Monitoramento da qualidade do processamento laboratorial no transplante autólogo de células progenitoras hematopoéticas
Carregando...
Data
2021-05-26
Autores
Lopes, Lia Beatriz Mantovani
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O Transplante de Células Tronco Hematopoética Autólogo (TCTH-auto) é terapia consolidada para tratamento de patologias onco-hematológicas e consiste em reestabelecer a função medular através da infusão intravenosa de células progenitoras hematopoéticas (CPH) do próprio paciente. Este estudo teve como objetivo caracterizar indicadores de qualidade na etapa de processamento laboratorial do TCTH-auto, através de análise retro-prospectiva de dados de pacientes submetidos ao Transplante de Medula Óssea do Hospital das Clínicas da Faculdade de Medicina de Botucatu (HCFMB). Neste contexto, a implementação de indicadores que permitam checar etapas do processo, são muito relevantes para sanar questões que encontramos durante e ao final deste. Os dados coletados compreendem o período de agosto de 2016 a dezembro de 2020, totalizando 107 pacientes, dos quais foram coletadas variáveis envolvidas no processo de TCTH-auto, e posteriormente submetidas à analises estatísticas convencionais e análises de dados avançadas, como técnicas de Machine Learning, árvore de decisão, Análise de Componentes Principais (PCA) e Análise de Discriminante Linear (LDA). As análises de Machine Learning nos permitiram identificar a partir do conjunto total de dados, um preditor da eficiência do processo de TCTH-auto, o CD34+ pré. Embora este também seja principal preditor do momento ideal da coleta por aférese,e isto já esteja bem consolidado na literatura e na prática clínica, os resultados do presente estudo validam o desempenho do serviço de transplantes. Diferencial foi estabelecido aplicando-se o algoritmo J48, culminando na obtenção de árvore de decisão preditiva que estabelece valores de corte. Quando a contagem de CPH no sangue periférico for >23 CD34/µL indica alta probabilidade da coleta ser eficiente, já valores de CD34 ≤23/µL indicam menor probabilidade de eficiência. A análise de Componentes Principais (PCA), indicou que as duas variáveis que apresentam maior correspondência com a eficiência da aférese são CD34+ pré e hematócrito (Ht). Utilizando-se a Análise de Discriminante Linear (LDA) elaborou-se a equação (score) para predição da eficiência, combinando valores absolutos de CD34+ pré e Ht, a qual pode ser aplicada especialmente quando os valores de CD34+ estiverem menores que 23 µL. Se o resultado da equação for ≤ -0,14 a probabilidade de eficiência na coleta é maior, no entanto se >-0,14 a probabilidade de eficiência é baixa. Este indicador possibilita a realização da coleta em pacientes com CD34+ <23µL. Outro indicador que pode ser adotado é o coeficiente de eficiência (CE) que possibilita avaliar a eficiência da leucaférese. Aplicando este indicador, observamos que temos obtido eficiências superiores às estimadas, de modo que podem ser estabelecidos maiores índices de estimativa de aférese. Diante do exposto, a identificação de potenciais indicadores de qualidade nas etapas desse processo que venham assegurar a qualidade do transplante de medula realizado no HCFMB, buscando a excelência e, tendo em vista o aumento de demanda neste serviço de abrangência regional, são de suma importância, entendendo-se que os resultados gerados por este estudo terão aplicabilidade imediata e poderão trazer melhorias na gestão da qualidade deste serviço de alta complexidade no âmbito do SUS.
The Autologous Hematopoietic Stem Cell Transplant (HSCT-auto) is a consolidated therapy for the treatment of onco-hematologic pathologies and consists of reestablishing the medullary function through the intravenous infusion of the patient's own hematopoietic progenitor cells (PHC). This study aimed to characterize quality indicators in the laboratory processing stage of the HSCT-auto, through the analysis of retrospective data from patients undergoing Bone Marrow Transplantation at Hospital das Clínicas, Faculty of Medicine of Botucatu (HCFMB). In this context, the implementation of indicators that allow checking stages of the process are very relevant to resolve issues we encountered during the process. The collected data cover the period from August 2016 to December 2020, totaling 107 patients, of which variables involved in the HSCT-auto process were collected, and later submitted to conventional and advanced statistical analysis, such as Machine Learning techniques, tree of decision, Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). Machine Learning analyzes allowed us to identify, from the total data set, an efficiency predictor of the HSCT-auto process, the CD34+ pre, which is the main predictor of the ideal moment of apheresis collection. Although this predictor is already well established in the literature and in clinical practice, the results validate its performance in our service. However, a differential was established by applying the J48 algorithm, resulting in a predictive decision tree that establishes cutoff values. When the CPH count in peripheral blood is >23 CD34/µL indicates a high probability of being efficient collection, while CD34 values ≤23/µL indicate a lower probability of efficiency. The Principal Component Analysis (PCA) indicated that the two variables that present the greatest correspondence with the efficiency of apheresis are CD34+ pre and hematocrit (Ht). Using Linear Discriminant Analysis (LDA), the equation (score) for efficiency prediction was elaborated, combining absolute values of CD34+ pre and Ht, which can be applied especially when CD34+ values are less than 23 µL. If the result of the equation is ≤ -0.14 the probability of efficiency in collection is higher, however if >-0.14 the probability of efficiency is low. This indicator makes it possible to carry out the collection in patients with CD34+ <23µL. Another indicator that can be adopted is the coefficient of efficiency (EC), which makes it possible to assess the efficiency of leukapheresis. Applying this indicator, we observe that we have obtained efficiencies higher than those estimated, so that higher apheresis estimation rates can be established. Given the above, the identification of potential quality indicators in the stages of this process that will ensure the quality of the bone marrow transplant performed at the HCFMB, seeking excellence and, in view of the increased demand in this service with regional coverage, is of paramount importance , it being understood that the results generated by this study will have immediate applicability and may bring improvements in the quality management of this highly complex service within the SUS.
The Autologous Hematopoietic Stem Cell Transplant (HSCT-auto) is a consolidated therapy for the treatment of onco-hematologic pathologies and consists of reestablishing the medullary function through the intravenous infusion of the patient's own hematopoietic progenitor cells (PHC). This study aimed to characterize quality indicators in the laboratory processing stage of the HSCT-auto, through the analysis of retrospective data from patients undergoing Bone Marrow Transplantation at Hospital das Clínicas, Faculty of Medicine of Botucatu (HCFMB). In this context, the implementation of indicators that allow checking stages of the process are very relevant to resolve issues we encountered during the process. The collected data cover the period from August 2016 to December 2020, totaling 107 patients, of which variables involved in the HSCT-auto process were collected, and later submitted to conventional and advanced statistical analysis, such as Machine Learning techniques, tree of decision, Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). Machine Learning analyzes allowed us to identify, from the total data set, an efficiency predictor of the HSCT-auto process, the CD34+ pre, which is the main predictor of the ideal moment of apheresis collection. Although this predictor is already well established in the literature and in clinical practice, the results validate its performance in our service. However, a differential was established by applying the J48 algorithm, resulting in a predictive decision tree that establishes cutoff values. When the CPH count in peripheral blood is >23 CD34/µL indicates a high probability of being efficient collection, while CD34 values ≤23/µL indicate a lower probability of efficiency. The Principal Component Analysis (PCA) indicated that the two variables that present the greatest correspondence with the efficiency of apheresis are CD34+ pre and hematocrit (Ht). Using Linear Discriminant Analysis (LDA), the equation (score) for efficiency prediction was elaborated, combining absolute values of CD34+ pre and Ht, which can be applied especially when CD34+ values are less than 23 µL. If the result of the equation is ≤ -0.14 the probability of efficiency in collection is higher, however if >-0.14 the probability of efficiency is low. This indicator makes it possible to carry out the collection in patients with CD34+ <23µL. Another indicator that can be adopted is the coefficient of efficiency (EC), which makes it possible to assess the efficiency of leukapheresis. Applying this indicator, we observe that we have obtained efficiencies higher than those estimated, so that higher apheresis estimation rates can be established. Given the above, the identification of potential quality indicators in the stages of this process that will ensure the quality of the bone marrow transplant performed at the HCFMB, seeking excellence and, in view of the increased demand in this service with regional coverage, is of paramount importance , it being understood that the results generated by this study will have immediate applicability and may bring improvements in the quality management of this highly complex service within the SUS.
Descrição
Palavras-chave
Células-tronco hematopoéticas, Indicadores de qualidade em assistência à saúde, Transplante autólogo, Hematopoietic stem cells, Health care quality indicators, Autologous transplantation