Chief Cushie ~MaryO~ Posted December 7, 2023 Chief Cushie Report Share Posted December 7, 2023 Abstract Introduction: Chronic exposure to excessive endogenous cortisol leads to brain changes in Cushing’s disease (CD). However, it remains unclear how CD affects large-scale functional networks (FNs) and whether these effects are reversible after treatment. This study aimed to investigate functional network changes of CD patients and their reversibility in a longitudinal cohort. Methods: Active CD patients (N = 37) were treated by transsphenoidal pituitary surgery and reexamined 3 months later. FNs were computed from resting-state fMRI data of the CD patients and matched normal controls (NCs, N = 37). A pattern classifier was built on the FNs to distinguish active CD patients from controls and applied to FNs of the CD patients at the 3-month follow-up. Two subgroups of endocrine-remitted CD patients were identified according to their classification scores, referred to as image-based phenotypically (IBP) recovered and unrecovered CD patients, respectively. The informative FNs identified by the classification model were compared between NCs, active CD patients, and endocrine-remitted patients as well as between IBP recovered and unrecovered CD patients to explore their functional network reversibility. Results: All 37 CD patients reached endocrine remission after treatment. The classification model identified three informative FNs, including cerebellar network (CerebN), fronto-parietal network (FPN), and default mode network. Among them, CerebN and FPN partially recovered toward normal at 3 months after treatment. Moreover, the informative FNs were correlated with 24-h urinary-free cortisol and emotion scales in CD patients. Conclusion: These findings suggest that CD patients have aberrant FNs that are partially reversible toward normal after treatment. Journal Section: Research Article Keywords: Cushing’s disease, Reversibility, Functional networks, Cortisol, Emotion Introduction Cushing’s disease (CD) is characterized by chronic exposure to excessive endogenous glucocorticoid most commonly caused by an adrenocorticotropic hormone (ACTH) pituitary adenoma [1, 2]. The CD is accompanied by multiple physical manifestations such as hypertension and osteoporosis, as well as various neuropsychiatric symptoms including memory lapses, attention deficits, executive function decline, emotional dysfunction, visual-spatial disability, and language defects [3‒14]. These neuropsychiatric symptoms are indicative of the effects of CD on the brain anatomy and function. Therefore, CD provides a unique and naturalistic model for investigating both the effects of hypercortisolism on the human brain and the reversibility of these effects after resolution of hypercortisolism. Recent studies have documented brain structural and metabolic abnormalities in CD patients with a variety of neuroimaging techniques, including structural magnetic resonance imaging (sMRI) [11, 12, 15‒24], diffusion tensor imaging [10, 25‒27], proton magnetic resonance spectroscopy [21, 28‒30], positron emission topography [21, 31], and arterial spin labeling [32]. These studies have shown that brain structural and metabolic abnormalities in CD patients can be partially restored after resolution of hypercortisolism [16, 18, 20‒22, 24, 32‒34], typically after transsphenoidal pituitary surgery (TSS), a safe and effective first-line treatment with a high endocrine remission rate [35, 36]. Several functional magnetic resonance imaging (fMRI) studies have also documented brain functional abnormalities in CD patients [37‒42]. Particularly, aberrant functional connectivity between the anterior cingulate cortex and the limbic network, as well as the lateral occipital cortex and the default mode network (DMN) was observed in endocrine-remitted CD patients after TSS treatment in a cross-sectional resting-state fMRI (rs-fMRI) study [40]. However, the causal effects of hypercortisolism on brain functional connectivity cannot be well investigated in CD patients only through the cross-sectional study. Additionally, the large-scale functional networks (FNs) of CD patients were not well investigated through univariate analyses in previous studies, which only examined one or few FNs in CD patients independently [37‒42]. The present study aims to jointly investigate a number of whole-brain large-scale intrinsic FNs and their reversibility due to hypercortisolism in CD patients based on rs-fMRI with a longitudinal design through multivariate analysis. Particularly, intrinsic FNs altered by CD were identified using a multivariate pattern classification model optimized by selecting intrinsic FNs informative for distinguishing CD patients from matched normal controls (NCs). The changes in these informative FNs of endocrine-remitted CD patients after TSS treatment were quantified at the 3-month follow-up with the established pattern classification model. Furthermore, changes in clinical measures, including serum cortisol, 24-h urinary-free cortisol (24hUFC), ACTH, self-rating depression scale (SDS), and self-rating anxiety scale (SAS), were detected between active and endocrine-remitted CD patients using pseudo paired t tests. Finally, the association between aberrant FNs and clinical measures was investigated in CD patients. Materials and Methods Participants In this study, 50 CD patients undergoing TSS, and 38 NCs with no history of glucocorticoid treatment were recruited at the Department of Neurosurgery, Peking Union Medical College Hospital. All these participants were assessed for depression and anxiety measured by the SDS and SAS, respectively [43]. The inclusion criteria for NCs were no past or present heart history of disease, atherosclerosis, hyperlipidemia, diabetes, neurological/psychiatric disorders, and claustrophobia. The exclusion criteria for both CD patients and NCs were past or present brain trauma, other neurological disorders, history of radiotherapy, or contraindications to MRI. Besides the inclusion and exclusion criteria, the quality of the imaging data was controlled as follows. No participant had head motion exceeding 2.0 mm translation in any of the three directions or exceeding 2.0o maximum rotation around any of the axes during rs-fMRI scanning [44]. Additionally, no participant had root-mean-square value of maximum frame-wise displacement greater than 0.3 mm [45]. After quality control of the imaging data, 37 CD patients and 37 sex-, age-, and education level-matched NCs were included in the study. The diagnosis of active CD was confirmed by experienced endocrinologists along with dynamic enhanced pituitary MRI, low- and high-dose dexamethasone suppression tests, and/or inferior petrosal sinus sampling in accordance with the latest clinical practice guidelines [46]. The 37 active CD patients were treated with TSS rather than radiotherapy. All of the 37 CD patients reached endocrine remission after treatment, which was confirmed by their normal serum cortisol (<5 µg/dL within 7 days of surgery) [46]. These patients were asked to revisit the hospital for reexamination 3 months after surgery, and all of them had no recurrence at the 3-month follow-up. Serum cortisol, 24hUFC, and ACTH were measured by direct chemiluminescence immunoassays in CD patients before surgery and at the 3-month follow-up (Siemens Healthcare Diagnostics Inc., USA). This study was approved by the Medical Ethics Committee of Peking Union Medical College Hospital, and written informed consent was obtained from all participants after explaining to them the nature of the study. Imaging Data Acquisition The MRI data were scanned by using an 8-channel phase-array head coil with a 3.0-Tesla MR scanner (Discovery MR750, General Electric) for all participants, including NCs, active CD patients, and endocrine-remitted CD patients without recurrence at the 3-month follow-up. The rs-fMRI data were acquired axially by using a gradient echo-planar imaging sequence, and the scanning parameters were 200 whole-brain volumes, 36 transverse slices with a thickness of 4 mm, in-plane resolution = 3.75 × 3.75 mm2, field of view = 240 × 240 mm2, flip angle = 90°, repetition time = 2,000 ms, and echo time = 30 ms. The extra high-resolution sagittal 3D T1-weighted data were acquired by using a brain volume sequence, and the scanning parameters were 172 slices with a thickness of 1.0 mm, in-place matrix = 512 × 512, field of view = 256 × 256 mm2, voxel size = 0.5 × 0.5 × 1.0 mm3, flip angle = 12°, repetition time = 7.2 ms, echo time = 3.2 ms, and inversion time = 400 ms. Imaging Data Preprocessing The rs-fMRI data were preprocessed as follows: (1) discarding the first four volumes of the fMRI data; (2) correction for slice timing; (3) 3D rigid-body correction for head motion to the middle frame of the data; (4) global 4D intensity scaling of the fMRI data to yield a mean value of 10,000; (5) nonlinear registration of the fMRI data to the MNI template with the deformation field obtained from its co-registered T1-weighted data using DARTEL within statistical parametric mapping (SPM12) software, with a resampled resolution of 3×3×3 mm3; (6) spatial smoothing with a 6-mm full-width at half maximum Gaussian kernel; (7) motion artifacts removal from fMRI data with ICA-AROMA; (8) regressing out averaged signals of white matter, cerebrospinal fluid, and whole brain; (9) temporal band-pass filtering (0.009–0.08 Hz). The preprocessing procedures were performed by using SPM12 software (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/). Identification of Informative FNs in Active CD Patients The flowchart for identifying informative FNs in active CD patients is shown in Figure 1. First, group information-guided independent component analysis was applied to rs-fMRI data of each participant from NCs, active CD patients, and endocrine-remitted CD patients at 3 months after treatment to extract subject-specific independent components (ICs), referred to as intrinsic FNs [47] (Fig. 1a). Specifically, group-level ICs were computed based on all participants from NC, active CD, and endocrine-remitted CD groups, by using the multivariate exploratory linear optimized decomposition into independent components (MELODIC) toolbox in FSL software (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/melodic). These group-level ICs were used as guidance information to compute subject-specific ICs of all individuals [47]. The number of ICs was empirically set to be 25, and therefore each individual was characterized by 25 FNs. Particularly, these FNs were restricted to gray matter in order to minimize the partial volume effects of cerebrospinal fluid and confounding effects on the estimated components, and to improve the sensitivity to the changes of blood-oxygen-level-dependent signals. Fig. 1. VIEW LARGEDOWNLOAD SLIDE Flowchart of the multivariate pattern classification method for distinguishing active CD patients from NCs, including data preparation (a), classification modeling (b), as well as identifying CD-associated ICs (c). CD, Cushing’s disease; active CD patients, CD patients before treatment; NCs, normal controls; rs-fMRI, resting-state functional magnetic resonance imaging; ICs, independent components; GIG-ICA, group information-guided ICA; SVM, support vector machine; LOOCV, leave-one-out cross-validation. Subsequently, a multivariate pattern classification method based on support vector machine (SVM) was applied to identify cross-sectional informative FNs, which were most discriminative in distinguishing active patients from NCs [48] (Fig. 1b). Specifically, sigmoid kernel SVM classifiers were built upon a subset of 25 FNs obtained via a forward selection technique to optimize the classification performance for differentiating active patients from NCs. The similarity between subjects in SVM classification was defined as the Riemannian distance of the subset of FNs on the Grassmann manifold [48, 49]. Initially, the forward component selection procedure built a classifier on each individual FN, and the performance of the classifier was estimated using leave-one-out cross-validation (LOOCV) so that each FN could be evaluated for its classification performance. The accuracy rate was chosen as the main metric for evaluating the classification performance. The FN with the best performance was selected to be included in the subsequent classification. Through combining the first selected FN and any one of the remaining FNs, classifiers were built upon all paired FNs which were evaluated based on the training data during the current outer round using an inner LOOCV procedure. The paired FNs with the best performance were selected to be included in the classification. The procedure was repeated to add more FNs in the classification one by one until a single classifier was built upon all available FNs. Accordingly, a subset of FNs with the best performance was deemed to be the final selected components in the classification, hereafter referred to as informative FNs. To avoid potential classification biases, a nested LOOCV procedure was applied to optimize the parameters of the sigmoid kernel SVM classifiers to improve the classification performance during the forward component selection procedure [48, 49]. Since different FNs might be selected in each training runs or each testing run during the nested LOOCV procedure, the informative FNs were selected as the best performing ones with higher frequency (selection frequency>0.5). Based on these informative FNs of 74 subjects (including 37 NCs and 37 active CD patients), the LOOCV classification model yielded 74 aggregated SVM classifiers with the nested LOOCV classifiers, respectively. Each aggregated classifier generated a classification score from its corresponding nested classifiers with a positive value indicating CD state and a negative value indicating healthy state. Finally, the classification performance was evaluated with metrics including classification accuracy, specificity, sensitivity, and the area under the receiver operating characteristic curve (AUROC) (Fig. 1c). Non-parametric permutation tests were adopted to examine the statistical significance of the classification performance. The classification rate for the null distribution was estimated by building sigmoid kernel SVM models upon cross-sectional informative FNs of all active CD patients and NCs with subject labels randomly permuted by using the LOOCV strategy. This procedure was repeated for 10,000 times. Finally, the null distribution of the classification rate based on permuted samples was obtained. Longitudinal Analyses of Informative FNs and Emotion Scales from Active to Endocrine-Remitted CD Patients To investigate the longitudinal functional connectivity changes, pseudo paired t tests between active and endocrine-remitted CD patients (10,000 permutations) were applied voxel-wisely to each of the informative FNs using statistical non-parametric mapping (SnPM) software (http://warwick.ac.uk/snpm). Brain regions with statistical significance within each informative FN were identified at a voxel-wise threshold of p < 0.01 and an extent threshold of 40 adjacent voxels (AlphaSim-corrected p < 0.01). Additionally, statistical analyses were performed to compare the IC’s z scores of FNs as well as emotion scales between any pair of NC, active CD, and endocrine-remitted CD groups to further examine the longitudinal brain functional connectivity changes. Particularly, a pseudo paired t test was applied to all IC’s z scores within each informative FN as well as SDS scores and SAS scores between active and endocrine-remitted CD patients (10,000 permutations). While a pseudo two-sample t test with age, sex, and years of school education as covariates was applied to all IC’s z scores within each informative FN as well as SDS scores and SAS scores between NCs and active CD patients and endocrine-remitted CD patients. Significant differences were determined at a false discovery rate (FDR) threshold of p < 0.05 after adjusting for multiple comparisons. Statistical Analyses of Informative FNs in Endocrine-Remitted CD Patients The established pattern classification model was applied to the FNs of the follow-up endocrine-remitted CD patients. Thus, each endocrine-remitted CD patient had a classification score that reflected the likelihood of the endocrine-remitted CD patient to be active CD or healthy state (a positive value indicating active CD state and a negative value indicating healthy state). Based on the follow-up classification scores, endocrine-remitted CD patients who were correctly classified as active CD patients before treatment were further stratified into two subgroups: subjects with negative classification scores, referred to as image-based phenotypically (IBP) recovered CD patients, and those with positive classification scores, referred to as IBP unrecovered CD patients. Additionally, statistical differences in the IC’s z scores within each of the informative FNs between the IBP recovered and unrecovered CD patients, were assessed to elucidate these endocrine-remitted CD patients’ brain recoveries in these informative FNs at 3 months after treatment. Specifically, a pseudo two-sample t test with age, sex, years of school education, and years of disease duration as covariates was applied to all IC’s z scores of each FN between IBP recovered and unrecovered CD patients, and significant differences were determined at an FDR threshold of p < 0.05 (10,000 permutations) after adjusting for multiple comparisons. Correlation Analyses between Informative FNs and Clinical Measures Correlation analyses were performed to investigate the relationship between informative FNs and clinical measures in all 37 CD patients. The clinical measures of interest were serum cortisol, 24hUFC, ACTH, SDS, and SAS. Specifically, the correlation between each clinical measure and the averaged IC’s z score of each informative FN of CD patients before treatment was computed using a general linear model with age, sex, years of school education, and years of disease duration as covariates. Significant correlations were determined at a threshold of p < 0.05 using FDR corrected for multiple comparisons. Additionally, the correlation between the changes of each clinical measure and the averaged IC’s z score of each informative FN for endocrine-remitted CD patients before and after treatment was computed by using a general linear model with age, sex, years of school education, years of disease duration, and this clinical measure before treatment as covariates. The change of the averaged IC’s z score of each informative FN for each CD patient was calculated as the value after treatment minus the value before treatment divided by the value before treatment, and the change of each clinical measure for each CD patient was calculated similarly. Significant correlations were identified at a threshold of p < 0.05 using FDR corrected for multiple comparisons. Results Demographics and Clinical Characteristics The demographic and clinical data, including age, sex, years of school education, hormones, and emotion scales, are summarized in Table 1. There were no significant differences in age, sex, and years of school education between NCs and CD patients before treatment or at the 3-month follow-up (p > 0.05). The hormone levels, including ACTH, 24hUFC, and serum cortisol, were significantly restored (lower to be precise) in endocrine-remitted CD patients at the 3-month follow-up compared to their pre-treatment levels (FDR-corrected p < 0.05). These CD patients reached endocrine remission confirmed by their normal serum cortisol (<5 µg/dL) within 7 days of surgery. The emotion scales, including SDS scores and SAS scores, were significantly improved (smaller to be precise) in endocrine-remitted CD patients at 3 months after treatment compared to their rating scales in active phase (FDR-corrected p < 0.05), and the SDS scores and SAS scores for these endocrine-remitted CD patients were comparable to those of NCs. There was also significant difference in SDS scores between endocrine-remitted CD patients and NCs (FDR-corrected p < 0.05), while no significant difference was found in SAS scores between endocrine-remitted CD patients and NCs (p = 0.70). These psychometric comparison results suggest that depressive symptoms were partially recovered in endocrine-remitted CD patients, while their anxiety symptoms were also not completely recovered. Table 1. Demographic and clinical data of the participants Characteristics NCs (N = 37) Active CDs (N = 37) Endocrine-remitted CDs (N = 37) p value Age, years 38.46±11.85 33.92±8.57 33.92±8.57 0.062a Sex (M/F) 10/27 8/29 8/29 0.83a Years of school education 12.84±3.53 13.27±3.11 13.27±3.11 0.55a ACTH, pg/mL - 75.70 (45.55, 103.25) 23 (10.33, 30.70) <0.01**b 24hUFC, μg/day - 582.34 (351.30, 991.56) 47.77 (14.41, 186.54) <0.01**b Serum cortisol, μg/dL - 26.58 (20.98, 31.84) 5.49 (1.75, 13.69) <0.01**b Depression (SDS) 38.72±7.45 53.99±9.20 45.54±10.24 <0.01**c <0.01**d Anxiety (SAS) 33.34±5.46 45.27±11.92 34.46±9.78 <0.01**c 0.70d Values for characteristics are presented as mean ± SD or median (25th percentiles, 75th percentiles) unless otherwise indicated. Group differences in age, years of school education, SDS, and SAS between NCs and CD patients before or at the 3-month follow-up were examined using pseudo two-sample t tests. Group differences in sex between NCs and the CD patients before treatment or at the 3-month follow-up were examined using a χ2 test. Group differences in ACTH, 24hUFC, serum cortisol, SDS, and SAS between CD patients before treatment and at the 3-month follow-up were examined using pseudo paired t tests. NCs, normal controls; CDs, patients with Cushing’s disease; ACTH, adrenocorticotropic hormone; 24hUFC, 24-h urinary-free cortisol; SDS, self-rating depression scale; SAS, self-rating anxiety scale; M, male; F, female; SD, standard deviations. **p < 0.01. aNCs versus active or endocrine-remitted CDs. bActive CDs versus endocrine-remitted CDs. cActive CDs versus NCs or endocrine-remitted CDs. dNCs versus endocrine-remitted CDs. Informative FNs in Active CD Patients Active CD patients were mostly different from the NCs in 3 out of 25 FNs (selection frequency>0.5), including cerebellar network (CerebN), fronto-parietal network (FPN), and DMN, as shown in Figure 2a and b. The classification models built upon these three informative FNs yielded an accuracy of 72% (sensitivity: 68%, specificity: 76%, AUROC: 0.81), as shown in Figure 2c. Non-parametric permutation tests demonstrated that the classification accuracy was promising and significant (p < 1.0e−04), as suggested by the histogram of permuted classification rates shown in Figure 2d. Particularly, 25 out of 37 (67%) CD patients were correctly classified as active CD patients before treatment. Fig. 2. VIEW LARGEDOWNLOAD SLIDE Three informative functional brain networks identified by the multivariate pattern classification method and the classification performance. a Three highly selected functional brain networks, including CerebN, FPN, and DMN, for differentiating active CD patients from NCs. b The frequency of the functional brain networks selected in the nested LOOCV experiments. c The receiver operating characteristic (ROC) curve (area under the ROC curve [AUROC] = 0.81) of the classification model built upon the selected most discriminative FNs. d The histogram of the classification rates of the permutation tests and the real classification rate. In panel (a), brain regions with significant functional connectivity were obtained by applying voxel-wise one-sample t tests to the IC’s z scores for each of the FNs across all active CD patients and NCs (p < 0.05, FWE corrected for multiple comparisons, and cluster size >400 voxels). CerebN, cerebellar network; FPN, fronto-parietal network; DMN, default mode network; CD, Cushing’s disease; Pres, CD patients before treatment (i.e., active CD patients); NCs, normal controls; FNs, functional networks; ICs, independent components; FWE, family-wise error; L, left; R, right. Changes in Informative FNs from Active to Endocrine-Remitted CD Patients Two out of the three informative FNs, i.e., CerebN and FPN other than DMN, exhibited significant functional connectivity changes in CD patients between active and endocrine-remitted states (Fig. 3a). Compared with their active state, the endocrine-remitted CD patients had significantly improved (increased to be precise) functional connectivity measured by IC’s z scores in both CerebN and FPN circuits at 3 months after treatment. These results indicate that the FNs of the endocrine-remitted CD patients partially recovered toward the NCs at 3 months after treatment (Fig. 3b). Fig. 3. VIEW LARGEDOWNLOAD SLIDE Two informative functional brain networks as well as emotion scales with significant longitudinal changes in CD patients before treatment and at the 3-month follow-up. a Brain regions with significant longitudinal changes in functional connectivity within circuits of CerebN and FPN for CD patients, identified using non-parametric permutation tests (AlphaSim-corrected p < 0.01). b, c Significantly different functional connectivity measured by IC’s z scores across voxels within circuits of CerebN and FPN as well as emotion scales measured by the self-rating depression scale (SDS) and self-rating anxiety scale (SAS) between any two of NCs, CD patients before the treatment (i.e., active CD patients), and endocrine-remitted CD patients at 3-month follow-up (FDR-corrected p < 0.05). A pseudo paired t test with age, sex, and years of school education as covariates was conducted to compare all IC’s z scores within each functional network as well as the SDS scores and SAS scores between CD patients before treatment and at the 3-month follow-up. While a pseudo two-sample t test with age, sex, and years of school education as covariates was conducted to compare IC’s z scores within each functional network as well as SDS scores and SAS scores between NCs and CD patients before treatment, and endocrine-remitted CD patients at 3-month follow-up. CerebN, cerebellar network; FPN, fronto-parietal network; CD, Cushing’s disease; Pres, CD patients before treatment; Posts, endocrine-remitted CD patients at 3-month follow-up; NCs, normal controls; ICs, independent components; FDR, false discovery rate. Changes in Informative FNs of Endocrine-Remitted CD Patients Among the endocrine-remitted CD patients who were correctly classified as active CD patient before treatment, 14 participants were classified as IBP-recovered patients, while 11 participants were classified as IBP-unrecovered patients. The IBP-recovered and -unrecovered CD patients were determined by using the established pattern classification model according to the opposite signs in their classification scores based on their follow-up rs-fMRI data at 3 months after treatment (Fig. 4b). The IBP recovered patients had better recovery of the impaired functional connectivity within the circuits of CerebN and FPN than the IBP-unrecovered patients, as shown in Figure 4a. Fig. 4. VIEW LARGEDOWNLOAD SLIDE Differences in functional connectivity measured by IC’s z scores across voxels within circuits of CerebN and FPN as well as classification scores between image-based phenotypically (IBP)-recovered and -unrecovered CD patients after treatment. In panel (a), statistical comparisons were performed using pseudo two-sample t tests with age, sex, years of school education, and years of disease duration as covariates (FDR-corrected p < 0.05). In panel (b), violin plots showed opposite signs in classification scores between IBP-recovered and -unrecovered CD patients. CerebN, cerebellar network; FPN, fronto-parietal network; CD, Cushing’s disease; CDs, patients with Cushing’s disease; ICs, independent components; FDR, false discovery rate. Relationship between Informative FNs and Clinical Measures Changes of 24hUFC for endocrine-remitted CD patients before and after treatment were negatively correlated with their changes of averaged IC’s z scores within the FPN circuits with statistical significance (r = −0.37, p = 0.020), as shown in Figure 5a. The emotion scales, including SDS and SAS, were significantly negatively correlated with the averaged IC’s z scores within the CerebN circuits in the active CD patients (r = −0.31, p < 0.042), as shown in Figure 5c and d. There was no significant correlation for other clinical measures. Fig. 5. VIEW LARGEDOWNLOAD SLIDE Correlations between clinical measures and averaged IC’s z scores of informative FNs in CD patients (FDR-corrected p < 0.05). a Scatter plot for the significantly negative correlation between changes in the averaged IC’s z scores of the FPN circuits and 24hUFC of these 37 endocrine-remitted CD patients. b Multi-slice view of the FPN circuits whose changes in the averaged z scores were significantly correlated with changes in 24hUFC for all 37 endocrine-remitted CD patients before and after treatment. c, d Scatter plots for the significantly negative correlations between the averaged IC’s z scores of the CerebN circuits, and the SDS scores and SAS scores in these 37 endocrine-remitted CD patients before treatment. In panel (a), the changes in the averaged IC’s z scores of the FPN circuits were adjusted by regressing out covariates including age, sex, years of school education, years of disease duration, and the pre-treatment 24hUFC. In panels (c) and (d), the averaged IC’s z scores of the CerebN circuits were adjusted by regressing out covariates including age, sex, years of school education, and years of disease duration. 24hUFC, 24-h urinary-free cortisol; SDS, self-rating depression scale; SAS, self-rating anxiety scale; FPN, fronto-parietal network; CerebN, cerebellar network; CD, Cushing’s disease; Pres, CD patients before treatment; Posts, endocrine-remitted CD patients at 3-month follow-up; ICs, independent components; FDR, false discovery rate. Discussion The present study investigated the large-scale FNs of CD patients before and after treatment based on longitudinal rs-fMRI data. To the best of our knowledge, this is the first study to characterize longitudinal large-scale functional brain network changes due to hypercortisolism in CD patients using multivariate analysis. Particularly, the active CD patients had aberrant functional connectivity within circuits of CerebN, FPN, and DMN, respectively. More importantly, the impaired functional connectivity within the circuits of the CerebN and FPN was partially recovered in the endocrine-remitted CD patients, respectively. The changes in 24hUFC of CD patients before and after treatment were correlated with their changes in the functional connectivity of the FPN circuits. In addition, the emotion scales, including SDS and SAS, were also correlated with the functional connectivity of the CerebN circuits in CD patients before treatment. Aberrant FNs in Active CD Patients The informative FNs identified by the multivariate method were able to distinguish active CD patients from NCs with an accuracy of 72% (sensitivity: 68%, specificity: 76%, AUROC: 0.81). The non-parametric permutation tests also suggested that the multivariate method performed well in differentiating active CD patients from NCs. The most frequently selected FNs (Fig. 2b), i.e., informative FNs, were CerebN, FPN, and DMN. The cross-sectional multivariate analyses have revealed that the active CD patients were mostly different from the NCs in the functional connectivity within 3 FNs out of 25 FNs, as shown in Figure 2a. The aforementioned cross-sectional results provided new insights into large-scale functional brain network abnormalities due to hypercortisolism in CD patients. Particularly, our study revealed that active CD patients had significantly disrupted functional connectivity within the cerebellum (Fig. 2a, 3b), and their emotional dysfunctions observed by the SDS and SAS were associated with the impaired functional connectivity within the cerebellum (Fig. 5c, d). Therefore, it was reasonable to speculate that the cognitive or emotional dysfunctions for active CD patients, documented in this study as well as numerous previous studies [3‒5, 7‒9, 11‒14, 50], might be closely related to the observed functional connectivity abnormalities in the cerebellum. Additionally, our study found that the functional connectivity within the FPN circuits was significantly reduced in active CD patients (Fig. 2a, 3b). It was postulated that cognitive impairments in active CD patients, reported in several early studies [6, 14, 51], might be associated with the observed functional connectivity abnormalities in the FPN circuits. While several recent studies reported that active CD patients had structural or metabolic abnormalities in two brain regions of the FPN, namely, the middle frontal gyrus and inferior parietal lobule [17, 21, 32] These local morphological or metabolic abnormalities might exacerbate the observed functional network (FPN) alterations in active CD patients. Moreover, our study found that the functional connectivity within the DMN circuits was vulnerable to the detrimental effects of hypercortisolism in active CD patients. Besides our finding, recent studies reported that active CD patients showed structural, metabolic, or spontaneous activity abnormalities in several brain regions of DMN, including the posterior cingulate cortex, precuneus, parahippocampal gyrus, ventral medial prefrontal cortex, superior frontal gyrus, inferior temporal gyrus, and lateral parietal cortex [21, 27, 30‒32, 38, 52] These local morphological, metabolic or activity abnormalities might exacerbate the newly discovered DMN impairments in active CD patients. Essentially, the functional, morphological, and metabolic abnormalities in regions of the DMN might be directly related to the adverse expressions of glucocorticoid receptor genes within these brain regions caused by excessive exposure to endogenous cortisol [53]. Reversible Impaired FNs in Endocrine-Remitted CD Patients after Treatment The longitudinal statistical analysis has revealed that the endocrine-remitted CD patients’ hormones, including ACTH, 24hUFC, and serum cortisol, maintained near-normal levels at 3 months after treatment, suggesting that these patients did not relapse according to the endocrine hormone levels (Table 1). Meanwhile, their functional connectivity within circuits of the FPN and CerebN was partially restored at the 3-month follow-up after resolution of hypercortisolism (Fig. 3b). Particularly, our combined longitudinal and cross-sectional study found that the functional connectivity within the FPN circuits in endocrine-remitted CD patients was partially restored after treatment. While a cross-sectional sMRI study reported that endocrine-remitted CD patients still had structural abnormalities in the FPN-related region, namely, the middle frontal gyrus [17]. Our study also found that the functional connectivity of the cerebellum in endocrine-remitted CD patients was partially restored after treatment (Fig. 3b). Besides this finding, two other cross-sectional sMRI studies reported that the structural abnormalities of the cerebellum in endocrine-remitted patients were present as well [16, 20]. Taken together, it was postulated that the reversibility of the observed functional connectivity impairments within circuits of the FPN and CerebN might be directly influenced by their local morphological abnormalities in endocrine-remitted CD patients. Moreover, our study uncovered that the IBP-recovered patients exhibited better recovery of the functional connectivity within circuits of the FPN and CerebN than the IBP-unrecovered ones, as shown in Figure 4a. This result demonstrated that different endocrine-remitted CD patients had different recovery levels for the impaired functional connectivity within circuits of these brain FNs. More importantly, our study further found that the recovered 24hUFC was associated with the improved functional connectivity within FPN circuits in endocrine-remitted CD patients at the 3-month follow-up after treatment (Fig. 5a). This finding indicated that chronic endogenous hypercortisolism in CD patients might be directly related to their FPN impairments. Strengths of This Study The combined longitudinal and cross-sectional analyses have confirmed that the brain functional network abnormalities in CD patients were partially reversible at 3 months after resolution of the hypercortisolism. Since the brain structural abnormalities in endocrine-remitted CD patients were not completely recovered [16], it merits further investigation how the brain structural and functional network recoveries couple with each other in a longitudinal design. The present study provided complementary information to existing neuroimaging studies of CD patients. The existing neuroimaging studies have reported that CD patients had brain volume loss in cortical and cerebellar regions, hippocampus, and amygdala, as well as enlarged ventricles. These structural abnormalities were partially recovered for endocrine-remitted CD patients after treatment [11, 15, 16, 18‒22, 24] or after resolution of the hypercortisolism [12, 18, 24]. CD patients also had reduced cortical thickness in many brain regions including superior frontal cortex, caudal middle frontal cortex, precentral gyrus, insula, precuneus, cuneus, caudal/rostral anterior cingulate gyrus, and posterior cingulate gyrus [17, 54]. In addition, disrupted white matter integrity was observed in CD patients throughout the brain including frontal lobe, temporal lobe, hippocampus, parahippocampal gyrus, cingulate cingulum, corpus callosum, uncinate fasciculus, and cerebellum [10, 25‒27]. Furthermore, metabolic abnormalities in CD patients have been reported in widely distributed brain regions [21, 28‒32], which could be almost completely restored after resolution of hypercortisolism. Besides aforementioned structural and metabolic abnormalities, functional abnormalities have also been reported in CD patients using fMRI [37‒42]. Particularly, abnormal functional activations in CD patients have been observed in the prefrontal cortex, superior/middle/inferior frontal gyrus, superior parietal lobule, superior/middle temporal gyrus, inferior occipital gyrus, rostral/dorsal anterior cingulate gyrus, anterior/middle/posterior hippocampus, amygdala, precuneus, cuneus, lingual gyrus, caudate body, pulvinar/lateral posterior nuclei of the thalamus, and substantia nigra using task fMRI [37‒39, 41]. Abnormal spontaneous functional activities measured by both the amplitude of low-frequency fluctuation and regional homogeneity for CD patients have been observed in the prefrontal cortex, occipital lobe, postcentral gyrus, posterior cingulate gyrus, precuneus, thalamus, and cerebellum [20]. The dysregulation of functional connectivity density of CD patients has been found primarily in the prefrontal cortex, lateral parietal cortex, anterior/posterior cingulate gyrus, and parahippocampal gyrus [55]. The abnormal functional connectivity for CD patients has also been observed between the prefrontal cortex and medial temporal lobe, ventromedial prefrontal cortex and posterior cingulate cortex, anterior cingulate gyrus and limbic network, and lateral occipital cortex and DMN using task fMRI or rs-fMRI [39, 40]. Limitations and Future Work This study has several limitations. First, the longitudinal sample size is not large enough due to the rarity of CD, which might lead to relatively low statistical power and potential biases. Second, our study mainly investigated the brain functional network reversibility of the CD. Studies of the CD’s structural reversibility may provide complementary information to the current study. Third, our study investigated the short-term (3 months) effects of hypercortisolism on large-scale functional brain networks in CD patients. Nevertheless, the long-term effects of hypercortisolism on large-scale functional brain networks remain unclear and merit further investigation. In future work, long-term follow-up data of the CD patients recruited in the current study will be collected to investigate the long-term dynamic changes of their impaired large-scale functional brain networks. Conclusion This is the first study to investigate large-scale functional brain networks and their reversibility in a longitudinal CD cohort by using multivariate analysis. The large-scale functional brain networks, including the CerebN, FPN, and DMN, were impaired due to elevated cortisol levels in active CD patients. More importantly, the impaired functional brain networks of these CD patients were partially restored when their hormone levels returned to normal at 3 months after treatment. The changes of the functional connectivity within the impaired FPN were correlated with changes of the 24hUFC in endocrine-remitted CD patients, while the functional connectivity within the impaired CerebN was closely associated with emotion dysfunctions in active CD patients. These findings suggest that pattern recognition techniques could help identify informative functional brain networks in CD patients, which may help open up novel avenues for their postoperative interventions and assessments after endocrine remission. Statement of Ethics This study confirmed to the Declaration of Helsinki and was approved by the Medical Ethics Committee of Peking Union Medical College Hospital (approval number S-424). Written informed consent was obtained from all participants. Conflict of Interest Statement All authors reported no financial interests or potential conflicts of interest. Funding Sources This study was supported in part by the China Postdoctoral Science Foundation (2020T130070, 2019M650567), and the Clinical Application Research of Capital Characteristic Fund from the Beijing Municipal Science and Technology Commission (Z151100004015099). Author Contributions Bing Xing, Feng Feng, and Yong Fan were involved in study conception and design. Bo Hou, Xiaopeng Guo, Yong Yao, and Ming Feng collected clinical and imaging data. Hewei Cheng, Lu Gao, and Rixing Jing performed data preparation and statistical analysis. Hewei Cheng, Lu Gao, Rixing Jing, Bing Xing, Feng Feng, and Yong Fan were involved in data interpretation. Hewei Cheng, Lu Gao, and Rixing Jing wrote the first draft of the manuscript. Hewei Cheng, Lu Gao, Rixing Jing, Bo Hou, Xiaopeng Guo, Zihao Wang, Ming Feng, Bing Xing, Feng Feng, and Yong Fan provided critical editing and revision of the manuscript for important intellectual content. All authors approved the final version of the manuscript. Additional Information Hewei Cheng and Lu Gao contributed equally to this work. Data Availability Statement All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author. References 1. Newell-Price J, Bertagna X, Grossman AB, Nieman LK. Cushing’s syndrome. Lancet. 2006;367(9522):1605–17. 2. Pivonello R, De Martino MC, De Leo M, Simeoli C, Colao A. Cushing’s disease: the burden of illness. Endocrine. 2017;56(1):10–8. 3. Whelan TB, Schteingart DE, Starkman MN, Smith A. Neuropsychological deficits in Cushing’s syndrome. J Nerv Ment Dis. 1980;168(12):753–7. 4. Forget H, Lacroix A, Somma M, Cohen H. Cognitive decline in patients with Cushing’s syndrome. J Int Neuropsychol Soc. 2000;6(1):20–9. 5. Starkman MN, Giordani B, Berent S, Schork MA, Schteingart DE. Elevated cortisol levels in Cushing’s disease are associated with cognitive decrements. Psychosom Med. 2001;63(6):985–93. 6. Tiemensma J, Kokshoorn NE, Biermasz NR, Keijser B-JSA, Wassenaar MJE, Middelkoop HAM, et al. Subtle cognitive impairments in patients with long-term cure of Cushing’s disease. J Clin Endocrinol Metab. 2010;95(6):2699–714. 7. Pivonello R, Simeoli C, De Martino MC, Cozzolino A, De Leo M, Iacuaniello D, et al. Neuropsychiatric disorders in Cushing’s syndrome. Front Neurosci. 2015;9:129. 8. Pires P, Santos A, Vives-Gilabert Y, Webb SM, Sainz-Ruiz A, Resmini E, et al. White matter involvement on DTI-MRI in Cushing’s syndrome relates to mood disturbances and processing speed: a case-control study. Pituitary. 2017;20(3):340–8. 9. Santos A, Resmini E, Pascual JC, Crespo I, Webb SM. Psychiatric symptoms in patients with Cushing’s syndrome: prevalence, diagnosis and management. Drugs. 2017;77(8):829–42. 10. Valassi E, Crespo I, Keevil BG, Aulinas A, Urgell E, Santos A, et al. Affective alterations in patients with Cushing’s syndrome in remission are associated with decreased BDNF and cortisone levels. Eur J Endocrinol. 2017;176(2):221–31. 11. Frimodt-Møller KE, Møllegaard Jepsen JR, Feldt-Rasmussen U, Krogh J. Hippocampal volume, cognitive functions, depression, anxiety, and quality of life in patients with Cushing syndrome. J Clin Endocrinol Metab. 2019;104(10):4563–77. 12. Kumar N, Jarial KDS, Bhansali A, Nehra R, Vyas S, Walia R. Neurocognitive functions and brain volume in patients with endogenous cushing’s syndrome before and after curative surgery. Indian J Endocrinol Metab. 2020;24(5):396–401. 13. Lin TY, Hanna J, Ishak WW. Psychiatric symptoms in Cushing’s syndrome: a systematic review. Innov Clin Neurosci. 2020;17(1–3):30–5. 14. Na S, Fernandes MA, Ioachimescu AG, Penna S. Neuropsychological and emotional functioning in patients with Cushing’s syndrome. Behav Neurol. 2020;2020:1–10. 15. Starkman MN, Giordani B, Gebarski SS, Berent S, Schork MA, Schteingart DE. Decrease in cortisol reverses human hippocampal atrophy following treatment of Cushing’s disease. Biol Psychiatry. 1999;46(12):1595–602. 16. Andela CD, van der Werff SJ, Pannekoek JN, van den Berg SM, Meijer OC, van Buchem MA, et al. Smaller grey matter volumes in the anterior cingulate cortex and greater cerebellar volumes in patients with long-term remission of Cushing’s disease: a case–control study. Eur J Endocrinol. 2013;169(6):811–9. 17. Crespo I, Esther G-M, Santos A, Valassi E, Yolanda V-G, De Juan-Delago M, et al. Impaired decision-making and selective cortical frontal thinning in Cushing’s syndrome. Clin Endocrinol. 2014;81(6):826–33. 18. Andela CD, van Haalen FM, Ragnarsson O, Papakokkinou E, Johannsson G, Santos A, et al. Mechanisms in endocrinology: cushing’s syndrome causes irreversible effects on the human brain: a systematic review of structural and functional magnetic resonance imaging studies. Eur J Endocrinol. 2015;173(1):R1–4. 19. Burkhardt T, Lüdecke D, Spies L, Wittmann L, Westphal M, Flitsch J. Hippocampal and cerebellar atrophy in patients with Cushing’s disease. Neurosurg Focus. 2015;39(5):E5–7. 20. Jiang H, Ren J, He NY, Liu C, Sun YH, Jian FF, et al. Volumetric magnetic resonance imaging analysis in patients with short-term remission of Cushing’s disease. Clin Endocrinol. 2017;87(4):367–74. 21. Bauduin SEEC, van der Wee NJA, van der Werff SJA. Structural brain abnormalities in Cushing’s syndrome. Curr Opin Endocrinol Diabetes Obes. 2018;25(4):285–9. 22. Jiang H, Liu C, Pan S-J, Ren J, He N-Y, Sun Y-H, et al. Reversible and the irreversible structural alterations on brain after resolution of hypercortisolism in Cushing’s disease. Steroids. 2019;151:108457. 23. Bauduin SEEC, Pal Z, Pereira AM, Meijer OC, Giltay EJ, Wee NJA, et al. Cortical thickness abnormalities in long-term remitted Cushing’s disease. Transl Psychiatry. 2020;10(1):239. 24. Hou B, Gao L, Shi L, Luo Y, Guo X, Young GS, et al. Reversibility of impaired brain structures after transsphenoidal surgery in Cushing’s disease: a longitudinal study based on an artificial intelligence-assisted tool. J Neurosurg. 2020;134(2):1–10. 25. van der Werff SJ, Andela CD, Nienke Pannekoek J, Meijer OC, van Buchem MA, Rombouts SA, et al. Widespread reductions of white matter integrity in patients with long-term remission of Cushing’s disease. Neuroimage Clin. 2014;4:659–67. 26. Pires P, Santos A, Vives-Gilabert Y, Webb S, Sainz-Ruiz A, Resmini E, et al. White matter alterations in the brains of patients with active, remitted, and cured Cushing syndrome: a DTI study. Am J Neuroradiol. 2015;36(6):1043–8. 27. Jiang H, He N-Y, Sun Y-H, Jian F-F, Bian L-G, Shen J-K, et al. Altered gray and white matter microstructure in Cushing’s disease: a diffusional kurtosis imaging study. Brain Res. 2017;1665:80–7. 28. Khiat A, Bard C, Lacroix A, Boulanger Y. Recovery of the brain choline level in treated Cushing’s patients as monitored by proton magnetic resonance spectroscopy. Brain Res. 2000;862(1–2):301–7. 29. Resmini E, Santos A, Gómez-Anson B, López-Mourelo O, Pires P, Vives-Gilabert Y, et al. Hippocampal dysfunction in cured Cushing’s syndrome patients, detected by 1H‐MR‐spectroscopy. Clin Endocrinol. 2013;79(5):700–7. 30. Crespo I, Santos A, Gómez-Ansón B, López-Mourelo O, Pires P, Vives-Gilabert Y, et al. Brain metabolite abnormalities in ventromedial prefrontal cortex are related to duration of hypercortisolism and anxiety in patients with Cushing’s syndrome. Endocrine. 2016;53(3):848–56. 31. Liu S, Wang Y, Xu K, Ping F, Li F, Wang R, et al. Voxel-based comparison of brain glucose metabolism between patients with Cushing’s disease and healthy subjects. Neuroimage Clin. 2018;17:354–8. 32. Cheng H, Gao L, Hou B, Feng F, Guo X, Wang Z, et al. Reversibility of cerebral blood flow in patients with Cushing’s disease after surgery treatment. Metabolism. 2020;104:154050. 33. Gao L, Liu L, Shi L, Luo Y, Wang Z, Guo X, et al. Dynamic changes of views on the brain changes of Cushing’s syndrome using different computer-assisted tool. Rev Endocr Metab Disord. 2020;21(1):185–200. 34. Piasecka M, Papakokkinou E, Valassi E, Santos A, Webb SM, de Vries F, et al. Psychiatric and neurocognitive consequences of endogenous hypercortisolism. J Intern Med. 2020;288(2):168–82. 35. Kelly DF. Transsphenoidal surgery for Cushing’s disease: a review of success rates, remission predictors, management of failed surgery, and Nelson’s Syndrome. Neurosurg Focus. 2007;23(3):E5. 36. Theodoropoulou M, Reincke M. Tumor-directed therapeutic targets in Cushing disease. J Clin Endocrinol Metab. 2019;104(3):925–33. 37. Maheu FS, Mazzone L, Merke DP, Keil MF, Stratakis CA, Pine DS, et al. Altered amygdala and hippocampus function in adolescents with hypercortisolemia: a functional magnetic resonance imaging study of Cushing syndrome. Dev Psychopathol. 2008;20(4):1177–89. 38. Langenecker SA, Weisenbach SL, Giordani B, Briceño EM, Guidotti Breting LM, Schallmo MP, et al. Impact of chronic hypercortisolemia on affective processing. Neuropharmacology. 2012;62(1):217–25. 39. Bas-Hoogendam JM, Andela CD, van der Werff SJA, Pannekoek JN, van Steenbergen H, Meijer OC, et al. Altered neural processing of emotional faces in remitted Cushing's disease. Psychoneuroendocrinology. 2015;59:134–46. 40. Van Der Werff SJ, Pannekoek JN, Andela CD, Meijer OC, Van Buchem MA, Rombouts SA, et al. Resting-state functional connectivity in patients with long-term remission of Cushing’s disease. Neuropsychopharmacology. 2015;40(8):1888–98. 41. Ragnarsson O, Stomby A, Dahlqvist P, Evang JA, Ryberg M, Olsson T, et al. Decreased prefrontal functional brain response during memory testing in women with Cushing’s syndrome in remission. Psychoneuroendocrinology. 2017;82:117–25. 42. Stomby A, Salami A, Dahlqvist P, Evang JA, Ryberg M, Bollerslev J, et al. Elevated resting-state connectivity in the medial temporal lobe and the prefrontal cortex among patients with Cushing’s syndrome in remission. Eur J Endocrinol. 2019;180(5):329–38. 43. Zung WW. A rating instrument for anxiety disorders. Psychosomatics. 1971;12(6):371–9. 44. Kim J-H, Lee J-M, Jo HJ, Kim SH, Lee JH, Kim ST, et al. Defining functional SMA and pre-SMA subregions in human MFC using resting state fMRI: functional connectivity-based parcellation method. NeuroImage. 2010;49(3):2375–86. 45. Power JD, Barnes KA, Snyder AZ, Schlaggar BL, Petersen SE. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. NeuroImage. 2012;59(3):2142–54. 46. Nieman LK, Biller BM, Findling JW, Murad MH, Newell-Price J, Savage MO, et al. Treatment of Cushing’s syndrome: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. 2015;100(8):2807–31. 47. Du Y, Fan Y. Group information guided ICA for fMRI data analysis. NeuroImage. 2013;69:157–97. 48. Fan Y, Liu Y, Wu H, Hao Y, Liu H, Liu Z, et al. Discriminant analysis of functional connectivity patterns on Grassmann manifold. NeuroImage. 2011;56(4):2058–67. 49. Li P, Jing RX, Zhao RJ, Ding ZB, Shi L, Sun HQ, et al. Electroconvulsive therapy-induced brain functional connectivity predicts therapeutic efficacy in patients with schizophrenia: a multivariate pattern recognition study. NPJ Schizophr. 2017;3(1):21–9. 50. Forget H, Lacroix A, Bourdeau I, Cohen H. Long-term cognitive effects of glucocorticoid excess in Cushing’s syndrome. Psychoneuroendocrinology. 2016;65:26–33. 51. Mauri M, Sinforiani E, Bono G, Vignati F, Berselli M, Attanasio R, et al. Memory impairment in Cushing’s disease. Acta Neurol Scand. 1993;87(1):52–5. 52. Jiang H, He NY, Sun YH, Jian FF, Bian LG, Shen JK, et al. Altered spontaneous brain activity in Cushing’s disease: a resting-state functional MRI study. Clin Endocrinol. 2017;86(3):367–76. 53. Amaya JM, Viho EM, Sips HC, Lalai RA, Sahut-Barnola I, Dumontet T, et al. Gene expression changes in the brain of a Cushing’s syndrome mouse model. J Neuroendocrinol. 2022;34(4):e13125. 54. Bauduin SEEC, van der Pal Z, Pereira AM, Meijer OC, Giltay EJ, van der Wee NJA, et al. Cortical thickness abnormalities in long-term remitted Cushing’s disease. Transl Psychiatry. 2020;10(1):293. 55. Wang X, Zhou T, Wang P, Zhang L, Feng S, Meng X, et al. Dysregulation of resting-state functional connectivity in patients with Cushing’s disease. Neuroradiology. 2019;61(8):911–20. © 2023 The Author(s). Published by S. Karger AG, Basel Open Access License / Drug Dosage / Disclaimer This article is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC). Usage and distribution for commercial purposes requires written permission. Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug. Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements. This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License . From https://karger.com/nen/article/doi/10.1159/000534789/869375/Reversibility-of-Impaired-Large-Scale-Functional 1 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.