A-145 Development of a Novel Algorithm Using Red Blood Cell Indices for Screening of Alpha Globin Gene Deletions

Abstract

Abstract Background Distinguishing between different causes of microcytic anemia is essential for management of anemic patients. More than 20 different RBC indices have been used to screen thalassemia trait with inconsistent results. Differentiation between iron deficiency anemia (IDA) and alpha-thalassemia is challenging based on RBC indices alone. Moreover, thalassemia and IDA may coexist. A highly sensitive screening algorithm is needed to select clinical cases for alpha globin gene analysis to reduce unnecessary testing and improve pre-test probabilities. Methods A retrospective study was performed to evaluate the reliability of 22 RBC indices (RBC, MCV, RDW, MCH, Mentzer Index, RDW Index, etc) and formulae in differentiating cases with 1–2 alpha globin gene deletions from negative cases, in a cohort of 90 individuals with confirmatory alpha globin gene analysis results. Results The number of patients with each alpha globin gene deletion pattern were: 37 patients with alpha-thalassemia trait, 21 silent carriers (one gene deletion), 31 patients with negative result and 1 patient with triplicated alpha globin gene. The mean RBC count in each group were 4.79 ± 0.65 (trait), 4.45 ± 0.73 (carrier), and 4.21 ± 0.84 (negative) (million/mm). The mean MCV in each group were 71.1 ± 4.8 (trait), 76.1 ± 5.9 (carrier), and 74.2 ± 9.3 (negative) (fL). The mean RDW in each group were 16.5 ± 3.0 (trait), 17.2 ± 3.4 (carrier), and 18.3 ± 4.9 (negative) (%). The mean MCH in each group were 22.0 ± 1.8 (trait), 23.8 ± 2.3 (carrier), and 23.5 ± 3.7 (negative) (pg). None of the RBC indices showed significant difference between carrier and negative groups, while 20 RBC indices showed significant difference (P < 0.05) between trait and negative groups. The sensitivities of the RBC indices in detecting 1–2 alpha globin gene deletions based on published cut-offs ranged between 3.4% and 81.0%, while the specificities ranged between 41.9% and 100%. A novel scoring algorithm was evaluated based on the combination of 12 selected RBC indices and formulae, with each index or formula contributing to a score of 0 or 1. The average value of the negative cases was used as the score cut-off value for each index or formula. The scoring algorithm gave a score distribution between 0 and 12, with lower scores indicating alpha globin gene deletions. It achieved a sensitivity of 96.6% in detecting 1–2 alpha globin gene deletions, with a negative likelihood ratio of 0.15. The negative likelihood ratio of the scoring algorithm outperformed each of the individual RBC indices at similar levels of sensitivities (>95%). Conclusion The novel algorithm is a highly sensitive tool for screening of alpha globin gene deletions, including one gene deletion. It is more effective at identifying negative cases compared to individual RBC indices. Implementation of the algorithm could efficiently reduce 10.0% (9/90) of the test volume of alpha globin gene analysis. A more stringent version of the algorithm that incorporates standard deviation in the cut-off values could potentially achieve nearly 100% sensitivity and reduce 5.6% (5/90) of the alpha globin gene analysis test volume.