Least Absolute Shrinkage And Selection Operator Analysis Research Articles

Radiomic-Based MRI for Classification of Solitary Brain Metastases Subtypes From Primary Lymphoma of the Central Nervous System.

Differential diagnosis of brain metastases subtype and primary central nervous system lymphoma (PCNSL) is necessary for treatment decisions. The application of machine learning facilitates the classification of brain tumors, but prior investigations into primary lymphoma and brain metastases subtype classification have been limited. To develop a machine-learning model to classify PCNSL, brain metastases with primary lung and non-lung origin. Retrospective. A total of 211 subjects with pathologically confirmed PCNSL or brain metastases (training cohort 168 and testing cohort 43). A 3.0 T axial contrast-enhanced T1-weighted spin-echo inversion recovery sequence (T1WI-CE), axial T2-weighted fluid-attenuation inversion recovery sequence (T2FLAIR) ASSESSMENT: Several machine-learning models (support vector machine, random forest, and K-nearest neighbors) were built with least absolute shrinkage and selection operator (LASSO) using features from T1WI-CE, T2FLAIR, and clinical. The model with the highest performance in the training cohort was selected to differentiate lesions in the testing cohort. Then, three radiologists conducted a two-round classification (with and without model reference) using images and clinical information from testing cohorts. Five-fold cross-validation was used for model evaluation and calibration. Model performance was assessed based on sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC). Twenty-five image features were selected by LASSO analysis. Random forest classifier was selected for its highest performance on the training set with an AUC of 0.73. After calibration, this model achieved an accuracy of 0.70 on the testing set. Accuracies of all three radiologists improved under model reference (0.49 vs. 0.70, 0.60 vs. 0.77, 0.58 vs. 0.72, respectively). The random forest model based on conventional MRI and clinical data can diagnose PCNSL and brain metastases subtypes (lung and non-lung origin). Model classification can help foster the diagnostic accuracy of specialists and streamline prognostication workflow. 4 TECHNICAL EFFICACY: Stage 2.

Establishment of a predictive model for purulent meningitis in preterm infants.

BackgroundPurulent meningitis (PM) is an important cause of mortality and morbidity in the newborn population throughout the world. The subtle of specific clinical signs and low success rates of lumbar puncture make diagnosis of PM more difficult in preterm than in older children. The objective of this study was to establish a predict model for preterm PM in hopes of helping clinicians develop new diagnostic and treatment strategies.MethodsPremature infants who were admitted to The First Affiliated Hospital of Zhengzhou University from September 2017 to March 2020 were enrolled in this study. All the patients underwent lumbar puncture. We collected data encompassing maternal diseases and neonatal clinical features. Cerebrospinal fluid (CSF) culture is the gold standard for diagnosing meningitis. The PM was diagnosed according to the diagnostic criteria. All statistical analyses were performed using R 3.63 (https://www.r-project.org/). Logistic regression and least absolute shrinkage and selection operator (LASSO) regression analyses were used to establish a risk prediction model of PM. The Brier score, calibration slope, and concordance (C)-index were used to verify the accuracy of prediction model.ResultsA total of 168 preterm infants were enrolled in this study, 80 boys and 88 girls, the gestational age (GA) was 26.43–36.86 weeks (32.45±2.79 weeks), the birth weight (BW) was 700–3,400 g (1,814.05±568.84 g). There were 77 preterm infants with PM while 91 without. We identified seven variables as independent risk factors for PM in preterm infants by LASSO analysis [the optimal λ was 0.080960, and log(λ) = −2.5138], including procalcitonin (PCT) on the 1st day after birth, prenatal glucocorticoid use, albumin, the 1-minute Apgar score, the use of non-invasive biphasic positive airway pressure, hemoglobin, and sex. These were used to construct a risk prediction nomogram and verified its accuracy. The Brier score was 0.17, the calibration slope was 0.966, and the concordance index was 0.82018.ConclusionsOur prediction model could predict the risk of PM in preterm infants. Using this prediction model, it may be able to provide reference to determine whether lumbar puncture is performed and whether antibiotics are applied as soon as possible.

Novel combination artificial neural network models could not outperform individual models for weather-based cashew yield prediction.

Cashew is an important cash crop which is ecologically sensitive, making it vulnerable to climate change. So, the present study compares the performance of stepwise linear regression (SLR), least absolute shrinkage and selection operator (LASSO), elastic net, and artificial neural network (ANN) individually against the ANN model combined with SLR, LASSO, elastic net, and principal components analysis (PCA) for prediction of cashew yield based on weather parameters. The model performances were evaluated using three approaches: (1) Taylor plot; (2) statistical metrics like coefficient of determination (R2), root mean square error (RMSE), and normalized RMSE (nRMSE); and (3) ranking followed by Kruskal-Wallis and Dunn's post hoc test. The results revealed that during calibration, the R2 and RMSE ranged from 0.486 to 0.999 and 2.184 to 88.040kgha-1, respectively, while RMSE and nRMSE varied from 3.561 to 242.704kgha-1 and 0.799 to 89.949%, respectively, during validation. Kruskal-Wallis and Dunn's post hoc test revealed LASSO as the best model which was at par with ELNET, SLR, and ELNET-ANN. So, these models can be used for cashew yield prediction for the study area well in advance.

Immune Response in Vitamin D Deficient Metastatic Colorectal Cancer Patients: A Player That Should Be Considered for Targeted Vitamin D Supplementation.

Simple SummaryIn this study, we investigated the role of vitamin D levels and the impact on immune response in patients with mCRC and propose a vitamin D cut-off level. Among several hematological, biochemical and immunological variables, we identified the neutrophils-to-lymphocytes ratio, CD4+ T lymphocytes and B lymphocytes as being closely related to vitamin D status with an impact on survival.Background: Vitamin D deficiency is a poor prognostic factor in metastatic colorectal cancer (mCRC); however, targeted supplementation trials have so far yielded limited results. We investigated clinical-laboratory parameters influencing vitamin D deficiency, with a particular focus on immune response, and the effect on survival. These parameters could help optimize targeted supplementation therapy. Methods: Association of plasma 25-hydroxyvitamin D (25(OH])D) with overall survival (OS) was assessed with the Hazard Ratio Smoothed Curve with Restricted Cubic Splines (HRSC-RCS) and maximally selected rank statistics (MSRS) in mCRC patients who underwent first-line chemotherapy. Several hematobiochemical variables were evaluated as predictors of vitamin D deficiency by means of Least Absolute Shrinkage and Selection Operator (LASSO) analysis. In a patient subset, peripheral lymphocyte subpopulations were also analyzed. Results: One hundred thirty-three mCRC patients were included. The median(m) baseline 25(OH)D was 10.8 ng/mL (range 3–53.4). HRSC-RCS revealed a linear association between 25(OH)D and OS. MSRS found 10 ng/mL as the optimal 25(OH)D cut-off. The median OS for 25(OH)D < 10 (n = 60) vs. > 10 ng/mL (n = 73) was 12.3 and 24.5 months, respectively (p = 0.002). The LASSO analysis identified high neutrophil-to-lymphocyte ratio (NLR > 3.5) as the strongest predictor of vitamin D deficiency (Odds Ratio 3.35, p 0.0009). Moreover, patients with low 25(OH)D levels (< 10 ng/mL) and high NLR (>3.5) had the shortest survival and patients with 25(OH)D >10 ng/mL and NLR <3.5 had the longest: mOS 8.1 and 28.1 months, respectively, HR 3.40 (1.76–6.59), p 0.0004. Besides the significant difference in NLR between 25(OH)D < and > 10 ng/mL patients (mNLR 3.6 vs. 2.9, p 0.03), the lymphocyte subpopulation analysis revealed that vitamin D deficiency was associated with high T- CD4+ (p = 0.04) and low B (p = 0.03) lymphocyte frequency. Conclusions: NLR is a powerful predictor of Vitamin D deficiency and can further help in stratifying prognosis. Vitamin D deficiency was associated with significant variations in peripheral immune cells. We hypothesize that integrated targeted interventions to both vitamin D and immune system would improve the prognosis of mCRC patients.

Serum metabolomic profiles associated with subclinical and clinical cardiovascular phenotypes in people with type 2 diabetes

BackgroundAtherosclerotic cardiovascular diseases (CVD) is the leading cause of death in diabetes, but the full range of biomarkers reflecting atherosclerotic burden and CVD risk in people with diabetes is unknown. Metabolomics may help identify novel biomarkers potentially involved in development of atherosclerosis. We investigated the serum metabolomic profile of subclinical atherosclerosis, measured using ankle brachial index (ABI), in people with type 2 diabetes, compared with the profile for symptomatic CVD in the same population.MethodsThe Edinburgh Type 2 Diabetes Study is a cohort of 1,066 individuals with type 2 diabetes. ABI was measured at baseline, years 4 and 10, with cardiovascular events assessed at baseline and during 10 years of follow-up. A panel of 228 metabolites was measured at baseline using nuclear magnetic resonance spectrometry, and their association with both ABI and prevalent CVD was explored using univariate regression models and least absolute shrinkage and selection operator (LASSO). Metabolites associated with baseline ABI were further explored for association with follow-up ABI and incident CVD.ResultsMean (standard deviation, SD) ABI at baseline was 0.97 (0.18, N = 1025), and prevalence of CVD was 35.0%. During 10-year follow-up, mean (SD) change in ABI was + 0.006 (0.178, n = 436), and 257 CVD events occurred. Lactate, glycerol, creatinine and glycoprotein acetyls levels were associated with baseline ABI in both univariate regression [βs (95% confidence interval, CI) ranged from − 0.025 (− 0.036, − 0.015) to − 0.023 (− 0.034, − 0.013), all p < 0.0002] and LASSO analysis. The associations remained nominally significant after adjustment for major vascular risk factors. In prospective analyses, lactate was nominally associated with ABI measured at years 4 and 10 after adjustment for baseline ABI. The four ABI-associated metabolites were all positively associated with prevalent CVD [odds ratios (ORs) ranged from 1.29 (1.13, 1.47) to 1.49 (1.29, 1.74), all p < 0.0002], and they were also positively associated with incident CVD [ORs (95% CI) ranged from 1.19 (1.02, 1.39) to 1.35 (1.17, 1.56), all p < 0.05].ConclusionsSerum metabolites relating to glycolysis, fluid balance and inflammation were independently associated with both a marker of subclinical atherosclerosis and with symptomatic CVD in people with type 2 diabetes. Additional investigation is warranted to determine their roles as possible etiological and/or predictive biomarkers for atherosclerotic CVD.

The Predictive Value of the Perivascular Adipose Tissue CT Fat Attenuation Index for Coronary In-stent Restenosis.

ObjectivesTo investigate the association between the perivascular adipose tissue (PVAT) fat attenuation index (FAI) derived from coronary computed tomography angiography (CCTA) and the prevalence of in-stent restenosis (ISR) in patients with coronary stent implantation.MethodsA total of 117 patients with previous coronary stenting referred for invasive coronary angiography (ICA) were enrolled in this retrospective observational analysis. All patients underwent CCTA between July 2016 and November 2021. The deep learning-based (DL-based) method was used to analyze and measure the peri-stent FAI value. Additionally, the relationship between hematological and biochemical parameters collected from all the patients was also explored. The least absolute shrinkage and selection operator (LASSO) method was applied to the most useful feature selection, and binary logistic regression was used to test the association between the selected features and ISR. The predictive performance for ISR of the identified subgroups was evaluated by calculating the area under the curve (AUC) of receiver operator curves plotted for each model. The Pearson correlation coefficient was used to assess the correlation of peri-stent FAI values with degrees of ISR.ResultsThe peri-stent FAI values in the ISR group were significantly higher than those in the non-ISR group (−78.1 ± 6.2 HU vs. −87.2 ± 7.3 HU, p < 0.001). The predictive ISR features based on the LASSO analysis were peri-stent FAI, high-density lipoprotein cholesterol (HDL-C), apolipoprotein A1 (ApoA1), and high-sensitivity c-reactive protein (hs-CRP), with an AUC of 0.849, 0.632, 0.620, and 0.569, respectively. Binary logistic regression analysis determined that peri-stent FAI was uniquely and independently associated with ISR after adjusting for other risk factors (odds ratio [OR] 1.403; 95% CI: 1.211 to 1.625; p < 0.001). In the subgroup analysis, the AUCs of the left anterior descending coronary artery (LAD), left circumflex coronary artery (LCx), and right coronary artery (RCA) stents groups were 0.80, 0.87, and 0.96, respectively. The Pearson's correlation coefficient indicated a term moderately correlation between ISR severity and peri-stent FAI values (r = 0.579, P < 0.001).ConclusionsThe peri-stent FAI can be used as an independently non-invasive biomarker to predict ISR risk and severity after stent implantation.

Machine Learning Based on Diffusion Kurtosis Imaging Histogram Parameters for Glioma Grading.

Glioma grading plays an important role in surgical resection. We investigated the ability of different feature reduction methods in support vector machine (SVM)-based diffusion kurtosis imaging (DKI) histogram parameters to distinguish glioma grades. A total of 161 glioma patients who underwent magnetic resonance imaging (MRI) from January 2017 to January 2020 were included retrospectively. The patients were divided into low-grade (n = 61) and high-grade (n = 100) groups. Parametric DKI maps were derived, and 45 features from the DKI maps were extracted semi-automatically for analysis. Three feature selection methods [principal component analysis (PCA), recursive feature elimination (RFE) and least absolute shrinkage and selection operator (LASSO)] were used to establish the glioma grading model with an SVM classifier. To evaluate the performance of SVM models, the receiver operating characteristic (ROC) curves of SVM models for distinguishing glioma grades were compared with those of conventional statistical methods. The conventional ROC analysis showed that mean diffusivity (MD) variance, MD skewness and mean kurtosis (MK) C50 could effectively distinguish glioma grades, particularly MD variance. The highest classification distinguishing AUC was found using LASSO at 0.904 ± 0.069. In comparison, classification AUC by PCA was 0.866 ± 0.061, and 0.899 ± 0.079 by RFE. The SVM-PCA model with the lowest AUC among the SVM models was significantly better than the conventional ROC analysis (z = 1.947, p = 0.013). These findings demonstrate the superiority of DKI histogram parameters by LASSO analysis and SVM for distinguishing glioma grades.

Analysis of Related Risk Factors of Microvascular Invasion in Hepatocellular Carcinoma.

Objective To forecast the onset of microvascular invasion (MVI) in patients with hepatoma by evaluating the preoperative aspartate aminotransferase-to-platelet ratio index (APRI), alpha-fetoprotein (AFP), neutrophil-to-lymphocyte ratio (NLR), and other clinicopathological data. Methods In this study, we retrospectively analysed the clinical data of 62 patients who received radical surgery for hepa toma from 2019 to 2021. Patients were separated into the MVI-negative group and the MVI-positive group according to the postoperative pathological diagnosis. The relationships between MVI and NLR, APRI, AFP, tumor size, and other clinical data were assessed using the univariate analysis, receiver operating characteristic (ROC) curve, least absolute shrinkage and selection operator (LASSO) analysis, and logistic analysis. Results The ROC curve determined that the cutoff values of NLR, platelet-to-lymphocyte ratio (PLR), and APRI were 1.520, 98, and 0.275, respectively. The univariate analysis showed that the MVI-positive result was associated with five factors: tumor size (χ2 = 10.620, p = 0.001), AFP (χ2 = 10.524, p = 0.001), Edmondson grade (χ2 = 20.736, p < 0.001), NLR (χ2 = 8.744, p = 0.003), and APRI (χ2 = 4.849, p = 0.028). The LASSO analysis indicated that the risk factors were the number of tumors, PLR, APRI, NLR, AFP, Edmondson grade, and tumor size. The multivariate logistic regression analysis showed that NLR ≥ 1.520 (OR 11.119, p = 0.006), APRI ≥ 0.275 (OR 12.515, p = 0.009), AFP ≥ 200 μg/mL (OR 7.823, p = 0.016), and tumor size > 3 cm (OR 7.689, p = 0.022) were independent risk factors for MVI in patients with hepatoma. Conclusion Preoperative NLR, APRI, AFP, and tumor size are reliable indicators for predicting the appearance of MVI in patients with hepatoma and are of great value in making detailed and reliable treatment protocols for these patients before surgery.

Identifying the hub gene and immune infiltration of Parkinson’s disease using bioinformatical methods

BackgroundParkinson’s disease (PD) is a common neurodegenerative disorder that affects 1%–2% of the population over 60 years old. Immune response dysfunction in the brain contributes to the occurrence and development of PD. This study aimed to uncover the potential diagnostic genes for PD and characterize the immune cell infiltrates. MethodsWe downloaded the microarray data of patients with PD samples from the Gene Expression Omnibus (GEO) database. Weighted Gene Co-Expression Network Analysis (WGCNA) was used to identify the modules linked to PD in the GSE20163 dataset. Meanwhile, differentially expressed genes (DEGs) between the healthy control samples and PD samples were also identified. Then the PD-related genes were integrated based on the genes in the key module and DEGs. Functional enrichment analysis was used to explore the molecular mechanisms of these PD-related genes. Protein-protein interaction (PPI) network and least absolute shrinkage and selection operator (LASSO) analysis were used to further screen candidate genes for PD. Gene set enrichment analysis (GSEA) was applied to explore the biological functions of these candidate genes. The infiltration of immune cells was detected by single-sample gene set enrichment analysis (ssGSEA) algorithm in the GSE20163 dataset, and Pearson analysis was used to investigate the correlation of candidate genes with immune cells and immune checkpoint proteins. The expression of candidate genes in clinical samples was verified by qPCR. ResultsAltogether, we found a unique gene module related to PD, where 109 DEGs were identified in the GSE20163 dataset. Following these results, we screened 68 genes associated with PD. Gene Expression Omnibus (GEO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses suggested that these genes were markedly enriched in the pathway of synthesis and transport of neurons. Three candidate genes (SLC18A2, CALB1, and SYNGR3) were further identified in PD patients through PPI network and LASSO analysis. The receiver operating characteristic (ROC) curve indicated that the three candidate genes had a good performance in distinguishing the PD samples from healthy control samples. The proportions of the aDC, DC, NK CD56dim cells, and follicular helper T cells (TFH) were obviously different between the healthy control and PD samples. Moreover, CTLA4, LAG3, CEACAM1, and CD27 were highly expressed in the PD group. GSEA analysis for candidate genes revealed that they were all closely related to the neurogenic disease. Additionally, the three candidate genes were all strongly correlated with the above immune cells and immune checkpoint proteins. The qPCR results validated the expression differences of SLC18A2 and SYNGR3 in the clinical PD and control samples. ConclusionThe three candidate genes may be a useful tool for diagnosing PD patients. These findings provide a reference for exploring new therapeutic targets and strategies for PD treatment.

Radiomics analysis of [18F]-fluoro-2-deoxyglucose positron emission tomography for the prediction of cervical lymph node metastasis in tongue squamous cell carcinoma.

This study aimed to create a predictive model for cervical lymph node metastasis (CLNM) in patients with tongue squamous cell carcinoma (SCC) based on radiomics features detected by [18F]-fluoro-2-deoxyglucose (18F-FDG) positron emission tomography (PET). A total of 40 patients with tongue SCC who underwent 18F-FDG PET imaging during their first medical examination were enrolled. During the follow-up period (mean 28months), 20 patients had CLNM, including six with late CLNM, whereas the remaining 20 patients did not have CLNM. Radiomics features were extracted from 18F-FDG PET images of all patients irrespective of metal artifact, and clinicopathological factors were obtained from the medical records. Late CLNM was defined as the CLNM that occurred after major treatment. The least absolute shrinkage and selection operator (LASSO) model was used for radiomics feature selection and sequential data fitting. The receiver operating characteristic curve analysis was used to assess the predictive performance of the 18F-FDG PET-based model and clinicopathological factors model (CFM) for CLNM. Six radiomics features were selected from LASSO analysis. The average values of the area under the curve (AUC), accuracy, sensitivity, and specificity of radiomics analysis for predicting CLNM from 18F-FDG PET images were 0.79, 0.68, 0.65, and 0.70, respectively. In contrast, those of the CFM were 0.54, 0.60, 0.60, and 0.60, respectively. The 18F-FDG PET-based model showed significantly higher AUC than that of the CFM. The 18F-FDG PET-based model has better potential for diagnosing CLNM and predicting late CLNM in patients with tongue SCC than the CFM.

Serum Inflammatory Biomarkers Contribute to the Prognosis Prediction in High-Grade Glioma.

BackgroundTo evaluate the prognostic value of serum inflammatory biomarkers and develop a risk stratification model for high-grade glioma (HGG) patients based on clinical, laboratory, radiological, and pathological factors.Materials and MethodsA retrospective study of 199 patients with HGG was conducted. Patients were divided into a training cohort (n = 120) and a validation cohort (n = 79). The effects of potential associated factors on the overall survival (OS) time were investigated and the benefits of serum inflammatory biomarkers in improving predictive performance was assessed. Univariable and multivariable Cox regression analyses, the least absolute shrinkage and selection operator (LASSO) regression analysis, and support vector machines (SVM) were used to select variables for the final nomogram model.ResultsAfter multivariable Cox, LASSO, and SVM analysis, in addition to 3 other clinico-pathologic factors, platelet-to-lymphocyte ratio (PLR) >144.4 (hazard ratio [HR], 2.05; 95% confidence interval [CI], 1.25–3.38; P = 0.005) were left for constructing the predictive model. The model with PLR exhibited a better predictive performance than that without them in both cohorts. The nomogram based on the model showed an excellent ability of discrimination in the entire cohort (C-index, 0.747; 95%CI, 0.706–0.788). The calibration curves showed good consistency between the predicted and observed survival probability.ConclusionOur study confirmed the prognostic value of serum inflammatory biomarkers including PLR and established a comprehensive scoring system for the OS prediction in HGG patients.

Identification of solute carrier family genes related to the prognosis and tumor-infiltrating immune cells of pancreatic ductal adenocarcinoma.

BackgroundPancreatic ductal adenocarcinoma (PDAC) has persisted as one of the worst prognostic tumors with a 5-year survival rate of lower than 6%. Although many studies have investigated PDAC, new biomarkers are required to ensure early diagnosis and predict the prognosis of PDAC.MethodsIn this study, we used bioinformatics methods to evaluate differences in the expression of solute carrier (SLC) family genes in tumors and non-tumors. A Kaplan-Meier analysis, least absolute shrinkage and selection operator (LASSO) analysis, and multivariate Cox proportional hazards regression analysis were used to evaluate the relationship between SLC genes and prognosis using The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) datasets. The prognostic signature was constructed depending on the risk score to assess the impact of multiple genes on the prognosis, receiver operating characteristic (ROC) curves and forest plot was constructed to assess the ability to predict the prognosis and effects of clinical variables in both high- and low-risk groups. Tumor-infiltrating immune cells were evaluated using Cell-type Identification by Estimating Relative Subsets of RNA Transcripts (CIBERSORT) in both high- and low-risk groups.ResultsIn 32 SLC genes, 9 were significantly associated with the OS after LASSO analysis. SLC19A3 (P=0.007), SLC25A39 (P=0.027), SLC39A11 (P=0.043) were significantly associated with prognosis and included into the prognostic model. CIBERSORT demonstrated that memory B cells (P=0.004), naive B cells (P=0.007), CD8 T cells (P=0.003), activated memory CD4 T cells (P=0.004), and activated NK cells (P=0.019) were significantly higher in the low-risk group. Gene set enrichment analysis (GSEA) showed that potential molecular mechanisms enriched in MYC and p53 signaling pathways.ConclusionsSLC19A3, SLC25A35, and SLC39A11 were significantly relative to the prognosis of PDAC and changed the tumor microenvironment, as well as the MYC and p53 signaling pathways. The SLC19A3 gene may represent a new tumor suppressor in PDAC.

Identification of Immune Cells and Key Genes associated with Alzheimer's Disease.

Alzheimer's disease (AD) is an age-related neurodegenerative disorder characterized by cognitive impairment and memory loss, for which there is no effective cure to date. In the past several years, numerous studies have shown that increased inflammation in AD is a major cause of cognitive impairment. This study aimed to reveal 22 kinds of peripheral immune cell types and key genes associated with AD. The prefrontal cortex transcriptomic data from Gene Expression Omnibus (GEO) database were collected, and CIBERSORT was used to assess the composition of 22 kinds of immune cells in all samples. Weighted gene co-expression network analysis (WGCNA) was used to construct gene co-expression networks and identified candidate module genes associated with AD. The least absolute shrinkage and selection operator (LASSO) and random forest (RF) models were constructed to analyze candidate module genes, which were selected from the result of WGCNA. The results showed that the immune infiltration in the prefrontal cortex of AD patients was different from healthy samples. Of all 22 kinds of immune cells, M1 macrophages were the most relevant cell type to AD. We revealed 10 key genes associated with AD and M1 macrophages by LASSO and RF analysis, including ARMCX5, EDN3, GPR174, MRPL23, RAET1E, ROD1, TRAF1, WNT7B, OR4K2 and ZNF543. We verified these 10 genes by logistic regression and k-fold cross-validation. We also validated the key genes in an independent dataset, and found GPR174, TRAF1, ROD1, RAET1E, OR4K2, MRPL23, ARMCX5 and EDN3 were significantly different between the AD and healthy controls. Moreover, in the 5XFAD transgenic mice, the differential expression trends of Wnt7b, Gpr174, Ptbp3, Mrpl23, Armcx5 and Raet1e are consistent with them in independent dataset. Our results provided potential therapeutic targets for AD patients.

The ratio of the max-to-mean coronary artery calcium score in the most calcified vessel is associated with the presence of coronary artery disease

PurposeThis study aimed to clarify the relationship between the severity of the calcium burden in the most calcified coronary vessel and coronary artery disease (CAD). MethodOf 2150 patients, 376 examined by both coronary computed tomographic angiography and invasive coronary angiography (ICA) within 30 days at Sun Yat-sen Memorial Hospital between March 2011 and July 2020 were included. Three coronary artery calcium scores (CACSs), including the Agatston score, volume score, and mass score, and other clinical variables were recorded. The ratio of max-to-mean CACS in the most calcified vessel (CACSmax:mean) was defined as the CACS in the most calcified vessel/average CACS of the four major epicardial coronary arteries. Logistic regression and least absolute shrinkage and selection operator (LASSO) analyses were performed to assess the relationship between CACSmax:mean and CAD. ResultsCACSmax:mean was higher in 81.1% of subjects diagnosed with CAD than in subjects without CAD. In multivariate logistic regression analysis, CACSmax:mean determined by the Agatston score, volumetric score, and mass score was associated with CAD. In LASSO analysis, Agatston scoremax:mean (not the total Agatston score or other CACSmax:mean) had the strongest correlation with CAD (β = 0.125). AUCs in the training set and the validation set were 0.811 and 0.789, respectively. Increased age, diabetes and hypertension correlated with higher Agatston scoremax:mean. ConclusionsIn addition to total CACS, CACSmax:mean may be a novel diagnostic parameter for CAD, showing the calcium burden severity.

Robust Rank Aggregation and Least Absolute Shrinkage and Selection Operator Analysis of Novel Gene Signatures in Dilated Cardiomyopathy

Objective: Dilated cardiomyopathy (DCM) is a heart disease with high mortality characterized by progressive cardiac dilation and myocardial contractility reduction. The molecular signature of dilated cardiomyopathy remains to be defined. Hence, seeking potential biomarkers and therapeutic of DCM is urgent and necessary.Methods: In this study, we utilized the Robust Rank Aggregation (RRA) method to integrate four eligible DCM microarray datasets from the GEO and identified a set of significant differentially expressed genes (DEGs) between dilated cardiomyopathy and non-heart failure. Moreover, LASSO analysis was carried out to clarify the diagnostic and DCM clinical features of these genes and identify dilated cardiomyopathy derived diagnostic signatures (DCMDDS).Results: A total of 117 DEGs were identified across the four microarrays. Furthermore, GO analysis demonstrated that these DEGs were mainly enriched in the regulation of inflammatory response, the humoral immune response, the regulation of blood pressure and collagen–containing extracellular matrix. In addition, KEGG analysis revealed that DEGs were mainly enriched in diverse infected signaling pathways. Moreover, Gene set enrichment analysis revealed that immune and inflammatory biological processes such as adaptive immune response, cellular response to interferon and cardiac muscle contraction, dilated cardiomyopathy are significantly enriched in DCM. Moreover, Least absolute shrinkage and selection operator (LASSO) analyses of the 18 DCM-related genes developed a 7-gene signature predictive of DCM. This signature included ANKRD1, COL1A1, MYH6, PERELP, PRKACA, CDKN1A, and OMD. Interestingly, five of these seven genes have a correlation with left ventricular ejection fraction (LVEF) in DCM patients.Conclusion: Our present study demonstrated that the signatures could be robust tools for predicting DCM in clinical practice. And may also be potential treatment targets for clinical implication in the future.

Machine Learning-Based Ultrasomics for Predicting Subacromial Impingement Syndrome Stages.

To determine the performance of machine learning (ML)-based ultrasomic analysis of subacromial impingement syndrome (SIS) stage evaluation. In this retrospective study, 324 patients with SIS were included. The SIS stage was evaluated with a Neer test. Regions of the musculi supraspinatus were manually segmented by an experienced radiologist. Then, 5936 ultrasomic features were extracted from the Ultrasomics Platform software. The Wilcoxon test was used to identify differentially expressed radiomic features. Then, these differentially expressed features were submitted to the least absolute shrinkage and selection operator (LASSO) for model construction. The area under the curve (AUC) of the receiver operating characteristic was used to evaluate the performance of the ultrasonic model for SIS stage evaluation. Finally, a total of 223 early-stage and 101 advanced-stage SIS patients were randomly divided into a training cohort (n=227) and a validation cohort (n=97). After feature-dimensionality reduction, a total of 28 radiomic features were submitted to LASSO analysis. Finally, 10 radiomic features were finally included for radiomics model construction. The AUC results showed that the ultrasomics model had moderate performance for SIS stage evaluation in both the training cohort (AUC=0.839) and the validation cohort (AUC=0.789). ML-derived ultrasomics can discriminate the SIS stage in patients with SIS. This noninvasive and low-cost approach may be helpful in the preliminary screening of shoulder pain.

Abstract 13327: Development and Validation of a Clinical Risk Score to Predict the Critical Illness in Hospitalized Patients With COVID-19

The Covid-19 pandemic has obligated healthcare systems to triage patients efficiently in order to utilize resources in the proper manner. Robust, validated clinical prediction tools are lacking that identify patients with coronavirus disease 2019 who are at the highest risk of mortality. A cohort of 661 positive COVID patients admitted to the Loyola hospital from 3/11/2020 to 10/12/2020 was used to validate previously published results with the composite outcome of ‘critically ill’, defined as invasive ventilation, ICU admission, or in-hospital death. Epidemiological, clinical, laboratory, and imaging variables ascertained at hospital admission were screened using Least Absolute Shrinkage and Selection Operator (LASSO) and logistic regression to construct a predictive risk score. Accuracy of the score was measured by the area under the receiver operating characteristic curve (AUC). Fifty-two (52) percent of our cohort had critical illness compared to 8% in the Liang paper (the first predictor model which was developed in China). Our population was older, had significantly higher abnormal chest x-ray, dyspnea, and number of comorbidities compared to their population. In addition, our population had higher values for NLR and LDH and lower direct bilirubin. Using the model coefficients presenting in Liang et al, the AUC was only 0.68 (95% CI: 0.64, 0.72), considerably lower than .88 presented in their publication. LASSO variable selection identified 8 variables (HR, SBP, fever, SpO2&lt;90, LDH, ferritin, D dimer, CRP on admission) for a multivariable model. In the multivariable model: SpO2 &lt;90, fever at admission, and increasing LDH were found to be statistically associated with our outcome. We finally added baseline predictors of age, sex, race, and number of comorbidities to the LASSO predictors which resulted in an AUC of 0.75 (95% CI: 0.71, 0.78). In our study, we found that three predictors (SpO2 &lt;90, fever, and increasing LDH on admission) were selected by the LASSO analysis to construct a predictive nomogram. The application of our model would help clinicians make a prompt decision to optimize patient stratification management. However, this quantitative tool needs to be validated by further large-scale prospective studies.

Identification of Immune-Related Genes for Establishment of Prognostic Index in Hepatocellular Carcinoma.

Background: Immune checkpoint inhibitor (ICI) therapy has been proved to be a promising therapy to many types of solid tumors. However, effective biomarker for estimating the response to ICI therapy and prognosis of hepatocellular carcinoma (HCC) patients remains underexplored. The aim of this study is to build a novel immune-related prognostic index based on transcriptomic profiles.Methods: Weighted gene co-expression network analysis (WGCNA) was conducted to identify immune-related hub genes that are differentially expressed in HCC cohorts. Next, univariate Cox regression analysis and least absolute shrinkage and selection operator (LASSO) analysis were used to detect hub genes associated to overall survival (OS). To validate the immune-related prognostic index, univariate and multivariate Cox regression analysis were performed. CIBERSORT and ESTIMATE were used to explore the tumor microenvironment and immune infiltration level.Results: The differential expression analysis detected a total of 148 immune-related genes, among which 25 genes were identified to be markedly related to overall survival in HCC patients. LASSO analysis yielded 10 genes used to construct the immune-related gene prognostic index (IRGPI), by which a risk score is computed to estimate low vs. high risk indicating the response to ICI therapy and prognosis. Further analysis confirmed that this immune-related prognostic index is an effective indicator to immune infiltration level, response to ICI treatment and OS. The IRGPI low-risk patients had better overall survival (OS) than IRGPI high-risk patients on two independent cohorts. Moreover, we found that IRGPI high-risk group was correlated with high TP53 mutation rate, immune-suppressing tumor microenvironment, and these patients acquired less benefit from ICI therapy. In contrast, IRGPI-low risk group was associated with low TP53 and PIK3CA mutation rate, high infiltration of naive B cells and T cells, and these patients gained relatively more benefit from ICI therapy.

A Prediction Model for Para-Aortic Lymph Node Metastasis in Cervical Cancer by Radiomics Analysis Using Pre-Treatment MRI Images of the Primary Tumor

Whole pelvis irradiation (WPI) is the standard radiation technique for locally advanced cervical cancer without para-aortic lymph node (PALN) metastasis. PALN is one of the most common late failure sites in patients treated with WPI. Although there are several reports about the utility of prophylactic extended-field irradiation (EFI), it is still controversial which patients benefit from prophylactic EFI. If we could predict a PALN metastasis from pre-treatment imaging data, it might help patients to select prophylactic EFI. This study aimed to construct a predictive model for the PALN metastasis in patients with cervical cancer by radiomics analysis using pretreatment magnetic resonance imaging (MRI) of the primary tumor.Data of 94 patients with cervical squamous cell carcinoma who underwent radiotherapy between 2003/10 and 2018/2 were split into two sets: 66 patients for the training/validation and 28 patients for testing. The PALN status was classified into two groups (positive or negative). Both the synchronous and metachronous PALN metastasis was classified as PALN positive. A total number of 9394 radiomics features per a patient image were extracted from T1- and T2-weighted MRI images. The set of candidate predictors were selected with the least absolute shrinkage and selection operator (LASSO) logistic regression and build predictive models with neural network classifiers were used. The precision, accuracy, and sensitivity by generating confusion matrices and the areas under the receiver operating characteristic curve (AUC) for each model were evaluated.By the LASSO analysis of the training/validation data, we found 9 radiomics features from T1-weighted MRI image and 61 radiomics features from T2-weighted MRI image for the classification. The accuracy, specificity, sensitivity, and AUC of the prediction model for the dataset in testing group were 67.9 %, 91.0%, 10.0%, and 0.60 with T1-weighted MRI image, 95.7%, 97.0%, 92.5%, and 0.98 with T2-weighted MRI image, 96.4%, 99.0%, 90.7%, and 0.99 with the combination of T1 and T2-weighted MRI images, respectively.We constructed a model to predict the PALN metastasis in patients with cervical cancer using pre-treatment MRI image-based radiomics and machine learning. The model based on T2-weighted image or combination of T1 and T2-wighted MRI image showed promising prediction accuracy. This model may be useful to select patients who benefit from prophylactic EFI.I. Nishibuchi: None. D. Kawahara: None. M. Kawamura: None. K. Kubo: None. N. Imano: None. Y. Takeuchi: None. A. Saito: None. Y. Murakami: None. Y. Nagata: None.

Radiomic Analysis for Pretreatment Prediction of Recurrence after Radiotherapy in Locally Advanced Cervical Cancer

<h3>Purpose/Objective(s)</h3> In earlier stage of locally advanced unresectable cervical squamous cell carcinoma, the treatment outcomes of surgery and definitive radiotherapy are comparable. However, the past study reported that one third of patients would experience recurrence. Predicting the treatment response and the long-term treatment outcome presents a major challenge for developing a more precise personalized care. Radiomics is a non-invasive and an emerging low-cost method to predict prognosis. In the current study, the recurrence of cervical cancer patients treated with radiotherapy was predicted using the radiomics features from the extended and shrink-uterus regions that extracted from pre-treatment T1- and T2-weighted MRI images. <h3>Materials/Methods</h3> Data of 90 tumor were split into two sets: 67 tumors for the training of models and 23 tumors for model testing. The treatment outcome was classified into two groups. The first group (group I) was patients with a recurrence. The second group (group II) was patients without a recurrence. A total number of 5022 radiomics features per a patient image were extracted from normalized and unnormalized T1- and T2-weighted MRI images. The set of candidate predictors were selected with the least absolute shrinkage and selection operator (LASSO) logistic regression and build predictive models with neural network classifiers were used. The precision, accuracy, and sensitivity by generating confusion matrices and the areas under the receiver operating characteristic curve (AUC) for each model were evaluated. <h3>Results</h3> By the LASSO analysis of the training data, we found 25 radiomics features from unnormalized T1-weighted (unT1w) MRI image and 4 radiomics features from unT2w MRI image for the classification. On the other hand, 11 radiomics features from normalized T1-weighted (nT1w) MRI image and 27 radiomics features from nT2w MRI image for the classification. The accuracy, specificity, sensitivity, and AUC of the prediction model for the dataset were 86.4%, 74.9%, 81.8%, and 0.89 with unT1w MRI image, 89.4%, 38.1%, 72.2%, and 0.69 with unT2w MRI image, 93.1, 81.6%, 88.7%, and 0.94 with the combination of unT1w and unT2w MRI images. On the other hand, the accuracy, specificity, sensitivity, and AUC of the prediction model for the dataset were 83.0%, 72.2%, 89.1%, and 0.90 with nT1w MRI image, 93.3%, 90.8%, 94.5%, and 0.96 with nT2w MRI image, 96.3, 92.8%, 98.9%, and 1.00 with the combination of nT1w and nT2w MRI images. <h3>Conclusion</h3> The radiomics features of the difference of the pixel number at center and peripheral region and the variation of the distribution were important factors in predicting the recurrence of the cervix cancer. The normalized method improved the accuracy of the prediction of the recurrence for cervix cancer.