Read Laura's Interview & Publication

Abnormal characteristic “round bottom flask” shape volume-based scattergram as a trigger to suspect persistent polyclonal B-cell lymphocytosis

Clinica Chimica Acta – https://doi.org/10.1016/j.cca.2020.10.015

What inspired your research and what did it cover?

Persistent polyclonal B-cell lymphocytosis (PPBL) is considered a rare entity. Some studies however have insinuated that its incidence might be higher. Under-diagnosis may result from the lack of diagnostic features, human skill, or a combination of both. Such diagnostic difficulties inspired the present study. The aim was to fully exploit the haematological data provided by Beckman Coulter DxH 800 analysers in combination with a novel machine learning (ML) approach to obtain

• new clues for PPBL detection and subsequent diagnosis, and
• an accurate distinction between this category and morphologically near indistinguishable splenic marginal zone lymphoma (SMZL).

This case involves a diagnostic challenge: benign and neoplastic categories present overlapping morphology. The research conducted within the framework of my PhD project addresses the assessment of lymphocytosis to classify the encounter as benign or neoplastic.

Which aspect(s) of your research work are you particularly excited about?

We described for the first time the specific pattern for PPBL cases as a “round bottom flask” shape scattergram. The consideration of such a pattern together with the morphological description, an especific pipeline for the laboratory workflow, and the proposed machine learning (ML) model, has allowed us to describe the largest Spanish PPBL series and one of the largest in the world.

Looking at the potential of your findings, what difference can they make?

Our particular environment is a patient-centred facility aimed at enhancing preventive medicine and delivering value-based clinical laboratory results. Facilitating early diagnosis and reinforcing the detection of benign conditions reduces unnecessary patient anxiety, as in the case of PPBL. Further, our results highlight that the proposed ML model can allow to increase PPBL detection and redefine its incidence in the general population.

AI-based models are emerging technologies likely to have a major impact on healthcare and the future of practice-integrated laboratory medicine. The full exploit of available laboratory data through machine learning would allow better interpretation, workflow and patient management. Additional phlebotomy is time-consuming, and uncomfortable for the patient. Accuracy and speed are key features to optimise the use of available specimen, maintain better turnaround times and high throughput in this highly demanding person-centred field.