cell RNA expression profiling. In this review, we present a history of stem-cell mobilization and update of recent innovations in novel mobilization strategies with a specific focus on the development of motixafortide, a long-acting CXCR4 inhibitor, as a novel HSPC mobilizing agent.

Peripheral Blood Allogeneic Stem Cell Mobilization: Can We Predict a Suboptimal Mobilization?

Allogeneic peripheral blood stem cells mobilization is now the basis of most stem cell transplants. In a very limited number of cases, mobilization is suboptimal leading to further collection procedures, to suboptimal cell doses infusion with delayed engraftment time, increased risks of transplant procedure and of related costs. To date we have no recognized and shared criteria for early estimating the probability of poor mobilization in healthy donors. We then analyzed allogeneic peripheral blood stem cell donations performed at the Fondazione Policlinico Universitario A.Gemelli IRCCS Hospital from January 2013 to December 2021 in order to identify premobilization factors associated with successful mobilization. The following data were collected: age, gender, weight, complete blood cell count at baseline, G-CSF dose, number of collection procedures, CD34+ cell count in peripheral blood on the first day of collection, CD34+ cell dose per kg body weight of recipient. Mobilization efficacy was defined according to the number of CD34+ cells in peripheral blood on day +5 of G-CSF administration. We classified donors as sub-optimal mobilizers or good mobilizers according to the achievement of the 50 CD34+ cell/μL threshold. We observed 30 suboptimal mobilizations in 158 allogeneic peripheral blood stem cell donations. Age and baseline white blood cell count were factors significantly associated with negative or positive impact on mobilization, respectively. We did not find significant differences in mobilization based on gender or G-CSF dose. Using cut-off values of 43 years and 5.5×10⁹/L WBC count, we built a suboptimal mobilization score: donors who reach 2, 1 or 0 points have a 46%, 16% or 4% probability of suboptimal mobilization, respectively. Our model explains 26% of the variability of mobilization confirming that most of the mobilization magnitude depends on genetically determined factors; however, suboptimal mobilization score is a simple tool providing an early assessment of mobilization efficacy before G-CSF administration begins in order to support allogeneic stem cells selection, mobilization and collection. Through a systematic review, we looked for confirmation of our findings. According to the published articles, all the variables we included in our model are confirmed to be strongly related to the success of mobilization. We believe that score system approach could be applied in clinical practice to assess the risk of mobilization failure at baseline allowing for a priori intervention.

Future stem cell analysis: progress and challenges towards state-of-the art approaches in automated cells analysis

Background and Aims

A microscopic image has been used in cell analysis for cell type identification and classification, cell counting and cell size measurement. Most previous research works are tedious, including detailed understanding and time-consuming. The scientists and researchers are seeking modern and automatic cell analysis approaches in line with the current in-demand technology.

Objectives

This article provides a brief overview of a general cell and specific stem cell analysis approaches from the history of cell discovery up to the state-of-the-art approaches.

Methodology

A content description of the literature study has been surveyed from specific manuscript databases using three review methods: manuscript identification, screening, and inclusion. This review methodology is based on Prism guidelines in searching for originality and novelty in studies concerning cell analysis.

Results

By analysing generic cell and specific stem cell analysis approaches, current technology offers tremendous potential in assisting medical experts in performing cell analysis using a method that is less laborious, cost-effective, and reduces error rates.

Conclusion

This review uncovers potential research gaps concerning generic cell and specific stem cell analysis. Thus, it could be a reference for developing automated cells analysis approaches using current technology such as artificial intelligence and deep learning.

Accurate Machine Learning Model to Diagnose Chronic Autoimmune Diseases Utilizing Information From B Cells and Monocytes

Heterogeneity and limited comprehension of chronic autoimmune disease pathophysiology cause accurate diagnosis a challenging process. With the increasing resources of single-cell sequencing data, a reasonable way could be found to address this issue. In our study, with the use of large-scale public single-cell RNA sequencing (scRNA-seq) data, analysis of dataset integration (3.1 × 10⁵ PBMCs from fifteen SLE patients and eight healthy donors) and cellular cross talking (3.8 × 10⁵ PBMCs from twenty-eight SLE patients and eight healthy donors) were performed to identify the most crucial information characterizing SLE. Our findings revealed that the interactions among the PBMC subpopulations of SLE patients may be weakened under the inflammatory microenvironment, which could result in abnormal emergences or variations in signaling patterns within PBMCs. In particular, the alterations of B cells and monocytes may be the most significant findings. Utilizing this powerful information, an efficient mathematical model of unbiased random forest machine learning was established to distinguish SLE patients from healthy donors via not only scRNA-seq data but also bulk RNA-seq data. Surprisingly, our mathematical model could also accurately identify patients with rheumatoid arthritis and multiple sclerosis, not just SLE, via bulk RNA-seq data (derived from 688 samples). Since the variations in PBMCs should predate the clinical manifestations of these diseases, our machine learning model may be feasible to develop into an efficient tool for accurate diagnosis of chronic autoimmune diseases.