Combined cobalamin and iron deficiency anemia: a diagnostic approach using a model based on age and homocysteine assessment

Abstract

Macrocytosis, the hallmark of cobalamin/folate deficiency anemia, is frequently absent. Clinicians have to be aware of coexisting conditions that can mask the macrocytosis expression of megaloblastic anemia, especially iron deficiency. The objective of this work was to investigate the degree of overlap between iron deficiency anemia (IDA) and cobalamin deficiency and to develop a predictive model for differentiating IDA from combined deficiency. A prospective case and control study was carried out to investigate vitamin B12 and folate status in iron deficiency anemia. A total of 658 patients were recruited, 41 of whom (6.2%) were excluded. The remaining 617 subjects consisted of 130 controls and 487 with IDA. Low vitamin B12 (LB12) was considered when serum vitamin B12 was ≤200pmol/L. High serum homocysteine (Hcy) was defined by Hcy >17μM/L. A multivariate analysis (including a logistic regression) was performed to develop a diagnostic model. Low vitamin B12 levels were found in 17.8% of IDA subjects. Ten out of 11 subjects (91%) with IDA and serum vitamin B12 (B12) ≤100pmol/L showed vitamin B12 deficiency. Moreover, vitamin B12 deficiency was demonstrated in 48% of cases with IDA and B12 between 101 and 150pmol/L and in 40% with IDA and B12 between 151 and 200pmol/, respectively. As a result of multivariate logistic analysis, neutrophil counts and age predicted subjects with vitamin B12 ≤200 and Hcy >17μmol/L, [Formula: see text]. Using the age of 60 as a cutoff, sensitivity was 91% (39 out of the 43 patients with vitamin B12 deficiency and IDA were identified). In summary, low vitamin B12 was found in 18% of patients with IDA. Vitamin B12 deficiency was demonstrated in many patients with LB12 and IDA. Age over 60years was used to separate patients with combined deficiency (sensitivity 91%). Therefore, for a diagnostic purpose, serum vitamin B12 should be evaluated in IDA patients over 60years. This diagnostic model needs to be validated in a different population.