From the marrow to the blood: optimising the diagnosis of iron deficiency in the setting of inflammation.

Abstract

Iron deficiency (ID) is a common condition with readily available treatment but can be challenging to diagnose. Traditional biomarkers of ID are acute-phase reactants, which complicate diagnosis in patients with co-existent inflammation. This study aimed to establish optimal biomarker diagnostic thresholds for ID diagnosis using bone marrow (BM) iron stores as the gold standard and the C-reactive protein (CRP) as an inflammatory marker. A cross-sectional study was carried out in the haematology department of a tertiary academic hospital. Patients undergoing BM biopsies for any reason were recruited for inclusion. Retrospective case finding was used to enrich the data for cases with confirmed BM ID. Laboratory markers including red cell indices, reticulocyte haemoglobin and iron studies were evaluated to establish optimal cut-offs for ID diagnosis. A CRP of >5 mg/L was used as a marker of inflammation. The study included 139 patients. Forty-two had BM ID, with a median serum ferritin (SF) of 48.5 μg/L. Ninety-six of 134 (72%) had inflammation with a CRP >5 mg/L. An SF of <80 μg/L had optimal sensitivity (69%) and specificity (94%) for ID diagnosis in the whole group (odds ratio 23.5; 95% confidence interval 4.3-129). In patients without inflammation, an SF cut-off of 80 μg/L had high sensitivity (93%) and specificity (96%). An SF <200 μg/L indicated ID in those with inflammation (sensitivity 78%, specificity 74%). A transferrin saturation of <13% in those with inflammation increased the diagnostic specificity (92%). The reticulocyte haemoglobin was unhelpful in diagnosing ID in this setting. In this hospital population, SF was the best parameter to diagnose ID, even in the presence of inflammation. The CRP was useful to identify populations with inflammation in whom higher SF thresholds could be used together with the transferrin saturation to accurately diagnose ID.