Predicting haemoglobin deferral using machine learning models: Can we use the same prediction model across countries?

BACKGROUND AND OBJECTIVES: Personalized donation strategies based on haemoglobin (Hb) prediction models may reduce Hb deferrals and hence costs of donation, meanwhile improving commitment of donors. We previously found that prediction models perform better in validation data with a high Hb deferral rate. We therefore investigate how Hb deferral prediction models perform when exchanged with other blood establishments. MATERIALS AND METHODS: Donation data from the past 5 years from random samples of 10,000 donors from Australia, Belgium, Finland, the Netherlands and South Africa were used to fit random forest models for Hb deferral prediction. Trained models were exchanged between blood establishments. Model performance was evaluated using the area under the precision-recall curve (AUPR). Variable importance was assessed using SHapley Additive exPlanations (SHAP) values. RESULTS: Across the validation datasets and exchanged models, the AUPR ranged from 0.05 to 0.43. Exchanged models performed similarly within validation datasets, irrespective of the origin of the training data. Apart from subtle differences, the importance of most predictor variables was similar in all trained models. CONCLUSION: Our results suggest that Hb deferral prediction models trained in different blood establishments perform similarly within different validation datasets, regardless of the deferral rate of their training data. Models learn similar associations in different blood establishments.

Intranasal administration of convalescent plasma protects against SARS-CoV-2 infection in hamsters.

BACKGROUND: Convalescent plasma (CP) transfusion is an early option for treating infections with pandemic potential, often preceding vaccine or antiviral drug rollout. Heterogenous findings from randomized clinical trials on transfusion of COVID-19 CP (CCP) have been reported. However, meta-analysis suggests that transfusion of high titer CCP is associated with a mortality benefit for COVID-19 outpatients or inpatients treated within 5 days after symptom onset, indicating the importance of early administration. METHODS: We tested if CCP is an effective prophylactic against SARS-CoV-2 infection by the intranasal administration of 25 µL CCP/nostril (i.e. 0.01-0.06 mg anti-RBD antibodies/kg) in hamsters exposed to infected littermates. FINDINGS: In this model, 40% of CCP treated hamsters were fully protected and 40% had significantly reduced viral loads, the remaining 20% was not protected. The effect seems dose-dependent because high-titer CCP from a vaccinated donor was more effective than low-titer CCP from a donation prior to vaccine rollout. Intranasal administration of human CCP resulted in a reactive (immune) response in hamster lungs, however this was not observed upon administration of hamster CCP. INTERPRETATION: We conclude that CCP is an effective prophylactic when used directly at the site of primary infection. This option should be considered in future prepandemic preparedness plans. FUNDING: Flanders Innovation & Entrepreneurship (VLAIO) and the Foundation for Scientific Research of the Belgian Red Cross Flanders.

An international comparison of haemoglobin deferral prediction models for blood banking.

BACKGROUND AND OBJECTIVES: Blood banks use a haemoglobin (Hb) threshold before blood donation to minimize donors' risk of anaemia. Hb prediction models may guide decisions on which donors to invite, and should ideally also be generally applicable, thus in different countries and settings. In this paper, we compare the outcome of various prediction models in different settings and highlight differences and similarities. MATERIALS AND METHODS: Donation data of repeat donors from the past 5 years of Australia, Belgium, Finland, the Netherlands and South Africa were used to fit five identical prediction models: logistic regression, random forest, support vector machine, linear mixed model and dynamic linear mixed model. Only donors with five or more donation attempts were included to ensure having informative data from all donors. Analyses were performed for men and women separately and outcomes compared. RESULTS: Within countries and overall, different models perform similarly well. However, there are substantial differences in model performance between countries, and there is a positive association between the deferral rate in a country and the ability to predict donor deferral. Nonetheless, the importance of predictor variables across countries is similar and is highest for the previous Hb level. CONCLUSION: The limited impact of model architecture and country indicates that all models show similar relationships between the predictor variables and donor deferral. Donor deferral is found to be better predictable in countries with high deferral rates. Therefore, such countries may benefit more from deferral prediction models than those with low deferral rates.

A novel competition ELISA for the rapid quantification of SARS-CoV-2 neutralizing antibodies in convalescent plasma.

BACKGROUND: COVID-19 convalescent plasma (CCP) ideally contains high titers of (neutralizing) anti-SARS-CoV-2 antibodies. Several scalable immunoassays for CCP selection have been developed. We designed an enzyme-linked immunosorbent assay (ELISA) that measures neutralizing antibodies (of all isotypes) in plasma by determining the level of competition between CCP and a mouse neutralizing antibody for binding to the receptor binding domain (RBD) of SARS-CoV-2. METHODS: Plasma was collected from 72 convalescent individuals and inhibition of viral infection was determined by plaque reduction neutralization (PRNT50). The level of neutralizing antibodies was measured in the novel competition ELISA and in a commercially available ELISA that measures inhibition of recombinant ACE2 binding to immobilized RBD. These results were compared with a high throughput chemiluminescent microparticle immunoassay (CMIA). RESULTS: The results from both ELISAs were correlating, in particular for high titer CCP (PRNT50 ≥ 1:160) (Spearman r = .73, p < .001). Moderate correlation was found between the competition ELISA and CMIA (r = .57 for high titer and r = .62 for low titer CCP, p < .001). Receiver operator characteristic analysis showed that the competition ELISA selected CCP with a sensitivity and specificity of 61% and 100%, respectively. However, discrimination between low and high titer CCP had a lower resolution (sensitivity: 34% and specificity: 89%). CONCLUSION: The competition ELISA screens for neutralizing antibodies in CCP by competition for just a single epitope. It exerts a sensitivity of 61% with no false identifications. These ELISA designs can be used for epitope mapping or for selection of CCP.

Are we underestimating reverse TRALI?

BACKGROUND: Transfusion-related acute lung injury (TRALI) is a rare but serious adverse transfusion reaction and is known to be related to anti-human leukocyte antigen (HLA) or anti-human neutrophil antigen (HNA) antibodies in donor plasma. In 2016, four of eight reported TRALI cases could not be explained by donor antibodies. It is assumed that fewer than 10% of TRALI cases are triggered by anti-HLA or anti-HNA antibodies in the patient's plasma (reverse TRALI). STUDY DESIGN AND METHODS: Three cases of red blood cell (RBC)-associated and one case of granulocyte-associated TRALI were investigated. Data were collected on the clinical aspects of the patient and the concerned blood product. Patient's HLA antibodies were determined and the implicated donor was contacted for HLA typing. The HLA antibody identification and strength were assessed using a bead assay (Luminex Single Antigen bead assay, Immucor). For HLA typing, a polymerase chain reaction-sequence-specific oligonucleotide method was used that also included Luminex detection. RESULTS: In three RBC-associated TRALI cases, HLA Class I and II antibodies found in the patient's plasma were specific for the HLA type of the transfused leukoreduced blood product. In a fourth case, HLA antibodies were found in a patient who developed TRALI after repeated granulocyte infusions. The HLA antibodies were directed against HLA antigens present on the donor WBCs. CONCLUSION: The diagnosis of reverse TRALI was retained in four cases, suggesting that reverse TRALI is more frequent than described in the literature, especially in patients with an increased risk for having HLA antibodies.

Hemoglobin screening in blood donors: a prospective study assessing the value of an invasive and a noninvasive point-of-care device for donor safety.

BACKGROUND: Low hemoglobin (Hb) levels are a common reason for whole blood donor deferral. As some of the deferred donors do not return to donate blood later, the development of accurate tools to reliably detect low Hb levels is therefore important for donor safety and retention. STUDY DESIGN AND METHODS: In a prospective study, 1483 whole blood donors were enrolled in three study arms. In each study arm the Hb results measured with a point-of-care testing device were compared with the venous Hb results obtained in our central laboratory. Both an invasive capillary Hb (Compolab) and a noninvasive spectroscopy Hb (Haemospect) were tested. For the latter, two different application methods for the positioning of the Digiclip were assessed. In a second phase the Compolab was tested in routine conditions. RESULTS: The Haemospect device failed to reliably identify donors with Hb levels below the European cutoffs (125 g/L for females and 135 g/L for males). In contrast, the Compolab identified 72.4% of first-time donors and 78.9% of regular donors with Hb levels below these cutoffs. In routine conditions, the Compolab identified the majority of donors with low Hb levels. CONCLUSION: Differences between Hb levels measured with invasive and noninvasive point-of-care devices and measured on venous samples exist. Implementing the Compolab at collection sites provides a high level of safety toward first-time donors.

Screening for HLA antibodies in plateletpheresis donors with a history of transfusion or pregnancy.

BACKGROUND: Transfusion-related acute lung injury (TRALI) is known as a life-threatening complication of transfusion. HLA and HNA antibodies have been associated with the immune pathway of TRALI. Since donors with a history of transfusion and/or pregnancy are presumed to have an increased risk of carrying such antibodies, we investigated the association of a history of transfusion or pregnancy with the occurrence of HLA alloimmunization in our donor population. STUDY DESIGN AND METHODS: A total of 1018 female plateletpheresis donors and male plateletpheresis donors with a history of transfusion were enrolled in the study. Included donors were systematically screened, using Luminex technology, for anti-HLA Class I and II. The association of donor history with HLA alloimmunization status was analyzed. RESULTS: The overall alloimmunization rate was 20.2%. In 0.0% of the nulliparous transfused female donors and in 1.3% of the transfused male donors, anti-HLA were detected. Thirty-one percent of the parous women versus 4.2% of the nulliparous women screened positive for anti-HLA. The rate of HLA alloimmunization increased with parity. CONCLUSION: Our data indicate that a history of transfusion is a minor risk factor for immunization against HLA antigens. In contrast, former pregnancies constitute a major risk factor for the development of HLA antibodies. Since HLA alloimmunization rate increases with parity, TRALI risk reduction measures should focus on this particular donor population. Repeated testing of female plateletpheresis donors after each pregnancy is implemented in our blood service.