Red Cell Volume Distribution Width as Another Biomarker

To the Editor,

We congratulate Nadar and Shaikh for their excellent review of the use of biomarkers in heart failure (HF), as well as the new biomarkers currently under investigation.[1]

Red blood cell distribution width (RDW) is a measure of the heterogeneity of distribution of red blood cell size. A high RDW implies a large variation in red blood cell (RBC) size (anisocytosis) and a low RDW implies a more homogeneous population of RBC sizes. RDW is routinely assessed as part of complete blood count and is calculated as RDW = (standard deviation/MCV) x 100, with reference values approximately 11–15%.[2] RDW in combination with mean corpuscular volume (MCV) has been used for the classification of anaemias.[3] RDW elevation is associated with conditions of impaired haematopoiesis, such as nutritional deficiencies (iron, folate, vitamin B12), some haemoglobinopathies, myelodysplastic syndrome, myelophthisic anaemia (e.g. neoplastic metastases to bone marrow) and liver impairment, as well as in conditions of increased red cell destruction, such as haemolysis, or when different populations of RBC are present, such as after blood transfusion.[4]

In recent years, many community cohort studies have shown that an increment in RDW is associated with all-cause mortality. The UK Biobank study (n=503,325; HR 3.10, 95% CI [2.57–3.74])[5] and the National Health and Nutrition Examination Survey (n=8175) showed that for every 1% increment in RDW, all-cause mortality risk increased by 22% (HR 1.2, 95% CI [1.15–1.30], p<0.001), and that even in non-anaemic participants it remained associated with mortality.[6] In addition, patients with high RDW are 1.8 times more likely to develop adverse events after cardiac surgery (OR 0.55, 95% CI [0.365–0.852], p=0.007).[7] In another cohort (n=16,631), after non-cardiac surgery, the area under the curve was 0.761 (95% CI [0.736–0.787]) using a cut-off value of RDW 15.7% with a specificity of 89.3% and a negative predictive value of 99% for predicting 30-day mortality.[8] In contrast, in another cohort (n=217,567) RDW >14% was associated with metabolic syndrome (OR 1.14; 95% CI [1.07–1.21]; p<0.0001).[9]

High RDW is an independent predictor for the development of anaemia during hospitalisation due to acute MI in patients without previous anaemia.[10] Moreover, RDW >14.9% is associated with increased major bleeding risk (HR 2.41, 95% [CI 1.15–5.02], p=0.02) in non-ST-segment elevation MI (NSTEMI). The addition of RDW to the Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA guidelines (CRUSADE) bleeding score had a significant integrated reclassification improvement of 10% (95% CI [6–19]; p=0.02).[11] Similarly, another study showed that RDW was a predictor of major bleeding and that with the addition of RDW as a continuous variable to the National Cardiovascular Data Registry risk model, net reclassification improvement increased by 17.3% (95% CI [6.7–28]; p=0.02).[12]

In the Cholesterol and Recurrent Events (CARE) study of patients with hyperlipidaemia and history of MI, baseline RDW was associated with increased risk of all-cause death per percent increase in RDW, and RDW in the highest quartile was associated with MI, stroke and HF.[13] In the Candesartan in Heart Failure: Assessment of Reduction in Mortality and Morbidity (CHARM) program (n=2,679) RDW was associated with morbidity and mortality (HR 1.17 per 1 SD increase, p<0.001) and in the Duke Databank (n=2,140) RDW was associated with all-cause mortality (HR 1.47 per 1-SD, p<0.001) in patients with advanced heart failure.[14]

Secondary analysis from the Justification for the Use of Statins in Prevention (JUPITER) trial revealed that RDW was associated with all-cause mortality.[15] In the United Investigators to Evaluate Heart Failure (UNITE-HF) registry, elevated troponin T was an independent predictor for all-cause mortality and hospitalisation for HF, and detectable troponin T was directly and independently related to increasing RDW.[16]

In the Pro-Brain Natriuretic Peptide Investigation of Dyspnea in the Emergency Department (PRIDE) registry, in patients with complaints of dyspnoea to the emergency department, RDW was independently associated with haemoglobin, use of loop diuretics and beta-blockers on presentation but not with nutritional deficiencies; in a multivariable analysis, RDW was a significant independent predictor of 1-year mortality and haemoglobin was not.[17] In patients with acute HF, those with high RDW received loop diuretics and oral anticoagulation more often, and high RDW was associated with increased all-cause mortality in patients with preserved left ventricular ejection fraction (LVEF).[18] In patients with acute HF, baseline RDW level was associated with a higher risk of death in long-term follow-up regardless of haemoglobin levels and anaemia status.[19]

In a study of patients with systolic HF, researchers analysed biomarkers of ineffective erythropoiesis, inflammation and undernutrition.[20] Patients with >15.2% RDW had low levels of iron, transferrin saturation, prealbumin, albumin, total cholesterol and estimated glomerular filtration rate, and high levels of soluble transferrin receptor, erythropoietin, interleukin-6 (IL-6), tumour necrosis factor-alpha receptor (TNF-R) I and TNF-R II, increased RDW was a predictor of all-cause mortality. The correlations between high RDW with inflammation, ineffective erythropoiesis, undernutrition and impaired renal function support the understanding of why high RDW is associated with adverse outcomes in HF.[20] The Study of Anemia in a Heart Failure Population (STAMINA-HFP) showed high RDW was predictive of mortality and hospitalisation. In addition, increasing RDW correlated with decreasing haemoglobin, increasing IL-6 and impaired iron mobilisation.[21]

In contrast, RDW is a parameter with a sensitivity of 94% for iron deficiency, and an RDW value within the reference interval can be used to exclude iron deficiency in cases in which the serum ferritin concentration does not accurately reflect the iron stores owing to severe tissue damage, as in inflammation or malignancy.[22] In the Ferinject Assessment in Patients with Iron Deficiency and Chronic HF (FAIR-HF) study, a subanalysis revealed that high RDW was associated with decreased transferrin saturation and increased C-reactive protein, and that treatment with IV ferric carboxymaltose in iron-deficient chronic HF patients decreased RDW (Table 1).[23]

Table 1: Studies Related to Red Cell Volume Distribution Width and its Association with Heart Failure and All-cause Mortality.

Author	Cut-off Value	Study Method and Setting	Results and Conclusion
Felker et al. 2007[14]	RDW 15.2	CHARM cohort (n=2,679) and Duke Databank cohort (n=2,140), endpoint all-cause mortality. Patients with chronic HF with reduced and preserved LVEF. Median follow-up 34 months	CHARM death or HF hospitalisation: n=952, mean RDW 15.2. Patients without death or HF hospitalisation had a mean RDW of 14.4. The multivariable model showed RDW was predictive of adverse outcomes (HR 1.17 per 1 SD increase, 95% CI [1.10–1.25]; p<0.0001) Duke Databank death: n=368. The multivariable model showed RDW remained associated with mortality (HR 1.29 per 1 SD increase, 95% CI [1.16–1.43]; p<0.0001)
Tonelli et al. 2008[13]	RDW ≥13.8	CARE study (n=4,111) Patients with hyperlipidaemia and a history of MI. Endpoint all-cause mortality and cardiovascular events. Median follow-up 59.7 months	RDW was associated with an increased risk of all-cause death (n=376; adjusted HR per % increase in RDW 1.14; 95% CI [1.05–1.24]; p=0.002) The adjusted risk of mortality in the highest quartile of RDW ≥13.8 was 1.78 (95% CI [1.28–2.47]; p=0.002) compared with the referent group (first quartile RDW 10.9–12.6%) RDW highest quartile had adjusted HR for experiencing MI of 1.43 (95% CI [1.03–1.99]; p=0.033) compared with the RDW lowest quartile RDW highest quartile had adjusted HR for experiencing the composite outcome of coronary death or nonfatal MI of 1.56 (95% CI [1.17–2.08]; p=0.001) compared with those with the lowest quartile Risk of stroke was increased in the highest quartile compared with the lowest quartile (adjusted HR 2.58, 95% CI [1.47–4.55]; p=0.004)
Pascual-Figal et al. 2009[19]	RDW >14.4	Consecutive patients hospitalised with acute HF (n=628). Endpoint all-cause mortality and anaemia status. Follow-up 38.1 months	Patients who died (n=209) had higher RDW 15 (13.8–16.1) versus 14.2 (13.3–15.3); p<0.001 and lower haemoglobin 12.3 ± 1.77 versus 12.8 ± 1.76; p=0.001. Multivariable Cox proportional hazards model RDW remained a risk factor (per %, HR 1.072, CI 95% [1.023–1.124]; p=0.004). Patients with RDW >14.4 (above the median) had lower survival time (log-rank <0.001) and higher risk of death in the long-term follow-up (HR 1.89, 95% CI [1.40–2.55]; p<0.001). Higher RDW was associated with higher risk of death in patients with anaemia (n=263) (per %, HR 1.057, 95% CI [1.006–1.112]; p=0.029) and without anaemia (per %, HR 1.287, 95% CI [1.147–1.445]; p<0.001)
Förhécz et al. 2009[20]	RDW ≥15.2	Patients with systolic HF (n=195). Primary endpoint all-cause mortality, hospital readmission due to worsening HF, assessment of 19 biochemical variables. Follow-up median of 14.5 months. T1 RDW ≤13.9%, T2 >13.9 – <15.2% and T3 ≥15.2%.	Died (n=43). RDW was higher in patients who died 15.9 (14.5–17.6) compared with those who lived 14.3 (13.6–15.3); p<0.001. In a multivariate model adjusted for NT-proBNP and other clinical covariates of HF death, RDW was an independent predictor of all-cause mortality (HR 1.61 per 1 SD increase; p<0.0001) Highest RDW T compared with lowest RDW T had high: soluble transferrin receptor 5.9 nmol/l (4.7–7.2) versus 2.3 (2.7–4.3; p<0.001); EPO 14.1 U/ml (7.9–25.7) versus 8.9 (4.9–15.2; p=0.002); IL-6 14.59 pg/ml (8.52–25.32) versus 6.62 (3.88–12.35, p=0.001); TNF-RI 6.93 ng/ml (4.22–10.68) versus 4.61 (3.42–6.69; p=0.007); TNF-RII 5.07 ng/ml (3.75–6.68) versus 3.4 (2.34–4.57; p<0.001). Also had low iron 10 μmol/l (6.64–13.7) versus 15.4 (10.9–19.4), transferrin saturation 16% (10–20) versus 23% (18–28), prealbumin 0.18 g/l (0.14–0.24) versus 0.26 (0.21–0.30, p<0.001), eGFR 55 (38–77.5) versus 75 (59–95), albumin 39 g/l (36–42) versus 43 (40.5–45; p<0.001), total cholesterol 3.62 mmol/l (2.95–4.14) versus 4.21 (3.81–5.31; p<0.001)
Adams et al. 2010[16]	RDW 15.2	UNITE-HF biomarker registry (n=254). Anaemia in outpatients with HF. Follow-up 1.9 ± 0.9 years. Endpoint all-cause mortality, hospitalisation due to HF, assessment of troponin T and indices of haematological function. Detectable troponin T (n=39). Median detectable troponin T 0.042 ng/ml; 59% (n=23) had values >0.03 ng/ml	Crude 1-year mortality 11.9% (n=55). Elevated troponin T was an independent predictor for all-cause mortality and all-cause hospitalisation for HF (HR 3.72, 95% CI [2.10–6.59], p<0.001) and (HR 3.88, 95% CI [2.43–6.19]; p<0.001) Direct relationship between RDW and elevated troponin T (RDW per unit increase OR 1.51, 95% CI [1.21–1.89], p<0.001). Patients with detectable troponin T compared with undetectable troponin had higher RDW 15.2 (14.2–17.2) versus 13.9 (13–15.1); p<0.001, had lower haemoglobin 11.9 (11.3–13.8) versus 13.6 (12.4–14.9); p=0.002, iron 55 μg/dl (44–72) versus 72 (50–99) p=0.009 and total iron binding capacity saturation 16.5 (12.5–21.5) versus 23 (15–30); p=0.005
Van Kimmenade et al. 2010[17]	RDW ≥14.9	PRIDE study (n=205). Consecutive patients with complaints of dyspnoea to the emergency department. Endpoint all-cause mortality. Follow-up of 1 year. RDW quartiles (Q): Q1 <13.8%, Q2 13.8–14.8%, Q3 14.9–16.2% and Q4 >16.2%	RDW was independently associated with haemoglobin (p<0.0019), the use of loop diuretics (p=0.006) and beta-blockers (p=0.015) on presentation, but not with folate ((p=0.59) or vitamin B12 deficiencies (p=0.99), recent transfusion (p=0.56) or C-reactive protein (p=0.57). Log-transformed RDW values independently predicted mortality (HR 1.03, 95% CI [1–1.06]; p=0.04), died (n=63). In a multivariable Cox’s proportional hazards analysis RDW was an independent predictor of 1-year outcome in acute HF (HR 1.03 per 1% increase in RDW; 95% CI [1.02–1.07]; p=0.04)
Van Craenenbroaeck et al. 2013[23]	RDW 14.7	Subanalysis of FAIR-HF study (n=415). Endpoint assess NYHA functional class and 6MWT after intravenous iron repletion. Follow-up 24 weeks	Baseline RDW was higher in anaemic RDW 15.2 (14–16.8%) versus non-anaemic patients RDW 14.2% (13.4–15.4%) p<0.0001. NYHA class III had higher RDW 14.75 (13.8–16.5) versus NYHA class II RDW 14 (13.2–15%) p<0.0001. Treatment with ferric carboxymaltose decreased RDW leves in patients with elevated base line C-reactive protein levels ≥3 mg/l (n=147, p for interaction 0.007), whereas in patients with normal baseline C-reactive protein levels this effect was not significant (p for interaction 0.17). The increase in 6MWT distance was significantly correlated with a decrease in RDW (r=−0.25; p<0.0001)
Horne et al. 2015[15]	Women: RDW 15.3 Men: RDW 14.9	Secondary analysis from JUPITER (n=17,197). Patients without a history of cardiovascular disease. Endpoint all-cause mortality. Median follow-up 1.9 years. Ts in women (n=6,568): RDW T1 13.1 (12.7–13.5); T3 15.3 (14.7–16.1) T in men (n=10,629): RDW T1 12.9 (12.6–13.2); T3 14.9 (14.4–15.8)	In a multivariable analysis, the RDW was associated with all-cause mortality (T3 versus T1: HR 1.46, 95% CI [1.12–1.89]; T2 versus T1: HR 0.97, 95% CI [0.74–1.28]; p=0.002)
Sotiropoulos et al. 2016[18]	RDW >16.6	Consecutive patients hospitalised due to acute HF without acute coronary syndrome or need of intensive care (n=402). Endpoint all-cause mortality at 1 year after admission. RDW Qs: Q1 12.2–14.2, Q2 14.3–15.2, Q3 15.3–16.6 and Q4 16.7–32.1	Increasing RDW quartile was an independent predictor of 1-year all-cause mortality (HR 1.66, 95% CI [1.02–2.8]), died (n=114) Kaplan–Meier analysis including all patients showed a graded increased probability of mortality with increasing quartile of RDW (p=0.0004) Patients with LVEF ≥50% showed a graded increased probability of mortality with rising RDW quartile (p=0.0195)

6MWT = 6-minute walk test; eGFR = estimated glomerular filtration rate; EPO = erythropoietin; HF = heart failure; IL-6 = interleukin-6; LVEF = left ventricular ejection fraction; Q = quartile; RDW = red blood cell distribution width; T = tertile; TNF-RI = tumour necrosis factor-alpha receptor I; TNF-RII = tumour necrosis factor-alpha receptor II.

Belonje et al. showed that in patients hospitalised for HF, those with higher erythropoietin levels at baseline were independently related to increased mortality at 18 months (HR 2.06, 95% CI [1.4–3.04]; p<0.01).[24] Ycas et al. showed that the largest changes in RDW (change in RDW >2%) were observed after following diagnoses of acute renal failure, septicaemia, acute post-haemorrhagic anaemia, pulmonary insufficiency and pleural effusion, suggesting that RDW is a biomarker of ineffective erythropoiesis and possibly hypoxaemia.[25] Hypoxia affects the regulation of erythropoiesis; hypoxia-induced factor-1-alpha may enhance or replace the effect of glucocorticoids on burst-forming unit-erythroid self-renewal and production of colony-forming unit-erythroid and the erythroblasts are enhanced approximately 170-fold.[26]

RDW is a parameter included in routine full blood count. It is feasible, quick and easy to obtain at the bedside, and is becoming a handy prognostic marker in patients with HF, indicating the advance of ineffective erythropoiesis, impaired ability to utilise available iron, inflammation and hypoxia. The European Society of Cardiology HF guidelines suggest the use of IV ferric carboxymaltose for patients with iron deficiency (serum ferritin <100 μg/l or ferritin 100–299 μg/l and transferrin saturation <20%) with an indication IIa-A.[27] RDW could also be used as a marker of response to iron substitution in patients with HF.