A TLR7-nanoparticle adjuvant promotes a broad immune response against heterologous strains of influenza and SARS-CoV-2.

The ideal vaccine against viruses such as influenza and SARS-CoV-2 must provide a robust, durable and broad immune protection against multiple viral variants. However, antibody responses to current vaccines often lack robust cross-reactivity. Here we describe a polymeric Toll-like receptor 7 agonist nanoparticle (TLR7-NP) adjuvant, which enhances lymph node targeting, and leads to persistent activation of immune cells and broad immune responses. When mixed with alum-adsorbed antigens, this TLR7-NP adjuvant elicits cross-reactive antibodies for both dominant and subdominant epitopes and antigen-specific CD8+ T-cell responses in mice. This TLR7-NP-adjuvanted influenza subunit vaccine successfully protects mice against viral challenge of a different strain. This strategy also enhances the antibody response to a SARS-CoV-2 subunit vaccine against multiple viral variants that have emerged. Moreover, this TLR7-NP augments antigen-specific responses in human tonsil organoids. Overall, we describe a nanoparticle adjuvant to improve immune responses to viral antigens, with promising implications for developing broadly protective vaccines.

Accurate Virus Identification with Interpretable Raman Signatures by Machine Learning (preprint)

Rapid identification of newly emerging or circulating viruses is an important first step toward managing the public health response to potential outbreaks. A portable virus capture device coupled with label-free Raman Spectroscopy holds the promise of fast detection by rapidly obtaining the Raman signature of a virus followed by a machine learning approach applied to recognize the virus based on its Raman spectrum, which is used as a fingerprint. We present such a machine learning approach for analyzing Raman spectra of human and avian viruses. A Convolutional Neural Network (CNN) classifier specifically designed for spectral data achieves very high accuracy for a variety of virus type or subtype identification tasks. In particular, it achieves 99% accuracy for classifying influenza virus type A vs. type B, 96% accuracy for classifying four subtypes of influenza A, 95% accuracy for differentiating enveloped and non-enveloped viruses, and 99% accuracy for differentiating avian coronavirus (infectious bronchitis virus, IBV) from other avian viruses. Furthermore, interpretation of neural net responses in the trained CNN model using a full-gradient algorithm highlights Raman spectral ranges that are most important to virus identification. By correlating ML-selected salient Raman ranges with the signature ranges of known biomolecules and chemical functional groups (for example, amide, amino acid, carboxylic acid), we verify that our ML model effectively recognizes the Raman signatures of proteins, lipids and other vital functional groups present in different viruses and uses a weighted combination of these signatures to identify viruses.

Accurate Virus Identification with Interpretable Raman Signatures by Machine Learning (preprint)

Rapid identification of newly emerging or circulating viruses is an important first step toward managing the public health response to potential outbreaks. A portable virus capture device coupled with label-free Raman Spectroscopy holds the promise of fast detection by rapidly obtaining the Raman signature of a virus followed by a machine learning approach applied to recognize the virus based on its Raman spectrum. In this paper, we present a machine learning analysis on Raman spectra of human and avian viruses. A Convolutional Neural Network (CNN) classifier specifically designed for spectral data achieves very high accuracy for a variety of virus type or subtype identification tasks. In particular, it achieves 99% accuracy for classifying influenza virus type A vs. type B, 96% accuracy for classifying four subtypes of influenza A, 95% accuracy for differentiating enveloped and non-enveloped viruses, and 99% for differentiating avian coronavirus (infectious bronchitis virus, IBV) from other avian viruses. Furthermore, interpretation of neural net responses in the trained CNN model using a full-gradient algorithm highlights Raman spectral ranges that are most important to virus identification. By correlating ML-selected salient Raman ranges with the signature ranges of known biomolecules and chemical functional groups (e.g. amide, amino acid, carboxylic acid) we verify that our ML model effectively recognizes the Raman signatures of proteins, lipids and other vital functional groups present in different viruses and uses a weighted combination of these signatures to identify viruses. The accurate and interpretable machine learning model developed for Raman virus identification presents promising potential in a real-time virus detection system. Significance Statement A portable micro-fluidic platform for virus capture promises rapid enrichment and label-free optical identification of viruses by Raman spectroscopy. A large Raman dataset collected on a variety of viruses enables the training of machine learning (ML) models capable of highly accurate and sensitive virus identification. The trained ML models can then be integrated with the portable device to provide real-time virus detection and identification capability. We validate this conceptual framework by presenting highly accurate virus type and subtype identification results using a convolutional neural network to classify Raman spectra of viruses.

Association Between Risk of VTE and Mortality in Patients with COVID-19 (preprint)

Background: Hospitalized patients with COVID-19 appeared high risk of venous thromboembolism (VTE), which exhibited the predictor of mortality in non-COVID-19 patients. Objectives: We aimed to investigate the association between risk of VTE with 30-day mortality in COVID-19 patients.Methods: In this retrospective cohort study, 1030 consecutive hospitalized patients with COVID-19 were recruited in two hospitals of Wuhan, China. We collected baseline data on demographics, SOFA parameters, and VTE risk assessment models (RAMs) including Padua Prediction Score (PPS), IMPROVE and Caprini RAM. The primary outcome of the study was 30-day mortality. Results: Thirty-day mortality increased progressively from 2% in patients at low risk of VTE to 63% in those at high risk defined by PPS. Similar findings were also observed for risk of VTE defined by IMPROVE and Caprini score. Progressive increases in VTE risk also were associated with higher SOFA score. Our findings showed that the presence of high risk of VTE was independently associated with 30-day mortality regardless of adjusted gender, smoking status and some comorbidities with hazard ratios of 29.19, 37.37, 20.60 for PPS, IMPROVE and Caprini RAM, respectively (P< 0.001 for all comparisons). Predictive accuracy of PPS (AUC, 0.900), IMPROVE (AUC, 0.917) or Caprini RAM (AUC, 0.861) as the risk of 30-day mortality was markedly well.Conclusions: The presence of high risk of VTE identifies a group of patients with COVID-19 at higher risk for 30-day mortality. Furthermore, there is higher accuracy of VTE RAMs to predict 30-day mortality in these patients.

Association between Risk of Venous Thromboembolism and Mortality in Patients with COVID-19 (preprint)

Background: Hospitalized patients with COVID-19 appeared high risk of venous thromboembolism (VTE) defined by some risk assessment models (RAMs), which exhibited the predictor of mortality in patients with non-COVID-19. We aimed to investigate the association between risk of VTE with 30-day mortality in COVID-19 patients.Methods: In this retrospective cohort study, 1030 consecutive hospitalized patients, between January 26, 2020 and March 29, 2020, aged 14–98 years with a confirmed diagnosis of COVID-19-related pneumonia were recruited from Jinyintan Hospital and Union Hosptial, Wuhan, China. We collected baseline data on demographics, SOFA parameters, and VTE RAMs including Padua Prediction Score (PPS), IMPROVE RAM and Caprini RAM. The primary outcome of the study was 30-day mortality. The secondary outcome was the length of stay in hospital. Findings: Thirty-day mortality increased progressively from 2% in patients at low risk of VTE to 63% in those at high risk of VTE defined by PPS ≥ 4. Similar findings were also observed for risk of VTE defined by IMPROVE score ³ 2 and Caprini score ³ 5. Progressive increases in VTE risk also were associated with higher SOFA score. Our findings showed that the presence of high risk of VTE was independently associated with 30-day mortality regardless of adjusted gender, smoking status and some comorbidities with hazard ratios of 29.19 (95% CI 15.76 - 54.05), 37.37 (95% CI 18.43 - 75.78), 20.60 (95% CI 11.41 - 37.19) for PPS, IMPROVE RAM and Caprini RAM, respectively (P < 0.001 for all comparisons). Predictive accuracy of PPS (AUC, 0.900; 95% CI 0.881 - 0.919), IMPROVE RAM (AUC, 0.917; 95% CI 0.898 - 0.936) or Caprini RAM (AUC, 0.861; 95% CI 0.840 - 0.882) as the risk of 30-day mortality was markedly well.Interpretation: The presence of high risk of VTE identifies a group of patients with COVID-19 at higher risk for 30-day mortality. Furthermore, there is higher accuracy of VTE RAMs to predict 30-day mortality in COVID-19 hospitalized patients.Funding Statement: This study is funded by the National Nature Science Foundation of China (81870062 to Jinjun Jiang, 81900038 to Shujing Chen).Declaration of Interests: The authors have no conflict of interest to disclose.Ethics Approval Statement: The study was approved by Jinyintan Hospital Ethics Committee (KY2020-06.01) and Union Hospital Ethics Committee (2020-0039). Written informed consent was waived by the Ethics Commission.

Lactate dehydrogenase, a Risk Factor of Severe COVID-19 Patients (preprint)

BACKGROUND The World Health Organization (WHO) has recently declared coronavirus disease 2019 (COVID-19) a public health emergency of global concern. Updated analysis of cases might help identify the characteristic and risk factors of the illness severity. METHODS We extracted data regarding 47 patients with confirmed COVID-19 from Renmin Hospital of Wuhan University between February 1 and February 18, 2020. The degree of severity of COVID-19 patients (severe vs. non-severe) was defined at the time of admission according to American Thoracic Society (ATS) guidelines for community-acquired pneumonia (CAP). RESULTS The median age was 64.91 years, 26 cases (55.31%) were male of which, and 70.83% were severe cases. Severe patients had higher APACHE II (9.92 vs 4.74) and SOFA (3.0 vs 1.0) scores on admission, as well as the higher PSI (86.13 vs 61.39), Curb-65 (1.14 vs 0.48) and CT semiquantitative scores (5.0 vs 2.0) when compared with non-severe patients. Among all univariable parameters, APACHE II, SOFA, lymphocytes, CRP, LDH, AST, cTnI, BNP, et al were significantly independent risk factors of COVID-19 severity. Among which, LDH was most positively related both with APACHE II (R = 0.682) and SOFA (R = 0.790) scores, as well as PSI (R = 0.465) and CT (R = 0.837) scores. To assess the diagnostic value of these selected parameters, LDH (0.9727) had maximum sensitivity (100.00%) and specificity (86.67%), with the cutoff value of 283. As a protective factor, lymphocyte counts less than 1.045 x 109 /L showed a good accuracy for identification of severe patients with AUC = 0.9845 (95%CI 0.959-1.01), the maximum specificity (91.30%) and sensitivity (95.24%). In addition, LDH was positively correlated with CRP, AST, BNP and cTnI, while negatively correlated with lymphocyte cells and its subsets, including CD3+, CD4+ and CD8+ T cells (P < 0.01). CONCLUSIONS This study showed that LDH coule be identified as a powerful predictive factor for early recognition of lung injury and severe COVID-19 cases. And importantly, lymphocyte counts, especially CD3+, CD4+, and CD8+ T cells in the peripheral blood of COVID-19 patients, which was relevant with serum LDH, were also dynamically correlated with the severity of the disease. FUNDING Key Project of Shanghai Municipal Health Bureau (2016ZB0202)