Federated learning enables big data for rare cancer boundary detection.

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.

Safety and Efficacy of Crizotinib in Combination with Temozolomide and Radiotherapy in Patients with Newly Diagnosed Glioblastoma: Phase Ib GEINO 1402 Trial.

BACKGROUND: MET-signaling and midkine (ALK ligand) promote glioma cell maintenance and resistance against anticancer therapies. ALK and c-MET inhibition with crizotinib have a preclinical therapeutic rationale to be tested in newly diagnosed GBM. METHODS: Eligible patients received crizotinib with standard radiotherapy (RT)/temozolomide (TMZ) followed by maintenance with crizotinib. The primary objective was to determine the recommended phase 2 dose (RP2D) in a 3 + 3 dose escalation (DE) strategy and safety evaluation in the expansion cohort (EC). Secondary objectives included progression-free (PFS) and overall survival (OS) and exploratory biomarker analysis. RESULTS: The study enrolled 38 patients. The median age was 52 years (33-76), 44% were male, 44% were MGMT methylated, and three patients had IDH1/2 mutation. In DE, DLTs were reported in 1/6 in the second cohort (250 mg/QD), declaring 250 mg/QD of crizotinib as the RP2D for the EC. In the EC, 9/25 patients (32%) presented grade ≥3 adverse events. The median follow up was 18.7 months (m) and the median PFS was 10.7 m (95% CI, 7.7-13.8), with a 6 m PFS and 12 m PFS of 71.5% and 38.8%, respectively. At the time of this analysis, 1 died without progression and 24 had progressed. The median OS was 22.6 m (95% CI, 14.1-31.1) with a 24 m OS of 44.5%. Molecular biomarkers showed no correlation with efficacy. CONCLUSIONS: The addition of crizotinib to standard RT and TMZ for newly diagnosed GBM was safe and the efficacy was encouraging, warranting prospective validation in an adequately powered, randomized controlled study.

Randomized Phase III Trial of Prophylactic Cranial Irradiation With or Without Hippocampal Avoidance for Small-Cell Lung Cancer (PREMER): A GICOR-GOECP-SEOR Study.

PURPOSE: Radiation dose received by the neural stem cells of the hippocampus during whole-brain radiotherapy has been associated with neurocognitive decline. The key concern using hippocampal avoidance-prophylactic cranial irradiation (HA-PCI) in patients with small-cell lung cancer (SCLC) is the incidence of brain metastasis within the hippocampal avoidance zone. METHODS: This phase III trial enrolled 150 patients with SCLC (71.3% with limited disease) to standard prophylactic cranial irradiation (PCI; 25 Gy in 10 fractions) or HA-PCI. The primary objective was the delayed free recall (DFR) on the Free and Cued Selective Reminding Test (FCSRT) at 3 months; a decrease of 3 points or greater from baseline was considered a decline. Secondary end points included other FCSRT scores, quality of life (QoL), evaluation of the incidence and location of brain metastases, and overall survival (OS). Data were recorded at baseline, and 3, 6, 12, and 24 months after PCI. RESULTS: Participants' baseline characteristics were well balanced between the two groups. The median follow-up time for living patients was 40.4 months. Decline on DFR from baseline to 3 months was lower in the HA-PCI arm (5.8%) compared with the PCI arm (23.5%; odds ratio, 5; 95% CI, 1.57 to 15.86; P = .003). Analysis of all FCSRT scores showed a decline on the total recall (TR; 8.7% v 20.6%) at 3 months; DFR (11.1% v 33.3%), TR (20.3% v 38.9%), and total free recall (14.8% v 31.5%) at 6 months, and TR (14.2% v 47.6%) at 24 months. The incidence of brain metastases, OS, and QoL were not significantly different. CONCLUSION: Sparing the hippocampus during PCI better preserves cognitive function in patients with SCLC. No differences were observed with regard to brain failure, OS, and QoL compared with standard PCI.

Voxel-based morphometry for the evaluation of patients with pharmacoresistant epilepsy with apparently normal MRI.

BACKGROUND AND PURPOSE: Magnetic resonance imaging (MRI) is essential in the diagnosis of pharmacoresistant epilepsy (PRE), because patients with lesions detected by MRI have a better prognosis after surgery. Focal cortical dysplasia (FCD) is one of the most frequent etiologies of PRE but can be difficult to identify by MRI. Voxel-based morphometric analysis programs, like the Morphometric Analysis Program (MAP), have been developed to help improve MRI detection. Our objective was to evaluate the clinical usefulness of MAP in patients with PRE and an apparently normal MRI. METHODS: We studied 70 patients with focal PRE and a nonlesional MRI. The 3DT1 sequence was processed with MAP, obtaining three z-score maps. Patients were classified as MAP+ if one or more z-score maps showed a suspicious area of brightness, and MAP- if the z-score maps did not show any suspicious areas. For MAP+ cases, a second-look MRI was performed with a dedicated inspection based on the MAP findings. The MAP results were correlated with the epileptogenic zone. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. RESULTS: Thirty-one percent of patients were classified as MAP+ and 69% were MAP-. Results showed a sensitivity of 0.57, specificity of 0.8, PPV of 0.91, and NPV of 0.35. In 19% of patients, an FCD was found in the second-look MRI after MAP. CONCLUSIONS: MAP was helpful in the detection of lesions in PRE patients with a nonlesional MRI, which could have important repercussions for the clinical management and postoperative prognosis of these patients.

RNA sequencing and Immunohistochemistry Reveal ZFN7 as a Stronger Marker of Survival than Molecular Subtypes in G-CIMP-negative Glioblastoma.

PURPOSE: Glioblastoma is the most aggressive brain tumor in adults and has few therapeutic options. The study of molecular subtype classifications may lead to improved prognostic classification and identification of new therapeutic targets. The Cancer Genome Atlas (TCGA) subtype classification has mainly been applied in U.S. clinical trials, while the intrinsic glioma subtype (IGS) has mainly been applied in European trials. EXPERIMENTAL DESIGN: From paraffin-embedded tumor samples of 432 patients with uniformly treated, newly diagnosed glioblastoma, we built tissue microarrays for IHC analysis and applied RNA sequencing to the best samples to classify them according to TCGA and IGS subtypes. RESULTS: We obtained transcriptomic results from 124 patients. There was a lack of agreement among the three TCGA classificatory algorithms employed, which was not solely attributable to intratumoral heterogeneity. There was overlapping of TCGA mesenchymal subtype with IGS cluster 23 and of TCGA classical subtype with IGS cluster 18. Molecular subtypes were not associated with prognosis, but levels of expression of 13 novel genes were identified as independent prognostic markers in glioma-CpG island methylator phenotype-negative patients, independently of clinical factors and MGMT methylation. These findings were validated in at least one external database. Three of the 13 genes were selected for IHC validation. In particular, high ZNF7 RNA expression and low ZNF7 protein expression were strongly associated with longer survival, independently of molecular subtypes. CONCLUSIONS: TCGA and IGS molecular classifications of glioblastoma have no higher prognostic value than individual genes and should be refined before being applied to clinical trials.

Pseudoaneurysm of the lingual artery as an unusual radiological finding of oropharyngeal carcinoma.

Pseudoaneurysms of the lingual artery are an extremely rare entity and often are consequence of neck surgery, trauma or inflammation (e.g., due to chemoradiotherapy or odontogenic infection), and may cause life-threatening bleeding. To our knowledge, this is the first report of buccal bleeding secondary to the presence of a previously undiagnosed oropharyngeal carcinoma with an associated lingual artery pseudoaneurysm.

Glioblastoma TCGA Mesenchymal and IGS 23 Tumors are Identifiable by IHC and have an Immune-phenotype Indicating a Potential Benefit from Immunotherapy.

PURPOSE: Molecular subtype classifications in glioblastoma may detect therapy sensitivities. IHC would potentially allow the identification of molecular subtypes in routine clinical practice. EXPERIMENTAL DESIGN: Formalin-fixed, paraffin-embedded tumor samples of 124 uniformly treated, newly diagnosed patients with glioblastoma were submitted to RNA sequencing, IHC, and immune-phenotyping to identify differences in molecular subtypes associated with treatment sensitivities. RESULTS: We detected high molecular and IHC overlapping of the The Cancer Genome Atlas (TCGA) mesenchymal subtype with instrinsic glioma subtypes (IGS) cluster 23 and of the TCGA classical subtype with IGS cluster 18. IHC patterns, gene fusion profiles, and immune-phenotypes varied across subtypes. IHC revealed that the TCGA classical subtype was identified by high expression of EGFR and low expression of PTEN, while the mesenchymal subtype was identified by low expression of SOX2 and high expression of two antibodies, SHC1 and TCIRG1, selected on the basis of RNA differential transcriptomic expression. The proneural subtype was identified by frequent positive IDH1 expression and high Olig2 and Ki67 expression. Immune-phenotyping showed that mesenchymal and IGS 23 tumors exhibited a higher positive effector cell score, a higher negative suppressor cell score, and lower levels of immune checkpoint molecules. The cell-type deconvolution analysis revealed that these tumors are highly enriched in M2 macrophages, resting memory CD4+ T cells, and activated dendritic cells, indicating that they may be ideal candidates for immunotherapy, especially with anti-M2 and/or dendritic cell vaccination. CONCLUSIONS: There is a subset of tumors, frequently classified as mesenchymal or IGS cluster 23, that may be identified with IHC and could well be optimal candidates for immunotherapy.

Misdiagnosis Worsens Prognosis in Subarachnoid Hemorrhage With Good Hunt and Hess Score.

Background and Purpose- Our aim was to describe variables associated with initial misdiagnosis of subarachnoid hemorrhage (SAH). We also analyzed the relationship of misdiagnosis with poor outcome and complications in good Hunt and Hess (HH) cases. Methods- In a prospective cohort of 401 patients with SAH, misdiagnosis was defined as failure to correctly identify, at first physician contact, a subsequently documented SAH; this meant no urgent radiological study and lumbar puncture was performed. Poor outcome was defined as modified Rankin Scale score 3 to 6 at 3-month follow-up. We recorded age, sex, hypertension, diabetes mellitus, current smoking, previous antithrombotic treatment, initial HH and radiological severity, presence of aneurysm, first therapeutic procedure, hydrocephalus, delayed cerebral ischemia (DCI), rebleeding, and procedure-related complications. Results- Misdiagnosis was confirmed in 104/401 (25.9%) patients, who also had a longer time-to-admission to hospital. Misdiagnosis was associated with less clinical and radiological severity, compared with a correct diagnosis; the 2 groups did not differ in age or cardiovascular risk factor profile. Poor outcome was registered in 167/401 patients (41.6%). Age, misdiagnosis, and greater clinical and radiological initial severity were independent predictors of poor outcome. In the 236 patients (58.8% of cohort) with HH 1-2, misdiagnosis was associated with poor outcome in univariate and multivariate analysis, respectively (odds ratio=3.89; 95% CI, 1.89-8.01). Delayed cerebral ischemia (odds ratio=2.47; 95% CI, 1.2-5.09) and procedure-related complications (odds ratio=2.27; 95% CI, 1.07-4.82) were independently associated with misdiagnosis. Conclusions- Misdiagnosis is an unresolved problem in SAH, and it is a missed opportunity for good outcome in patients with HH 1-2. The poor outcome is partially explained by a higher risk of delayed cerebral ischemia and procedure-related complications in misdiagnosed patients. There is a need to improve the diagnostic strategy in patients reporting only a headache (HH 1-2) after SAH.

The RANO Leptomeningeal Metastasis Group proposal to assess response to treatment: lack of feasibility and clinical utility and a revised proposal.

BACKGROUND: A scorecard to evaluate magnetic resonance imaging (MRI) findings during the course of leptomeningeal metastases (LM) has been proposed by the Response Assessment in Neuro-Oncology (RANO) group. METHODS: To explore the feasibility of the Leptomeningeal Assessment in Neuro-Oncology (LANO) scorecard, cerebrospinal MRIs of 22 patients with LM from solid tumors were scored by 10 neuro-oncologists and 9 neuroradiologists at baseline and at follow-up after treatment. Raters were blinded for clinical data including treatment. Agreement between raters of single items was evaluated using a Krippendorff alpha coefficient. Agreement between numerical parameters such as scores for changes between baseline and follow-up and total scores was evaluated by determining the intraclass coefficient of correlation. RESULTS: Most raters experienced problems with the instructions of the scorecard. No acceptable alpha concordance coefficient was obtained for the rating of single items at baseline or follow-up. The most concordant ratings were obtained for spinal nodules. The concordances were worst for brain linear leptomeningeal enhancement and cranial nerve enhancement. Discordance was less prominent among neuroradiologists than among neuro-oncologists. High variability was also observed for evaluating changes between baseline and follow-up and for total scores. CONCLUSIONS: Assessing response of LM by MRI remains challenging. Central imaging review is therefore indispensable for clinical trials. Based on the present results, we propose a new, simplified scorecard that will require validation using a similar approach as pursued here. The main challenges are to define measurable versus nonmeasurable (target) lesions and measures of change that allow assessment of response.

Macrovascular Networks on Contrast-Enhanced Magnetic Resonance Imaging Improves Survival Prediction in Newly Diagnosed Glioblastoma.

A higher degree of angiogenesis is associated with shortened survival in glioblastoma. Feasible morphometric parameters for analyzing vascular networks in brain tumors in clinical practice are lacking. We investigated whether the macrovascular network classified by the number of vessel-like structures (nVS) visible on three-dimensional T1-weighted contrast⁻enhanced (3D-T1CE) magnetic resonance imaging (MRI) could improve survival prediction models for newly diagnosed glioblastoma based on clinical and other imaging features. Ninety-seven consecutive patients (62 men; mean age, 58 ± 15 years) with histologically proven glioblastoma underwent 1.5T-MRI, including anatomical, diffusion-weighted, dynamic susceptibility contrast perfusion, and 3D-T1CE sequences after 0.1 mmol/kg gadobutrol. We assessed nVS related to the tumor on 1-mm isovoxel 3D-T1CE images, and relative cerebral blood volume, relative cerebral flow volume (rCBF), delay mean time, and apparent diffusion coefficient in volumes of interest for contrast-enhancing lesion (CEL), non-CEL, and contralateral normal-appearing white matter. We also assessed Visually Accessible Rembrandt Images scoring system features. We used ROC curves to determine the cutoff for nVS and univariate and multivariate cox proportional hazards regression for overall survival. Prognostic factors were evaluated by Kaplan-Meier survival and ROC analyses. Lesions with nVS > 5 were classified as having highly developed macrovascular network; 58 (60.4%) tumors had highly developed macrovascular network. Patients with highly developed macrovascular network were older, had higher volumeCEL, increased rCBFCEL, and poor survival; nVS correlated negatively with survival (r = -0.286; p = 0.008). On multivariate analysis, standard treatment, age at diagnosis, and macrovascular network best predicted survival at 1 year (AUC 0.901, 83.3% sensitivity, 93.3% specificity, 96.2% PPV, 73.7% NPV). Contrast-enhanced MRI macrovascular network improves survival prediction in newly diagnosed glioblastoma.

Is a pretreatment radiological staging system feasible for suggesting the optimal extent of resection and predicting prognosis in glioblastoma? An observational study.

To assess the value of resection in glioblastoma based on pre-surgical tumor characteristics and a subsequent staging system. The lack of a staging system for glioblastoma hinders the analysis of treatment outcome. We classified 292 uniformly treated glioblastoma patients as stage I, II, or III based on tumor size, location, and eloquence and then analyzed the impact of the extent of resection. We classified 62% of patients as stage I, 25.3% as stage II, and 12.7% as stage III. Gross total resection (GTR) was performed mainly in stage I rather than stage II or III patients (79.2% vs. 14.6% vs. 6.3%; P < 0.001). Overall survival (OS) was 17.7, 14.6, and 10.8 months for stage I, II, and III patients, respectively (P = 0.005). Longer OS was significantly associated with greater extent of resection, younger age, KPS ≥ 70%, MGMT methylation, lower stage, and tumor ≤ 5 cm. In the subgroups of stage I (P = 0.04) and stage II (P < 0.001)-but not stage III-patients, GTR and partial resection (PR) were associated with longer OS. We constructed several multivariable models including different variables, and greater extent of resection, smaller tumor size, and MGMT methylation consistently emerged as independent markers of longer OS. This staging system provides a feasible tool for comparison of results. We confirmed the value of partial resection in stage I and II patients, in contrast to some reports suggesting that biopsy only is sufficient when gross total resection cannot be safely achieved.

Post-Operative Localization of Deep Brain Stimulation Electrodes in the Subthalamus Using Transcranial Sonography.

OBJECTIVES: The correct positioning of deep brain stimulation electrodes determines the success of surgery. In this study, we attempt to validate transcranial sonography (TCS) as a method for early postoperative confirmation of electrode location in the subthalamic nucleus (STN). MATERIALS AND METHODS: Nineteen patients diagnosed with Parkinson's disease were enrolled in the study. Postoperative TCS was applied to measure the distance between the implanted electrodes and the third ventricle in the axial plane. Whether the electrodes were positioned within or outside the substantia nigra (SN) was evaluated through measurements in the coronal plane. The obtained metrics through TCS were compared with those from postoperative computed tomography (CT) and magnetic resonance imaging (MRI). RESULTS: A statistically significant correlation between distances from electrode to third ventricle by TCS and CT/MRI (r = 0.75, p < 0.01) was observed. Distances from third ventricle to electrodes tips were different when sonographically they showed to be inside or outside the SN (p < 0.01). A cut-off value of 8.85mm in these distances was the most sensitive (100%) and specific (90.5%) to predict if electrodes were positioned inside the SN (CI 95% 0.81-10.30, p = 0.001). CONCLUSIONS: Transcranial sonography is a useful technique to reliably identify targeted positioning of deep brain stimulation electrodes in or out of the SN.

Pseudoprogression as an adverse event of glioblastoma therapy.

We explored predictive factors of pseudoprogression (PsP) and its impact on prognosis in a retrospective series of uniformly treated glioblastoma patients. Patients were classified as having PsP, early progression (eP) or neither (nP). We examined potential associations with clinical, molecular, and basal imaging characteristics and compared overall survival (OS), progression-free survival (PFS), post-progression survival (PPS) as well as the relationship between PFS and PPS in the three groups. Of the 256 patients studied, 56 (21.9%) were classified as PsP, 70 (27.3%) as eP, and 130 (50.8%) as nP. Only MGMT methylation status was associated to PsP. MGMT methylated patients had a 3.5-fold greater possibility of having PsP than eP (OR: 3.48; 95% CI: 1.606-7.564; P = 0.002). OS was longer for PsP than eP patients (18.9 vs. 12.3 months; P = 0.0001) but was similar for PsP and nP patients (P = 0.91). OS was shorter-though not significantly so-for PsP than nP patients (OS: 19.5 vs. 27.9 months; P = 0.63) in methylated patients. PPS was similar for patients having PsP, eP or nP (PPS: 7.2 vs. 5.4 vs. 6.7; P = 0.43). Neurological deterioration occurred in 64.3% of cases at the time they were classified as PsP and in 72.8% of cases of eP (P = 0.14). PsP confounds the evaluation of disease and does not confer a survival advantage in glioblastoma.

Geometrical Measures Obtained from Pretreatment Postcontrast T1 Weighted MRIs Predict Survival Benefits from Bevacizumab in Glioblastoma Patients.

BACKGROUND: Antiangiogenic therapies for glioblastoma (GBM) such as bevacizumab (BVZ), have been unable to extend survival in large patient cohorts. However, a subset of patients having angiogenesis-dependent tumors might benefit from these therapies. Currently, there are no biomarkers allowing to discriminate responders from non-responders before the start of the therapy. METHODS: 40 patients from the randomized GENOM009 study complied the inclusion criteria (quality of images, clinical data available). Of those, 23 patients received first line temozolomide (TMZ) for eight weeks and then concomitant radiotherapy and TMZ. 17 patients received BVZ+TMZ for seven weeks and then added radiotherapy to the treatment. Clinical variables were collected, tumors segmented and several geometrical measures computed including: Contrast enhancing (CE), necrotic, and total volumes; equivalent spherical CE width; several geometric measures of the CE 'rim' geometry and a set of image texture measures. The significance of the results was studied using Kaplan-Meier and Cox proportional hazards analysis. Correlations were assessed using Spearman correlation coefficients. RESULTS: Kaplan-Meier and Cox proportional hazards analysis showed that total, CE and inner volume (p = 0.019, HR = 4.258) and geometric heterogeneity of the CE areas (p = 0.011, HR = 3.931) were significant parameters identifying response to BVZ. The group of patients with either regular CE areas (small geometric heterogeneity, median difference survival 15.88 months, p = 0.011) or those with small necrotic volume (median survival difference 14.50 months, p = 0.047) benefited substantially from BVZ. CONCLUSION: Imaging biomarkers related to the irregularity of contrast enhancing areas and the necrotic volume were able to discriminate GBM patients with a substantial survival benefit from BVZ. A prospective study is needed to validate our results.

Bevacizumab and temozolomide versus temozolomide alone as neoadjuvant treatment in unresected glioblastoma: the GENOM 009 randomized phase II trial.

We sought to determine the impact of bevacizumab on reduction of tumor size prior to chemoradiotherapy in unresected glioblastoma patients. Patients were randomized 1:1 to receive temozolomide (TMZ arm) or temozolomide plus bevacizumab (TMZ + BEV arm). In both arms, neoadjuvant treatment was temozolomide (85 mg/m(2), days 1-21, two 28-day cycles), concurrent radiation plus temozolomide, and six cycles of adjuvant temozolomide. In the TMZ + BEV arm, bevacizumab (10 mg/kg) was added on days 1 and 15 of each neoadjuvant cycle and on days 1, 15 and 30 of concurrent treatment. The primary endpoint was investigator-assessed response to neoadjuvant treatment. Secondary endpoints included progression-free survival (PFS), overall survival (OS), and the impact on outcome of MGMT methylation in tumor and serum. One hundred and two patients were included; 43 in the TMZ arm and 44 in the TMZ + BEV arm were evaluable for response. Results favored the TMZ + BEV arm in terms of objective response (3 [6.7 %] vs. 11 [22.9 %]; odds ratio 4.2; P = 0.04). PFS and OS were longer in the TMZ + BEV arm, though the difference did not reach statistical significance. MGMT methylation in tumor, but not in serum, was associated with outcome. More patients experienced toxicities in the TMZ + BEV than in the TMZ arm (P = 0.06). The combination of bevacizumab plus temozolomide is more active than temozolomide alone and may well confer benefit in terms of tumor shrinkage in unresected patients albeit at the expense of greater toxicity.

Sunitinib administered prior to radiotherapy in patients with non-resectable glioblastoma: results of a phase II study.

Sunitinib is a tyrosine kinase inhibitor with direct anti-tumor and anti-angiogenesis activity targeting VEGFR 1-2, PDGFR α-ß, c-kit, bFGF, (CSF-1), FLT3 and RET. The present trial examined the activity of sunitinib in 12 patients with newly diagnosed, non-resectable glioblastoma. Patients (≤75 years of age with performance status [PS] ≥2 and minimental status [MMS] ≥25) were treated post-biopsy with sunitinib 37.5 mg daily for 8 weeks pre-radiotherapy, during radiotherapy (60 Gy, 6 weeks) and post-radiotherapy until disease progression. The primary endpoints were overall response rate (ORR; RANO criteria) after 8 weeks of sunitinib and patient tolerance. Secondary endpoints were percentage of patients free of neurological deterioration pre-radiotherapy, percentage of patients completing radiotherapy, progression-free survival (PFS), overall survival (OS), and 1-year survival. A Simon 2-stage design (12 →20) based on ORR was applied to calculate the number of patients needed to detect at least 10 % response with α error of 0.05 and ß error of 0.10. The trial was closed because it did not meet minimal activity criteria. ORR was 0 % with only 1/12 patients (8.3 %) achieving stable disease after sunitinib treatment. No patient showed reduction in gadolinium enhancement. The most frequent G3/4 toxicities were fatigue (24.9 %) and diarrhea (16.6 %); one patient died of a CNS hemorrhage; 10/12 patients (83.3 %) deteriorated neurologically before radiation therapy; median PFS was 7.7 weeks (95 % CI: 7.2-8.2); median OS was 12.8 weeks (95 % CI: 0.5-23.8 weeks); 1-year survival was 0 %. Sunitinib has no activity as monotherapy in glioblastoma, and further investigation of its efficacy in this setting is unwarranted.

Compatibility between 3T 1H SV-MRS data and automatic brain tumour diagnosis support systems based on databases of 1.5T 1H SV-MRS spectra.

OBJECT: This study demonstrates that 3T SV-MRS data can be used with the currently available automatic brain tumour diagnostic classifiers which were trained on databases of 1.5T spectra. This will allow the existing large databases of 1.5T MRS data to be used for diagnostic classification of 3T spectra, and perhaps also the combination of 1.5T and 3T databases. MATERIALS AND METHODS: Brain tumour classifiers trained with 154 1.5T spectra to discriminate among high grade malignant tumours and common grade II glial tumours were evaluated with a subsequently-acquired set of 155 1.5T and 37 3T spectra. A similarity study between spectra and main brain tumour metabolite ratios for both field strengths (1.5T and 3T) was also performed. RESULTS: Our results showed that classifiers trained with 1.5T samples had similar accuracy for both test datasets (0.87 ± 0.03 for 1.5T and 0.88 ± 0.03 for 3.0T). Moreover, non-significant differences were observed with most metabolite ratios and spectral patterns. CONCLUSION: These results encourage the use of existing classifiers based on 1.5T datasets for diagnosis with 3T (1)H SV-MRS. The large 1.5T databases compiled throughout many years and the prediction models based on 1.5T acquisitions can therefore continue to be used with data from the new 3T instruments.