Author + information
- Thomas K. Jones, MD∗ ()
- Division of Pediatric Cardiology, University of Washington School of Medicine, Seattle Children’s Hospital, Seattle, Washington
- ↵∗Address for correspondence:
Dr. Thomas K. Jones, University of Washington School of Medicine, Seattle Children’s Hospital, 4800 Sand Point Way, North East, Seattle, Washington 98105.
One of the most gratifying achievements in medicine over the past several decades has been the success of surgical and transcatheter treatment of complex congenital heart disease. The improvement in quality survival to adulthood for these patients rivals other breakthroughs in modern medicine including prematurity, childhood leukemia, and other chronic diseases of childhood (1–3). Those who care for children and adults with complex congenital heart disease should feel a sense of accomplishment in these improved outcomes. However, improved survival for our patients has come at the cost of other time-related and dynamic risks, among which is the burden of foreign tissue in the heart, specifically bioprosthetic heart valves and conduits within the right ventricular outflow tract (RVOT). Among the hazards patients with foreign tissue in the RVOT face, infective endocarditis (IE) remains one of the most concerning and least well understood. Risk factors for IE in this population are difficult to quantify, and the available published reports can best be described as nascent and ambiguous (4).
In this issue of JACC: Cardiovascular Interventions, Sharma et al. (5), report a systematic review of the published reports of IE in patients with bovine jugular vein valves compared with other valve types. Their indictment: bovine jugular valves, independent of implantation technique, are associated with a higher cumulative risk of IE compared with other valve types. The authors approach their prosecution of bovine venous valves with a systematic review of IE reported in the published data occurring in patients with surgical or transcatheter valves implanted within the RVOT, comparing the overall incidence of infection in patients with bovine jugular venous valves to all other valve types. By the authors’ own admission, there are significant limitations to their study design that deserve further scrutiny and debate before a verdict can be fairly rendered.
The authors should be congratulated for organizing and performing the hard work of accumulating and analyzing the available medical reports that relate to the troublesome topic of IE in patients whose surgical or transcatheter treatment included bioprosthetic valves or tissue conduits within the RVOT. They then performed a meta-analysis of published articles pertaining to the broad area of RVOT intervention to test their hypothesis that the cow is guilty as charged. Meta-analysis has proven to be a powerful tool when properly applied to challenging clinical questions where prospective, controlled clinical data are lacking. By combining the accumulated experience in the published data, such analysis can offer insight about a treatment effect not possible with smaller studies where a myriad of potential biases may exist. However, meta-analysis, just like the clinical studies on which it is based, is by no means free of the same potential for bias. To explore this concern further, let us use the study limitations described by the authors themselves to determine whether these biases are so overwhelming as to call the verdict into question.
Firstly, the authors state that, “There is an obvious difference in case ascertainment given that the Melody studies are prospective while the surgical series are retrospective.” The authors go on to admit these same surgical series likely under-report IE events that did not result in surgical intervention. Even more importantly, and fundamental to a rational debate on IE risk, the authors acknowledge their inability to determine the time point at which IE occurred in follow-up. Reporting a time-dependent event like IE only in terms of a crude cumulative incidence severely limits insight into a deeper understanding of risk factors for the disparate populations lumped together in this analysis. Furthermore, the analysis of IE over time presented in Figure 3 can be viewed simply as a graphic portrayal of the case ascertainment bias already admitted by the authors (5). This inability to account for IE time-to-event on a patient-specific level severely limits any conclusions about the relative risk factors for IE among the diverse valve types studied.
The authors also recognize the serious limitation imposed by the lack of a uniform definition of valve-related endocarditis in the publications they analyzed. The lack of uniformly applied diagnostic criteria across these case series further clouds the issue, especially when strict definitions including positive imaging data were required in the prospectively enrolled Melody valve (Medtronic, Dublin, Ireland) clinical studies, whereas several of the surgical bioprosthetic valve studies make no mention whatsoever of how the diagnosis of endocarditis was established or confirmed.
Inherent in this systematic review, by the authors own admission, is the high probability of patient overlap, that is double or even triple counting individual cases of endocarditis. The author’s solution was to perform an analysis, not further specified, which would avoid overlap but would maximize patient numbers, a seemingly contradictory strategy. The difficult task of accurately identifying patient-level data to accurately account for unique cases of IE would have required the authors to contact every primary or corresponding author of each analyzed publication to obtain the patient-level data required to adequately overcome this important source of potential bias. A recent systematic review by McElhinney (4) of reports of IE after implantation of the Melody valve highlights the significant issue of duplicate case reporting and the painstaking work required to accurately identify unique cases suitable for further rigorous statistical analysis.
Finally, the most important limitation of this study, and one not defined as such by the authors, is the vast heterogeneity of the populations studied. Patients who are candidates for a bioprosthetic pulmonary valve replacement are quite different from those receiving a surgical Contegra conduit (Medtronic) or transcatheter pulmonary valve replacement (TPVR). Pulmonary valve replacement with a bioprosthesis is more commonly performed in an orthotopic position in older patients with widely patent native RVOTs where blood flow profiles are optimized when compared with patients treated with surgical RVOT conduits or TPVR who are often younger and whose conduits may lie in heterotopic positions making them prone to greater blood flow turbulence, a decades old, well-known pathophysiological factor for increased IE risk (6). Similarly, TPVR is almost exclusively performed in patients with existing dysfunctional RVOT conduits or biosprosthetic valves where additional comorbidities related to their prior surgical procedures and higher burden of prosthetic material already exists. Even within specific subsets of RVOT conduit types including Contegra conduits, wide variation in the incidence of IE rates has been reported, prompting commentary that greater insight about risk factors can only be achieved using multi-institutional matched-patient comparisons, an approach not attempted nor feasible in the present study. Indeed, a fundamental caveat of meta-analysis suggests the pooling of individual data may not be appropriate when high heterogeneity is evident (7,8).
Despite the author’s genuine attempt at resolving these important study limitations through sensitivity analysis and other statistical manipulations, underlying questions about this systematic review remain, namely, publication bias and small-study effect, use of raw IE incidence rates rather than actuarial analysis of freedom from valve endocarditis over time, and the significant heterogeneity of the patient populations reported. For direct comparisons of IE rates to be meaningful, there must be strict, standardized definitions of diagnoses and outcomes, prospectively enrolled patient populations that are similar, and most importantly, the recognition that endocarditis is a time-related adverse outcome that is neither linear nor predictable. In the absence of consistent reports that adhere to these principles, extreme care is called for when attempting to generalize from disparate single-center experiences using systematic reviews of incomplete data.
Therefore, the pertinent question of whether bovine jugular vein valves impose a significantly greater ongoing hazard of endocarditis compared with other valve types remains open to honest debate. Thus far, the trial has ended in a hung jury. In the meantime, a more meaningful and useful way to reframe the debate would be to consider the global risk-to-benefit ratio of each therapeutic option within the context of like-patient groups. An infant facing their first RVOT conduit implantation with at least 1 conduit replacement, and then a TPVR before adulthood, faces fundamentally different, ongoing, and cumulative risks that are far different than a 25-year-old undergoing their first surgical bioprosthetic pulmonary valve implantation. In the context of the care of the former patient, even if bovine jugular vein tissue is associated with a higher infection risk, these valves may still prove to be of greater overall benefit for that patient population. At present, we do not have enough undisputed testimony to pass judgment on the cow. Any future indictment must adhere to principles of fairness outlined in the preceding text and take into account the extenuating circumstances under which our defendant must perform.
↵∗ Editorials published in JACC: Cardiovascular Interventions reflect the views of the authors and do not necessarily represent the views of JACC: Cardiovascular Interventions or the American College of Cardiology.
Dr. Jones has received research grant support from Edwards Lifesciences and Medtronic; and is a consultant for Medtronic.
- 2017 American College of Cardiology Foundation
- Chamberlain L.J.,
- Fernandes S.M.,
- Saynina O.,
- et al.
- Hunger S.P.,
- Lu X.,
- Devidas M.,
- et al.
- McElhinney D.B.
- Sharma A.,
- Cote A.T.,
- Hosking M.C.K.,
- Harris K.C.
- Robard S.
- Morales D.L.S.,
- Khan M.S.
- Greco T.,
- Zangrillo A.,
- Biondi A.,
- Landoni G.