Search

Background and Objectives: Combining knowledge of clinical pathologists and deep learning models is a growing trend in morphological analysis of cells circulating in blood to add objectivity, accuracy, and speed in diagnosing hematological and non-hematological diseases. However, the variability in staining protocols across different laboratories can affect the color of images and performance of automatic recognition models. The objective of this work is to develop, train and evaluate a new system for the normalization of color staining of peripheral blood cell images, so that it transforms images from different centers to map the color staining of a reference center (RC) while preserving the structural morphological features. Methods: The system has two modules, GAN1 and GAN2. GAN1 uses the PIX2PIX technique to fade original color images to an adaptive gray, while GAN2 transforms them into RGB normalized images. Both GANs have a similar structure, where the generator is a U-NET convolutional neural network with ResNet and the discriminator is a classifier with ResNet34 structure. Digitally stained images were evaluated using GAN metrics and histograms to assess the ability to modify color without altering cell morphology. The system was also evaluated as a pre-processing tool before cells undergo a classification process. For this purpose, a CNN classifier was designed for three classes: abnormal lymphocytes, blasts and reactive lymphocytes. Results: Training of all GANs and the classifier was performed using RC images, while evaluations were conducted using images from four other centers. Classification tests were performed before and after applying the stain normalization system. The overall accuracy reached a similar value around 96% in both cases for the RC images, indicating the neutrality of the normalization model for the reference images. On the contrary, it was a significant improvement in the classification performance when applying the stain normalization to the other centers. Reactive lymphocytes were the most sensitive to stain normalization, with true positive rates (TPR) increasing from 46.3% - 66% for the original images to 81.2% - 97.2% after digital staining. Abnormal lymphocytes TPR ranged from 31.9% - 95.7% with original images to 83% - 100% with digitally stained images. Blast class showed TPR ranges of 90.3% - 94.4% and 94.4% - 100%, for original and stained images, respectively. Conclusions: The proposed GAN-based normalization staining approach improves the performance of classifiers with multicenter data sets by generating digitally stained images with a quality similar to the original images and adaptability to a reference staining standard. The system requires low computation cost and can help improve the performance of automatic recognition models in clinical settings. This work is part of a research project funded by the Ministry of Science and Innovation of Spain, with reference PID2019-104087RB-I00.

The purpose of this study was to investigate whether differential phosphorylation states of blood markers can identify patients with LRRK2 Parkinson's disease (PD). We assessed phospho(P)-Ser-935-LRRK2 and P-Ser-473-AKT levels in peripheral blood cells from patients with G2019S LRRK2-associated PD (L2PD, n = 31), G2019S LRRK2 non-manifesting carriers (L2NMC, n = 26), idiopathic PD (iPD, n = 25), and controls (n = 40, total n = 122). We found no differences at P-Ser-935-LRRK2 between groups but detected a specific increase of P-Ser-473-AKT levels in all G2019S carriers, either L2PD or L2NMC, absent in iPD. Although insensitive to LRRK2 inhibition, our study identifies P-Ser-473-AKT as an endogenous candidate biomarker for peripheral inflammation in G2019S carriers using accessible blood cells. ANN NEUROL 2022;92:888-894.

[eng] Clinical pathologists have learned to identify morphological qualitative features to characterise the different normal cells, as well as the abnormal cell types whose presence in peripheral blood is the evidence of serious haematological diseases. A drawback of visual morphological analysis is that is time consuming, requires well-trained personnel and is prone to intra-observer variability, which is particularly true when dealing with blast cells. Indeed, subtle interclass morphological differences exist for leukaemia types, which turns into low specificity scores in the routine screening. They are well-known the difficulties that clinical pathologists have in the discrimination among different blasts and the subjectivity associated with their morphological recognition. The general objective of this thesis is the automatic recognition of different types of blast cells circulating in peripheral blood in acute leukaemia using digital image processing and machine learning techniques. In order to accomplish this objective, this thesis starts with a discrimination among normal mononuclear cells, reactive lymphocytes and three types of leukemic cells using traditional machine learning techniques and hand-crafted features obtained from cell segmentation. In the second part of the thesis, a new predictive system designed with two serially connected convolutional neural networks is developed for the diagnosis of acute leukaemia. This system was proved to distinguish neoplastic (leukaemia) and non-neoplastic (infections) diseases, as well as recognise the leukaemia lineage. Furthermore, it was evaluated for its integration in a real-clinical setting. This thesis also contributes in advancing the state of the art of the automatic recognition of acute leukaemia by providing a more realistic approach which reflects the real-life complexity of acute leukaemia diagnosis.

Background: The aim of the present study was to evaluate the association between prosthetic joint infection (PJI) after joint arthroplasty and the length of red blood cell (RBC) storage, timing of RBC transfusion, and the number of RBC units transfused.Study design and Methods: All patients who underwent a primary or revision joint artrhoplasty between January 2000 and December 2012 were retrospectively reviewed. For this study, only patients who received RBC transfusions during the day of the surgery (early transfusion group) or within the first 4 days after surgery (late transfusion group) were included.Results: A total of 9906 patients were reviewed. In the early transfusion group (n=1153, 11.6%), patients receiving 1 or 2 RBC units (3.5% vs 6.3%, P=0.041), 3 or 4 RBC (1.3% vs 13.3%, P=0.004) or ≥5 RBC units (5.0% vs 37.5%, P=0.026) had a higher PJI rate only when >50% of RBC units transfused had been stored >14 days. In the late transfusion group (n=920, 9.3%) these differences were not significant. Early transfusion of RBCs stored >14 days was an independent variable associated with an increased risk of PJI (OR:2.50, 95%CI:1.44-4.33)Conclusion: Transfusion of RBC within the first 6h after joint arthroplasty was an independent variable associated with PJI risk when RBC units are stored >14 days. The rate of PJI increased with the number of old RBC units transfused within this critical period.

[eng] Clinical pathologists identify visually many morphological features to characterize the different normal cells, as well as the abnormal cell types whose presence in peripheral blood is the evidence of serious diseases. Disadvantages of visual morphological analysis are that it is time consuming, needs expertise to perform an objective review of the smears and is prone to inter-observer variability. Also, most of the morphological descriptions are given in qualitative terms and there is a lack of quantitative measures. The general objective of this thesis is the automatic recognition of normal and dysplastic cells circulating in blood in myelodysplastic syndromes using convolutional neural networks and digital image processing techniques. In order to accomplish this objective, this work starts with the design and development of a Mysql database to store information and images from patients and the development of a first classifier of four groups of cells, using convolutional neural networks as feature extractors. Then, a high- quality dataset of around 17,000 images of normal blood cells is compiled and used for the development of a recognition system of eight groups of blood cells. In this work, we compare two transfer learning approaches to find the best to classify the different cell types. In the second part of the thesis, a new convolutional neural network model for the diagnosis of myelodysplastic syndromes is developed. This model was validated by means of a proof of concept. It is considered among the first models that have been built for diagnosis support. The final work of the thesis is the integration of two convolutional networks in a modular system for the automatic classification of normal and abnormal cells. The methodology and models developed constitute a step forward to the implementation of a modular system to recognize automatically all cell types in a real setup in the laboratory., [spa] Los especialistas de laboratorio identifican visualmente muchas características morfológicas para identificar las diferentes células normales, así como los tipos de células anormales, cuya presencia en sangre periférica es evidencia de enfermedades graves. Algunas de las desventajas del análisis morfológico visual incluyen que toma mucho tiempo, necesita experiencia para realizar una revisión objetiva de los frotis y es propenso a la variabilidad entre observadores. Además, la mayoría de las descripciones morfológicas se proporcionan en términos cualitativos. Debido a lo expuesto anteriormente, es necesario establecer medidas cuantitativas. El objetivo general de esta tesis es el reconocimiento automático de células normales y células displásicas circulantes en sangre en síndromes mielodisplásicos mediante redes neuronales convolucionales y técnicas de procesamiento digital de imágenes. Para lograr este objetivo, este trabajo comenzó con el diseño y desarrollo de una base de datos Mysql para almacenar información e imágenes de pacientes y el desarrollo de un primer clasificador de cuatro grupos de células, utilizando redes neuronales convolucionales como extractores de características. Luego, se compila un conjunto de datos de alta calidad de alrededor de 17.000 imágenes de células sanguíneas normales y se utiliza para el desarrollo de un sistema de reconocimiento de ocho grupos de células sanguíneas. En este trabajo, comparamos dos enfoques de aprendizaje por transferencia para encontrar el mejor para clasificar los diferentes tipos de células. En la segunda parte de la tesis se desarrolla un nuevo modelo de red neuronal convolucional para el diagnóstico de síndromes mielodisplásicos. Este modelo fue validado mediante prueba de concepto. Se considera uno de los primeros modelos que se han construido para apoyar el diagnóstico. El trabajo final de la tesis es la integración de dos redes convolucionales en un sistema modular para la clasificación automática de células normales y anormales. La metodología y los modelos desarrollados constituyen un paso adelante hacia la implementación de un sistema modular para reconocer automáticamente todos los tipos de células en una configuración real en el laboratorio.

Des de l'any 1989 en que es va realitzar el primer trasplantament de cèl·lules hematopoiètiques procedents de la sang de cordó umbilical (SCU) a un pacient amb anèmia de Falconi, ha anat creixent l'interès per les característiques immunològiques de la sang de cordó umbilical. Un dels principals avantatges que presenta la SCU és la baixa incidència en la malaltia de l'injert contra l'hoste (GVHD) que es desencadena després del trasplantament entre pacients no emparentats. Fins el moment tots els treballs han demostrat que els limfòcits T de SCU presenten un fenotip CD45RA+ que indicaria en un principi la manca d'exposició prèvia a antígens i aquesta característica ha comportat que es considerés als limfòcits naive de SCU com immunològicament immadurs i amb una resposta peculiar enfront a diversos estímuls. L'ús de diferents condicions experimentals i sistemes d'activació han donat lloc a resultats no comparables i a vegades fins i tot contradictoris. Per una banda, l'ús de tècniques de separació que comportin la pèrdua sel·lectiva de poblacions cel·lulars o produeixin pertorbacions funcionals, i per l'altra la utilització de diversos agents per activar les cèl·lules que intervenen a diferents nivells de la cascada de senyalització, han donat lloc a resultats difícils d'interpretar. L'objectiu general d'aquest treball ha estat de caracteritzar la població majoritària CD4+CD45RA+ de SCU utilitzant un anticòs monoclonal anti-CD3 humà (OKT3), que permet una activació de tots els limfòcits independentment de la seva especificitat.Per abordar aquest objectiu, es va obtenir la població majoritària CD4+CD45RA+ de la sang de cordó umbilical i de sang perifèrica d'adult, utilitzant una selecció negativa, que consistia en l'obtenció dels limfòcits CD3+ a partir de les cèl·lules mononuclears obtingudes per gradient de Ficoll, i una segona selecció negativa a partir dels limfòcits CD3+ per poder obtenir la població CD4+CD45RA+ purificada. El grau de purificació es va determinar per citometria de flux.L'estudi fenotípic i funcional dels limfòcits CD4+CD45RA+ de sang de cordó umbilical va demostrar que aquesta població limfocitària presentava característiques que la diferenciava de la població CD4+CD45RA+ de la sang perifèrica d'adult. Així s'observà, que els limfòcits naive de SCU presentaven una major tendència a entrar en apoptosi espontània comparat amb la SPA, i que aquesta mort podia ser en part revertida per la presència d'interleuquina 2 (IL2). Una activació via TCR/CD3 amb una sola senyal anti-CD3 portà als limfòcits naive de SCU a activar-se, dividir-se i finalment a morir per apoptosi. Aquesta activació anava acompanyada d'una fosforilació diferencial en Tyr de proteïnes implicades en els primers processos de la cascada de senyalització via TCR, transformació blàstica, expressió de marcadors d'activació en superfície i divisió cel·lular. La coestimulació via CD28 feia que els dos tipus cel·lulars assolissin uns graus d'activació similars, excepte en l'coexpressió de CD45RO que només s'observava amb la SCU.No sabem si la resposta de les cèl·lules naive observada in vitro pot ser sobrepassada in vivo per altres factors i tampoc coneixem el paper de les cèl·lules de SCU en l'expansió perifèrica després del trasplantament. Després del trasplantament amb cèl·lules de SCU, la primera onada de reconstitució de cèl·lules T s'origina per les cèl·lules T donants madures presents en l'injert. L'estimulació de les cèl·lules de SCU a través del receptor per l'antigen sense coestimulació podria induir un estímul toleritzant mitjançant l'activació subòptima i la posterior eliminació d'una gran part de les cèl·lules activades. En aquest cas, l'apoptosi és possiblement sel·lectiva per cèl·lules reactives donants, fenomen que es contempla com la base de la tolerància en el trasplantament i que podria explicar l'aparent reducció del GVHD en els individus trasplantats amb SCU.

In previous studies, we have demonstrated the inhibition of CD4 expression in rat lymphocytes treated with phorbol myristate acetate (PMA) by antisense oligonucleotides (AS-ODNs) directed against the AUG start region of the cd4 gene. The aim of the present study was to inhibit CD4 expression in lymphocytes without promoting CD4 synthesis and to determine the effect of this inhibition on CD4 1 T cell function. Four 21-mer ODNs against the rat cd4 gene (AS-CD4-1 to AS-CD4-4) were used. Surface CD4 expression was measured by immunofluorescence staining and flow cytometry, and mRNA CD4 expression was measured by RT-PCR. T CD4 1 cell function was determined by specific and unspecific proliferative response of rat-primed lymphocytes. After 24 hours of incubation, AS-CD4-2 and AS-CD4-4 reduced lymphocyte surface CD4 expression by 40%. This effect remained for 72 hours and was not observed on other surface molecules, such as CD3, CD5, or CD8. CD4 mRNA expression was reduced up to 40% at 24 hours with AS-CD4-2 and AS-CD4-4. After 48 hours treatment, CD4 mRNA decreased up to 27% and 29% for AS-CD4-2 and AS-CD4-4, respectively. AS-CD4-2 and AS-CD4-4 inhibited T CD4 1 cell proliferative response upon antigen-specific and unspecific stimuli. Therefore, AS-ODNs against CD4 molecules inhibited surface and mRNA CD4 expression, under physiologic turnover and, consequently, modulate T CD4 1 cell reactivity.

DNA methylation could be reversible and mouldable by environmental factors, such as dietary exposures. The objective was to analyse whether an intervention with two Mediterranean diets, one rich in extra-virgin olive oil (MedDiet + EVOO) and the other one in nuts (MedDiet + nuts), was influencing the methylation status of peripheral white blood cells (PWBCs) genes. A subset of 36 representative individuals were selected within the PREvención con DIeta MEDiterránea (PREDIMED-Navarra) trial, with three intervention groups in high cardiovascular risk volunteers: MedDiet + EVOO, MedDiet + nuts, and a low-fat control group. Methylation was assessed at baseline and at five-year follow-up. Ingenuity pathway analysis showed routes with differentially methylated CpG sites (CpGs) related to intermediate metabolism, diabetes, inflammation, and signal transduction. Two CpGs were specifically selected: cg01081346-CPT1B/CHKB-CPT1B and cg17071192-GNAS/GNASAS, being associated with intermediate metabolism. Furthermore, cg01081346 was associated with PUFAs intake, showing a role for specific fatty acids on epigenetic modulation. Specific components of MedDiet, particularly nuts and EVOO, were able to induce methylation changes in several PWBCs genes. These changes may have potential benefits in health; especially those changes in genes related to intermediate metabolism, diabetes, inflammation and signal transduction, which may contribute to explain the role of MedDiet and fat quality on health outcomes. This work was supported by CIBERobn (CB12/03/30002 to J.A.M.) and Ministerio de Economía y Competitividad (AGL2013-45554-R to J.A.M. and F.I.M.) and Red PREDIMED-RETIC RD06/0045. A.A. was supported by a grant from Centre for Nutrition Research (Universidad de Navarra) until august 2016, and from that day, by a “Formación de Profesorado Universitario” fellowship from Ministerio de Educación, Cultura y Deporte (FPU15/02790).

Granulocyte colony-stimulating factor is the most commonly used cytokine for the mobilization of hematopoietic progenitor cells from healthy donors for allogeneic stem cell transplantation. Although the administration of this cytokine is considered safe, knowledge about its long-term effects, especially in hematopoietic progenitor cells, is limited. On this background, the aim of our study was to analyze whether or not granulocyte colony-stimulating factor induces changes in gene and microRNA expression profiles in hematopoietic progenitor cells from healthy donors, and to determine whether or not these changes persist in the long-term. For this purpose, we analyzed the whole genome expression profile and the expression of 384 microRNA in CD34(+) cells isolated from peripheral blood of six healthy donors, before mobilization and at 5, 30 and 365 days after mobilization with granulocyte colony-stimulating factor. Six microRNA were differentially expressed at all time points analyzed after mobilization treatment as compared to the expression in samples obtained before exposure to the drug. In addition, 2424 genes were also differentially expressed for at least 1 year after mobilization. Of interest, 109 of these genes are targets of the differentially expressed microRNA also identified in this study. These data strongly suggest that granulocyte colony-stimulating factor modifies gene and microRNA expression profiles in hematopoietic progenitor cells from healthy donors. Remarkably, some changes are present from early time-points and persist for at least 1 year after exposure to the drug. This effect on hematopoietic progenitor cells has not been previously reported.

Background: High red blood cell distribution width (RDW) is associated with worse outcome in diverse scenarios, including in critical illness. The Sabadell score (SS) predicts in-hospital survival after ICU discharge. We aimed to determine RDW’s association with survival after ICU discharge and whether RDW can improve the accuracy of the SS. Design: Retrospective cohort study. Setting: general ICU at a university hospital. Patients: We included all patients discharged to wards from January 2010 to October 2016. Methods: We analyzed associations between RDW and variables recorded on admission (age, comorbidities, severity score), during the ICU stay (treatments, complications, length of stay (LOS)), and at ICU discharge (SS). The primary outcome was hospital mortality. Statistical analysis included multivariable logistic regression and receiver operating characteristic curve (ROC) analyses. Results: We discharged 3366 patients to wards; median ward LOS was 7 [4–13] days; ward mortality was 5.2%. Mean RDW at ICU discharge was 15.4 ± 2.5%. Ward mortality was higher at each quartile of RDW (0.7%, 2.9%, 7.5%, 10.3%; area under ROC 0.81). A logistic regression model with Sabadell score obtained an excellent accuracy for ward mortality (area under ROC 0.863), and the addition of RDW slightly improved accuracy (AUROC 0.890, p