ABSTRACT
BACKGROUND AND AIMS: Since the beginning of the COVID-19 pandemic in early 2020, >290 million people were infected by SARS-CoV-2 and >5.4 million have died from or with COVID-19 (https://coronavirus.jhu.edu/). Patients with chronic health conditions such as end-stage kidney disease (ESKD) experience particularly high morbidity and mortality because of COVID-19. ESKD patients on hemodialysis are widely vaccinated for hepatitis B (HBV) and seroconversion is routinely measured. This practice presents a rare opportunity to study immune function on a wide scale. It can be reasonably assumed that patients who are able to produce a vaccinal or post-HBV antibodies titers have a better immune function than those who are unable to mount such a serological response. We aim to jointly analyze results of SARS-CoV-2 RT-PCR and hepatitis B serology to determine if presence of vaccinal or post-HBV antibodies is associated with likelihood of developing COVID-19 infection. METHOD: Patients who were tested for COVID-19 at Fresenius Medical Care North America dialysis clinics from May 2020 to September 2020 were included in this analysis. HBV infection/vaccination status, demographic parameters and clinical parameters were obtained from the medical record. Nasopharyngeal swab specimen was tested via RT-PCR to detect presence of SARS-CoV-2. Patients were categorized as having good immune function or poor immune function based on vaccinal and post-HBV sero-status. Patients who were vaccinated against HBV but did not seroconvert were considered to have poor immune function. On the other hand, patients who mounted vaccinal or post-HBV antibodies were considered to have good immune function. Univariate and multivariate logistic regression were utilized to study the association between immune function and other demographic, anthropometric and clinical parameters on the likelihood of not being diagnosed with COVID-19. Four models were constructed: Model 1: unadjusted;Model 2: adjusted for age. Model 3: adjusted for age, gender, race, ethnicity, body mass index (BMI). Model 4: adjusted for parameters in model 3 and dialysis vintage (in years), diabetes and congestive heart failure (CHF). RESULTS: 11 870 patients were included in this analysis. 54% patients were male, 33% were Black, 24% of the patients were Hispanic, 69% had diabetes and 22% had CHF. Patients were 61.2 ± 14.4 years old with dialysis vintage of 3.9 ± 3.9 years, BMI of 29.6 ± 9.7 kg/m2 and eKt/V 1.5 ± 0.3. Of these patients, 21% had poor immune function and 79% had good immune function. Results of the logistic regression models are shown in Table 1. In the unadjusted model, poor immune function was associated with an increased likelihood of being diagnosed with COVID-19. In models, 2, 3 and 4 age, vintage and presence of diabetes were all significantly associated with a higher likelihood of being diagnosed with COVID-19. However, poor immune function was not a significant predictor of COVID-19 diagnosis in the adjusted models. CONCLUSION: Patients who have vaccinal or post-HBV antibodies did not have a lower likelihood of COVID-19 compared with patients who were unable to mount an adequate vaccinal or post-HBV antibody response. Response to HBV vaccination or infection may not be adequate to characterize a patient as having good immune response. Other factors that are routinely measured in hemodialysis patients, which may allow us to make inferences about a patient's immune function should be explored.
ABSTRACT
Medical imaging such as computed tomography (CT) plays a critical role in the global fight against COVID19. Computer-aided platforms have emerged to help radiologists diagnose and track disease prognosis. In this paper, we introduce an automated deep-learning segmentation model, which builds upon the current U-net model, however, leverages the strengths of long and short skip connections. We complemented the long skip connections with a cascaded dilated convolution module that learns multi-scale context information, compensates the reduction in receptive fields, and reduces the disparity between encoded and decoded features. The short connections are considered in utilizing residual blocks as the basic building blocks for our model. They ease the training process, reduce the degradation problem, and propagate the low fine details. This enables the model to perform well in capturing smaller regions of interest. Furthermore, each residual block is followed by a squeeze and excitation unit, which stimulates informative features and suppresses less important ones, thus improving the overall feature representation. After extensive experimentation with a dataset of 1705 COVID-19 axial CT images, we demonstrate that performance gains can be achieved when deep learning modules are integrated with the basic U-net model. Experimental results show that our model outperformed the basic U-net and ResDUnet model by 8.1% and 1.9% in dice similarity, respectively. Our model provided a dice similarity measure of 85.3%, with a slight increase in trainable parameters, thus demonstrating a huge potential for use in the clinical domain.
ABSTRACT
BACKGROUND AND AIMS: Dialysis patients are at higher risk for severe acute respiratory syndrome coronavirus (SARS-CoV-2) infection. Longevity of antibody response to SARS-CoV-2 infection remains unclear. It is reported that maintenance hemodialysis (MHD) patients can mount an antibody response that is similar in intensity and timing to the non-dialysis population. We aim to investigate the prevalence and persistence of antibodies in hemodialysis patients. METHOD: We measured IgG and IgM antibodies in MHD patients as part of a quality improvement project. Four New York City dialysis clinics participated in this study. Strict policy of RT-PCR testing was implemented in clinics for patients with signs and symptoms of Coronavirus Disease 2019 (COVID-19). Initial antibody testing was done on June 10 and July 13, 2020 (phase 1) and retesting was done for previously positive patients between December 9 and 17, 2020 (phase 2). Upon obtaining verbal consent, 3.5 ml of pre-dialysis blood samples were taken via vascular access. SARS-CoV-2 antibodies were determined using the emergency use authorized Diazyme DZ-Lite SARS-CoV-2 IgM / IgG CLIA assays with 100% sensitivity and 98% specificity. Detection of formed immune-complexes is achieved with N-(4-amino-butyl)-N-ethylisoluminol;the luminescence signal is reported as units per ml (AU/ml), values ≥ 1.00 AU/ml are considered as 'reactive' and < 1.00 AU/ml as 'non-reactive.' RESULTS: A total of 429 MHD patients were studied in phase 1. Antibodies were present in 130 (30.3%) and only 55 patients with Covid-19 diagnosis confirmed by RTPCR test were reactive for IgG antibodies. The time to antibody testing was 73 days (median 77;range 30-111) days. In the phase 2 antibody testing, IgG antibodies were only detected in 47 patients (85.5%) 242 days (median 245, range 204 to 268) after clinical diagnosis of Covid-19. Between the two phases of antibody testing, the luminescence signal declined by 40.9 AU/mL (95% confidence interval 31.5 to 50.3) from 54.1±45.3 to 13.2±20.9 AU/mL (P<0.0001 by paired t-test;Figure 1). In univariate logistic regression, a higher number of days between clinical diagnosis of COVID-19 and the second antibody measurement was associated with a lower seropositivity rate (odds ratio 0.929, 95% confidence interval 0.864 to 0.998, P=0.044). Antibody persistence was not associated with age, gender, race, and ethnicity. CONCLUSION: We observed that about 6 out of 7 MHD patients maintain SARSCoV-2 antibodies over 6-9 months but there is a significant decline of IgG level. The time between clinical diagnosis and IgG testing was associated with IgG decline. Follow up study to understand antibody dynamics in MHD population is a crucial step once vaccines become available.