Prediction of pathologic complete response to neoadjuvant chemotherapy using machine learning models in patients with breast cancer.

BACKGROUND: The aim of this study was to develop a machine learning (ML) based model to accurately predict pathologic complete response (pCR) to neoadjuvant chemotherapy (NAC) using pretreatment clinical and pathological characteristics of electronic medical record (EMR) data in breast cancer (BC). METHODS: The EMR data from patients diagnosed with early and locally advanced BC and who received NAC followed by curative surgery were reviewed. A total of 16 clinical and pathological characteristics was selected to develop ML model. We practiced six ML models using default settings for multivariate analysis with extracted variables. RESULTS: In total, 2065 patients were included in this analysis. Overall, 30.6% (n = 632) of patients achieved pCR. Among six ML models, the LightGBM had the highest area under the curve (AUC) for pCR prediction. After hyper-parameter tuning with Bayesian optimization, AUC was 0.810. Performance of pCR prediction models in different histology-based subtypes was compared. The AUC was highest in HR+HER2- subgroup and lowest in HR-/HER2- subgroup (HR+/HER2- 0.841, HR+/HER2+ 0.716, HR-/HER2 0.753, HR-/HER2- 0.653). CONCLUSIONS: A ML based pCR prediction model using pre-treatment clinical and pathological characteristics provided useful information to predict pCR during NAC. This prediction model would help to determine treatment strategy in patients with BC planned NAC.

Nanotopography-Promoted Formation of Axon Collateral Branches of Hippocampal Neurons.

Axon collateral branches, as a key structural motif of neurons, allow neurons to integrate information from highly interconnected, divergent networks by establishing terminal boutons. Although physical cues are generally known to have a comprehensive range of effects on neuronal development, their involvement in axonal branching remains elusive. Herein, it is demonstrated that the nanopillar arrays significantly increase the number of axon collateral branches and also promote their growth. Immunostaining and biochemical analyses indicate that the physical interactions between the nanopillars and the neurons give rise to lateral filopodia at the axon shaft via cytoskeletal changes, leading to the formation of axonal branches. This report, demonstrates that nanotopography regulates axonal branching, and provides a guideline for the design of sophisticated neuron-based devices and scaffolds for neuro-engineering.

Accelerated Development of Hippocampal Neurons and Limited Adhesion of Astrocytes on Negatively Charged Surfaces.

This work examines the development of primary neurons and astrocytes on thoroughly controlled functional groups. Negatively charged surfaces presenting carboxylate (COO-) or sulfonate (SO3-) groups prove beneficial to neuronal behavior, in spite of their supposed repulsive electrostatic interactions with cellular membranes. The adhesion and survival of primary hippocampal neurons on negatively charged surfaces are comparable to or slightly better than those on positively charged (poly-d-lysine-coated) surfaces, and neuritogenesis and neurite outgrowth are accelerated on COO- and SO3- surfaces. Moreover, such favorable influences of the negatively charged surfaces are only seen in neurons but not for astrocytes. Our results indicate that the in vitro developmental behavior of primary hippocampal neurons is sophisticatedly modulated by angstrom-sized differences in chemical structure or the charge density of the surface. We believe that this work provides new implications for understanding neuron-material interfaces as well as for establishing new ways to fabricate neuro-active surfaces.

Tissue-based metabolic labeling of polysialic acids in living primary hippocampal neurons.

The posttranslational modification of neural cell-adhesion molecule (NCAM) with polysialic acid (PSA) and the spatiotemporal distribution of PSA-NCAM play an important role in the neuronal development. In this work, we developed a tissue-based strategy for metabolically incorporating an unnatural monosaccharide, peracetylated N-azidoacetyl-D-mannosamine, in the sialic acid biochemical pathway to present N-azidoacetyl sialic acid to PSA-NCAM. Although significant neurotoxicity was observed in the conventional metabolic labeling that used the dissociated neuron cells, neurotoxicity disappeared in this modified strategy, allowing for investigation of the temporal and spatial distributions of PSA in the primary hippocampal neurons. PSA-NCAM was synthesized and recycled continuously during neuronal development, and the two-color labeling showed that newly synthesized PSA-NCAMs were transported and inserted mainly to the growing neurites and not significantly to the cell body. This report suggests a reliable and cytocompatible method for in vitro analysis of glycans complementary to the conventional cell-based metabolic labeling for chemical glycobiology.

Identification of feedback loops in neural networks based on multi-step Granger causality.

MOTIVATION: Feedback circuits are crucial network motifs, ubiquitously found in many intra- and inter-cellular regulatory networks, and also act as basic building blocks for inducing synchronized bursting behaviors in neural network dynamics. Therefore, the system-level identification of feedback circuits using time-series measurements is critical to understand the underlying regulatory mechanism of synchronized bursting behaviors. RESULTS: Multi-Step Granger Causality Method (MSGCM) was developed to identify feedback loops embedded in biological networks using time-series experimental measurements. Based on multivariate time-series analysis, MSGCM used a modified Wald test to infer the existence of multi-step Granger causality between a pair of network nodes. A significant bi-directional multi-step Granger causality between two nodes indicated the existence of a feedback loop. This new identification method resolved the drawback of the previous non-causal impulse response component method which was only applicable to networks containing no co-regulatory forward path. MSGCM also significantly improved the ratio of correct identification of feedback loops. In this study, the MSGCM was testified using synthetic pulsed neural network models and also in vitro cultured rat neural networks using multi-electrode array. As a result, we found a large number of feedback loops in the in vitro cultured neural networks with apparent synchronized oscillation, indicating a close relationship between synchronized oscillatory bursting behavior and underlying feedback loops. The MSGCM is an efficient method to investigate feedback loops embedded in in vitro cultured neural networks. The identified feedback loop motifs are considered as an important design principle responsible for the synchronized bursting behavior in neural networks.

Evaluating the Risk of Paroxysmal Atrial Fibrillation in Noncardioembolic Ischemic Stroke Using Artificial Intelligence-Enabled ECG Algorithm.

Background: The identification of latent atrial fibrillation (AF) in patients with ischemic stroke (IS) attributed to noncardioembolic etiology may have therapeutic implications. An artificial intelligence (AI) model identifying the electrocardiographic signature of AF present during normal sinus rhythm (NSR; AI-ECG-AF) can identify individuals with a high likelihood of paroxysmal AF (PAF) with NSR electrocardiogram (ECG). Objectives: Using AI-ECG-AF, we aimed to compare the PAF risk between noncardioembolic IS subgroups and general patients of a university hospital after controlling for confounders. Further, we sought to compare the risk of PAF among noncardioembolic IS subgroups. Methods: After training AI-ECG-AF with ECG data of university hospital patients, model inference outputs were obtained for the control group (i.e., general patient population) and NSRs of noncardioembolic IS patients. We conducted multiple linear regression (MLiR) and multiple logistic regression (MLoR) analyses with inference outputs (for MLiR) or their binary form (set at threshold = 0.5 for MLoR) used as dependent variables and patient subgroups and potential confounders (age and sex) set as independent variables. Results: The number of NSRs inferenced for the control group, cryptogenic, large artery atherosclerosis (LAA), and small artery occlusion (SAO) strokes were 133,340, 133, 276, and 290, respectively. The regression analyses indicated that patients with noncardioembolic IS had a higher PAF risk based on AI-ECG-AF relative to the control group, after controlling for confounders with the "cryptogenic" subgroup having the highest risk (odds ratio [OR] = 1.974, 95% confidence interval [CI]: 1.371-2.863) followed by the "LAA" (OR = 1.592, 95% CI: 1.238-2.056) and "SAO" subgroups (OR = 1.400, 95% CI: 1.101-1.782). Subsequent regression analyses failed to illustrate the differences in PAF risk based on AI-ECG-AF among noncardioembolic IS subgroups. Conclusion: Using AI-ECG-AF, we found that noncardioembolic IS patients had a higher PAF risk relative to the general patient population. The results from our study imply the need for more vigorous cardiac monitoring in noncardioembolic IS patients. AI-ECG-AF can be a cost-effective screening tool to identify high-risk noncardioembolic IS patients of PAF on-the-spot to be candidates for receiving additional prolonged cardiac monitoring. Our study highlights the potential of AI in clinical practice.

Contribution of Extracellular Matrix Component Landscapes in the Adult Subventricular Zone to the Positioning of Neural Stem/Progenitor Cells.

Neurogenesis persists in restricted regions of the adult brain, including the subventricular zone (SVZ). Adult neural stem cells (NSCs) in the SVZ proliferate and give rise to new neurons and glial cells depending on intrinsic and environmental cues. Among the multiple factors that contribute to the chemical, physical, and mechanical components of the neurogenic niche, we focused on the composition of the extracellular matrix (ECM) of vasculature and fractones in the SVZ. The SVZ consists of ECM-rich blood vessels and fractones during development and adulthood, and adult neural stem/progenitor cells (NS/PCs) preferentially attach to the laminin-rich basal lamina. To examine the ECM preference of adult NS/PCs, we designed a competition assay using cell micropatterning. Although both laminin and collagen type IV, which are the main components of basal lamina, act as physical scaffolds, adult NS/PCs preferred to adhere to laminin over collagen type IV. Interestingly, the ECM preference of adult NS/PCs could be manipulated by chemokines such as stromal-derived factor 1 (SDF1) and α6 integrin. As SDF1 re-routes NSCs and their progenitors toward the injury site after brain damage, these results suggest that the alteration in ECM preferences may provide a molecular basis for contextdependent NS/PC positioning.

Multimodal deep learning models for the prediction of pathologic response to neoadjuvant chemotherapy in breast cancer.

The achievement of the pathologic complete response (pCR) has been considered a metric for the success of neoadjuvant chemotherapy (NAC) and a powerful surrogate indicator of the risk of recurrence and long-term survival. This study aimed to develop a multimodal deep learning model that combined clinical information and pretreatment MR images for predicting pCR to NAC in patients with breast cancer. The retrospective study cohort consisted of 536 patients with invasive breast cancer who underwent pre-operative NAC. We developed a deep learning model to fuse high-dimensional MR image features and the clinical information for the pretreatment prediction of pCR to NAC in breast cancer. The proposed deep learning model trained on all datasets as clinical information, T1-weighted subtraction images, and T2-weighted images shows better performance with area under the curve (AUC) of 0.888 as compared to the model using only clinical information (AUC = 0.827, P < 0.05). Our results demonstrate that the multimodal fusion approach using deep learning with both clinical information and MR images achieve higher prediction performance compared to the deep learning model without the fusion approach. Deep learning could integrate pretreatment MR images with clinical information to improve pCR prediction performance.

Generative Model for Proposing Drug Candidates Satisfying Anticancer Properties Using a Conditional Variational Autoencoder.

Deep learning-based molecular generative models have successfully identified drug candidates with desired properties against biological targets of interest. However, syntactically invalid molecules generated from a deep learning-generated model hinder the model from being applied to drug discovery. Herein, we propose a conditional variational autoencoder (CVAE) as a generative model to propose drug candidates with the desired property outside a data set range. We train the CVAE using molecular fingerprints and corresponding GI50 (inhibition of growth by 50%) results for breast cancer cell lines instead of training with various physical properties for each molecule together. We confirm that the generated fingerprints, not included in the training data set, represent the desired property using the CVAE model. In addition, our method can be used as a query expansion method for searching databases because fingerprints generated using our method can be regarded as expanded queries.

Astrocyte-Encapsulated Hydrogel Microfibers Enhance Neuronal Circuit Generation.

Astrocytes, the most representative glial cells in the brain, play a multitude of crucial functions for proper neuronal development and synaptic-network formation, including neuroprotection as well as physical and chemical support. However, little attention has been paid, in the neuroregenerative medicine and related fields, to the cytoprotective incorporation of astrocytes into neuron-culture scaffolds and full-fledged functional utilization of encapsulated astrocytes for controlled neuronal development. In this article, a 3D neurosupportive culture system for enhanced induction of neuronal circuit generation is reported, where astrocytes are confined in hydrogel microfibers and protected from the outside. The astrocyte-encapsulated microfibers significantly accelerate the neurite outgrowth and guide its directionality, and enhance the synaptic formation, without any physical contact with the neurons. This astrocyte-laden system provides a pivotal culture scaffold for advanced development of cell-based therapeutics for neural injuries, such as spinal cord injury.

Slow-Wave Recordings From Micro-Sized Neural Clusters Using Multiwell Type Microelectrode Arrays.

OBJECTIVE: The use of microelectrode array (MEA) recordings is a very effective neurophysiological method because it is able to continuously and noninvasively obtain the spatiotemporal information of electrical activity from many neurons constituting a neural network. Very recently, studies have been published that used MEAs for the measurement of a low-frequency component of electrical activity as an indicator of diverse activity of cultured neurons. The occurrence of low-frequency activities has electrophysiological information that does not include the information from fast spikes. However, there is no in vitro experimental model suitable for measuring the low-frequency activities (slow-waves) for further study. METHODS: Neural clusters consisting of dozens of neurons were placed directly onto each electrode of an MEA from which fast spikes and slow-waves were measured. RESULTS: We obtained sufficient data on the early development patterns of the slow-waves and the spikes measured from many independent neural clusters confirming that the slow-waves occurred first before the emergence of the spikes in the neural clusters. We also showed that changes in the occurrence frequency of the slow-waves for synaptic blockers were measured from a large number of independent cultures. CONCLUSION: Microsized neural cluster arrays, which can be combined with conventional MEAs, are suitable for multiple simultaneous recordings of slow-waves. SIGNIFICANCE: Our technology provides a simple but useful method to study the generation of a low-frequency component of the electrical activity in cultured neural networks that are not yet well known as well as to expand the use of conventional MEAs.

A monitoring system for axonal growth dynamics using micropatterns of permissive and Semaphorin 3F chemorepulsive signals.

Neurons reach their correct targets by directional outgrowth of axons, which is mediated by attractive or repulsive cues. Growing axons occasionally cross a field of repulsive cues and stop at intermediate targets on the journey to their final destination. However, it is not well-understood how individual growth cones make decisions, and pass through repulsive territory to reach their permissive target regions. We developed a microcontact printing culture system that could trap individual axonal tips in a permissive dot area surrounded by the repulsive signal, semaphorin 3F (Sema3F). Axons of rat hippocampal neurons on the Sema3F/PLL dot array extended in the checkboard pattern with a significantly slow growth rate. The detailed analysis of the behaviors of axonal growth cones revealed the saccadic dynamics in the dot array system. The trapped axonal tips in the permissive area underwent growth cone enlargement with remarkably spiky filopodia, promoting their escape from the Sema3F constraints with straight extension of axons. This structured axonal growth on the dot pattern was disrupted by increased inter-dot distance, or perturbing intracellular signaling machineries. These data indicate that axons grow against repulsive signals by jumping over the repulsive cues, depending on contact signals and intracellular milieu. Our study suggests that our dot array culture system can be used as a screening system to easily and efficiently evaluate ECM or small molecule inhibitors interfering growth cone dynamics leading to controlling axonal growth.

Pioneering Effects and Enhanced Neurite Complexity of Primary Hippocampal Neurons on Hierarchical Neurotemplated Scaffolds.

In this work, the use of scaffolds is reported, templated from live neurons as an advanced culture platform for primary neurons. Hippocampal neurons cultured on neurotemplated scaffolds exhibit an affinity for templated somas, revealing a preference for micrometric structures amidst nanotopographical features. It is also reported, for the first time, that neurite complexity can be topographically controlled by increasing the density of nanometric features on neurotemplated scaffolds. Neurotemplated scaffolds are versatile, hierarchical topographies that feature biologically relevant structures, in both form and scale, and capture the true complexity of an in vivo environment. The introduction and implementation of neurotemplated scaffolds is sure to advance research in the fields of neurodevelopment, network development, and neuroregeneration.

Characterization of Axonal Spikes in Cultured Neuronal Networks Using Microelectrode Arrays and Microchannel Devices.

OBJECTIVE: Axonal propagation has a pivotal role in information processing in the brain. However, there has been little experimental study due to the difficulty of isolation of axons and recording their signals. Here, we developed dual chamber neuronal network interconnected with axons by integrating microchannel devices with microelectrode arrays (MEAs) to investigate axonal signals in developmental stage. METHODS: The device was composed of two chambers and microchannels between them, and hippocampal neurons were cultured in both chambers. Neuronal activity was recorded for four weeks. RESULTS: Large axonal signal was detected in microchannels, which were 137.0 ± 8.5 µV at 14 days in vitro (DIV). It was significantly larger than those in chambers with a similar range of signal-to-noise ratio. Detection efficiency of axonal spikes was analyzed by calculating the number of active electrodes over time. We found that active electrodes were detected earlier and their number increased faster in microchannels than those in chambers. Finally, we estimated the axonal conduction velocity and 73% of axons had the velocity in range of 0.2-0.5 m/s at 14 DIV. By estimating the velocity over the cultivation period, we observed that axonal conduction velocity increased linearly over time. CONCLUSION: Using MEAs and microchannel devices, we successfully detected large axonal signals and analyzed their detection efficiency and conduction velocity. We first showed the gradual increase in conduction velocity depending on cultivation days. SIGNIFICANCE: The developed microchannel device integrated MEA may be applicable for the studies of axonal conduction in cultured neuronal networks.

Cell-Type Dependent Effect of Surface-Patterned Microdot Arrays on Neuronal Growth.

Surface micropatterns have been widely used as chemical cues to control the microenvironment of cultured neurons, particularly for neurobiological assays and neurochip designs. However, the cell-type dependency on the interactions between neurons and underlying micropatterns has been rarely investigated despite the inherent differences in the morphology of neuronal types. In this study, we used surface-printed microdot arrays to investigate the effect of the same micropatterns on the growth of mouse spinal interneuron, mouse hippocampal neurons, and rat hippocampal neurons. While mouse hippocampal neurons showed no significantly different growth on control and patterned substrates, we found the microdot arrays had different effects on early neuronal growth depending on the cell type; spinal interneurons tended to grow faster in length, whereas hippocampal neurons tended to form more axon collateral branches in response to the microdot arrays. Although there was a similar trend in the neurite length and branch number of both neurons changed across the microdot arrays with the expanded range of size and spacing, the dominant responses of each neuron, neurite elongation of mouse spinal interneurons and branching augmentation of rat hippocampal neurons were still preserved. Therefore, our results demonstrate that the same design of micropatterns could cause different neuronal growth results, raising an intriguing issue of considering cell types in neural interface designs.

Synaptic compartmentalization by micropatterned masking of a surface adhesive cue in cultured neurons.

Functions of neuronal circuit are fundamentally modulated by its quality and quantity of connections. Assessment of synapse, the basic unit for a neuronal connection, is labor-intensive and time-consuming in conventional culture systems, due to the small size and the spatially random distribution. In the present study, we propose a novel 'synapse compartmentalization' culture system, in which synapses are concentrated at controlled locations. We fabricated a negative dot array pattern by coating the entire surface with poly-l-lysine (PLL) and subsequent microcontact printing of 1) substrates which mask positive charge of PLL (Fc, BSA and laminin), or 2) a chemorepulsive protein (Semaphorin 3F-Fc). By combination of physical and biological features of these repulsive substrates, functional synapses were robustly concentrated in the PLL-coated dots. This synapse compartmentalization chip can be combined with the various high-throughput assay formats based on the synaptic morphology and function. Therefore, this quantifiable and controllable dot array pattern by microcontact printing will be potential useful for bio-chip platforms for the high-density assays used in synapse-related neurobiological studies.

In vitro neurite guidance effects induced by polylysine pinstripe micropatterns with polylysine background.

Engineered culture substrates with chemical neurite guidance cues have been used for studying the mechanism of axon pathfinding at cellular level. In this study, we designed a novel poly-l-lysine (PLL) micropattern ("pinstripe micropattern") to investigate how the same biomolecules with slightly different surface concentration can affect in vitro neuronal growth. The pinstripe micropattern was fabricated by stamping PLL on a PLL-coated glass coverslip, which resulted in denser PLL lines and a less-dense PLL background. There were two effects of the substrate on cultured primary hippocampal neuron: neurite initiation and growth cone turning. Although the whole surface was permissive for neurite outgrowth, we observed that the growth direction of neurites had a strong tendency to follow the stamped PLL line patterns with PLL background. However, the micropattern did not affect the spreading of cell body on the substrate. According to these investigations, we concluded that the PLL pinstripe pattern with PLL background, which had the step difference of polylysine concentrations, would be very useful for designing novel cell assays for the investigation of neurite guidance mechanisms, and suggested it as a new design method for controlling the direction of neurite growth on in vitro neural network.

Effects of ECM protein micropatterns on the migration and differentiation of adult neural stem cells.

The migration and differentiation of adult neural stem cells (aNSCs) are believed to be strongly influenced by the spatial distribution of extracellular matrix (ECM) proteins in the stem cell niche. In vitro culture platform, which involves the specific spatial distribution of ECM protein, could offer novel tools for better understanding of aNSC behavior in the spatial pattern of ECM proteins. In this work, we applied soft-lithographic technique to design simple and reproducible laminin (LN)-polylysine cell culture substrates and investigated how aNSCs respond to the various spatial distribution of laminin, one of ECM proteins enriched in the aNSC niche. We found that aNSC preferred to migrate and attach to LN stripes, and aNSC-derived neurons and astrocytes showed significant difference in motility towards LN stripes. By changing the spacing of LN stripes, we were able to control the alignment of neurons and astrocytes. To the best of our knowledge, this is the first time to investigate the differential cellular responses of aNSCs on ECM protein (LN) and cell adhesive synthetic polymer (PDL) using surface micropatterns. Our findings would provide a deeper understanding in astrocyte-neuron interactions as well as ECM-stem cell interactions.

Surface-printed microdot array chips for the quantification of axonal collateral branching of a single neuron in vitro.

Precise and quantitative control of extracellular signalling cues using surface-engineered chips has facilitated various neurobiological assays in vitro. Although the formation of axon collateral branches is important for the establishment and refinement of the neuronal connections during the development and regeneration, surface designs for controlling branch phenotypes have been rarely proposed. In this work, we fabricated a surface-printed microdot array for controlling axon branch formation. Following the culture of hippocampal neurons on a 5 µm dot array patterned by micro-contact printing of poly-d-lysine, we found that most axon collateral branches were initiated from axonal regions on a microdot and terminated on neighboring dots. In addition, the length of branches increased as the spacing between dots increased. Surprisingly, other morphological features were not significantly different from the neurons cultured on a conventional unpatterned surface. Further investigation of this phenomenon indicated that the branch-forming machineries, such as actin patches, were focused on the dot. According to these investigations, we concluded that discontinuous adhesion spots given by dot arrays arranged the branching formation on the expectable location and direction. Therefore, microdot arrays will be applicable as the surface design parameter of bio-chip platforms to reduce branching complexity and quantize branching formation for the simple and easy assay in neurobiological studies.

High-efficiency blue phosphorescent organic light-emitting diodes using a carbazole and carboline-based host material.

A novel bipolar host 9-(4-(9H-pyrido[2,3-b]indol-9-yl)phenyl)-9H-3,9'-bicarbazole (pBCb2Cz) was prepared for high efficiency blue phosphorescent organic light-emitting diodes (PhOLEDs), a high triplet energy (ET) material, employing electron-deficient α-carboline. pBCb2Cz (ET = 2.93 eV) was effective as a host material for FIrpic- and FCNIrpic-based blue PhOLEDs, and highest quantum efficiencies of 23.0 and 16.2%, respectively, were achieved.