Algorithm of differential diagnosis of anemia involving laboratory medicine specialists to advance diagnostic excellence

Reticulocyte production index

The automated hematology analyzers measure hemoglobin concentration as part of the complete blood count (CBC) panel. We follow the WHO guidelines in defining anemia as stated in the Introduction. At the same time or as a next step, reticulocyte number should be determined (Figure 1). This is a significant difference between our algorithm and some other guidelines, where reticulocyte count may not or may be considered to be measured later [8], [9], [10], [11], [12], [13], [14], [15]. Due to the simplicity, low cost, and importance in all further anemia subgrouping, the absolute reticulocyte number should be assessed, and the derived reticulocyte percentage (Ret%) – which is sometimes wrongly referred to as reticulocyte count – and reticulocyte production index (RPI) should be calculated at this step. Ret% is the percentage of the absolute reticulocyte number and the red blood cell number. RPI is a marker adjusted for the level of anemia and is a more accurate reflection of erythropoiesis. RPI is calculated from reticulocyte percentage (Ret %), hematocrit (Hct), and a maturation time, based on the

equation [16], where maturation time is 1.0 day for Hct 35–40 %, 1.5 days for Hct 25–35 %, 2.0 days for Hct 15–25 %, and 2.5 days for Hct < 15 % [17].

Note, that the bone marrow response to anemia requires days to two weeks to increase reticulocyte number and RPI.

Cut-off value for RPI for optimal bone marrow response of accelerated RBC production varies in the literature between 2 and 3 %. We found two references for cut-off value of 2 % [8, 9], and two other references for 2.5 % [18, 19]. Five references suggested that optimal bone marrow response is at RPI >3 % while inadequate response is at RPI <2 %, leaving an undefined zone of RPI between 2 and 3% [20], [21], [22], [23], [24]. RPI cut-off is suggested at 3 % by three references [25], [26], [27]. In order for the algorithm to be sensitive for hemolysis, and to avoid an undefined “grey zone” of RPI, we accepted the cut-off value of 2 %. We also included a “runaway ramp” for cases of RPI above 2 % when causes listed for cases below 2 % of RPI should also be considered. Based on RPI, continue with Figure 2 (≤2 %) or Figure 3 (>2 %).

Anemia with RPI ≤ 2 %

If the RPI is ≤2 %, further classification of anemia is based on the mean red cell volume (MCV) (Figure 2). If MCV is <80 fL, the anemia is microcytic. Since iron deficiency is the most common cause of microcytic anemias, it is advised to continue the investigation towards iron deficiency anemia (IDA). IDA is likely if ferritin concentration is low, serum iron concentration is low, transferrin concentration is high, and transferrin saturation is low. As serum iron concentration is influenced by several physiological factors (diurnal rhythm, monthly cycle in women, stress, etc.), serum iron concentration alone is not acceptable to confirm iron deficiency. As the reference range of these parameters may vary according to age, sex, race, and method, results outside the reference range used in each laboratory are indicative. We also note that neither low ferritin with normal transferrin saturation, nor borderline low ferritin with low transferrin saturation exclude IDA [1].

If iron deficiency is not confirmed, calculation of the Mentzer index is recommended; automatic calculation can be set up in the laboratory information system (LIS). The Mentzer index is an easily accessible quotient of the MCV (in fL) and RBC count (in millions per microliter or × 10¹²/L). If the Mentzer index is below 13, hemoglobin electrophoresis or hemoglobin analysis by HPLC should be performed to measure the HbF and HbA₂ ratios of total hemoglobins. If the HbF >1 % or HbA₂ >3.5 %, beta thalassemia is likely, and a genetic background investigation is warranted. For this purpose, sequencing of the most frequently abnormal regions of the beta globin gene in the actual population is recommended. If HbF ≤1 % and HbA₂ ≤3.5 %, alpha thalassemia and other hemoglobinopathies are considered. Genetic testing is also recommended to identify these cases.

If the Mentzer index is ≥13, serum creatinine concentration should be determined. Based on serum creatinine, age, and gender, eGFR is calculated [28]. If eGFR is ≥60 mL/min/1.73 m², chronic inflammation or other chronic diseases (e.g., autoimmune disease, chronic endocrine disease) or malignancy are likely to contribute to anemia. In such cases, it is recommended to determine the soluble transferrin receptor (sTfR) concentration in serum, from which the sTfR/ferritin index can be calculated (automatically by the LIS) by dividing the sTfR concentration (mg/L) by the logarithm of the ferritin concentration (ng/mL). The sTfR/ferritin index is superior to sTfR (AUC 0.87 vs. 0.74, p<0.0001) [29]. The suggested cut-off value varies between 1.03 and 1.8 in the literature, we suggest the average 1.5 for cut-off value [29], [30], [31], [32], [33]. If the sTfR/ferritin is >1.5, iron deficiency associated with chronic disease is likely; if it is ≤1.5, anemia associated with chronic disease without iron deficiency is likely. After excluding chronic disease, chronic inflammation and solid tumor, bone marrow examination should be performed to search for sideroblastic anemia [34]. In this case, a smear of bone marrow aspirate should be stained with Prussian blue for ring sideroblasts. Ring sideroblasts are erythroblasts with at least five siderotic granules in the cytoplasm, occupying at least one-third of the circumference of the nucleus [34]. The presence of ring sideroblasts confirm sideroblastic anemia. In the absence of ring sideroblasts, anemia due to other bone marrow failure is likely.

If eGFR is <60 mL/min/1.73 m², renal failure is possible. Moderately decreased eGFR values are recommended to be confirmed by cystatin-based eGFR calculation [28]. In this case, check for iron deficiency by determining the proportion of hypochromic RBCs or the reticulocyte Hgb (CHr), depending on the type of hematology analyzer. Iron deficiency anemia associated to chronic kidney disease can be confirmed if the hypochromic RBC is >6 % (HYPO% on ADVIA Siemens analyzer), or reticulocyte Hgb is <29 pg (all analyzers) [10], or transferrin saturation is <20 % and ferritin is <100 ng/mL. If these conditions are not met, iron deficiency is unlikely, and the diagnosis is simply anemia associated to chronic kidney disease.

The first step in the investigation of normocytic anemia (MCV 80–100 fL) is the measurement of serum creatinine and calculation of eGFR. If eGFR is <60 mL/min/1.73 m², the renal anemia of the patient should be investigated. If the patient with normocytic anemia has an eGFR ≥60 mL/min/1.73 m², the patient should be assessed for chronic inflammation, chronic disease (e.g., autoimmune disease, chronic endocrine disease), or malignancy. If any of them is present, consider anemia is due to iron deficiency associated with the chronic disease and investigate as described above. If chronic inflammation, chronic diseases, and malignancy are excluded, we evaluate whether iron deficiency is combined with vitamin B12/folic acid deficiency. In combined iron and B12/folic acid deficiency, the ferritin concentration is low, the serum iron concentration is low, the serum transferrin concentration is high, the transferrin saturation is low, and the vitamin B12 or folic acid concentration is low. If this combined deficiency cannot be confirmed, evaluate the bone marrow smear with panoptic staining assessing for infiltrating malignant cells (leukemias, multiple myeloma, bone marrow metastasis of solid tumor) and hypoplasia/aplasia (aplastic anemia, pure red cell aplasia). For a more precise indication of a bone marrow test, it is advised to look for blasts in the peripheral smear, which is readily available from the original blood sample. However, circulating tumor cells of a solid tumor are so few that they are not detectable by a standard peripheral blood smear analysis.

In macrocytic anemia (MCV >100 fL), serum vitamin B12 and folic acid concentrations are first to be measured. If the folic acid concentration is close to the lower limit of the reference range, measure the folic acid content of the red blood cells, as this may be the only detectable sign of folic acid deficiency. In case of borderline low vitamin B12, measurement of homocysteine and methylmalonic acid (MMA) may be suggested, elevation of either suggests B12 deficiency [35]. However, borderline low vitamin B12 levels usually do not result in anemia. If vitamin B12 and folic acid deficiencies are excluded, check for recently recommended/administered medications of the patient that may cause macrocytic anemia, such as antagonists of folic acid (methotrexate), purine (6-mercaptopurine, azathioprine, acyclovir) and pyrimidine (fluorouracil, zidovudine), ribonucleotide reductase inhibitors (cytosine arabinoside, hydroxyurea), alkylating agents (cyclophosphamide) and tyrosine kinase inhibitors (sunitinib, imatinib). In the absence of such medications, evaluate the shape of macrocytes in peripheral smear. Causes of round macrocytes include liver disease, alcoholism, or thyroid disease. If these are ruled out, anemia is due to other bone marrow failure. If the macrocytes in the peripheral smear are oval, bone marrow tests, including morphological examination of the smear and MDS-directed cytogenetic and FISH studies may reveal and confirm MDS. If MDS is ruled out, macrocytic anemia is associated with other bone marrow failure, including acquired forms such as primary myelofibrosis or inherited diseases (e.g., Fanconi anemia, Diamond-Blackfan anemia).

Anemia with RPI >2 %

If RPI is greater than 2 % (Figure 3), check patient’s medical history for recently administered new medication (iron, vitamin B12, folic acid) for a previously confirmed deficiency anemia. In these cases, peripheral reticulocytosis is a normal bone marrow response. However, in the absence of supplementation, look for signs of intravascular hemolysis. Common markers for intravascular hemolysis are low haptoglobin, high LDH, and elevated unconjugated bilirubin levels; LDH and bilirubin being less specific for hemolysis. Haptoglobin alone showed high sensitivity and specificity (83 and 96 %, respectively) in distinguishing hemolytic and non-hemolytic conditions, providing 87 % probability of predicting hemolytic disease when the serum haptoglobin level falls below 25 mg/dL (0.25 g/L) [36]. Combining LDH and haptoglobin only slightly increases sensitivity (92 %) and does not increase specificity (90 %) for ruling out hemolysis [37]. Low haptoglobin may result from hepatic insufficiency and in the rare cases of congenital ahaptoglobinemia. Therefore, we suggest that a combination of LDH and haptoglobin is only necessary with a history of hepatic insufficiency.

If the results do not suggest hemolysis, the guiding diagnosis is acute anemia after blood loss, with a time lag of 1–2 days after hemorrhage [6].

If hemolysis is confirmed, a urinalysis with urine dipstick and sediment examination should be performed to detect hemoglobinuria. Hemoglobinuria is defined as detectable hemoglobin on the urine test strip exam without red blood cells in the urine sediment. Hemoglobinuria increases the risk of acute renal failure. Rarely, hemoglobin casts are present.

In addition to urinalysis, hemolysis necessitates further testing including peripheral smear staining and evaluation, and a Coombs test (direct antiglobulin test, DAT) with polyvalent reagent. A positive Coombs test with polyvalent reagent confirms immune hemolysis, and the Coombs test should also be performed with the monovalent reagent. If IgG antibodies and C3d complement are detected by the DAT, and spherocytes are detected in the smear, warm antibody autoimmune hemolytic anemia (w-AIHA) is confirmed. Antibodies of warm AIHA may develop idiopathically (primary) or may be secondary due to transfusion complication, drugs (fludarabine, interferon-alpha), autoimmune disease (SLE), lymphoproliferative diseases (CLL, NHL), allogeneic stem cell transplantation, organ transplantation, or viral infections (hepatitis, CMV). If the Coombs test with monovalent reagent shows no IgG coverage, but C3d coverage of RBCs, and the RBCs in the smear are agglutinated, then cold agglutinin is the cause of hemolysis, and the cold agglutinin titer should be determined. This phenomenon may occur in IgM gammopathies, Waldenström macroglobulinemia, Mycoplasma infection, EBV, CMV, Chlamydia infection, chronic lymphoproliferative B cell disease, lymphoma, paroxysmal cold hemoglobinuria.

If the Coombs test with polyvalent reagent provides a negative result, calculate the Mentzer index. If the Mentzer index is <13, screening for thalassemia is recommended. For Mentzer index ≥13, further testing is based on the dominant pathological morphology of RBC in the peripheral smear.

In the presence of spherocytes in the smear and the patient has a history of burns or large skin deficiencies, the diagnosis is hemolysis associated with thermal injury. If medical history excludes thermal injury, congenital spherocytosis is suspected, and a flow cytometric test with eosin-5-maleimide staining and/or an osmotic resistance test of red blood cells is recommended. In case of a positive test result, it is reasonable to test for mutations in the suspected genes using molecular biology methods. If both tests are negative, check medical history for snakebite or insect bite to confirm hemolysis due to these causes. If medical history does not include bites, we should consider other diseases causing spherocytosis in addition to Coombs negativity, such as autoimmune hemolysis (with warm or cold antibodies), hemolytic complication after transfusion, AB0 neonatal hemolytic disease, bartonellosis, C. perfringens infection, Rh-null phenotype.

In case of the high rate of elliptocytes or stomatocytes in the smear, hereditary elliptocytosis or stomatocytosis is considered and genetic testing is warranted. If contracted, ‘bite’ or ‘blister’ RBCs and Heinz bodies in the RBCs are identified in the smear (with supravital staining), a glucose-6-phosphate dehydrogenase or glutathione cycle enzyme deficiency is likely; targeted genetic testing can confirm the enzyme deficiency.

The presence of sickle cells in the smear suggests sickle cell anemia and hemoglobin electrophoresis or HPLC can determine the proportion of the abnormal variant of hemoglobin. However, regardless the extent of the abnormality, targeted genetic testing with sequencing of the beta-globin gene is required to detect the characteristic mutation.

If the smear shows polychromasia without other reproducible morphological abnormalities, consider infection or drug exposure-induced hemolysis; the underlying enzyme defects (pyruvate kinase, hexokinase, or other rare enzyme deficiencies) can be verified by targeted genetic testing.

If hemolysis was not initiated by infection or drug administration, test for the absence of the glycosylphosphatidylinositol (GPI) anchor molecule by flow cytometry in both white blood cells and RBCs. If the absence of the GPI anchor is confirmed, the diagnosis is paroxysmal nocturnal hemoglobinuria. If the flow cytometry test is negative, other causes such as liver disease, kidney disease, hemolysis due to extreme physical activity (exercise-induced [march] hemoglobinuria) should be considered.

If unusual RBC inclusions are detected in the smear, an infectious background is likely: most common disorders are malaria, babesiosis, and Bartonella infection.

Coarse dense basophilic stippling in several RBCs suggest lead poisoning; measure serum lead level. Zinc protoporphyrin levels may indicate chronic lead toxicity. If lead poisoning is excluded, consider 5′ nucleotidase enzyme deficiency, which can be confirmed by targeted molecular biology testing.

If the proportion of fragmented cells (schistocytes) in the smear is over 1 %, consider mechanical damage to the red blood cells by cardiac protheses or arterial grafts. Mechanical damage to the RBCs due to increased shear stress is the most widely occurring etiology of cardiac prostheses-related hemolytic anemia (CPHA) [38]. In the absence of such implants, malignant hypertension and vasculitis should be considered. Once these have been excluded, consider and investigate for thrombotic microangiopathies: disseminated intravascular coagulation (DIC), thrombotic thrombocytopenic purpura (TTP), hemolytic uremic syndrome (HUS), atypical hemolytic uremic syndrome (aHUS), hemolysis, elevated liver enzymes and low platelets (HELLP) syndrome.