Current Classification Schemes Research Articles

Automated Prognostic Assessment of Endometrial Hyperplasia for Progression Risk Evaluation Using Artificial Intelligence

Endometrial hyperplasia is a precursor to endometrial cancer, characterized by excessive proliferation of glands that is distinguishable from normal endometrium. Current classifications define 2 types of EH, each with a different risk of progression to endometrial cancer. However, these schemes are based on visual assessments and, therefore, subjective, possibly leading to overtreatment or undertreatment. In this study, we developed an automated artificial intelligence tool (ENDOAPP) for the measurement of morphologic and cytologic features of endometrial tissue using the software Visiopharm. The ENDOAPP was used to extract features from whole-slide images of PAN-CK+–stained formalin-fixed paraffin-embedded tissue sections from 388 patients diagnosed with endometrial hyperplasia between 1980 and 2007. Follow-up data were available for all patients (mean = 140 months). The most prognostic features were identified by a logistic regression model and used to assign a low-risk or high-risk progression score. Performance of the ENDOAPP was assessed for the following variables: images from 2 different scanners (Hamamatsu XR and S60) and automated placement of a region of interest versus manual placement by an operator. Then, the performance of the application was compared with that of current classification schemes: WHO94, WHO20, and EIN, and the computerized-morphometric risk classification method: D-score. The most significant prognosticators were percentage stroma and the standard deviation of the lesser diameter of epithelial nuclei. The ENDOAPP had an acceptable discriminative power with an area under the curve of 0.765. Furthermore, strong to moderate agreement was observed between manual operators (intraclass correlation coefficient: 0.828) and scanners (intraclass correlation coefficient: 0.791). Comparison of the prognostic capability of each classification scheme revealed that the ENDOAPP had the highest accuracy of 88%-91% alongside the D-score method (91%). The other classification schemes had an accuracy between 83% and 87%. This study demonstrated the use of computer-aided prognosis to classify progression risk in EH for improved patient treatment.

Observation of spin-momentum locked surface states in amorphous Bi2Se3.

Crystalline symmetries have played a central role in the identification and understanding of quantum materials. Here we investigate whether an amorphous analogue of a well known three-dimensional strong topological insulator has topological properties in the solid state. We show that amorphous Bi2Se3 thin films host a number of two-dimensional surface conduction channels. Our angle-resolved photoemission spectroscopy data are consistent with a dispersive two-dimensional surface state that crosses the bulk gap. Spin-resolved photoemission spectroscopy shows this state has an anti-symmetric spin texture, confirming the existence of spin-momentum locked surface states. We discuss these experimental results in light of theoretical photoemission spectra obtained with an amorphous topological insulator tight-binding model, contrasting it with alternative explanations. The discovery of spin-momentum locked surface states in amorphous materials opens a new avenue to characterize amorphous matter, and triggers the search for an overlooked subset of quantum materials outside of current classification schemes.

Choroidal morphologic and vascular features in patients with unilateral idiopathic epiretinal membranes: An optical coherence tomography analysis integrated with assessment of retinal layers.

Integrated analysis of retinal and choroidal morphologic and vascular features is urgently needed to examine whether and how these two elements interact with each other, thus contributing to visual impairment in patients with idiopathic epiretinal membranes (iERMs). An observational retrospective study consisting of 181 patients diagnosed with unilateral iERM between August 2019 and July 2022 was carried out at Peking University Third Hospital. All patients underwent a standardized set of ophthalmologic examinations, including EDI-OCT and OCTA scanning, and were subsequently categorized into four stages according to current classification schemes based on their OCT findings. Altogether, 15 qualitative and quantitative parameters of both the retina (full-layer, inner and outer layers) and choroid were identified. The results revealed variations in the choroidal vascularity index (CVI) among different stages of iERMs (p < 0.001) for the first time. Distributions of retinal parameters across four stages of iERMs were validated. Correlation analysis between choroidal and retinal parameters showed that the CVI was associated with both inner and outer retinal morphologic biomarkers. Functional damage to retinal integrity was determined to be a strong contributor to visual acuity reduction in iERMs. This study complemented our present understanding of posterior segment structural and vascular alterations in iERMs.

CNSC-05. ELECTROPHYSIOLOGICAL PATTERNS OF GLIOMA-INDUCED NEURONAL NETWORK REMODELING ARE CONSERVED ACROSS TUMOR SUBTYPE

Abstract Recent evidence indicates that diffuse gliomas engage with neurons at the single-unit and circuit level through differing mechanisms. Certain malignant gliomas form glioma-neuron excitatory glutamatergic synapses and modulate neuron-neuron synapses through activity-dependent paracrine signaling, while others establish glioma-glioma connections via tumor microtubes. It is therefore possible that diffuse gliomas remodel neuronal circuits in a defined and predictable manner and demonstrate distinct electrophysiological profiles with prognostic and therapeutic significance. Here we apply machine learning principles in 140 patients across glioma subtypes to uncover unique electrophysiological features non-invasively via magnetoencephalography (discovery dataset) followed by feature validation using subdural electrocorticography (validation dataset). Following spatial-temporal registration, we fit an elastic net logistic regression classifier to distinguish between power spectra arising from glioma-remodeled cortex and within-subject control conditions. Model significance was determined non-parametrically by re-training each model 1,000 times with randomly permuted class labels and testing the true phi coefficient against the null distribution. In the discovery dataset, we were able to classify glioma infiltration based on tumor intrinsic neuronal activity (p &amp;lt; 0.05) in 127 patients (90.7%). We identified 30 electrophysiological features which revealed increased power in the delta range (1-4 Hz) and decreased power in the beta range (12-20 Hz) as a unique signature of glioma remodeling (p &amp;lt; 0.05) which was preserved in the validation dataset as well as across WHO 2021 diffuse glioma subtypes. In order to identify gene expression programs and signaling mechanisms that may contribute to glioma-induced remodeling but are potentially not identified in the current clinical classification scheme, we assessed targeted, next generation sequencing and DNA mutations as covariates, which again demonstrated the significance of the delta-beta spectral features. These data support converging mechanisms of glioma-induced neuronal network remodeling across tumor subtypes, setting the stage for novel therapies such as neuromodulation.

Associations between fully-automated, 3D-based functional analysis of the left atrium and classification schemes in atrial fibrillation.

BackgroundAtrial fibrillation (AF) has been linked to left atrial (LA) enlargement. Whereas most studies focused on 2D-based estimation of static LA volume (LAV), we used a fully-automatic convolutional neural network (CNN) for time-resolved (CINE) volumetry of the whole LA on cardiac MRI (cMRI). Aim was to investigate associations between functional parameters from fully-automated, 3D-based analysis of the LA and current classification schemes in AF.MethodsWe retrospectively analyzed consecutive AF patients who underwent cMRI on 1.5T systems including a stack of oblique-axial CINE series covering the whole LA. The LA was automatically segmented by a validated CNN. In the resulting volume-time curves, maximum, minimum and LAV before atrial contraction were automatically identified. Active, passive and total LA emptying fractions (LAEF) were calculated and compared to clinical classifications (AF Burden score (AFBS), increased stroke risk (CHA2DS2VASc≥2), AF type (paroxysmal/persistent), EHRA score, and AF risk factors). Moreover, multivariable linear regression models (mLRM) were used to identify associations with AF risk factors.ResultsOverall, 102 patients (age 61±9 years, 17% female) were analyzed. Active LAEF (LAEF_active) decreased significantly with an increase of AFBS (minimal: 44.0%, mild: 36.2%, moderate: 31.7%, severe: 20.8%, p<0.003) which was primarily caused by an increase of minimum LAV. Likewise, LAEF_active was lower in patients with increased stroke risk (30.7% vs. 38.9%, p = 0.002). AF type and EHRA score did not show significant differences between groups. In mLRM, a decrease of LAEF_active was associated with higher age (per year: -0.3%, p = 0.02), higher AFBS (per category: -4.2%, p<0.03) and heart failure (-12.1%, p<0.04).ConclusionsFully-automatic morphometry of the whole LA derived from cMRI showed significant relationships between LAEF_active with increased stroke risk and severity of AFBS. Furthermore, higher age, higher AFBS and presence of heart failure were independent predictors of reduced LAEF_active, indicating its potential usefulness as an imaging biomarker.

GRB 201104A: A “Repetitive” Short Gamma-Ray Burst?

Gamma-ray bursts (GRBs) are divided into short gamma-ray bursts (SGRBs) and long gamma-ray bursts (LGRBs) based on the bimodal distribution of their durations. LGRBs and SGRBs are typically characterized by different statistical characteristics. Nevertheless, there are some samples that challenge such a framework, such as GRB 060614, a long-duration burst with short-burst characteristics. Furthermore, GRBs are generally considered to be an event with no periodic or repetitive behavior, since the progenitors usually undergo destructive events, such as massive explosions or binary compact star mergers. In this work, we investigated Fermi data for possible quasiperiodic oscillations and repetitive behaviors of GRBs using timing analysis methods and report a special event GRB 201104A, which is a long-duration burst with the characteristics of an SGRB, and it exhibits a “repetitive” behavior. We propose that such a situation may arise from lensed SGRBs and attempt to verify it by Bayesian inference. In addition, we extend the spectral analysis to Bayesian inference. In spite of the existence of at least two distinct time periods with a nearly identical spectrum, there is no strong evidence that they result from a lensing GRB. Taking the gravitational-lensing scenario out of consideration, a long burst would resemble a short burst in its repetitive behavior, which presents a challenge for the current classification scheme.

Recent evolution of ancient Arctic leech relatives: systematics of Acanthobdellida

AbstractAcanthobdellida gnaw into the sides of salmonid fishes in frigid Arctic lakes and rivers, latching on with fearsome facial hooks. Sister to leeches, they are an ancient lineage with two described species. Unfortunately, Acanthobdellida are rarely collected, leading to a paucity of literature despite their unique morphology. Populations range from Eurasia to Alaska (USA), but few specimens of Acanthobdella peledina are represented in molecular studies, and no molecular data exist for Paracanthobdella livanowi, making their taxonomic position difficult to assess. We use phylogenetics and morphology to determine whether allopatric populations of A. peledina are distinct species and assess the current classification scheme used for Acanthobdellida. We produce a new suborder, Acanthobdelliformes, to match the taxonomy within Hirudinea. Scanning electron micrographs indicate species-level differences in the anterior sucker and facial hooks; molecular phylogenetics mirrors this divergence between species. We assign both species to the family Acanthobdellidae and abandon the family Paracanthobdellidae. Alaskan and European A. peledina populations are morphologically similar, but appear phylogenetically divergent. Our data strongly suggest that members of the order Acanthobdellida diverged relatively recently in their ancient history, but based on genetic distance, this divergence appears to pre-date the most recent cycles of glaciation.

Mitochondrial genomic data are effective at placing mycoheterotrophic lineages in plant phylogeny.

Fully mycoheterotrophic plants can be difficult to place in plant phylogeny due to elevated substitution rates associated with photosynthesis loss. This potentially limits the effectiveness of downstream analyses of mycoheterotrophy that depend on accurate phylogenetic inference. Although mitochondrial genomic data sets are rarely used in plant phylogenetics, theory predicts that they should be resilient to long-branch artefacts, thanks to their generally slow evolution, coupled with limited rate elevation in heterotrophs. We examined the utility of mitochondrial genomes for resolving contentious higher-order placements of mycoheterotrophic lineages in two test cases: monocots (focusing on Dioscoreales) and Ericaceae. We find Thismiaceae to be distantly related to Burmanniaceae in the monocot order Dioscoreales, conflicting with current classification schemes based on few gene data sets. We confirm that the unusual Afrothismia is related to Taccaceae-Thismiaceae, with a corresponding independent loss of photosynthesis. In Ericaceae we recovered the first well supported relationships among its five major lineages: mycoheterotrophic Ericaceae are not monophyletic, as pyroloids are inferred to be sister to core Ericaceae, and monotropoids to arbutoids. Genes recovered from mitochondrial genomes collectively resolved previously ambiguous mycoheterotroph higher-order relationships. We propose that mitochondrial genomic data should be considered in standardised gene panels for inferring overall plant phylogeny.

Incomplete lineage sorting and reticulate evolution mask species relationships in Brunelliaceae, an Andean family with rapid, recent diversification.

To date, phylogenetic relationships within the monogeneric Brunelliaceae have been based on morphological evidence, which does not provide sufficient phylogenetic resolution. Here we use target-enriched nuclear data to improve our understanding of phylogenetic relationships in the family. We used the Angiosperms353 toolkit for targeted recovery of exonic regions and supercontigs (exons + introns) from low copy nuclear genes from 53 of 70 species in Brunellia, and several outgroup taxa. We removed loci that indicated biased inference of relationships and applied concatenated and coalescent methods to infer Brunellia phylogeny. We identified conflicts among gene trees that may reflect hybridization or incomplete lineage sorting events and assessed their impact on phylogenetic inference. Finally, we performed ancestral-state reconstructions of morphological traits and assessed the homology of character states used to define sections and subsections in Brunellia. Brunellia comprises two major clades and several subclades. Most of these clades/subclades do not correspond to previous infrageneric taxa. There is high topological incongruence among the subclades across analyses. Phylogenetic reconstructions point to rapid species diversification in Brunelliaceae, reflected in very short branches between successive species splits. The removal of putatively biased loci slightly improves phylogenetic support for individual clades. Reticulate evolution due to hybridization and/or incomplete lineage sorting likely both contribute to gene-tree discordance. Morphological characters used to define taxa in current classification schemes are homoplastic in the ancestral character-state reconstructions. While target enrichment data allows us to broaden our understanding of diversification in Brunellia, the relationships among subclades remain incompletely understood.

FedHGCDroid: An Adaptive Multi-Dimensional Federated Learning for Privacy-Preserving Android Malware Classification.

With the popularity of Android and its open source, the Android platform has become an attractive target for hackers, and the detection and classification of malware has become a research hotspot. Existing malware classification methods rely on complex manual operation or large-volume high-quality training data. However, malware data collected by security providers contains user privacy information, such as user identity and behavior habit information. The increasing concern for user privacy poses a challenge to the current malware classification scheme. Based on this problem, we propose a new android malware classification scheme based on Federated learning, named FedHGCDroid, which classifies malware on Android clients in a privacy-protected manner. Firstly, we use a convolutional neural network and graph neural network to design a novel multi-dimensional malware classification model HGCDroid, which can effectively extract malicious behavior features to classify the malware accurately. Secondly, we introduce an FL framework to enable distributed Android clients to collaboratively train a comprehensive Android malware classification model in a privacy-preserving way. Finally, to adapt to the non-IID distribution of malware on Android clients, we propose a contribution degree-based adaptive classifier training mechanism FedAdapt to improve the adaptability of the malware classifier based on Federated learning. Comprehensive experimental studies on the Androzoo dataset (under different non-IID data settings) show that the FedHGCDroid achieves more adaptability and higher accuracy than the other state-of-the-art methods.

Classification of acute diverticulitis in the left colon with ultrasonography and computed tomography: Value of current severity classification schemes

Background and aimsThe current management of acute diverticulitis of the left colon (ADLC) requires tests with high prognostic value. This paper analyzes the usefulness of ultrasonography (US) in the initial diagnosis of ADLC and the validity of current classifications schemes for ADLC. PatientsThis retrospective observational study included patients with ADLC scheduled to undergo US or computed tomography (CT) following a clinical algorithm. According to the imaging findings, ADLC was classified as mild, locally complicated, or complicated. We analyzed the efficacy of US in the initial diagnosis and the reasons why CT was used as the first-line technique. We compared the findings with published classifications schemes for ADLC. ResultsA total of 311 patients were diagnosed with acute diverticulitis; 183 had ADLC, classified at imaging as mild in 104, locally complicated in 60, and complicated in 19. The diagnosis was reached by US alone in 98 patients, by CT alone in 77, and by combined US and CT in 8. The main reasons for using CT as the first-line technique were the radiologist’s lack of experience in abdominal US and the unavailability of a radiologists on call. Six patients diagnosed by US were reexamined by CT, but the classification changed in only three. None of the published classification schemes included all the imaging findings. ConclusionsUS should be the first-line imaging technique in patients with suspected ADLC. Various laboratory and imaging findings are useful in establishing the prognosis of ADLC. New schemes to classify the severity of ADLC are necessary to ensure optimal clinical decision making.

Foveal Triangular Fibrocartilage Complex Tears: Recognition of a Combined Tear Pattern

Triangular fibrocartilage complex (TFCC) injuries are categorized most commonly by the Palmer and the Atzei and Luchetti classifications. Combined tears are reported less frequently, and the classification of these tears is unclear. In this study, we present a series of patients with combined central disc and foveal TFCC tears that do not fit into either of these classifications. Seventeen patients from 2 institutions presented with a combined central and foveal TFCC tear, confirmed by wrist arthroscopy between September 2017 and December 2020. Patient demographics, injury mechanism, associated injuries, clinical findings, magnetic resonance imaging results, surgeon description of tears, and treatment rendered were evaluated. Patients were predominantly female (76%). Mean age was 40 years with a bimodal age distribution (7/17 patients <35 and 10/17 >45 years). Eleven patients (65%) sustained the injury from a fall. All had some degree of distal radioulnar joint (DRUJ) instability on clinical examination and almost half (8/17) had a history of a distal radius or ulnar styloid fracture during the initial injury. The foveal tear was confirmed by direct visual assessment and a positive hook test result; a suction test result commonly was positive as well. All patients underwent debridement of the central tear, and 16 of 17 underwent foveal TFCC repair with 1 having an irreparable tear. We present a series of patients with combined central disc and foveal TFCC tears, a pattern that does not fit current classification schemes. All patients exhibited some degree of instability of the DRUJ and almost half had a history of distal radius or ulnar styloid fracture. Identification of a central tear of the TFCC, together with even subtle DRUJ instability, should lead to consideration of a coexistent foveal tear. Diagnostic IV.

Prevalence of Crimean-Congo Hemorrhagic Fever Virus among Livestock and Ticks in Zhambyl Region, Kazakhstan, 2017.

Crimean-Congo hemorrhagic fever (CCHF) is a highly fatal zoonotic disease endemic to Kazakhstan. Previous work estimated the seroprevalence of CCHF virus (CCHFV) among livestock owners in the Zhambyl region of southern Kazakhstan at 1.2%. To estimate CCHFV seroprevalence among cattle and sheep, we selected 15 villages with known history of CCHFV circulation (endemic) and 15 villages without known circulation (nonendemic) by cluster sampling with probability proportional to livestock population size. We collected whole blood samples from 521 sheep and 454 cattle from randomly selected households within each village and collected ticks found on the animals. We tested livestock blood for CCHFV-specific IgG antibodies by ELISA; ticks were screened for CCHFV RNA by real-time reverse transcription polymerase chain reaction and CCHFV antigen by antigen-capture ELISA. We administered questionnaires covering animal demographics and livestock herd characteristics to an adult in each selected household. Overall weighted seroprevalence was 5.7% (95% CI: 3.1, 10.3) among sheep and 22.5% (95% CI: 15.8, 31.2) among cattle. CCHFV-positive tick pools were found on two sheep (2.4%, 95% CI: 0.6, 9.5) and three cattle (3.8%, 95% CI: 1.2, 11.5); three CCHFV-positive tick pools were found in nonendemic villages. Endemic villages reported higher seroprevalence among sheep (15.5% versus 2.8%, P < 0.001) but not cattle (25.9% versus 20.1%, P = 0.42). Findings suggest that the current village classification scheme may not reflect the geographic distribution of CCHFV in Zhambyl and underscore that public health measures must address the risk of CCHF even in areas without a known history of circulation.

Characterization of Supernovae Based on the Spectral–Temporal Energy Distribution: Two Possible SN Ib Subtypes

A quantitative data-driven comparison among supernovae (SNe) based on their spectral time series combined with multiband photometry is presented. We use an unsupervised random forest algorithm as a metric on a set of 82 well-documented SNe representing all the main spectroscopic types, in order to embed these in an abstract metric space reflecting shared correlations between the objects. We visualize the resulting metric space in 3D, revealing strong agreement with the current spectroscopic classification scheme. The embedding splits Type Ib supernovae into two groups, with one subgroup exhibiting broader, less prominent, higher-velocity lines than the other, possibly suggesting a new SN Ib subclass is required. The method could be to classify newly discovered SNe according to their distance from known event groups, or ultimately to devise a new, spectral–temporal classification scheme. Such an embedding could also depend on hidden parameters that may perhaps be physically interpretable.

The Meaning of Words Related to Orofacial Pain and Headache Conditions: The Need for a Single and Unified Classification Scheme in a People-Centered Language

The Meaning of Words Related to Orofacial Pain and Headache Conditions: The Need for a Single and Unified Classification Scheme in a People-Centered Language

Grasses through space and time: An overview of the biogeographical and macroevolutionary history of Poaceae

AbstractGrasses are widespread on every continent and are found in all terrestrial biomes. The dominance and spread of grasses and grassland ecosystems have led to significant changes in Earth′s climate, geochemistry, and biodiversity. The abundance of DNA sequence data, particularly chloroplast sequences, and advances in placing grass fossils within the family allows for a reappraisal of the family′s origins, timing, and geographic spread and the factors that have promoted diversification. We reconstructed a time‐calibrated grass phylogeny and inferred ancestral areas using chloroplast DNA sequences from nearly 90% of extant grass genera. With a few notable exceptions, the phylogeny is well resolved to the subtribal level. The family began to diversify in the Early–Late Cretaceous (crown age of 98.54 Ma) on West Gondwana before the complete split between Africa and South America. Vicariance from the splitting of Gondwana may be responsible for the initial divergence in the family. However, Africa clearly served as the center of origin for much of the early diversification of the family. With this phylogenetic, temporal, and spatial framework, we review the evolution and biogeography of the family with the aim to facilitate the testing of biogeographical hypotheses about its origins, evolutionary tempo, and diversification. The current classification of the family is discussed with an extensive review of the extant diversity and distribution of species, molecular and morphological evidence supporting the current classification scheme, and the evidence informing our understanding of the biogeographical history of the family.

Precision Medicine in Inflammatory Bowel Diseases: Challenges and Considerations for the Path Forward

Precision Medicine in Inflammatory Bowel Diseases: Challenges and Considerations for the Path Forward

AI-Based Pipeline for Classifying Pediatric Medulloblastoma Using Histopathological and Textural Images.

Pediatric medulloblastomas (MBs) are the most common type of malignant brain tumors in children. They are among the most aggressive types of tumors due to their potential for metastasis. Although this disease was initially considered a single disease, pediatric MBs can be considerably heterogeneous. Current MB classification schemes are heavily reliant on histopathology. However, the classification of MB from histopathological images is a manual process that is expensive, time-consuming, and prone to error. Previous studies have classified MB subtypes using a single feature extraction method that was based on either deep learning or textural analysis. Here, we combine textural analysis with deep learning techniques to improve subtype identification using histopathological images from two medical centers. Three state-of-the-art deep learning models were trained with textural images created from two texture analysis methods in addition to the original histopathological images, enabling the proposed pipeline to benefit from both the spatial and textural information of the images. Using a relatively small number of features, we show that our automated pipeline can yield an increase in the accuracy of classification of pediatric MB compared with previously reported methods. A refined classification of pediatric MB subgroups may provide a powerful tool for individualized therapies and identification of children with increased risk of complications.

Forthcoming Issues

Recent advances in molecular diagnostics have led to better understanding of glioma tumorigenesis and biology. Numerous glioma biomarkers with diagnostic, prognostic, and predictive value have been identified. Although some of these markers are already part of the routine clinical management of glioma patients, data regarding others are limited and difficult to apply routinely. In addition, multiple methods for molecular subclassification have been proposed either together with or as an alternative to the current morphologic classification and grading scheme. This article reviews the literature regarding glioma biomarkers and offers a few practical suggestions.

Risk stratification of gastrointestinal stromal tumors by Nanostring gene expression profiling.

The risk assessment classification schemes for gastrointestinal stromal tumors (GIST) include tumor site, size, mitotic count and variably tumor rupture. Heterogeneity in high-risk GIST poses limitations for current classification schemes. This study aims to demonstrate the clinical utility of risk stratification by gene expression profiling (GEP) using Nanostring technology. Fifty-six GIST cases were analyzed using a 231 gene expression panel. GEP results were correlated with clinical and pathological data. The prognostic performance was assessed in 34 patients with available survival data using ROC curves, Kaplan-Meier survival curves and compared with traditional risk assessment schemes. Volcano plot analysis identified seven genes with significantly higher expression (FDR < .0.05) in high-risk than in non-high-risk tumors, namely TYMS, CDC2, TOP2A, CCNA2, E2F1, PCNA, and BIRC5. Together, these transcripts exhibited significantly higher expression in high-risk tumors than in intermediate (P < 0.01), low (P < 0.001), and very low (P = 0.01) risk tumors. Receiver-operating characteristic curve analysis demonstrated area under the curve (AUC) to be 0.858 for the separation of high-risk and non-high-risk tumors. Kaplan-Meier survival analysis demonstrated improved risk stratification (log-rank test P < 0.001) compared to the current risk assessment classification (P = 0.231). In addition to current clinical and histology-based risk classification for patients with GIST, gene expression may offer complementary prognostic information.