Multimodal imaging measures in the prediction of clinical response to deep brain stimulation for refractory depression: A machine learning approach.

OBJECTIVES: This study compared machine learning models using unimodal imaging measures and combined multi-modal imaging measures for deep brain stimulation (DBS) outcome prediction in treatment resistant depression (TRD). METHODS: Regional brain glucose metabolism (CMRGlu), cerebral blood flow (CBF), and grey matter volume (GMV) were measured at baseline using 18F-fluorodeoxy glucose (18F-FDG) positron emission tomography (PET), arterial spin labelling (ASL) magnetic resonance imaging (MRI), and T1-weighted MRI, respectively, in 19 patients with TRD receiving subcallosal cingulate (SCC)-DBS. Responders (n = 9) were defined by a 50% reduction in HAMD-17 at 6 months from the baseline. Using an atlas-based approach, values of each measure were determined for pre-selected brain regions. OneR feature selection algorithm and the naïve Bayes model was used for classification. Leave-out-one cross validation was used for classifier evaluation. RESULTS: The performance accuracy of the CMRGlu classification model (84%) was greater than CBF (74%) or GMV (74%) models. The classification model using the three image modalities together led to a similar accuracy (84%0 compared to the CMRGlu classification model. CONCLUSIONS: CMRGlu imaging measures may be useful for the development of multivariate prediction models for SCC-DBS studies for TRD. The future of multivariate methods for multimodal imaging may rest on the selection of complementing features and the developing better models.Clinical Trial Registration: ClinicalTrials.gov (#NCT01983904).

The phenotypic spectrum of KCNT1: a new family with variable epilepsy syndromes including mild focal epilepsy.

BACKGROUND AND OBJECTIVE: Pathogenic variants in KCNT1 have been associated with severe forms of epilepsy, typically sleep-related hypermotor epilepsy or epilepsy of infancy with migrating focal seizures. To show that pathogenic variants in KCNT1 can be associated with mild extra-frontal epilepsy, we report a KCNT1 family with a wide spectrum of phenotypes ranging from developmental and epileptic encephalopathy to mild focal epilepsy without cognitive regression and not consistent with sleep-related hypermotor epilepsy. METHODS: A large Canadian family of Caucasian descent including 9 affected family members was recruited. Family members were phenotyped by direct interview and review of existing medical records. Clinical epilepsy gene panel analysis and exome sequencing were performed. RESULTS: Phenotypic information was available for five family members of which two had developmental and epileptic encephalopathy and three had normal development and focal epilepsy with presumed extra-frontal onset. All three had predominantly nocturnal seizures that did not show hyperkinetic features. All three reported clusters of seizures at night with a feeling of being unable to breathe associated with gasping for air, choking and/or repetitive swallowing possibly suggesting insular or opercular involvement. Genetic analysis identified a heterozygous KCNT1 c.2882G > A, p.Arg961His variant that was predicted to be deleterious. DISCUSSION: This family demonstrates that the phenotypic spectrum associated with KCNT1 pathogenic variants is broader than previously assumed. Our findings indicate that variants in KCNT1 can be associated with mild focal epilepsy and should not be excluded during variant interpretation in such patients based solely on gene-disease validity.

Impact of ictal subtraction SPECT and PET in presurgical evaluation.

OBJECTIVE: To assess the relative contribution of ictal subtraction single-photon emission computed tomography (ISSPECT) and 18 F-fluorodeoxyglucose positron emission tomography computed tomography (PET) in epilepsy surgery decision making. MATERIALS AND METHODS: A retrospective 3-year study of consecutive patients with resistant focal epilepsy who underwent ISSPECT and PET to evaluate to what extent these modalities influence decisions in epilepsy surgery and outcomes. ISSPECT imaging was performed in 106 patients and 58 (55%) had PET also. The clinical consensus (ClinC) was the final arbiter for decisions. Post-surgical outcomes were collected from follow-up clinics. Non-parametric statistics were used to assess association and logistic regression to evaluate prediction of outcomes. RESULTS: Of 106 patients, 60 were males (57%). MRI was non-lesional in 46 (43%). Concordance with ClinC was seen in 80 patients (76%) for ISSPECT, in 46 patients (79%) for PET, and in 37 patients (64%) for ISSPECT + PET. Fifty-six patients (53%) were planned for intracranial video-electroencephalography monitoring (IVEM). Those with ClinC-PET concordance were likely to proceed to IVEM (p = 0.02). ClinC-PET concordance and ClinC-ISSPECT concordance did not predict decision to proceed to surgery, but VEM-MRI concordance did in lesional cases (p = 0.018). Forty-five (42%) underwent surgery of which 29 had minimum follow-up for 1 year (mean, 20 months; SD, 8) and 22 (76%) had Engel class I outcomes. ClinC-ISSPECT concordance (p = 0.024) and VEM-MRI concordance (p = 0.016) predicted Engel class I outcomes. CONCLUSION: Those with ClinC-PET concordance were more likely to proceed with IVEM. ClinC-ISSPECT concordance and VEM-MRI concordance predicted good surgical outcomes.

Metabolic activity in subcallosal cingulate predicts response to deep brain stimulation for depression.

Subcallosal cingulate (SCC) deep brain stimulation (DBS) is a promising therapy for treatment-resistant depression (TRD), but response rates in open-label studies were not replicated in a large multicenter trial. Identifying biomarkers of response could improve patient selection and outcomes. We examined SCC metabolic activity as both a predictor and marker of SCC DBS treatment response. Brain glucose metabolism (CMRGlu) was measured with [18F] FDG-PET at baseline and 6 months post DBS in 20 TRD patients in a double-blind randomized controlled trial where two stimulation types (long pulse width (LPW) n = 9 and short pulse width (SPW) n = 11) were used. Responders (n = 10) were defined by a ≥48% reduction in Hamilton Depression Rating Scale scores after 6 months. The response rates were similar with five responders in each stimulation group: LPW (55.6%) and SPW (44.5%). First, differences in SCC CMRGlu in responders and non-responders were compared at baseline. Then machine learning analysis was performed with a leave-one-out cross-validation using a Gaussian naive Bayes classifier to test whether baseline CMRGlu in SCC could categorize responders. Finally, we compared 6-month change in metabolic activity with change in depression severity. All analyses were controlled for age. Baseline SCC CMRGlu was significantly higher in responders than non-responders. The machine learning analysis predicted response with 80% accuracy. Furthermore, reduction in SCC CMRGlu 6 months post DBS correlated with symptom improvement (r(17) = 0.509; p = 0.031). This is the first evidence of an image-based treatment selection biomarker that predicts SCC DBS response. Future studies could utilize SCC metabolic activity for prospective patient selection.

Clinical utility of PET/CT in the evaluation of head and neck squamous cell carcinoma with an unknown primary: a prospective clinical trial.

BACKGROUND: Metastatic head and neck squamous cell carcinoma with an unknown primary is an uncommon but important problem. PET/CT, as an adjunct to diagnosis, is potentially useful but has never been studied in a prospective, single-blinded clinical trial. METHODS: In all, 20 subjects with cervical metastases from an unknown head and neck primary were enrolled in a prospective clinical trial. A standard protocol was used in both clinic and operating room (OR). Study surgeons were blinded to the PET/CT result upon completion of the standard work-up. RESULTS: PET/CT increased the detection of a primary site from 25% to 55% (5 vs 11 subjects). This difference was statistically and clinically significant (p = .03, McNemar's test). There was 1 false negative PET/CT scan. CONCLUSIONS: An unknown primary should be diagnosed only after a complete head and neck examination, flexible endoscopy, and CT or MRI. PET/CT performed prior to panendoscopy will increase the diagnostic yield in the unknown head and neck primary population, leading to more targeted, and less morbid, treatment.