Pb2518: optimizing the detection of 2,3- diphosphoglycerate in dried blood spots of patients with sickle cell disease: untargeted metabolomics within the genomed4all project

Topic: 26. Sickle cell disease Background: Increased levels of 2,3-diphosphoglycerate (2,3-DPG), a metabolite in red blood cells affecting the oxygen affinity of hemoglobin, may be related to the clinical variability observed in sickle cell disease (SCD). It is well known that untargeted metabolomics obtains more extensive information about the total metabolome than quantitative methods, leading to increasing interest from research groups. Currently, from 1000 patients clinical data, untargeted metabolomics- and other omics- are being collected within the GenoMed4All project to develop Artificial Intelligence based deep learning algorithms to improve prediction of disease course in SCD. We previously showed that 2,3-DPG can be detected by untargeted metabolomics using direct infusion high resolution mass spectrometry (DI-HRMS). Moreover, we showed that by correcting for detection biases based on hematocrit (Ht) by adding (1-HtHt)to outcomes, correlations between untargeted and targeted quantitative detection of 2,3-DPG improved (r=0.391; p=0.001 to 0.569; p<0.001). This suggests that Ht is an important factor when detecting 2,3-DPG by untargeted metabolomics. Aims: This study aims to improve our understanding of the effect of Ht in untargeted metabolomics to improve the correlation between untargeted and targeted detection of 2,3-DPG. Methods: 2,3-DPG was quantified by liquid chromatography mass spectrometry in snap frozen blood samples and detected in dried blood spots (DBS) by DI-HRMS of 35 patients with SCD and 29 healthy controls. Furthermore, DBS with a Ht between 2 and 90% were prepared from four healthy controls by removing the buffy coat and mixing red blood cells with different volumes of plasma. 20 internal standard (IS) metabolites, serving as quality controls, and 2,3-DPG were detected by DI-HRMS. Statistical analysis were performed by Spearman’s correlation coefficients (SPSS v27). Results: In DBS with Ht ranging from 2-90%, no linearity was observed with the detection of 2,3-DPG using untargeted metabolomics. A bell-shaped curve starting with a positive correlation followed by a negative correlation with Ht was observed (Fig. 1A, r=0.060, n.s.). Additional factors besides Ht thereby affect the detection of 2,3-DPG. All IS metabolites negatively correlated with Ht (-0.681 to -0.955; p<0.001), indicating that with increasing Ht other factors also compete with the detection of IS metabolites. By correcting the peak intensities of 2,3-DPG with the sum of all IS metabolites the correlation with Ht improved (r=0.589, p=0.002). Furthermore, adding this correction of the sum of IS metabolites to the previously established correction factor (1-HtHt)resulted in further improvement of the correlation between untargeted and targeted detection of 2,3-DPG (Fig. 1B, r=0.610, p<0.001). Summary/Conclusion: Detection of 2,3-DPG by untargeted metabolomics is biased by Ht levels in a multifactorial way. This can be improved by correcting for Ht and ion suppression based on the sum of 20 IS metabolites. The established improvement and further amelioration of the correlation between untargeted and targeted 2,3-DPG detection likely add value to the use of untargeted metabolomics within the field of SCD, and potentially other diseases characterized by hemolytic anemia. Funding/acknowledgement: Horizon2020, GenoMed4All, Agios Pharmaceuticals Inc., ERN-EuroBloodNet.Keywords: Hematocrit, Hemoglobinopathy, Sickle cell disease, Hemolytic anemia