Whole-brain structural connectivity predicts cognitive deficits in pretreated glioma patients

doi:10.21203/rs.3.rs-3135042/v1

Download PDF

Case Report

Whole-brain structural connectivity predicts cognitive deficits in pretreated glioma patients

https://doi.org/10.21203/rs.3.rs-3135042/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Purpose

Glioma patients frequently suffer from cognitive dysfunction potentially caused by tumor invasion or treatment effects. We hypothesized that cognitive functioning in pretreated glioma patients critically depends on the maintained structural connectivity of multiple brain networks.

Methods

The study included 121 pretreated glioma patients (median age, 52 years; median ECOG score 1; CNS WHO grade 3 or 4) who had biopsy or resection plus chemoradiation. Cognitive performance was assessed by ten tests in five main cognitive domains 14 (1-214) months after therapy initiation. Hybrid amino acid PET/MRI using the tracer O-(2-[¹⁸F]fluoroethyl)-L-tyrosine, a network-based cortical parcellation, and advanced tractography methods were used to generate whole-brain fiber count-weighted connectivity matrices. The matrices were applied to cross-validated machine learning-based models to identify fiber connections and underlying networks predictive for cognitive performance in the evaluated domains.

Results

Compared to healthy controls (n = 121), the cognitive scores were significantly lower in nine cognitive tests. The models well predicted 7/10 scores (median correlation coefficient, 0.47; range, 0.39–0.57) from 24–230 (0.5–4.6%) of the possible connections/edges, 84% were between nodes of different networks. Critically involved cortical regions/nodes (≥ 10 adjacent predictive edges) included predominantly left-sided nodes of the visual, somatomotor, dorsal/ventral attention, and default mode networks. Highly critical nodes (≥ 15–20 edges) included the default-mode network’s left temporal and bilateral posterior cingulate cortex.

Conclusions

These results suggest that the cognitive performance of pretreated glioma patients is strongly related to structural connectivity between multiple brain networks and depends on the integrity of known network hubs also involved in other neurological disorders.

glioma

PET/MR hybrid imaging

structural connectivity

brain networks

cognitive functions

The survival prognosis of glioma patients mainly depends on molecular and histomorphologic tumor features that determine the growth rate and pattern of invasion into the normal brain tissue [1, 2]. After therapy, patients often experience disturbances in cognitive function [3–6], potentially related to the structural damage imposed on the brain by the tumor itself or to therapeutic interventions. The pattern of brain lesions in glioma patients is heterogeneous, as it comprises brain infiltration by residual or recurrent tumor growth, neurosurgical resection of cortical or subcortical brain tissue, and radiation- or chemotherapy-induced damage of white matter [7]. To date, basic and clinical research in neuro-oncology has primarily targeted identifying selected structures at risk, such as the language pathways (neurosurgery) [8] or the hippocampus (radiation therapy) [9]. However, a growing body of evidence indicates that the long-term outcome of higher-order cognitive functioning in glioma patients depends on the preserved integrity of several distributed brain networks rather than individual cortical regions or white-matter tracts [8, 10–12].

We here hypothesized that the leading cause of cognitive deterioration in glioma patients is the damage of cortical regions and their adjacent white matter fiber tracts. Therefore, whole-brain structural connectomes were constructed in a group of pretreated patients with World Health Organization (WHO) Central Nervous System (CNS) grade 3 or 4 gliomas. Advanced structural and diffusion-weighted MRI [13] as well as amino acid PET [14] imaging techniques were used to identify fiber tracts and structural brain lesions. In addition, a recent functional parcellation of the cortex [15] and tractography tools capable of fiber tracking within or close to tumor- or treatment-related lesions were applied [16, 17]. The individual connectomes were used to develop a predictive machine learning-based model [18] that identified brain networks, cortical regions, and connecting fiber tracts critical for performance in different cognitive domains.

Patient characteristics

From February 2018 to September 2020, we prospectively evaluated 121 pretreated glioma patients (73 males, 48 females; mean age, 51.6 ± 11.6 years) who had undergone multimodal therapy, including resection, radiotherapy, alkylating chemotherapy, or combinations thereof (Table S1 (Online Resource 1)). Patients were referred for 3T hybrid PET/MR imaging using the radiolabeled amino acid O-(2-[¹⁸F]fluoroethyl)-L-tyrosine (¹⁸F-FET) at different time points after completion of first-line therapy (median time, 14 months; range, 1-214 months). The referrals were made to assess residual metabolic active tumor [19] after surgery, evaluate response to adjuvant chemotherapy, or to differentiate tumor relapse from treatment-related changes.

The inclusion criteria comprised an ECOG (Eastern Cooperative Oncology Group) [20] performance score of 0–1, absence of major depression, no seizures at presentation, and fluency in the German language. Patients were screened and registered for the study by phone calls, reviewed on the day of imaging, and were included in the study after providing informed written consent according to the Declaration of Helsinki. The local ethics committee approved the protocol (17–365). As detailed in Table S1 (Online Resource 1), patients had either received tumor resection (n = 108) or biopsy (n = 13), and the majority (n = 100) had undergone at least one series of local radiotherapy (60 ± 2 Gy in 92%) at a median interval of 13 months (range, 2-213 months) between the start of radiotherapy and imaging. Fourteen patients had two radiotherapy series. According to the 2016 WHO classification [1], n = 81 patients had a CNS WHO grade 4 glioblastoma, n = 28 patients a CNS WHO grade 3 astrocytoma, and n = 12 patients a CNS WHO grade 3 oligodendroglioma. A total of 81 patients (67%) had mild neurological (48%) or other symptoms (19%) without requiring assistance for personal needs. All patients except one were right-handed. Based on clinical status, serial MRI findings, and ¹⁸F-FET PET results, the diagnosis of a glioma residual or relapse was made in 58 of 121 patients.

Hybrid PET/MR imaging

Simultaneous PET/MR imaging was performed on a 3T hybrid PET/MR scanner (Siemens Trim-TRIO/BrainPET, Siemens Medical Systems, Erlangen, Germany) equipped with a couch-mounted birdcage-like quadrature transmitter head coil and an 8-channel receiver coil. The PET insert consisted of 72 rings (axial field-of-view, 19.2 cm; center spatial resolution, 3 mm FWHM). The PET data were corrected for random and scatter coincidences, dead time, attenuation (based on a T1-weighted anatomical MRI scan), as well as motion artifacts and reconstructed by OPOSEM (ordered Poisson ordinary subset expectation maximization) reconstruction (2 subsets, 32 iterations) [21]. All patients underwent a dynamic PET scan from 0 to 50 min post-injection of 3 MBq of ¹⁸F-FET per kg of body. For clinical purposes, the summed activity from 20–40 min post-injection and the time-activity curves were evaluated according to established protocols [14].

The MRI protocol [13] comprised a 3D high-resolution T1-weighted magnetization-prepared rapid acquisition gradient echo (MPRAGE) native scan, a contrast-enhanced MPRAGE scan recorded after injection of gadolinium-based contrast agent, a T2-weighted sampling perfection with application-optimized contrasts using different flip angle evolution (SPACE) scan and a T2-weighted fluid-attenuated inversion recovery (T2/FLAIR) scan. High-angular-resolution diffusion imaging (HARDI) measurements were acquired using a diffusion-weighted double-echo echo-planar imaging (EPI) sequence (55 slices; TR = 8000 ms; TE = 112 ms; b-values = 0, 13 interleaved volumes and b = 2700 s/mm², 120 gradient directions; voxel size = 2.4×2.4×2.4 mm³). Afterward, a non-diffusion weighted (b = 0) volume was acquired with a reversed phase-encoding direction needed for the EPI distortion correction.

Cognitive performance

Cognitive performance was assessed on the day of imaging and based on ten cognitive tests selected from a more extensive test battery developed for the 1000BRAINS study of healthy subjects [13]. This set of tests could be finished within 20–30 minutes by most of the patients.

As shown in Table S2 (Online Resource 1), the Trail-Making tests A and B were used to asses processing speed/attention and executive function/concept shifting [22]. Semantic word fluency (imagined shopping tour), language processing (number transcoding test), verbal working memory (digit span forward/backward) and verbal episodic memory (word list immediate/delayed recall) were tested by the DemTect dementia screening instrument [23]. Visual-spatial working memory was tested with the Corsi Block-Tapping test (forward/backward) [24]. All scores were compared to those of a group of n = 121 healthy subjects [13] that were matched for age, gender, and educational level (http://uis.unesco.org/sites/default/files/documents/international-standard-classification-of-education-1997-en_0.pdf).

Whole-brain structural connectome

The main steps for determining the whole-brain structural connectome and prediction modeling of cognitive performance are depicted in Fig. 1. We used the tractography imaging pipeline based on the GitHub-fork MRtrix3Tissue (https://3tissue.github.io), a recently developed modification of the fiber-tracking software MRtrix3 (https://www.mrtrix.org) based on constrained spherical deconvolution (CSD) [25]. CSD-based fiber mapping assumes that the diffusion-weighted MRI signal results from the spherical convolution of a response function with the underlying FOD function [26]. The novel single-shell 3-tissue constrained spherical deconvolution (SS3T-CSD) method generates estimates of white matter fiber orientation distribution functions (FODs) as bias-free as possible, even within different compartments infiltrated by the tumor [16, 27].

The HARDI data underwent image preprocessing following published recommendations (https://osf.io/ht7zv) and comprised corrections for EPI distortion, eddy current, motion distortion, and bias field. Fiber tracking was based on Anatomically-Constrained Tractography (ACT), which poses physiological restrictions on the behavior of healthy neuronal fibers in terms of their propagation and termination [17]. These assumptions were lifted in the area of pathologic tissue by masking out the entire lesioned areas in order to allow detection of any residual fiber connections in the structurally altered brain tissue. Resection cavities were manually contoured by a radiation oncologist (M.K.), the T1-contrast-enhancing lesions and T2/FLAIR hyperintensities were automatically segmented using the deep-learning-based software HD-GLIO-AUTO (https://github.com/NeuroAI-HD/HD-GLIO-AUTO), and ¹⁸F-FET PET segmentation was implemented by an FSL (https://fsl.fmrib.ox.ac.uk) custom script using a tumor-to-brain ratio (TBR) of 1.6 (20–40 minutes summed activity) as the lower threshold [28]. All segmentations were visually inspected, manually corrected, and added to form a composite lesion mask.

Functional cortical regions (= nodes) were defined using the Schaefer-Yeo Atlas [15] which comprises 100 nodes (50 in each hemisphere) belonging to seven networks: visual, somatomotor, dorsal attention, ventral attention, limbic, frontal control, and default mode network. The number of fibers connecting any two cortical regions (= weighted edge) were entered into a 100x100 structural connectivity matrix, resulting in 4950 possible edges (symmetric matrix, no self-connection of nodes), see Figs. 1 and 2.

Connectome-based predictive modeling

The relationship between structural brain connectivity and cognitive functions was analyzed by an established method for connectome-based predictive modeling (CPM) [18]. All steps were performed in Matlab (Matlab R2022a, MathWorks, Natick, MA, USA). The CPM protocol (Fig. 1) comprises four parts. For feature selection (i), each fiber count (edge) in the connectivity matrix was related to any of the cognitive test scores using Spearman's rank correlation, and only significant (p < 0.001) edges were selected. Next, summary connectivity values (ii) were calculated from the selected edges by separately summing the edge values of negative and positive signs.

For model construction (iii), linear regressions between the cognitive scores and the summary connectivity scores were computed. Finally, the model’s generalizability and predictive power (iv) were evaluated by leave-one-out cross-validation and permutation analysis. In addition, binary matrices were constructed for each test, labelling the predictive edges present in both 90% of the cross-validation models and in those models with significant (p < 0.05) predictive power (validated edges). From these binary matrices, the critical network nodes for each cognitive test were identified from their degree, i.e. the number of adjacent validated edges.

Cognitive outcome and fiber count distribution

The results of the cognitive test scores in glioma patients and healthy individuals are shown in Table S2 (Online Resource 1). With the exception of the number transcoding test, glioma patients performed significantly worse than healthy individuals on all tests, with relative performance losses ranging from 8–72%. As shown in Fig. 2, the majority of the fiber connections were located within the ipsilateral nodes of the different networks. The median nodal fiber count for the prevailing intra-hemispheric connections was lower in the affected than in the unaffected, contralateral hemisphere: median fiber number in left-sided lesions, 41 (range, 0-4035 fibers) vs. 52 (range, 0-4849 fibers); median fiber number in right-sided lesions, 40 (range, 0-4013 fibers) vs. 47 (range, 0-5265 fibers), representing an average fiber loss of 15–20% per node in the affected hemisphere.

Connectome-based predictive modeling

In the first step of the CPM analysis, 2770 edges (node-to-node fiber counts) with a significant correlation to any of the ten cognitive test scores (predictive edges) were identified. In the vast majority (98%), the sign of the correlation indicated a lower performance for lower fiber counts (Table S3, Online Resource 1). Therefore, these edges were exclusively used. A median number of 254 predictive edges (range, 32–542) was selected for linear regression modeling between the connectivity values and the cognitive performance scores. All linear models were significant (p < 0.001), and most of them had a coefficient of determination (R²) in the range between 0.35 and 0.45 (Tables S3, Fig. S1, Online Resource 1). The results for the correlation and permutation analyses (100 permutations) between the predicted scores from the leave-one-out cross-validation models and the actual scores are illustrated in Table S4 (Online Resource 1) as well as in Table 1 and Fig. 3. The models well predicted 7/10 scores (median correlation coefficient, 0.47; range, 0.39–0.57) from 24–230 (0.5–4.6%) of the possible connections/edges, underpinning the potential generalizability of the developed models for most of the cognitive tests.

The binary matrices of the validated edges for the cognitive tests are shown in Fig. 4A-C (and Fig. S2A-D, Online Resource 1), together with an anatomical representation of the validated edges and their adjacent nodes. The validated edges followed a pattern of mainly left intra-hemispheric as well as inter-hemispherical connections. Of note, the majority (overall 830/991 = 84%, Table S4, Online Resource 1) of the validated edges were between nodes of different networks (inter-network connections, median 104, range 19–193) rather than between nodes of the same network (median 19, range 5–41).

A detailed heatmap of the network nodes and their degrees of adjacent validated edges are depicted in Fig. 4D. The distribution of critical nodes (degree of ≥ 10 adjacent validated edges) varied considerably between domains. Critical nodes for attention/processing speed (TMT-A) and executive function/concept shifting (TMT-B) were mainly located in the left visual and somatomotor networks, but also included nodes from the bilateral dorsal attention and default mode networks. For performance in visual working memory (Corsi Block Tapping), mainly nodes of the right dorsal attention and bilateral default mode networks proved critical and were predominantly connected by inter-hemispherical fibers. Nodes and edges critically involved in verbal semantic memory (Word List, immediate and delayed recall) were almost exclusively left-sided and included nodes from the visual, somatomotor, and default-mode networks. Highly critical nodes (degree of 15–20 adjacent validated edges) included the default-mode network nodes in the left temporal/parietal and bilateral posterior cingulate cortex in four to five of the seven predictive models.

Table 1

Correlation and permutation test results for the predicted vs. observed scores from the cross-validated CPM-models
Cognitive	Cognitive	Correlation	P-Value	P-Value
Domains	Test	Coefficient (r)	Correlation	Permutation
Attention, processing speed	Trail-Making Test A	0.388	1.088⋅10^-05 ***	0.01 *
Executive function, concept shifting	Trail-Making Test B	0.470	5.330⋅10^-08 ***	0.01 *
Language, semantic word fluency	Imagined Shopping Tour	0.481	2.346⋅10^-08 ***	0.01 *
Language processing	Number Transcoding	0.411	2.817⋅10^-06 ***	0.01 *
Verbal working memory	Digit Span Fw	0.123	0.180	0.13
Verbal working memory	Digit Span Bw	0.132	0.150	0.16
Visuo-spatial working memory	Corsi Block Tapping Fw	0.202	0.027 *	0.11
Visuo-spatial working memory	Corsi Block Tapping Bw	0.433	7.211⋅10^-06 ***	0.01 *
Verbal episodic memory	Word List, Immediate Recall	0.569	9.346⋅10^-12 ***	0.01 *
Verbal episodic memory	Word List, Delayed Recall	0.525	6.487⋅10^-10 ***	0.01 *

CPM, Connectivity-based predictive modeling; Fw, Forward; Bw, Backward; * p < 0.05, *** p < 0.001

As demonstrated in recent randomized trials and in the present study, glioma patients are at high risk of developing cognitive decline during their disease [5, 29]. While the relationship between brain damage and neurologic function is generally well established for the so-called eloquent primary cortical areas and their associated fiber tracts, the underlying causes of cognitive deterioration in brain tumor patients remain poorly understood [4, 30, 31]. We found that in pretreated glioma patients, disturbed structural connectivity between different brain networks is associated with reduced cognitive performance in most domains. Critical nodes included several cortical regions known as connective hubs, such as the bilateral precuneus and posterior cingulate cortex [32].

Prediction of cognitive deficits from the affection of individual fiber tracts

Studies regarding the relationship between white matter alterations and neurologic or cognitive deficits in glioma patients have, so far, focused on the perioperative setting and single, anatomically defined tracts and their impact on motor function, speech and selected higher cognitive functions. These include central callosal fibers (supplementary motor area syndrome) [33], left inferior fronto-occipital/inferior longitudinal/uncinate fascicles (aphasia) [34, 35], right frontal aslant tract (executive functions, attention shift, verbal fluency) [36], right superior longitudinal fascicle/fronto-striatal tract/orbitofrontal cortex (mentalizing/visuospatial function) [37–40], and right inferior fronto-striatal tract/inferior frontal gyrus (inference control processes) [41]. Regarding radiation therapy, it was observed that radiation-induced changes in the hippocampus, parahippocampal cingulate, and other structures of the medial temporal lobe were associated with decreases in verbal memory, verbal fluency and visuospatial memory [42, 43].

Whole-brain connectome: the importance of networks and hubs

Although the integrity of single fiber tracts appears to have a measurable impact on different aspects of cognitive functioning, most higher brain functions are probably supported by more general organizational principles governing the information flow in the brain. Several brain regions which are centrally located and structurally highly connected have been identified and termed the “rich club” [32]. Of note, functional brain networks also show rich club phenomena [44]. Despite differences in node topography and network nomenclature, evidence exists that most of the densely connected hubs are located in the posterior part of the default mode network (bilateral precuneus/posterior cingulate cortex and inferior parietal regions) [45, 46]. Also, structural rich club connections were highly represented among inter-RSN connections, especially between the default mode, salience (ventral attention), and frontoparietal control networks [46].

These findings and the availability of advanced MR imaging techniques in the clinic have led to more network-oriented approaches for investigating the dependence of cognitive performance on structural connectivity in perioperative glioma patients. Liu et al. [47] constructed a whole-brain structural connectome from deterministic tractography in preoperative glioma patients. They observed that local (node) efficiency, a measure for communication strength among the first neighbors of a node, was generally reduced in tumor patients and particularly related to memory function in temporal tumors and to information processing speed in frontal tumors. Zhang et al. [8] applied navigated transcranial magnetic stimulation (nTMS) and whole-brain deterministic tractography in left-sided tumors and observed a correlation between the average node degree (and other connectome properties) of the left-hemispheric/nTMS-positive networks and the degree of aphasia. A network-level approach was also adopted by Mrah et al. [48], who found lesions or disconnections of the fronto-parietal (control) network to be most predictive for postoperative deterioration in cognitive set-shifting.

In the present study, most of the validated predictive connections were those between different RSNs rather than within single RSNs. Particularly critical nodes included the default mode network's left temporal and bilateral posterior cingulate cortex which may be in part due to the language-dependency of most tests. These findings fit well with the view that structural connections between RSNs form the backbone of functional connectivity, enabling higher cognitive processes any may imply that cognitive decline in treated brain tumor patients shares a common mechanism with other major psychiatric and neurological disorders where the rich club nodes were also found to be predominantly involved [49], such as the precuneus/posterior cingulate cortex in Alzheimer’s disease [50].

Limitations

This study included patients with substantial variability in treatment intensity and time between treatment initiation and imaging and neurocognitive assessment, and each patient was observed only once, such that longitudinal observations were not available. On the other hand, while group analyses are usually challenging to perform in this constellation, a rich, diverse pattern of structural damage may have facilitated the construction of a predictive model for cognition performance. From a methodological point of view, applying an atlas-based parcellation created from healthy subjects may be questionable because a functional restructuring of the brain, including shifts and deformations of cortical nodes, may have occurred in the patients. Finally, fiber tractography always approximates the actual white matter structure since even the most advanced methods may fail, especially in structurally altered brain tissue.

In summary, the present results suggest that the cognitive performance of treated glioma patients is strongly related to the structural connectivity between rather than within multiple brain networks and the integrity of known network hubs. This mirrors a pattern observed for other major neurological disorders.

ACT: Anatomically-Constrained Tractography; CNS: Central Nervous System; CPM: connectome-based predictive modeling; CSD: constrained spherical deconvolution; ECOG: Eastern Cooperative Oncology Group; EPI: echo-planar imaging; ¹⁸F-FET: O-(2-[¹⁸F]fluoroethyl)-L-tyrosine; FLAIR: fluid-attenuated inversion recovery; FOD: fiber orientation distribution function; FoV: field-of-view; HARDI: high-angular-resolution diffusion imaging; IDH: isocitrate dehydrogenase; MPRAGE: magnetization prepared rapid acquisition gradient echo; nTMS: navigated transcranial magnetic stimulation; OPOSEM: ordered Poisson ordinary subset expectation maximization; R²: coefficient of determination; RSN: resting-state network; SPACE: sampling perfection with application-optimized contrasts using different flip angle evolution; SS3T-CSD: single-shell 3-tissue constrained spherical deconvolution; TE: echo time; TR: repetition time; WHO: World Health Organization

Funding

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.

Authorship Contribution

MF and MK analyzed and interpreted most of the data. PL, KJL and NJS developed the MR and PET imaging techniques. SC provided the control group data. GS and CPF performed the patient examinations. FMM, CWS, MIR, KJL, GRF, NG and MK recruited the patients and conceptualized the study. All authors contributed to the article and approved the submitted version.

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of University of Cologne (14.11.2017, 17-365).

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Acknowledgments

The technical assistance of N. Judov, T. Plum, S. Schaden, and L. Tellmann is gratefully acknowledged

Louis DN, Perry A, Reifenberger G, von Deimling A, Figarella-Branger D, Cavenee WK, Ohgaki H, Wiestler OD, Kleihues P, Ellison DW: The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol 2016, 131(6):803-820.
Louis DN, Perry A, Wesseling P, Brat DJ, Cree IA, Figarella-Branger D, Hawkins C, Ng HK, Pfister SM, Reifenberger G et al: The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol 2021, 23(8):1231-1251.
Bosma I, Vos MJ, Heimans JJ, Taphoorn MJ, Aaronson NK, Postma TJ, van der Ploeg HM, Muller M, Vandertop WP, Slotman BJ et al: The course of neurocognitive functioning in high-grade glioma patients. Neuro Oncol 2007, 9(1):53-62.
Dallabona M, Sarubbo S, Merler S, Corsini F, Pulcrano G, Rozzanigo U, Barbareschi M, Chioffi F: Impact of mass effect, tumor location, age, and surgery on the cognitive outcome of patients with high-grade gliomas: a longitudinal study. Neurooncol Pract 2017, 4(4):229-240.
Gilbert MR, Dignam JJ, Armstrong TS, Wefel JS, Blumenthal DT, Vogelbaum MA, Colman H, Chakravarti A, Pugh S, Won M et al: A randomized trial of bevacizumab for newly diagnosed glioblastoma. N Engl J Med 2014, 370(8):699-708.
Satoer D, Visch-Brink E, Smits M, Kloet A, Looman C, Dirven C, Vincent A: Long-term evaluation of cognition after glioma surgery in eloquent areas. J Neurooncol 2014, 116(1):153-160.
Friedrich M, Farrher E, Caspers S, Lohmann P, Lerche C, Stoffels G, Filss C, Weiss Lucas C, Ruge MI, Langen KJ et al: Alterations in white matter fiber density associated with structural MRI and metabolic PET lesions following multimodal therapy in glioma patients. Front Oncol 2022, 14.
Zhang H, Ille S, Sogerer L, Schwendner M, Schroder A, Meyer B, Wiestler B, Krieg SM: Elucidating the structural-functional connectome of language in glioma-induced aphasia using nTMS and DTI. Hum Brain Mapp 2022, 43(6):1836-1849.
Le Fevre C, Cheng X, Loit MP, Keller A, Cebula H, Antoni D, Thiery A, Constans JM, Proust F, Noel G: Role of hippocampal location and radiation dose in glioblastoma patients with hippocampal atrophy. Radiat Oncol 2021, 16(1):112.
Dadario NB, Brahimaj B, Yeung J, Sughrue ME: Reducing the Cognitive Footprint of Brain Tumor Surgery. Front Neurol 2021, 12:711646.
Duffau H: Why brain radiation therapy should take account of the individual structural and functional connectivity: Toward an irradiation "a la carte". Crit Rev Oncol Hematol 2020, 154:103073.
Herbet G, Duffau H: Revisiting the Functional Anatomy of the Human Brain: Toward a Meta-Networking Theory of Cerebral Functions. Physiol Rev 2020, 100(3):1181-1228.
Caspers S, Moebus S, Lux S, Pundt N, Schutz H, Muhleisen TW, Gras V, Eickhoff SB, Romanzetti S, Stocker T et al: Studying variability in human brain aging in a population-based German cohort-rationale and design of 1000BRAINS. Front Aging Neurosci 2014, 6:149.
Langen KJ, Stoffels G, Filss C, Heinzel A, Stegmayr C, Lohmann P, Willuweit A, Neumaier B, Mottaghy FM, Galldiks N: Imaging of amino acid transport in brain tumours: Positron emission tomography with O-(2-[(18)F]fluoroethyl)-L-tyrosine (FET). Methods 2017, 130:124-134.
Schaefer A, Kong R, Gordon EM, Laumann TO, Zuo XN, Holmes AJ, Eickhoff SB, Yeo BTT: Local-Global Parcellation of the Human Cerebral Cortex from Intrinsic Functional Connectivity MRI. Cereb Cortex 2018, 28(9):3095-3114.
Aerts H, Dhollander T, Marinazzo D: Evaluating the performance of 3-tissue constrained spherical deconvolution pipelines for within-tumor tractography. bioRxiv preprint 2019.
Smith RE, Tournier JD, Calamante F, Connelly A: Anatomically-constrained tractography: improved diffusion MRI streamlines tractography through effective use of anatomical information. Neuroimage 2012, 62(3):1924-1938.
Shen X, Finn ES, Scheinost D, Rosenberg MD, Chun MM, Papademetris X, Constable RT: Using connectome-based predictive modeling to predict individual behavior from brain connectivity. Nat Protoc 2017, 12(3):506-518.
Langen KJ, Galldiks N, Hattingen E, Shah NJ: Advances in neuro-oncology imaging. Nat Rev Neurol 2017, 13(5):279-289.
Oken MM, Creech RH, Tormey DC, Horton J, Davis TE, McFadden ET, Carbone PP: Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol 1982, 5(6):649-655.
Herzog H, Langen KJ, Weirich C, Rota Kops E, Kaffanke J, Tellmann L, Scheins J, Neuner I, Stoffels G, Fischer K et al: High resolution BrainPET combined with simultaneous MRI. Nuklearmedizin 2011, 50(2):74-82.
Tombaugh TN: Trail Making Test A and B: normative data stratified by age and education. Arch Clin Neuropsychol 2004, 19(2):203-214.
Kalbe E, Kessler J, Calabrese P, Smith R, Passmore AP, Brand M, Bullock R: DemTect: a new, sensitive cognitive screening test to support the diagnosis of mild cognitive impairment and early dementia. Int J Geriatr Psychiatry 2004, 19(2):136-143.
Berch DB, Krikorian R, Huha EM: The Corsi block-tapping task: methodological and theoretical considerations. Brain Cogn 1998, 38(3):317-338.
Tournier JD, Smith R, Raffelt D, Tabbara R, Dhollander T, Pietsch M, Christiaens D, Jeurissen B, Yeh CH, Connelly A: MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation. Neuroimage 2019, 202:116137.
Tournier JD, Calamante F, Connelly A: Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution. Neuroimage 2007, 35(4):1459-1472.
Dhollander T, Connelly A: A novel iterative approach to reap the benefits of multi-tissue CSD from just single-shell (+b=0) diffusion MRI data. Proc Intl Soc Mag Reson Med 2016, 24:3010.
Pauleit D, Floeth F, Hamacher K, Riemenschneider MJ, Reifenberger G, Muller HW, Zilles K, Coenen HH, Langen KJ: O-(2-[18F]fluoroethyl)-L-tyrosine PET combined with MRI improves the diagnostic assessment of cerebral gliomas. Brain 2005, 128(Pt 3):678-687.
Wefel JS, Cloughesy T, Zazzali JL, Zheng M, Prados M, Wen PY, Mikkelsen T, Schiff D, Abrey LE, Yung WK et al: Neurocognitive function in patients with recurrent glioblastoma treated with bevacizumab. Neuro Oncol 2011, 13(6):660-668.
Habets EJJ, Hendriks EJ, Taphoorn MJB, Douw L, Zwinderman AH, Vandertop WP, Barkhof F, De Witt Hamer PC, Klein M: Association between tumor location and neurocognitive functioning using tumor localization maps. J Neurooncol 2019, 144(3):573-582.
Hendriks EJ, Habets EJJ, Taphoorn MJB, Douw L, Zwinderman AH, Vandertop WP, Barkhof F, Klein M, De Witt Hamer PC: Linking late cognitive outcome with glioma surgery location using resection cavity maps. Hum Brain Mapp 2018, 39(5):2064-2074.
van den Heuvel MP, Sporns O: Network hubs in the human brain. Trends Cogn Sci 2013, 17(12):683-696.
Tuncer MS, Fekonja LS, Ott S, Pfnur A, Karbe AG, Engelhardt M, Faust K, Picht T, Coburger J, Duhrsen L et al: Role of interhemispheric connectivity in recovery from postoperative supplementary motor area syndrome in glioma patients. J Neurosurg 2022:1-10.
Ille S, Engel L, Kelm A, Meyer B, Krieg SM: Language-Eloquent White Matter Pathway Tractography and the Course of Language Function in Glioma Patients. Front Oncol 2018, 8:572.
Prasse G, Meyer HJ, Scherlach C, Maybaum J, Hoffmann A, Kasper J, Karl Fehrenbach M, Wilhelmy F, Meixensberger J, Hoffmann KT et al: Preoperative language tract integrity is a limiting factor in recovery from aphasia after glioma surgery. Neuroimage Clin 2022, 37:103310.
Landers MJF, Meesters SPL, van Zandvoort M, de Baene W, Rutten GM: The frontal aslant tract and its role in executive functions: a quantitative tractography study in glioma patients. Brain Imaging Behav 2022, 16(3):1026-1039.
Liu D, Liu Y, Hu X, Hu G, Yang K, Xiao C, Hu J, Li Z, Zou Y, Chen J et al: Alterations of white matter integrity associated with cognitive deficits in patients with glioma. Brain Behav 2020:e01639.
Nakajima R, Kinoshita M, Miyashita K, Okita H, Genda R, Yahata T, Hayashi Y, Nakada M: Damage of the right dorsal superior longitudinal fascicle by awake surgery for glioma causes persistent visuospatial dysfunction. Sci Rep 2017, 7(1):17158.
Nakajima R, Kinoshita M, Okita H, Yahata T, Matsui M, Nakada M: Neural Networks Mediating High-Level Mentalizing in Patients With Right Cerebral Hemispheric Gliomas. Front Behav Neurosci 2018, 12:33.
Nakajima R, Yordanova YN, Duffau H, Herbet G: Neuropsychological evidence for the crucial role of the right arcuate fasciculus in the face-based mentalizing network: A disconnection analysis. Neuropsychologia 2018, 115:179-187.
Puglisi G, Howells H, Sciortino T, Leonetti A, Rossi M, Conti Nibali M, Gabriel Gay L, Fornia L, Bellacicca A, Vigano L et al: Frontal pathways in cognitive control: direct evidence from intraoperative stimulation and diffusion tractography. Brain 2019, 142(8):2451-2465.
Chapman CH, Zhu T, Nazem-Zadeh M, Tao Y, Buchtel HA, Tsien CI, Lawrence TS, Cao Y: Diffusion tensor imaging predicts cognitive function change following partial brain radiotherapy for low-grade and benign tumors. Radiother Oncol 2016, 120(2):234-240.
Tringale KR, Nguyen TT, Karunamuni R, Seibert T, Huynh-Le MP, Connor M, Moiseenko V, Gorman MK, Marshall A, Tibbs MD et al: Quantitative Imaging Biomarkers of Damage to Critical Memory Regions Are Associated With Post-Radiation Therapy Memory Performance in Brain Tumor Patients. Int J Radiat Oncol Biol Phys 2019, 105(4):773-783.
Grayson DS, Ray S, Carpenter S, Iyer S, Dias TG, Stevens C, Nigg JT, Fair DA: Structural and functional rich club organization of the brain in children and adults. PLoS One 2014, 9(2):e88297.
Horn A, Ostwald D, Reisert M, Blankenburg F: The structural-functional connectome and the default mode network of the human brain. Neuroimage 2014, 102 Pt 1:142-151.
van den Heuvel MP, Sporns O: An anatomical substrate for integration among functional networks in human cortex. J Neurosci 2013, 33(36):14489-14500.
Liu Y, Yang K, Hu X, Xiao C, Rao J, Li Z, Liu D, Zou Y, Chen J, Liu H: Altered Rich-Club Organization and Regional Topology Are Associated With Cognitive Decline in Patients With Frontal and Temporal Gliomas. Front Hum Neurosci 2020, 14:23.
Mrah S, Descoteaux M, Wager M, Bore A, Rheault F, Thirion B, Mandonnet E: Network-level prediction of set-shifting deterioration after lower-grade glioma resection. J Neurosurg 2022:1-9.
de Lange SC, Scholtens LH, Alzheimer's Disease Neuroimaging I, van den Berg LH, Boks MP, Bozzali M, Cahn W, Dannlowski U, Durston S, Geuze E et al: Shared vulnerability for connectome alterations across psychiatric and neurological brain disorders. Nat Hum Behav 2019, 3(9):988-998.
Yokoi T, Watanabe H, Yamaguchi H, Bagarinao E, Masuda M, Imai K, Ogura A, Ohdake R, Kawabata K, Hara K et al: Involvement of the Precuneus/Posterior Cingulate Cortex Is Significant for the Development of Alzheimer's Disease: A PET (THK5351, PiB) and Resting fMRI Study. Front Aging Neurosci 2018, 10:304.

No competing interests reported.

OnlineResourceJNeurooncolsubmitted.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Whole-brain structural connectivity predicts cognitive deficits in pretreated glioma patients

Status:

Version 1

Abstract

Purpose

Methods

Results

Conclusions

Figures

Introduction

Patients and methods

Patient characteristics

Hybrid PET/MR imaging

Cognitive performance

Whole-brain structural connectome

Connectome-based predictive modeling

Results

Cognitive outcome and fiber count distribution

Connectome-based predictive modeling

Discussion

Prediction of cognitive deficits from the affection of individual fiber tracts

Whole-brain connectome: the importance of networks and hubs

Limitations

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1