In leukemia patients, minimal residual disease (MRD) is the level of leukemia cells that remain at a certain point in treatment. MRD monitoring has become routine clinical practice in the front-line treatment of virtually all childhood acute lymphoblastic leukemias (ALL) and in many adult patients with ALL. MRD is the most important prognostic factor for acute lymphoblastic leukemia and its persistence after induction/consolidation therapy is the main cause of relapse.
MRD can be estimated in several ways. The most effective and widely used methods include multiparameter flow cytometry (MFC), quantitative polymerase chain reaction (qPCR), and high-yeld next-generation sequencing (NGS). Each method has its advantages and disadvantages, but, in general, newer techniques tend to be superior. In general, in clinical trials in the USA and in some Asian countries there is a preference for MFC, while in Europe qPCR is frequently seen. The NGS technique is still new and, despite being already used in several important centers, it still does not enjoy the full recognition that the other two do. The main reason for this is probably the lack of relevant studies, with a large number of cases, which should come naturally with time.
The Boldrini Children's Center is pioneering the introduction of the NGS method in Brazil, an initiative that began around 2018 and continues to this day. We have participated in the introduction of the NGS trial for MRD at Boldrini practically since its inception (see papers below).
At the Center, reports are currently prepared dually, containing both qPCR and NGS results. The idea is to eventually abandon qPCR in favor of NGS. In addition to being less laborious, the NGS method allows the monitoring of virtually all cells of interest, unlike qPCR, which can only monitor cells previously seen in the patient, generally upon diagnosis.
In 2008, computer engineer Ariosto Siqueira Silva completed his doctorate at Boldrini Children's Center, with a thesis on computational methods for simulating biological processes, with applications to three-dimensional simulation of tumor development metabolism. This work continued in his new job, at the H. Lee Moffit Cancer Center and Research Institute, USA, and gave rise to a new software, EMMA (Ex vivo Mathematical Myeloma Advisor), capable of quickly predicting the clinical response of multiple myeloma patients to 31 medications, using fresh bone marrow aspirates.
His doctoral advisor at Boldrini, Dr José Andrés Yunes, has been in contact with Dr Silva, and is currently coordinating a project financed by the Brazilian Ministry of Health, within the scope of the National Support Program for Oncologic Attendance (Pronon), to use the same type of technology here in Brazil, with some differences. Instead of multiple myeloma, acute lymphocytic leukemia. Instead of adult cancer, pediatric cancer. Instead of fresh aspirates, patient-derived xenografts. The number of medications is larger: instead of 31 drugs, 62 substances, the majority of which are already used for cancer treatments in adults, but without significant studies on their use in children and adolescents. A master's thesis has already been defended within the scope of this project. Interaction with Dr Silva and his group at the Moffit Center continues. And, alongside analyzes to predict patient response, a battery of genomic assays, including gene expression (RNASeq), gene panels, methylome, etc., will be performed to cross-reference response data with genomic data, trying to identify relevant genes and potential drug targets.
I have been involved with this project, while still at Scylla, helping the team to transpose the EMMA software to Boldrini installations. Now, as a full time researcher at Unicamp, I intend to help in the analysis of the data generated, searching for correlations between the various results, in order to draw significant scientific conclusions.
Gene fusion occurs when, due to genomic rearrangements, such as translocations, deletions and inversions of large stretches, two independent genes are placed together in the genome, forming a single unit, often with the loss of pieces of the original genes. Fused genes have been found to be prevalent in all major types of human cancer. Gene fusion detection therefore plays a prominent role in the diagnosis, prognosis and treatment of cancer.
Traditional methods for detecting gene fusions involve mapping short reads onto a reference genome or transcriptome. However, these methods generally generate a high number of false positives and require extensive filtering steps. The addition of a directed assembly subroutine can alleviate this problem, but there is no consolidated strategy for selecting the reads to be used in this assembly. We therefore intend to study strategies for selecting readings for this type of montage, evaluating their effectiveness in this context. For example, techniques that analyze structural variants could be useful with the data we are interested in, originating from human cancer cells.
Alignment-free methods are an alternative, faster approach, which makes them attractive for a clinical setting. However, their application for gene fusion detection remains largely unexplored. Along these lines, we intend to investigate the applicability of alignment-free methods for detecting gene fusion. To this end, we intend to evaluate the alignment-free tools with the best performance in practice and identify which of them would be effective in the role of gene fusion detection.
This line of research is mainly conducted by my doctoral student Lucas Peres Oliveira, who has a FAPESP scholarship from 2023 to 2027 to carry out this work at the Institute of Computing.
Joao Meidanis
Last modified: Tue Sep 17 16:47:06 -03 2024
by JM