American Journal of Therapeutics 0, 1–5 (2015)
Platelet Reactivity Unit in Predicting Risk of Bleeding in Patients Undergoing Coronary Artery Bypass Graft Surgery Zaid Altheeb, MD,1* Ahmad Sbitan, MD,2 Martin Shabiah, BS,1 Vincent Debari, PhD,2 Aiman Hamdan, MD,1 Mahesh Bikkina, MD,1 Fayez Shamoon, MD,2 and Wilbert S. Aronow, MD3
Bleeding is a common complication of cardiac surgery, accounting for a significant proportion of the total transfusions performed in the United States and Europe. The relationship between platelet reactivity, bleeding, and other adverse events after coronary artery bypass graft surgery (CABGS) has been incompletely characterized. This study investigated the relationship between platelet reactivity and bleeding as a clinical outcome after successful CABGS. A total of 238 patients who underwent CABGS were retrospectively followed for postoperative bleeding. Platelet reactivity unit (PRU) values for all patients were obtained preoperatively to assess the platelet reactivity. The data showed that a range of 180–200 PRU suggests the likelihood of bleeding after CABGS (P 5 0.004), with a statistically significant association only for dual antiplatelet therapy with aspirin and clopidogrel. In conclusion, by using PRU values as a method to assess platelet reactivity and antiplatelet responsiveness, our findings suggest that it may be possible to stratify patients undergoing CABGS for the risk of postoperative bleeding particularly patients on dual antiplatelet therapy. Keywords: coronary artery bypass graft surgery, platelet reactivity unit
INTRODUCTION Bleeding is a common complication of cardiac surgery, accounting for a significant proportion of the total transfusions performed in the United States and Europe.1 The response to antiplatelet therapy varies from one patient to another. Hence, platelet reactivity tests have recently been proven useful in predicting long-term outcomes in the setting of acute coronary syndromes. Tailoring antiplatelet therapy based on
platelet assay results should be considered in patients undergoing major surgeries such as coronary artery bypass graft surgery (CABGS). Previous studies determined the risk of adverse outcomes related to blood clots in patients receiving dual antiplatelet therapy after percutaneous coronary interventions (PCIs).2 In these studies, platelet reactivity was tested before PCI procedures. However, the relationship between platelet reactivity and bleeding after CABGS has been incompletely characterized. This study investigated the relationship between platelet reactivity and bleeding as a clinical outcome after successful CABGS.
1
Department of Medicine, Saint Joseph’s Regional Medical Center/New York Medical College, Paterson, NJ; 2Department of Medicine, Saint Michael’s Medical Center/New York Medical College, Newark, NJ; and 3Cardiology Division, Westchester Medical Center/New York Medical College, Valhalla, NY. The authors have no conflicts of interest to declare. *Address for correspondence: Department of Medicine, Saint Joseph’s Regional Medical Center/New York Medical College, 703 Main St, Paterson, NJ 07503. E-mail: [email protected]
METHODS Between January 2011 and December 2013, we retrospectively enrolled 238 patients who underwent CABGS in a 700-bed teaching hospital in northern New Jersey. All elements of the study were collected retrospectively. Data collection based on obtaining preoperative platelet reactivity unit (PRU) values and
1075–2765 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.americantherapeutics.com
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Altheeb et al
2
postoperative outcomes for patients who had CABGS was done. Age, sex, race, body mass index (BMI), medications, and comorbidities were also considered in the study. We used the hospital electronic medical records and operative reports in the data collection phase. The total number of patients evaluated in this study was 238. Antiplatelet therapy was stopped 5 days before surgery, and none of the patients were on anticoagulant medications. Postoperative bleeding from surgical site has been reported in 43 (18%) of the patients, whereas 195 (82%) patients did not have bleeding after the surgery. Consequently, bleeding was further divided into minor and major depending on the number of blood units transfused postoperatively, as major bleeding defined by more than 4 units of blood transfused. Among patients with bleeding, minor and major bleeding have occurred in 41 (95%) and 2 (5%) patients, respectively (Figure 1). For this study, a was set at 0.05; thus, a P value ,0.05 was required for statistical significance. Because of the retrospective nature of the study, the measure of effect size used was the odds ratio (OR). Continuous variables were tested for normality with the D’Agostino–Pearson omnibus normality test. If both groups of 2 group comparisons were normally distributed, the t test for independent assortment (“unpaired” t test) was used. For nonnormally distributed data, the Mann–Whitney test was used. Continuous variables were subjected to receiver
operating characteristic curve analysis to determine cutoff points. Categorical data were tested for significance with Fisher exact test. Among baseline characteristics, variables that differed by ,5a (P , 0.25) were considered as potential confounders and were used to adjust the univariate OR by multivariable logistic regression. Data were analyzed using Prism (GraphPad Corp, San Diego, CA) and SPSS (IBM Corp, Armonk, NY).
RESULTS The demographic and baseline characteristics of subjects are illustrated in (Table 1). Patients were divided into 2 groups: patients who had postoperative bleeding and who had no bleeding. Males and females with ages between 45 and 75 (mean: 68.7 6 8 SD for patients with bleeding and mean: 67.5 6 9.9 SD for patients with no bleeding) who underwent CABGS were included in the study. The BMI for subjects was estimated for both groups: median (range): 30 (25–38) and 29 (25–38), respectively. Subjects were of variable ethnicities, had multiple comorbidities, and on different medications including statins, b-blockers, calcium channel blockers, and angiotensin-converting enzyme inhibitors/angiotensin receptor blockers. Among baseline characteristics, variables that differed by ,5a (P , 0.25) were considered as potential confounders and were used to adjust the univariate OR by multivariable logistic regression. Figure 2 shows the OR and 95% CI for the postoperative bleeding as the values of PRU. The range of 180–200 PRU suggests the likelihood of bleeding after CABGS. The association between antiplatelets (aspirin at dose of 81 and 325 mg, clopidogrel 75 mg, and aspirin with clopidogrel) and bleeding is given in Figure 3. The univariate data suggest a statistically significant association only for dual therapy with aspirin and clopidogrel. After adjustment for confounding covariates, there are few significant associations, these being for aspirin and clopidogrel together after adjustment for dyslipidemia and diabetes. Interestingly, a negative association in subjects with coronary artery disease suggests a protective effect in that population.
DISCUSSION
FIGURE 1. Flowchart of study population. American Journal of Therapeutics (2015) 0(0)
The major finding of this study is that platelet reactivity as measured by PRU during the preoperative period seems to be predictive of postoperative bleeding risk after CABGS, including major and minor bleeding. The definitions of antiplatelet responsiveness have been relatively arbitrary in previous studies using PRU to examine the association between www.americantherapeutics.com
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Platelet Reactivity Unit
3
Table 1. Baseline demographics and characteristics. Characteristics Age (mean 6 1 SD) Gender (female/male) Body mass index, median (IQR) Ethnicity/race Black Hispanic White Other Black versus all other Hispanic versus all other White versus all other Dyslipidemia Diabetes mellitus Chronic kidney disease Coronary artery disease Atrial fibrillation Statins Angiotensin-converting enzyme inhibitors/angiotensin receptor blockers b-blockers Calcium channel blockers
antiplatelet therapy response variability and postoperative bleeding outcomes. The variable response to antiplatelet therapy has led to the use of platelet function tests to monitor the effects of antiplatelet drugs in cardiovascular diseases. The goal is to guide antiplatelet
Bleeding
No bleeding
P
68.7 6 8 16/27 30 (25–38)
67.5 6 9.9 63/132 29 (25–38)
0.473 0.593 0.902
4/43 7/43 31/43 1/43
29/195 52/195 103/195 11/195
38/43 5/43 5/43 18/43 2/43 30/43 6/43
100/195 85/195 71/195 50/195 8/195 129/195 28/195
0.466 0.176 0.017 ,0.0001 ,0.0001 1 0.041 1 0.723 0.108
19/43 13/43
55/195 52/195
0.047 0.706
therapy to the optimal dose for the prevention or treatment of thrombosis while minimizing hemorrhagic side effects.3 Thus far, certain previous case–control studies using Vienna ab initio simulation package 15, 16 and lipoteichoic acid 15 that measured platelet
FIGURE 2. The OR and 95% CIs for bleeding risk at certain values of PRU. www.americantherapeutics.com
American Journal of Therapeutics (2015) 0(0)
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
4
Altheeb et al
FIGURE 3. The association between antiplatelets (aspirin at dose of 81 and 325 mg, clopidogrel 75 mg, and aspirin and clopidogrel) and bleeding.
activity after PCI/stent found that patients with stent thrombosis have incomplete P2Y12 receptor inhibition and higher posttreatment platelet reactivity.4 Provided that previous observational studies in PCI patients showed a significant relationship between stent thrombosis and the presence of high post-antiplatelet reactivity before the thrombotic event, it is not known whether American Journal of Therapeutics (2015) 0(0)
the high platelet reactivity found in these patients is simply a result of their presentation rather than the etiology of the stent thrombosis.5,6 As the relationship between platelet reactivity and bleeding and other adverse events after CABGS has been incompletely characterized, we conducted a retrospective analysis on a cohort of patients who www.americantherapeutics.com
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Platelet Reactivity Unit
underwent off-pump CABGS and followed the postoperative events to identify a clinically based threshold to determine the relative rates of bleeding from surgical site among patients with high and lower reactivity. We observed that the PRU value had a clinically significant predictive value for risk of bleeding at certain range particularly patients on dual antiplatelet therapy. However, there was no optimal cutoff threshold for the identification of high-risk patients as most of the bleeding events occurred in patients with reactivity around this range (Figure 2). Platelet function assessment was performed before the bleeding events, and therefore our data support the hypothesis that low postoperative platelet reactivity is not an innocent bystander but is itself an etiologic factor in post-CABGS bleeding. The relatively small number of patients, the contribution of other factors (eg, comorbidities, BMI, age, medications, race, and intraoperative mechanical factors) to the bleeding events we observed cannot be excluded. Nevertheless, given our adjustments to the confounders, we demonstrated the significance of these factors in the study design. Interestingly, a negative association in subjects with coronary artery disease suggests a protective effect in that population (Figure 3). Our findings suggest that it may be possible to stratify patients undergoing CABGS at risk for postoperative bleeding, particularly patients who are on dual antiplatelet therapy using a point-of-care platelet function and reactivity assessment performed during the preoperative hospitalization time. However, a large randomized clinical trial must evaluate whether the benefits of holding preoperative antiplatelet therapy
www.americantherapeutics.com
5
in high responders outweigh any potential risk of increased bleeding events and whether such an approach is effective in reducing risk of bleeding. In conclusion, by using PRU values as a method to assess platelet reactivity and antiplatelet responsiveness, our findings suggest that it may be possible to stratify patients undergoing CABGS for the risk of postoperative bleeding particularly patients on dual antiplatelet therapy.
REFERENCES 1. Bracey AW, Grigore AM, Nussmeier NA. Impact of platelet testing on presurgical screening and implications for cardiac and noncardiac surgical procedures. Am J Cardiol. 2006;98:25–32. 2. Kottke-Marchant K. Importance of platelets and platelet response in acute coronary syndromes. Clev Clin J Med. 2009;76:2–7. 3. Michelson AD, Frelinger AL III, Furman MI. Current options in platelet function testing. Am J Cardiol. 2006;98: 4N–10N. 4. Price MJ, Coleman JL, Steinhubl SR, et al. Onset and offset of platelet inhibition after high-dose clopidogrel loading and standard daily therapy measured by a point-of-care assay in healthy volunteers. Am J Cardiol. 2006;98:681–684. 5. Price MJ, Endemann S, Gollapudi RR, et al. Prognostic significance of post-clopidogrel platelet reactivity assessed by a point-of-care assay on thrombotic events after drugeluting stent implantation. Eur Heart J. 2008;29:992–1000. 6. Stone GW, Witzenbichler B, Weisz G, et al. Platelet reactivity and clinical outcomes after coronary artery implantation of drug-eluting stents (ADAPT-DES). Lancet. 2013; 382:614–623.
American Journal of Therapeutics (2015) 0(0)
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Platelet Reactivity Unit in Predicting Risk of Bleeding in Patients Undergoing Coronary Artery Bypass Graft Surgery Zaid Altheeb, MD,1* Ahmad Sbitan, MD,2 Martin Shabiah, BS,1 Vincent Debari, PhD,2 Aiman Hamdan, MD,1 Mahesh Bikkina, MD,1 Fayez Shamoon, MD,2 and Wilbert S. Aronow, MD3
Bleeding is a common complication of cardiac surgery, accounting for a significant proportion of the total transfusions performed in the United States and Europe. The relationship between platelet reactivity, bleeding, and other adverse events after coronary artery bypass graft surgery (CABGS) has been incompletely characterized. This study investigated the relationship between platelet reactivity and bleeding as a clinical outcome after successful CABGS. A total of 238 patients who underwent CABGS were retrospectively followed for postoperative bleeding. Platelet reactivity unit (PRU) values for all patients were obtained preoperatively to assess the platelet reactivity. The data showed that a range of 180–200 PRU suggests the likelihood of bleeding after CABGS (P 5 0.004), with a statistically significant association only for dual antiplatelet therapy with aspirin and clopidogrel. In conclusion, by using PRU values as a method to assess platelet reactivity and antiplatelet responsiveness, our findings suggest that it may be possible to stratify patients undergoing CABGS for the risk of postoperative bleeding particularly patients on dual antiplatelet therapy. Keywords: coronary artery bypass graft surgery, platelet reactivity unit
INTRODUCTION Bleeding is a common complication of cardiac surgery, accounting for a significant proportion of the total transfusions performed in the United States and Europe.1 The response to antiplatelet therapy varies from one patient to another. Hence, platelet reactivity tests have recently been proven useful in predicting long-term outcomes in the setting of acute coronary syndromes. Tailoring antiplatelet therapy based on
platelet assay results should be considered in patients undergoing major surgeries such as coronary artery bypass graft surgery (CABGS). Previous studies determined the risk of adverse outcomes related to blood clots in patients receiving dual antiplatelet therapy after percutaneous coronary interventions (PCIs).2 In these studies, platelet reactivity was tested before PCI procedures. However, the relationship between platelet reactivity and bleeding after CABGS has been incompletely characterized. This study investigated the relationship between platelet reactivity and bleeding as a clinical outcome after successful CABGS.
1
Department of Medicine, Saint Joseph’s Regional Medical Center/New York Medical College, Paterson, NJ; 2Department of Medicine, Saint Michael’s Medical Center/New York Medical College, Newark, NJ; and 3Cardiology Division, Westchester Medical Center/New York Medical College, Valhalla, NY. The authors have no conflicts of interest to declare. *Address for correspondence: Department of Medicine, Saint Joseph’s Regional Medical Center/New York Medical College, 703 Main St, Paterson, NJ 07503. E-mail: [email protected]
METHODS Between January 2011 and December 2013, we retrospectively enrolled 238 patients who underwent CABGS in a 700-bed teaching hospital in northern New Jersey. All elements of the study were collected retrospectively. Data collection based on obtaining preoperative platelet reactivity unit (PRU) values and
1075–2765 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.americantherapeutics.com
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Altheeb et al
2
postoperative outcomes for patients who had CABGS was done. Age, sex, race, body mass index (BMI), medications, and comorbidities were also considered in the study. We used the hospital electronic medical records and operative reports in the data collection phase. The total number of patients evaluated in this study was 238. Antiplatelet therapy was stopped 5 days before surgery, and none of the patients were on anticoagulant medications. Postoperative bleeding from surgical site has been reported in 43 (18%) of the patients, whereas 195 (82%) patients did not have bleeding after the surgery. Consequently, bleeding was further divided into minor and major depending on the number of blood units transfused postoperatively, as major bleeding defined by more than 4 units of blood transfused. Among patients with bleeding, minor and major bleeding have occurred in 41 (95%) and 2 (5%) patients, respectively (Figure 1). For this study, a was set at 0.05; thus, a P value ,0.05 was required for statistical significance. Because of the retrospective nature of the study, the measure of effect size used was the odds ratio (OR). Continuous variables were tested for normality with the D’Agostino–Pearson omnibus normality test. If both groups of 2 group comparisons were normally distributed, the t test for independent assortment (“unpaired” t test) was used. For nonnormally distributed data, the Mann–Whitney test was used. Continuous variables were subjected to receiver
operating characteristic curve analysis to determine cutoff points. Categorical data were tested for significance with Fisher exact test. Among baseline characteristics, variables that differed by ,5a (P , 0.25) were considered as potential confounders and were used to adjust the univariate OR by multivariable logistic regression. Data were analyzed using Prism (GraphPad Corp, San Diego, CA) and SPSS (IBM Corp, Armonk, NY).
RESULTS The demographic and baseline characteristics of subjects are illustrated in (Table 1). Patients were divided into 2 groups: patients who had postoperative bleeding and who had no bleeding. Males and females with ages between 45 and 75 (mean: 68.7 6 8 SD for patients with bleeding and mean: 67.5 6 9.9 SD for patients with no bleeding) who underwent CABGS were included in the study. The BMI for subjects was estimated for both groups: median (range): 30 (25–38) and 29 (25–38), respectively. Subjects were of variable ethnicities, had multiple comorbidities, and on different medications including statins, b-blockers, calcium channel blockers, and angiotensin-converting enzyme inhibitors/angiotensin receptor blockers. Among baseline characteristics, variables that differed by ,5a (P , 0.25) were considered as potential confounders and were used to adjust the univariate OR by multivariable logistic regression. Figure 2 shows the OR and 95% CI for the postoperative bleeding as the values of PRU. The range of 180–200 PRU suggests the likelihood of bleeding after CABGS. The association between antiplatelets (aspirin at dose of 81 and 325 mg, clopidogrel 75 mg, and aspirin with clopidogrel) and bleeding is given in Figure 3. The univariate data suggest a statistically significant association only for dual therapy with aspirin and clopidogrel. After adjustment for confounding covariates, there are few significant associations, these being for aspirin and clopidogrel together after adjustment for dyslipidemia and diabetes. Interestingly, a negative association in subjects with coronary artery disease suggests a protective effect in that population.
DISCUSSION
FIGURE 1. Flowchart of study population. American Journal of Therapeutics (2015) 0(0)
The major finding of this study is that platelet reactivity as measured by PRU during the preoperative period seems to be predictive of postoperative bleeding risk after CABGS, including major and minor bleeding. The definitions of antiplatelet responsiveness have been relatively arbitrary in previous studies using PRU to examine the association between www.americantherapeutics.com
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Platelet Reactivity Unit
3
Table 1. Baseline demographics and characteristics. Characteristics Age (mean 6 1 SD) Gender (female/male) Body mass index, median (IQR) Ethnicity/race Black Hispanic White Other Black versus all other Hispanic versus all other White versus all other Dyslipidemia Diabetes mellitus Chronic kidney disease Coronary artery disease Atrial fibrillation Statins Angiotensin-converting enzyme inhibitors/angiotensin receptor blockers b-blockers Calcium channel blockers
antiplatelet therapy response variability and postoperative bleeding outcomes. The variable response to antiplatelet therapy has led to the use of platelet function tests to monitor the effects of antiplatelet drugs in cardiovascular diseases. The goal is to guide antiplatelet
Bleeding
No bleeding
P
68.7 6 8 16/27 30 (25–38)
67.5 6 9.9 63/132 29 (25–38)
0.473 0.593 0.902
4/43 7/43 31/43 1/43
29/195 52/195 103/195 11/195
38/43 5/43 5/43 18/43 2/43 30/43 6/43
100/195 85/195 71/195 50/195 8/195 129/195 28/195
0.466 0.176 0.017 ,0.0001 ,0.0001 1 0.041 1 0.723 0.108
19/43 13/43
55/195 52/195
0.047 0.706
therapy to the optimal dose for the prevention or treatment of thrombosis while minimizing hemorrhagic side effects.3 Thus far, certain previous case–control studies using Vienna ab initio simulation package 15, 16 and lipoteichoic acid 15 that measured platelet
FIGURE 2. The OR and 95% CIs for bleeding risk at certain values of PRU. www.americantherapeutics.com
American Journal of Therapeutics (2015) 0(0)
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
4
Altheeb et al
FIGURE 3. The association between antiplatelets (aspirin at dose of 81 and 325 mg, clopidogrel 75 mg, and aspirin and clopidogrel) and bleeding.
activity after PCI/stent found that patients with stent thrombosis have incomplete P2Y12 receptor inhibition and higher posttreatment platelet reactivity.4 Provided that previous observational studies in PCI patients showed a significant relationship between stent thrombosis and the presence of high post-antiplatelet reactivity before the thrombotic event, it is not known whether American Journal of Therapeutics (2015) 0(0)
the high platelet reactivity found in these patients is simply a result of their presentation rather than the etiology of the stent thrombosis.5,6 As the relationship between platelet reactivity and bleeding and other adverse events after CABGS has been incompletely characterized, we conducted a retrospective analysis on a cohort of patients who www.americantherapeutics.com
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Platelet Reactivity Unit
underwent off-pump CABGS and followed the postoperative events to identify a clinically based threshold to determine the relative rates of bleeding from surgical site among patients with high and lower reactivity. We observed that the PRU value had a clinically significant predictive value for risk of bleeding at certain range particularly patients on dual antiplatelet therapy. However, there was no optimal cutoff threshold for the identification of high-risk patients as most of the bleeding events occurred in patients with reactivity around this range (Figure 2). Platelet function assessment was performed before the bleeding events, and therefore our data support the hypothesis that low postoperative platelet reactivity is not an innocent bystander but is itself an etiologic factor in post-CABGS bleeding. The relatively small number of patients, the contribution of other factors (eg, comorbidities, BMI, age, medications, race, and intraoperative mechanical factors) to the bleeding events we observed cannot be excluded. Nevertheless, given our adjustments to the confounders, we demonstrated the significance of these factors in the study design. Interestingly, a negative association in subjects with coronary artery disease suggests a protective effect in that population (Figure 3). Our findings suggest that it may be possible to stratify patients undergoing CABGS at risk for postoperative bleeding, particularly patients who are on dual antiplatelet therapy using a point-of-care platelet function and reactivity assessment performed during the preoperative hospitalization time. However, a large randomized clinical trial must evaluate whether the benefits of holding preoperative antiplatelet therapy
www.americantherapeutics.com
5
in high responders outweigh any potential risk of increased bleeding events and whether such an approach is effective in reducing risk of bleeding. In conclusion, by using PRU values as a method to assess platelet reactivity and antiplatelet responsiveness, our findings suggest that it may be possible to stratify patients undergoing CABGS for the risk of postoperative bleeding particularly patients on dual antiplatelet therapy.
REFERENCES 1. Bracey AW, Grigore AM, Nussmeier NA. Impact of platelet testing on presurgical screening and implications for cardiac and noncardiac surgical procedures. Am J Cardiol. 2006;98:25–32. 2. Kottke-Marchant K. Importance of platelets and platelet response in acute coronary syndromes. Clev Clin J Med. 2009;76:2–7. 3. Michelson AD, Frelinger AL III, Furman MI. Current options in platelet function testing. Am J Cardiol. 2006;98: 4N–10N. 4. Price MJ, Coleman JL, Steinhubl SR, et al. Onset and offset of platelet inhibition after high-dose clopidogrel loading and standard daily therapy measured by a point-of-care assay in healthy volunteers. Am J Cardiol. 2006;98:681–684. 5. Price MJ, Endemann S, Gollapudi RR, et al. Prognostic significance of post-clopidogrel platelet reactivity assessed by a point-of-care assay on thrombotic events after drugeluting stent implantation. Eur Heart J. 2008;29:992–1000. 6. Stone GW, Witzenbichler B, Weisz G, et al. Platelet reactivity and clinical outcomes after coronary artery implantation of drug-eluting stents (ADAPT-DES). Lancet. 2013; 382:614–623.
American Journal of Therapeutics (2015) 0(0)
Copyright Ó 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
