GA-UNet: A Lightweight Ghost and Attention U-Net for Medical Image Segmentation.

U-Net has demonstrated strong performance in the field of medical image segmentation and has been adapted into various variants to cater to a wide range of applications. However, these variants primarily focus on enhancing the model's feature extraction capabilities, often resulting in increased parameters and floating point operations (Flops). In this paper, we propose GA-UNet (Ghost and Attention U-Net), a lightweight U-Net for medical image segmentation. GA-UNet consists mainly of lightweight GhostV2 bottlenecks that reduce redundant information and Convolutional Block Attention Modules that capture key features. We evaluate our model on four datasets, including CVC-ClinicDB, 2018 Data Science Bowl, ISIC-2018, and BraTS 2018 low-grade gliomas (LGG). Experimental results show that GA-UNet outperforms other state-of-the-art (SOTA) models, achieving an F1-score of 0.934 and a mean Intersection over Union (mIoU) of 0.882 on CVC-ClinicDB, an F1-score of 0.922 and a mIoU of 0.860 on the 2018 Data Science Bowl, an F1-score of 0.896 and a mIoU of 0.825 on ISIC-2018, and an F1-score of 0.896 and a mIoU of 0.853 on BraTS 2018 LGG. Additionally, GA-UNet has fewer parameters (2.18M) and lower Flops (4.45G) than other SOTA models, which further demonstrates the superiority of our model.

Conditional generative adversarial network driven radiomic prediction of mutation status based on magnetic resonance imaging of breast cancer.

BACKGROUND: Breast Cancer (BC) is a highly heterogeneous and complex disease. Personalized treatment options require the integration of multi-omic data and consideration of phenotypic variability. Radiogenomics aims to merge medical images with genomic measurements but encounter challenges due to unpaired data consisting of imaging, genomic, or clinical outcome data. In this study, we propose the utilization of a well-trained conditional generative adversarial network (cGAN) to address the unpaired data issue in radiogenomic analysis of BC. The generated images will then be used to predict the mutations status of key driver genes and BC subtypes. METHODS: We integrated the paired MRI and multi-omic (mRNA gene expression, DNA methylation, and copy number variation) profiles of 61 BC patients from The Cancer Imaging Archive (TCIA) and The Cancer Genome Atlas (TCGA). To facilitate this integration, we employed a Bayesian Tensor Factorization approach to factorize the multi-omic data into 17 latent features. Subsequently, a cGAN model was trained based on the matched side-view patient MRIs and their corresponding latent features to predict MRIs for BC patients who lack MRIs. Model performance was evaluated by calculating the distance between real and generated images using the Fréchet Inception Distance (FID) metric. BC subtype and mutation status of driver genes were obtained from the cBioPortal platform, where 3 genes were selected based on the number of mutated patients. A convolutional neural network (CNN) was constructed and trained using the generated MRIs for mutation status prediction. Receiver operating characteristic area under curve (ROC-AUC) and precision-recall area under curve (PR-AUC) were used to evaluate the performance of the CNN models for mutation status prediction. Precision, recall and F1 score were used to evaluate the performance of the CNN model in subtype classification. RESULTS: The FID of the images from the well-trained cGAN model based on the test set is 1.31. The CNN for TP53, PIK3CA, and CDH1 mutation prediction yielded ROC-AUC values 0.9508, 0.7515, and 0.8136 and PR-AUC are 0.9009, 0.7184, and 0.5007, respectively for the three genes. Multi-class subtype prediction achieved precision, recall and F1 scores of 0.8444, 0.8435 and 0.8336 respectively. The source code and related data implemented the algorithms can be found in the project GitHub at https://github.com/mattthuang/BC_RadiogenomicGAN . CONCLUSION: Our study establishes cGAN as a viable tool for generating synthetic BC MRIs for mutation status prediction and subtype classification to better characterize the heterogeneity of BC in patients. The synthetic images also have the potential to significantly augment existing MRI data and circumvent issues surrounding data sharing and patient privacy for future BC machine learning studies.

MAGP2 promotes osteogenic differentiation during fracture healing through its crosstalk with the ß-catenin pathway.

Osteogenic differentiation is important for fracture healing. Microfibrial-associated glycoprotein 2 (MAGP2) is found to function as a proangiogenic regulator in bone formation; however, its role in osteogenic differentiation during bone repair is not clear. Here, a mouse model of critical-sized femur fracture was constructed, and the adenovirus expressing MAGP2 was delivered into the fracture site. Mice with MAGP2 overexpression exhibited increased bone mineral density and bone volume fraction (BV/TV) at Day 14 postfracture. Within 7 days postfracture, overexpression of MAGP2 increased collagen I and II expression at the fracture callus, with increasing chondrogenesis. MAGP2 inhibited collagen II level but elevated collagen I by 14 days following fracture, accompanied by increased endochondral bone formation. In mouse osteoblast precursor MC3T3-E1 cells, MAGP2 treatment elevated the expression of osteoblastic factors (osterix, BGLAP and collagen I) and enhanced ALP activity and mineralization through activating ß-catenin signaling after osteogenic induction. Besides, MAGP2 could interact with lipoprotein receptor-related protein 5 (LRP5) and upregulated its expression. Promotion of osteogenic differentiation and ß-catenin activation mediated by MAGP2 was partially reversed by LRP5 knockdown. Interestingly, ß-catenin/transcription factor 4 (TCF4) increased MAGP2 expression probably by binding to MAGP2 promoter. These findings suggest that MAGP2 may interact with ß-catenin/TCF4 to enhance ß-catenin/TCF4's function and activate LRP5-activated ß-catenin signaling pathway, thus promoting osteogenic differentiation for fracture repair. mRNA sequencing identified the potential targets of MAGP2, providing novel insights into MAGP2 function and the directions for future research.

Direct Knockdown of PDZ and LIM Domain 1 Using an Adenoviral Delivery System Accelerates Osteogenesis and Fracture Healing in Mice.

Substantial advances have been made in understanding the role of partial PDZ and LIM domain family's proteins in skeletal-related diseases. Yet, little is known about the effect of PDZ and LIM Domain 1 (Pdlim1) on osteogenesis and fracture repair. This study aimed to investigate whether direct gene delivery using an adenovirus vector carrying Pdlim1 (Ad-oePdlim1) or encoding shRNA-Pdlim1 (Ad-shPdlim1) could affect the osteogenic activity of preosteoblastic MC3T3-E1 cells in vitro, and influence the fracture healing of mice in vivo. We found that Ad-shPdlim1 transfection contributed to the calcified nodule formation in MC3T3-E1 cells. Downregulation of Pdlim1 enhanced the alkaline phosphatase activity and increased the expression of osteogenic markers (Runt-related transcription factor 2 [Runx2], collagen type I alpha 1 chain [Col1A1], osteocalcin [OCN], and osteopontin [OPN]). Further analysis indicated that Pdlim1 knockdown could activate ß-catenin signaling, as evidenced by the accumulation of ß-catenin in the nucleus and the increase levels of downstream regulators such as Lef1/Tcf7, axis inhibition protein 2, cyclin D1, and SRY-box transcription factor 9. By contrast, Pdlim1 overexpression resulted in inhibition of the osteogenic activity of MC3T3-E1 cells. In vivo, at day 3 after fracture,Ad-shPdlim1 adenovirus particles were injected into the fracture site of the femur of mice, and the process of fracture healing was evaluated by X-ray, micro-computed tomography and histological examination. Local injection of Ad-shPdlim1 promoted the early cartilage callus formation, restored bone mineral density, and accelerated cartilaginous ossification, with the upregulation of osteogenic gene (Runx2, Col1A1, OCN, and OPN) expression and activation of ß-catenin signaling. Thus, we concluded that inhibition of Pdlim1 contributed to osteogenesis and fracture healing by activating the ß-catenin signaling pathway.

Combined therapy of local recombinant human interferon α1b injection and acupuncture on verruca vulgaris: a retrospective study.

Verruca vulgaris is always stubborn to treat. We applied a new combined therapy of local recombinant human interferon alpha 1b (rhIFNα1b) injection plus acupuncture on verruca vulgaris recently to evaluate the efficacy and safety of the combined therapy. The retrospective study was conducted in The First Hospital of China Medical University from 2018 to 2020. Patients with verruca vulgaris were included. Combined therapy with local rhIFNα1b injection plus acupuncture was set as treatment group, rhIFNα1b injection and carbon dioxide (CO2) laser were set as control groups. A total of 2415 patients were included in the study. The cure rates were 81.85%, 85.93%, and 100% in combined group, rhIFNα1b group, and CO2 laser group, separately. All lesions cured in combined group were located on hands or feet, while majority of lesions cured in other groups were located on other sites. For patients with medium/big single lesion or 6-9 lesions, less treatment times were needed in combined group than rhIFNα1b group. For patients with small single, two to five or more than ten lesions, the treatment times of combined group and rhIFNα1b group were comparable. All patients complained of pain in varying degrees when local injection or laser irradiation. Compared with CO2 laser group, more fever, less swelling or scar was reported in combined group. In conclusion, combined therapy of local rhIFNα1b plus acupuncture was beneficial for verruca vulgaris with limited adverse effects. The therapy was more acceptable by younger female patients with verruca vulgaris.

Supramolecular salicylic acid ameliorates rosacea-like eruptions by suppressing NLRP3-mediated inflammasome activation in mice.

BACKGROUND: Rosacea is a chronic inflammatory skin disease with immunological dysfunction. Supramolecular salicylic acid (SSA) has the properties of keratolytic, antibacterial, and anti-inflammatory. However, the mechanism of SSA in the treatment of rosacea is still unclear. OBJECTIVE: To investigate the efficiencies and molecular mechanisms of SSA in rosacea. METHODS: Forty mice were randomly divided into four groups (10 in each group): control, LL-37, LL-37 + azelaic acid (AzA), and LL-37 + SSA. Forty µl LL-37 (320 µM) was administered intradermally into the dorsal skin of the mice in the latter 3 groups every 12 h and 4 times altogether (0 h, 12 h, 24 h, 36 h). Twenty % AzA was applied on the eruptions after the first and third LL-37 injection (0 h, 24 h) in LL-37 + AzA group, while 30 % SSA was applied after the first injection (0 h) in LL-37 + SSA group. The redness score and redness area were evaluated. The skin barrier function was measured by the transepidermal water loss (TEWL) and pH. The infiltration of inflammatory cells was evaluated by hematoxylin-eosin staining, and the inflammatory biomarkers were analyzed by RT-PCR and immunohistochemistry. RESULTS: SSA alleviated LL-37-induced rosacea-like inflammation. The increased TEWL and pH induced by LL-37 were also reversed by SSA. In addition, SSA reduced inflammatory cell infiltration and suppressed the production of Toll-like receptor 2, Matrix metallopeptidase 9, kallikrein 5, LL-37 associated with rosacea, and inhibited LL-37-induced NOD-like receptor family, pyrin domain containing 3 (NLRP3)-mediated inflammasome activation in mice. CONCLUSIONS: Our findings indicated that SSA ameliorated LL-37-induced rosacea-like lesions by suppressing NLRP3-mediated inflammasome activation in mice.

Identification of novel candidate genes and predicted miRNAs in atopic dermatitis patients by bioinformatic methods.

Atopic dermatitis (AD) is a common, chronic inflammatory dermatosis with relapsing eruptions. Our study used bioinformatics to find novel candidate differentially expressed genes (DEGs) and predicted miRNAs between AD patients and healthy controls. The Mesh term "atopic dermatitis" was retrieved to obtain DEGs in GEO datasets. DEGs between AD patients and healthy controls were analyzed using GEO2R. Overlapping DEGs between different datasets were obtained with use of Draw Venn software. GO and KEGG enrichment analyses were conducted by the use of DAVID. STRING and miRWalk were used to individually analyze PPI networks, interactions of candidate genes and predicted miRNAs. A total of 571 skin samples, as retrieved from 9 databases were assessed. There were 225 overlapping DEGs between lesioned skin samples of AD patients and that of healthy controls. Nineteen nodes and 160 edges were found in the largest PPI cluster, consisting of 17 up-regulated and 2 down-regulated nodes. Two KEGG pathways were identified, including the cell cycle (CCNB1, CHEK1, BUB1B, MCM5) and p53 (CCNB1, CHEK1, GTSE1) pathways. There were 56 nodes and 100 edges obtained in the miRNA-target gene network, with has-miR-17-5p targeted to 4 genes and has-miR-106b-5p targeted to 3 genes. While these findings will require further verification as achieved with experiments involving in vivo and in vitro modles, these results provided some initial insights into dysfunctional inflammatory and immune responses associated with AD. Such information offers the potential to develop novel therapeutic targets for use in preventing and treating AD.

Activation of aryl hydrocarbon receptor ameliorates rosacea-like eruptions in mice and suppresses the TLR signaling pathway in LL-37-induced HaCaT cells.

BACKGROUND: Toll-like receptor (TLR) 2, along with some chemokines, were found to be overexpressed in rosacea patients. Aryl hydrocarbon Receptor (AhR) activation inhibited the inflammatory responses triggered by TLR activation. The current study was conducted to evaluate the underlying mechanisms of AhR activation in rosacea models. MATERIALS AND METHODS: Seven-week-old female BALB/c mice received twice daily intradermal injections of LL-37 for 2 consecutive days. Thirty minutes after the second LL-37 injection, 1% or 0.5% AhR agonist benvitimod was administrated topically once per day for 3 consecutive days. HaCaT cells were treated with different concentrations of LL-37 and benvitimod, and were further infected with lentivirus to over-express TLR2. Expressions of TLR2, CCL5, CXCL9, CXCL10 and CXCL11 were evaluated using qRT-PCR, Western Blot or ELISA. RESULTS: AhR activation ameliorated LL-37-induced rosacea-like eruptions in mice by reductions in redness scores, redness areas and dermal inflammatory cell infiltrates. Elevated expressions of TLR2 and chemokines (CCL5, CXCL9, CXCL10 and CXCL11) following LL-37 treatment were decreased by AhR activation. In HaCaT cells receiving LL-37, TLR2 and the four chemokines were up-regulated, and levels of these chemokines were further enhanced after over-expressing TLR2. At 8 h after an administration of 10 µM benvitimod, gene expressions of TLR2 and the four chemokines in LL-37 treated HaCat cells were decreased, while their protein expressions were decreased for 24 h. CONCLUSION: AhR activation is beneficial in treating rosacea in a LL-37-induced rosacea mouse model and involves a suppression of the TLR signaling pathway in an HaCaT cell model of rosacea.

Application of benvitimod on psoriasis: A systematic review and meta-analysis of randomized controlled trials. A systematic review of benvitimod on psoriasis.

BACKGROUND: We conducted this systematic review to clarify the efficacy and safety of benvitimod on psoriasis. METHODS: We searched the databases of PubMed, China National Knowledge infrastructure, Cochrane Library, Embase, Web of science to identify randomized controlled trials (RCTs) of benvitimod on psoriasis up to April 2021. RESULTS: Five RCTs of benvitimod on psoriasis were included. A total of 1237 patients were included. 0.5% or 1.0% benvitimod was applied topically once or twice a day. More patients in benvitimod group achieved PASI 90, PASI 75, PASI 50 and BSA reduction than placebo at Week 12. More patients in benvitimod group achieved PGA 0 or 1 than placebo since Week 6. There were no statistical significances in efficacies of benvitimod at different concentrations and frequencies. CONCLUSIONS: Benvitimod was effective and safe for psoriasis. More RCTs with high qualities are needed to further verify the current conclusion.

Identification of novel candidate genes in rosacea by bioinformatic methods.

BACKGROUND: Rosacea is a chronic inflammatory skin disease whose psychological consequences severely affect patient's quality of life. OBJECTIVE: To identify candidate genes of rosacea for potential development of new target therapies. METHODS: Gene Expression Omnibus datasets were retrieved to obtain differentially expressed genes (DEGs) between rosacea patients and healthy controls. Gene ontology (GO) analyses were used to identify functions of candidate genes. Related signaling pathways of DEGs were analyzed using Kyoto Encyclopedia of Genes and Genomes (KEGG) and gene set enrichment analysis. Protein-protein interaction (PPI) networks were applied using search tools for the retrieval of interacting genes/proteins and modulations involving PPI networks were evaluated with use of the MCODE app. RESULTS: Samples from 19 rosacea patients and 10 healthy controls of dataset GSE65914 were enrolled. A total of 215 DEGs, 115 GO terms and 6 KEGG pathways were identified. A total of 182 nodes and 456 edges were enriched in PPI networks. Maximal clusters showed 15 central nodes and 96 edges. The toll-like receptor (TLR) signaling pathway was the most significant pathway detected and 5 DEGs were identified as candidate genes which included TLR2, C-C motif chemokine (CCL) 5, C-X-C motif chemokine ligand (CXCL) 9, CXCL10 and CXCL11. The results were verified in rosacea patients with use of real-time polymerase chain reaction and immunohistochemistry. Cell-type enrichment analysis revealed 8 lymphocytes that were enriched in rosacea patients. CONCLUSIONS: The results suggest that both innate and adaptive immune responses were involved in the etiology of rosacea. Five DEGs in the TLR signaling pathway may serve as potential therapeutic target genes.

A predictive tool for the assessment of right ventricular dysfunction in non-high-risk patients with acute pulmonary embolism.

BACKGROUND: Rapid and accurate identification of right ventricular (RV) dysfunction is essential for decreasing mortality associated with acute pulmonary embolism (PE), particularly for non-high-risk patients without hypotension on admission. This study aimed to develop a rapid and accurate tool for predicting the risk of RV dysfunction in non-high-risk patients with acute PE. METHODS: The medical records of non-high-risk patients with acute PE admitted to Shengjing Hospital of China Medical University between January 2011 and May 2020 were retrospectively analysed. The primary outcome of this study was RV dysfunction within 24 h after admission. The enrolled patients were randomized into training or validation sets as a ratio of 2:1. In the training set, a nomogram was developed, and the consistency was corroborated in the validation set. The areas under the receiver operating characteristic curves (AUCs) and 95% confidence intervals (CIs) were calculated. RESULTS: A total of 845 patients were enrolled, including 420 men and 425 women with an average age of 60.05 ± 15.43 years. Right ventricular dysfunction was identified in 240 patients (28.40%). The nomogram for RV dysfunction included N-terminal pro-brain natriuretic peptide, cardiac troponin I, and ventricular diameter ratios, which provided AUC values of 0.881 in the training dataset (95% confidence interval (CI): 0.868-0.898, p < 0.001) and 0.839 in the validation set (95% CI: 0.780-0.897, p < 0.001). The predictive tool was published as a web-based calculato ( https://gaoyzcmu.shinyapps.io/APERVD/ ). CONCLUSIONS: The combination of CT and laboratory parameters forms a predictive tool that may facilitate the identification of RV dysfunction in non-high-risk patients with acute PE.

Antimicrobial photodynamic therapy in skin wound healing: A systematic review of animal studies.

Bacterial infection is a common wound complication that can significantly delay healing. Classical local therapies for infected wounds are expensive and are frequently ineffective. One alternative therapy is photodynamic therapy (PDT). We conducted a systematic review to clarify whether PDT is useful for bacteria-infected wounds in animal models. PubMed and Medline were searched for articles on PDT in infected skin wounds in animals. The language was limited to English. Nineteen articles met the inclusion criteria. The overall study methodological quality was moderate, with a low-moderate risk of bias. The animal models were mice and rats. The wounds were excisional, burn, and abrasion wounds. Wound size ranged from 6 mm in diameter to 1.5 × 1.5 cm2 . Most studies inoculated the wounds with Pseudomonas aeruginosa or methicillin-resistant Staphylococcus aureus. Eleven and 17 studies showed that the PDT of infected wounds significantly decreased wound size and bacterial counts, respectively. Six, four, and two studies examined the effect of PDT on infected wound-cytokine levels, wound-healing time, and body weight, respectively. Most indicated that PDT had beneficial effects on these variables. PDT accelerated bacteria-infected wound healing in animals by promoting wound closure and killing bacteria.

TLR3 is involved in paraquat-induced acute renal injury.

AIMS: To investigate the role of Toll-like receptor 3 (TLR3) in mouse paraquat-induced acute renal injury. MATERIALS AND METHODS: Acute renal injury was established in C57BL/6J mice by intraperitoneal injection of paraquat (28 mg/kg). The mice were also injected intraperitoneally with TLR3 agonist poly I:C (20 mg/kg) or TLR3/dsRNA complex inhibitor (1 mg) 1 h before paraquat exposure. At 72 hour post paraquat exposure, the mice were sacrificed and the blood and renal tissues were collected to examine TLR3 expression in renal tissues, pathological injury in renal tissues, renal function, inflammation, and cell apoptosis. KEY FINDINGS: After paraquat exposure, TLR3 expression in mouse renal tissues was significantly increased, and pathological changes to the renal tissues and remarkable renal impairment were present. Compared to the paraquat group, the poly I:C group showed no significant difference in renal pathology, renal function, inflammation, or cell apoptosis. However, TLR3 inhibitor treatment significantly alleviated injury to the renal tissues, improved renal function, inhibited NF-κB activation, suppressed the infiltration of neutrophils, and lessened the expression of IL-1ß, TNF-α, and keratinocyte chemoattractant (KC) in renal tissues. TLR3 inhibitor treatment also suppressed the activation of caspase-8 and caspase-3 and reduced apoptosis in the renal tissues. SIGNIFICANCE: Paraquat exposure significantly upregulates TLR3 expression in renal tissues, and activation of the TLR3 signaling pathway is an important contributor to paraquat nephrotoxicity. TLR3 activation exacerbates inflammation and cell apoptosis in renal tissues by activating NF-κB and caspase-8, thus promoting paraquat-induced acute renal injury.

Topical photodynamic therapy with 5-aminolevulinic acid in Chinese patients with Rosacea.

Background: Rosacea is difficult to cure and frequently recurs. Topical photodynamic therapy (PDT) has been tentatively used, with only preliminary results reported. Objective: To evaluate the efficacy and safety of topical PDT in Chinese patients with rosacea. Methods & Materials: Seventeen participants with rosacea were treated three times using 5-aminolevulinic acid (ALA)-PDT at intervals of 7-10 days. Papule and pustule numbers, erythema severity, telangiectasia severity, physician's global assessment (PGA) score (1 [best]-6), and patient satisfaction score (0-3 [highest]) were assessed. Rosacea improvement and the total effective rate were calculated. Stratum corneum hydration and sebum levels, and the melanin index (MI) and erythema index (EI) were measured non-invasively. Results: After three treatments with ALA-PDT, the total effective rate (≥50% improvement) was 64.71%, mean PGA score was 2.88 ± 0.93, and mean patient satisfaction score was 1.71 ± 0.69. The EI significantly decreased 1 month after the final treatment (from 468 ± 80.61 to 439 ± 77.78 for the forehead and from 507.65 ± 92.51 to 483.27 ± 78.32 for the nasal ala). Four participants received three additional treatments. They achieved 50-74% improvement after three treatments and ≥75% improvement after six treatments. Conclusion: ALA-PDT is safe and effective for treating rosacea.

MyD88 gene knockout attenuates paraquat-induced acute lung injury.

OBJECTIVE: This study investigated the role of myeloid differentiation factor 88 (MyD88) in paraquat-induced acute lung injury (ALI). METHODS: C57BL mice were divided into the control group, paraquat group, MyD88 knockout (KO) group, and MyD88 KO plus paraquat group. At 48h after paraquat poisoning, serum and lung tissues were collected. ELISA was employed to detect tumor necrosis factor-α (TNF-α) and interleukine-1ß (IL-1ß) contents in serum. Lung tissues were processed for hematoxylin-eosin staining, followed by histological scoring. PCR was performed to detect the mRNA expression of MyD88, TNF-α, and IL-1ß in the lungs. Immunofluorescence staining was done to evaluate the expression and distribution of MyD88 and nuclear factor κB (NF-κB) in the lungs. Western blotting was conducted to detect the protein level of toll-like receptor (TLR) 4, TLR9, MyD88, and NF-κB in the lungs. RESULTS: Paraquat poisoning significantly increased serum inflammatory cytokines, as well as MyD88, TLR4, TLR9, and NF-κB, and resulted in ALI. After MyD88 KO, the levels of inflammatory cytokines and NF-κB decreased markedly, and ALI was also attenuated although TLR4 and TLR9 expression continued at an elevated level. CONCLUSION: MyD88 mediates paraquat-induced ALI, and MyD88 gene knockout may attenuate paraquat-induced ALI and reduce the production of proinflammatory cytokines.

[Molecular genetic analysis for the A3 alleles].

To study four A(3) subgroup samples identified by serologic tests, among which two belong to a family, three were A(3) subgroup, one was A(3)B subgroup. All four samples were genotyped by PCR-SSP method, and the nucleotide sequences of Exon 6, Exon 7 and part introns at the ABO locus for these samples were detected by ABI Prism 3100 DNA sequencer. Comparison with the consensus of A101 was performed. The results showed that haplotypes of two A(3) subgroups were common A102 allele and O1-2 allele, and haplotypes of one A(3) subgroup were common A102 allele and rare O(1v)-4 allele. Unexpectedly, a synonymous substitution 838C-->T had been found in A allele of the A(3)B subgroup sample, which predict a Leu280Phe alteration. The results suggested that molecular genetic background of the A(3) phenotypes is polymorphic. Possibly, the missense mutation 838C-->T is the molecular genetic basis of A(3)B subgroup that lead to low activity of the glycosyltransferases.