Abstract
Introduction
Chronic heart failure (HF) is a frequent cardiovascular disease with increasing incidence and prevalence [2, 17]. In Europe, near 5% of all hospitalizations are related to heart failure [3, 7]. Risk stratification is critical for the management of HF patients [11, 22]. Several biomarkers and multi-biomarker approach have been proposed. However, they are expensive, time consuming and biological variability is also important.
Red blood cell distribution width (RDW) is a routine parameter of hematologic tests that measures the variability of the red blood cell size [4]. It is inexpensive, readily available marker, and reported in automatic cell counts. It has long been used to differentiate the etiology of anemia [18]. More recently, RDW emerged as a powerful predictor of poor outcome in the general population [24, 31], autoimmune disease [34], liver disease [19], respiratory diseases [23], stroke [15], critical illness [8, 35], metabolic diseases [30], in patients with coronary artery disease [28], in both acute and chronic heart failure [1, 5, 29], and is associated with inflammation [9].
Many studies have explored the prognostic value of RDW in HF, varying in the time for its determination, variables used to adjust the model and the follow-up duration. a recent meta-analysis [13] showed that HF patients with higher values of RDW have a poorer prognosis. However, short follow-up time duration (<2-years) may underestimate the RDW prognostic value while studies with models not adjusted for Nt-ProBNP tends to overestimate it [13]. The sampling time is also important. If done during or soon after acute HF decompensation body congestion might interfere with the values of the rdw. Anemia is also an important confounding factor.
The mechanistic association of the rdw with the cardiovascular risk remains poorly understood. Possibly high rdw could simply reflect inflammation as a part of the “cardio-renal anemia syndrome.”
We though to evaluate in stable HF outpatients with reduced ejection fraction (HFrEF), on optimized therapy and in the dry state, the prognostic value of the RDW regarding the 3-year survival and this association with anemia, Nt-ProBNP, and the traditional risk factors.
Material and methods
A total of 233 consecutive patients followed-up in a single specialized outpatient Heart Failure Clinic in a University Hospital, between May 2010 and October 2013, were recruited to the present study according to the inclusion criteria.
Patients were included if they had heart failure with reduced ejection fraction (HFrEF) with EF <40% by trans-thoracic echocardiography. They should have finished the nurse led HF education program and achieved optimal therapy and clinically stable. Optimal therapy was defined according to the ESC guidelines [21] and classified by two senior HF specialists (SL, LS, and NL). Clinical stability defined as no hospital admission or HF decompensation (with extra consultation in the HF clinic), no change in the NYHA class and the furosemide dosage during a six-month period. Thus, ensuring the dry-state, clinical stability, and optimal medical treatment. The baseline timing for the clinical follow-up was considered the appointment at the HF Unit when the team considered that the therapeutics was optimized. Them the patient continued the planned and scheduled routine observation in the Unit. The study is conformed to the principles outlined in the 1964 Declaration of Helsinki. The protocol approved by the institutional review board and written consent obtained from all patients.
Baseline characteristics
At the baseline visit several parameters covering clinical, laboratory, electrocardiographic and echocardiographic domains entered into a dataset. Demographic and clinical characteristics from medical charts including age, gender, medical history (comorbidities and etiology), treatment and doses, New York Heart Association (NYHA) functional class, laboratory (hemoglobin, renal function including estimated glomerular filtration rate- -GFR defined by the MDRD method), electrolytes, Nt-ProBNP), heart rhythm, and LVEF.
Laboratory parameters
Blood samples collected, as the routine protocol of the Unit, and processed in a Central Laboratory of the Lisboa North Hospital Centre. Hemoglobin, hematocrit, and RDW values were determined using the automated blood analyzer Advia 2120. Mean corpuscular hemoglobin concentration (MCHC) calculated. Anemia defined according to World Health Organization criteria (hemoglobin <12 g/dL in men and <11.5 g/dL in women). Nt-ProBNP determined with Roche Elecsys@ ProBNPimmunoassay.
Follow-up and clinical outcome data
The endpoint was all-cause mortality. Patients were assessed at scheduled clinical visits, in the Heart Failure Unit, in a 3-years follow-up. The regular schedule included a minimum of quarterly visits to the Unit. Those who did not attend were contacted with a telephone call to ensure adherence. All-cause death was the main outcome. The circumstances of the incident determined from the hospital records and/or interviews with the relatives. However, we were not able to discriminate the correct cause of death in some patients, because of out of the hospital death.
Objective
The main purpose was to evaluate RDW as a survival marker in stable, optimally medicatedoutpatients with HFrEF.
The secondary objectives were to evaluate the interrelationship of RDW with comorbidities,biomarkers, renal function and therapeutics.
Statistics
Statistical analysis was carried out using the statistical package SPSS® 22.0 for Windows. Continuous variables expressed as mean±standard deviation or median and interquartile range (IQR) depending on whether the distribution was normal. Categorical variables expressed as percentages. Comparative analysis between variables was carried out using the required tests for each variable (dichotomous or continuous) and according to whether the distribution was normal (chi-square test, student’s t-test, Mann–Whitney U test). The variation between the RDW tertiles determined by One-Way ANOVA variance analysis and with the Kruskal-Wallis test. Significance considered at p < 0.05. The relationship between RDW and the other parameter assessed with the Spearman correlation coefficient. The accuracy for predicting death determined by the area under the Receiver Operating Characteristic (ROC) curve, and the optimal cutoff point calculated using Youden-statistics. Because this method of dichotomization is calculated from the study population (in which will be applied as a prognosticator) and is “outcome associated” [33] it would induce bias while analyzing RDW prognostic value. Dichotomization of variables using percentiles is not an ‘outcome-based method’, has less bias, and values are more arbitrary and reproducible [33]. The RDW was categorized according to the tertiles, and the higher risk tertile was also used for survival analysis.
The association between RDW and mortality among the different groups was analyzed using Kaplan–Meier survival curves comparing logarithmic ranges (log-rank test) and using the Cox regression. Survival time calculated from the baseline visit until the date of death or last follow-up. Because the first and second tertiles survival curves are superimposed after adjustment for confounders, those tertiles were merged. Survival plotted, and univariable and multivariable (covariate-adjusted) Cox proportional hazards regression models were used to estimate effects for each covariate, with simultaneous adjustment for potential confounders in multivariable models. Proportional hazards assumptions verified. Significant covariates in the univariate analysis, Online Supporting Table S1 and clinically meaningful included for adjustment in multivariable models. Model adjusted for age, sex, anemia (WHO classification), NYHA class, BMI (by 1 kg/m2 increase), LVEF (by 1% increase), atrial fibrillation, chronic obstructive pulmonary disease (COPD), hypertensive and ischemic etiology, GFR (cut-off 60 ml/min/m2), Nt-ProBNP (optimal cutoff value from the ROC curve-4270 pg/ml) and treatment combinations of ACEi/ARB+beta-blockers and ACEi/ARB+BB+ spironolactone.
Results
We included 233 patients, 71.7% were male, the average age of 68.1±11.2. The median overall survival time was 953 days, and there were 80 (34.3%) events. In Table 1 are shown patients demographics, clinical characteristics, and treatment. a third of the patients were in NYHA class III, 39% were ischemic, and the median Nt-ProBNP 1440 pg/ml. Anemia, as defined by the WHO classification, was present in 19.3% of the patients. The majority received ACEi/ARB (94%), and b-blocker (83.3%); and 59.2% on spironolactone. a third had resynchronization therapy.
RDW interrelationship with hemoglobin, renal function, and Nt-ProBNP
RDW correlated significantly with hemoglobin (r = –0.210; p = 0.002), MCHC (r = –0.425 p < 0.001), creatinine (r = 0.321 p = 0.002), GFR (r = –0.301 p = 0.001) and Nt-ProBNP (r = –0.485 p < 0.001).
RDW tertiles, demographics and therapeutics (Table 1)
Greater the RDW tertile greater the proportion of patients in NYHA III class (p < 0.001), with anemia (13.6–15–30.6%; p = 0.014), atrial fibrillation (p = 0.017) and the Nt-ProBNP (p < 0.001). Renal function was worst with the RDW tertile. Regarding therapeutics, greater the tertile, lower the proportion of patients on ACEi/ARB (p < 0.001), and a greater daily dose of furosemide (p < 0.001).
Survival analysis
Patients who died had greater RDW (14.3±1.5 vs. 15.7±2.1; p < 0.001). The ROC curve (Fig. 1) for death was significantly associated with RDW (AUC = 0.73; 95% CI 0.67–0.79; p < 0.001). The optimal cutoff value was RDW >15.1 (Youden index 0.3961), with a sensitivity and specificity of 60.3 and 79.4, respectively.
Greater the tertile of the RDW greater the 3-year death rate (17.3–25–61.1%; p < 0.001), shorter the survival time (p < 0.001) and greater the risk of death (Hazard-ratio [HR] 2.46 p < 0.001).
Patients in T1 and T2 tertiles had a similar risk of death, and the adjusted survival curve (Fig. 2) were almost superimposed.
The subgroup of patients in the upper tertile (RDW> = 15.3) Table 2 had increased the risk of death (odds-ratio [OR] = 3.1; 95% CI 2.1–4.6; p < 0.001), lower survival time (Log-Rank p = 0.001) and increased adjusted risk of death (HR = 1.54; 95% CI 1.13–2.09; p = 0.006).
The RDW>15.1 (cutpoint calculated from the ROC curve) was also associated with significant greater risk of death (20.1% vs. 59.5%; OR = 2.9; 95% CI 2.0–4.17;p < 0.001), and worse adjusted survival (HR = 2.10; 95% CI 1.13–3.93; p < 0.001).
Discussion
The present work includes a cohort of outpatients with severe SHF that achieved optimal HF therapy and were clinical stable. To ensure clinical stability before and after enrollment and, the best possible dry state of the patients, it was defined a six months period without acute decompensation with or without hospitalization, no change of the NYHA class and no dose change of furosemide therapy. We showed that RDW values are significant predictors of the 3-years survival, even after adjusting for the significant confounders, including anemia, Nt-ProBNP and 12 other significant clinical, laboratory, therapeutic and LV function.
RDW has progressively emerged as a powerful predictor in HF [1, 29], however, is not exclusive for HF [8, 35]. Several previous studies have shown RDW prognostic significant in a broad HF population. However, many are short-term studies that can underestimate the effect [13] while others do not adjust the models for BNP or Nt-ProBNP, which can overestimate the rdw prognostic value [13]. Our results showed that greater values of RDW were associated with increased risk of death and lower survival time, independently of the degree of anemia, optimal therapy, LV function, rhythm, body constitution and Nt-ProBNP. The evaluation time of the predictors of prognosis is of extreme importance, as it could be influenced by several variables that still unstable after decompensations, or during drug titration. We chose to evaluate patients after finishing the nurse led education program and achieving optimal therapy (medical and devices). This time-point is crucial in HF natural history because patients should be re-stratified, as some will not benefit further from hospital follow-up and could be transferred to the community medicine [25].
Several potential mechanisms have been proposed to explain the association of RDW with the prognosis in patients with HF. It reflects the erythrocyte size variability, and when elevated defines the state of anisocytosis [20]. Originally, it was used to the diagnostic workup in normocytic anemia. However, RDW may be associated with the inflammatory state, inadequate production of erythropoietin, functional iron availability, renal function, and nutritional state [10, 25–27]. Anemia is frequent in HF, and is associated with poor prognosis [27], and was present only in 19% of our patients, but the cohort was constituted with patients with optimized medical therapy. Present data confirms that RDW prognostic value supersedes that of the presence of anemia. RDW correlated with hemoglobin, renal function, and Nt-ProBNP. Thus, RDW may be a consequence of the chronic inflammatory state in HF, and be as part of the “cardio-renal anemia syndrome” [16, 32]. RDW may represent an integrative measure of multiple physiopathological processes that occurs in HF, explaining the association with survival. RDW is not a measurable molecule, but rather a statistical concept [27]. Nevertheless, all these parameters were included in the multivariable Cox model, and RDW remained a significant independent predictor of survival.
Clinical implications
A biomarker to be considered for routine measurement it should be evaluated across a range of patients, performed easily, reflect a relevant clinical, pathophysiological process and be useful for caregivers [13]. These potential physiological mechanisms were not assessed in our study, and will require further evaluation. However, our study confirms the independent and solid prognostic value of RDW in stable outpatients with HFrEF, who had achieved optimal HF therapy, about classic and strong prognosticators, including anemia and Nt-ProBNP.
RDW is inexpensive, routinely reported together with blood count and with no extra charge. The potential value of this biomarker should be taken into consideration in the comprehensive management of patients with HF.
Study limitations
Our study has several limitations, first as a single center study limits generalization. However, the cohort represents well the outpatient HF Clinic. Second, we did not assess parameters that could characterize inflammation and iron metabolism that could be the link between rdw and survival. Also, the physiopathological mechanisms were not explored, and will require future attention.
Conclusion
We have shown that the RDW determined during the dry-state in clinically stable outpatients with HFrEF and on optimal medical therapy is a powerful and independent predictor of the long-term survival.