We read with interest the article by Bultink et al reporting that ferritin <50 μg/liter in combination with soluble transferrin receptor (sTfR) >2.50 mg/liter was 100% sensitive and 97% specific for the detection of iron deficiency in patients with rheumatoid arthritis (RA) (1). The difficulty in distinguishing anemia caused by chronic disease from that due to iron deficiency in patients with inflammatory arthritis has been much discussed. Examination of bone marrow stained with Perls' Prussian blue remains the gold standard, and some authors have advocated use of this procedure in the initial investigation of severe anemia in RA (2). However, bone marrow aspiration is both time-consuming and uncomfortable for patients.
Once a diagnosis of iron deficiency is established, an underlying cause must be sought (3). The main purpose of this search is to exclude life-threatening diseases such as peptic ulcer and gastrointestinal carcinoma. Such investigations are often incomplete (4), however, and even when they are thorough, the cause of anemia may never be established, and management may remain unaltered.
We undertook a review of the laboratory results and case notes of patients with inflammatory arthritis who, over the past 3 years, had undergone bone marrow examination to determine the nature of their anemia. No patient was taking iron, and causes of anemia other than chronic disease and iron deficiency had been excluded. Results of the last studies of hemoglobin (Hgb), mean corpuscular volume (MCV), erythrocyte sedimentation rate, and serum ferritin performed prior to marrow aspiration were recorded. Iron deficiency was diagnosed in cases in which stainable marrow iron was absent.
The Rheumatology Department at King's Mill Centre provides care for ∼2,000 patients with inflammatory arthritis. Fourteen of these patients had undergone bone marrow aspiration for the investigation of anemia (12 women, 2 men, median age 64 years, mean Hgb value 9.9 gm/dl). The specificity of serum ferritin (<50 μg/liter) was 100%, but 2 of 4 patients with iron-deficient bone marrow had serum ferritin levels >50 μg/liter (sensitivity 50%). The diagnosis of iron deficiency did not determine treatment in either of the 2 patients with iron-deficiency anemia whose serum ferritin levels were >50 μg/liter.
In the first case, a 34-year-old woman with RA developed intermittent normocytic anemia. Initially, no clinical features suggested a cause for iron deficiency. Poor compliance with disease-modifying antirheumatic drugs had led her to rely on nonsteroidal antiinflammatory drugs (NSAIDs). The onset of indigestion prompted performance of an upper gastrointestinal endoscopy and small bowel biopsy; results from both procedures were normal. No further changes in management were made subsequent to bone marrow aspiration.
The second case involved a man with longstanding RA complicated by Felty's syndrome and prior hemorrhage from a duodenal ulcer in whom normocytic anemia developed during treatment with methotrexate and diclofenac. Because he was experiencing dyspepsia, the latter drug was discontinued and was substituted with a proton pump inhibitor before bone marrow aspiration. He was then hospitalized for septic arthritis, and his anemia subsequently resolved.
A third case illustrates the development of iron deficiency shortly after examination of marrow aspirate revealed normal iron stores. The patient was a 27-year-old woman with psoriatic arthropathy. Analysis of bone marrow aspirate revealed normal iron stores (Hgb 10.3 gm/dl, MCV 78 fl, and serum ferritin 29 μg/liter). Six months later, these values were unchanged, except the level of serum ferritin had fallen to 7 μg/liter, consistent with iron deficiency. No clinical features suggested blood loss or malabsorption.
We conclude that the diagnosis of iron deficiency remains difficult in patients with inflammatory arthritis. Introduction of an estimation of sTfR into routine clinical practice would increase the specificity of criteria based on serum Hgb and ferritin <50 μg/liter but may be helpful in only a small group of patients with ferritin levels in the normal range. We found the gold standard of bone marrow aspiration to be both practical and acceptable. Increasing use of proton pump inhibitors or cyclooxygenase 2–specific antiinflammatory drugs in place of nonselective NSAIDs should reduce the incidence of iron deficiency in these patients and consequently the number of bone marrow examinations needed. Determination of sTfR levels may prove helpful in the continued monitoring of patients in whom anemia of chronic disease has been diagnosed. A definitive diagnosis of iron deficiency does not always lead to changes in treatment, especially in premenopausal women, and a prospective multicenter study would be required to determine the cost–benefit ratio of screening for iron deficiency in this group.