Artificial intelligence-based analysis of retinal fluid volume dynamics in neovascular age-related macular degeneration and association with vision and atrophy.

BACKGROUND/OBJECTIVES: To characterise morphological changes in neovascular age-related macular degeneration (nAMD) during anti-angiogenic therapy and explore relationships with best-corrected visual acuity (BCVA) and development of macular atrophy (MA). SUBJECTS/METHODS: Post-hoc analysis of the phase III HARBOR trial. SD-OCT scans from 1097 treatment-naïve nAMD eyes were analysed. Volumes of intraretinal cystoid fluid (ICF), subretinal hyperreflective material (SHRM), subretinal fluid (SRF), pigment epithelial detachment (PED) and cyst-free retinal volume (CFRV) were measured by deep-learning model. Volumes were analysed by treatment regimen, macular neovascularisation (MNV) subtypes and topographic location. Associations of volumetric features with BCVA and MA development were quantified at month 12/24. RESULTS: Differences in feature volume changes by treatment regimens and MNV subtypes were observed. Each additional 100 nanolitre unit (AHNU) of residual ICF, SHRM and CFRV at month 1 in the fovea was associated with deficits of 10.3, 7.3 and 12.2 letters at month 12. Baseline AHNUs of ICF, CFRV and PED were associated with increased odds of MA development at month 12 by 10%, 4% and 3%. While that of SRF was associated with a decrease in odds of 5%. Associations at month 24 were similar to those at month 12. CONCLUSION: Eyes with different MNV subtypes showed distinct trajectories of feature volume response to treatment. Higher baseline volumes of ICF or PED and lower baseline volume of SRF were associated with higher likelihoods of MA development over 24 months. Residual intraretinal fluid, including ICF and CFRV, along with SHRM were predictors of poor visual outcomes.

Intraretinal Hyper-Reflective Foci Are Almost Universally Present and Co-Localize With Intraretinal Fluid in Diabetic Macular Edema.

Purpose: In diabetic macular edema (DME), hyper-reflective foci (HRF) has been linked to disease severity and progression. Using an automated approach, we aimed to investigate the baseline distribution of HRF in DME and their co-localization with cystoid intraretinal fluid (IRF). Methods: Baseline spectral-domain optical coherence tomography (SD-OCT) volume scans (N = 1527) from phase III clinical trials YOSEMITE (NCT03622580) and RHINE (NCT03622593) were segmented using a deep-learning-based algorithm (developed using B-scans from BOULEVARD NCT02699450) to detect HRF. The HRF count and volume were assessed. HRF distributions were analyzed in relation to best-corrected visual acuity (BCVA), central subfield thickness (CST), and IRF volume in quartiles, and Diabetic Retinopathy Severity Scores (DRSS) in groups. Co-localization of HRF with IRF was calculated in the central 3-mm diameter using the en face projection. Results: HRF were present in most patients (up to 99.7%). Median (interquartile range [IQR]) HRF volume within the 3-mm diameter Early Treatment Diabetic Retinopathy Study ring was 1964.3 (3325.2) pL, and median count was 64.0 (IQR = 96.0). Median HRF volumes were greater with decreasing BCVA (nominal P = 0.0109), and increasing CST (nominal P < 0.0001), IRF (nominal P < 0.0001), and DRSS up to very severe nonproliferative diabetic retinopathy (nominal P < 0.0001). HRF co-localized with IRF in the en face projection. Conclusions: Using automated HRF segmentation of full SD-OCT volumes, we observed that HRF are a ubiquitous feature in DME and exhibit relationships with BCVA, CST, IRF, and DRSS, supporting a potential link to disease severity. The spatial distribution of HRF closely followed that of IRF.

Machine Learning to Predict Response to Ranibizumab in Neovascular Age-Related Macular Degeneration.

Purpose: Neovascular age-related macular degeneration (nAMD) shows variable treatment response to intravitreal anti-VEGF. This analysis compared the potential of different artificial intelligence (AI)-based machine learning models using OCT and clinical variables to accurately predict at baseline the best-corrected visual acuity (BCVA) at 9 months in response to ranibizumab in patients with nAMD. Design: Retrospective analysis. Participants: Baseline and imaging data from patients with subfoveal choroidal neovascularization secondary to age-related macular dengeration. Methods: Baseline data from 502 study eyes from the HARBOR (NCT00891735) prospective clinical trial (monthly ranibizumab 0.5 and 2.0 mg arms) were pooled; 432 baseline OCT volume scans were included in the analysis. Seven models, based on baseline quantitative OCT features (Least absolute shrinkage and selection operator [Lasso] OCT minimum [min], Lasso OCT 1 standard error [SE]); on quantitative OCT features and clinical variables at baseline (Lasso min, Lasso 1SE, CatBoost, RF [random forest]); or on baseline OCT images only (deep learning [DL] model), were systematically compared with a benchmark linear model of baseline age and BCVA. Quantitative OCT features were derived by a DL segmentation model on the volume images, including retinal layer volumes and thicknesses, and retinal fluid biomarkers, including statistics on fluid volume and distribution. Main Outcome Measures: Prognostic ability of the models was evaluated using coefficient of determination (R2) and median absolute error (MAE; letters). Results: In the first cross-validation split, mean R2 (MAE) of the Lasso min, Lasso 1SE, CatBoost, and RF models was 0.46 (7.87), 0.42 (8.43), 0.45 (7.75), and 0.43 (7.60), respectively. These models ranked higher than or similar to the benchmark model (mean R2, 0.41; mean MAE, 8.20 letters) and better than OCT-only models (mean R2: Lasso OCT min, 0.20; Lasso OCT 1SE, 0.16; DL, 0.34). The Lasso min model was selected for detailed analysis; mean R2 (MAE) of the Lasso min and benchmark models for 1000 repeated cross-validation splits were 0.46 (7.7) and 0.42 (8.0), respectively. Conclusions: Machine learning models based on AI-segmented OCT features and clinical variables at baseline may predict future response to ranibizumab treatment in patients with nAMD. However, further developments will be needed to realize the clinical utility of such AI-based tools. Financial Disclosures: Proprietary or commercial disclosure may be found after the references.

Anti-VEGF-resistant subretinal fluid is associated with better vision and reduced risk of macular atrophy.

BACKGROUND/AIM: To evaluate relationships between subretinal fluid (SRF), macular atrophy (MA) and visual outcomes in ranibizumab-treated neovascular age-related macular degeneration (nAMD). METHODS: This post hoc HARBOR trial (NCT00891735) analysis included ranibizumab-treated (0.5 or 2.0 mg, monthly or as-needed, all treatment arms pooled) eyes with nAMD and baseline (screening, baseline and week 1) SRF. SRF presence, SRF thickness (0, >0-50, >50-100 and >100 µm) and subretinal fluid volume (SRFV) were determined by spectral domain optical coherence tomography (SD-OCT). Best-corrected visual acuity (BCVA) was assessed. MA was identified using fluorescein angiograms and colour fundus photographs, as well as SD-OCT. RESULTS: Seven hundred eighty-five of 1097 eyes met analysis criteria. In eyes without baseline MA, residual versus no SRF at month (M) 3 was associated with lower MA rates at M12 (5.1% vs 22.1%) and M24 (13.3% vs 31.2%) (both p<0.0001); MA percentages at M12/M24 were similar among patients with residual SRF at M6. Higher baseline SRFV was associated with a lower MA rate. Greater mean BCVA was observed with residual SRF of any thickness (>0-50 µm, 71.2 letters; >50-100 µm, 71.3 letters; >100 µm, 69.2 letters) versus no SRF (63.6 letters), but the change in BCVA from baseline to M12 or M24 was the same for eyes with or without treatment-resistant subretinal fluid (TR-SRF) at M3 or M6. CONCLUSION: TR-SRF was not detrimental to vision outcomes over 2 years, regardless of thickness. MA rates were significantly higher without TR-SRF.

Accuracy of a Machine-Learning Algorithm for Detecting and Classifying Choroidal Neovascularization on Spectral-Domain Optical Coherence Tomography.

BACKGROUND: To evaluate the performance of a machine-learning (ML) algorithm to detect and classify choroidal neovascularization (CNV), secondary to age-related macular degeneration (AMD) on spectral-domain optical coherence tomography (SD-OCT) images. METHODS: Baseline fluorescein angiography (FA) and SD-OCT images from 1037 treatment-naive study eyes and 531 fellow eyes, without advanced AMD from the phase 3 HARBOR trial (NCT00891735), were used to develop, train, and cross-validate an ML pipeline combining deep-learning-based segmentation of SD-OCT B-scans and CNV classification, based on features derived from the segmentations, in a five-fold setting. FA classification of the CNV phenotypes from HARBOR was used for generating the ground truth for model development. SD-OCT scans from the phase 2 AVENUE trial (NCT02484690) were used to externally validate the ML model. RESULTS: The ML algorithm discriminated CNV absence from CNV presence, with a very high accuracy (area under the receiver operating characteristic [AUROC] = 0.99), and classified occult versus predominantly classic CNV types, per FA assessment, with a high accuracy (AUROC = 0.91) on HARBOR SD-OCT images. Minimally classic CNV was discriminated with significantly lower performance. Occult and predominantly classic CNV types could be discriminated with AUROC = 0.88 on baseline SD-OCT images of 165 study eyes, with CNV from AVENUE. CONCLUSIONS: Our ML model was able to detect CNV presence and CNV subtypes on SD-OCT images with high accuracy in patients with neovascular AMD.

Author Correction: Deep learning algorithm predicts diabetic retinopathy progression in individual patients.

A Correction to this paper has been published: https://doi.org/10.1038/s41746-020-00365-5.

Deep learning algorithm predicts diabetic retinopathy progression in individual patients.

The global burden of diabetic retinopathy (DR) continues to worsen and DR remains a leading cause of vision loss worldwide. Here, we describe an algorithm to predict DR progression by means of deep learning (DL), using as input color fundus photographs (CFPs) acquired at a single visit from a patient with DR. The proposed DL models were designed to predict future DR progression, defined as 2-step worsening on the Early Treatment Diabetic Retinopathy Diabetic Retinopathy Severity Scale, and were trained against DR severity scores assessed after 6, 12, and 24 months from the baseline visit by masked, well-trained, human reading center graders. The performance of one of these models (prediction at month 12) resulted in an area under the curve equal to 0.79. Interestingly, our results highlight the importance of the predictive signal located in the peripheral retinal fields, not routinely collected for DR assessments, and the importance of microvascular abnormalities. Our findings show the feasibility of predicting future DR progression by leveraging CFPs of a patient acquired at a single visit. Upon further development on larger and more diverse datasets, such an algorithm could enable early diagnosis and referral to a retina specialist for more frequent monitoring and even consideration of early intervention. Moreover, it could also improve patient recruitment for clinical trials targeting DR.

Deep Learning Predicts OCT Measures of Diabetic Macular Thickening From Color Fundus Photographs.

Purpose: To develop deep learning (DL) models for the automatic detection of optical coherence tomography (OCT) measures of diabetic macular thickening (MT) from color fundus photographs (CFPs). Methods: Retrospective analysis on 17,997 CFPs and their associated OCT measurements from the phase 3 RIDE/RISE diabetic macular edema (DME) studies. DL with transfer-learning cascade was applied on CFPs to predict time-domain OCT (TD-OCT)-equivalent measures of MT, including central subfield thickness (CST) and central foveal thickness (CFT). MT was defined by using two OCT cutoff points: 250 µm and 400 µm. A DL regression model was developed to directly quantify the actual CFT and CST from CFPs. Results: The best DL model was able to predict CST ≥ 250 µm and CFT ≥ 250 µm with an area under the curve (AUC) of 0.97 (95% confidence interval [CI], 0.89-1.00) and 0.91 (95% CI, 0.76-0.99), respectively. To predict CST ≥ 400 µm and CFT ≥ 400 µm, the best DL model had an AUC of 0.94 (95% CI, 0.82-1.00) and 0.96 (95% CI, 0.88-1.00), respectively. The best deep convolutional neural network regression model to quantify CST and CFT had an R2 of 0.74 (95% CI, 0.49-0.91) and 0.54 (95% CI, 0.20-0.87), respectively. The performance of the DL models declined when the CFPs were of poor quality or contained laser scars. Conclusions: DL is capable of predicting key quantitative TD-OCT measurements related to MT from CFPs. The DL models presented here could enhance the efficiency of DME diagnosis in tele-ophthalmology programs, promoting better visual outcomes. Future research is needed to validate DL algorithms for MT in the real-world.

Machine learning-powered antibiotics phenotypic drug discovery.

Identification of novel antibiotics remains a major challenge for drug discovery. The present study explores use of phenotypic readouts beyond classical antibacterial growth inhibition adopting a combined multiparametric high content screening and genomic approach. Deployment of the semi-automated bacterial phenotypic fingerprint (BPF) profiling platform in conjunction with a machine learning-powered dataset analysis, effectively allowed us to narrow down, compare and predict compound mode of action (MoA). The method identifies weak antibacterial hits allowing full exploitation of low potency hits frequently discovered by routine antibacterial screening. We demonstrate that BPF classification tool can be successfully used to guide chemical structure activity relationship optimization, enabling antibiotic development and that this approach can be fruitfully applied across species. The BPF classification tool could be potentially applied in primary screening, effectively enabling identification of novel antibacterial compound hits and differentiating their MoA, hence widening the known antibacterial chemical space of existing pharmaceutical compound libraries. More generally, beyond the specific objective of the present work, the proposed approach could be profitably applied to a broader range of diseases amenable to phenotypic drug discovery.

Inhibition of EGF Uptake by Nephrotoxic Antisense Drugs In Vitro and Implications for Preclinical Safety Profiling.

Antisense oligonucleotide (AON) therapeutics offer new avenues to pursue clinically relevant targets inaccessible with other technologies. Advances in improving AON affinity and stability by incorporation of high affinity nucleotides, such as locked nucleic acids (LNA), have sometimes been stifled by safety liabilities related to their accumulation in the kidney tubule. In an attempt to predict and understand the mechanisms of LNA-AON-induced renal tubular toxicity, we established human cell models that recapitulate in vivo behavior of pre-clinically and clinically unfavorable LNA-AON drug candidates. We identified elevation of extracellular epidermal growth factor (EGF) as a robust and sensitive in vitro biomarker of LNA-AON-induced cytotoxicity in human kidney tubule epithelial cells. We report the time-dependent negative regulation of EGF uptake and EGF receptor (EGFR) signaling by toxic but not innocuous LNA-AONs and revealed the importance of EGFR signaling in LNA-AON-mediated decrease in cellular activity. The robust EGF-based in vitro safety profiling of LNA-AON drug candidates presented here, together with a better understanding of the underlying molecular mechanisms, constitutes a significant step toward developing safer antisense therapeutics.

Innovative Strategies to Develop Chemical Categories Using a Combination of Structural and Toxicological Properties.

Interest is increasing in the development of non-animal methods for toxicological evaluations. These methods are however, particularly challenging for complex toxicological endpoints such as repeated dose toxicity. European Legislation, e.g., the European Union's Cosmetic Directive and REACH, demands the use of alternative methods. Frameworks, such as the Read-across Assessment Framework or the Adverse Outcome Pathway Knowledge Base, support the development of these methods. The aim of the project presented in this publication was to develop substance categories for a read-across with complex endpoints of toxicity based on existing databases. The basic conceptual approach was to combine structural similarity with shared mechanisms of action. Substances with similar chemical structure and toxicological profile form candidate categories suitable for read-across. We combined two databases on repeated dose toxicity, RepDose database, and ELINCS database to form a common database for the identification of categories. The resulting database contained physicochemical, structural, and toxicological data, which were refined and curated for cluster analyses. We applied the Predictive Clustering Tree (PCT) approach for clustering chemicals based on structural and on toxicological information to detect groups of chemicals with similar toxic profiles and pathways/mechanisms of toxicity. As many of the experimental toxicity values were not available, this data was imputed by predicting them with a multi-label classification method, prior to clustering. The clustering results were evaluated by assessing chemical and toxicological similarities with the aim of identifying clusters with a concordance between structural information and toxicity profiles/mechanisms. From these chosen clusters, seven were selected for a quantitative read-across, based on a small ratio of NOAEL of the members with the highest and the lowest NOAEL in the cluster (< 5). We discuss the limitations of the approach. Based on this analysis we propose improvements for a follow-up approach, such as incorporation of metabolic information and more detailed mechanistic information. The software enables the user to allocate a substance in a cluster and to use this information for a possible read- across. The clustering tool is provided as a free web service, accessible at http://mlc-reach.informatik.uni-mainz.de.

Automated and reproducible read-across like models for predicting carcinogenic potency.

Several qualitative (hazard-based) models for chronic toxicity prediction are available through commercial and freely available software, but in the context of risk assessment a quantitative value is mandatory in order to be able to apply a Margin of Exposure (predicted toxicity/exposure estimate) approach to interpret the data. Recently quantitative models for the prediction of the carcinogenic potency have been developed, opening some hopes in this area, but this promising approach is currently limited by the fact that the proposed programs are neither publically nor commercially available. In this article we describe how two models (one for mouse and one for rat) for the carcinogenic potency (TD50) prediction have been developed, using lazar (Lazy Structure Activity Relationships), a procedure similar to read-across, but automated and reproducible. The models obtained have been compared with the recently published ones, resulting in a similar performance. Our aim is also to make the models freely available in the near future thought a user friendly internet web site.

lazar: a modular predictive toxicology framework.

lazar (lazy structure-activity relationships) is a modular framework for predictive toxicology. Similar to the read across procedure in toxicological risk assessment, lazar creates local QSAR (quantitative structure-activity relationship) models for each compound to be predicted. Model developers can choose between a large variety of algorithms for descriptor calculation and selection, chemical similarity indices, and model building. This paper presents a high level description of the lazar framework and discusses the performance of example classification and regression models.

Combinatorial QSAR modeling of human intestinal absorption.

Intestinal drug absorption in humans is a central topic in drug discovery. In this study, we use a broad selection of machine learning and statistical methods for the classification and numerical prediction of this key end point. Our data set is based on a selection of 458 small druglike compounds with FDA approval. Using easily available tools, we calculated one- to three-dimensional physicochemical descriptors and used various methods of feature selection (best-first backward selection, correlation analysis, and decision tree analysis). We then used decision tree induction (DTI), fragment-based lazy-learning (LAZAR), support vector machine classification, multilayer perceptrons, random forests, k-nearest neighbor and Naïve Bayes analysis to model absorption ratios and binary classification (well-absorbed and poorly absorbed compounds). Best performance for classification was seen with DTI using the chi-squared analysis interaction detector (CHAID) algorithm, yielding corrected classification rate of 88% (Matthews correlation coefficient of 75%). In numeric predictions, the multilayer perceptron performed best, achieving a root mean squared error of 25.823 and a coefficient of determination of 0.6. In line with current understanding is the importance of descriptors such as lipophilic partition coefficients (log P) and hydrogen bonding. However, we are able to highlight the utility of gravitational indices and moments of inertia, reflecting the role of structural symmetry in oral absorption. Our models are based on a diverse data set of marketed drugs representing a broad chemical space. These models therefore contribute substantially to the molecular understanding of human intestinal drug absorption and qualify for a generalized use in drug discovery and lead optimization.

Collaborative development of predictive toxicology applications.

OpenTox provides an interoperable, standards-based Framework for the support of predictive toxicology data management, algorithms, modelling, validation and reporting. It is relevant to satisfying the chemical safety assessment requirements of the REACH legislation as it supports access to experimental data, (Quantitative) Structure-Activity Relationship models, and toxicological information through an integrating platform that adheres to regulatory requirements and OECD validation principles. Initial research defined the essential components of the Framework including the approach to data access, schema and management, use of controlled vocabularies and ontologies, architecture, web service and communications protocols, and selection and integration of algorithms for predictive modelling. OpenTox provides end-user oriented tools to non-computational specialists, risk assessors, and toxicological experts in addition to Application Programming Interfaces (APIs) for developers of new applications. OpenTox actively supports public standards for data representation, interfaces, vocabularies and ontologies, Open Source approaches to core platform components, and community-based collaboration approaches, so as to progress system interoperability goals.The OpenTox Framework includes APIs and services for compounds, datasets, features, algorithms, models, ontologies, tasks, validation, and reporting which may be combined into multiple applications satisfying a variety of different user needs. OpenTox applications are based on a set of distributed, interoperable OpenTox API-compliant REST web services. The OpenTox approach to ontology allows for efficient mapping of complementary data coming from different datasets into a unifying structure having a shared terminology and representation.Two initial OpenTox applications are presented as an illustration of the potential impact of OpenTox for high-quality and consistent structure-activity relationship modelling of REACH-relevant endpoints: ToxPredict which predicts and reports on toxicities for endpoints for an input chemical structure, and ToxCreate which builds and validates a predictive toxicity model based on an input toxicology dataset. Because of the extensible nature of the standardised Framework design, barriers of interoperability between applications and content are removed, as the user may combine data, models and validation from multiple sources in a dependable and time-effective way.

Development of decision tree models for substrates, inhibitors, and inducers of p-glycoprotein.

In silico classification of new compounds for certain properties is a useful tool to guide further experiments or compound selection. Interaction of new compounds with the efflux pump P-glycoprotein (P-gp) is an important drug property determining tissue distribution and the potential for drug-drug interactions. We present three datasets on substrate, inhibitor, and inducer activities for P-gp (n = 471) obtained from a literature search which we compared to an existing evaluation of the Prestwick Chemical Library with the calcein-AM assay (retrieved from PubMed). Additionally, we present decision tree models of these activities with predictive accuracies of 77.7 % (substrates), 86.9 % (inhibitors), and 90.3 % (inducers) using three algorithms (CHAID, CART, and C4.5). We also present decision tree models of the calcein-AM assay (79.9 %). Apart from a comprehensive dataset of P-gp interacting compounds, our study provides evidence of the efficacy of logD descriptors and of two algorithms not commonly used in pharmacological QSAR studies (CART and CHAID).