Development of a single-step plasma iron detection method

Iron deficiency anaemia (IDA) is the most common cause of anaemia worldwide that affects almost two billion people in many developing countries including Malaysia. The gold standard for identifying iron deficiency is a direct test of bone marrow iron. Bone marrow aspiration is too invasive in nature for routine use. Therefore, peripheral whole blood and serum/plasma are used to assess iron status through haematological and biochemistry tests, respectively. Iron status quantification can be limited in some areas due to high cost and lack of access to these analysers. Therefore, a low-cost and efficient technique was designed to detect iron status using human plasma. Currently, no single definitive diagnosis can assess iron status effectively except bone marrow iron. Although ferritin is a common practice, it can be confounded by inflammation and required haemoglobin level to detect iron deficiency anaemia. To devise a technique for field studies, plasma iron (PI) was chosen due to its simplicity without involving multiple steps like ferritin. Firstly, a recipe to rapidly induce iron release from human plasma was identified, comprising 720 mM citric acid, 20 mM ascorbic acid, 100 mM thiourea and 3 mM ferrozine. Next, a total of 190 samples were collected, i.e. 10 inflammation (Infla), 31 iron deficiency with and without anaemia (IDwwoA), 114 normal iron (NI) and 35 iron overload (IOL), respectively. These samples were subjected to PI screening using the freeze-dried version of the concocted recipe. By comparing the current technique (termed Prototype_PI) to autoanalyser (termed Cobas_PI), Prototype_PI and Cobas_PI across all samples were ranged from 148.3-2744.4 μg/L and 184.0-2918.0 μg/L, respectively with 72.1-157.4% of recoveries. Only nine samples were found to be beyond 80-120% of the acceptance range. Both methods correlated well with a Spearman rho coefficient of 0.967. In Passing-Bablok analysis, both methods did not differ by any constant or proportional error but has random error, with residual standard deviation (RSD) of 61.5 μg/L across all samples. The Bland-Altman’s limit of agreement (LoA) was -239.7 to 104.8 μg/L with a mean difference of -67.5 μg/L. Concordance (CCC) and intraclass correlation coefficient (ICC) of 0.980 and 0.994, respectively, indicating a good agreement between two methods. Across iron status, each group indicated good agreement with values more than 0.9 for Spearman rho coefficient, CCC and ICC. LoAs were -156.3 to 65.6 μg/L (IDwwoA), -225.8 to 86.5 μg/L (NI) and -336.0 to 129.3 μg/L (IOL) with a mean difference of -45.35 μg/L, -69.65 μg/L and -103.32 μg/L, respectively. Similarly, neither constant nor proportional error found across iron status, indicating random error contributed to the difference between both methods. As compared to Cobas_PI, Prototype_PI has a sensitivity of 87.5% (91.7%) and a specificity of 97.1% (96.8%) in diagnosis of IDwwoA in male (female), respectively. However, the ability of Prototype_PI to diagnose IOL in male (female) was reported to have lower sensitivity, i.e. 71.4% (male) and 80.8% (female) but 100% specificity, respectively. By comparing to ferritin level, both Prototype_PI and Cobas_PI found to have moderate sensitivity and specificity. This project concluded that Prototype_PI could screen PI successfully as comparable to Cobas_PI for diagnosis of IDwwoA but less accurate in IOL screening. Further justification has to be done by performing double-blind study