Very preterm neonates (born before 32 weeks of gestation) are more exposed to critical glycemic variations in the first few days after delivery. Neonatal glycemic instability, defined as a series of hypo- and/or hyperglycemic events, seems to be associated with increased mortality rate, as well as impaired neurological development. Previous studies have shown a relation between glucose values and cerebral hemodynamics. Glucose is one of the most important cerebral metabolites; therefore, its decrease or excessive increase can lead to important cerebral hemodynamics variation which can affect brain development. The aim of this work is to investigate whether glycemic events in very preterm neonates during the first days of life can modify and modulate the brain patterns of resting state functional connectivity. The dataset, unique in its kind, consists of 47 very preterm neonates whose glycemic and hemodynamics data were collected at the Neonatal Intensive Care Unit of the University Hospital of Padova during the first week of life, between March 2020 and June 2023. The glycemic signal was acquired through a continuous glucose monitoring (CGM) device for a minimum of 20 hours to a maximum of 6 days; during this period, the brain hemodynamic signal was also acquired with diffuse optical tomography (DOT). The DOT device (NTS Gowerlabs) is composed of 8 sources and 8 detectors, for a total of 64 channels localized on the parietal, motor, superior occipital, superior temporal and posterior frontal brain areas. The glycemic signal was analyzed to identify and characterize all the glycemic events and the duration of the euglycemic intervals. From this analysis, 38 neonates were initially excluded based on the following criteria: presence of critical temporal gaps within the CGM signal, presence of hypo- or hyperglycemia events before the first euglycemic interval or if the total time outside the euglycemia range was below the 25th percentile computed on the entire selected population. For the cerebral hemodynamic analysis, two five-minute intervals were identified in the first and last euglycemic window, choosing the 5 min windows having the least number of bad channels, fewer motion artifacts, and adequate coverage of the cortical surface by the channel array. DOT signals were pre-processed, removing bad channels, and paying particular attention to the correction of motion artifacts through the implementation of a strategy based on the Wavelet filter. The resting state functional connectivity analysis (rsFC) was then performed using a seed-based strategy, assessing correlations between 9 Regions of Interest (ROI) identified based on the international 10-20 EEG system. The differences in rsFC matrices were then linked to various glycemic metrics extracted from the CGM data (e.g., glycemic variability, number of hypoglycemic events, duration of glycemic events, etc.). At the end of the exploratory analysis, 24 statistically significant correlations were found for the HbO signal and 39 for HbR. In most cases, ROIs in the parietal region (P3, Pz, P4) were involved, suggesting a higher likelihood of potential brain injuries or distress in that area. After adjusting for multiple comparisons, only two statistically significant positive correlations remained between changes in rsFC for HbR in F3-P3 and, respectively, the Absolute Average Rate of Change and the Coefficient of Variation of the CGM signal. The presence of a strong positive correlation for these two parameters suggests a relationship between glycemic variability and differences in brain connectivity in the F3-P3 ROI pair, concerning the analyzed population. These results may support future studies on glucose control and its impact on cerebral hemodynamics, to assess its effects in the short and long term in relation to the neurodevelopment of very preterm neonates.
I neonati molto prematuri (nati prima delle 32 settimane di gestazione) sono più esposti a variazioni glicemiche critiche nei primi giorni dopo il parto. L’instabilità glicemica neonatale, definita come una serie di eventi ipo- e/o iperglicemici, sembra essere associata ad un incremento del tasso di mortalità ed un compromesso sviluppo neurologico e cognitivo. Alcuni studi hanno dimostrato una relazione tra i valori della concentrazione di glucosio e dell’emodinamica cerebrale. Il glucosio è il più importante metabolita cerebrale; di conseguenza, la sua diminuzione o eccessivo incremento può provocare variazioni dell’emodinamica cerebrale che possono influire nello sviluppo encefalico. L’obiettivo principale di questa tesi consiste nell’indagare se gli eventi glicemici in neonati molto prematuri durante i primi giorni di vita possono modificare e modulare i pattern della connettività funzionale cerebrale a riposo. Il dataset a disposizione, unico nel suo genere, consiste in 47 neonati prematuri i cui dati glicemici ed emodinamici sono stati acquisiti presso la terapia intensiva neonatale dell’ospedale Universitario di Padova durante la prima settimana di vita, tra marzo 2020 e giugno 2023. Il segnale glicemico è stato rilevato attraverso l’utilizzo di un sensore di monitoraggio continuo del glucosio (CGM) per un minimo di 20 ore ed un massimo di 6 giorni; durante questo periodo, l’emodinamica cerebrale è stata acquisita tramite tomografia ottica diffusa (DOT). Il dispositivo utilizzato per eseguire la DOT (NTS Gowerlabs) è composto da 8 sorgenti e 8 rilevatori, per un totale di 64 canali posizionati sulla zona cerebrale parietale, motoria, occipitale superiore, temporale superiore e frontale posteriore. Il segnale glicemico è stato analizzato per identificare e classificare gli eventi glicemici, nonché per calcolare la durata degli intervalli di euglicemia. Da questa analisi, 38 pazienti sono stati esclusi poiché il loro tracciato CGM presentava alcune anomalie. Per l’analisi dell’emodinamica cerebrale, due finestre da cinque minuti sono state identificate all’interno del primo ed ultimo intervallo di euglicemia rilevato, scegliendo la finestra con la migliore qualità del segnale. Il segnale DOT è stato pre-processato rimuovendo i canali rumorosi e poco informativi e prestando particolare attenzione all’attenuazione degli artefatti da movimento attraverso l’applicazione di una strategia basata sul filtraggio Wavelet. L’analisi della connettività cerebrale funzionale a riposo (rsFC) è stata effettuata utilizzando una strategia seed based valutando la correlazione tra 9 ROI identificate sulla base del sistema internazionale 10-20 dell’EEG. La differenza delle matrici di rsFC è stata messa in relazione con alcune metriche glicemiche estratte dai tracciati CGM (e.g., variabilità glicemica, numero di ipoglicemie, durata degli eventi glicemici, etc.). Al termine dell’analisi esplorativa si sono riscontrate 24 correlazioni statisticamente significative per il segnale HbO e 39 per HbR. Nella maggior parte sono coinvolte ROI appartenenti alla regione parietale (P3, Pz, P4) presupponendo una maggiore presenza di possibili lesioni o sofferenze cerebrali in quella zona. Dopo aver effettuato la correzione per confronti multipli, rimangono due correlazioni positive statisticamente significative tra l’alterazione della rsFC per HbR in F3-P3 con, rispettivamente, l’Absolute Average Rate of Change e il Coefficiente di Variazione del segnale CGM. Questi risultati fanno presupporre che vi sia una relazione tra la variabilità glicemica e la differenza della connettività cerebrale nella coppia di ROI F3-P3, per quanto riguarda la popolazione analizzata. I risultati ottenuti in questa tesi possono supportare futuri studi relativi al controllo glicemico e al suo impatto sull’emodinamica cerebrale, per valutarne l’effetto in relazione al neurosviluppo dei neonati molto prematuri.
Relazione tra connettività funzionale cerebrale a riposo e variazioni glicemiche in neonati molto prematuri: uno studio mediante monitoraggio continuo del glucosio e spettroscopia nel vicino infrarosso
BIANCO, GIACOMO
2022/2023
Abstract
Very preterm neonates (born before 32 weeks of gestation) are more exposed to critical glycemic variations in the first few days after delivery. Neonatal glycemic instability, defined as a series of hypo- and/or hyperglycemic events, seems to be associated with increased mortality rate, as well as impaired neurological development. Previous studies have shown a relation between glucose values and cerebral hemodynamics. Glucose is one of the most important cerebral metabolites; therefore, its decrease or excessive increase can lead to important cerebral hemodynamics variation which can affect brain development. The aim of this work is to investigate whether glycemic events in very preterm neonates during the first days of life can modify and modulate the brain patterns of resting state functional connectivity. The dataset, unique in its kind, consists of 47 very preterm neonates whose glycemic and hemodynamics data were collected at the Neonatal Intensive Care Unit of the University Hospital of Padova during the first week of life, between March 2020 and June 2023. The glycemic signal was acquired through a continuous glucose monitoring (CGM) device for a minimum of 20 hours to a maximum of 6 days; during this period, the brain hemodynamic signal was also acquired with diffuse optical tomography (DOT). The DOT device (NTS Gowerlabs) is composed of 8 sources and 8 detectors, for a total of 64 channels localized on the parietal, motor, superior occipital, superior temporal and posterior frontal brain areas. The glycemic signal was analyzed to identify and characterize all the glycemic events and the duration of the euglycemic intervals. From this analysis, 38 neonates were initially excluded based on the following criteria: presence of critical temporal gaps within the CGM signal, presence of hypo- or hyperglycemia events before the first euglycemic interval or if the total time outside the euglycemia range was below the 25th percentile computed on the entire selected population. For the cerebral hemodynamic analysis, two five-minute intervals were identified in the first and last euglycemic window, choosing the 5 min windows having the least number of bad channels, fewer motion artifacts, and adequate coverage of the cortical surface by the channel array. DOT signals were pre-processed, removing bad channels, and paying particular attention to the correction of motion artifacts through the implementation of a strategy based on the Wavelet filter. The resting state functional connectivity analysis (rsFC) was then performed using a seed-based strategy, assessing correlations between 9 Regions of Interest (ROI) identified based on the international 10-20 EEG system. The differences in rsFC matrices were then linked to various glycemic metrics extracted from the CGM data (e.g., glycemic variability, number of hypoglycemic events, duration of glycemic events, etc.). At the end of the exploratory analysis, 24 statistically significant correlations were found for the HbO signal and 39 for HbR. In most cases, ROIs in the parietal region (P3, Pz, P4) were involved, suggesting a higher likelihood of potential brain injuries or distress in that area. After adjusting for multiple comparisons, only two statistically significant positive correlations remained between changes in rsFC for HbR in F3-P3 and, respectively, the Absolute Average Rate of Change and the Coefficient of Variation of the CGM signal. The presence of a strong positive correlation for these two parameters suggests a relationship between glycemic variability and differences in brain connectivity in the F3-P3 ROI pair, concerning the analyzed population. These results may support future studies on glucose control and its impact on cerebral hemodynamics, to assess its effects in the short and long term in relation to the neurodevelopment of very preterm neonates.| File | Dimensione | Formato | |
|---|---|---|---|
|
Bianco_Giacomo.pdf
accesso aperto
Dimensione
10.32 MB
Formato
Adobe PDF
|
10.32 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/55250