Design and optimization of a computer-assisted tool for erythrocyte phenotyping: in search of a new biomarker for autism spectrum disorders

The morphology of a red blood cell (RBC) in physiological conditions (discocyte) is of primary importance for its main physiological role (i.e., the transport of respiratory gases to and from the tissues). The erythrocyte membrane-cortical cytoskeleton-complex structure, ensuring both shape resilience and marked physiological deformability, also allows erythrocytes to undergo peculiar shape changes, from spherical globes (spherocytes), concave shapes (stomatocytes), cells provided with spines (echinocytes) or with a central crest (knizocytes). Several pathological conditions are associated with characteristic RBC shape alterations. For instance, typical thorny red cells (acanthocytes) are prevalent in neuroacanthocytosis, a group of rare genetic diseases; hereditary spherocytosis, elliptocytosis, and stomatocytosis are RBC disorders resulting from mutations in genes encoding various membrane and skeletal proteins; codocytes are a common occurrence in beta- thalassemia. Furthermore, aberrant erythrocytic shapes are observed in renal and liver disease, and in case of hemoglobinopathies and toxemias. Leptocytes, as well as other abnormal erythrocyte shapes, are reported in patients diagnosed with Rett syndrome, an X-linked genetically determined neurodevelopmental disorder. Abnormal RBC shapes have been described also in patients diagnosed with autism spectrum disorders (ASD). ASD represent a complex set of neurodevelopmental disorders, characterized by social and behavioral impairments. These disorders have a prevalent genetic etiology; however, epigenetically acting environmental factors (e.g., immune dysregulation, pollutants) seem to play a key role in the development of the disease. This condition is associated with a high social impact and strong suggestions of a dramatically rising prevalence in the general pediatric population over the last decades. In the present thesis work, we started to investigate the potential value of the RBC shape as a biomarker candidate for an early ASD diagnosis, illustrating and discussing the results achieved so far within this broad research project. One of these results is the optimization of a new protocol, based on the acupuncture method, for blood samples preparation aimed at SEM (Scanning Electron Microscopy) morphological analysis of RBC. Indeed, to reach the final research goal it is necessary to carry out many bloods drawn from young children in a practical way. The novel preparation and imaging method, both minimally invasive and cheap, consists in aspirating by a capillary tube (preloaded with anticoagulant solution) a drop of peripheral blood obtained from a prick in a human subject’s fingertip. Samples are subsequently processed by ad hoc protocols and imaged at SEM. Another goal successfully achieved through this work is the favorable opinion and approval from local ethics committee. Preliminary results from blood samples analysis at SEM confirm the presence of morphologically abnormal erythrocytes in ASD patients in a greater extent than in healthy volunteers. Nevertheless, given the small sample size searched to date, more accurate information and a high caseload of subjects are still needed. The last and most consistent part of this thesis work illustrates development and validation of a new computer-assisted tool for erythrocyte phenotyping in SEM micrographs. Indeed, achieving our final research goal involves the analysis of many blood samples, each of which needing a time-consuming manual search, count and classification of a large number of RBC. In this context a software capable to semi-automatically associate a distribution of erythrocytes among the relevant morphological classes to each examined patient would represent a powerful tool towards a more rapid and reproducible analysis of RBC morphology. Furthermore, observations from studying of intra-operator variability of the manual RBC morphological analysis by SEM carried out in this thesis work suggest the need of a high number of replicate counts, supporting the key relevance of implementing reliable computer-assisted algorithms and machine learning systems for our research project. Our own software, developed using Microsoft Visual Studio 2010 as a working environment and the C++ language as a programming language, has shown to be able to detect RBC in SEM micrographs and to intercept erythrocyte morphological categories, albeit with different degrees of efficiency. This new software represents a promising and powerful tool with many potential applications, from early ASD diagnostic to basic research in hematology. This thesis work presents the initial phases of an ambitious research project about erythrocyte morphological changes in pediatric ASD patients, laying solid foundations for its continuation. Indeed, further investigations are needed to establish the statistical significance of the differences about erythrocyte morphological alterations in blood micro-samples from ASD patients and neurotypical controls. Moreover, our work will continue validating the data - obtained by our software - about the distribution of erythrocytes among several morphometric classes in SEM micrographs from peripheral blood samples. These aspects are of seminal importance in the research aimed at discovering potential novel biomarkers for ASD diagnosis based on the SEM analysis of peripheral blood micro-samples.