Doxorubicin (DOX) is a potent chemotherapeutic agent routinely administered in the treatment of several pediatric malignancies, including acute lymphoblastic leukemia (ALL). Despite its efficacy to improve the outlook of cancer patients, doxorubicin is known to cause a wide spectrum of cardiac complications that can seriously affect the quality of life of ALL survivors. Structural changes can occur in myocardial tissue after exposure to doxorubicin-based chemotherapy. These changes are frequently accompanied by alterations in left ventricular (LV) geometry and function. Myocardial damage can progress silently for many years before the manifestation of clinical symptoms and become apparent even long time after completion of treatment. Higher cumulative doses of this agent increase the risk for late cardiac complications. The feasibility of CMR imaging has been established and a variety of software tools for 3D geometric modeling of the heart have been developed to assess ejection fraction, wall thicknesses, end-systolic and end-diastolic volumes. Finite element (FE) modelling and inverse material parameter identification strategies were then introduced to take into account the passive mechanical behavior of the myocardial tissue. In light of the above, we undertook a study to assess the asymptomatic changes in LV structure and function in a group of long-term survivors of childhood ALL treated with low to moderate doses of doxorubicin therapy. Given the high potential of numerical cardiac modeling, this evaluation was conducted using a FE model-based approach that integrates the subject-specific shape and motion of the ventricular chamber directly from imaging data. Eighty-four ALL survivors (23±7 years old) were prospectively enrolled and stratified into two different groups, designated as standard-risk (SR, n=19) and high-risk groups, according to their risk of tumor recurrence. Subjects treated for high-risk ALL were further divided into two groups depending upon whether they did (HRdex, n=36) or did not (HR, n=45) receive the protective therapy (dexrazoxane) in an attempt to reduce the likelihood of late cardiotoxicity. Furthermore, for purposes of comparison, 10 healthy volunteers (HV, 22±4 years), with no prior history of cancer or cardiac pathologies and similar in age to the survivors, were used as controls. As a part of the investigation of late-onset cardiotoxicity, all subjects underwent CMR imaging, transthoracic echocardiography, and exercise stress testing. From the acquired CMR data, a trained operator extracted the subject-specific shape and motion of the LV using a guide-point framework (CIM v8.1, University of Auckland, New Zealand). The inner and outer borders of the LV walls were semi-automatically drawn from six guidepoints placed by the operator at end-systole and then manually corrected for mis-registration errors. From this tracing, ejection fraction, stroke volume, myocardial mass, wall thickness, end-diastolic and end-systolic volumes were computed for each of the study participants. After that, inter- and intraobserver repeatability of the segmentation results were quantified by intra-class coefficients (ICC) on four reconstructions of 15 leukemia survivors, each by four trained operators, and on three reconstructions of the same subject, each by a single operator. For a subset of the leukemia survivor population (n=48), subject-specific data about LV wall motion and geometry at diastasis were incorporated into a model of the passive LV mechanics and used to estimate the global stiffness parameter C1 by means of an inverse materials parameters identification strategy. This biomechanical parameter was initially calculated by assuming physiologically realistic values of 0.75, 1.0, and 1.25 kPa as pressure loading constraints for all participants. Once the subject-specific LV pressures (at rest and during peak exercise) became available, we incorporated such data in the model and repeated the FE analysis. The resulting values of C1 were reported and compared between groups for each of the five loading scenarios. Statistical comparisons were performed by one-way analysis of variance (ANOVA) on global and regional geometrical parameters, and two-way repeated-measures ANOVA on time-dependent geometrical parameters. Ultimately, a sensitivity analysis was conducted to evaluate the dependence of the hyperelastic property on the diastasis frame and the pressure load. In our experience, inter-observer repeatability was good for regional geometrical parameters (ICC, 0.60-0.74) and excellent for global geometrical parameters (ICC, 0.75-1.00), while intra-observer repeatability was excellent for both regional and global parameters (ICC, 0.75-1.00). Groups had similar LV function values. No significant differences were observed among the four study groups in ejection fraction, stroke volume, mass, end-diastolic or end-systolic volumes. Some differences were detected in epicardial volume only between end-systole and diastasis phases (p<0.01) among the HRdex and HV or SR groups, and among the HV and HR groups. Interestingly, the hyperelastic property for standard pressures was slightly lower in the HR group when compared against the HRdex or SR group, and also when compared against the control group (p<0.05). In contrast, no appreciable difference could be noted in the hyperelastic property for intra-ventricular pressures at rest (p>0.5) and peak exercise (p>0.6). From this analysis, it is clear that the global and regional geometrical parameters are not sufficiently sensitive to capture the subtle changes induced by late doxorubicin cardiotoxicity in LV structure and function. Nevertheless, the time-dependent global parameters provided preliminary evidence that early diastole was more affected by doxorubicin exposure than systole or late diastole. The smaller hyperelastic property in the high-risk group suggested a myocardial tissue more prone to dilatation if increased intra-ventricular pressure is applied than in the other study groups. This finding is consistent with what has been observed in long-term survivors of adult leukemia treated with high-dose doxorubicin-based chemotherapy. It should be noted, however, that these estimates could account only for the geometrical effects of myocardial remodeling on the passive ventricular mechanics since the subject-specific LV cavity pressure was not included in these simulations. Similar results were obtained when applying intra-ventricular pressures at rest and peak exercise. In conclusion, this study demonstrated that the subclinical cardiotoxicity of doxorubicin can be non-invasively assessed through the mechanical behavior analysis of the LV on CMR images. Additional investigations will be necessary to confirm our results and draw a firm conclusion about the long-term prognostic significance of alterations in myocardial stiffness and their relationship with ALL risk status in this or similar cohort of survivors. In future works, we hope to include the FE simulation of the left ventricular mechanics during systole and early diastole in order to better capture the changes in stiffness across groups. To the same purpose, it might be convenient to increase the temporal resolution of the image data in both leukemia survivors and control subjects.
La doxorubicina è un potente agente chemioterapico comunemente utilizzato nel trattamento di svariati tumori dell’età pediatrica, tra cui la leucemia linfoblastica acuta (LLA). Nonostante i benefici apportati in termini di aspettativa di vita ai pazienti affetti da LLA, la doxorubicina causa un ampio spettro di complicanze cardiache che possono compromettere sensibilmente la qualità di vita dei sopravvissuti alla leucemia. I cambiamenti strutturali che possono verificarsi nel tessuto miocardico a seguito dell’esposizione a doxorubicina sono spesso accompagnati da alterazioni della geometria e della funzione del ventricolo sinistro (VS). Il danno miocardico può progredire silenziosamente per anni, per poi manifestarsi senza alcun preavviso anche molto tempo dopo il termine del trattamento con doxorubicina. È stato dimostrato che dosi cumulative più elevate di doxorubicina aumentano il rischio di sviluppare complicanze cardiache a lungo termine. La cardio-risonanza magnetica (CRM) è stata validata come strumento per valutare la funzione ventricolare, mentre svariati software per la modellazione geometrica del VS sono stati sviluppati per misurare frazione di eiezione, volumi telesistolico e telediastolico, e spessore della parete ventricolare. La modellazione agli elementi finiti (EF) della meccanica ventricolare sinistra e la tecnica dell’analisi inversa per l'identificazione dei parametri dei materiali sono state in seguito introdotte per analizzare il comportamento meccanico passivo del tessuto miocardico. In questo contesto, abbiamo intrapreso uno studio volto ad analizzare le alterazioni asintomatiche della geometria e della funzione ventricolare cardiaca in un gruppo di giovani sopravvissuti a LLA infantile che erano stati sottoposti in precedenza a chemioterapia adiuvante con doxorubicina (dosi cumulative da basse a moderate). Considerato l’elevato potenziale della modellazione numerica della biomeccanica ventricolare, tale analisi è stata eseguita utilizzando un modello ad elementi finiti del ventricolo sinistro ottenuto a partire da dati estratti da immagini di risonanza magnetica dei singoli soggetti. Ai fini del presente studio, 84 adulti sopravvissuti a LLA infantile (23±7 anni) sono stati reclutati prospetticamente e ripartiti in funzione del rischio di recidiva in due gruppi distinti, denominati rispettivamente “rischio standard” (SR, n=19) e “rischio elevato”. Dopodiché, i soggetti classificati ad alto rischio di ricaduta sono stati ulteriormente divisi in due sottogruppi a seconda che avessero ricevuto (HRdex, n = 36) o meno (HR, n = 45) la terapia cardioprotettiva (dexrazoxane) atta a ridurre il rischio di danni cardiaci a lungo termine. Infine, una coorte di 10 volontari sani (HV, 22±4 anni), simili per età ai sopravvissuti alla leucemia e senza alcuna storia pregressa di tumore o cardiomiopatia, è stato utilizzata come gruppo di controllo. Come parte dell’analisi della cardiotossicità tardiva dopo chemioterapia, i partecipanti allo studio sono stati sottoposti a risonanza magnetica cardiaca, ecocardiografia transtoracica e test da sforzo. Dalle immagini di risonanza magnetica ottenute, un operatore opportunamente formato ha estratto le informazioni della forma e del movimento del ventricolo sinistro utilizzando un set di marcatori fiduciali (CIM v8.1, Università di Auckland, Nuova Zelanda). I bordi endocardici e epicardici del miocardio ventricolare sono stati tracciati in modo semiautomatico a partire da sei punti fiduciali posizionati dall’operatore in corrispondenza della telesistole e poi corretti manualmente al fine di ridurre gli eventuali errori di segmentazione. Da tale segmentazione, frazione di eiezione, gittata cardiaca, massa, spessore della parete, e volumi telesistolico e telediastolico sono stati calcolati per ognuno dei soggetti in esame. Ripetibilità inter e intraosservatore dei risultati della segmentazione sono state quantificate con coefficienti di correlazione intraclasse (ICC) su 4 ricostruzioni di 15 sopravvissuti a ALL, ciascuna eseguita da 4 operatori differenti, e su 3 ricostruzioni dello stesso soggetto eseguite tutte da un unico operatore. Per un sottogruppo di sopravvissuti a ALL (n=48), le informazioni della forma e del movimento ventricolare in diastasi sono state incorporate in un modello della meccanica ventricolare passiva e, successivamente, utilizzate per calcolare la proprietà iperelastica C1 tramite analisi inversa agli EF. Tale parametro biomeccanico è stato inizialmente stimato imponendo tre valori di pressione fisiologicamente realistici (0.75, 1.0 e 1.25 kPa) come condizioni di carico per tutti i partecipanti. Una volta ottenute le pressioni ventricolari specifiche per i soggetti in esame (a riposo e durante l’esercizio massimale), l’analisi meccanica è stata ripetuta incorporando tali dati nel modello. Per ciascuno dei cinque scenari di carico, le stime risultanti sono state riportate e confrontata tra i gruppi. I confronti statistici sono stati effettuati mediante analisi unidirezionale della varianza (ANOVA) nel caso dei parametri globali e regionali, e mediante ANOVA a 2 vie con misure ripetute nel caso dei parametri geometrici dipendenti dal tempo. Infine, è stata condotta un'analisi di sensibilità per valutare la dipendenza di C1 dalla diastasi e dal carico di pressione. Nel nostro studio, la ripetibilità interosservatore dei risultati della segmentazione è stata buona per i parametri regionali (ICC, 0.60-0.74) ed eccellente per i parametri globali (ICC, 0.75-1.00), mentre la ripetibilità intraosservatore è stata eccellente sia per i parametri regionali che per quelli globali (ICC, 0.75-1.00). Nessuna significativa differenza è emersa dal confronto tra i quattro gruppi di partecipanti allo studio per quanto riguarda frazione di eiezione, gittata cardiaca, massa miocardica, o volumi telesistolico e telediastolico. Differenze statisticamente significative sono state riscontrate nel volume epicardico, unicamente tra telesistole e diastasi (p<0.01), tra i gruppi HRdex e HV o SR e tra i gruppi HV e HR. È interessante notare, inoltre, che la proprietà iperelastica per pressioni standard era leggermente più piccola nel gruppo HR rispetto al gruppo HRdex o SR e anche rispetto alla coorte di controllo (p<0.05). Nessuna differenza apprezzabile è stata, invece, riscontrata nella proprietà iperelastica per pressioni ventricolari in condizioni di riposo (p>0.5) e di esercizio fisico massimale (p>0.6). Dall’analisi precedente risulta chiaro che i parametri geometrici (globali e regionali) non sono sufficientemente sensibili per rilevare le lievi alterazioni causate dalla cardiotossicità tardiva della doxorubicina nella geometria e funzione del ventricolo sinistro. Ciononostante, le variabili globali tempo-dipendenti hanno dimostrato che l'esposizione alla doxorubicina ha un effetto più deleterio sulla fase precoce della diastole che sulla sistole. L’esistenza di un C1 più piccolo nel gruppo ad alto rischio di recidiva indica un miocardio più incline alla dilatazione quando una pressione intraventricolare più elevata viene applicata, rispetto agli altri gruppi. Questo risultato è coerente con quanto documentato per i sopravvissuti a lungo termine a leucemia in età adulta che erano stati trattati con doxorubicina ad alte dosi. Va evidenziato, tuttavia, che le nostre stime di rigidità riflettono unicamente gli effetti geometrici del rimodellamento del miocardio sulla meccanica ventricolare passiva. Infatti, la pressione intraventricolare sinistra del soggetto non era stata applicata in queste simulazioni. Risultati simili sono stati ottenuti incorporando nel modello pressioni intraventricolari a riposo e durante il massimo esercizio fisico. In conclusione, la ricerca che abbiamo intrapreso ha mostrato che la cardiotossicità subclinica dopo terapia con doxorubicina può essere valutata in maniera non invasiva mediante analisi del comportamento meccanico del ventricolo sinistro con geometria ottenuta da immagini medicali. Tuttavia, ulteriori analisi sono necessarie per confermare i risultati fin qui raccolti e trarre una conclusione definitiva circa il significato prognostico a lungo termine di alterazioni della rigidità miocardica e del loro legame con lo stato di rischio di LLA in questa o in una coorte similare. In studi futuri, sarebbe auspicabile includere la simulazione numerica della meccanica ventricolare sinistra durante un ciclo cardiaco completo in modo da meglio analizzare i cambiamenti della rigidità tra i quattro gruppi di studio. Inoltre, per la stessa ragione, sarebbe utile incrementare la risoluzione temporale delle immagini sia nei sopravvissuti di leucemia infantile sia nei soggetti di controllo.
Subtle changes in hyperelastic properties of myocardium with cardiotoxicity remodeling from cardiac magnetic resonance
GAMBA, MARIANNA
2018/2019
Abstract
Doxorubicin (DOX) is a potent chemotherapeutic agent routinely administered in the treatment of several pediatric malignancies, including acute lymphoblastic leukemia (ALL). Despite its efficacy to improve the outlook of cancer patients, doxorubicin is known to cause a wide spectrum of cardiac complications that can seriously affect the quality of life of ALL survivors. Structural changes can occur in myocardial tissue after exposure to doxorubicin-based chemotherapy. These changes are frequently accompanied by alterations in left ventricular (LV) geometry and function. Myocardial damage can progress silently for many years before the manifestation of clinical symptoms and become apparent even long time after completion of treatment. Higher cumulative doses of this agent increase the risk for late cardiac complications. The feasibility of CMR imaging has been established and a variety of software tools for 3D geometric modeling of the heart have been developed to assess ejection fraction, wall thicknesses, end-systolic and end-diastolic volumes. Finite element (FE) modelling and inverse material parameter identification strategies were then introduced to take into account the passive mechanical behavior of the myocardial tissue. In light of the above, we undertook a study to assess the asymptomatic changes in LV structure and function in a group of long-term survivors of childhood ALL treated with low to moderate doses of doxorubicin therapy. Given the high potential of numerical cardiac modeling, this evaluation was conducted using a FE model-based approach that integrates the subject-specific shape and motion of the ventricular chamber directly from imaging data. Eighty-four ALL survivors (23±7 years old) were prospectively enrolled and stratified into two different groups, designated as standard-risk (SR, n=19) and high-risk groups, according to their risk of tumor recurrence. Subjects treated for high-risk ALL were further divided into two groups depending upon whether they did (HRdex, n=36) or did not (HR, n=45) receive the protective therapy (dexrazoxane) in an attempt to reduce the likelihood of late cardiotoxicity. Furthermore, for purposes of comparison, 10 healthy volunteers (HV, 22±4 years), with no prior history of cancer or cardiac pathologies and similar in age to the survivors, were used as controls. As a part of the investigation of late-onset cardiotoxicity, all subjects underwent CMR imaging, transthoracic echocardiography, and exercise stress testing. From the acquired CMR data, a trained operator extracted the subject-specific shape and motion of the LV using a guide-point framework (CIM v8.1, University of Auckland, New Zealand). The inner and outer borders of the LV walls were semi-automatically drawn from six guidepoints placed by the operator at end-systole and then manually corrected for mis-registration errors. From this tracing, ejection fraction, stroke volume, myocardial mass, wall thickness, end-diastolic and end-systolic volumes were computed for each of the study participants. After that, inter- and intraobserver repeatability of the segmentation results were quantified by intra-class coefficients (ICC) on four reconstructions of 15 leukemia survivors, each by four trained operators, and on three reconstructions of the same subject, each by a single operator. For a subset of the leukemia survivor population (n=48), subject-specific data about LV wall motion and geometry at diastasis were incorporated into a model of the passive LV mechanics and used to estimate the global stiffness parameter C1 by means of an inverse materials parameters identification strategy. This biomechanical parameter was initially calculated by assuming physiologically realistic values of 0.75, 1.0, and 1.25 kPa as pressure loading constraints for all participants. Once the subject-specific LV pressures (at rest and during peak exercise) became available, we incorporated such data in the model and repeated the FE analysis. The resulting values of C1 were reported and compared between groups for each of the five loading scenarios. Statistical comparisons were performed by one-way analysis of variance (ANOVA) on global and regional geometrical parameters, and two-way repeated-measures ANOVA on time-dependent geometrical parameters. Ultimately, a sensitivity analysis was conducted to evaluate the dependence of the hyperelastic property on the diastasis frame and the pressure load. In our experience, inter-observer repeatability was good for regional geometrical parameters (ICC, 0.60-0.74) and excellent for global geometrical parameters (ICC, 0.75-1.00), while intra-observer repeatability was excellent for both regional and global parameters (ICC, 0.75-1.00). Groups had similar LV function values. No significant differences were observed among the four study groups in ejection fraction, stroke volume, mass, end-diastolic or end-systolic volumes. Some differences were detected in epicardial volume only between end-systole and diastasis phases (p<0.01) among the HRdex and HV or SR groups, and among the HV and HR groups. Interestingly, the hyperelastic property for standard pressures was slightly lower in the HR group when compared against the HRdex or SR group, and also when compared against the control group (p<0.05). In contrast, no appreciable difference could be noted in the hyperelastic property for intra-ventricular pressures at rest (p>0.5) and peak exercise (p>0.6). From this analysis, it is clear that the global and regional geometrical parameters are not sufficiently sensitive to capture the subtle changes induced by late doxorubicin cardiotoxicity in LV structure and function. Nevertheless, the time-dependent global parameters provided preliminary evidence that early diastole was more affected by doxorubicin exposure than systole or late diastole. The smaller hyperelastic property in the high-risk group suggested a myocardial tissue more prone to dilatation if increased intra-ventricular pressure is applied than in the other study groups. This finding is consistent with what has been observed in long-term survivors of adult leukemia treated with high-dose doxorubicin-based chemotherapy. It should be noted, however, that these estimates could account only for the geometrical effects of myocardial remodeling on the passive ventricular mechanics since the subject-specific LV cavity pressure was not included in these simulations. Similar results were obtained when applying intra-ventricular pressures at rest and peak exercise. In conclusion, this study demonstrated that the subclinical cardiotoxicity of doxorubicin can be non-invasively assessed through the mechanical behavior analysis of the LV on CMR images. Additional investigations will be necessary to confirm our results and draw a firm conclusion about the long-term prognostic significance of alterations in myocardial stiffness and their relationship with ALL risk status in this or similar cohort of survivors. In future works, we hope to include the FE simulation of the left ventricular mechanics during systole and early diastole in order to better capture the changes in stiffness across groups. To the same purpose, it might be convenient to increase the temporal resolution of the image data in both leukemia survivors and control subjects.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi-MARIANNA GAMBA.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
2.56 MB
Formato
Adobe PDF
|
2.56 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/149077