Secondary Logo

Journal Logo

Adult Circulatory Support

Red Cell Distribution Width Predicts 90 Day Mortality in Continuous-Flow Left Ventricular Assist Device Patients

Truby, Lauren K.*; Sridharan, Lakshmi*; Flores, Raul J.*; Garan, A. Reshad*; Jennings, Douglas*; Yuzefpolskaya, Melana*; Takeda, Koji; Takayama, Hiroo; Naka, Yoshifumi; Colombo, Paolo C.*; Topkara, Veli K.*

Author Information
doi: 10.1097/MAT.0000000000000803


Continuous-flow left ventricular assist device (CF-LVAD) therapy has become standard of care in patients with end-stage heart failure for bridge to transplantation (BTT) and destination therapy (DT) indications. After approval of this technology for transplant ineligible patients, nationwide utilization of CF-LVAD therapy has exponentially increased despite a relatively stagnant number of transplants performed within the same time period.1,2 Even with recent advances in the pump design and patient care, CF-LVAD therapy remains associated with significant morbidity and mortality, particularly in the early postimplantation period, highlighting the importance of proper patient selection and identification of high-risk individuals before surgery. Several clinical risk prediction models have been developed to determine mortality risk in CF-LVAD candidates, including HeartMate II risk score (HMRS) and model of end-stage liver disease excluding INR (MELD-Xi).3–5 However, discriminatory ability of commonly used risk scores is relatively modest suggesting that previously unexplored factors can contribute to the mortality in this population.3,6,7

Red cell distribution width (RDW) is a measure of the variation in circulating red blood cell (RBC) volume and routinely reported in automated complete blood counts.8 Red cell distribution width has been traditionally used for differential diagnosis of anemia, with normal values ranging between 11% and 14%. Growing lines of evidence suggest that RDW may serve as a prognostic indicator in healthy individuals as well as specific patient populations including malignancy, renal failure, and cardiovascular disease: an association independent of anemia.9 In patients with chronic ambulatory heart failure, RDW was among the strongest predictors of mortality after adjustment for widely used indices of disease severity.10,11 Red cell distribution width has also been validated as a marker of adverse outcomes in patients with acute decompensated heart failure (HF) and those undergoing coronary artery bypass grafting surgery.12–14 However, prognostic utility of RDW in CF-LVAD candidates is just beginning to be explored, and one recent small, single-center study suggests that a higher preimplant RDW is associated with higher postimplant mortality as well as increased risk of infection.15 Similarly, the objective of the current study was to evaluate the ability of preimplant RDW to predict postimplant mortality, but in addition, to evaluate its ability to identify high-risk patients as compared with previously validated risk scoring systems.


Study Population

All adult patients (age ≥18 years) who underwent CF-LVAD implantation at Columbia University Medical Center between September 2004 and December 2015 for BTT or DT indications were included in this retrospective analysis (n = 409). Analysis was restricted to patients receiving contemporary Food and Drug Administration (FDA) approved devices (HeartMate II [HM II; Thoratec Corp, Pleasington, CA, and St. Jude Medical, St. Paul, MN] or HeartWare VAD [HVAD; HeartWare, Framingham, MA]). Those receiving total artificial hearts or investigational CF-LVADs were excluded. Data collection included patient demographics, comorbid conditions, and laboratory data obtained before device implantation. Outcome variables included early complications and mortality after CF-LVAD support. This study was approved by the Institutional Review Board at Columbia University.

Statistical Analysis

Continuous variables were defined as mean and standard deviations and compared by means of independent t-test. Categoric variables were summarized as percentages and were compared by Pearson χ2 test or Fisher exact test when less than five outcomes expected per cell. Patients were categorized into low and high RDW groups based on the median RDW value of the study population. Postimplantation survival was assessed using Kaplan–Meier analysis with log-rank comparison between groups. Patients who were transplanted, explanted for myocardial recovery, or transferred to other centers were censored from the analysis at the time of these events.

Univariate predictors of 90 day mortality after CF-LVAD implantation were determined using logistic regression analysis. Multivariable stepwise logistic regression analysis with backward selection method (entry and exit significance level of 0.10) was used to identify independent predictors of 90 day mortality after CF-LVAD implantation. Discriminatory power of RDW, HMRS, and MELD-Xi scores was assessed using receiver operating characteristic (ROC) curve analysis. Area under the curve (AUC), optimal cut-off points, sensitivity, and specificity for all risk scores were summarized. Area under the curves for RDW, HMRS, and MELD-Xi scores were compared using the Wilcoxon statistic. Mortality data for HMRS and MELD-Xi risk groups were summarized using established cut-off levels.3,4 A two-tailed p value ≤0.05 were considered statistically significant for all comparisons. Data were analyzed with the use of IBM SPSS Statistics software, version 22.0 (IBM Corp, Armonk, NY).


Baseline Characteristics

Median baseline RDW in CF-LVAD population was 15.8% (Figure 1). Two cohorts were generated based on median RDW: RDW ≤15.8% (n = 204) and RDW >15.8% (n = 205). Baseline demographics and medical history of patients in both groups were summarized and compared in Table 1. Patients with RDW >15.8% were significantly older than those with RDW ≤15.8% (58.6 ± 13.0 vs. 55.9 ± 13.8 years; p = 0.040), were more likely to have ischemic cardiomyopathy (50.5% vs. 39.0%; p = 0.020) and to have had a previous sternotomy (32.8% vs. 22.0%; p = 0.014). There was no difference in prior use of temporary mechanical circulatory support devices (MCSDs). Patients with higher RDW had greater HMRS (1.67 ± 0.74 vs. 1.42 ± 0.94; p = 0.003) and MELD-Xi score (15.9 ± 4.5 vs. 14.4 ± 4.2; p < 0.001).

Table 1.
Table 1.:
Baseline Characteristics of Patients With Continuous-Flow LVAD Implantation
Figure 1.
Figure 1.:
Distribution of preoperative RDW in CF-LVAD recipients. CF-LVAD, continuous-flow left ventricular assist device; IQR, interquartile range; RDW, red cell distribution width.

Preimplant laboratory values of patients in the low and high RDW groups were summarized in Table 2. Patients with a high RDW were more anemic (hemoglobin [Hgb], 10.6 ± 1.8 vs. 11.9 ± 2.1; p < 0.001) with more prominent microcytosis (MCV, 84.9 ± 7.7 vs. 88.7 ± 5.9), and had lower mean corpuscular Hgb and mean corpuscular Hgb concentrations. Total bilirubin was higher in patients with elevated RDW (1.67 ± 2.21 vs. 1.29 ± 0.96; p = 0.025), driven mostly by an increase in direct hyperbilirubinemia (0.73 ± 1.67 vs. 0.40 ± 0.41; p = 0.006). Albumin was noted to be significantly lower in patients with high RDWs (3.4 ± 0.6 vs. 3.7 ± 0.5; p < 0.001). Other cell lines including leukocytes and platelets were not significantly different among groups. Mean creatinine of the overall cohort was 1.46 ± 0.70 and did not differ significantly between groups (1.50 ± 0.67 in the high RDW group versus 1.41 ± 0.72 in the low RDW group; p = 0.162).

Table 2.
Table 2.:
Laboratory Values of Patients With Continuous-Flow LVAD Implantation

Association of Red Cell Distribution Width With Mortality on Continuous-Flow Left Ventricular Assist Device Support

Kaplan–Meier analysis demonstrated significantly lower rates of survival on CF-LVAD support in patients with elevated preimplant RDW compared with those with low preimplant RDW (Figure 2). As shown, survival curves separated within the first 90 days after device implantation. Ninety day mortality was 9.0% (n = 37) in the study cohort and was significantly higher in the high RDW group compared with the low RDW group (12.7% vs. 5.4%; p = 0.009). Higher mortality rates were observed in both nonischemic (14.3% vs. 5.4%; p = 0.021) and ischemic subgroups (11.1% vs. 4.9%; p = 0.125), although the latter did not reach statistical significance because of limited sample size. Univariable logistic regression demonstrated temporary MCSD use before CF-LVAD (odds ratio [OR], 2.05; p = 0.040), white blood cell (WBC) count (OR, 1.10; p = 0.014), Hgb (OR, 0.81; p = 0.017), albumin (OR, 0.37; p = 0.001), RDW (OR, 1.21; p = 0.001), HMRS (OR, 1.63; p = 0.008), and MELD-Xi score (OR, 1.11; p = 0.003) as significant predictors of 90 day mortality after CF-LVAD implantation (Table 3). Multivariable analysis showed high RDW (OR, 1.16; p = 0.010) and low albumin (OR, 0.45; p = 0.011) as independent predictors of 90 day mortality after CF-LVAD implantation. Red cell distribution width remained a significant predictor for 90 day mortality after adjusting for HMRS (OR, 1.21; p = 0.001) as well as MELD-Xi score (OR, 1.18; p = 0.003) (Table 3).

Table 3.
Table 3.:
Predictors of 90 Day Mortality in CF-LVAD Patients
Figure 2.
Figure 2.:
Survival on CF-LVAD support based on preoperative RDW levels. CF-LVAD, continuous-flow left ventricular assist device; RDW, red cell distribution width.

To gain insights into potential mechanisms responsible for increased 90 day mortality observed in CF-LVAD patients with high RDW, we investigated the relationship between preimplant RDW with markers of inflammation and iron metabolism in a limited number of patients with available data (Table 4). As shown, we identified a weak positive correlation between C-reactive protein (CRP) with RDW levels (r = 0.234; p = 0.038) and a weak negative correlation between transferrin saturation with RDW levels (r = −0.278; p = 0.016). Patients with high preimplantation RDW had significantly lower iron levels and transferrin saturation compared with those with low RDW. There were no significant differences found in the nutritional markers examined. Similarly, we found no correlation between RDW levels with markers of hepatocellular injury including aspartate amino transferase (r = −0.027; p = 0.581) and amino alanine transferase (r = −0.050; p = 0.309).

Table 4.
Table 4.:
Relationship Between Preimplant RDW With Markers of Inflammation, Iron Metabolism, and Nutrition in CF-LVAD Patients

Discriminatory Power of Red Cell Distribution Width on 90 Day Post–Continuous-Flow Left Ventricular Assist Device Mortality

The discriminatory power of RDW, HMRS, and MELD-Xi for 90 day mortality in the study population is displayed in Figure 3. The AUCs for RDW, HMRS, and MELD-Xi scores were 0.632, 0.662, and 0.642, respectively. Statistical comparison of AUCs between the RDW versus HMRS (p = 0.622), RDW versus MELD-Xi score (p = 0.872), and HMRS versus MELD-Xi score (p = 0.741) was nonsignificant. The sensitivity of RDW (70.3%) was the same as HMRS (70.3%) and greater than MELD-Xi (62.2%). The specificity of RDW (54.8%) was lower than HMRS (57.3%) and MELD-Xi (63.2%).

Figure 3.
Figure 3.:
Receiver operating characteristics curve analysis: predictors of 90 day mortality. AUC, area under the curve; HMRS, HeartMate II risk score; MELD-Xi, model of end-stage liver disease excluding INR; RDW, red cell distribution width.

When each scoring system was broken into ordinal categories, 90 day mortality was significantly different among groups in all three risk scoring models (RDW, HMRS, and MELD-Xi) (Figure 4A). When RDW (ordinal categories by cut-off 16.0, high versus low) was added to HMRS and MELD-Xi, RDW provided further discriminatory power, most significantly among those in the high-risk category of HMRS (8.3% 90 day mortality with high HMRS, low RDW versus 27.3% 90 day mortality with high HMRS, high RDW) (Figure 4B).

Figure 4.
Figure 4.:
Risk of 90 day mortality after CF-LVAD implantation (A) based on preoperative RDW, MELD-Xi, and HMRS and (B) based on preoperative MELD-Xi and HMRS adjusted for preoperative RDW. *p < 0.05, †p < 0.10. CF-LVAD, continuous-flow left ventricular assist device; HMRS, HeartMate II risk score; MELD-Xi, model of end-stage liver disease excluding INR; RDW, red cell distribution width.

Association of Red Cell Distribution Width With Post–Continuous-Flow Left Ventricular Assist Device Complications

To determine potential role of CF-LVAD complications on the association between RDW and 90 day mortality, early postimplantation complications between high- and low-RDW cohorts were comparatively analyzed (Figure 5). As shown, patients with high RDW had significantly higher risk of right ventricular (RV) failure requiring right ventricular assist device implantation (12.9% vs. 7.0%; p = 0.045) and acute renal failure requiring continuous venovenous hemodialysis (3.3% vs. 11.9%; p = 0.001). Risk of early device thrombosis requiring pump exchange, stroke, or reoperation for bleeding was comparable between the groups.

Figure 5.
Figure 5.:
Risk of CF-LVAD complications in the early postimplantation period. *p < 0.05. CF-LVAD, continuous-flow left ventricular assist device; RDW, red cell distribution width.

Effect of Continuous-Flow Left Ventricular Assist Device Support on Red Cell Distribution Width Levels

To investigate whether RDW elevation observed in patients with advanced heart failure is reversible with mechanical unloading, RDW levels at predetermined time points were serially examined (see Supplemental Figure 1, Supplemental Digital Content, As shown, RDW levels demonstrated a further elevation early after CF-LVAD implantation followed by a gradual linear decrease with levels below preimplantation levels by 1 year of device support.


The current study evaluated preimplant RDW as a prognostic indicator in advanced heart failure patients undergoing CF-LVAD implantation, in an era in which application of this technology is rapidly growing. Major findings are as follows: 1) preimplant RDW is a significant predictor of 90 day mortality in CF-LVAD patients; 2) discriminatory ability of RDW alone is comparable to traditional LVAD risk scores which require computation of multiple individual risk factors; 3) effect of RDW on CF-LVAD mortality is independent of traditional LVAD risk scores; and 4) elevated RDW is associated with increased risk of RV failure and acute kidney injury in the early postimplantation setting. Taken together, these findings suggest that RDW—a readily available, inexpensive laboratory measure—can help guide decision making for patient selection for CF-LVAD therapy.

The HMRS and MELD-Xi represent contemporary methods of risk stratification in CF-LVAD candidates and have been validated successfully in single-center cohorts. The HMRS, developed by Cowger et al.,3 uses age, albumin, creatinine, INR, and center volume to risk stratify patients. Although HMRS is a feasible tool for the clinician to use, its discriminatory function remains relatively modest in validation cohorts.3,6,7 Similarly, the MELD-Xi scoring system, originally developed to aid in listing for liver transplantation, has been successfully applied as a prognostic tool for LVAD and heart transplantation candidates.4,16,17 Results of the current analysis in more than 400 CF-LVAD patients suggest that neither scoring systems has satisfactory discriminatory function for mortality, limiting their utility for patient selection. Discriminatory ability of HMRS and MELD-Xi was comparable to RDW alone, which does not require a calculator. Modest discrimination is not unexpected because risk scores are generally intended to be simple for the use of busy clinicians and do not always account for all variables that may affect mortality on CF-LVAD support, such as RDW, previous sternotomy, and use of temporary MCSD.

Our analysis identified RDW as a robust, independent predictor of 90 day mortality on CF-LVAD support, which is in agreement with the conclusions of the recent publication by Miller et al.15 The potential of RDW as a surrogate marker for illness and inflammation has been well described, particularly in cardiovascular disease and critical illness.18–21 Felker et al.10 were the first to report on the utility of RDW as a prognostic factor in patients with chronic systolic heart failure. Analysis of 2,679 patients from Candesartan in Heart Failure – Assessment of Reduction in Morbidity and Mortality program followed by a replication cohort analysis from the Duke Clinical Data registry has demonstrated RDW as one of the strongest prognostic markers of morbidity and mortality in an ambulatory heart failure cohort, after adjusting for well-accepted heart failure risk factors such as left ventricular ejection fraction, New York Heart Association class, and renal function.10 These findings were replicated in other chronic heart failure cohorts as well as in patients presenting with acute decompensated heart failure, in which RDW elevation during hospital stay portends a worse prognosis.11,13,14 Of note, the median RDWs in the aforementioned chronic HF studies were 14.4% and 14.2%, respectively, as compared with 15.8% in the current study, demonstrating the relatively high overall acuity of CF-LVAD candidates.

Mechanisms for association of RDW with heart failure mortality remain a subject of debate. Red cell distribution width values become elevated in presence of increased RBC destruction or ineffective erythropoiesis of bone marrow. Nutritional deficiencies (iron, vitamin B12, and folic acid) and impaired iron mobilization may contribute to RDW elevation.11 Red cell distribution width may also reflect an underlying inflammatory state, as proinflammatory cytokines—commonly activated in heart failure—inhibit erythropoietin-induced RBC maturation.22,23 We found a significant correlation between preimplant CRP and RDW in a limited number of patients with available data, suggesting that inflammation may play a role in RDW elevation. This finding is consistent with previous reports, which established a relationship between RDW and inflammatory markers.24,25 We also noted reduced iron and transferrin saturation in CF-LVAD candidates with high RDW. Although precise mechanisms remain to be elucidated, RDW is likely an integrative marker of multiple pathophysiologic processes commonly observed in heart failure patients.

Mechanical unloading with CF-LVAD leads to favorable changes in the structure of failing myocardium termed as reverse remodeling, which may allow for device explantation in selected candidates.26 Elevated RDW was associated with impaired reverse remodeling in patients undergoing cardiac resynchronization therapy, suggesting that RDW may serve as a marker of left ventricular remodeling.27 We noted an increase in RDW early after CF-LVAD implantation followed by gradual decrease to levels below preimplant values by 1 year of device support, suggesting that RDW is a modifiable risk factor. CF-LVAD support leads to improvements in proinflammatory cytokine levels, which may account for the trends observed in RDW.28 However, hemolysis commonly observed in patients supported with CF-LVAD may lead to an elevation in RDW and limit its utility as a biomarker in the postimplantation setting.

Study Limitations

This study represents a retrospective analysis of a large single-institutional clinical data registry with known limitations inherent to such design. We were unable to derive an internal validation cohort because of our small number of outcome events and sample size; therefore, our findings require external validation by other implanting centers. Small event rate limited the number of covariates that could be included in the final prediction model. We were also unable to study predictive ability of RDW in different device types because only a small portion of our patients received centrifugal LVAD. Finally, our ability to explore potential mechanisms of RDW elevation was limited due to missing data in inflammatory markers and iron metabolism that were not routinely tested in the preimplantation period.

In conclusion, RDW is an independent predictor of 90 day mortality on CF-LVAD support. Red cell distribution width as an inexpensive, readily available marker can be incorporated into risk stratification in CF-LVAD candidates. Future research is needed to confirm relationship between RDW and increased mortality in heart failure patients and determine biologic mechanisms underlying this association.


1. Stehlik J, Edwards LB, Kucheryavaya AY, et al.; International Society of Heart and Lung Transplantation: The Registry of the International Society for Heart and Lung Transplantation: 29th official adult heart transplant report–2012. J Heart Lung Transplant 2012.31: 1052–1064.
2. Kirklin JK, Naftel DC, Stevenson LW, et al. INTERMACS database for durable devices for circulatory support: First annual report. J Heart Lung Transplant 2008.27: 1065–1072.
3. Cowger J, Sundareswaran K, Rogers JG, et al. Predicting survival in patients receiving continuous flow left ventricular assist devices: The HeartMate II risk score. J Am Coll Cardiol 2013.61: 313–321.
4. Yang JA, Kato TS, Shulman BP, et al. Liver dysfunction as a predictor of outcomes in patients with advanced heart failure requiring ventricular assist device support: Use of the Model of End-stage Liver Disease (MELD) and MELD eXcluding INR (MELD-XI) scoring system. J Heart Lung Transplant 2012.31: 601–610.
5. Cowger JA, Castle L, Aaronson KD, et al. The HeartMate II risk score: An adjusted score for evaluation of all continuous-flow left ventricular assist devices. ASAIO J 2016.62: 281–285.
6. Thomas SS, Nahumi N, Han J, et al. Pre-operative mortality risk assessment in patients with continuous-flow left ventricular assist devices: Application of the HeartMate II risk score. J Heart Lung Transplant 2014.33: 675–681.
7. Adamo L, Nassif M, Tibrewala A, et al. The HeartMate risk score predicts morbidity and mortality in unselected left ventricular assist device recipients and risk stratifies INTERMACS class 1 patients. JACC Heart Fail 2015.3: 283–290.
8. Ryan DH. Kaushansky K, Lichtman MA, Prchal JT, et al. Examination of blood cells. In Williams Hematology, 2015, pp. 9th ed. New York, NY, McGraw-Hill Education, 19–24.
9. Patel KV, Ferrucci L, Ershler WB, Longo DL, Guralnik JM. Red blood cell distribution width and the risk of death in middle-aged and older adults. Arch Intern Med 2009.169: 515–523.
10. Felker GM, Allen LA, Pocock SJ, et al.; CHARM Investigators: Red cell distribution width as a novel prognostic marker in heart failure: Data from the CHARM Program and the Duke Databank. J Am Coll Cardiol 2007.50: 40–47.
11. Allen LA, Felker GM, Mehra MR, et al. Validation and potential mechanisms of red cell distribution width as a prognostic marker in heart failure. J Card Fail 2010.16: 230–238.
12. Warwick R, Mediratta N, Shaw M, et al. Red cell distribution width and coronary artery bypass surgery. Eur J Cardiothorac Surg 2013.43: 1165–1169.
13. Uemura Y, Shibata R, Takemoto K, et al. Elevation of red blood cell distribution width during hospitalization predicts mortality in patients with acute decompensated heart failure. J Cardiol 2016.67: 268–273.
14. Ferreira JP, Girerd N, Arrigo M, et al. Enlarging red blood cell distribution width during hospitalization identifies a very high-risk subset of acutely decompensated heart failure patients and adds valuable prognostic information on top of hemoconcentration. Medicine (Baltimore) 2016.95: e3307.
15. Miller PE, Houston BA, Schneider AL, et al. Associations of preimplant red blood cell distribution width with clinical outcomes among individuals with left ventricular assist devices. ASAIO J 2016.62: 677–683.
16. Kim MS, Kato TS, Farr M, et al. Hepatic dysfunction in ambulatory patients with heart failure: Application of the MELD scoring system for outcome prediction. J Am Coll Cardiol 2013.61: 2253–2261.
17. Grimm JC, Shah AS, Magruder JT, et al. MELD-XI score predicts early mortality in patients after heart transplantation. Ann Thorac Surg 2015.100: 1737–1743.
18. Ku NS, Kim HW, Oh HJ, et al. Red blood cell distribution width is an independent predictor of mortality in patients with gram-negative bacteremia. Shock 2012.38: 123–127.
19. Meynaar IA, Knook AH, Coolen S, et al. Red cell distribution width as predictor for mortality in critically ill patients. Neth J Med 2013.71: 488–493.
20. Hunziker S, Celi LA, Lee J, Howell MD. Red cell distribution width improves the simplified acute physiology score for risk prediction in unselected critically ill patients. Crit Care 2012.16: R89.
21. Guimarães PO, Sun JL, Kragholm K, et al.; MURDOCK Horizon 1 Cardiovascular Study Investigators: Association of standard clinical and laboratory variables with red blood cell distribution width. Am Heart J 2016.174: 22–28.
22. Deswal A, Petersen NJ, Feldman AM, Young JB, White BG, Mann DL. Cytokines and cytokine receptors in advanced heart failure: An analysis of the cytokine database from the Vesnarinone trial (VEST). Circulation 2001.103: 2055–2059.
23. Jelkmann W. Proinflammatory cytokines lowering erythropoietin production. J Interferon Cytokine Res 1998.18: 555–559.
24. Lippi G, Targher G, Montagnana M, Salvagno GL, Zoppini G, Guidi GC. Relation between red blood cell distribution width and inflammatory biomarkers in a large cohort of unselected outpatients. Arch Pathol Lab Med 2009.133: 628–632.
25. Förhécz Z, Gombos T, Borgulya G, Pozsonyi Z, Prohászka Z, Jánoskuti L. Red cell distribution width in heart failure: Prediction of clinical events and relationship with markers of ineffective erythropoiesis, inflammation, renal function, and nutritional state. Am Heart J 2009.158: 659–666.
26. Kato TS, Chokshi A, Singh P, et al. Effects of continuous-flow versus pulsatile-flow left ventricular assist devices on myocardial unloading and remodeling. Circ Heart Fail 2011.4: 546–553.
27. Rickard J, Kumbhani DJ, Gorodeski EZ, et al. Elevated red cell distribution width is associated with impaired reverse ventricular remodeling and increased mortality in patients undergoing cardiac resynchronization therapy. Congest Heart Fail 2012.18: 79–84.
28. Ahmad T, Wang T, O’Brien EC, et al. Effects of left ventricular assist device support on biomarkers of cardiovascular stress, fibrosis, fluid homeostasis, inflammation, and renal injury. JACC Heart Fail 2015.3: 30–39.

left ventricular assist device; prediction; risk score

Supplemental Digital Content

Copyright © 2018 by the ASAIO