Toward the Creation of Diagnostic Algorithms for Congenital Hemolytic Anemia Using Advanced Parameters from Modern Hematology Analyzers

Introduction. Congenital hemolytic anemias (CHAs) are a group of disorders caused by defects of red blood cells membrane, metabolism, erythroid maturation, or abnormal hemoglobin. Due to their rarity and heterogeneity diagnosis is challenging and requires specific tools available only in reference centers; therefore, some cases may remain undiagnosed or misdiagnosed. Modern hematology analyzers provide new erythrocytes (RBC) and reticulocytes (RET) parameters; however, few studies focused on these parameters and CHAs. Algorithms have been published on Sysmex XE and XN analyzers only for hereditary spherocytosis (HS) (Mullier, 2011; Persijn, 2012; Bobée, 2018; Sottiaux, 2020) and pyruvate kinase deficiency (PKD) (Bobée, 2018)

Aim. We aimed to create a comprehensive algorithm for screening and differential diagnosis of the most common forms of CHA (HS, PKD, and congenital dyserythropoietic anemia type II - CDAII) by using RBC and RET parameters provided by Sysmex XN. An additional algorithm for patients with methemoglobinemia due to NADH diaphorase deficiency or unstable hemoglobins (MetHb) was defined thanks to a particular scatterplot of the WBC differential channel observed in these patients.

Methods. Between 2020 and 2023, 500 consecutive patients were analyzed on Sysmex XN. Among them 305 had a confirmed diagnosis of CHA (106 HS, 21 PKD, 12 CDAII; 10 stomatocytosis, 118 hemoglobinopathies; 38 other CHAs), 6 had MetHb and 189 had acquired anemias (50 autoimmune hemolytic anemias; 32 myelodysplastic syndromes; 107 other anemias). When possible, patients were stratified based on the severity of anemia (Hb≤12 and Hb>12 g/dL). Parameters with significant differences (Mann Whitney U-test, p-value<0.05) and with the highest diagnostic power to identify patients with HS, PK, CDAII or MetHb (assessed by ROC curve) were combined into algorithms to discriminate each disease. Algorithms were validated on 1965 routine samples analyzed from Dec 2023 to Jan 2024, including 14 HS, 1 CDAII and 1 PKD newly diagnosed patients.

Results. All MetHb patients showed low values of NE-SFL, LY-Y, and MO-Y (mean fluorescence of NEUT, LYMPH and MONO respectively). All CDAII patients showed increased values in RDW-SD, hyperchromic (HyperHe) and macrocytic (MacroR) RBCs, and a decrease in Reticulocyte Production Index (RPI) and RET maturation indices (IRF, MFR). PKD patients showed high RET count (x10^9/L) and increased IRF combined with low values of hypochromic (HypoHe) and microcytic (MicroR) RBCs. HS patients had high RET count without equally elevated IRF along with increased MicroR and low HypoHe, leading to increased values of RET/IRF and MicroR/HypoHe ratios.

Below are the algorithms for MetHb, CDAII, PKD, HS listed by evaluation order and the diagnostic performances obtained (SE sensitivity; SP specificity; NPV negative predictive value; PPV positive predictive value). Square brackets group all the conditions that must be satisfied.

MetHb: [NE-SFL<42.0; LY-Y<57.0; MO-Y<90; NRBC<5.0%; WBC>1.00x10^9/L] (SE 1.00, SP 1.00, NPV 1.00, PPV 1.00)

CDAII: [RDW-SD>52.9; MacroR>3.8%; HyperHe>0.5%; MFR<9.5%; RPI<1.7; IRF<14.2%; Hgb<13.0; NRBC>0.0%; PLT>120x10^9/L] (SE=1.00, SP=0.99, NPV 1.00, PPV 0.92)

PKD with Hb≤12g/dL: [IRF>50%] OR [RET>95; IRF>25.0%; MicroR<1.7%;HypoHe<0.3%]; PKD with Hb>12g/dL: [RET>135; IRF>15.5%; MicroR<1.9%; HypoHe<0.2%] (SE 0.86, SP 0.99; NPV 0.99, PPV 0.78)

HS: Precondition:[RET>75; RET/IRF≥7.7] HS with Hb≤12g/dL: [RET>150; MicroR/HypoHe 1.5 to 5.0; MicroR>2.0%; NRBC<5.0%;] OR [MicroR/HypoHe>5.0; MicroR>1.5%] HS with Hb>12: [RET/IRF>15.0] OR [RET>190; RET/IRF 9.0 to 15.0; MicroR≥2.5%; MicroR/HypoHe>5.0] (SE 1.00, SP 0.96 NPV 1.00, PPV 0.87)

The excellent diagnostic performances have been confirmed on the 1965 samples used for validation; all new patients with HS, CDAII and PKD were correctly classified. The algorithms published by the other authors, tested on the 500 patients of this study, showed lower SE for both HS (0.85 to 0.92) and PKD (0.14)

Conclusions. For the first time we present a comprehensive algorithm for HS, CDAII, PKD and MetHb with excellent performance and higher diagnostic accuracy than previously published algorithms for HS and PKD. It can represent a simple, costless, and efficient first-line screening of the most common CHAs even outside reference centers, contributing to early diagnosis and better management of these patients.