Transfusion最新文献

Background: In this pilot trial, we tested the feasibility of conducting a randomized controlled trial in cardiac surgery patients using extended-, 100% plasma-, cold-stored, and long-distance-shipped platelets (CSPs).

Study design and methods: We conducted a single center, controlled, double-blind pilot study in adult patients undergoing elective redo or complex cardiothoracic surgery. Patients were allocated in a week-based block randomization scheme to receive either room temperature-stored platelets (RTPs) or CSPs shipped from a Texas-based blood center and stored between 10 and 14 days. The primary outcome was defined as the feasibility of recruitment and accrual. Several other secondary endpoints were assessed. All platelet units were screened for aggregates using RTP release criteria.

Results: In a post-hoc "as treated" analysis, 15 patients received RTPs, and 9 received CSPs (including 3 who received both). We found that most CSPs (58%) were not usable for transfusion due to the presence of aggregates. This resulted in an excess of subjects receiving RTPs; consequently, the final nine transfused participants were allocated to receive CSPs without randomization. We accrued 0.7 evaluable subjects/month of active enrollment, which was below our desired primary outcome feasibility target of ≥1.2. One death within 28 days occurred in the RTP transfusion group, while none occurred in the CSPs group. In vitro testing yielded contradictory results.

Conclusion: Due to slow recruitment and the abundance of aggregates in CSPs, this pilot trial does not support the feasibility of the study protocol.

Background: A national transfusion history sharing service, such as the alloantibody exchange, should provide standardized data. However, red cell antigen profiles frequently exist as unstructured text. To reduce human curation and improve scalability, we evaluated whether large language models (LLMs) could standardize free-text antigen profiles.

Study design and methods: Using an academic center's antigen profiles, we compared open-weight LLMs run locally and commercial LLM accessed via the internet. Outputs were constrained to a prespecified format ("schema-guided decoding") and retries were allowed ("error-aware retries"). The primary outcome was an exact match of the antigen profile; secondary outcomes included antigen-level agreement, categorized error types, and processing cost. Differences were tested with Cochran's Q and pairwise McNemar tests with Holm correction.

Results: Of 908 deidentified red cell antigen profiles, a total of 10,547 antigens with 88 distinct antigens were noted. Performance differed significantly across LLMs (Cochran's Q = 3014.6, df = 10, p < 0.001). Commercial LLMs outperformed open source LLMs. Open-source models with a higher number of parameters outperformed smaller models. The top LLM matched 92.5% of the antigen profiles; antigen-level agreement reached 99.4%. Estimated API costs for commercial LLM were modest for the studied workload.

Discussion: Commercial LLMs can standardize unstructured antigen profiles with high accuracy when paired with schema-guided decoding and error-aware retries. The high accuracy shown by the LLMs suggests they would offer significant scalability while reducing manual curation for a national transfusion history sharing service like the alloantibody exchange. They represent a foundational technology for enabling such a service.

Background: Specimens contaminated with intravenous (IV) fluids can lead to considerable measurement errors in complete blood counts (CBCs), posing challenges for laboratory operations and clinical decision-making. There is no gold-standard method for retrospectively identifying these events, making it difficult to target quality improvement initiatives or optimize laboratory detection protocols.

Study design and methods: This study aimed to develop and validate machine learning (ML) models to retrospectively identify IV fluid contamination in CBC results at scale across two institutions. The models were trained on simulated contamination in CBCs using prior, current, and post hemoglobin concentrations, platelet counts, and white blood cell counts, then validated against expert-reviewed datasets. Real-world applicability was assessed using 1 year's worth of CBC results from each institution.

Results: The models effectively discriminated contaminated from non-contaminated results on real-world datasets, with areas under the receiver operating characteristic curve of 0.972 and 0.957, and areas under the precision-recall curve of 0.723 and 0.619. In 1 year of CBC data, ~2% of inpatient CBC trios were predicted as contaminated across both institutions. 6%-9% of inpatient transfusions for which a CBC trio was available were deemed potentially unnecessary using a rule set validated by expert chart review.

Discussion: The findings support the feasibility of using ML to identify IV fluid contamination in CBC results efficiently and effectively. Further work, including prospective real-world evaluations, targeted quality improvement initiatives, and development of real-time detection models, is necessary before realizing the potential benefits to patient safety, laboratory operations, and patient blood management.

Background: Characterization of platelet gel (PG) from different blood sources and preparation methods remains incomplete.

Study design and methods: This study compared the weight, growth factor (GF) content, and gelation dynamics (via rotational thromboelastometry, ROTEM) of allogeneic PG prepared from adult blood (AB) and cord blood (CB) platelet concentrates (PC). ABPC were suspended in 100% plasma or 35% plasma/65% platelet additive solution (PAS), while CBPC were suspended in 100% plasma. PG was prepared in commercial BioNest D mini-bags using a method with calcium gluconate, locally made cryoprecipitate, and thrombin (CT), or a reference method with calcium gluconate and commercial batroxobin (BA). PG was analyzed at 30 min and 6 h.

Results: At 30 min, PG weights were higher with CT than BA across all conditions. With 5 mL ABPC in 100% plasma, weights were 7.20 ± 1.07 g (CT) versus 2.72 ± 0.19 g (BA); with 10 mL ABPC in PAS, 9.49 ± 2.86 g (CT) versus 3.87 ± 3.1 g (BA); and with 4.09 ± 0.45 mL CBPC, 5.62 ± 1.21 g (CT) versus 2.04 ± 0.78 g (BA). All PGs lost mass by 6 h, but CT consistently retained more weight. GF percent retention in the PG at 6 h was higher with CT. ROTEM results showed differences in clotting time and comparable values of maximum clot firmness between PG from AB and CB.

Discussion: The CT method is a promising alternative for producing allogeneic PG using affordable, locally sourced reagents.