Open AccessArticle

Detecting Methotrexate in Pediatric Patients Using Artificial Neural Networks

Alejandro Medina Santiago

^1,2,*,†

Jorge Iván Bermúdez Rodríguez

^2,3,*,†

Jorge Antonio Orozco Torres

^2,3,†

Julio Alberto Guzmán Rabasa

⁴

José Manuel Villegas Izaguirre

⁵

and

Gladys Falconi Alejandro

⁶

Instituto Nacional de Astrofísica, Óptica y Electrónica, Coordinación de Ciencias Computacionales—Conahcyt, Santa María Tonanzintla, San Andres Cholula, Puebla 72840, Mexico

Universidad de Ciencia y Tecnología Descartes, Avenida Ciprés 480, Tuxtla Gutiérrez 29065, Mexico

Tecnológico Nacional de México, Instituto Tecnológico de Tuxtla Gutiérrez, Carretera Panamericana Km. 1080, Tuxtla Gutiérrez 29050, Mexico

⁴

Tecnológico Nacional de México, IT Hermosillo, Av. Tecnológico y Periférico Poniente S/N, Hermosillo 83170, Mexico

⁵

Facultad de Ciencias de la Ingeniería y Tecnología, Universidad Autónoma de Baja California, Boulevard Universitario #1000, Unidad Valle de las Palmas, Tijuana 21500, Mexico

⁶

Secretaria Académica, Universidad Politécnica del Golfo de México, Carretera Federal Malpaso—El Bellote Km. 171 / Monte Adentro, Paraíso, Tabasco 86600, Mexico

Authors to whom correspondence should be addressed.

^†

These authors contributed equally to this work.

Appl. Sci. 2025, 15(1), 306; https://doi.org/10.3390/app15010306

Submission received: 31 May 2024 / Revised: 23 November 2024 / Accepted: 12 December 2024 / Published: 31 December 2024

(This article belongs to the Special Issue Artificial Intelligence in Medical Diagnostics: Second Edition)

Download

Browse Figures

Figure 1
Proposed methodology based on Kendall’s Life Cycle Management [<a href="#B22-applsci-15-00306" class="html-bibr">22</a>]. "> Figure 2
Ishikawa diagram for methotrexate toxicity manifestation in ALL patients. "> Figure 3
Normalized clinical data objectives. "> Figure 4
Patient data. "> Figure 5
Backpropagation neural network design. "> Figure 6
Intelligent diagnosis in pediatric patients with acute lymphoblastic leukemia to predict intoxication by methotrexate. "> Figure 7
Mean square error. "> Figure 8
Gradient learning curves and neural network validation. "> Figure 9
Linear regression graphic. "> Figure 10
The graph depicts the results generated by the neural network in the processing of the pediatric patient data sheet. "> Figure 11
AUC curve. ">

Review Reports Versions Notes

Abstract

Methotrexate is an antimetabolic agent with proliferative and immunosuppressive activity. It has been demonstrated to be an effective treatment for acute lymphoblastic leukemia (ALL) in children. However, there is evidence of an association between methotrexate and toxicity risks, which influences the personalization of treatment, particularly in the case of childhood ALL. This article presents the development and implementation of an algorithm based on artificial neural networks to detect methotrexate toxicity in pediatric patients with acute lymphoblastic leukemia. The algorithm utilizes historical clinical and laboratory data, with an effectiveness of 99% in the tests performed with the patient dataset. The use of neural networks in medicine is often linked to disease diagnosis systems. However, neural networks are not only capable of recognizing examples but also hold very important information. For this reason, one of the main areas of application of neural networks is the interpretation of medical data. In this article, we diagnose, with the application of neural networks in medicine, a concrete example: detecting methotrexate in its early stages in pediatric patients.

Keywords:

methotrexate; artificial neural networks; acute lymphoblastic leukemia (ALL); detection system

1. Introduction

Acute lymphoblastic leukemia (ALL) is the most frequent cancer in children under 15 years old, with a high incidence between 2 and 5 years of age. The treatment for this disease has made important progress, which is why the concept of mortality has changed, and it is now considered curable. Methotrexate is an important component in the treatment protocols of these patients. It has been proposed that the function of enzymes and transporters involved in the pathway of MTX may be altered by genetic polymorphisms that substantially influence the kinetics and response to therapy with high doses of MTX.

High-dose methotrexate (HD-MTX) plays a significant part in the therapy of acute lymphoblastic leukemia (ALL) in many treatment procedures around the world. However, there is a large variability in the pharmacokinetics and toxicity of the drug [1]. Polymorphisms in the MTX pathway genes influence the kinetics and response to high-dose MTX therapy in childhood ALL.

Acute leukemias are the most common group of cancer in children, where acute lymphoblastic leukemia (ALL) represents 80% of the cases. Although the etiology is uncertain, there are predisposing factors, such as genetic, viral, and environmental factors. The clinical manifestations, on occasion, are produced by the existence of malignant cells in the bone marrow, such as in anemia, thrombocytopenia, and leukopenia [1].

The diagnosis is made by morphological, cytogenetic, and molecular analysis of the bone marrow aspirate. The treatment takes an average of two years. The expectation of cure for children with ALL has improved exponentially thanks to new drugs and in recent years to a treatment adapted to the risk of patients. Currently, the percentage of cured patients with ALL is 75% [2].

Many studies have been carried out on methotrexate and its application in the treatment of ALL: the researchers in [3] investigated the influence of the polymorphisms C677T and A1298C on the gene for methylenetetrahydrofolate reductase (MTHFR) on the toxicity induced by MTX during the treatment of children with ALL.

Performing a regression analysis in patients with ALL, it was found that the ALL subtype was the strongest predictor of MTXPG accumulation, which shows that the genetic variation acquired in the ALL cells themselves had a stronger influence on the accumulation of MTXPG [4]. Children with ALL with multiple common polymorphisms in SLCO1B1 who received high doses of MTX in tests were associated with MTX clearance [5,6,7].

The authors of [8,9,10] found that variants of the SLC19A1, SLC22A8, MTR, and MTHFD1 genes were shown to be associated with the development of acute toxicity after HD-MTX treatment.

The results in [11,12,13] show evidence for the contribution of pharmacogenetics to the toxicity of high-dose MTX and plasma MTX concentrations 48 hours after treatment in patients with ALL or non-Hodgkin lymphoma. The influence of MTHFR C677T on MTX-related toxicities and plasma levels was confirmed.

The authors of [14,15] found that the TT genotype of the MTHFR gene is associated with an increase in MTX toxicity and the relapse rate. They consider the dose adjustment of MTX in the treatment protocols, depending on the genotype of the patient.

The study [16] identified that the 677TT genotype of the MTHFR gene is associated with decreased clearance of MTX and that genetic polymorphisms in the folate pathway and SLC19A1 were associated with MTX toxicity.

Five variants in four genes (MTHFR, ABCB1, ABCC2, and TYMS) were shown to be associated with toxicity, and they also found a statistically significant association between MTHFR rs1801133 and anemia in the consolidation phase [17].

By investigating key genes involved in the MTX pathway, the authors of [18] detected genetic polymorphisms associated with MTX pharmacokinetics, toxicity, and outcomes. The strongest association was observed between the SLCO1B1 rs4149056 SNP and MTX pharmacokinetic parameters such as clearance C_24h and AUC_0–48h.

The aim of the study [19] was to analyze the influence of genetic polymorphisms in RFC1 (reduced folate carrier), MS (methionine synthase), and MTRR (methionine synthase reductase) on the occurrence of adverse effects from MTX therapy, such as hematological disorders, hepatotoxicity, and nephrotoxicity. Polymorphisms in genes coding drug-metabolizing enzymes may cause individual differences in the effectiveness and toxicity of many medications, including cytostatics.

In ALL [20], too many stem cells become a type of white blood cell called lymphocytes. These lymphocytes are also called lymphoblasts or leukemic cells. There are three types of lymphocytes: (1) B lymphocytes produce antibodies to help fight infections, (2) T lymphocytes (3) natural aggressor cells attack cancer cells or viruses. The factors that increase the risk of contracting a disease are called risk factors. Having a risk factor does not mean that you are going to get cancer, and not having a risk factor does not mean that you are not going to get cancer. Possible risk factors for ALL include the following aspects: having a brother with leukemia, having been exposed to X-rays before birth, being exposed to radiation, having had previous treatment with chemotherapy or other medications that weaken the immune system, and having certain genetic disorders such as Down syndrome; possible signs of childhood ALL include fever and bruises [21].

MTX is an antimetabolic that has anti-proliferative and immunosuppressive activity by competitively inhibiting the enzyme di-hydrofolate reductase (DFR), a key enzyme in the metabolism of folic acid that regulates the amount of intracellular folate available for the synthesis of proteins and nucleic acids. It prevents the formation of tetrahydrofolate necessary for the synthesis of nucleic acids. It catalyzes the reduction of 5.10 methylene tetrahydrofolate to 5 methyl tetrahydrofolate, a form in which endogenous folate circulates, which is the donor of methyl groups necessary for the conversion of homocysteine to methionine during protein synthesis. This mainly affects cells that are in the phase of the cell cycle [1].

Mucositis is caused by the rapid proliferation of epithelial cells, causing the oropharynx and gastrointestinal tract to be particularly vulnerable to treatment with MTX. The combination of intensive chemotherapy and radiotherapy increases the damage of this drug to the level of the mucous membranes. Many factors are involved in the pathophysiology of mucositis, beginning with damage to connective tissue and blood vessels in the submucosa. This is followed by the release of proinflammatory cytokines and reactive oxygen species that exacerbate vascular damage and turn the process into a vicious circle that leads to tissue destruction. Bacterial colonization of ulcers increases tissue damage by the inflammation that is generated to control the infection.

MTX poisoning can be avoided in the following way: it should not be administered in patients with renal or hepatic deficiency or both and also in those with a pleural effusion or ascites to avoid plasma concentrations of MTX, which are a predictive value for such toxicity. In addition, constant monitoring to reverse the undesirable effects of the medication should be undertaken.

The article is organized as follows: Section 2 describes the methodology used for the development of the project and is in turn divided into subsections that describe the identification of the problem, the determination of the necessary requirements, the needs analysis, the design of the system, as well as development and documentation. Section 3 describes the results obtained from the design and training of the proposed neural network and is also divided into subsections that present the graphical interface for data recording, the methotrexate validation neural network, as well as the performance results. In Section 4, the discussion related to the research project is addressed. Finally, Section 5 presents the conclusions of the work.

2. Methodology

This section presents the methodology used for the project development, and the analysis is based on Kendall System Analysis and Design, which proposes seven phases, which, for our analysis, are presented in Figure 1 [22]. The Kendall and Kendall methodology for information system development is structured into seven fundamental phases that guide the creation of effective and user-adapted systems. The first phase, the Feasibility Study, aims to assess the project’s technical, economic, and operational viability. During this phase, the need for an intelligent diagnostic system for acute lymphoblastic leukemia (ALL) in pediatric patients is evaluated, considering available resources and the project’s feasibility within the context of the Tuxtla Gutiérrez Pediatric Hospital. The second phase, Requirement Analysis, focuses on defining and documenting the system requirements. An exhaustive collection of personal and clinical patient data are examined, alongside the functionalities required for the mobile application, ensuring that these requirements are clear and detailed for subsequent implementation.

In the third phase, System Design, a detailed design of the system is developed, including architecture, interfaces, and necessary processes. The design of the mobile application for data collection and the neural network for data analysis are developed, specifying the characteristics and integration of each component.

The fourth phase, Development, involves constructing the system according to the established design. This phase includes programming the mobile application for data capture and storage and implementing the neural network in Matlab 2023a, integrating both components to ensure accurate data analysis.

During the fifth phase, Testing, the system’s functionality is verified to ensure it meets the specified requirements. Extensive testing of the mobile application is performed to ensure its functionality, and the performance of the neural network is evaluated with test data, adjusting the system as needed to enhance accuracy and effectiveness.

The sixth phase, Implementation, focuses on deploying the system in the real environment. The application is installed on hospital devices, ensuring proper integration and functionality of the intelligent diagnostic system and providing training to end-users to ensure efficient use.

Finally, the seventh phase, Maintenance and Evaluation, involves continuous monitoring of the system to correct errors, perform updates, and ensure its proper functioning over time. User feedback is collected, and periodic evaluations of the system are conducted to ensure it continues to meet established objectives and adapt to any new requirements.

The seven phases will be described below.

2.1. Problem Identification

In order to understand the problem´s cause, an analysis is conducted through the Ishikawa diagram (Figure 2), in which the possible causes of methotrexate toxicity manifestation in patients with ALL are determined.

2.2. Determine Necessary Requirements

The needs and requirements of the system were provided by the staff of the Centro Regional de Alta Especialidades Pediátricas Hospital de Especialidades Pediátricas de Tuxtla Gutiérrez. A list of the required patient information is presented; Table 1 shows the sociodemographic data of the patients admitted to the pediatric hospital, Table 2 presents the subjective clinical data that oncologists use to determine the presence of ALL in patients, and finally, Table 3 shows the objective clinical data that allow the determination the presence of ALL, as well as the manifestation of symptoms for methotrexate intoxication.

2.3. Needs Analysis

In this phase, the collected information was concentrated for analysis. In order to realize the design and training of the network, it was necessary to separate the data that possess specific information. A normalization method was used for all the collected data (Table 1, Table 2 and Table 3) to design and train the neural network, which uses min-max normalization method. This method transforms each data value so that the minimum value in the dataset is mapped to 0 and the maximum value is mapped to 1. For a value x in a dataset, the min-max normalization formula is:

x_{n o r m} = \frac{x - x_{m i n}}{x_{m a x} - x_{m i n}}

(1)

where:

$x_{n o r m}$ = is the normalized value.
x = is the original data value
$x_{m i n}$ = is the minimum value in the dataset
$x_{m a x}$ = is the maximum value in the dataset.

This transformation ensures that intermediate values are proportionally distributed within the [0, 1] range. The application of this normalization process is detailed in Table 1, Table 2 and Table 3, which show how the raw data were scaled to fit within this range. Table 4 presents the “clinical data objectives” to determine the symptomatology present in the patients. Figure 3 illustrates the graphs depicting patient behavior based on the severity and quantity of symptoms exhibited.

2.4. System Design

To collect the personal and clinical data of patients, shown in Figure 3, admitted to the Tuxtla Gutiérrez Pediatric Hospital, a human–machine interface was developed (Figure 6).

The diagram presented in Figure 4 shows the different patient data collected, which are patient’s registry, clinical data, family data, nutritional values, location of infections, blood analysis, morphological classification, and the patients risk factors for ALL.

Figure 5 depicts the implemented neural architecture; the network comprises 25 neurons in the initial hidden layer (input layer), 15 and 10 neurons in the subsequent hidden layers, and a single neuron in the output layer (result of data processing). For data collection, we developed a graphic interface in order to collect the patient data and train the network, and this interface is shown in Figure 6.

2.5. Development and Documentation

With the patient characteristics, it is possible to know their ALL condition, as well as diagnose the risk of poisoning due to the administration of methotrexate according to the criteria established by the World Health Organization (WHO). In this instance, we put forth of a diagnostic ALL support tool as a potential alternative in which traditional methods have not had the desired results. For the diagnostic system, a graphical user interface (GUIDE) programmed in Matlab is used and shown in Figure 6.

Figure 6. Intelligent diagnosis in pediatric patients with acute lymphoblastic leukemia to predict intoxication by methotrexate.

3. Results

The tests performed on the graphical user interface for the registration of the patient’s personal and clinical data, as well as the results of the design and training of the proposed neural network, are presented.

3.1. Graphical Interface for Data Registration

The graphical user interface, developed in Matlab GUIDE (Figure 6), is divided into four sections. In the first section, the data received are displayed. In the second section, the user can select either the option for the diagnosis of ALL or the option for the diagnosis of intoxication. The third section shows both the data received and the training data. Finally, the fourth section presents the training of the neural network and provides the option to exit the system. The graphical user interface (GUI) receives the patient’s clinical data, which are stored in an Excel file for the purpose of diagnosis.

3.2. Neuronal Methotrexate Validation Network

The previously described backpropagation neural network design comprises 10 input data points and four hidden layers of 1, 10, 15, and 25 neurons (Figure 5). Once the network had been designed, the mean square error was calculated in order to evaluate its performance. A portion of the neural network training code, which was designed in accordance with the aforementioned specifications, is presented herein:

t=xlsread(‘baseDatosRnafinal.xlsx’, ‘Hoja8’, ‘A3:BG1’);
msgbox(“Data have been loaded correctly”);
net=newff(minmax(p),[25,15,10,1], ‘logsig’, ‘tansig’, ‘tansig’, ‘purelin’, ‘trainlm’);
net.trainparam.show = 10;
net.trainparam.lr = 0.05;
net.trainparam.epochs = 50;
net.trainparam.goal = 1 $\times 10^{- 5}$ ;
net=train(net, p, t);
net=init(net);
net, tr=train(net, p, t);
a=sim(net, p);

The mean square error was reduced to

3.10 \times 10^{- 11}

after 38 iterations, indicating that the propagation error was effectively eliminated with each iteration. Figure 7, Figure 8 and Figure 9 present the mean square error, the learning curve, and the linear regression curve, respectively. The latter indicates the amount of data that move away from the input data. All data are included in the line of linear regression, indicating that no data loss occurs during the training process.

3.3. Performance Evaluation: TP, TN, FP, and FN

Acute lymphoblastic leukemia (ALL) is a disease that primarily affects children under the age of 15. In this study, a graphical user interface (GUI) was developed to collect data from the patient and their family environment, providing a comprehensive view of the patient. The design and implementation of artificial neural network algorithms with supervised training enabled the determination of the levels of toxicity associated with the administration of methotrexate in patients with acute lymphoblastic leukemia at the Pediatric High Specialty Hospital of the State of Chiapas. Additionally, statistical patterns established by the World Health Organization and criteria defined by hospital medical specialists were used. Figure 10 depicts the classified information generated by the ANN, which facilitates post-processing knowledge of the same. This graph illustrates the trend according to the range of analysis. The X axis contains information pertaining to the various ranges issued for processing, extending from the minimum to the maximum patients in a segment of the data sheet. The Y axis refers to the number of patients immersed within the ranges indicated. This allowed for an accurate evaluation of symptoms according to the morphological and risk classification of the disease in each patient. This classification will allow physicians to make decisions regarding necessary safety measures during the first 48 hours following methotrexate administration, which promises to significantly improve treatment follow-up, prevent common complications such as mucositis, and ultimately contribute to increasing patients’ life expectancy.

The ability to accurately identify methotrexate levels in patients using this tool can also reduce the costs associated with treatment monitoring by optimizing the management of medical resources and minimizing the need for unnecessary interventions.

To evaluate the performance of the prediction model, an ROC (Receiver Operating Characteristic) curve analysis was conducted. The observed and expected values were used to calculate the true positive rate (TPR) and false positive rate (FPR) and generate the ROC curve. The following code was used for this analysis: Applsci 15 00306 i001

The area under the curve (AUC) obtained is a metric that indicates the model’s ability to distinguish between positive and negative classes. A higher AUC reflects better model performance. In this case, the calculated AUC value provides an indication of the model’s performance in predicting the toxicity levels of methotrexate in patients with ALL; see Figure 11.

4. Discussion

It is possible to create a graphical user interface for data collection from patients under 15 years of age, and this allows us to have a complete picture of the current situation of the patient.

The development and training of the neural network contains the information for hospital admissions of the patients, as well as their family data, which allow us to determine the toxicity levels in methotrexate administration in patients with ALL who are younger than 15 years old, as observed in the learning curve in Figure 7, with statistical patterns established by the World Health Organization or by the doctors themselves. Patient differences in the pathophysiology of MTX neurotoxicity are complex because inhibiting DFR increases adenosine and homocysteine. Adenosine dilates cerebral blood vessels, delays the release of neurotransmitters in the pre-synaptic junctions, modifies the response after the synapse, and slows the neuronal discharge, which is manifested by headache, anorexia, vomiting, nausea, arterial hypertension, poisoning, vertigo, aphasia, agitation, lethargy, seizures, sensory depression, and coma. This neurotoxicity is usually associated with MTX-induced neurotoxicity or renal failure that leads to a poor outcome like in the sample shown by the authors of [23].

5. Conclusions

Acute lymphoblastic leukemia is a condition suffered by children under 15 years old, with treatment presenting substantial progress to improve the life of children suffering from this disease.

It was possible to design a graphical user interface to collect patient and family data to have a complete vision of the patient’s current situation.

The development and training of the neural network allow determining the levels of toxicity from methotrexate administration in patients with ALL with the help of statistical patterns established by the World Health Organization or by the physicians themselves, allowing access to the symptom characteristics depending on the morphological and risk classification of the ALL suffered by the patient.

Once this classification is available, it will help the doctor decide the safety measures that must be taken within of the first 48 hours after methotrexate administration.

This development will help improve treatment monitoring and prevent frequent complications such as mucositis, as well as increase the life expectancy of patients. It will help reduce costs in the treatment follow-up by identifying the levels of methotrexate in patients.

In conclusion, this article represents the impact on a vulnerable population due to economic, family, mobility, and housing situations at the time of having a patient with these characteristics.

Author Contributions

Conceptualization, A.M.S., J.A.O.T. and J.I.B.R.; methodology, A.M.S., J.A.O.T. and J.I.B.R.; software, A.M.S., J.A.O.T. and J.I.B.R.; validation, A.M.S., J.A.O.T., J.I.B.R., J.A.G.R., J.M.V.I. and G.F.A.; formal analysis, A.M.S., J.A.O.T. and J.I.B.R.; investigation, A.M.S., J.A.O.T. and J.I.B.R.; resources, A.M.S.; data curation, A.M.S., J.A.O.T. and J.I.B.R.; writing—original draft preparation, A.M.S., J.A.O.T. and J.I.B.R.; writing—review and editing, A.M.S., J.A.O.T., J.I.B.R., J.A.G.R., J.M.V.I. and G.F.A.; visualization, A.M.S., J.A.O.T., J.I.B.R., J.A.G.R., J.M.V.I. and G.F.A.; supervision, A.M.S.; project administration, A.M.S.; funding acquisition, A.M.S., J.A.O.T., J.I.B.R., J.A.G.R., J.M.V.I. and G.F.A. All authors have read and agreed to the published version of the manuscript.

Funding

The research was funded by the authors and some of the institutions where the authors themselves are employed.

Institutional Review Board Statement

The necessity for ethical review and approval was waived for this study due to the fact that patients had consented to the use of their data and had provided verbal consent to this use. The direct interaction with patients is not the responsibility of our department.

Informed Consent Statement

Informed verbal consent has been obtained from patients for publication of this article. The direct interaction with patients is not the responsibility of our department.

Data Availability Statement

No public involvement occurred in any aspect of this research, We comply with the code of ethics required and requested at the time.

Acknowledgments

We are infinitely grateful to the patients, patients’ relatives, administrators, researchers, doctors, nurses, technicians, and all the direct and indirect actors who carried out this wonderful study; God bless them all, and for those who are no longer with us, we will always remember them with much love. We extend our gratitude to Néstor Rodolfo García Chong for his invaluable contributions to the creation of the dataset, as well as to our former PhD student in technology development, Jorge Iván Bermúdez Rodríguez. Life is a paradigm in which the individual may exist in one moment and not in another. It hurts my soul to know that each patient and his or her family suffer, and I ask God for comfort and refuge in His arms. We are grateful for the multidisciplinary linkage that generated the collaboration between different areas.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:

ALL	Acute Lymphoblastic Leukemia
MTX	Methotrexate
HD-MTX	High-Dose Methotrexate
C677T	Polymorphism
A129C	Polymorphism
MTHFR	Methylenetetrahydrofolate Reductase
MTXPG	Methotrexate Polyglutamate
SLCO1B1, SLC19A1, SLC22A8 and MTHFD1	Genes were shown to be associated with the
	development of acute toxicity after
	HD-MTX treatment
MTHFR C677T	Common changes in MTHFR
MTHFR, ABCB1, ABCC2, and TYMS	Variants in four genes (MTHFR, ABCB1, ABCC2, and
	TYMS) were shown to be associated with toxicity;
	they also found a statistically significant association
	between MTHFR rs1801133 and anemia
	in the consolidation phase
MTHFR rs1801133	Frequency of the c677t polymorphism
SLCO1B1 rs4149056	Polymorphism
C_24h, AUC_48h	MTX pharmacokinetic parameters
RFC1	Reduce Folate Carrier
MTRR	Methionine Synthase Reductase
DFR	Enzyme Di-Hydrofolate Reductase
ANN	Artificial Neural Network
Matlab GUIDE	Matlab Graphical User Interface
GUIDE	Graphical User Interface
WHO	World Health Organization (WHO)
GUI	Graphical User Interface

References

Jaime-Fagundo, J.C.; Forrellat-Barrios, M.; Arencibia-Núñez, A. Urgencias hematológicas. III. Toxicidad por metotrexato. Rev. Cuba. Hematol. Inmunol. Hemoter. 2012, 28, 246–252. [Google Scholar]
Pérez, L.; Shirley, R. Identificación de Alteraciones Biocelulares en la Mucosa Oral en Pacientes con Leucemia Linfoblástica Aguda Post Tratamiento Poliquimioterápico en el Instituto de Oncohematologia Paolo Belli Mediante Estudios Citológicos. Ph.D. Thesis, Universidad Mayor de San Andrés, Ciudad de La Paz, Estado Plurinacional de Bolivia, 2016. [Google Scholar]
Mandal, P.; Samaddar, S.; Chandra, J.; Parakh, N.; Goel, M. Adverse effects with intravenous methotrexate in children with acute lymphoblastic leukemia/lymphoma: A retrospective study. Indian J. Hematol. Blood Transfus. 2020, 36, 498–504. [Google Scholar] [CrossRef]
Di Francia, R.; Crisci, S.; De Monaco, A.; Cafiero, C.; Re, A.; Iaccarino, G.; De Filippi, R.; Frigeri, F.; Corazzelli, G.; Micera, A.; et al. Response and Toxicity to Cytarabine Therapy in Leukemia and Lymphoma: From Dose Puzzle to Pharmacogenomic Biomarkers. Cancers 2021, 13, 966. [Google Scholar] [CrossRef]
Karpa, V.; Kalinderi, K.; Fidani, L.; Tragiannidis, A. Association of microRNA Polymorphisms with Toxicities Induced by Methotrexate in Children with Acute Lymphoblastic Leukemia. Hematol. Rep. 2023, 15, 634–650. [Google Scholar] [CrossRef] [PubMed]
Vora, A. Childhood leukaemia: An update. Paediatr. Child Health 2016, 26, 51–56. [Google Scholar] [CrossRef]
Dong, Y.; Shi, O.; Zeng, Q.; Lu, X.; Wang, W.; Li, Y.; Wang, Q. Leukemia incidence trends at the global, regional, and national level between 1990 and 2017. Exp. Hematol. Oncol. 2020, 9, 14. [Google Scholar] [CrossRef] [PubMed]
Zazuli, Z.; Irham, L.M.; Adikusuma, W.; Sari, N.M. Identification of Potential Treatments for Acute Lymphoblastic Leukemia through Integrated Genomic Network Analysis. Pharmaceuticals 2022, 15, 1562. [Google Scholar] [CrossRef] [PubMed]
de Oliveira Henz, P.; Pinhatti, A.V.; Gregianin, L.J.; Martins, M.; Curra, M.; de Araújo, B.V.; Dalla Costa, T. Population Pharmacokinetic Model of Methotrexate in Brazilian Pediatric Patients with Acute Lymphoblastic Leukemia. Pharm. Res. 2023, 40, 1777–1787. [Google Scholar] [CrossRef] [PubMed]
Csordas, K.; Lautner-Csorba, O.; Semsei, A.F.; Harnos, A.; Hegyi, M.; Erdelyi, D.J.; Eipel, O.T.; Szalai, C.; Kovacs, G.T. Associations of novel genetic variations in the folate-related and ARID5B genes with the pharmacokinetics and toxicity of high-dose methotrexate in paediatric acute lymphoblastic leukaemia. Br. J. Haematol. 2017, 166, 410–420. [Google Scholar] [CrossRef] [PubMed]
Liao, C.; Nie, J.; Xu, X.J.; Zhang, J.Y.; Xu, W.Q.; Song, H.; Tang, Y.M. The effect of the plasma methotrexate concentration during high-dose methotrexate therapy in childhood acute lymphoblastic leukemia. Leuk. Lymphoma 2024, 65, 91–99. [Google Scholar] [CrossRef] [PubMed]
Xu, M.; Wu, S.; Wang, Y.; Zhao, Y.; Wang, X.; Wei, C.; Liu, X.; Hao, F.; Hu, C. Association between high-dose methotrexate-induced toxicity and polymorphisms within methotrexate pathway genes in acute lymphoblastic leukemia. Front. Pharmacol. 2022, 13, 1003812. [Google Scholar] [CrossRef]
Zhan, M.; Sun, Y.; Zhou, F.; Wang, H.; Chen, Z.; Yan, L.; Li, X. Population pharmacokinetics of methotrexate in paediatric patients with acute lymphoblastic leukaemia and malignant lymphoma. Xenobiotica 2022, 52, 265–273. [Google Scholar] [CrossRef]
Zhou, Y.; He, H.; Ding, L.; Wang, T.; Liu, X.; Zhang, M.; Fu, J. Effects of gene polymorphisms on delayed MTX clearance, toxicity, and metabolomic changes after HD-MTX treatment in children with acute lymphoblastic leukemia. Eur. J. Pediatr. 2024, 183, 581–590. [Google Scholar] [CrossRef] [PubMed]
Ramalingam, R.; Kaur, H.; Scott, J.X.; Sneha, L.M.; Arunkumar, G.; Srinivasan, A.; Paul, S.F. Evaluation of cytogenetic and molecular markers with MTX-mediated toxicity in pediatric acute lymphoblastic leukemia patients. Cancer Chemother. Pharmacol. 2022, 89, 393–400. [Google Scholar] [CrossRef] [PubMed]
Al-Sheikh, A.; Yousef, A.M.; Alshamaseen, D.; Farhad, R. Effects of thymidylate synthase polymorphisms on toxicities associated with high-dose methotrexate in childhood acute lymphoblastic leukemia. Cancer Chemother. Pharmacol. 2021, 87, 379–385. [Google Scholar] [CrossRef]
Zgheib, N.K.; Akra-Ismail, M.; Aridi, C.; Mahfouz, R.; Abboud, M.R.; Solh, H.; Muwakkit, S.A. Genetic polymorphisms in candidate genes predict increased toxicity with methotrexate therapy in Lebanese children with acute lymphoblastic leukemia. Pharmacogenet. Genom. 2014, 24, 387–396. [Google Scholar] [CrossRef] [PubMed]
Radtke, S.; Zolk, O.; Renner, B.; Paulides, M.; Zimmermann, M.; Möricke, A.; Stanulla, M.; Schrappe, M.; Langer, T. Germline genetic variations in methotrexate candidate genes are associated with pharmacokinetics, toxicity, and outcome in childhood acute lymphoblastic leukemia. Blood 2013, 121, 5145–5153. [Google Scholar] [CrossRef] [PubMed]
Niedzielska, E.; Węcławek-Tompol, J.; Matkowska-Kocjan, A.; Chybicka, A. The influence of genetic RFC1, MS and MTHFR polymorphisms on the risk of acute lymphoblastic leukemia relapse in children and the adverse effects of methotrexate. Adv. Clin. Exp. Med. 2013, 22, 579–584. [Google Scholar]
Mohar, A.; Frías-Mendivil, M.; Suchil-Bernal, L.; Mora-Macías, T.; Garza, J.G. Descriptive epidemiology of cancer at the Instituto Nacional de Cancerología of Mexico. Salud Publica Mex. 1997, 39, 253–258. [Google Scholar] [CrossRef] [PubMed]
Martínez, B.; Sánchez, C.; Torres, L.; Sánchez, V. Relevancia de la morfología a través del tiempo en el diagnóstico de la leucemia linfoide aguda. Rev. Colomb. Cancerol. 2006, 10, 291–294. [Google Scholar]
Espinoza Yovera, M. Implementación de un Sistema Informático para Mejorar la Gestión de Operaciones en la Empresa GRN MAQUINARIAS SAC en el año 2021. 2022. Available online: https://repositorio.upci.edu.pe/handle/upci/548 (accessed on 21 December 2024).
Faganel Kotnik, B.; Grabnar, I.; Bohanec Grabar, P.; Dolžan, V.; Jazbec, J. Association of genetic polymorphism in the folate metabolic pathway with methotrexate pharmacokinetics and toxicity in childhood acute lymphoblastic leukaemia and malignant lymphoma. Eur. J. Clin. Pharmacol. 2011, 67, 993–1006. [Google Scholar] [CrossRef] [PubMed]

Figure 1. Proposed methodology based on Kendall’s Life Cycle Management [22].

Figure 2. Ishikawa diagram for methotrexate toxicity manifestation in ALL patients.

Figure 3. Normalized clinical data objectives.

Figure 4. Patient data.

Figure 5. Backpropagation neural network design.

Figure 7. Mean square error.

Figure 8. Gradient learning curves and neural network validation.

Figure 9. Linear regression graphic.

Figure 10. The graph depicts the results generated by the neural network in the processing of the pediatric patient data sheet.

Figure 11. AUC curve.

Table 1. Sociodemographic data of the patients.

No.	1	2	3	4	5	6	7	8	9	10
File	21,864	22,097	22,443	18,940	18,941	20,544	20,508	21,302	20,455	21,057
Entry Date	15/08/2014	23/09/2014	03/11/2014	11/07/2013	20/06/2013	17/02/2014	17/02/2014	04/06/2014	11/02/2014	18/05/2014
Exit Date	28/08/2014	29/10/2014	06/11/2014	11/08/2013	29/06/2013	20/03/2014	18/03/2014	09/07/2014	24/03/2014	01/06/2014
Date of Birth	24/01/2010	26/02/2009	04/11/2009	30/12/2011	05/10/2011	14/09/2011	21/12/2011	1/08/2011	27/06/2010	27/01/2010
Today/Date	02/09/2017	02/09/2017	02/09/2017	02/09/2017	02/09/2017	02/09/2017	02/09/2017	02/09/2017	02/09/2017	02/09/2017
Age	7	8	7	5	5	5	5	6	7	7
Gender	M	F	F	F	M	M	M	M	F	M
Last Name	FR	RL	GL	GL	AL	PG	SP	SG	AL	PM
Names	AG	SD	D	E	JH	NR	IL	S	SS	AE
Child Number	1	4	1	1	6	2	1	1	5	2
Municipality	T	M	O	B	M	C	C	O	C	T
Father’s Occupation	Driver	Farmer	Farmer	Farmer	Farmer	Employee	security guard	Employee	Mechanic	Merchant
Mother’s Occupation	Housewife	Deceased	Housewife	Housewife	Farmer	Housewife	Housewife	Housewife	Merchant	Merchant
Total Number of Children	7	2	1	5	2	2	1	5	2	2

Table 2. Subject Clinical of the Patient.

Subjective Clinical Data
Age of the Father	Age of the Mother	Fever	Bleeding	Cephalea or Paleness or Anemic Cor	Abdominal Pain and Hypereosinophilia	Bone Pain or Increase in Volume	Weight (kg)
31	31	Yes	Yes	Yes	No	No	14
24	19	Yes	No	Yes	No	No	17.5
55	46	Yes	Yes	Yes	No	No	18
31	31	Yes	Yes	Yes	No	No	9.5
30	22	Yes	No	No	No	Yes	9.5
27	23	Yes	No	Yes	No	No	11
20	20	Yes	No	No	No	Yes	11.5
43	36	Yes	Yes	Yes	No	Yes	10.6
34	28%	Yes	No	Yes	No	No	9
43	40	Yes	No	Yes	No	No	16

Table 3. Clinical data that determine MTX toxicity.

Clinical Data That determine Symptoms of Acuate Lymphoblastic Leukemia
Waterlow Index (weigth)	74%	102%	107%	90%	90%	87%	100%	100%	98%	97%
Weight Percentile (WHO)	1.65 (−) DE	0.73 (−) DE	0.08 (−) DE	0.70 (−) DE	1.57 (−) DE	1.57 (−) DE	0.77 (−) DE	2.15 (−) DE	3.26 (−) DE	0.47 (−) DE
Size (cm)	111.4	115	95	79	80	90	85	79.7	75	104
Waterlow Index (size)	104%	102%	95%	98%	94%	99%	96%	85%	75%	98%
Percentile (size) WHO	0.96 (+) DE	0.46 (+) DE	0.94 (−) DE	0.72 (−) DE	1.70 (−) DE	0.46 (−) DE	1.19 (−) DE	4.0 (−) DE	6.1 (−) DE	0.33 (−) DE
Sup. Corp. (F. simplified)	0.61	0.72	0.72	0.45	0.45	0.5	0.52	0.49	0.43	0.67
Sup. Corp Haycock	0.65	0.74	0.73	0.46	0.46	0.52	0.53	0.49	0.44	0.68
IMC	11.28	12.85	16.33	15.2	14.8	13.5	15.9	16.66	16	14.7
IMC [22]	P 1	P 10	P 70	P 13	P 7	P 5	P 35	P 66	P 51	P 28
Nutritional Index (Shukla)	77%	89%	103%	87%	88%	85%	96%	91%	79%	96%
Weigh–length–size relationship (WHO 06-07)	3.1 (−) DE	1.68 (−) DE	0.75 (−) DE	0.5 (−) DE	1.13 (−) DE	1.9 (−) DE	0.22 (−) DE	0.07 (−) DE	0.38 (−) DE	0.38 (−) DE
Nutritional Assessment	Moderate Malnutrition	Mild Malnutrition	Eutrophic	Mild Malnutrition	Mild Malnutrition	Moderate Malnutrition	Eutrophic	Eutrophic	Moderate Malnutrition	Eutrophic
Max. Temp. First 24 h	36.7	37.9	37.1	36.6	38.5	38.5	38.5	38	37.5	38
Adenomegaly	No	No	Inguinal	Inguinal	Inguinal, Axillary	Cervicals, Inguinal	Cervicals	Retroauricular; Cervicals	Retroauricular; Cervicals	Cervicals
Hepatomegaly	Yes	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Splenomegaly	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes
Hemorrhagic Manifestations	Gingivorrhagia	Gingivorrhagia	Epistaxis	Wet, Purple	No	No	No	Petechia	No	No
Infection at Admission	Si	Si	No	No	Si	Si	Si	No	No	No

Table 4. Normalized clinical data objectives.

Patient Number	1	2	3	4	5	6	7	8
Waterlow Index (weight)	0.22222222	0.00306306	0.32132132	0.27027027	0.27027027	0.26126126	0.3003003	0.3003003
Weight Percentile (WHO)	−0.125	−0.05530303	−0.00606061	−0.0530303	−0.11893939	−0.11893939	−0.05833333	−0.16287879
Size (cm)	0.68765432	0.70987654	0.64814815	0.48765432	0.49382716	0.55555556	0.52469136	0.49197531
Waterlow Index (size)	0.80620155	0.79069767	0.73643411	0.75968992	0.72868217	0.76744186	0.74418605	0.65891473
Percentile (size) WHO	0.11566265	0.05542169	−0.11325301	−0.08674699	−0.20481928	−0.05542169	−0.14337349	−0.48192771
Sup.Corp (F.simplicada)	0.37888199	0.44720497	0.44720497	0.27950311	0.27950311	0.31055901	0.32298137	0.30434783
Sup.Corp Haycock	0.41139241	0.46835443	0.46202532	0.29113924	0.29113924	0.32911392	0.33544304	0.31012658
IMC	0.18989899	0.21632997	0.27491582	0.25589226	0.24915825	0.22727273	0.26767677	0.27946128
IMC [23]	0.01010101	0.1010101	0.70707071	0.13131313	0.07070707	0.05050505	0.35353535	0.66666667
Nutritional Index (Shukla)	0.39487179	0.45641026	0.52820513	0.44615385	0.45128205	0.43589744	0.49230769	0.46666667
Weight–length–size relationship (WHO 06-07)	−0.37214886	−0.20168067	0.09003601	−0.06002401	−0.13565426	−0.22809124	−0.02641056	0.00840336
Nutritional Assessment	1	1	1	1	1	0.5	1	1
Max. Temp. First 24 h	0.9175	0.9475	0.9275	0.915	0.9625	0.9625	0.9625	0.95
Adenomegaly	0	0	1	1	1	1	1	1
Hepatomegaly	1	0	1	1	1	1	1	1
Splenomegaly	1	1	1	1	1	1	1	0
Hemorrhagic Manifestations	1	1	1	1	0	0	0	1
Infection at Admission	1	1	0	0	1	1	1	0

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Share and Cite

MDPI and ACS Style

Santiago, A.M.; Rodríguez, J.I.B.; Torres, J.A.O.; Rabasa, J.A.G.; Izaguirre, J.M.V.; Alejandro, G.F. Detecting Methotrexate in Pediatric Patients Using Artificial Neural Networks. Appl. Sci. 2025, 15, 306. https://doi.org/10.3390/app15010306

AMA Style

Santiago AM, Rodríguez JIB, Torres JAO, Rabasa JAG, Izaguirre JMV, Alejandro GF. Detecting Methotrexate in Pediatric Patients Using Artificial Neural Networks. Applied Sciences. 2025; 15(1):306. https://doi.org/10.3390/app15010306

Chicago/Turabian Style

Santiago, Alejandro Medina, Jorge Iván Bermúdez Rodríguez, Jorge Antonio Orozco Torres, Julio Alberto Guzmán Rabasa, José Manuel Villegas Izaguirre, and Gladys Falconi Alejandro. 2025. "Detecting Methotrexate in Pediatric Patients Using Artificial Neural Networks" Applied Sciences 15, no. 1: 306. https://doi.org/10.3390/app15010306

APA Style

Santiago, A. M., Rodríguez, J. I. B., Torres, J. A. O., Rabasa, J. A. G., Izaguirre, J. M. V., & Alejandro, G. F. (2025). Detecting Methotrexate in Pediatric Patients Using Artificial Neural Networks. Applied Sciences, 15(1), 306. https://doi.org/10.3390/app15010306

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Menu