Journal of the American College of Cardiology Ó 2014 by the American College of Cardiology Foundation Published by Elsevier Inc.
EDITORIAL COMMENT
Deconstructing Stent Polymers* John A. Bittl, MD Ocala, Florida
Vol. 63, No. 4, 2014 ISSN 0735-1097/$36.00 http://dx.doi.org/10.1016/j.jacc.2013.10.016
1-year myocardial infarction (OR: 0.68, 95% BCI: 0.52 to 0.88) and lower long-term definite or probable ST (OR: 0.60, 95% BCI: 0.40 to 0.89) than BMS. Lastly, BP-BES have lower rates of long-term myocardial infarction (OR: 0.73, 95% BCI: 0.59 to 0.99) and lower definite or probable ST (OR: 0.52, 95% BCI: 0.31 to 0.81) than DP-PES. The robust analysis thus creates the following hierarchy, which lists stents from most to least preferred: DP-CoCr-EES > BP-BES > BMS or first-generation DP-PES
Jacques Derrida delivered the lecture, “Structure, Sign, and Play in the Discourse of the Human Sciences,” at Johns Hopkins on October 21, 1966, and launched the philosophical movement known as deconstruction (1). Deconstruction, which diverges from the traditional philosophical search for universal truths, breaks down the hierarchy of privileged concepts that exists in word pairs called binaries. See page 299
A binary is a set of related terms such as rationality/emotion, essence/existence, speech/writing, ordin the setting of sponsored trialsdcase/control, in which the first term is perceived to be favored over the second. Deconstruction, which questions the existence of privileged concepts and replaces the structure of binaries with “a full free play of meaning,” comprises a nearly incomprehensible approach for philosophical discourse (2) and probably irrelevant process for most clinical investigation but might cast a fresh view of the evidence for using bioabsorbable polymers (BPs) in favor of durable polymers (DPs), or no polymer at all, on coronary stents. Current study. In a study reported in this issue of the Journal, Palmerini et al. (3) describe a Bayesian network meta-analysis of 89 clinical trials that compares the relative safety of bioabsorbable-polymer biolimus-eluting stents (BP-BES), durable-polymer drug-eluting stents (DP-DES), and bare-metal stents (BMS). The analytical methods for this analysis are complex, but the main results are straightforward. Although BP-BES, DP-DES, and BMS are equivalent for most endpoints, several differences are observed. First, BP-BES have higher rates of 1-year definite stent thrombosis (ST) (posterior median odds ratio [OR]: 2.44, 95% Bayesian credible intervals [95% BCI]: 1.30 to 4.76) and higher long-term definite ST (OR: 1.92, 95% BCI: 1.02 to 3.45) than durable-polymer cobalt-chromium everolimus-eluting stents (DP-CoCr-EES). Second, BP-BES have lower rates of
*Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology. From the Munroe Heart and Vascular Institute, Munroe Regional Medical Center, Ocala, Florida. Dr. Bittl has reported that he has no relationships relevant to the contents of this paper to disclose.
Bayesian perspective. The Bayesian meta-analysis, which requires no adjustments for multiple simultaneous comparisons, generates credible intervals (BCIs) that differ from traditional confidence intervals (CIs) in several ways (4). The traditional approach defines probability as a limit as the number of observations hypothetically approaches infinity and thus measures a frequency. If a large number of traditional frequentist 95% CIs could be generated, 95% of them should contain the true OR, but the actual probability that a particular CI contains the true OR cannot be stated (4). In other words, the probability that a given 95% CI contains the true OR is either 0 or 1. It either does or it does not. The Bayesian meta-analysis generates a 95% BCI that has a 95% probability of containing the true OR, a characteristic that is commonly but erroneously attributed to the traditional frequentist CI (4). For example, the 95% BCI of 1.30 to 4.76 for 1-year ST after BP-BES compared with DPCoCr-EES has a 95% chance of containing the true OR, confirming that the higher rate of ST after BP-BES is believable and virtually certain. The possibility that the 2 stents have the same risk of ST is extremely small. According to the definition of the 95% BCI, there is only a 2.5% chance that the odds of ST after BP-BES compared with DP-CoCr-EES is less than the lower boundary of the credible interval. The overall results of the network metaanalysis suggest that the risk of ST is related to stent design, on the basis of the assumption that patients in the randomized stent trials were sampled from similar populations. Polymers. The Biomatrix BP-BES and Nobori stents (Terumo Corporation, Tokyo, Japan) have a BP comprising polyglycolic acid/polylactic acid (PGLA). The Promus and Xience stents (Boston Scientific, Natick, Massachusetts) use a DP made of a fluorinated polymer consisting of polyvinylidene fluoride (PVDF) and hexafluoropropylene (HFP) monomers. The Endeavor stents (Medtronic, Minneapolis, Minnesota) have a DP made of phosphorylcholine (PC), and the Endeavor Resolute stent (Medtronic) has a DP made of the proprietary BioLinx polymer, consisting of a hydrophobic C10 component, a hydrophilic polyvinylpyrrolidinone and a C19 composite component. Inflammation. The development of BPs was based on the premise that they cause less vascular inflammation and fibrin
JACC Vol. 63, No. 4, 2014 February 4, 2014:308–9
deposition than DPs, but different polymer types might actually cause similar vascular responses. van der Giessen et al. (5) described the reaction of porcine coronary arteries to 5 different BPs, including PGLA, polycaprolactone, polyhydroxybutyrate valerate, polyorthoester, and polyethyleneoxide/polybutylene terephthalate and 3 DPs including polyurethane, silicone, and polyethylene terephthalate. The biodegradable polycaprolactone, polyhydroxybutyrate valerate, and polyorthoester and the nonbiodegradable polyurethane and silicone evoked extensive inflammatory responses and fibrocellular proliferation. Less but still severe responses were observed for the biodegradable PGLA and polyethyleneoxide/ polybutylene terephthalate and for the nonbiodegradable polyethylene terephthalate. The investigators concluded that all polymers have the potential to cause vascular inflammation (5). Thromboresistance. Several studies suggest that the surface potential, wettability, and smoothness of certain polymers might confer thromboresistance. One study suggests that PVDF-HEP and PC have equivalent thromboresistance (6). Several studies suggest that VDF-grafted polymers are less likely than untreated polymers to activate platelets (7,8). A study by Kolandaivelu et al. (9) has shown that the risk of acute stent thrombosis in high-risk interventional settings is reduced by the presence of a fluoropolymer. With a Chandler loop model, the investigators have observed that the rate of thrombosis of DP-CoCr-EES in silico is lower than that for matched BMS and remains insensitive to incomplete deployment. Comparing fluoropolymers with stent metals, Eppihimer et al. (10) have found less platelet adhesion on the platinum chromium (PtCr) of the fourth-generation PtCr-EES than on PVDFHFP polymers. Deconstruction. The titular goal of the current study was to compare “bioabsorbable polymer-based versus durable polymer-based drug-eluting stents,” but the authors of the current report (3) make no definitive conclusions about the superiority of BPs versus DPs and “refrain from giving recommendations about the use of a specific device.” The investigators cannot be faulted for a cautious interpretation of their own data, but in rejecting the possibility of a favored polymer or device after such a thorough analysis, the investigators present an aporeia (Greek, impasse) and, like Derrida, invoke the philosophy of Gorgias the Sophist, who wrote (2): Nothing exists. If it exists, it cannot be known. If it can be known, it cannot be communicated. If it can be communicated, it cannot be understood. Denouement. The weight of evidence in the current analysis (3) supports conclusions about polymer types and devices. The large sample size, absence of heterogeneity
Bittl Deconstructing Stents
309
across multiple comparisons, the stability of results across several sensitivity analyses, and consistency between direct and indirect estimates constitute robust scientific evidence to demonstrate that BPs do not outperform fluoropolymer DPs. The analysis in the current report (3) suggests that the DP-CoCr-EES and their PtCr analogs are the preferred devices for coronary stenting in current practice. Scientific progress is incremental. Complex datasets often require advanced methods of analysis. The Bayesian approach of network meta-analysis allows statistical inferences to be made when direct comparisons do not exist. A “discriminating, philosophical and fearless” approach borrowed from literary criticism (11) helps to translate trial evidence into clinical practice to keep the field moving forward. Reprint requests and correspondence: Dr. John A. Bittl, Ocala Heart Institute, Interventional Cardiology, 1221 Southeast 5th Street, Ocala, Florida 34471. E-mail: [email protected].
REFERENCES
1. Derrida J. Structure, sign, and play in the discourse of the human sciences. In: Writing and Difference. London: Routledge; 1967: 351–70. 2. Markos L. Derrida and Deconstruction. In: Burgstaller A, editor. Great Minds of the Western Intellectual Tradition. Chantilly, Virginia: The Teaching Company, 2000:38–42. 3. Palmerini T, Biondi-Zoccai G, Della Riva D, et al. Clinical outcomes with bioabsorbale polymer- versus durable polymer-based drug-eluting and bare-metal stents: evidence from a comprehensive network metaanalysis. J Am Coll Cardiol 2014;63:299–307. 4. Spiegelhalter DJ, Abrams KR, Myles JP. Bayesian Approaches to Clinical Trials and Health Care Evaluations. Chichester, England: Wiley, 2004. 5. van der Giessen WJ, Lincoff AM, Schwartz RS, et al. Marked inflammatory sequelae to implantation of biodegradable and nonbiodegradable polymers in porcine coronary arteries. Circulation 1996;94: 1690–7. 6. Chin-Quee SL, Hsu SH, Nguyen-Ehrenreich K-L, et al. Endothelial cell recovery, acute thrombogenicity, and monocyte adhesion and activation on fluorinated copolymer and phosphorylcholine polymer stent coatings. Biomaterials 2010;31:648–57. 7. Lin JC, Tiong SL, Chen CY. Surface characterization and platelet adhesion studies on fluorocarbons prepared by plasma-induced graft polymerization. J Biomater Sci Polym Ed 2000;11:701–14. 8. Liu T-Y, Lin W-C, Huang L-Y, Chen S-Y, Yang M- C. Surface characteristics and hemocompatibility of PAN/PVDF blend membranes. Polymer Adv Technol 2005;16:413–9. 9. Kolandaivelu K, Swaminathan R, Gibson WJ, et al. Stent thrombogenicity early in high-risk interventional settings is driven by stent design and deployment and protected by polymer-drug coatings. Circulation 2011;123:1400–9. 10. Eppihimer MJ, Sushkova N, Grimsby JL, et al. Impact of stent surface on thrombogenecity and vascular healing: a comparative analysis of metallic and polymeric surfaces. Circ Cardiovasc Interv 2013;6:370–7. 11. Lowell JR. Our Contributors: Edgar Allen Poe. Philadelphia, PA: George Rex Graham, 1845:49–53. Key Words: Bayesian analysis - coronary stent(s) intervention - randomized trials.
-
percutaneous coronary
EDITORIAL COMMENT
Deconstructing Stent Polymers* John A. Bittl, MD Ocala, Florida
Vol. 63, No. 4, 2014 ISSN 0735-1097/$36.00 http://dx.doi.org/10.1016/j.jacc.2013.10.016
1-year myocardial infarction (OR: 0.68, 95% BCI: 0.52 to 0.88) and lower long-term definite or probable ST (OR: 0.60, 95% BCI: 0.40 to 0.89) than BMS. Lastly, BP-BES have lower rates of long-term myocardial infarction (OR: 0.73, 95% BCI: 0.59 to 0.99) and lower definite or probable ST (OR: 0.52, 95% BCI: 0.31 to 0.81) than DP-PES. The robust analysis thus creates the following hierarchy, which lists stents from most to least preferred: DP-CoCr-EES > BP-BES > BMS or first-generation DP-PES
Jacques Derrida delivered the lecture, “Structure, Sign, and Play in the Discourse of the Human Sciences,” at Johns Hopkins on October 21, 1966, and launched the philosophical movement known as deconstruction (1). Deconstruction, which diverges from the traditional philosophical search for universal truths, breaks down the hierarchy of privileged concepts that exists in word pairs called binaries. See page 299
A binary is a set of related terms such as rationality/emotion, essence/existence, speech/writing, ordin the setting of sponsored trialsdcase/control, in which the first term is perceived to be favored over the second. Deconstruction, which questions the existence of privileged concepts and replaces the structure of binaries with “a full free play of meaning,” comprises a nearly incomprehensible approach for philosophical discourse (2) and probably irrelevant process for most clinical investigation but might cast a fresh view of the evidence for using bioabsorbable polymers (BPs) in favor of durable polymers (DPs), or no polymer at all, on coronary stents. Current study. In a study reported in this issue of the Journal, Palmerini et al. (3) describe a Bayesian network meta-analysis of 89 clinical trials that compares the relative safety of bioabsorbable-polymer biolimus-eluting stents (BP-BES), durable-polymer drug-eluting stents (DP-DES), and bare-metal stents (BMS). The analytical methods for this analysis are complex, but the main results are straightforward. Although BP-BES, DP-DES, and BMS are equivalent for most endpoints, several differences are observed. First, BP-BES have higher rates of 1-year definite stent thrombosis (ST) (posterior median odds ratio [OR]: 2.44, 95% Bayesian credible intervals [95% BCI]: 1.30 to 4.76) and higher long-term definite ST (OR: 1.92, 95% BCI: 1.02 to 3.45) than durable-polymer cobalt-chromium everolimus-eluting stents (DP-CoCr-EES). Second, BP-BES have lower rates of
*Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology. From the Munroe Heart and Vascular Institute, Munroe Regional Medical Center, Ocala, Florida. Dr. Bittl has reported that he has no relationships relevant to the contents of this paper to disclose.
Bayesian perspective. The Bayesian meta-analysis, which requires no adjustments for multiple simultaneous comparisons, generates credible intervals (BCIs) that differ from traditional confidence intervals (CIs) in several ways (4). The traditional approach defines probability as a limit as the number of observations hypothetically approaches infinity and thus measures a frequency. If a large number of traditional frequentist 95% CIs could be generated, 95% of them should contain the true OR, but the actual probability that a particular CI contains the true OR cannot be stated (4). In other words, the probability that a given 95% CI contains the true OR is either 0 or 1. It either does or it does not. The Bayesian meta-analysis generates a 95% BCI that has a 95% probability of containing the true OR, a characteristic that is commonly but erroneously attributed to the traditional frequentist CI (4). For example, the 95% BCI of 1.30 to 4.76 for 1-year ST after BP-BES compared with DPCoCr-EES has a 95% chance of containing the true OR, confirming that the higher rate of ST after BP-BES is believable and virtually certain. The possibility that the 2 stents have the same risk of ST is extremely small. According to the definition of the 95% BCI, there is only a 2.5% chance that the odds of ST after BP-BES compared with DP-CoCr-EES is less than the lower boundary of the credible interval. The overall results of the network metaanalysis suggest that the risk of ST is related to stent design, on the basis of the assumption that patients in the randomized stent trials were sampled from similar populations. Polymers. The Biomatrix BP-BES and Nobori stents (Terumo Corporation, Tokyo, Japan) have a BP comprising polyglycolic acid/polylactic acid (PGLA). The Promus and Xience stents (Boston Scientific, Natick, Massachusetts) use a DP made of a fluorinated polymer consisting of polyvinylidene fluoride (PVDF) and hexafluoropropylene (HFP) monomers. The Endeavor stents (Medtronic, Minneapolis, Minnesota) have a DP made of phosphorylcholine (PC), and the Endeavor Resolute stent (Medtronic) has a DP made of the proprietary BioLinx polymer, consisting of a hydrophobic C10 component, a hydrophilic polyvinylpyrrolidinone and a C19 composite component. Inflammation. The development of BPs was based on the premise that they cause less vascular inflammation and fibrin
JACC Vol. 63, No. 4, 2014 February 4, 2014:308–9
deposition than DPs, but different polymer types might actually cause similar vascular responses. van der Giessen et al. (5) described the reaction of porcine coronary arteries to 5 different BPs, including PGLA, polycaprolactone, polyhydroxybutyrate valerate, polyorthoester, and polyethyleneoxide/polybutylene terephthalate and 3 DPs including polyurethane, silicone, and polyethylene terephthalate. The biodegradable polycaprolactone, polyhydroxybutyrate valerate, and polyorthoester and the nonbiodegradable polyurethane and silicone evoked extensive inflammatory responses and fibrocellular proliferation. Less but still severe responses were observed for the biodegradable PGLA and polyethyleneoxide/ polybutylene terephthalate and for the nonbiodegradable polyethylene terephthalate. The investigators concluded that all polymers have the potential to cause vascular inflammation (5). Thromboresistance. Several studies suggest that the surface potential, wettability, and smoothness of certain polymers might confer thromboresistance. One study suggests that PVDF-HEP and PC have equivalent thromboresistance (6). Several studies suggest that VDF-grafted polymers are less likely than untreated polymers to activate platelets (7,8). A study by Kolandaivelu et al. (9) has shown that the risk of acute stent thrombosis in high-risk interventional settings is reduced by the presence of a fluoropolymer. With a Chandler loop model, the investigators have observed that the rate of thrombosis of DP-CoCr-EES in silico is lower than that for matched BMS and remains insensitive to incomplete deployment. Comparing fluoropolymers with stent metals, Eppihimer et al. (10) have found less platelet adhesion on the platinum chromium (PtCr) of the fourth-generation PtCr-EES than on PVDFHFP polymers. Deconstruction. The titular goal of the current study was to compare “bioabsorbable polymer-based versus durable polymer-based drug-eluting stents,” but the authors of the current report (3) make no definitive conclusions about the superiority of BPs versus DPs and “refrain from giving recommendations about the use of a specific device.” The investigators cannot be faulted for a cautious interpretation of their own data, but in rejecting the possibility of a favored polymer or device after such a thorough analysis, the investigators present an aporeia (Greek, impasse) and, like Derrida, invoke the philosophy of Gorgias the Sophist, who wrote (2): Nothing exists. If it exists, it cannot be known. If it can be known, it cannot be communicated. If it can be communicated, it cannot be understood. Denouement. The weight of evidence in the current analysis (3) supports conclusions about polymer types and devices. The large sample size, absence of heterogeneity
Bittl Deconstructing Stents
309
across multiple comparisons, the stability of results across several sensitivity analyses, and consistency between direct and indirect estimates constitute robust scientific evidence to demonstrate that BPs do not outperform fluoropolymer DPs. The analysis in the current report (3) suggests that the DP-CoCr-EES and their PtCr analogs are the preferred devices for coronary stenting in current practice. Scientific progress is incremental. Complex datasets often require advanced methods of analysis. The Bayesian approach of network meta-analysis allows statistical inferences to be made when direct comparisons do not exist. A “discriminating, philosophical and fearless” approach borrowed from literary criticism (11) helps to translate trial evidence into clinical practice to keep the field moving forward. Reprint requests and correspondence: Dr. John A. Bittl, Ocala Heart Institute, Interventional Cardiology, 1221 Southeast 5th Street, Ocala, Florida 34471. E-mail: [email protected].
REFERENCES
1. Derrida J. Structure, sign, and play in the discourse of the human sciences. In: Writing and Difference. London: Routledge; 1967: 351–70. 2. Markos L. Derrida and Deconstruction. In: Burgstaller A, editor. Great Minds of the Western Intellectual Tradition. Chantilly, Virginia: The Teaching Company, 2000:38–42. 3. Palmerini T, Biondi-Zoccai G, Della Riva D, et al. Clinical outcomes with bioabsorbale polymer- versus durable polymer-based drug-eluting and bare-metal stents: evidence from a comprehensive network metaanalysis. J Am Coll Cardiol 2014;63:299–307. 4. Spiegelhalter DJ, Abrams KR, Myles JP. Bayesian Approaches to Clinical Trials and Health Care Evaluations. Chichester, England: Wiley, 2004. 5. van der Giessen WJ, Lincoff AM, Schwartz RS, et al. Marked inflammatory sequelae to implantation of biodegradable and nonbiodegradable polymers in porcine coronary arteries. Circulation 1996;94: 1690–7. 6. Chin-Quee SL, Hsu SH, Nguyen-Ehrenreich K-L, et al. Endothelial cell recovery, acute thrombogenicity, and monocyte adhesion and activation on fluorinated copolymer and phosphorylcholine polymer stent coatings. Biomaterials 2010;31:648–57. 7. Lin JC, Tiong SL, Chen CY. Surface characterization and platelet adhesion studies on fluorocarbons prepared by plasma-induced graft polymerization. J Biomater Sci Polym Ed 2000;11:701–14. 8. Liu T-Y, Lin W-C, Huang L-Y, Chen S-Y, Yang M- C. Surface characteristics and hemocompatibility of PAN/PVDF blend membranes. Polymer Adv Technol 2005;16:413–9. 9. Kolandaivelu K, Swaminathan R, Gibson WJ, et al. Stent thrombogenicity early in high-risk interventional settings is driven by stent design and deployment and protected by polymer-drug coatings. Circulation 2011;123:1400–9. 10. Eppihimer MJ, Sushkova N, Grimsby JL, et al. Impact of stent surface on thrombogenecity and vascular healing: a comparative analysis of metallic and polymeric surfaces. Circ Cardiovasc Interv 2013;6:370–7. 11. Lowell JR. Our Contributors: Edgar Allen Poe. Philadelphia, PA: George Rex Graham, 1845:49–53. Key Words: Bayesian analysis - coronary stent(s) intervention - randomized trials.
-
percutaneous coronary
