Brain Connectivity and Cognitive Flexibility in Nonirradiated Adult Survivors of Childhood Leukemia

Brain Connectivity and Cognitive Flexibility in Nonirradiated Adult Survivors of Childhood Leukemia Abstract Background This study aimed to assess functional and structural brain connectivity in adult childhood leukemia survivors and the link with cognitive functioning and previously identified risk factors such as intrathecal methotrexate dose and age at start of therapy. Methods Thirty-one nonirradiated adult childhood leukemia survivors and 35 controls underwent cognitive testing and multimodal magnetic resonance imaging (resting state functional MRI, T1-weighted, diffusion-weighted, and myelin water imaging [MWI]). Analyses included dual regression, voxel-based morphometry, advanced diffusion, and MWI modeling techniques besides stepwise discriminant function analysis to identify the most affected executive cognitive domain. Correlations with discrete intrathecal MTX doses and (semi)continuous variables were calculated using Spearman’s rank and Pearson’s correlation, respectively. All correlation tests were two-sided. Positive and negative T-contrasts in functional and structural MRI analysis were one-sided. Results Survivors demonstrated lower functional connectivity between the default mode network (DMN) and inferior temporal gyrus (ITG; P < .008). Additionally, we observed higher fractional anisotropy (FA; P = .04) and lower orientation dispersion index (ODI; P = .008) at the left centrum semiovale, which could—given that several fiber bundles cross this region—suggest selective reduced integrity of the respective white matter tracts. Set shifting reaction time, a measure of cognitive flexibility, was mostly impaired and correlated with lower FA (r = –0.53, P = .003) and higher ODI (r = 0.40, P = .04) in survivors but not with DMN-ITG connectivity. There were no statistically significant differences between survivors and controls in WM or GM volume, nor was there a statistically significant correlation between imaging measurements and age at start of therapy or intrathecal methotrexate dose. Conclusions Adult, nonirradiated childhood leukemia survivors show altered brain connectivity, which is linked with cognitive flexibility. Central nervous system (CNS)–directed prophylaxis to prevent early relapse has improved survival rates in childhood leukemia (1–3), increasing the attention for long-term sequelae due to such prophylactic treatment. The detrimental impact of cranial irradiation on the developing brain (1,2,4–14) has resulted in its successful replacement by intrathecal and high-dose intravenous methotrexate. However, long-term neurocognitive deficits remain a clinical concern even in nonirradiated survivors, albeit to a milder extent. Because executive functioning appeared mostly affected in nonirradiated childhood cancer survivors (1–3,7,11,15–22), white matter (WM) tracts are suggested to be especially at risk for chemotherapy-induced damage (23) and related processing speed, which has been confirmed by imaging studies. Neuro-imaging findings in nonirradiated survivors of childhood leukemia range from gross calcifications and WM hyperintensities to more subtle changes in WM volumes (22,24–32). While voxel-based morphometry (VBM) can be applied to study volumetric differences, several techniques exist to study microstructural WM alterations. One imaging technique is diffusion MRI (dMRI), which probes the random motion of water molecules, directed by axonal membranes and modulated by the myelin sheath (33). Building on dMRI data, the diffusion tensor imaging (DTI) (34,35) model quantifies mean diffusivity (MD) and fractional anisotropy (FA). Because intact white matter exhibits highly directional diffusion, decreased FA may indicate injury to a particular tract (36). Previous studies in childhood leukemia survivors have, however, reported inconsistent findings with regard to FA measures. Some authors confirmed statistically significantly reduced FA in nonirradiated childhood leukemia survivors (5,27,30), whereas others did not (9,26,37) or reported even increased FA (38). Admittedly, some assumptions of the DTI model could be inadequate for biological tissue. Different time points after treatment also obscure the DTI parameters, which complicates comparisons between studies. Furthermore, dMRI might not be sensitive enough to detect the more subtle effects of chemotherapy compared with cranial irradiation. Hence, advanced dMRI models may clarify the aforementioned conflicting findings. One such model is diffusion kurtosis imaging (DKI), which quantifies non-Gaussian diffusion, believed to reflect tissue compartmentalization (39,40). More recently, neurite orientation dispersion and density imaging (NODDI) (41) was developed, an approach modeling the contribution of cerebrospinal fluid (Fiso), axonal vs extra-axonal tissue fraction (Vic) to the diffusion signal. It also models the axonal orientation dispersion index (ODI), which is inversely related to FA. Given its pivotal role in signal transmission and protection of fiber bundles, recent techniques focus on myelin content. One such technique is myelin water imaging (MWI), estimating the myelin water fraction (MWF), a potential marker for myelin content (42). Although demonstrating promising results in several neurological conditions (43,44), none of these listed techniques have been implemented yet in the study of chemotherapy-induced brain damage in childhood leukemia survivors. Besides potentially altered structural brain connectivity, little is known about the effect of chemotherapeutics on functional brain connectivity in acute lymphoblastic leukemia (ALL) survivors. Typically, functional connectivity can be studied by using resting state functional MRI (rsfMRI). This technique registers spontaneous blood oxygenation level–dependent (BOLD) signal fluctuations across the brain. Regions that show synchronized BOLD amplitude fluctuations in the absence of an explicit task (45–47) are assumed to constitute functionally linked “resting state networks” (48–51). Given the evidence for executive impairments in childhood leukemia survivors, two functional networks are of particular interest, the default mode network (DMN) and the frontoparietal network (FPN; also referred to as central executive network). The DMN comprises the precuneus, posterior cingulate, medial frontal, middle temporal, and lateral parietal regions (48,49,52,53). The FPN entails the lateral prefrontal cortex, anterior cingulate cortex, inferior parietal lobule, and superior temporal region. It is presumed to play an important role in attention control and set shifting and is highly interconnected with the DMN (53–56). Although several studies have investigated functional connectivity in adult breast cancer patients (57–60), connectivity in child survivors of leukemia was assessed in only one study so far. Specifically, Kesler et al. reported diffuse altered brain connectivity compared with healthy controls, which was related to lower efficiency and slower information processing (61). Given that previous studies reported mainly attentional and executive difficulties in survivors, the aim of our study was to determine the most vulnerable brain regions for potential neurotoxicity in leukemia survivors. To investigate this in detail, the entire brain was investigated using multiple modalities. Our hypotheses were as follows: 1) altered functional brain connectivity in attention-related networks in survivors (investigated using rsfMRI), 2) structural gray and white matter brain changes (using VBM, dMRI, and MWI), and 3) the prediction of cognitive performance based on statistically significantly altered regions (ie, as the result of the neuroimaging analyses of the rsfMRI, VBM, dMRI, or MWI measurements), previously identified risk factors such as intrathecal dose of methotrexate (32,62,63), and age at start of therapy (2,11,15,21,64,65). Methods Participants Thirty-one adult, nonirradiated childhood leukemia survivors (mean age at diagnosis = 6.4 years, range = 17–30 years), treated between 1994 and 2004 according to the European Organisation for Research and Treatment of Cancer 58881 or 58951 protocol (66,67) at the University Hospital of Leuven, were included. Thirty-five control participants were recruited via family members of the survivors and announcements. Demographics of study participants are listed in Table 1. A detailed description of the survivors’ therapeutic background and comparison between treatment protocols can be found in Supplementary Tables 1 and 2 (available online). Each participant signed a written informed consent that was approved by the ethical committee (dossier No. B322201419664) of the University Hospital of Leuven. Table 1. Demographic data of study participants Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    * Statistical analysis was based on two-sided, independent-sample t test. † Socio-economic index score was calculated by multiplying parental occupation scale value by a weight of 5 and parental education scale by 3, and then summing these products as described earlier (69). High score indicates high socio-economic status. Because this index differed statistically significantly between survivors and controls, this parameter was applied as covariate in subsequent analyses. ‡ Scores on the Beck Depression Inventory and State-Trait Anxiety Inventory are presented as raw scores. High score indicates a higher problem rate. § Statistical analysis was based on two-sided, one-way analysis of covariance, with group (survivors vs controls) as an independent factor and a parental socioeconomic status as a covariate. Table 1. Demographic data of study participants Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    * Statistical analysis was based on two-sided, independent-sample t test. † Socio-economic index score was calculated by multiplying parental occupation scale value by a weight of 5 and parental education scale by 3, and then summing these products as described earlier (69). High score indicates high socio-economic status. Because this index differed statistically significantly between survivors and controls, this parameter was applied as covariate in subsequent analyses. ‡ Scores on the Beck Depression Inventory and State-Trait Anxiety Inventory are presented as raw scores. High score indicates a higher problem rate. § Statistical analysis was based on two-sided, one-way analysis of covariance, with group (survivors vs controls) as an independent factor and a parental socioeconomic status as a covariate. Data Acquisition Magnetic Resonance Imaging Magnetic resonance (MR) scans were acquired in UZ Leuven using a 3T Philips (Achieva) MR-scanner and a 32-channel phased-array head coil. A high-resolution T1-weighted image was acquired for anatomical reference and VBM (MPRAGE, resolution 0.98×0.98 × 1.2 mm3, TE/TR= 4.6 ms/9636 ms, FOV = 192×250×250, 1.2 mm slice thickness, 160 slices). For diffusion data, an echo-planar, multishell imaging scheme was used, consisting of b-values 700, 1000, and 2800 s/mm2, applied along 25, 40, and 75 uniformly distributed directions, respectively, complemented by 10 b = 0 images (voxel size 2.5×2.5×2.5 mm3, TR/TE = 7800 ms/90 ms, FOV= 240×240×125, 50 axial slices, parallel imaging factor 2) (68). For MWI, a 3D GraSE sequence was acquired (FOV 230×80×90 mm, slice thickness 5 mm with 2.5 mm overlap, reconstructed voxel size 0.96×0.96×2.5 mm3, 32 slices, ETL = 32, TE = 10, 20, … , 320 ms, TR = 1 s, EPI factor 3, acceleration factor 2). The resting-state fMRI images were acquired using a T2*-weighted echo-planar imaging (EPI) sequence (resolution 4×4×4 mm3, TE/TR = 33 ms/1700 ms; FOV = 230×120×230 mm; 4 mm slice thickness, 30 slices were assembled in seven minutes), with whole-cerebrum coverage. Two-hundred fifty dynamics of rsfMRI were acquired. Participants were instructed to lay still, close their eyes, and feel at ease, but not to sleep. Image data of one control participants were excluded due to technical problems during the storage process. See Supplementary Table 3 (available online) for image acquisition details. Image Preprocessing Please see the Supplementary Materials (available online) for details. After preprocessing and quality control, the final sample sizes were 30 survivors/33 controls for dMRI analysis, 29 survivors/32 controls for MWI analysis, 31 survivors/34 controls for VBM, and 28 survivors/33 controls for rsfMRI. Neuropsychological Assessment Neuropsychological assessments, including the Amsterdam Neuropsychological Test battery (ANT) (70,71), were performed (for a detailed description, see [65] and Supplementary Table 4, available online). In this study, we focused on the ANT, given the existing evidence for specific executive impairments in childhood cancer survivors (1–3,15). Statistical Analysis For dMRI and MWI data, statistical analyses were performed voxelwise with the Statistical Parametric Mapping toolbox (SPM), using univariate t tests with group, age, and socioeconomic status (SES) as covariates. The contrasts survivors > controls and controls > survivors were thresholded at a P value of less than .001, uncorrected for multiple comparisons. Clusters with family-wise error–corrected P values of less than .05 (Random Field Theory as implemented in SPM) were deemed statistically significant. From clusters yielding statistically significant differences during imaging experiments, average metric values adjusted for age and SES were extracted. The rsfMRI spatial maps obtained from dual regression were tested for differences between groups using FSL randomized nonparametric permutation testing. Similar to the structural analyses, given the multiple voxelwise tests for each component, the analysis was corrected for multiple comparisons across voxels by family-wise error rate (FWER) correction (permutation testing with 5000 permutations as implemented in FSL). Finally, an additional Bonferroni correction was applied, as six group comparisons of interest were performed (3 networks [DMN, left FPN, right FPN]×2 group comparisons [higher vs lower connectivity in the survivor group]). This resulted in an adjusted (ie, more stringent) threshold for the P values (αcorrected = .008 (= .05/6)). Mean time series within regions showing a statistically significant group difference were calculated for each subject and correlated (Pearson correlation) with the mean time series of the associated component network (ie, DMN, left FPN, or right FPN). This correlation was used as a functional connectivity index (72) and correlated with behavior (Figure 1). Statistical analysis of performances on the ANT tasks included two-sided, one-way analysis of covariance (ANCOVA) with group (survivor vs control) as the independent factor and an index of parental socioeconomic status (69) as the covariate. Reaction times were transformed using a logarithmic transformation. Stepwise discriminant function analysis (DFA) was applied using the Wilks Lambda test to demonstrate which variable of executive functioning best distinguished survivors from controls. Outcome measurements from dMRI, MWI, or fMRI significantly differing between survivors and controls were correlated with 1) the neurocognitive measurement best discriminating controls from survivors and 2) predictors of outcome, that is, age at start of therapy and total intrathecal dose of methotrexate. The statistical significance of these correlations was assessed using two-sided Pearson (ANT outcome measurements, age at start of therapy) or Spearman (total dose intrathecal methotrexate) correlation. All correlation tests were two-sided, with a statistical significance level α of .05. Results Functional Connectivity Data After Bonferroni and multiple-voxel correction, dual regression analysis for the DMN resulted in a statistically significantly lower connectivity in survivors between the DMN and the inferior temporal gyrus (ITG; P <.008) (Figure 2). This region of the ITG included 38 voxels. No regions of hyperconnectivity were encountered in survivors. Voxel-wise comparison in connectivity with the FPN did not show any statistically significant differences between both groups. Figure 1. View largeDownload slide Selection of resting state networks. After acquisition and preprocessing, functional resting state networks were selected based on visual inspection by the neuroradiologist. A–C) Right frontoparietal network. D–F) Left frontoparietal network. G–I) Default mode network. A, D, G) Sagittal plane. B, E, H) Coronal plane. C, F, I) Transverse plane. L= left; R = right. Figure 1. View largeDownload slide Selection of resting state networks. After acquisition and preprocessing, functional resting state networks were selected based on visual inspection by the neuroradiologist. A–C) Right frontoparietal network. D–F) Left frontoparietal network. G–I) Default mode network. A, D, G) Sagittal plane. B, E, H) Coronal plane. C, F, I) Transverse plane. L= left; R = right. Figure 2. View largeDownload slide Functional connectivity measurements in survivors vs controls. For all subjects, functional connectivity between the attentional networks (ie, default mode network [DMN], left and right frontoparietal network) and other brain areas was assessed. After Bonferroni and multiple-voxel correction, hypoconnectivity was detected between the DMN and inferior temporal gyrus in survivors (n = 28), compared with controls (n = 33). The group-average DMN is depicted in orange, while the region with hypoconnectivity in survivors is marked in blue. No regions of hyperconnectivity were encountered in survivors. Figure 2. View largeDownload slide Functional connectivity measurements in survivors vs controls. For all subjects, functional connectivity between the attentional networks (ie, default mode network [DMN], left and right frontoparietal network) and other brain areas was assessed. After Bonferroni and multiple-voxel correction, hypoconnectivity was detected between the DMN and inferior temporal gyrus in survivors (n = 28), compared with controls (n = 33). The group-average DMN is depicted in orange, while the region with hypoconnectivity in survivors is marked in blue. No regions of hyperconnectivity were encountered in survivors. Structural Connectivity Data Voxel-based morphometry did not reveal statistically significant differences between survivors and controls in WM and GM volume. However, in a region covering the left centrum semiovale, survivors expressed higher FA than controls (210 voxels, peak T59 = 5.03, cluster PFWE = .04) and lower ODI (165 voxels, peak T59 = 4.26, cluster PFWE = .008). These clusters were largely overlapping and cover areas that are crossed by WM fibers from the superior longitudinal fasciculus (SLF), corticospinal tract (CST), and corpus callosum (CC), as illustrated in Figure 3. A trend for similar changes was also found in the right hemisphere, yet this did not reach statistical significance at cluster level. Other measures describing tissue microstructure (MD, MK, Vic, Fiso, MWF) did not reveal differences between both groups (all PFWE > .05). Figure 3. View largeDownload slide Functional connectivity measurements in survivors vs controls . Clusters of voxels in which survivors (n = 30) and controls (n = 33) differ, overlaid on the fractional anisotropy (FA) template. A) The red-yellow cluster indicates high FA, while the blue clusters indicate low orientation dispersion index (ODI) in patients compared with controls. In the left hemisphere, clusters from both FA and ODI are present and overlap. B) Tractography, seeded from the cluster in the left hemisphere. Figure 3. View largeDownload slide Functional connectivity measurements in survivors vs controls . Clusters of voxels in which survivors (n = 30) and controls (n = 33) differ, overlaid on the fractional anisotropy (FA) template. A) The red-yellow cluster indicates high FA, while the blue clusters indicate low orientation dispersion index (ODI) in patients compared with controls. In the left hemisphere, clusters from both FA and ODI are present and overlap. B) Tractography, seeded from the cluster in the left hemisphere. Executive Functioning Results of cognitive testing were published earlier (65) and revealed that processing speed of FA4L (Th1), MSL (Th1), SAD (Th; focused attention, divided attention, and sustained attention, respectively), set shifting and flexibility SSV (Tc1), and reaction times on working memory SSV (Th3-Th1) and cognitive flexibility SSV (Tc3-Tc1) differed statistically significantly between survivors and controls. DFA demonstrated that set shifting reaction time best discriminated controls from survivors (Wilks Lambda = .87, chi-square (1) = 6.89, P = .009; data not shown). Correlation Between Executive Functioning and Connectivity Data Higher reaction times on the flexibility task corresponded statistically significantly with lower FA (r = –0.53, P = .003) and higher ODI (r = 0.40, P = .04) in survivors, but not in controls (r = –0.34, P = .05; r = 0.28, P = .12, respectively). Lower RSFC index was not correlated with higher reaction time in survivors or in controls (r = –0.08 and –0.08, P = .92 and .69, respectively). For the entire participant group (ie, combination of survivors and controls), no statistically significant correlations between imaging and neurocognitive measurements were encountered. The results are depicted in Table 2 and Figure 4. Finally, neither age at start of therapy nor total dose of intrathecal methotrexate was correlated with FA, ODI, or RSFC (all |r| < 0.27, P > .05). Table 2. Correlations between imaging measurements and set shifting reaction times and predictors of outcome Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  * Statistical significance was tested using two-sided Pearson correlation. DMN = default mode network; FA = fractional anisotropy; ITG = inferior temporal gyrus; ODI = orientation dispersion index; RSFC = resting state functional connectivity. † Statistical significance was tested using two-sided Spearman correlation. Table 2. Correlations between imaging measurements and set shifting reaction times and predictors of outcome Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  * Statistical significance was tested using two-sided Pearson correlation. DMN = default mode network; FA = fractional anisotropy; ITG = inferior temporal gyrus; ODI = orientation dispersion index; RSFC = resting state functional connectivity. † Statistical significance was tested using two-sided Spearman correlation. Figure 4. View largeDownload slide Correlation between diffusion magnetic resonance imaging (dMRI) and cognitive measurements. Correlations were calculated using Pearson’s correlation. Error bars indicate minimal and maximal values, respectively. Numbers of survivors were 30 and 28 for diffusion tensor imaging (DTI) and functional MRI (fMRI) measurements, respectively. Numbers of controls were 33 and 33 for DTI and fMRI measurements, respectively. A–C) Linear regression curves. D–F) Boxplot. Statistical significance was tested using two-sided Pearson correlation. FA = fractional anisotropy; ODI = orientation dispersion index; rsfMRI = resting state functional magnetic resonance imaging; RT SSV = Reaction time (ms) on Set Shifting Visual (attentional computerized task Amsterdam Neuropsychological Tasks). Figure 4. View largeDownload slide Correlation between diffusion magnetic resonance imaging (dMRI) and cognitive measurements. Correlations were calculated using Pearson’s correlation. Error bars indicate minimal and maximal values, respectively. Numbers of survivors were 30 and 28 for diffusion tensor imaging (DTI) and functional MRI (fMRI) measurements, respectively. Numbers of controls were 33 and 33 for DTI and fMRI measurements, respectively. A–C) Linear regression curves. D–F) Boxplot. Statistical significance was tested using two-sided Pearson correlation. FA = fractional anisotropy; ODI = orientation dispersion index; rsfMRI = resting state functional magnetic resonance imaging; RT SSV = Reaction time (ms) on Set Shifting Visual (attentional computerized task Amsterdam Neuropsychological Tasks). Discussion The present study uniquely combines neurocognitive data with structural and functional brain connectivity in nonirradiated childhood survivors on average 15 years after treatment. In these survivors, we report altered WM microstructure at the left centrum semiovale and reduced functional brain connectivity. In addition, measures describing WM microstructure are statistically significantly associated with cognitive performance. There were no differences between survivors and controls in WM or GM volume, or in FPN connectivity. Finally, we did not find a statistically significant correlation between age at start of therapy or intrathecal methotrexate dose and imaging measurements. Altered functional connectivity of the DMN was earlier reported after cancer treatment (57–59,61) and confirms the hypothesis of the DMN as a potential marker for chemotherapy-induced brain injury (73). In this study, a model-free (independent component) analysis was used, in contrast to the study of Kesler and colleagues (who used a seed-based analysis for 90 atlas-based regions). Using this analysis, connectivity between the DMN and other brain regions could be investigated globally. More specifically, the statistically significant difference in connectivity between the DMN and ITG in this study specifically could also suggest altered DMN-FPN (between-network) connectivity, as the ITG region is also part of the FPN. Reduced connectivity of the DMN with other brain areas can explain the observation that cognitive flexibility was mostly affected. Indeed, the DMN plays a pivotal role in allocating brain energy resources to relevant areas and thus in flexibly switching between attentional processes (74), although we could not correlate reaction times during a cognitive flexibility task with DMN-ITG connectivity. Chemosensitivity of the WM has been reported extensively in vivo (5,9,30,37,75–77) and in vitro (78). WM vulnerability might explain both the statistically significant dMRI and fMRI results. In this view, the study of Van den Heuvel et al. is of particular interest. They demonstrated that structural connections within resting state networks—such as the DMN—reflect the connectivity architecture of the brain. More specifically, CC and long associative fibers were identified to play a commending role in connecting the DMN with other brain areas (47). Interestingly, we found altered dMRI measurements at the centrum semiovale, a region where the CST, CC, and SLF—a long associative tract—intercross. Admittedly, reduced WM integrity is most often linked to lower FA and higher ODI measurements. However, the centrum semiovale is a region of so-called crossing fibers. The multidirectional diffusion in such regions a priori leads to a low anisotropy at voxel level (36). High FA and low ODI in the centrum semiovale of survivors can therefore be explained by the selective reduced integrity of one or more crossing fiber bundles, increasing overall anisotropy. Based on our data, we cannot disentangle which bundle(s) is/are most affected, but neurobiological and developmental mechanisms suggest involvement of SLF and/or CC fibers. The sensitivity of these tracts to chemotherapy (5,25,27,29,36) vs the robustness of the CST (79) was indeed described earlier. Retrogenesis —degeneration processes mirroring the order of acquisition in normal development—implies that the latest myelinated fiber tracts are more vulnerable to degeneration (23,36,80,81). This process might be explained by immature myelination of later developing tracts at the time of therapy, as an intact myelin sheath not only facilitates nerve impulse propagation, but also protects axonal fibers against noxious influences (81,82). We therefore believe that the higher FA and lower ODI at the centrum semiovale can be attributed to alterations in the SLF and/or CC. Interestingly, the FA of the SLF was earlier demonstrated to correlate with cognitive flexibility (83), as was the case in our survivors but less so in our control group. This discrepancy could be due to a reduction in crossing fibers in survivors, reestablishing the normal correlation between set shifting reaction time and dMRI measurements that gets obscured in this region for controls because of the multiple crossing fibers. Our results suggest the left hemisphere to be more affected than the right. Neurobiologically, the left hemisphere demonstrates a higher connectivity status (84) and therefore has higher metabolic needs (85). This might result in higher susceptibility of this hemisphere to aging and neurotoxicity (38,79,86). Moreover, Waber et al. reported that age at treatment onset for leukemia patients was related to left hemisphere– vs right hemisphere–dominant deficits. More specifically, patients younger than 36 months at the start of treatment displayed more impairments in right hemisphere–dominant functions, whereas older children showed more deficits related to left hemisphere–dominant functions (87), as is the case in our study. The mean age at diagnosis in our survivor cohort (ie, 6.4 years of age) confirms this observation. We could not confirm the established vulnerability of younger children to chemotherapy (2,11,15,21,61,64,88). However, the fact that the youngest children received a milder treatment resulting from a better prognosis might explain why we could not correlate age at start of therapy with imaging measurements. Nevertheless, the presumed vulnerability of the SLF and/or CC indirectly points to the importance of the maturation status of the brain at the time of chemotherapy. Also, we did not demonstrate the earlier established correlation between total intrathecal methotrexate dose and outcome (30,32,65). This could be due to homogeneity in methotrexate doses. To account for the interference of SES, this parameter was included as a covariate. Also, neurocognitive and imaging data were correlated at the subgroup level (ie, survivors or controls), as chemotherapy-induced brain alterations possibly result in two different populations. In that case, considering both cohorts as a homogenous group would not yield statistically significant findings. Additionally, stringent statistical correction might explain why we found only one region statistically significantly differing between controls and survivors in both DTI and fMRI measurements. Alternatively, subtle findings possibly imply a rather mild impact of chemotherapy on the developing brain after one or two decades, which is in accordance with earlier observations where nonirradiated survivors demonstrated more attenuated deficits than irradiated ones (9). To conclude, we confirm altered DMN connectivity in nonirradiated survivors of childhood leukemia, possibly explained by structural WM sequelae, as evidenced by altered FA and ODI values in the left centrum semiovale. Evidence from the literature and the correlation between imaging and neurocognitive results suggest the SLF and CC—passing through the centrum semiovale—to be especially at risk for chemotherapy-induced damage and indirectly confirm the vulnerability of the young brain to chemotherapy-induced neurotoxicity. Future studies are needed to evaluate the clinical applicability of functional and structural brain connectivity to detect sequelae of chemotherapy during childhood. Funding This study was supported by the charity-based Olivia Hendrickx Research Fund (www.olivia.be). Notes Affiliations of authors: Department of Radiology, University Hospital Leuven, Leuven, Belgium (TB, CS, SD); Icometrix, Leuven, Belgium (TB); Department of Child and Adolescent Psychiatry, KU Leuven, University Psychiatric Centre Leuven, Leuven, Belgium (IE, JL); Laboratory of Biological Psychology, KU Leuven, Leuven, Belgium (IE, RD); Department of Pediatrics, Pediatric Hemato-Oncology, KU Leuven, University Hospital Leuven, Leuven, Belgium (CS, AU, JL); Department of Imaging and Pathology, KU Leuven, Leuven, Belgium (SD). The funder had no role in the design of the study; the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication. The authors have no conflicts of interest to disclose. We are grateful to the participants who contributed to this study and to Elise Bossuyt, Charlotte van Soest, Trui Vercruysse, Linde Van den Wyngaert, and Dorothee Vercruysse, who have made this work possible. The authors are grateful to the pediatric oncology team for their dedicated care for childhood cancer patients. We are grateful to Prof. Dr. Veerle Labarque for clinical advice and Prof. Dr. Stefaan van Gool for logistic and intellectual support. Finally, we thank Prof. Dr. Stefan Sunaert for sharing his knowledge and insights as an expert neuroradiologist. References 1 Bisen-Hersh EB, Hineline PN, Walker EA. Disruption of learning processes by chemotherapeutic agents in childhood survivors of acute lymphoblastic leukemia and preclinical models. J Cancer.  2011; 2: 292– 301. http://dx.doi.org/10.7150/jca.2.292 Google Scholar CrossRef Search ADS PubMed  2 Butler RW, Mulhern RK. Neurocognitive interventions for children and adolescents surviving cancer. J Pediatr Psychol.  2005; 30( 1): 65– 78. http://dx.doi.org/10.1093/jpepsy/jsi017 Google Scholar CrossRef Search ADS PubMed  3 Mariotto AB, Rowland JH, Yabroff KR, et al.   Long-term survivors of childhood cancers in the United States. Cancer Epidemiol Biomarkers Prev.  2009; 18( 4): 1033– 1040. http://dx.doi.org/10.1158/1055-9965.EPI-08-0988 Google Scholar CrossRef Search ADS PubMed  4 Lockwood K, Bell T, Colegrove R. Long-term effects of cranial radiation therapy on attention functioning in survivors of childhood leukemia. J Pediatr Psychol.  1999; 24( 1): 55– 66. http://dx.doi.org/10.1093/jpepsy/24.1.55 Google Scholar CrossRef Search ADS   5 Aukema EJ, Caan MW, Oudhuis N, et al.   White matter fractional anisotropy correlates with speed of processing and motor speed in young childhood cancer survivors. Int J Radiat Oncol Biol Phys.  2009; 74( 3): 837– 843. http://dx.doi.org/10.1016/j.ijrobp.2008.08.060 Google Scholar CrossRef Search ADS PubMed  6 D'Angio GJ. Pediatric cancer in perspective: Cure is not enough. Cancer.  1975; 35(3 suppl): 866– 870. Google Scholar CrossRef Search ADS   7 Waber DP. Central nervous system late effects: A new frontier? Pediatr Blood Cancer.  2011; 57( 3): 355– 356. http://dx.doi.org/10.1002/pbc.23206 Google Scholar CrossRef Search ADS PubMed  8 Daams M, Schuitema I, van Dijk BW, et al.   Long-term effects of cranial irradiation and intrathecal chemotherapy in treatment of childhood leukemia: A MEG study of power spectrum and correlated cognitive dysfunction. BMC Neurol.  2012; 12: 84. http://dx.doi.org/10.1186/1471-2377-12-84 Google Scholar CrossRef Search ADS PubMed  9 Schuitema I, Deprez S, Van Hecke W, et al.   Accelerated aging, decreased white matter integrity, and associated neuropsychological dysfunction 25 years after pediatric lymphoid malignancies. J Clin Oncol.  2013; 31( 27): 3378– 3388. Google Scholar CrossRef Search ADS PubMed  10 Van Dongen-Melman JE, De Groot A, Van Dongen JJ, et al.   Cranial irradiation is the major cause of learning problems in children treated for leukemia and lymphoma: A comparative study. Leukemia.  1997; 11( 8): 1197– 1200. Google Scholar CrossRef Search ADS PubMed  11 Askins MA, Moore BD3rd. Preventing neurocognitive late effects in childhood cancer survivors. J Child Neurol.  2008; 23( 10): 1160– 1171. http://dx.doi.org/10.1177/0883073808321065 Google Scholar CrossRef Search ADS PubMed  12 Anderson VA, Godber T, Smibert E, et al.   Cognitive and academic outcome following cranial irradiation and chemotherapy in children: A longitudinal study. Br J Cancer.  2000; 82( 2): 255– 262. Google Scholar CrossRef Search ADS PubMed  13 Campbell LK, Scaduto M, Sharp W, et al.   A meta-analysis of the neurocognitive sequelae of treatment for childhood acute lymphocytic leukemia. Pediatr Blood Cancer.  2007; 49( 1): 65– 73. http://dx.doi.org/10.1002/pbc.20860 Google Scholar CrossRef Search ADS PubMed  14 Duffner PK. Long-term effects of radiation therapy on cognitive and endocrine function in children with leukemia and brain tumors. Neurologist.  2004; 10( 6): 293– 310. http://dx.doi.org/10.1097/01.nrl.0000144287.35993.96 Google Scholar CrossRef Search ADS PubMed  15 Iyer NS, Balsamo LM, Bracken MB, et al.   Chemotherapy-only treatment effects on long-term neurocognitive functioning in childhood ALL survivors: A review and meta-analysis. Blood.  2015; 126( 3): 346– 353. http://dx.doi.org/10.1182/blood-2015-02-627414 Google Scholar CrossRef Search ADS PubMed  16 Buizer AI, de Sonneville LM, Veerman AJ. Effects of chemotherapy on neurocognitive function in children with acute lymphoblastic leukemia: A critical review of the literature. Pediatr Blood Cancer.  2009; 52( 4): 447– 454. http://dx.doi.org/10.1002/pbc.21869 Google Scholar CrossRef Search ADS PubMed  17 Cheung YT, Krull KR. Neurocognitive outcomes in long-term survivors of childhood acute lymphoblastic leukemia treated on contemporary treatment protocols: A systematic review. Neurosci Biobehav Rev.  2015; 53: 108– 120. http://dx.doi.org/10.1016/j.neubiorev.2015.03.016 Google Scholar CrossRef Search ADS PubMed  18 Mennes M, Stiers P, Vandenbussche E, et al.   Attention and information processing in survivors of childhood acute lymphoblastic leukemia treated with chemotherapy only. Pediatr Blood Cancer.  2005; 44( 5): 478– 486. Google Scholar CrossRef Search ADS PubMed  19 Kanellopoulos A, Andersson S, Zeller B, et al.   Neurocognitive outcome in very long-term survivors of childhood acute lymphoblastic leukemia after treatment with chemotherapy only. Pediatr Blood Cancer.  2016; 63( 1): 133– 138. http://dx.doi.org/10.1002/pbc.25690 Google Scholar CrossRef Search ADS PubMed  20 Krull KR, Brinkman TM, Li C, et al.   Neurocognitive outcomes decades after treatment for childhood acute lymphoblastic leukemia: A report from the St Jude lifetime cohort study. J Clin Oncol.  2013; 31( 35): 4407– 4415. http://dx.doi.org/10.1200/JCO.2012.48.2315 Google Scholar CrossRef Search ADS PubMed  21 Caron JE, Krull KR, Hockenberry M, et al.   Oxidative stress and executive function in children receiving chemotherapy for acute lymphoblastic leukemia. Pediatr Blood Cancer.  2009; 53( 4): 551– 556. http://dx.doi.org/10.1002/pbc.22128 Google Scholar CrossRef Search ADS PubMed  22 Krull KR, Cheung YT, Liu W, et al.   Chemotherapy pharmacodynamics and neuroimaging and neurocognitive outcomes in long-term survivors of childhood acute lymphoblastic leukemia. J Clin Oncol.  2016; 34( 22): 2644– 2653. Google Scholar CrossRef Search ADS PubMed  23 Ahles TA. Brain vulnerability to chemotherapy toxicities. Psychooncology . 2012; 21( 11): 1141– 1148. http://dx.doi.org/10.1002/pon.3196 Google Scholar CrossRef Search ADS PubMed  24 Carey ME, Haut MW, Reminger SL, et al.   Reduced frontal white matter volume in long-term childhood leukemia survivors: A voxel-based morphometry study. Am J Neuroradiol.  2008; 29( 4): 792– 797. http://dx.doi.org/10.3174/ajnr.A0904 Google Scholar CrossRef Search ADS   25 Reddick WE, Glass JO, Johnson DP, et al.   Voxel-based analysis of T2 hyperintensities in white matter during treatment of childhood leukemia. Am J Neuroradiol.  2009; 30( 10): 1947– 1954. http://dx.doi.org/10.3174/ajnr.A1733 Google Scholar CrossRef Search ADS   26 Genschaft M, Huebner T, Plessow F, et al.   Impact of chemotherapy for childhood leukemia on brain morphology and function. PLoS One.  2013; 8( 11): e78599. Google Scholar CrossRef Search ADS PubMed  27 Morioka S, Morimoto M, Yamada K, et al.   Effects of chemotherapy on the brain in childhood: Diffusion tensor imaging of subtle white matter damage. Neuroradiology.  2013; 55( 10): 1251– 1257. http://dx.doi.org/10.1007/s00234-013-1245-7 Google Scholar CrossRef Search ADS PubMed  28 Kingma A, van Dommelen RI, Mooyaart EL, et al.   Slight cognitive impairment and magnetic resonance imaging abnormalities but normal school levels in children treated for acute lymphoblastic leukemia with chemotherapy only. J Pediatr.  2001; 139( 3): 413– 420. http://dx.doi.org/10.1067/mpd.2001.117066 Google Scholar CrossRef Search ADS PubMed  29 Pastura G, Doering T, Gasparetto EL, et al.   Exploratory analysis of diffusion tensor imaging in children with attention deficit hyperactivity disorder: Evidence of abnormal white matter structure. Atten Defic Hyperact Disord.  2015; 8( 2): 65– 71. Google Scholar CrossRef Search ADS PubMed  30 ElAlfy M, Ragab I, Azab I, et al.   Neurocognitive outcome and white matter anisotropy in childhood acute lymphoblastic leukemia survivors treated with different protocols. Pediatr Hematol Oncol.  2014; 31( 2): 194– 204. http://dx.doi.org/10.3109/08880018.2013.871763 Google Scholar CrossRef Search ADS PubMed  31 Hill DE, Ciesielski KT, Hart BL, et al.   MRI morphometric and neuropsychological correlates of long-term memory in survivors of childhood leukemia. Pediatr Blood Cancer.  2004; 42( 7): 611– 617. http://dx.doi.org/10.1002/pbc.20004 Google Scholar CrossRef Search ADS PubMed  32 Montour-Proulx I, Kuehn SM, Keene DL, et al.   Cognitive changes in children treated for acute lymphoblastic leukemia with chemotherapy only according to the Pediatric Oncology Group 9605 protocol. J Child Neurol.  2005; 20( 2): 129– 133. http://dx.doi.org/10.1177/08830738050200020901 Google Scholar CrossRef Search ADS PubMed  33 Beaulieu C, Allen PS. Determinants of anisotropic water diffusion in nerves. Magnet Res Med.  1994; 31( 4): 394– 400. http://dx.doi.org/10.1002/mrm.1910310408 Google Scholar CrossRef Search ADS   34 Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magnet Res Ser B.  1994; 103( 3): 247– 254. http://dx.doi.org/10.1006/jmrb.1994.1037 Google Scholar CrossRef Search ADS   35 Basser PJ. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR Biomed.  1995; 8( 7–8): 333– 344. http://dx.doi.org/10.1002/nbm.1940080707 Google Scholar CrossRef Search ADS PubMed  36 Deprez S, Billiet T, Sunaert S, et al.   Diffusion tensor MRI of chemotherapy-induced cognitive impairment in non-CNS cancer patients: A review. Brain Imaging Behav.  2013; 7( 4): 409– 435. Google Scholar CrossRef Search ADS PubMed  37 Khong PL, Leung LH, Fung AS, et al.   White matter anisotropy in post-treatment childhood cancer survivors: Preliminary evidence of association with neurocognitive function. J Clin Oncol.  2006; 24( 6): 884– 890. http://dx.doi.org/10.1200/JCO.2005.02.4505 Google Scholar CrossRef Search ADS PubMed  38 Edelmann MN, Krull KR, Liu W, et al.   Diffusion tensor imaging and neurocognition in survivors of childhood acute lymphoblastic leukaemia. Brain.  2014; 137(Pt 11): 2973– 2983. Google Scholar CrossRef Search ADS   39 Jensen JH, Helpern JA, Ramani A, et al.   Diffusional kurtosis imaging: The quantification of non-gaussian water diffusion by means of magnetic resonance imaging. Magnet Res Med.  2005; 53( 6): 1432– 1440. http://dx.doi.org/10.1002/mrm.20508 Google Scholar CrossRef Search ADS   40 Jensen JH, Helpern JA. MRI quantification of non-Gaussian water diffusion by kurtosis analysis. NMR Biomed.  2010; 23( 7): 698– 710. http://dx.doi.org/10.1002/nbm.1518 Google Scholar CrossRef Search ADS PubMed  41 Zhang H, Schneider T, Wheeler-Kingshott CA, et al.   NODDI: Practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage.  2012; 61( 4): 1000– 1016. http://dx.doi.org/10.1016/j.neuroimage.2012.03.072 Google Scholar CrossRef Search ADS PubMed  42 MacKay A, Whittall K, Adler J, et al.   In vivo visualization of myelin water in brain by magnetic resonance. Magnet Res Med.  1994; 31( 6): 673– 677. http://dx.doi.org/10.1002/mrm.1910310614 Google Scholar CrossRef Search ADS   43 de Kouchkovsky I, Fieremans E, Fleysher L, et al.   Quantification of normal-appearing white matter tract integrity in multiple sclerosis: A diffusion kurtosis imaging study. J Neurol.  2016; 263( 6): 1146– 1155. http://dx.doi.org/10.1007/s00415-016-8118-z Google Scholar CrossRef Search ADS PubMed  44 Caverzasi E, Papinutto N, Castellano A, et al.   Neurite orientation dispersion and density imaging color maps to characterize brain diffusion in neurologic disorders. J Neuroimaging.  2016; 26( 5): 494– 498. Google Scholar CrossRef Search ADS PubMed  45 Damoiseaux JS. Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci U S A.  2006; 103( 37): 13848– 13853. http://dx.doi.org/10.1073/pnas.0601417103 Google Scholar CrossRef Search ADS PubMed  46 Greicius MD, Krasnow B, Reiss AL, et al.   Functional connectivity in the resting brain: A network analysis of the default mode hypothesis. Proc Natl Acad Sci U S A.  2003; 100( 1): 253– 258. http://dx.doi.org/10.1073/pnas.0135058100 Google Scholar CrossRef Search ADS PubMed  47 van den Heuvel MP, Mandl RC, Kahn RS, et al.   Functionally linked resting-state networks reflect the underlying structural connectivity architecture of the human brain. Hum Brain Mapp.  2009; 30( 10): 3127– 3141. http://dx.doi.org/10.1002/hbm.20737 Google Scholar CrossRef Search ADS PubMed  48 Raichle ME. The brain's default mode network. Annu Rev Neurosci.  2015; 38: 433– 447. http://dx.doi.org/10.1146/annurev-neuro-071013-014030 Google Scholar CrossRef Search ADS PubMed  49 Shulman GL, Fiez JA, Corbetta M, et al.   Common blood flow changes across visual tasks: II. Decreases in cerebral cortex. J Cogn Neurosci.  1997; 9( 5): 648– 663. Google Scholar CrossRef Search ADS PubMed  50 Kwong KK, Belliveau JW, Chesler DA, et al.   Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci U S A.  1992; 89( 12): 5675– 5679. http://dx.doi.org/10.1073/pnas.89.12.5675 Google Scholar CrossRef Search ADS PubMed  51 Frahm J, Bruhn H, Merboldt KD, et al.   Dynamic MR imaging of human brain oxygenation during rest and photic stimulation. J Magn Reson Imaging.  1992; 2( 5): 501– 505. http://dx.doi.org/10.1002/jmri.1880020505 Google Scholar CrossRef Search ADS PubMed  52 Raichle ME, MacLeod AM, Snyder AZ, et al.   A default mode of brain function. Proc Natl Acad Sci U S A.  2001; 98( 2): 676– 682. http://dx.doi.org/10.1073/pnas.98.2.676 Google Scholar CrossRef Search ADS PubMed  53 Fox MD, Snyder AZ, Vincent JL, et al.   The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A.  2005; 102( 27): 9673– 9678. http://dx.doi.org/10.1073/pnas.0504136102 Google Scholar CrossRef Search ADS PubMed  54 Spreng RN, Stevens WD, Chamberlain JP, et al.   Default network activity, coupled with the frontoparietal control network, supports goal-directed cognition. Neuroimage.  2010; 53( 1): 303– 317. http://dx.doi.org/10.1016/j.neuroimage.2010.06.016 Google Scholar CrossRef Search ADS PubMed  55 Spreng RN, Sepulcre J, Turner GR, et al.   Intrinsic architecture underlying the relations among the default, dorsal attention, and frontoparietal control networks of the human brain. J Cogn Neurosci.  2013; 25( 1): 74– 86. http://dx.doi.org/10.1162/jocn_a_00281 Google Scholar CrossRef Search ADS PubMed  56 Smallwood J, Brown K, Baird B, et al.   Cooperation between the default mode network and the frontal-parietal network in the production of an internal train of thought. Brain Res.  2012; 1428: 60– 70. http://dx.doi.org/10.1016/j.brainres.2011.03.072 Google Scholar CrossRef Search ADS PubMed  57 Bruno J, Hosseini SM, Kesler S. Altered resting state functional brain network topology in chemotherapy-treated breast cancer survivors. Neurobiol Dis.  2012; 48( 3): 329– 338. http://dx.doi.org/10.1016/j.nbd.2012.07.009 Google Scholar CrossRef Search ADS PubMed  58 Kesler SR, Wefel JS, Hosseini SM, et al.   Default mode network connectivity distinguishes chemotherapy-treated breast cancer survivors from controls. Proc Natl Acad Sci U S A.  2013; 110( 28): 11600– 11605. http://dx.doi.org/10.1073/pnas.1214551110 Google Scholar CrossRef Search ADS PubMed  59 Kesler SR, Watson CL, Blayney DW. Brain network alterations and vulnerability to simulated neurodegeneration in breast cancer. Neurobiol Aging.  2015; 36( 8): 2429– 2442. http://dx.doi.org/10.1016/j.neurobiolaging.2015.04.015 Google Scholar CrossRef Search ADS PubMed  60 Hampson JP, Zick SM, Khabir T, et al.   Altered resting brain connectivity in persistent cancer related fatigue. Neuroimage Clin.  2015; 8: 305– 313. http://dx.doi.org/10.1016/j.nicl.2015.04.022 Google Scholar CrossRef Search ADS PubMed  61 Kesler SR, Gugel M, Pritchard-Berman M, et al.   Altered resting state functional connectivity in young survivors of acute lymphoblastic leukemia. Pediatr Blood Cancer.  2014; 61( 7): 1295– 1299. http://dx.doi.org/10.1002/pbc.25022 Google Scholar CrossRef Search ADS PubMed  62 Steinberg S, Hartmann R, Wisniewski S, et al.   (Late sequelae of CNS recurrence of acute lymphoblastic leukemia in childhood). Klin Padiatr.  1998; 210( 4): 200– 206. Google Scholar CrossRef Search ADS PubMed  63 Iuvone L, Mariotti P, Colosimo C, et al.   Long-term cognitive outcome, brain computed tomography scan, and magnetic resonance imaging in children cured for acute lymphoblastic leukemia. Cancer.  2002; 95( 12): 2562– 2570. http://dx.doi.org/10.1002/cncr.10999 Google Scholar CrossRef Search ADS PubMed  64 von der Weid N, Mosimann I, Hirt A, et al.   Intellectual outcome in children and adolescents with acute lymphoblastic leukaemia treated with chemotherapy alone: Age- and sex-related differences. Eur J Cancer.  2003; 39( 3): 359– 365. http://dx.doi.org/10.1016/S0959-8049(02)00260-5 Google Scholar CrossRef Search ADS PubMed  65 Elens I, Deprez S, Danckaerts M, et al.   Neurocognitive sequelae in adult childhood-leukemia survivors related to levels of phosphorylated Tau. J Natl Cancer Inst.  2017; 109( 7):djw321. 66 Renard M, Suciu S, Bertrand Y, et al.   Second neoplasm in children treated in EORTC 58881 trial for acute lymphoblastic malignancies: Low incidence of CNS tumours. Pediatr Blood Cancer.  2011; 57( 1): 119– 125. http://dx.doi.org/10.1002/pbc.23083 Google Scholar CrossRef Search ADS PubMed  67 Domenech C, Suciu S, De Moerloose B, et al.   Dexamethasone (6 mg/m2/day) and prednisolone (60 mg/m2/day) were equally effective as induction therapy for childhood acute lymphoblastic leukemia in the EORTC CLG 58951 randomized trial. Haematologica.  2014; 99( 7): 1220– 1227. http://dx.doi.org/10.3324/haematol.2014.103507 Google Scholar CrossRef Search ADS PubMed  68 Poot DH, den Dekker AJ, Achten E, et al.   Optimal experimental design for diffusion kurtosis imaging. IEEE Trans Med Imaging.  2010; 29( 3): 819– 829. http://dx.doi.org/10.1109/TMI.2009.2037915 Google Scholar CrossRef Search ADS PubMed  69 Miatton M, De Wolf D, François K, et al.   Neuropsychological performance in school-aged children with surgically corrected congenital heart disease. J Pediatr.  2007; 151( 1): 73– 78, 78.e1. Google Scholar CrossRef Search ADS PubMed  70 de Sonneville LMJ. Amsterdam Neuropsychological Tasks: a computer-aided assessment program. In den Brinker BPML, Beek PJ, Brand AN, Maarse SJ, Mulder LJM, eds. Cognitive Ergonomics, Clinical Assessment and Computer-Aided Learning: Computers in Psychology. Lisse: Swets & Zeitlinger BV;1999:187–203. 71 Van Der Elst W, Van Boxtel M, Van Breukelen G, et al.   Rey’s verbal learning test: Normative data for 1855 healthy participants aged 24–81 years and the influence of age, sex, education, and mode of presentation. J Int Neuropsychol Soc.  2005; 11: 290– 302. Google Scholar CrossRef Search ADS PubMed  72 Uddin LQ, Kelly AM, Biswal BB, et al.   Functional connectivity of default mode network components: correlation, anticorrelation, and causality. Hum Brain Mapp.  2009; 30( 2): 625– 637. http://dx.doi.org/10.1002/hbm.20531 Google Scholar CrossRef Search ADS PubMed  73 Kesler SR. Default mode network as a potential biomarker of chemotherapy-related brain injury. Neurobiol Aging.  2014; 35(suppl 2): S11– S19. Google Scholar CrossRef Search ADS PubMed  74 Sambataro F, Murty VP, Callicott JH, et al.   Age-related alterations in default mode network: Impact on working memory performance. Neurobiol Aging.  2010; 31( 5): 839– 852. http://dx.doi.org/10.1016/j.neurobiolaging.2008.05.022 Google Scholar CrossRef Search ADS PubMed  75 Deprez S, Amant F, Smeets A, et al.   Longitudinal assessment of chemotherapy-induced structural changes in cerebral white matter and its correlation with impaired cognitive functioning. J Clin Oncol.  2012; 30( 3): 274– 281. http://dx.doi.org/10.1200/JCO.2011.36.8571 Google Scholar CrossRef Search ADS PubMed  76 Porto L, Preibisch C, Hattingen E, et al.   Voxel-based morphometry and diffusion-tensor MR imaging of the brain in long-term survivors of childhood leukemia. Eur Radiol.  2008; 18( 11): 2691– 2700. http://dx.doi.org/10.1007/s00330-008-1038-2 Google Scholar CrossRef Search ADS PubMed  77 Dellani PR, Eder S, Gawehn J, et al.   Late structural alterations of cerebral white matter in long-term survivors of childhood leukemia. J Magn Reson Imaging.  2008; 27( 6): 1250– 1255. http://dx.doi.org/10.1002/jmri.21364 Google Scholar CrossRef Search ADS PubMed  78 Dietrich J, Han R, Yang Y, et al.   CNS progenitor cells and oligodendrocytes are targets of chemotherapeutic agents in vitro and in vivo. J Biol.  2006; 5( 7): 22. http://dx.doi.org/10.1186/jbiol50 Google Scholar CrossRef Search ADS PubMed  79 Perry ME, McDonald CR, Hagler DJ, et al.   White matter tracts associated with set-shifting in healthy aging. Neuropsychologia.  2009; 47( 13): 2835– 2842. http://dx.doi.org/10.1016/j.neuropsychologia.2009.06.008 Google Scholar CrossRef Search ADS PubMed  80 Braak H, Braak E. Development of Alzheimer-related neurofibrillary changes in the neocortex inversely recapitulates cortical myelogenesis. Acta Neuropathol.  1996; 92( 2): 197– 201. http://dx.doi.org/10.1007/s004010050508 Google Scholar CrossRef Search ADS PubMed  81 Reisberg B, Franssen EH, Hasan SM, et al.   Retrogenesis: Clinical, physiologic, and pathologic mechanisms in brain aging, Alzheimer's and other dementing processes. Eur Arch Psychiatry Clin Neurosci.  1999; 249(suppl 3): 28– 36. Google Scholar CrossRef Search ADS PubMed  82 Rodriguez M. A function of myelin is to protect axons from subsequent injury: Implications for deficits in multiple sclerosis. Brain.  2003; 126(Pt 4): 751– 752. Google Scholar CrossRef Search ADS   83 Urger SE, De Bellis MD, Hooper SR, et al.   The superior longitudinal fasciculus in typically developing children and adolescents: Diffusion tensor imaging and neuropsychological correlates. J Child Neurol.  2015; 30( 1): 9– 20. http://dx.doi.org/10.1177/0883073813520503 Google Scholar CrossRef Search ADS PubMed  84 Parker GJ, Luzzi S, Alexander DC, et al.   Lateralization of ventral and dorsal auditory-language pathways in the human brain. Neuroimage.  2005; 24( 3): 656– 666. http://dx.doi.org/10.1016/j.neuroimage.2004.08.047 Google Scholar CrossRef Search ADS PubMed  85 Lord LD, Expert P, Huckins JF, et al.   Cerebral energy metabolism and the brain's functional network architecture: An integrative review. J Cereb Blood Flow Metab.  2013; 33( 9): 1347– 1354. http://dx.doi.org/10.1038/jcbfm.2013.94 Google Scholar CrossRef Search ADS PubMed  86 Kesler SR, Gugel M, Huston-Warren E, et al.   Atypical structural connectome organization and cognitive impairment in young survivors of acute lymphoblastic leukemia. Brain Connect.  2016; 6( 4): 273– 282. http://dx.doi.org/10.1089/brain.2015.0409 Google Scholar CrossRef Search ADS PubMed  87 Waber D, Bernstein J, Kammerer B, et al.   Neuropsychological diagnostic profiles of children who received CNS treatment for acute lymphoblastic-leukemia: The systemic approach to assessment. Dev Neuropsychol.  1992; 8( 1): 28. Google Scholar CrossRef Search ADS   88 Kesler SR, Gugel M, Pritchard-Berman M, et al.   Altered resting state functional connectivity in young survivors of acute lymphoblastic leukemia. Pediatric Blood Cancer.  2014; 61( 7): 1295– 1299. http://dx.doi.org/10.1002/pbc.25022 Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JNCI: Journal of the National Cancer Institute Oxford University Press

Brain Connectivity and Cognitive Flexibility in Nonirradiated Adult Survivors of Childhood Leukemia

Loading next page...
 
/lp/ou_press/brain-connectivity-and-cognitive-flexibility-in-nonirradiated-adult-qvKrwvWj8U
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
0027-8874
eISSN
1460-2105
D.O.I.
10.1093/jnci/djy009
Publisher site
See Article on Publisher Site

Abstract

Abstract Background This study aimed to assess functional and structural brain connectivity in adult childhood leukemia survivors and the link with cognitive functioning and previously identified risk factors such as intrathecal methotrexate dose and age at start of therapy. Methods Thirty-one nonirradiated adult childhood leukemia survivors and 35 controls underwent cognitive testing and multimodal magnetic resonance imaging (resting state functional MRI, T1-weighted, diffusion-weighted, and myelin water imaging [MWI]). Analyses included dual regression, voxel-based morphometry, advanced diffusion, and MWI modeling techniques besides stepwise discriminant function analysis to identify the most affected executive cognitive domain. Correlations with discrete intrathecal MTX doses and (semi)continuous variables were calculated using Spearman’s rank and Pearson’s correlation, respectively. All correlation tests were two-sided. Positive and negative T-contrasts in functional and structural MRI analysis were one-sided. Results Survivors demonstrated lower functional connectivity between the default mode network (DMN) and inferior temporal gyrus (ITG; P < .008). Additionally, we observed higher fractional anisotropy (FA; P = .04) and lower orientation dispersion index (ODI; P = .008) at the left centrum semiovale, which could—given that several fiber bundles cross this region—suggest selective reduced integrity of the respective white matter tracts. Set shifting reaction time, a measure of cognitive flexibility, was mostly impaired and correlated with lower FA (r = –0.53, P = .003) and higher ODI (r = 0.40, P = .04) in survivors but not with DMN-ITG connectivity. There were no statistically significant differences between survivors and controls in WM or GM volume, nor was there a statistically significant correlation between imaging measurements and age at start of therapy or intrathecal methotrexate dose. Conclusions Adult, nonirradiated childhood leukemia survivors show altered brain connectivity, which is linked with cognitive flexibility. Central nervous system (CNS)–directed prophylaxis to prevent early relapse has improved survival rates in childhood leukemia (1–3), increasing the attention for long-term sequelae due to such prophylactic treatment. The detrimental impact of cranial irradiation on the developing brain (1,2,4–14) has resulted in its successful replacement by intrathecal and high-dose intravenous methotrexate. However, long-term neurocognitive deficits remain a clinical concern even in nonirradiated survivors, albeit to a milder extent. Because executive functioning appeared mostly affected in nonirradiated childhood cancer survivors (1–3,7,11,15–22), white matter (WM) tracts are suggested to be especially at risk for chemotherapy-induced damage (23) and related processing speed, which has been confirmed by imaging studies. Neuro-imaging findings in nonirradiated survivors of childhood leukemia range from gross calcifications and WM hyperintensities to more subtle changes in WM volumes (22,24–32). While voxel-based morphometry (VBM) can be applied to study volumetric differences, several techniques exist to study microstructural WM alterations. One imaging technique is diffusion MRI (dMRI), which probes the random motion of water molecules, directed by axonal membranes and modulated by the myelin sheath (33). Building on dMRI data, the diffusion tensor imaging (DTI) (34,35) model quantifies mean diffusivity (MD) and fractional anisotropy (FA). Because intact white matter exhibits highly directional diffusion, decreased FA may indicate injury to a particular tract (36). Previous studies in childhood leukemia survivors have, however, reported inconsistent findings with regard to FA measures. Some authors confirmed statistically significantly reduced FA in nonirradiated childhood leukemia survivors (5,27,30), whereas others did not (9,26,37) or reported even increased FA (38). Admittedly, some assumptions of the DTI model could be inadequate for biological tissue. Different time points after treatment also obscure the DTI parameters, which complicates comparisons between studies. Furthermore, dMRI might not be sensitive enough to detect the more subtle effects of chemotherapy compared with cranial irradiation. Hence, advanced dMRI models may clarify the aforementioned conflicting findings. One such model is diffusion kurtosis imaging (DKI), which quantifies non-Gaussian diffusion, believed to reflect tissue compartmentalization (39,40). More recently, neurite orientation dispersion and density imaging (NODDI) (41) was developed, an approach modeling the contribution of cerebrospinal fluid (Fiso), axonal vs extra-axonal tissue fraction (Vic) to the diffusion signal. It also models the axonal orientation dispersion index (ODI), which is inversely related to FA. Given its pivotal role in signal transmission and protection of fiber bundles, recent techniques focus on myelin content. One such technique is myelin water imaging (MWI), estimating the myelin water fraction (MWF), a potential marker for myelin content (42). Although demonstrating promising results in several neurological conditions (43,44), none of these listed techniques have been implemented yet in the study of chemotherapy-induced brain damage in childhood leukemia survivors. Besides potentially altered structural brain connectivity, little is known about the effect of chemotherapeutics on functional brain connectivity in acute lymphoblastic leukemia (ALL) survivors. Typically, functional connectivity can be studied by using resting state functional MRI (rsfMRI). This technique registers spontaneous blood oxygenation level–dependent (BOLD) signal fluctuations across the brain. Regions that show synchronized BOLD amplitude fluctuations in the absence of an explicit task (45–47) are assumed to constitute functionally linked “resting state networks” (48–51). Given the evidence for executive impairments in childhood leukemia survivors, two functional networks are of particular interest, the default mode network (DMN) and the frontoparietal network (FPN; also referred to as central executive network). The DMN comprises the precuneus, posterior cingulate, medial frontal, middle temporal, and lateral parietal regions (48,49,52,53). The FPN entails the lateral prefrontal cortex, anterior cingulate cortex, inferior parietal lobule, and superior temporal region. It is presumed to play an important role in attention control and set shifting and is highly interconnected with the DMN (53–56). Although several studies have investigated functional connectivity in adult breast cancer patients (57–60), connectivity in child survivors of leukemia was assessed in only one study so far. Specifically, Kesler et al. reported diffuse altered brain connectivity compared with healthy controls, which was related to lower efficiency and slower information processing (61). Given that previous studies reported mainly attentional and executive difficulties in survivors, the aim of our study was to determine the most vulnerable brain regions for potential neurotoxicity in leukemia survivors. To investigate this in detail, the entire brain was investigated using multiple modalities. Our hypotheses were as follows: 1) altered functional brain connectivity in attention-related networks in survivors (investigated using rsfMRI), 2) structural gray and white matter brain changes (using VBM, dMRI, and MWI), and 3) the prediction of cognitive performance based on statistically significantly altered regions (ie, as the result of the neuroimaging analyses of the rsfMRI, VBM, dMRI, or MWI measurements), previously identified risk factors such as intrathecal dose of methotrexate (32,62,63), and age at start of therapy (2,11,15,21,64,65). Methods Participants Thirty-one adult, nonirradiated childhood leukemia survivors (mean age at diagnosis = 6.4 years, range = 17–30 years), treated between 1994 and 2004 according to the European Organisation for Research and Treatment of Cancer 58881 or 58951 protocol (66,67) at the University Hospital of Leuven, were included. Thirty-five control participants were recruited via family members of the survivors and announcements. Demographics of study participants are listed in Table 1. A detailed description of the survivors’ therapeutic background and comparison between treatment protocols can be found in Supplementary Tables 1 and 2 (available online). Each participant signed a written informed consent that was approved by the ethical committee (dossier No. B322201419664) of the University Hospital of Leuven. Table 1. Demographic data of study participants Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    * Statistical analysis was based on two-sided, independent-sample t test. † Socio-economic index score was calculated by multiplying parental occupation scale value by a weight of 5 and parental education scale by 3, and then summing these products as described earlier (69). High score indicates high socio-economic status. Because this index differed statistically significantly between survivors and controls, this parameter was applied as covariate in subsequent analyses. ‡ Scores on the Beck Depression Inventory and State-Trait Anxiety Inventory are presented as raw scores. High score indicates a higher problem rate. § Statistical analysis was based on two-sided, one-way analysis of covariance, with group (survivors vs controls) as an independent factor and a parental socioeconomic status as a covariate. Table 1. Demographic data of study participants Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    Demographic data  Survivors (n = 31)  Controls (n = 35)  P  Females, No. (%)  17 (54.8)  17 (48.5)  --  Age (range), y  20.9 (17–29)  21.5 (16–30)  .49*  Parental socioeconomic status (range) †  41.2 (13–64.5)  50.3 (19–66)  .004*  Beck Depression Inventory (range) ‡  5.4 (0–21)  3.9 (0–15)  .17§   Minimal, No.  29  33     Mild, No.  1  2     Moderate, No.  1  0     Severe, No.  0  0    State-Trait Anxiety Inventory‡ (range)  35.2 (20–57)  34.7 (20–56)  .14§  Neurological/cognitive history, No.      --   Attention deficit disorder  3  2     Dyslexia  3  1     History of drug abuse  1  1     Meningitis  3  0     Trauma capitis  1  2     Epilepsy during treatment  1       Epilepsy since treatment  1       Asymptomatic history of familial Huntington    1    * Statistical analysis was based on two-sided, independent-sample t test. † Socio-economic index score was calculated by multiplying parental occupation scale value by a weight of 5 and parental education scale by 3, and then summing these products as described earlier (69). High score indicates high socio-economic status. Because this index differed statistically significantly between survivors and controls, this parameter was applied as covariate in subsequent analyses. ‡ Scores on the Beck Depression Inventory and State-Trait Anxiety Inventory are presented as raw scores. High score indicates a higher problem rate. § Statistical analysis was based on two-sided, one-way analysis of covariance, with group (survivors vs controls) as an independent factor and a parental socioeconomic status as a covariate. Data Acquisition Magnetic Resonance Imaging Magnetic resonance (MR) scans were acquired in UZ Leuven using a 3T Philips (Achieva) MR-scanner and a 32-channel phased-array head coil. A high-resolution T1-weighted image was acquired for anatomical reference and VBM (MPRAGE, resolution 0.98×0.98 × 1.2 mm3, TE/TR= 4.6 ms/9636 ms, FOV = 192×250×250, 1.2 mm slice thickness, 160 slices). For diffusion data, an echo-planar, multishell imaging scheme was used, consisting of b-values 700, 1000, and 2800 s/mm2, applied along 25, 40, and 75 uniformly distributed directions, respectively, complemented by 10 b = 0 images (voxel size 2.5×2.5×2.5 mm3, TR/TE = 7800 ms/90 ms, FOV= 240×240×125, 50 axial slices, parallel imaging factor 2) (68). For MWI, a 3D GraSE sequence was acquired (FOV 230×80×90 mm, slice thickness 5 mm with 2.5 mm overlap, reconstructed voxel size 0.96×0.96×2.5 mm3, 32 slices, ETL = 32, TE = 10, 20, … , 320 ms, TR = 1 s, EPI factor 3, acceleration factor 2). The resting-state fMRI images were acquired using a T2*-weighted echo-planar imaging (EPI) sequence (resolution 4×4×4 mm3, TE/TR = 33 ms/1700 ms; FOV = 230×120×230 mm; 4 mm slice thickness, 30 slices were assembled in seven minutes), with whole-cerebrum coverage. Two-hundred fifty dynamics of rsfMRI were acquired. Participants were instructed to lay still, close their eyes, and feel at ease, but not to sleep. Image data of one control participants were excluded due to technical problems during the storage process. See Supplementary Table 3 (available online) for image acquisition details. Image Preprocessing Please see the Supplementary Materials (available online) for details. After preprocessing and quality control, the final sample sizes were 30 survivors/33 controls for dMRI analysis, 29 survivors/32 controls for MWI analysis, 31 survivors/34 controls for VBM, and 28 survivors/33 controls for rsfMRI. Neuropsychological Assessment Neuropsychological assessments, including the Amsterdam Neuropsychological Test battery (ANT) (70,71), were performed (for a detailed description, see [65] and Supplementary Table 4, available online). In this study, we focused on the ANT, given the existing evidence for specific executive impairments in childhood cancer survivors (1–3,15). Statistical Analysis For dMRI and MWI data, statistical analyses were performed voxelwise with the Statistical Parametric Mapping toolbox (SPM), using univariate t tests with group, age, and socioeconomic status (SES) as covariates. The contrasts survivors > controls and controls > survivors were thresholded at a P value of less than .001, uncorrected for multiple comparisons. Clusters with family-wise error–corrected P values of less than .05 (Random Field Theory as implemented in SPM) were deemed statistically significant. From clusters yielding statistically significant differences during imaging experiments, average metric values adjusted for age and SES were extracted. The rsfMRI spatial maps obtained from dual regression were tested for differences between groups using FSL randomized nonparametric permutation testing. Similar to the structural analyses, given the multiple voxelwise tests for each component, the analysis was corrected for multiple comparisons across voxels by family-wise error rate (FWER) correction (permutation testing with 5000 permutations as implemented in FSL). Finally, an additional Bonferroni correction was applied, as six group comparisons of interest were performed (3 networks [DMN, left FPN, right FPN]×2 group comparisons [higher vs lower connectivity in the survivor group]). This resulted in an adjusted (ie, more stringent) threshold for the P values (αcorrected = .008 (= .05/6)). Mean time series within regions showing a statistically significant group difference were calculated for each subject and correlated (Pearson correlation) with the mean time series of the associated component network (ie, DMN, left FPN, or right FPN). This correlation was used as a functional connectivity index (72) and correlated with behavior (Figure 1). Statistical analysis of performances on the ANT tasks included two-sided, one-way analysis of covariance (ANCOVA) with group (survivor vs control) as the independent factor and an index of parental socioeconomic status (69) as the covariate. Reaction times were transformed using a logarithmic transformation. Stepwise discriminant function analysis (DFA) was applied using the Wilks Lambda test to demonstrate which variable of executive functioning best distinguished survivors from controls. Outcome measurements from dMRI, MWI, or fMRI significantly differing between survivors and controls were correlated with 1) the neurocognitive measurement best discriminating controls from survivors and 2) predictors of outcome, that is, age at start of therapy and total intrathecal dose of methotrexate. The statistical significance of these correlations was assessed using two-sided Pearson (ANT outcome measurements, age at start of therapy) or Spearman (total dose intrathecal methotrexate) correlation. All correlation tests were two-sided, with a statistical significance level α of .05. Results Functional Connectivity Data After Bonferroni and multiple-voxel correction, dual regression analysis for the DMN resulted in a statistically significantly lower connectivity in survivors between the DMN and the inferior temporal gyrus (ITG; P <.008) (Figure 2). This region of the ITG included 38 voxels. No regions of hyperconnectivity were encountered in survivors. Voxel-wise comparison in connectivity with the FPN did not show any statistically significant differences between both groups. Figure 1. View largeDownload slide Selection of resting state networks. After acquisition and preprocessing, functional resting state networks were selected based on visual inspection by the neuroradiologist. A–C) Right frontoparietal network. D–F) Left frontoparietal network. G–I) Default mode network. A, D, G) Sagittal plane. B, E, H) Coronal plane. C, F, I) Transverse plane. L= left; R = right. Figure 1. View largeDownload slide Selection of resting state networks. After acquisition and preprocessing, functional resting state networks were selected based on visual inspection by the neuroradiologist. A–C) Right frontoparietal network. D–F) Left frontoparietal network. G–I) Default mode network. A, D, G) Sagittal plane. B, E, H) Coronal plane. C, F, I) Transverse plane. L= left; R = right. Figure 2. View largeDownload slide Functional connectivity measurements in survivors vs controls. For all subjects, functional connectivity between the attentional networks (ie, default mode network [DMN], left and right frontoparietal network) and other brain areas was assessed. After Bonferroni and multiple-voxel correction, hypoconnectivity was detected between the DMN and inferior temporal gyrus in survivors (n = 28), compared with controls (n = 33). The group-average DMN is depicted in orange, while the region with hypoconnectivity in survivors is marked in blue. No regions of hyperconnectivity were encountered in survivors. Figure 2. View largeDownload slide Functional connectivity measurements in survivors vs controls. For all subjects, functional connectivity between the attentional networks (ie, default mode network [DMN], left and right frontoparietal network) and other brain areas was assessed. After Bonferroni and multiple-voxel correction, hypoconnectivity was detected between the DMN and inferior temporal gyrus in survivors (n = 28), compared with controls (n = 33). The group-average DMN is depicted in orange, while the region with hypoconnectivity in survivors is marked in blue. No regions of hyperconnectivity were encountered in survivors. Structural Connectivity Data Voxel-based morphometry did not reveal statistically significant differences between survivors and controls in WM and GM volume. However, in a region covering the left centrum semiovale, survivors expressed higher FA than controls (210 voxels, peak T59 = 5.03, cluster PFWE = .04) and lower ODI (165 voxels, peak T59 = 4.26, cluster PFWE = .008). These clusters were largely overlapping and cover areas that are crossed by WM fibers from the superior longitudinal fasciculus (SLF), corticospinal tract (CST), and corpus callosum (CC), as illustrated in Figure 3. A trend for similar changes was also found in the right hemisphere, yet this did not reach statistical significance at cluster level. Other measures describing tissue microstructure (MD, MK, Vic, Fiso, MWF) did not reveal differences between both groups (all PFWE > .05). Figure 3. View largeDownload slide Functional connectivity measurements in survivors vs controls . Clusters of voxels in which survivors (n = 30) and controls (n = 33) differ, overlaid on the fractional anisotropy (FA) template. A) The red-yellow cluster indicates high FA, while the blue clusters indicate low orientation dispersion index (ODI) in patients compared with controls. In the left hemisphere, clusters from both FA and ODI are present and overlap. B) Tractography, seeded from the cluster in the left hemisphere. Figure 3. View largeDownload slide Functional connectivity measurements in survivors vs controls . Clusters of voxels in which survivors (n = 30) and controls (n = 33) differ, overlaid on the fractional anisotropy (FA) template. A) The red-yellow cluster indicates high FA, while the blue clusters indicate low orientation dispersion index (ODI) in patients compared with controls. In the left hemisphere, clusters from both FA and ODI are present and overlap. B) Tractography, seeded from the cluster in the left hemisphere. Executive Functioning Results of cognitive testing were published earlier (65) and revealed that processing speed of FA4L (Th1), MSL (Th1), SAD (Th; focused attention, divided attention, and sustained attention, respectively), set shifting and flexibility SSV (Tc1), and reaction times on working memory SSV (Th3-Th1) and cognitive flexibility SSV (Tc3-Tc1) differed statistically significantly between survivors and controls. DFA demonstrated that set shifting reaction time best discriminated controls from survivors (Wilks Lambda = .87, chi-square (1) = 6.89, P = .009; data not shown). Correlation Between Executive Functioning and Connectivity Data Higher reaction times on the flexibility task corresponded statistically significantly with lower FA (r = –0.53, P = .003) and higher ODI (r = 0.40, P = .04) in survivors, but not in controls (r = –0.34, P = .05; r = 0.28, P = .12, respectively). Lower RSFC index was not correlated with higher reaction time in survivors or in controls (r = –0.08 and –0.08, P = .92 and .69, respectively). For the entire participant group (ie, combination of survivors and controls), no statistically significant correlations between imaging and neurocognitive measurements were encountered. The results are depicted in Table 2 and Figure 4. Finally, neither age at start of therapy nor total dose of intrathecal methotrexate was correlated with FA, ODI, or RSFC (all |r| < 0.27, P > .05). Table 2. Correlations between imaging measurements and set shifting reaction times and predictors of outcome Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  * Statistical significance was tested using two-sided Pearson correlation. DMN = default mode network; FA = fractional anisotropy; ITG = inferior temporal gyrus; ODI = orientation dispersion index; RSFC = resting state functional connectivity. † Statistical significance was tested using two-sided Spearman correlation. Table 2. Correlations between imaging measurements and set shifting reaction times and predictors of outcome Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  Measurement  FA   ODI   RSFC DMN-ITG   r  P  r  P  r  P  Survivors              Set shifting reaction time*  –0.53  .003  0.40  .04  –0.08  .69  Age at start therapy*  –0.10  .61  0.29  .12  0.30  .12  Intrathecal methotrexate†  –0.26  .16  0.05  .79  –0.09  .64  Controls              Set shifting reaction time*  –0.34  .05  0.28  .12  –0.08  .92  * Statistical significance was tested using two-sided Pearson correlation. DMN = default mode network; FA = fractional anisotropy; ITG = inferior temporal gyrus; ODI = orientation dispersion index; RSFC = resting state functional connectivity. † Statistical significance was tested using two-sided Spearman correlation. Figure 4. View largeDownload slide Correlation between diffusion magnetic resonance imaging (dMRI) and cognitive measurements. Correlations were calculated using Pearson’s correlation. Error bars indicate minimal and maximal values, respectively. Numbers of survivors were 30 and 28 for diffusion tensor imaging (DTI) and functional MRI (fMRI) measurements, respectively. Numbers of controls were 33 and 33 for DTI and fMRI measurements, respectively. A–C) Linear regression curves. D–F) Boxplot. Statistical significance was tested using two-sided Pearson correlation. FA = fractional anisotropy; ODI = orientation dispersion index; rsfMRI = resting state functional magnetic resonance imaging; RT SSV = Reaction time (ms) on Set Shifting Visual (attentional computerized task Amsterdam Neuropsychological Tasks). Figure 4. View largeDownload slide Correlation between diffusion magnetic resonance imaging (dMRI) and cognitive measurements. Correlations were calculated using Pearson’s correlation. Error bars indicate minimal and maximal values, respectively. Numbers of survivors were 30 and 28 for diffusion tensor imaging (DTI) and functional MRI (fMRI) measurements, respectively. Numbers of controls were 33 and 33 for DTI and fMRI measurements, respectively. A–C) Linear regression curves. D–F) Boxplot. Statistical significance was tested using two-sided Pearson correlation. FA = fractional anisotropy; ODI = orientation dispersion index; rsfMRI = resting state functional magnetic resonance imaging; RT SSV = Reaction time (ms) on Set Shifting Visual (attentional computerized task Amsterdam Neuropsychological Tasks). Discussion The present study uniquely combines neurocognitive data with structural and functional brain connectivity in nonirradiated childhood survivors on average 15 years after treatment. In these survivors, we report altered WM microstructure at the left centrum semiovale and reduced functional brain connectivity. In addition, measures describing WM microstructure are statistically significantly associated with cognitive performance. There were no differences between survivors and controls in WM or GM volume, or in FPN connectivity. Finally, we did not find a statistically significant correlation between age at start of therapy or intrathecal methotrexate dose and imaging measurements. Altered functional connectivity of the DMN was earlier reported after cancer treatment (57–59,61) and confirms the hypothesis of the DMN as a potential marker for chemotherapy-induced brain injury (73). In this study, a model-free (independent component) analysis was used, in contrast to the study of Kesler and colleagues (who used a seed-based analysis for 90 atlas-based regions). Using this analysis, connectivity between the DMN and other brain regions could be investigated globally. More specifically, the statistically significant difference in connectivity between the DMN and ITG in this study specifically could also suggest altered DMN-FPN (between-network) connectivity, as the ITG region is also part of the FPN. Reduced connectivity of the DMN with other brain areas can explain the observation that cognitive flexibility was mostly affected. Indeed, the DMN plays a pivotal role in allocating brain energy resources to relevant areas and thus in flexibly switching between attentional processes (74), although we could not correlate reaction times during a cognitive flexibility task with DMN-ITG connectivity. Chemosensitivity of the WM has been reported extensively in vivo (5,9,30,37,75–77) and in vitro (78). WM vulnerability might explain both the statistically significant dMRI and fMRI results. In this view, the study of Van den Heuvel et al. is of particular interest. They demonstrated that structural connections within resting state networks—such as the DMN—reflect the connectivity architecture of the brain. More specifically, CC and long associative fibers were identified to play a commending role in connecting the DMN with other brain areas (47). Interestingly, we found altered dMRI measurements at the centrum semiovale, a region where the CST, CC, and SLF—a long associative tract—intercross. Admittedly, reduced WM integrity is most often linked to lower FA and higher ODI measurements. However, the centrum semiovale is a region of so-called crossing fibers. The multidirectional diffusion in such regions a priori leads to a low anisotropy at voxel level (36). High FA and low ODI in the centrum semiovale of survivors can therefore be explained by the selective reduced integrity of one or more crossing fiber bundles, increasing overall anisotropy. Based on our data, we cannot disentangle which bundle(s) is/are most affected, but neurobiological and developmental mechanisms suggest involvement of SLF and/or CC fibers. The sensitivity of these tracts to chemotherapy (5,25,27,29,36) vs the robustness of the CST (79) was indeed described earlier. Retrogenesis —degeneration processes mirroring the order of acquisition in normal development—implies that the latest myelinated fiber tracts are more vulnerable to degeneration (23,36,80,81). This process might be explained by immature myelination of later developing tracts at the time of therapy, as an intact myelin sheath not only facilitates nerve impulse propagation, but also protects axonal fibers against noxious influences (81,82). We therefore believe that the higher FA and lower ODI at the centrum semiovale can be attributed to alterations in the SLF and/or CC. Interestingly, the FA of the SLF was earlier demonstrated to correlate with cognitive flexibility (83), as was the case in our survivors but less so in our control group. This discrepancy could be due to a reduction in crossing fibers in survivors, reestablishing the normal correlation between set shifting reaction time and dMRI measurements that gets obscured in this region for controls because of the multiple crossing fibers. Our results suggest the left hemisphere to be more affected than the right. Neurobiologically, the left hemisphere demonstrates a higher connectivity status (84) and therefore has higher metabolic needs (85). This might result in higher susceptibility of this hemisphere to aging and neurotoxicity (38,79,86). Moreover, Waber et al. reported that age at treatment onset for leukemia patients was related to left hemisphere– vs right hemisphere–dominant deficits. More specifically, patients younger than 36 months at the start of treatment displayed more impairments in right hemisphere–dominant functions, whereas older children showed more deficits related to left hemisphere–dominant functions (87), as is the case in our study. The mean age at diagnosis in our survivor cohort (ie, 6.4 years of age) confirms this observation. We could not confirm the established vulnerability of younger children to chemotherapy (2,11,15,21,61,64,88). However, the fact that the youngest children received a milder treatment resulting from a better prognosis might explain why we could not correlate age at start of therapy with imaging measurements. Nevertheless, the presumed vulnerability of the SLF and/or CC indirectly points to the importance of the maturation status of the brain at the time of chemotherapy. Also, we did not demonstrate the earlier established correlation between total intrathecal methotrexate dose and outcome (30,32,65). This could be due to homogeneity in methotrexate doses. To account for the interference of SES, this parameter was included as a covariate. Also, neurocognitive and imaging data were correlated at the subgroup level (ie, survivors or controls), as chemotherapy-induced brain alterations possibly result in two different populations. In that case, considering both cohorts as a homogenous group would not yield statistically significant findings. Additionally, stringent statistical correction might explain why we found only one region statistically significantly differing between controls and survivors in both DTI and fMRI measurements. Alternatively, subtle findings possibly imply a rather mild impact of chemotherapy on the developing brain after one or two decades, which is in accordance with earlier observations where nonirradiated survivors demonstrated more attenuated deficits than irradiated ones (9). To conclude, we confirm altered DMN connectivity in nonirradiated survivors of childhood leukemia, possibly explained by structural WM sequelae, as evidenced by altered FA and ODI values in the left centrum semiovale. Evidence from the literature and the correlation between imaging and neurocognitive results suggest the SLF and CC—passing through the centrum semiovale—to be especially at risk for chemotherapy-induced damage and indirectly confirm the vulnerability of the young brain to chemotherapy-induced neurotoxicity. Future studies are needed to evaluate the clinical applicability of functional and structural brain connectivity to detect sequelae of chemotherapy during childhood. Funding This study was supported by the charity-based Olivia Hendrickx Research Fund (www.olivia.be). Notes Affiliations of authors: Department of Radiology, University Hospital Leuven, Leuven, Belgium (TB, CS, SD); Icometrix, Leuven, Belgium (TB); Department of Child and Adolescent Psychiatry, KU Leuven, University Psychiatric Centre Leuven, Leuven, Belgium (IE, JL); Laboratory of Biological Psychology, KU Leuven, Leuven, Belgium (IE, RD); Department of Pediatrics, Pediatric Hemato-Oncology, KU Leuven, University Hospital Leuven, Leuven, Belgium (CS, AU, JL); Department of Imaging and Pathology, KU Leuven, Leuven, Belgium (SD). The funder had no role in the design of the study; the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication. The authors have no conflicts of interest to disclose. We are grateful to the participants who contributed to this study and to Elise Bossuyt, Charlotte van Soest, Trui Vercruysse, Linde Van den Wyngaert, and Dorothee Vercruysse, who have made this work possible. The authors are grateful to the pediatric oncology team for their dedicated care for childhood cancer patients. We are grateful to Prof. Dr. Veerle Labarque for clinical advice and Prof. Dr. Stefaan van Gool for logistic and intellectual support. Finally, we thank Prof. Dr. Stefan Sunaert for sharing his knowledge and insights as an expert neuroradiologist. References 1 Bisen-Hersh EB, Hineline PN, Walker EA. Disruption of learning processes by chemotherapeutic agents in childhood survivors of acute lymphoblastic leukemia and preclinical models. J Cancer.  2011; 2: 292– 301. http://dx.doi.org/10.7150/jca.2.292 Google Scholar CrossRef Search ADS PubMed  2 Butler RW, Mulhern RK. Neurocognitive interventions for children and adolescents surviving cancer. J Pediatr Psychol.  2005; 30( 1): 65– 78. http://dx.doi.org/10.1093/jpepsy/jsi017 Google Scholar CrossRef Search ADS PubMed  3 Mariotto AB, Rowland JH, Yabroff KR, et al.   Long-term survivors of childhood cancers in the United States. Cancer Epidemiol Biomarkers Prev.  2009; 18( 4): 1033– 1040. http://dx.doi.org/10.1158/1055-9965.EPI-08-0988 Google Scholar CrossRef Search ADS PubMed  4 Lockwood K, Bell T, Colegrove R. Long-term effects of cranial radiation therapy on attention functioning in survivors of childhood leukemia. J Pediatr Psychol.  1999; 24( 1): 55– 66. http://dx.doi.org/10.1093/jpepsy/24.1.55 Google Scholar CrossRef Search ADS   5 Aukema EJ, Caan MW, Oudhuis N, et al.   White matter fractional anisotropy correlates with speed of processing and motor speed in young childhood cancer survivors. Int J Radiat Oncol Biol Phys.  2009; 74( 3): 837– 843. http://dx.doi.org/10.1016/j.ijrobp.2008.08.060 Google Scholar CrossRef Search ADS PubMed  6 D'Angio GJ. Pediatric cancer in perspective: Cure is not enough. Cancer.  1975; 35(3 suppl): 866– 870. Google Scholar CrossRef Search ADS   7 Waber DP. Central nervous system late effects: A new frontier? Pediatr Blood Cancer.  2011; 57( 3): 355– 356. http://dx.doi.org/10.1002/pbc.23206 Google Scholar CrossRef Search ADS PubMed  8 Daams M, Schuitema I, van Dijk BW, et al.   Long-term effects of cranial irradiation and intrathecal chemotherapy in treatment of childhood leukemia: A MEG study of power spectrum and correlated cognitive dysfunction. BMC Neurol.  2012; 12: 84. http://dx.doi.org/10.1186/1471-2377-12-84 Google Scholar CrossRef Search ADS PubMed  9 Schuitema I, Deprez S, Van Hecke W, et al.   Accelerated aging, decreased white matter integrity, and associated neuropsychological dysfunction 25 years after pediatric lymphoid malignancies. J Clin Oncol.  2013; 31( 27): 3378– 3388. Google Scholar CrossRef Search ADS PubMed  10 Van Dongen-Melman JE, De Groot A, Van Dongen JJ, et al.   Cranial irradiation is the major cause of learning problems in children treated for leukemia and lymphoma: A comparative study. Leukemia.  1997; 11( 8): 1197– 1200. Google Scholar CrossRef Search ADS PubMed  11 Askins MA, Moore BD3rd. Preventing neurocognitive late effects in childhood cancer survivors. J Child Neurol.  2008; 23( 10): 1160– 1171. http://dx.doi.org/10.1177/0883073808321065 Google Scholar CrossRef Search ADS PubMed  12 Anderson VA, Godber T, Smibert E, et al.   Cognitive and academic outcome following cranial irradiation and chemotherapy in children: A longitudinal study. Br J Cancer.  2000; 82( 2): 255– 262. Google Scholar CrossRef Search ADS PubMed  13 Campbell LK, Scaduto M, Sharp W, et al.   A meta-analysis of the neurocognitive sequelae of treatment for childhood acute lymphocytic leukemia. Pediatr Blood Cancer.  2007; 49( 1): 65– 73. http://dx.doi.org/10.1002/pbc.20860 Google Scholar CrossRef Search ADS PubMed  14 Duffner PK. Long-term effects of radiation therapy on cognitive and endocrine function in children with leukemia and brain tumors. Neurologist.  2004; 10( 6): 293– 310. http://dx.doi.org/10.1097/01.nrl.0000144287.35993.96 Google Scholar CrossRef Search ADS PubMed  15 Iyer NS, Balsamo LM, Bracken MB, et al.   Chemotherapy-only treatment effects on long-term neurocognitive functioning in childhood ALL survivors: A review and meta-analysis. Blood.  2015; 126( 3): 346– 353. http://dx.doi.org/10.1182/blood-2015-02-627414 Google Scholar CrossRef Search ADS PubMed  16 Buizer AI, de Sonneville LM, Veerman AJ. Effects of chemotherapy on neurocognitive function in children with acute lymphoblastic leukemia: A critical review of the literature. Pediatr Blood Cancer.  2009; 52( 4): 447– 454. http://dx.doi.org/10.1002/pbc.21869 Google Scholar CrossRef Search ADS PubMed  17 Cheung YT, Krull KR. Neurocognitive outcomes in long-term survivors of childhood acute lymphoblastic leukemia treated on contemporary treatment protocols: A systematic review. Neurosci Biobehav Rev.  2015; 53: 108– 120. http://dx.doi.org/10.1016/j.neubiorev.2015.03.016 Google Scholar CrossRef Search ADS PubMed  18 Mennes M, Stiers P, Vandenbussche E, et al.   Attention and information processing in survivors of childhood acute lymphoblastic leukemia treated with chemotherapy only. Pediatr Blood Cancer.  2005; 44( 5): 478– 486. Google Scholar CrossRef Search ADS PubMed  19 Kanellopoulos A, Andersson S, Zeller B, et al.   Neurocognitive outcome in very long-term survivors of childhood acute lymphoblastic leukemia after treatment with chemotherapy only. Pediatr Blood Cancer.  2016; 63( 1): 133– 138. http://dx.doi.org/10.1002/pbc.25690 Google Scholar CrossRef Search ADS PubMed  20 Krull KR, Brinkman TM, Li C, et al.   Neurocognitive outcomes decades after treatment for childhood acute lymphoblastic leukemia: A report from the St Jude lifetime cohort study. J Clin Oncol.  2013; 31( 35): 4407– 4415. http://dx.doi.org/10.1200/JCO.2012.48.2315 Google Scholar CrossRef Search ADS PubMed  21 Caron JE, Krull KR, Hockenberry M, et al.   Oxidative stress and executive function in children receiving chemotherapy for acute lymphoblastic leukemia. Pediatr Blood Cancer.  2009; 53( 4): 551– 556. http://dx.doi.org/10.1002/pbc.22128 Google Scholar CrossRef Search ADS PubMed  22 Krull KR, Cheung YT, Liu W, et al.   Chemotherapy pharmacodynamics and neuroimaging and neurocognitive outcomes in long-term survivors of childhood acute lymphoblastic leukemia. J Clin Oncol.  2016; 34( 22): 2644– 2653. Google Scholar CrossRef Search ADS PubMed  23 Ahles TA. Brain vulnerability to chemotherapy toxicities. Psychooncology . 2012; 21( 11): 1141– 1148. http://dx.doi.org/10.1002/pon.3196 Google Scholar CrossRef Search ADS PubMed  24 Carey ME, Haut MW, Reminger SL, et al.   Reduced frontal white matter volume in long-term childhood leukemia survivors: A voxel-based morphometry study. Am J Neuroradiol.  2008; 29( 4): 792– 797. http://dx.doi.org/10.3174/ajnr.A0904 Google Scholar CrossRef Search ADS   25 Reddick WE, Glass JO, Johnson DP, et al.   Voxel-based analysis of T2 hyperintensities in white matter during treatment of childhood leukemia. Am J Neuroradiol.  2009; 30( 10): 1947– 1954. http://dx.doi.org/10.3174/ajnr.A1733 Google Scholar CrossRef Search ADS   26 Genschaft M, Huebner T, Plessow F, et al.   Impact of chemotherapy for childhood leukemia on brain morphology and function. PLoS One.  2013; 8( 11): e78599. Google Scholar CrossRef Search ADS PubMed  27 Morioka S, Morimoto M, Yamada K, et al.   Effects of chemotherapy on the brain in childhood: Diffusion tensor imaging of subtle white matter damage. Neuroradiology.  2013; 55( 10): 1251– 1257. http://dx.doi.org/10.1007/s00234-013-1245-7 Google Scholar CrossRef Search ADS PubMed  28 Kingma A, van Dommelen RI, Mooyaart EL, et al.   Slight cognitive impairment and magnetic resonance imaging abnormalities but normal school levels in children treated for acute lymphoblastic leukemia with chemotherapy only. J Pediatr.  2001; 139( 3): 413– 420. http://dx.doi.org/10.1067/mpd.2001.117066 Google Scholar CrossRef Search ADS PubMed  29 Pastura G, Doering T, Gasparetto EL, et al.   Exploratory analysis of diffusion tensor imaging in children with attention deficit hyperactivity disorder: Evidence of abnormal white matter structure. Atten Defic Hyperact Disord.  2015; 8( 2): 65– 71. Google Scholar CrossRef Search ADS PubMed  30 ElAlfy M, Ragab I, Azab I, et al.   Neurocognitive outcome and white matter anisotropy in childhood acute lymphoblastic leukemia survivors treated with different protocols. Pediatr Hematol Oncol.  2014; 31( 2): 194– 204. http://dx.doi.org/10.3109/08880018.2013.871763 Google Scholar CrossRef Search ADS PubMed  31 Hill DE, Ciesielski KT, Hart BL, et al.   MRI morphometric and neuropsychological correlates of long-term memory in survivors of childhood leukemia. Pediatr Blood Cancer.  2004; 42( 7): 611– 617. http://dx.doi.org/10.1002/pbc.20004 Google Scholar CrossRef Search ADS PubMed  32 Montour-Proulx I, Kuehn SM, Keene DL, et al.   Cognitive changes in children treated for acute lymphoblastic leukemia with chemotherapy only according to the Pediatric Oncology Group 9605 protocol. J Child Neurol.  2005; 20( 2): 129– 133. http://dx.doi.org/10.1177/08830738050200020901 Google Scholar CrossRef Search ADS PubMed  33 Beaulieu C, Allen PS. Determinants of anisotropic water diffusion in nerves. Magnet Res Med.  1994; 31( 4): 394– 400. http://dx.doi.org/10.1002/mrm.1910310408 Google Scholar CrossRef Search ADS   34 Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magnet Res Ser B.  1994; 103( 3): 247– 254. http://dx.doi.org/10.1006/jmrb.1994.1037 Google Scholar CrossRef Search ADS   35 Basser PJ. Inferring microstructural features and the physiological state of tissues from diffusion-weighted images. NMR Biomed.  1995; 8( 7–8): 333– 344. http://dx.doi.org/10.1002/nbm.1940080707 Google Scholar CrossRef Search ADS PubMed  36 Deprez S, Billiet T, Sunaert S, et al.   Diffusion tensor MRI of chemotherapy-induced cognitive impairment in non-CNS cancer patients: A review. Brain Imaging Behav.  2013; 7( 4): 409– 435. Google Scholar CrossRef Search ADS PubMed  37 Khong PL, Leung LH, Fung AS, et al.   White matter anisotropy in post-treatment childhood cancer survivors: Preliminary evidence of association with neurocognitive function. J Clin Oncol.  2006; 24( 6): 884– 890. http://dx.doi.org/10.1200/JCO.2005.02.4505 Google Scholar CrossRef Search ADS PubMed  38 Edelmann MN, Krull KR, Liu W, et al.   Diffusion tensor imaging and neurocognition in survivors of childhood acute lymphoblastic leukaemia. Brain.  2014; 137(Pt 11): 2973– 2983. Google Scholar CrossRef Search ADS   39 Jensen JH, Helpern JA, Ramani A, et al.   Diffusional kurtosis imaging: The quantification of non-gaussian water diffusion by means of magnetic resonance imaging. Magnet Res Med.  2005; 53( 6): 1432– 1440. http://dx.doi.org/10.1002/mrm.20508 Google Scholar CrossRef Search ADS   40 Jensen JH, Helpern JA. MRI quantification of non-Gaussian water diffusion by kurtosis analysis. NMR Biomed.  2010; 23( 7): 698– 710. http://dx.doi.org/10.1002/nbm.1518 Google Scholar CrossRef Search ADS PubMed  41 Zhang H, Schneider T, Wheeler-Kingshott CA, et al.   NODDI: Practical in vivo neurite orientation dispersion and density imaging of the human brain. Neuroimage.  2012; 61( 4): 1000– 1016. http://dx.doi.org/10.1016/j.neuroimage.2012.03.072 Google Scholar CrossRef Search ADS PubMed  42 MacKay A, Whittall K, Adler J, et al.   In vivo visualization of myelin water in brain by magnetic resonance. Magnet Res Med.  1994; 31( 6): 673– 677. http://dx.doi.org/10.1002/mrm.1910310614 Google Scholar CrossRef Search ADS   43 de Kouchkovsky I, Fieremans E, Fleysher L, et al.   Quantification of normal-appearing white matter tract integrity in multiple sclerosis: A diffusion kurtosis imaging study. J Neurol.  2016; 263( 6): 1146– 1155. http://dx.doi.org/10.1007/s00415-016-8118-z Google Scholar CrossRef Search ADS PubMed  44 Caverzasi E, Papinutto N, Castellano A, et al.   Neurite orientation dispersion and density imaging color maps to characterize brain diffusion in neurologic disorders. J Neuroimaging.  2016; 26( 5): 494– 498. Google Scholar CrossRef Search ADS PubMed  45 Damoiseaux JS. Consistent resting-state networks across healthy subjects. Proc Natl Acad Sci U S A.  2006; 103( 37): 13848– 13853. http://dx.doi.org/10.1073/pnas.0601417103 Google Scholar CrossRef Search ADS PubMed  46 Greicius MD, Krasnow B, Reiss AL, et al.   Functional connectivity in the resting brain: A network analysis of the default mode hypothesis. Proc Natl Acad Sci U S A.  2003; 100( 1): 253– 258. http://dx.doi.org/10.1073/pnas.0135058100 Google Scholar CrossRef Search ADS PubMed  47 van den Heuvel MP, Mandl RC, Kahn RS, et al.   Functionally linked resting-state networks reflect the underlying structural connectivity architecture of the human brain. Hum Brain Mapp.  2009; 30( 10): 3127– 3141. http://dx.doi.org/10.1002/hbm.20737 Google Scholar CrossRef Search ADS PubMed  48 Raichle ME. The brain's default mode network. Annu Rev Neurosci.  2015; 38: 433– 447. http://dx.doi.org/10.1146/annurev-neuro-071013-014030 Google Scholar CrossRef Search ADS PubMed  49 Shulman GL, Fiez JA, Corbetta M, et al.   Common blood flow changes across visual tasks: II. Decreases in cerebral cortex. J Cogn Neurosci.  1997; 9( 5): 648– 663. Google Scholar CrossRef Search ADS PubMed  50 Kwong KK, Belliveau JW, Chesler DA, et al.   Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. Proc Natl Acad Sci U S A.  1992; 89( 12): 5675– 5679. http://dx.doi.org/10.1073/pnas.89.12.5675 Google Scholar CrossRef Search ADS PubMed  51 Frahm J, Bruhn H, Merboldt KD, et al.   Dynamic MR imaging of human brain oxygenation during rest and photic stimulation. J Magn Reson Imaging.  1992; 2( 5): 501– 505. http://dx.doi.org/10.1002/jmri.1880020505 Google Scholar CrossRef Search ADS PubMed  52 Raichle ME, MacLeod AM, Snyder AZ, et al.   A default mode of brain function. Proc Natl Acad Sci U S A.  2001; 98( 2): 676– 682. http://dx.doi.org/10.1073/pnas.98.2.676 Google Scholar CrossRef Search ADS PubMed  53 Fox MD, Snyder AZ, Vincent JL, et al.   The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A.  2005; 102( 27): 9673– 9678. http://dx.doi.org/10.1073/pnas.0504136102 Google Scholar CrossRef Search ADS PubMed  54 Spreng RN, Stevens WD, Chamberlain JP, et al.   Default network activity, coupled with the frontoparietal control network, supports goal-directed cognition. Neuroimage.  2010; 53( 1): 303– 317. http://dx.doi.org/10.1016/j.neuroimage.2010.06.016 Google Scholar CrossRef Search ADS PubMed  55 Spreng RN, Sepulcre J, Turner GR, et al.   Intrinsic architecture underlying the relations among the default, dorsal attention, and frontoparietal control networks of the human brain. J Cogn Neurosci.  2013; 25( 1): 74– 86. http://dx.doi.org/10.1162/jocn_a_00281 Google Scholar CrossRef Search ADS PubMed  56 Smallwood J, Brown K, Baird B, et al.   Cooperation between the default mode network and the frontal-parietal network in the production of an internal train of thought. Brain Res.  2012; 1428: 60– 70. http://dx.doi.org/10.1016/j.brainres.2011.03.072 Google Scholar CrossRef Search ADS PubMed  57 Bruno J, Hosseini SM, Kesler S. Altered resting state functional brain network topology in chemotherapy-treated breast cancer survivors. Neurobiol Dis.  2012; 48( 3): 329– 338. http://dx.doi.org/10.1016/j.nbd.2012.07.009 Google Scholar CrossRef Search ADS PubMed  58 Kesler SR, Wefel JS, Hosseini SM, et al.   Default mode network connectivity distinguishes chemotherapy-treated breast cancer survivors from controls. Proc Natl Acad Sci U S A.  2013; 110( 28): 11600– 11605. http://dx.doi.org/10.1073/pnas.1214551110 Google Scholar CrossRef Search ADS PubMed  59 Kesler SR, Watson CL, Blayney DW. Brain network alterations and vulnerability to simulated neurodegeneration in breast cancer. Neurobiol Aging.  2015; 36( 8): 2429– 2442. http://dx.doi.org/10.1016/j.neurobiolaging.2015.04.015 Google Scholar CrossRef Search ADS PubMed  60 Hampson JP, Zick SM, Khabir T, et al.   Altered resting brain connectivity in persistent cancer related fatigue. Neuroimage Clin.  2015; 8: 305– 313. http://dx.doi.org/10.1016/j.nicl.2015.04.022 Google Scholar CrossRef Search ADS PubMed  61 Kesler SR, Gugel M, Pritchard-Berman M, et al.   Altered resting state functional connectivity in young survivors of acute lymphoblastic leukemia. Pediatr Blood Cancer.  2014; 61( 7): 1295– 1299. http://dx.doi.org/10.1002/pbc.25022 Google Scholar CrossRef Search ADS PubMed  62 Steinberg S, Hartmann R, Wisniewski S, et al.   (Late sequelae of CNS recurrence of acute lymphoblastic leukemia in childhood). Klin Padiatr.  1998; 210( 4): 200– 206. Google Scholar CrossRef Search ADS PubMed  63 Iuvone L, Mariotti P, Colosimo C, et al.   Long-term cognitive outcome, brain computed tomography scan, and magnetic resonance imaging in children cured for acute lymphoblastic leukemia. Cancer.  2002; 95( 12): 2562– 2570. http://dx.doi.org/10.1002/cncr.10999 Google Scholar CrossRef Search ADS PubMed  64 von der Weid N, Mosimann I, Hirt A, et al.   Intellectual outcome in children and adolescents with acute lymphoblastic leukaemia treated with chemotherapy alone: Age- and sex-related differences. Eur J Cancer.  2003; 39( 3): 359– 365. http://dx.doi.org/10.1016/S0959-8049(02)00260-5 Google Scholar CrossRef Search ADS PubMed  65 Elens I, Deprez S, Danckaerts M, et al.   Neurocognitive sequelae in adult childhood-leukemia survivors related to levels of phosphorylated Tau. J Natl Cancer Inst.  2017; 109( 7):djw321. 66 Renard M, Suciu S, Bertrand Y, et al.   Second neoplasm in children treated in EORTC 58881 trial for acute lymphoblastic malignancies: Low incidence of CNS tumours. Pediatr Blood Cancer.  2011; 57( 1): 119– 125. http://dx.doi.org/10.1002/pbc.23083 Google Scholar CrossRef Search ADS PubMed  67 Domenech C, Suciu S, De Moerloose B, et al.   Dexamethasone (6 mg/m2/day) and prednisolone (60 mg/m2/day) were equally effective as induction therapy for childhood acute lymphoblastic leukemia in the EORTC CLG 58951 randomized trial. Haematologica.  2014; 99( 7): 1220– 1227. http://dx.doi.org/10.3324/haematol.2014.103507 Google Scholar CrossRef Search ADS PubMed  68 Poot DH, den Dekker AJ, Achten E, et al.   Optimal experimental design for diffusion kurtosis imaging. IEEE Trans Med Imaging.  2010; 29( 3): 819– 829. http://dx.doi.org/10.1109/TMI.2009.2037915 Google Scholar CrossRef Search ADS PubMed  69 Miatton M, De Wolf D, François K, et al.   Neuropsychological performance in school-aged children with surgically corrected congenital heart disease. J Pediatr.  2007; 151( 1): 73– 78, 78.e1. Google Scholar CrossRef Search ADS PubMed  70 de Sonneville LMJ. Amsterdam Neuropsychological Tasks: a computer-aided assessment program. In den Brinker BPML, Beek PJ, Brand AN, Maarse SJ, Mulder LJM, eds. Cognitive Ergonomics, Clinical Assessment and Computer-Aided Learning: Computers in Psychology. Lisse: Swets & Zeitlinger BV;1999:187–203. 71 Van Der Elst W, Van Boxtel M, Van Breukelen G, et al.   Rey’s verbal learning test: Normative data for 1855 healthy participants aged 24–81 years and the influence of age, sex, education, and mode of presentation. J Int Neuropsychol Soc.  2005; 11: 290– 302. Google Scholar CrossRef Search ADS PubMed  72 Uddin LQ, Kelly AM, Biswal BB, et al.   Functional connectivity of default mode network components: correlation, anticorrelation, and causality. Hum Brain Mapp.  2009; 30( 2): 625– 637. http://dx.doi.org/10.1002/hbm.20531 Google Scholar CrossRef Search ADS PubMed  73 Kesler SR. Default mode network as a potential biomarker of chemotherapy-related brain injury. Neurobiol Aging.  2014; 35(suppl 2): S11– S19. Google Scholar CrossRef Search ADS PubMed  74 Sambataro F, Murty VP, Callicott JH, et al.   Age-related alterations in default mode network: Impact on working memory performance. Neurobiol Aging.  2010; 31( 5): 839– 852. http://dx.doi.org/10.1016/j.neurobiolaging.2008.05.022 Google Scholar CrossRef Search ADS PubMed  75 Deprez S, Amant F, Smeets A, et al.   Longitudinal assessment of chemotherapy-induced structural changes in cerebral white matter and its correlation with impaired cognitive functioning. J Clin Oncol.  2012; 30( 3): 274– 281. http://dx.doi.org/10.1200/JCO.2011.36.8571 Google Scholar CrossRef Search ADS PubMed  76 Porto L, Preibisch C, Hattingen E, et al.   Voxel-based morphometry and diffusion-tensor MR imaging of the brain in long-term survivors of childhood leukemia. Eur Radiol.  2008; 18( 11): 2691– 2700. http://dx.doi.org/10.1007/s00330-008-1038-2 Google Scholar CrossRef Search ADS PubMed  77 Dellani PR, Eder S, Gawehn J, et al.   Late structural alterations of cerebral white matter in long-term survivors of childhood leukemia. J Magn Reson Imaging.  2008; 27( 6): 1250– 1255. http://dx.doi.org/10.1002/jmri.21364 Google Scholar CrossRef Search ADS PubMed  78 Dietrich J, Han R, Yang Y, et al.   CNS progenitor cells and oligodendrocytes are targets of chemotherapeutic agents in vitro and in vivo. J Biol.  2006; 5( 7): 22. http://dx.doi.org/10.1186/jbiol50 Google Scholar CrossRef Search ADS PubMed  79 Perry ME, McDonald CR, Hagler DJ, et al.   White matter tracts associated with set-shifting in healthy aging. Neuropsychologia.  2009; 47( 13): 2835– 2842. http://dx.doi.org/10.1016/j.neuropsychologia.2009.06.008 Google Scholar CrossRef Search ADS PubMed  80 Braak H, Braak E. Development of Alzheimer-related neurofibrillary changes in the neocortex inversely recapitulates cortical myelogenesis. Acta Neuropathol.  1996; 92( 2): 197– 201. http://dx.doi.org/10.1007/s004010050508 Google Scholar CrossRef Search ADS PubMed  81 Reisberg B, Franssen EH, Hasan SM, et al.   Retrogenesis: Clinical, physiologic, and pathologic mechanisms in brain aging, Alzheimer's and other dementing processes. Eur Arch Psychiatry Clin Neurosci.  1999; 249(suppl 3): 28– 36. Google Scholar CrossRef Search ADS PubMed  82 Rodriguez M. A function of myelin is to protect axons from subsequent injury: Implications for deficits in multiple sclerosis. Brain.  2003; 126(Pt 4): 751– 752. Google Scholar CrossRef Search ADS   83 Urger SE, De Bellis MD, Hooper SR, et al.   The superior longitudinal fasciculus in typically developing children and adolescents: Diffusion tensor imaging and neuropsychological correlates. J Child Neurol.  2015; 30( 1): 9– 20. http://dx.doi.org/10.1177/0883073813520503 Google Scholar CrossRef Search ADS PubMed  84 Parker GJ, Luzzi S, Alexander DC, et al.   Lateralization of ventral and dorsal auditory-language pathways in the human brain. Neuroimage.  2005; 24( 3): 656– 666. http://dx.doi.org/10.1016/j.neuroimage.2004.08.047 Google Scholar CrossRef Search ADS PubMed  85 Lord LD, Expert P, Huckins JF, et al.   Cerebral energy metabolism and the brain's functional network architecture: An integrative review. J Cereb Blood Flow Metab.  2013; 33( 9): 1347– 1354. http://dx.doi.org/10.1038/jcbfm.2013.94 Google Scholar CrossRef Search ADS PubMed  86 Kesler SR, Gugel M, Huston-Warren E, et al.   Atypical structural connectome organization and cognitive impairment in young survivors of acute lymphoblastic leukemia. Brain Connect.  2016; 6( 4): 273– 282. http://dx.doi.org/10.1089/brain.2015.0409 Google Scholar CrossRef Search ADS PubMed  87 Waber D, Bernstein J, Kammerer B, et al.   Neuropsychological diagnostic profiles of children who received CNS treatment for acute lymphoblastic-leukemia: The systemic approach to assessment. Dev Neuropsychol.  1992; 8( 1): 28. Google Scholar CrossRef Search ADS   88 Kesler SR, Gugel M, Pritchard-Berman M, et al.   Altered resting state functional connectivity in young survivors of acute lymphoblastic leukemia. Pediatric Blood Cancer.  2014; 61( 7): 1295– 1299. http://dx.doi.org/10.1002/pbc.25022 Google Scholar CrossRef Search ADS PubMed  © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Journal

JNCI: Journal of the National Cancer InstituteOxford University Press

Published: Mar 5, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off