In the last decades, the research for the neurobiological bases of mental illnesses has involved multidisciplinary fields, including psychopathology, genetics, functional and structural brain imaging as well as their integration (e.g. imaging-genetics). Brain imaging has undoubtedly played a leading role in allowing the study of functional and structural aspects of brain structures in vivo, making use of numerous methods of image acquisition including the powerful and non-invasive Magnetic Resonance Imaging (MRI). MRI is a radiological technique that in medicine allows to study anatomical and functional aspects of the human body by exploiting the principle of Nuclear Magnetic Resonance. MRI represents the main method of investigation in brain imaging, with a history dating back to the early 1980s, when for the first time MRI studies were performed on patients suffering from bipolar disorder (Rangel- Guerra et al. ,1983) and on schizophrenic patients (Smith et al. ,1984), and which continues with constant technological improvements and increasingly complex fields of use. In the present experimental thesis, 3T MRI has been used to investigate the complex determinants of structural brain development. The MRI study involved a unique sample of 59 young couples of twins, of which 25 monozygotic (MZ) and 34 dizygotic (DZ) from Northern Italy. The experimental work concerned 1) the development of algorithms for the estimation of brain morphological measures and 2) the implementation of statistical methods for twin designs, with the ultimate goal of estimating the genetic and environmental impacts on a set of brain morphological measures during neurodevelopment. To this end, sophisticated region-based and voxel-based neuroanatomical analysis techniques were applied, which allowed to extract the morphological measures of interest. These measures (phenotypes) were then compared within each pair of twins and between MZ and DZ couples to estimate genetic and environmental effects. The structural MRI images were processed using powerful and widely used neuroimaging softwares, FreeSurfer and SPM 12. The first branch of investigation involved, through a region-based approach, the estimation of heritability, i.e. the portion of inter-individual variance linked to genetic differences, of regional neuroanatomical parameters. Furthermore, the univariate ACE twin model was developed to identify the extent to which genetic factors and environmental factors (shared and unique) influenced the investigated parameter, respectively. The neuroanatomical similarity / diversity of twin couples was investigated using an alternative approach to the classical method, based on multiple linear regression. This phase of analysis has involved the creation of the multiple generalized linear regression model in order to evaluate the relative influence of zygosity, age and sex (independent variables) on intra-couple differences in terms of neuroanatomy (dependent variables). Furthermore, a complementary voxel-based approach was used for automatic classification of zygosity. This analysis involved the implementation of Machine Learning algorithms, in the software NeuroMiner, capable of predicting the twins' zygosity starting from the intra-couple punctual difference (voxel by voxel) of gray and white matter tissue maps. This method identifies the brain clusters that can help differentiate between MZ and DZ twins based on the similarity in tissue volume, thus providing information on genetic influences on the local volumes of gray and white matter. Such investigations represent an additional piece to the understanding of the relative role of genes and environment on intra-couple neuroanatomical similarity during development. Another research topic of the thesis regarded the preliminary identification of brain regional markers of psychotic risk. Indeed, twins represent the ideal sample for exploring genetic and environmental effects on such markers. Information concerning the psychotic risk of each subject was obtained on the basis of the ERIraos-CL scale. We identified 13 couples discordant for psychotic risk and selected 13 age-matched singletons, chosen from the twin pairs without psychotic risk, considered as healthy controls. Discordant twins and healthy subjects were compared in terms of neuroanatomical parameters in order to 1) identify the brain regions characterized by anatomical alterations due to the risk of manifesting psychosis, 2) speculating about the genetic / environmental determinants of these alterations. Algorithms have been created to carry out the following statistical comparisons for each measure: (1) comparison between high-risk twins and low-risk twins of discordant couples (2) comparison between high-risk twins and control subjects (3) comparison between low-risk twins and control subjects. Heritability analyses showed that the regional volume is significantly heritable (p-value <0.05) in a larger number of brain regions (equal to 19) compared to the other morphological parameters analyzed. Therefore, it is reasonable to hypothesize that volume is the morphological measure most widely influenced by genetic factors. On the contrary, the mean curvature emerged as the least heritable measure, in fact it is significantly inherited only in three cortical regions (frontal pole of the left hemisphere, lingual cortex of the right hemisphere and triangular part of the left inferior frontal gyrus). Thus, except in these three brain areas, it is likely that the mean curvature is less influenced by genetic factors and more by environmental factors. Furthermore, correlation analysis for surface area, cortical thickness and volume parameters showed a significant correlation in both groups of zygosity greater in the brain regions of the right hemisphere than the brain regions of the left hemisphere, while regarding mean curvature correlation analysis showed for MZ pairs a greater correlation in brain areas of the right hemisphere than left hemisphere and for DZ pairs a greater correlation in brain areas of the left hemisphere than the right one. The ACE model highlighted a contribution of common environment, shared by the twins on the volume of many subcortical regions, which unlike the cortical regions seem to be less affected by genetic and unique environmental factors. The results obtained by the different approaches show that the volume of white matter in the cerebellar cortex is defined by genetic factors rather than by environmental factors. In addition, the surface area value of the mean temporal gyrus of the left hemisphere is also largely attributable to genetic factors. Furthermore, the analyses revealed an important aspect regarding the impact of development on intra-couple similarity. In fact, irrespective of the zygosity of the couple, some brain regions showed increased neuroanatomical similarity with age, other brain regions showed reduced similarity with age. This result could presumably be due to genetic or environmental factors that come into play during growth and do influence the twins in a similar way causing greater similarity, or influence them differently, accentuating diversity. The preliminary analyses on neuroanatomical alterations in the presence of psychotic risk allowed to identify the mean curvature and the volume as the parameters most affected by alterations in the presence of nascent psychosis. Given that for these two parameters heritability scores were found to be low and high respectively, this result corroborates the thesis that psychopathology is largely influenced by both environmental and genetic factors. Indeed, it has been observed that in at-risk twins these parameters have higher values than in the control twins. Differences in the values were evident both in the twins that are considered to be at psychotic risk for the ERIraos-CL scale and in their siblings who form each discordant couple. The obtained results could represent neurobiological markers for the risk of developing psychosis, allowing the early recognition of subjects at risk and the planning of adequate preventive strategies. Regarding the analyses implemented in NeuroMiner (a Machine Learning software), from the comparison between the results obtained using the intra-couple punctual difference (voxel by voxel) of gray and white matter tissue maps emerges a higher accuracy of the classifier that bases the prediction of zygosity on the punctual difference of white matter tissue maps than the classifier that bases the prediction of zygosity on the punctual difference of gray matter tissue maps. The classifier appeared able to identify similarities / differences in white matter values and to attribute the correct zygosity to new images. The complementary voxel-based classification analysis suggested the presence of significant genetic influences in the local white matter volumes, on the basis of which it was possible to predict the zygosity of a new pair. In particular, the volume of white matter in the cerebellar cortex was more similar in the MZ twins than in the DZ twins and determinant in the prediction. The results obtained in this thesis work provide quantitative information regarding the extent to which genes and environment influence brain morphology in the delicate phase of development. This discrimination is of fundamental importance in the clinical context, for the understanding of pathologies associated with brain morphological alterations. The presence of a relevant environmental impact on the expression of the parameter can be useful for the development of preventive, personalized and innovative therapeutic strategies.
Negli ultimi decenni la ricerca per le basi neurobiologiche sulle malattie mentali ha coinvolto ambiti multidisciplinari, che comprendono la psicopatologia, la genetica, il brain imaging funzionale e strutturale e la loro integrazione (ad esempio, il genetic imaging). Il brain imaging ha avuto senza dubbio un ruolo di prim’ordine consentendo lo studio di aspetti funzionali e strutturali di strutture cerebrali in vivo avvalendosi di numerosi metodi di acquisizione delle immagini tra cui il potente e non invasivo Imaging a Risonanza Magnetica (MRI). L’MRI è una tecnica radiologica che in medicina consente di studiare aspetti anatomici e funzionali del corpo umano sfruttando il principio della Risonanza Magnetica Nucleare. L’MRI rappresenta il principale metodo di indagine in brain imaging, con una storia che risale agli inizi degli anni ’80, quando per la prima volta furono eseguiti studi MRI su pazienti affetti da disturbo bipolare (Rangel- Guerra et al. ,1983) e su pazienti schizofrenici (Smith et al. ,1984), e che continua con costanti miglioramenti tecnologici e campi di utilizzo sempre più complessi. Nella presente tesi sperimentale, l’MRI a 3T è stata utilizzata per investigare i complessi determinanti dello sviluppo della struttura celebrale. Lo studio MRI ha coinvolto un campione unico composto da ben 59 giovani coppie di gemelli, dei quali 25 monozigoti (MZ) e 34 dizigoti (DZ) provenienti dal Nord Italia. Il lavoro sperimentale ha riguardato 1) l’implementazione di algoritmi per la stima di parametri morfologici celebrali e 2) l’implementazione di metodi statistici gemellari, con l’obiettivo ultimo di stimare l’impatto genetico ed ambientale su un certo numero di parametri morfologici celebrali nel neurosviluppo. A tale scopo sono state applicate sofisticate tecniche di analisi neuroanatomica region-based e voxel-based, che hanno consentito di estrarre le misure morfologiche di interesse. Queste misure (fenotipi) sono state poi confrontate in ogni coppia di gemelli e tra coppie di gemelli MZ e DZ per stimare gli effetti genetici ed ambientali. Le immagini MRI strutturali sono state analizzate usando software ampiamente utilizzati in neuroimaging quali FreeSurfer e SPM 12. Il primo ramo di indagine ha previsto, mediante un approccio region-based, la stima dell’ereditabilità, intesa come porzione di varianza inter-individuale legata a differenze genetiche, di parametri neuroanatomici regionali. Inoltre, è stato sviluppato il modello gemellare univariato ACE al fine di identificare in che misura, rispettivamente, i fattori genetici e i fattori ambientali (condivisi e unici) influenzino i valori del parametro investigato. La similarità / diversità neuroanatomica delle coppie è stata analizzata applicando un approccio regionale alternativo al metodo classico, basato sulla regressione lineare multipla. Questa fase di analisi ha previsto la creazione del modello di regressione lineare generalizzata multipla per valutare l’influenza relativa delle variabili zigosità, età e sesso (variabili indipendenti) sulle differenze intra-coppia in termini di neuroanatomia (variabili dipendenti). Inoltre, è stato usato un approccio complementare voxel-based per la classificazione automatica della zigosità. Questa analisi ha previsto l’implementazione di algoritmi di Machine Learning, nel software NeuroMiner, in grado di predire la zigosità dei gemelli a partire da un’immagine ottenuta dalla differenza puntuale (voxel per voxel) intra-coppia delle mappe tessutali di materia grigia e bianca. Tale metodo identifica cluster celebrali che svolgono un ruolo determinante nella differenziazione tra gemelli MZ e DZ basandosi sulla similarità nei volumi locali di materia grigia e bianca, quindi forniscono informazioni sulle influenze genetiche su tali volumi. Tali indagini rappresentano un tassello aggiuntivo alla comprensione del ruolo relativo di geni e ambiente sulla similarità neuroanatomica intra-coppia durante lo sviluppo. Un altro tema di ricerca di questa tesi ha riguardato l’identificazione preliminare di marker celebrali di rischio psicotico. Infatti, i gemelli rappresentano il campione ideale per esplorare gli effetti genetici e ambientali su tali marker. Le informazioni riguardanti il rischio psicotico di ciascun soggetto sono state ottenute sulla base della scala ERIraos-CL. Sono state identificate 13 coppie discordanti per rischio psicotico e 13 gemelli di età comparabile, scelti tra le coppie concordanti per l’assenza di rischio, considerati come soggetti di controllo. I gemelli discordanti e i soggetti di controllo sono stati confrontati in termini di parametri neuroanatomici per 1) identificare le regioni celebrali caratterizzate da alterazioni anatomiche dovute al rischio di manifestare psicosi, 2) fare ipotesi circa i determinanti genetici / ambientali di queste alterazioni. Sono stati sviluppati algoritmi per effettuare i seguenti confronti statistici per ogni parametro: (1) Confronto tra i valori nei gemelli ad alto rischio e nei gemelli a basso rischio delle coppie discordanti (2) Confronto tra i valori nei gemelli ad alto rischio e nei soggetti di controllo (3) Confronto tra i valori nei gemelli a basso rischio e nei soggetti di controllo. Le analisi di ereditabilità hanno mostrato che il volume regionale è significativamente ereditabile (p-value<0.05) in un numero maggiore di regioni celebrali (pari a 19) rispetto agli altri parametri morfologici analizzati. Pertanto, è ragionevole ipotizzare che il volume sia il parametro morfologico più largamente influenzato da fattori genetici. Al contrario, la curvatura media è risultato essere il parametro meno ereditabile infatti è significativamente (p-value<0.05) ereditabile solo in tre regioni corticali (polo frontale dell’emisfero sinistro, corteccia linguale dell’emisfero destro e parte triangolare della circonvoluzione frontale inferiore sinistra). Infatti, eccetto che in queste tre aree celebrali, è presumibile che la curvatura media assuma valori meno influenzati da fattori genetici e maggiormente da fattori ambientali. Inoltre, l’analisi di correlazione per i parametri area superficiale, spessore corticale e volume, ha mostrato una correlazione significativa in entrambi i gruppi di zigosità maggiore nelle regioni celebrali dell’emisfero destro rispetto alle regioni celebrali dell’emisfero sinistro, mentre l’analisi di correlazione della curvatura media ha mostrato per i MZ una maggiore correlazione nelle aree celebrali dell’emisfero destro rispetto all’emisfero sinistro e per i DZ una maggiore correlazione nelle aree celebrali dell’emisfero sinistro rispetto al destro. Il modello ACE ha messo in luce un contributo dell’ambiente condiviso tra i gemelli sul volume di molte regioni sottocorticali che, diversamente dalle regioni corticali, sembrano essere meno affette da fattori genetici e dai fattori ambientali unici. I risultati ottenuti mediante i diversi approcci mostrano che il volume di materia bianca nella corteccia cerebellare è definito dal patrimonio genetico più che da fattori ambientali. In aggiunta, anche il valore di area superficiale della circonvoluzione temporale media dell’emisfero sinistro è attribuibile in larga parte a fattori genetici. Inoltre, le analisi hanno messo in luce un aspetto importante riguardo l’impatto dello sviluppo sulla similarità intra-coppia. Infatti, indipendentemente dalla zigosità della coppia, alcune regioni celebrali hanno mostrato un aumento della similarità neuroanatomica con l’età, altre regioni celebrali hanno mostrato una riduzione della similarità con l’età. Questo risultato, potrebbe presumibilmente essere attribuibile a fattori di tipo genetico oppure a fattori di tipo ambientale che entrano in gioco durante la crescita che potrebbero condizionare i gemelli allo stesso modo causando una maggiore similarità, oppure condizionarli in maniera differente accentuandone la diversità. Le analisi preliminari sulle alterazioni neuroanatomiche in presenza di rischio psicotico, hanno permesso di individuare la curvatura media ed il volume come parametri maggiormente affetti da alterazioni in presenza di psicosi nascente. Dato che per questi due parametri i punteggi di ereditabilità sono risultati essere rispettivamente bassi e alti, questo risultato avvalora la tesi che la psicopatologia è largamente influenzata sia da fattori ambientali che da fattori genetici. Infatti, è stato osservato che nei gemelli a rischio, questi parametri assumono valori maggiori rispetto ai valori nei gemelli di controllo. Le anomalie nei valori sono risultate evidenti sia nei gemelli che sono considerati a rischio psicotico per la scala ERIraos-CL che nei fratelli gemelli di questi soggetti che costituiscono ciascuna coppia discordante. I risultati ottenuti potrebbero rappresentare marker neurobiologici per il rischio di sviluppare psicosi, permettendo il riconoscimento precoce di soggetti a rischio e la pianificazione di strategie preventive adeguate. Nell’ambito delle analisi implementate nel software di Machine Learning NeuroMiner, dal confronto tra i risultati ottenuti utilizzando le immagini di differenza puntuale intra-coppia di mappe tessutali di materia grigia e di materia bianca emerge un’accuratezza maggiore del classificatore che basa la predizione di zigosità sulle immagini di differenza puntuale intra-coppia di mappe tessutali di materia bianca. Il classificatore è apparso in grado di individuare similarità / differenze nei valori di materia bianca e di attribuire a nuove immagini la corretta zigosità. Questa complementare analisi di classificazione di tipo voxel-based ha suggerito la presenza di influenze genetiche significative nei volumi locali di materia bianca, sulla base dei quali è stato possibile predire la zigosità di una nuova coppia. In particolare, il volume di materia bianca nella corteccia cerebellare è risultato più simile nei gemelli MZ rispetto ai gemelli DZ e determinante nella predizione. I risultati ottenuti nel presente lavoro di tesi forniscono informazioni di tipo quantitativo riguardanti la misura in cui geni e ambiente influenzano la morfologia nella delicata fase dello sviluppo celebrale. Tale discriminazione è di fondamentale importanza in ambito clinico, per la comprensione di patologie associate ad alterazioni morfologiche celebrali. La presenza di un impatto ambientale rilevante sull’espressione del parametro può essere utile per lo sviluppo di strategie terapeutiche preventive, personalizzate e innovative.
Metodi statistici di analisi dell'impatto genetico e ambientale sul neurosviluppo. Studio MRI a 3T su coppie di gemelli
AMBROSINO, ADRIANA
2018/2019
Abstract
In the last decades, the research for the neurobiological bases of mental illnesses has involved multidisciplinary fields, including psychopathology, genetics, functional and structural brain imaging as well as their integration (e.g. imaging-genetics). Brain imaging has undoubtedly played a leading role in allowing the study of functional and structural aspects of brain structures in vivo, making use of numerous methods of image acquisition including the powerful and non-invasive Magnetic Resonance Imaging (MRI). MRI is a radiological technique that in medicine allows to study anatomical and functional aspects of the human body by exploiting the principle of Nuclear Magnetic Resonance. MRI represents the main method of investigation in brain imaging, with a history dating back to the early 1980s, when for the first time MRI studies were performed on patients suffering from bipolar disorder (Rangel- Guerra et al. ,1983) and on schizophrenic patients (Smith et al. ,1984), and which continues with constant technological improvements and increasingly complex fields of use. In the present experimental thesis, 3T MRI has been used to investigate the complex determinants of structural brain development. The MRI study involved a unique sample of 59 young couples of twins, of which 25 monozygotic (MZ) and 34 dizygotic (DZ) from Northern Italy. The experimental work concerned 1) the development of algorithms for the estimation of brain morphological measures and 2) the implementation of statistical methods for twin designs, with the ultimate goal of estimating the genetic and environmental impacts on a set of brain morphological measures during neurodevelopment. To this end, sophisticated region-based and voxel-based neuroanatomical analysis techniques were applied, which allowed to extract the morphological measures of interest. These measures (phenotypes) were then compared within each pair of twins and between MZ and DZ couples to estimate genetic and environmental effects. The structural MRI images were processed using powerful and widely used neuroimaging softwares, FreeSurfer and SPM 12. The first branch of investigation involved, through a region-based approach, the estimation of heritability, i.e. the portion of inter-individual variance linked to genetic differences, of regional neuroanatomical parameters. Furthermore, the univariate ACE twin model was developed to identify the extent to which genetic factors and environmental factors (shared and unique) influenced the investigated parameter, respectively. The neuroanatomical similarity / diversity of twin couples was investigated using an alternative approach to the classical method, based on multiple linear regression. This phase of analysis has involved the creation of the multiple generalized linear regression model in order to evaluate the relative influence of zygosity, age and sex (independent variables) on intra-couple differences in terms of neuroanatomy (dependent variables). Furthermore, a complementary voxel-based approach was used for automatic classification of zygosity. This analysis involved the implementation of Machine Learning algorithms, in the software NeuroMiner, capable of predicting the twins' zygosity starting from the intra-couple punctual difference (voxel by voxel) of gray and white matter tissue maps. This method identifies the brain clusters that can help differentiate between MZ and DZ twins based on the similarity in tissue volume, thus providing information on genetic influences on the local volumes of gray and white matter. Such investigations represent an additional piece to the understanding of the relative role of genes and environment on intra-couple neuroanatomical similarity during development. Another research topic of the thesis regarded the preliminary identification of brain regional markers of psychotic risk. Indeed, twins represent the ideal sample for exploring genetic and environmental effects on such markers. Information concerning the psychotic risk of each subject was obtained on the basis of the ERIraos-CL scale. We identified 13 couples discordant for psychotic risk and selected 13 age-matched singletons, chosen from the twin pairs without psychotic risk, considered as healthy controls. Discordant twins and healthy subjects were compared in terms of neuroanatomical parameters in order to 1) identify the brain regions characterized by anatomical alterations due to the risk of manifesting psychosis, 2) speculating about the genetic / environmental determinants of these alterations. Algorithms have been created to carry out the following statistical comparisons for each measure: (1) comparison between high-risk twins and low-risk twins of discordant couples (2) comparison between high-risk twins and control subjects (3) comparison between low-risk twins and control subjects. Heritability analyses showed that the regional volume is significantly heritable (p-value <0.05) in a larger number of brain regions (equal to 19) compared to the other morphological parameters analyzed. Therefore, it is reasonable to hypothesize that volume is the morphological measure most widely influenced by genetic factors. On the contrary, the mean curvature emerged as the least heritable measure, in fact it is significantly inherited only in three cortical regions (frontal pole of the left hemisphere, lingual cortex of the right hemisphere and triangular part of the left inferior frontal gyrus). Thus, except in these three brain areas, it is likely that the mean curvature is less influenced by genetic factors and more by environmental factors. Furthermore, correlation analysis for surface area, cortical thickness and volume parameters showed a significant correlation in both groups of zygosity greater in the brain regions of the right hemisphere than the brain regions of the left hemisphere, while regarding mean curvature correlation analysis showed for MZ pairs a greater correlation in brain areas of the right hemisphere than left hemisphere and for DZ pairs a greater correlation in brain areas of the left hemisphere than the right one. The ACE model highlighted a contribution of common environment, shared by the twins on the volume of many subcortical regions, which unlike the cortical regions seem to be less affected by genetic and unique environmental factors. The results obtained by the different approaches show that the volume of white matter in the cerebellar cortex is defined by genetic factors rather than by environmental factors. In addition, the surface area value of the mean temporal gyrus of the left hemisphere is also largely attributable to genetic factors. Furthermore, the analyses revealed an important aspect regarding the impact of development on intra-couple similarity. In fact, irrespective of the zygosity of the couple, some brain regions showed increased neuroanatomical similarity with age, other brain regions showed reduced similarity with age. This result could presumably be due to genetic or environmental factors that come into play during growth and do influence the twins in a similar way causing greater similarity, or influence them differently, accentuating diversity. The preliminary analyses on neuroanatomical alterations in the presence of psychotic risk allowed to identify the mean curvature and the volume as the parameters most affected by alterations in the presence of nascent psychosis. Given that for these two parameters heritability scores were found to be low and high respectively, this result corroborates the thesis that psychopathology is largely influenced by both environmental and genetic factors. Indeed, it has been observed that in at-risk twins these parameters have higher values than in the control twins. Differences in the values were evident both in the twins that are considered to be at psychotic risk for the ERIraos-CL scale and in their siblings who form each discordant couple. The obtained results could represent neurobiological markers for the risk of developing psychosis, allowing the early recognition of subjects at risk and the planning of adequate preventive strategies. Regarding the analyses implemented in NeuroMiner (a Machine Learning software), from the comparison between the results obtained using the intra-couple punctual difference (voxel by voxel) of gray and white matter tissue maps emerges a higher accuracy of the classifier that bases the prediction of zygosity on the punctual difference of white matter tissue maps than the classifier that bases the prediction of zygosity on the punctual difference of gray matter tissue maps. The classifier appeared able to identify similarities / differences in white matter values and to attribute the correct zygosity to new images. The complementary voxel-based classification analysis suggested the presence of significant genetic influences in the local white matter volumes, on the basis of which it was possible to predict the zygosity of a new pair. In particular, the volume of white matter in the cerebellar cortex was more similar in the MZ twins than in the DZ twins and determinant in the prediction. The results obtained in this thesis work provide quantitative information regarding the extent to which genes and environment influence brain morphology in the delicate phase of development. This discrimination is of fundamental importance in the clinical context, for the understanding of pathologies associated with brain morphological alterations. The presence of a relevant environmental impact on the expression of the parameter can be useful for the development of preventive, personalized and innovative therapeutic strategies.File | Dimensione | Formato | |
---|---|---|---|
Tesi_di_Laurea_Magistrale_ADRIANA_AMBROSINO-finale.pdf
accessibile in internet per tutti
Dimensione
4.69 MB
Formato
Adobe PDF
|
4.69 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/10589/151030