Synergistic HNO3-H2SO4-NH3 upper tropospheric particle formation.

New particle formation in the upper free troposphere is a major global source of cloud condensation nuclei (CCN)1-4. However, the precursor vapours that drive the process are not well understood. With experiments performed under upper tropospheric conditions in the CERN CLOUD chamber, we show that nitric acid, sulfuric acid and ammonia form particles synergistically, at rates that are orders of magnitude faster than those from any two of the three components. The importance of this mechanism depends on the availability of ammonia, which was previously thought to be efficiently scavenged by cloud droplets during convection. However, surprisingly high concentrations of ammonia and ammonium nitrate have recently been observed in the upper troposphere over the Asian monsoon region5,6. Once particles have formed, co-condensation of ammonia and abundant nitric acid alone is sufficient to drive rapid growth to CCN sizes with only trace sulfate. Moreover, our measurements show that these CCN are also highly efficient ice nucleating particles-comparable to desert dust. Our model simulations confirm that ammonia is efficiently convected aloft during the Asian monsoon, driving rapid, multi-acid HNO3-H2SO4-NH3 nucleation in the upper troposphere and producing ice nucleating particles that spread across the mid-latitude Northern Hemisphere.

Hidden hotspots of amphibian biodiversity in China.

Identifying and protecting hotspots of endemism and species richness is crucial for mitigating the global biodiversity crisis. However, our understanding of spatial diversity patterns is far from complete, which severely limits our ability to conserve biodiversity hotspots. Here, we report a comprehensive analysis of amphibian species diversity in China, one of the most species-rich countries on Earth. Our study combines 20 y of field surveys with new molecular analyses of 521 described species and also identifies 100 potential cryptic species. We identify 10 hotspots of amphibian diversity in China, each with exceptional species richness and endemism and with exceptional phylogenetic diversity and phylogenetic endemism (based on a new time-calibrated, species-level phylogeny for Chinese amphibians). These 10 hotspots encompass 59.6% of China's described amphibian species, 49.0% of cryptic species, and 55.6% of species endemic to China. Only four of these 10 hotspots correspond to previously recognized biodiversity hotspots. The six new hotspots include the Nanling Mountains and other mountain ranges in South China. Among the 186 species in the six new hotspots, only 9.7% are well covered by protected areas and most (88.2%) are exposed to high human impacts. Five of the six new hotspots are under very high human pressure and are in urgent need of protection. We also find that patterns of richness in cryptic species are significantly related to those in described species but are not identical.

Five Porous Complexes Constructed from a Racemic Ligand: Synthesis, Chiral Self-Assembly, Iodine Adsorption, and Desorption Properties.

Herein, a chiral bispyridyl ligand (L) was designed and synthesized using a Schiff base condensation reaction, followed by a 1,3-H shift. Five complexes, [Zn2L2(OAc)4] (1), {[CdLCl2(DMF)]·4H2O}n (2), [Cd2L2I4]·4H2O (3), {[CdL2(H2O)2](NO3)2·2CH3OH}n (4), and [Hg2L2Cl4]·2DMF (5), were synthesized and characterized upon its reaction with Zn(II), Cd(II), or Hg(II) ions, respectively. X-ray crystallography shows that the organic compound exists as a racemic ligand with equal amounts of its R- and S-isomers, and all of the synthesized complexes exhibit heterochiral self-assembly via a chiral self-discrimination process. Complexes 1, 3, and 5 exist as centrosymmetric binuclear metallamacrocycles, while complexes 2 and 4 exist as 1D looped-chain coordination polymers. Inspired by the assembled structures of the five complexes, I2 adsorption/desorption measurements for the complexes were carried out. The results show that complexes 1 and 5 exhibit good adsorption capacities toward I2 in n-hexane and in water, respectively.

Coordination-Driven Heterochiral Self-Assembly: Construction of Cd(II) Coordination Polymers with Sorption Behaviors for Iodine and Dyes.

A racemic bispyridyl ligand (L) was synthesized via a Schiff base condensation reaction. Four Cd(II) complexes, {[CdL2Cl2]·2DMF}n (1), [CdLI2]n (2), {[CdL2Br2]·4H2O}n (3), and {[CdL2(H2O)2](NO3)2·2CH3OH·8H2O}n (4), were synthesized and further characterized based on this ligand. Single-crystal structures show that the coordination-driven assembly of the bispyridyl ligand with Cd(II) salts bearing different counteranions can lead to multidimensional coordination polymers via a heterochiral self-discrimination process. Complex 1 exists as a one-dimensional (1D) looped chain polymer, and complex 2 exists as a 1D zigzag chain polymer. Complex 3 is a 2D grid coordination polymer, and complex 4 exists as a 3D framework polymer. Furthermore, the iodine sorption capacities of the four complexes were investigated in the solution of n-hexane and water as well as in the iodine steam. The dye sorption behaviors were investigated in water, which showed that complex 2 exhibited good adsorption for crystal violet (CV), while complex 4 had good adsorption capability toward direct yellow 4 (DY).

Enhancing Nanofiltration Selectivity of Metal-Organic Framework Membranes via a Confined Interfacial Polymerization Strategy.

Development of well-constructed metal-organic framework (MOF) membranes can bring about breakthroughs in nanofiltration (NF) performance for water treatment applications, while the relatively loose structures and inevitable defects usually cause low rejection capacity of MOF membranes. Herein, a confined interfacial polymerization (CIP) method is showcased to synthesize polyamide (PA)-modified NF membranes with MOF nanosheets as the building blocks, yielding a stepwise transition from two-dimensional (2D) MOF membranes to polyamide NF membranes. The CIP process was regulated by adjusting the loading amount of piperazine (PIP)-grafted MOF nanosheets on substrates and the additional content of free PIP monomers distributed among the nanosheets, followed by the reaction with trimesoyl chloride in the organic phase. The prepared optimal membrane exhibited a high Na2SO4 rejection of 98.4% with a satisfactory water permeance of 37.4 L·m-2·h-1·bar-1, which could be achieved by neither the pristine 2D MOF membranes nor the PA membranes containing the MOF nanosheets as the conventional interlayer. The PA-modified MOF membrane also displayed superior stability and enhanced antifouling ability. This CIP strategy provides a novel avenue to develop efficient MOF-based NF membranes with high ion-sieving separation performance for water treatment.

Molecular Understanding of the Enhancement in Organic Aerosol Mass at High Relative Humidity.

The mechanistic pathway by which high relative humidity (RH) affects gas-particle partitioning remains poorly understood, although many studies report increased secondary organic aerosol (SOA) yields at high RH. Here, we use real-time, molecular measurements of both the gas and particle phase to provide a mechanistic understanding of the effect of RH on the partitioning of biogenic oxidized organic molecules (from α-pinene and isoprene) at low temperatures (243 and 263 K) at the CLOUD chamber at CERN. We observe increases in SOA mass of 45 and 85% with increasing RH from 10-20 to 60-80% at 243 and 263 K, respectively, and attribute it to the increased partitioning of semi-volatile compounds. At 263 K, we measure an increase of a factor 2-4 in the concentration of C10H16O2-3, while the particle-phase concentrations of low-volatility species, such as C10H16O6-8, remain almost constant. This results in a substantial shift in the chemical composition and volatility distribution toward less oxygenated and more volatile species at higher RH (e.g., at 263 K, O/C ratio = 0.55 and 0.40, at RH = 10 and 80%, respectively). By modeling particle growth using an aerosol growth model, which accounts for kinetic limitations, we can explain the enhancement in the semi-volatile fraction through the complementary effect of decreased compound activity and increased bulk-phase diffusivity. Our results highlight the importance of particle water content as a diluting agent and a plasticizer for organic aerosol growth.

New chromones from the roots of Saposhnikovia divaricata (Turcz.) Schischk with anti-inflammatory activity.

Fifteen new chromones, sadivamones A-E (1-5), cimifugin monoacetate (6), sadivamones F-N (7-15), together with fifteen known chromones (16-30), were isolated from the ethyl acetate portions of 70% ethanol extract of Saposhnikovia divaricata (Turcz.) Schischk roots. The structures of the isolates were determined using 1D/2D NMR data and electron circular dichroism (ECD) calculations. Meanwhile, LPS induced RAW264.7 inflammatory cell model was used to determine the potential anti-inflammatory activity of all the isolated compounds in vitro. The results showed that compounds 2, 8, 12-13, 18, 20-22, 24, and 27 significantly inhibited the production of lipopolysaccharide (LPS)-induced NO in macrophages. To determine the signaling pathways involved in the suppression of NO production by compounds 8, 12 and 13, we investigated ERK and c-Jun N-terminal protein kinase (JNK) expression by western blot analysis. Further mechanistic studies demonstrated that compounds 12 and 13 inhibited the phosphorylation of ERK and the activation of ERK and JNK signaling in RAW264.7 cells via MAPK signaling pathways. Taken together, compounds 12 and 13 may be valuable candidates for the treatment of inflammatory diseases.

Exploring the role of Hmox1 in ferroptosis and immune infiltration during renal ischemia-reperfusion injury.

Ischemia-reperfusion injury (IRI) is inevitable in kidney transplantations and, as a complex pathophysiological process, it can be greatly impacted by ferroptosis and immune inflammation. Our study aimed to identify the biomarkers of renal IRI (RIRI) and elucidate their relationship with immune infiltration. In this study, the GSE148420 database was used as a training set to analyze differential genes and overlap them with ferroptosis-related genes to identify hub genes using a protein-protein interaction (PPI) network, the least absolute shrinkage and selection operator (LASSO), and random forest algorithm (RFA). We verified the hub gene and ferroptosis-related phenotypes in a verification set and animal experiments involving unilateral IRI with contralateral nephrectomy in rats. Gene set enrichment analysis (GSEA) of single genes was conducted according to the hub gene to predict related endogenous RNAs (ceRNAs) and drugs to establish a network. Finally, we used the Cibersort to analyze immunological infiltration and conducted Spearman's correlation analysis. We identified 5456 differential genes and obtained 26 ferroptosis-related differentially expressed genes. Through PPI, LASSO, and RFA, Hmox1 was identified as the only hub gene and its expression levels were verified using verification sets. In animal experiments, Hmox1 was verified as a key biomarker. GSEA of single genes revealed the seven most related pathways, and the ceRNAs network included 138 mRNAs and miRNAs. We predicted 11 related drugs and their three-dimensional structural maps. Thus, Hmox1 was identified as a key biomarker and regulator of ferroptosis in RIRI and its regulation of ferroptosis was closely related to immune infiltration.

Phenylpropanoids from Solanum capsicoides and their anti-inflammatory activity.

Two new phenylpropanoids, 4-O-(1''-O-cis-caffeoyl)-ß-glucopyran osyl-1-allyl-3-methoxy-benzene (1), 4'-O-(1''-O-cis-caffeoyl)-ß-glucopyranosyl-hydroxymegastigm-4-en-3-one (2), together with nine known compounds were obtained from the leaves of Solanum capsicoides. Their structures were elucidated based on spectroscopic methods, and comparing spectral data with those in literature. Meanwhile, their anti-inflammatory activities were evaluated on (LPS)-induced RAW 246.7 cells, and 1, 9, and 10 showed better inhibitory effects with IC50 values of 17.19 ± 1.12, 18.15 ± 0.47, and 19.8 ± 0.95 µM, respectively.

Combined angiogenesis and PD-1 inhibition for immunomodulatory TNBC: concept exploration and biomarker analysis in the FUTURE-C-Plus trial.

BACKGROUND: Immune checkpoint inhibitors had a great effect in triple-negative breast cancer (TNBC); however, they benefited only a subset of patients, underscoring the need to co-target alternative pathways and select optimal patients. Herein, we investigated patient subpopulations more likely to benefit from immunotherapy and inform more effective combination regimens for TNBC patients. METHODS: We conducted exploratory analyses in the FUSCC cohort to characterize a novel patient selection method and actionable targets for TNBC immunotherapy. We investigated this in vivo and launched a phase 2 trial to assess the clinical value of such criteria and combination regimen. Furthermore, we collected clinicopathological and next-generation sequencing data to illustrate biomarkers for patient outcomes. RESULTS: CD8-positivity could identify an immunomodulatory subpopulation of TNBCs with higher possibilities to benefit from immunotherapy, and angiogenesis was an actionable target to facilitate checkpoint blockade. We conducted the phase II FUTURE-C-Plus trial to assess the feasibility of combining famitinib (an angiogenesis inhibitor), camrelizumab (a PD-1 monoclonal antibody) and chemotherapy in advanced immunomodulatory TNBC patients. Within 48 enrolled patients, the objective response rate was 81.3% (95% CI, 70.2-92.3), and the median progression-free survival was 13.6 months (95% CI, 8.4-18.8). No treatment-related deaths were reported. Patients with CD8- and/or PD-L1- positive tumors benefit more from this regimen. PKD1 somatic mutation indicates worse progression-free and overall survival. CONCLUSION: This study confirms the efficacy and safety of the triplet regimen in immunomodulatory TNBC and reveals the potential of combining CD8, PD-L1 and somatic mutations to guide clinical decision-making and treatments. TRIAL REGISTRATION: ClinicalTrials.gov: NCT04129996 . Registered 11 October 2019.

Radiomics model based on shear-wave elastography in the assessment of axillary lymph node status in early-stage breast cancer.

OBJECTIVES: To develop and validate an ultrasound elastography radiomics nomogram for preoperative evaluation of the axillary lymph node (ALN) burden in early-stage breast cancer. METHODS: Data of 303 patients from hospital #1 (training cohort) and 130 cases from hospital #2 (external validation cohort) between Jun 2016 and May 2019 were enrolled. Radiomics features were extracted from shear-wave elastography (SWE) and corresponding B-mode ultrasound (BMUS) images. The minimum redundancy maximum relevance and least absolute shrinkage and selection operator algorithms were used to select ALN status-related features. Proportional odds ordinal logistic regression was performed using the radiomics signature together with clinical data, and an ordinal nomogram was subsequently developed. We evaluated its performance using C-index and calibration. RESULTS: SWE signature, US-reported LN status, and molecular subtype were independent risk factors associated with ALN status. The nomogram based on these variables showed good discrimination in the training (overall C-index: 0.842; 95%CI, 0.773-0.879) and the validation set (overall C-index: 0.822; 95%CI, 0.765-0.838). For discriminating between disease-free axilla (N0) and any axillary metastasis (N + (≥ 1)), it achieved a C-index of 0.845 (95%CI, 0.777-0.914) for the training cohort and 0.817 (95%CI, 0.769-0.865) for the validation cohort. The tool could also discriminate between low (N + (1-2)) and heavy metastatic ALN burden (N + (≥ 3)), with a C-index of 0.827 (95%CI, 0.742-0.913) in the training cohort and 0.810 (95%CI, 0.755-0.864) in the validation cohort. CONCLUSION: The radiomics model shows favourable predictive ability for ALN staging in patients with early-stage breast cancer, which could provide incremental information for decision-making. KEY POINTS: • Radiomics analysis helps radiologists to evaluate the axillary lymph node status of breast cancer with accuracy. • This multicentre retrospective study showed that radiomics nomogram based on shear-wave elastography provides incremental information for risk stratification. • Treatment can be given with more precision based on the model.

Exploitation of Amine Groups Cooped up in Polyamide Nanofiltration Membranes to Achieve High Rejection of Micropollutants and High Permeance of Divalent Cations.

To enhance the use of nanofiltration in the production of quality drinking water, particularly through the efficient removal of micropollutants yet still preserving essential minerals, the targeted nanofiltration membranes (NFMs) are required to have small pore dimensions coupled with a high, net-negative charge density. Herein, after the formation of a separation layer using piperazine interfacially polymerized with trimesoyl chloride, the exploitation of residual amine groups was systematically investigated by different diacyl chlorides in an organic milieu, which caused the upper part of the final separation layer to be denser and highly negatively charged. Hence, this protocol offers a novel means to fabricate NFMs simultaneously endowed with a low molecular cutoff (MWCO) of 145-238 Da and a reduced rejection of MgCl2 (48%-80%) as well as a competitive water permeance. Those features are ideally applicable to the goal of removing small micropollutants while preserving mineral ions, as needed for the energy-efficient production of safe, quality drinking water. Furthermore, an attempt was made to correlate MWCO with MgCl2 rejection, which provides some insights on the nexus of the electrostatic effects constrained by size exclusion. The significance of residual amine groups and the modification environment was unveiled, and this method paves a new avenue for designing functional NFMs.

Modulating the Asymmetry of the Active Layer in Pursuit of Nanofiltration Selectivity via Differentiating Interfacial Reactions of Piperazine.

Nanofiltration (NF), highly prospective for drinking water treatment, faces a challenge in simultaneously removing emerging contaminants while maintaining mineral salts, particularly divalent cations. To overcome this challenge, NF membranes possessing small pores concomitant with highly negatively charged surfaces were synthesized via a two-step fabrication strategy. The key is to generate a polyamide active layer having a loose and carboxyl group-abundant segment on top and a dense barrier segment underneath. This was achieved by restrained interfacial polymerization between trimesoyl chloride and partly protonated piperazine to form a highly depth-heterogeneous polyamide network, followed by second amidation in an organic environment to remove untethered polyamide fragments and associate malonyl chlorides with reserved amine groups to introduce more negative charges. Most importantly, on first-principle engineering the spatial architecture of the polyamide layer, amplifying asymmetric charge distribution was paired with the thinning of the vertical structure. The optimized membrane exhibits high salt/organic rejection selectivity and water permeance superior to most NF membranes reported previously. The rejections of eight emerging contaminants were in the range of 66.0-94.4%, much higher than the MgCl2 rejection of 41.1%. This new fabrication strategy, suitable for various diacyl chlorides, along with the new membranes so produced, offers a novel option for NF in potable water systems.

Molecular Design of the Polyamide Layer Structure of Nanofiltration Membranes by Sacrificing Hydrolyzable Groups toward Enhanced Separation Performance.

Nanofiltration (NF) is an effective technology for removing trace organic contaminants (TrOCs), while the inherent trade-off effect between water permeance and solute rejections hinders its widespread application in water treatment. Herein, we propose a novel scheme of "monomers with sacrificial groups" to regulate the microstructure of the polyamide active layer via introducing a hydrolyzable ester group onto piperazine to control the diffusion and interfacial polymerization process. The achieved benefits include narrowing the pore size, improving the interpore connectivity, enhancing the microporosity, and reducing the active layer thickness, which collectively realized the simultaneous improvement of water permeance and enhancement of TrOCs rejection performance. The resulting membranes were superior to both the control and commercial membranes, especially in water-TrOCs selectivity. The effects of using the new monomers on the membrane physicochemical properties were systematically studied, and underlying mechanisms for the enhanced separation performance were further revealed by simulating the polymerization process through density functional theory calculation and measuring the trans-interface diffusion rate of monomers. This study demonstrates a novel promising NF membrane synthesis strategy by designing the structure of reaction monomers for achieving excellent rejection of TrOCs with a low energy input in water treatment.

Boosting the Performance of Nanofiltration Membranes in Removing Organic Micropollutants: Trade-Off Effect, Strategy Evaluation, and Prospective Development.

In view of the high risks brought about by organic micropollutants (OMPs), nanofiltration (NF) processes have been playing a vital role in advanced water and wastewater treatment, owing to the high membrane performance in rejection of OMPs, permeation of water, and passage of mineral salts. Though numerous studies have been devoted to evaluating and technically enhancing membrane performance in removing various OMPs, the trade-off effect between water permeance and water/OMP selectivity for state-of-the-art membranes remains far from being understood. Knowledge of this effect is significant for comparing and guiding membrane development works toward cost-efficient OMP removal. In this work, we comprehensively assessed the performance of 88 NF membranes, commercialized or newly developed, based on their water permeance and OMP rejection data published in the literature. The effectiveness and underlying mechanisms of various modification methods in tailoring properties and in turn performance of the mainstream polyamide (PA) thin-film composite (TFC) membranes were quantitatively analyzed. The trade-off effect was demonstrated by the abundant data from both experimental measurements and machine learning-based prediction. On this basis, the advancement of novel membranes was benchmarked by the performance upper-bound revealed by commercial membranes and lab-made PA membranes. We also assessed the potentials of current NF membranes in selectively separating OMPs from inorganic salts and identified the future research perspectives to achieve further enhancement in OMP removal and salt/OMP selectivity of NF membranes.

The effect of different carbon sources on biofouling in membrane fouling simulators: microbial community and implications.

Biofouling is a problem affecting the operation of nanofiltration systems due to the complexity of the carbon matrix affecting bacteria and biofilm growth. This study used membrane fouling simulators to investigate the effects of five different carbon sources on the biofouling of nanofiltration membranes. For all the carbon sources analyzed, the increase in pressure drop was most accelerated for acetate. The use of acetate as the single carbon source produced less adenosine triphosphate but more extracellular polymers than glucose. The microbial community was analyzed using 16 s rRNA. The use of more than a single carbon source produced an increase in bacteria diversity even at similar concentrations. The relative abundance of proteobacteria was the highest at the phylum level (95%) when a single carbon source was added. Additionally, it was found that the use of different carbon sources produced a shift in the microbial community, affecting the biofouling and pressure drop on membranes.

Chemical Constituents of the Roots of Schisandra chinensis.

One new phenylpropanoid schineolignin D (1), one new sesquiterpene (-)-(7S,10S)-3,11,12,13-tetrahydroxycalamenene (2), one new quinic acid 4-(E)-O-coumaroylquinic acid ethyl ester (3), and seven known compounds 4-10 were separated from the roots of Schisandra chinensis. The chemical structures of all compounds were characterized by NMR spectroscopic experiments. All compounds were assessed for their neuroprotective effects on PC12 cell lines induced by H2 O2 . Compounds 1, 3-4, and 7 showed statistically significant neuroprotective activities with the negative control group at 12.5 µM.

Two new quinic acid derivatives from the leaves of Schisandra chinensis.

Two new (1-2) and three known quinic acid derivatives (3-5) were isolated from the leaves of Schisandra chinensis (Turcz) Baill. The structures of the compounds were determined by spectroscopic methods, especially the NMR techniques, and also by comparison with reported data in the literature. The cytotoxicity activities of these compounds were evaluated on human tumor cell lines LN229 and three of them showed a certain activity.

[Cloning and prokaryotic expression of FcLEA-D29 gene of Fritillaria cirrhosa].

Fritillaria cirrhosa, the most valuable source of the precious Chinese medicinal material Fritillariae Cirrhosae Bulbus, suffers from various stresses during growth which influence the yield and quality of the medicinal part. This study aims to explore the genes related to stress resistance in this medicinal species. To be specific, based on the transcriptome data of F. cirrhosa, a gene encoding the late embryogenesis abundant(LEA) protein was obtained, which was named as FcLEA-D29. The gene sequence and protein structure were analyzed with bioinformatics methods and qRT-PCR was used to detect the expression of the gene in different tissues and in response to temperature stress. The low-temperature tolerance of FcLEA-D29 was verified by the prokaryotic expression system. The results showed that FcLEA-D29 contained an open reading frame of 777 bp, encoding 258 amino acids. Multiple sequence alignment revealed that FcLEA-D29 protein belonged to LEA-D29 subfamily of LEA3 family. qRT-PCR results showed that FcLEA-D29 was specifically expressed in bulbs and responded to low temperature. The strain with the recombinant plasmid demonstrated better growth status than the control strain in the instance of low temperature stress, suggesting that FcLEA-D29 may play a role in bulb development and low temperature response of F. cirrhosa. This study laid a basis for further research on the role of FcLEA-D29 in the accumulation of secondary metabolites in F. cirrhosa, especially alkaloids, under low temperature and provided evidence for the low-temperature adaptation of F. cirrhosa.

Deep learning with convolutional neural network in the assessment of breast cancer molecular subtypes based on US images: a multicenter retrospective study.

OBJECTIVES: To evaluate the prediction performance of deep convolutional neural network (DCNN) based on ultrasound (US) images for the assessment of breast cancer molecular subtypes. METHODS: A dataset of 4828 US images from 1275 patients with primary breast cancer were used as the training samples. DCNN models were constructed primarily to predict the four St. Gallen molecular subtypes and secondarily to identify luminal disease from non-luminal disease based on the ground truth from immunohistochemical of whole tumor surgical specimen. US images from two other institutions were retained as independent test sets to validate the system. The models' performance was analyzed using per-class accuracy, positive predictive value (PPV), and Matthews correlation coefficient (MCC). RESULTS: The model achieved good performance in identifying the four breast cancer molecular subtypes in the two test sets, with accuracy ranging from 80.07% (95% CI, 76.49-83.23%) to 97.02% (95% CI, 95.22-98.16%) and 87.94% (95% CI, 85.08-90.31%) to 98.83% (95% CI, 97.60-99.43) for the two test cohorts for each sub-category, respectively. In terms of 4-class weighted average MCC, the model achieved 0.59 for test cohort A and 0.79 for test cohort B. Specifically, the DCNN also yielded good diagnostic performance in discriminating luminal disease from non-luminal disease, with a PPV of 93.29% (95% CI, 90.63-95.23%) and 88.21% (95% CI, 85.12-90.73%) for the two test cohorts, respectively. CONCLUSION: Using pretreatment US images of the breast cancer, deep learning model enables the assessment of molecular subtypes with high diagnostic accuracy. TRIAL REGISTRATION: Clinical trial number: ChiCTR1900027676 KEY POINTS: • Deep convolutional neural network (DCNN) helps clinicians assess tumor features with accuracy. • Multicenter retrospective study shows that DCNN derived from pretreatment ultrasound imagine improves the prediction of breast cancer molecular subtypes. • Management of patients becomes more precise based on the DCNN model.