Author + information
- †Department of Medico-Surgical Sciences and Biotechnologies, Sapienza University of Rome, Latina, Italy
- ‡Department of Cardiovascular, Respiratory, Geriatric and Nephrologic Sciences, Umberto I Hospital, Sapienza University of Rome, Rome, Italy
- ↵∗Reprint requests and correspondence:
Dr. Giuseppe Biondi-Zoccai, Department of Medico-Surgical Sciences and Biotechnologies, Sapienza University of Rome, Corso Della Repubblica 79, 04100 Latina, Italy.
Ignorance is the curse of God, knowledge the wing wherewith we fly to heaven.
—Henry VI, Part II, Act 4, Scene 7, William Shakespeare (1564–1616)
Despite ongoing refinements in techniques, devices, and ancillary medical care, the outlook of patients undergoing cardiac surgery is still suboptimal, especially when comorbidities are present (1). Risk prediction is an intense field of research, yet prognostic studies often provide conflicting results, undermining clinicians' confidence when weighing a given risk factor for decision making or when trying to amend it (2). The Society of Thoracic Surgeons score and European System for Cardiac Operative Risk Evaluation II score are 2 of the most common surgical risk scores (1,3). Yet only some clinical variables are shared by the 2 scores, with several of them being unique to only 1 of the 2. This is clear evidence of the challenges inherent in everyday risk prognostication in cardiac surgery.
Several explanations can be tentatively advanced to elucidate such limitations in the internal and external validity of prognostic studies. First, differences in study design, setting, and patient selection criteria may lead to heterogeneous findings. Second, disparate strategies for patient management and follow-up may influence the direction and magnitude of effect estimates. Finally, diagnostic thresholds, outcome definitions, and statistical approaches may well change the study results, as well as their precision and robustness. This is why debate still persists on how to best identify patients at risk in several important clinical scenarios, not limited to cardiothoracic surgery (4).
Renal artery stenosis in patients undergoing cardiac surgery represents a paradigm in this sense. There are several pieces of evidence suggesting that renal artery stenosis may be associated with unfavorable renal and cardiovascular outcomes (Fig. 1), especially when both renal vessels are diseased (5,6). Yet the validity of these data is suboptimal, and conflicts in such reports are apparent (7,8). Moreover, what should a clinician do when renal artery stenosis is recognized and cardiac surgery is imminent or mandatory? Before addressing this question, it is better to clarify what we mean when we state that a patient has renal artery stenosis and why this could negatively affect the patient.
Renal artery stenosis can be defined in terms of etiology, being most commonly due to atherosclerosis and much more rarely to fibromuscular dysplasia or inflammatory diseases such as Takayasu arteritis (9,10). In addition, a distinction is necessary between renal artery stenosis diagnosed using morphologic techniques (luminologic or tomographic), such as angiography and computed tomography, and that diagnosed using modalities that combine both morphologic and functional features, such as duplex ultrasound, nuclear scan, and magnetic resonance imaging. Although purely morphologic techniques tend to be more objective and reproducible, they only indirectly estimate blood flow (11,12). Conversely, imaging modalities including functional evaluation may bear more clinical relevance for decision makers because they provide a more comprehensive assessment of the pathophysiology of the renal circulation. Finally, novel approaches based on invasive renal artery pressure gradient assessment should also be borne in mind (13).
This issue of the Journal brings a new piece to the puzzle, encouraging us to guess at a solution. Philip et al. (14) exploited a large administrative institutional database of 37,000 patients undergoing cardiac surgery between 2000 and 2010 at the Cleveland Clinic, concomitantly querying a diagnostic vascular database. They thus provide data on 714 patients undergoing first-time open-chest cardiac surgery and also receiving renal duplex ultrasound within 90 days of surgery. They found that atherosclerotic renal artery stenosis was rather common, recognized in 29% of the patients, and was even more common in those with old age, female sex, low body mass index, peripheral artery disease, and low level of high-density lipoprotein (Table 1). They analyzed this large dataset using sophisticated statistical techniques, including bootstrap resampling and machine learning tools, to test the hypothesis that renal artery stenosis may be associated with an increased risk for renal injury and ensuing clinical complications. Specifically, they focused on post-operative change in estimated glomerular filtration rate, need for renal replacement therapy, prolonged intensive care unit stay, and death. Notably, the presence of unilateral or bilateral renal artery stenosis was not significantly and independently associated with any of these endpoint. This finding was reinforced by the fact that even bilateral renal artery stenosis was not, per se, an independent predictor of unfavorable outcomes. Indeed, although unadjusted analyses did suggest the potential association between bilateral renal artery stenosis and a late decline in glomerular filtration rate, this did not hold true after adjustment and was not accompanied by concomitant detrimental impact on other outcomes.
In fact, depending on the statistical method used, low post-operative glomerular filtration rate was independently predicted instead by age, hypertension, diabetes, endocarditis, and need for aortic surgery or coronary artery bypass grafting without an internal thoracic artery. Renal replacement therapy was more common in those with low pre-operative hematocrit, low platelet count, insulin-dependent diabetes, need for ventricular support, or aortic surgery. Prolonged intensive care unit stay was predicted by female sex, need for prior ventricular support, aortic surgery, type of surgery, and low cholesterol level. Finally, early death was more common in the elderly, those with diabetes, congestive heart failure, low pre-operative hematocrit, and tricuspid or aortic valve surgery. Conversely, late death was predicted by advanced age, prior congestive heart failure, and low pre-operative glomerular filtration rate.
Although these results might appear counterintuitive to clinicians too preoccupied with luminology, it is reasonable to assume that the kidney, an organ that receives very high blood flow in proportion to its requirements, as testified by its low oxygen extraction in comparison to other organs (e.g., the heart) and tissues, is much more protected from overall decreases in blood flow due to renal artery stenosis thanks to increases in renal oxygen extraction (9,10).
This work has many important strengths. First, the presence of sophisticated statistical co-plots discloses how the association between renal artery stenosis and outcomes is likely largely mediated by confounders, including low pre-operative hematocrit, low platelet count, low cholesterol, and type of surgery (Fig. 1). Second, no confounding due to pre-emptive percutaneous or surgical revascularization is present, because all patients were only managed medically for their renal artery stenosis. Moreover, the fact that the investigators have made publicly available the R code for their refined analyses enables straightforward replication and validation of their work.
However, these remarkable findings are typical of one of the best cardiovascular centers worldwide and thus might not be easily replicated elsewhere. In addition, as many as 15% duplex ultrasound scans were performed after surgery. Although the overall results were confirmed after excluding patients with these post-operative scans, the ideal scenario would have been to rely only on pre-operative imaging tests. Glomerular filtration rate was estimated using the Modification of Diet in Renal Disease formula, which may be less than reliable post-operatively because it is based on serum creatinine, whose perioperative kinetics may be unpredictable. Although formal missing data imputation (involving up to 22% of data) may be considered a strength rather than a weakness, obviously perfectly complete data would have been the best-case scenario. Whereas statistical power was formally appraised and appeared adequate (>90%), larger samples with even greater precision could be obtained with patient-level analyses in the future. Indeed, 714 patients may not enough to perform so many multivariate analysis, especially for rare dichotomous outcomes (15).
Nonetheless, these findings, in conjunction with those comparing percutaneous revascularization versus best medical therapy for renal artery stenosis, make the compelling case that renal artery stenosis, when found in patients candidate for cardiac surgery, should not be treated (16).
In conclusion, now that ignorance of the impact of renal artery stenosis has been reasonably defeated and the best quality evidence is finally available, it appears evident that renal artery stenosis should no longer be considered per se an important prognostic factor when cardiac surgery is envisioned, nor should it be preemptively treated using percutaneous or surgical means except in very selected cases. Nonetheless, the fact that renal artery stenosis is so common in patients with severe cardiovascular disease and so frequently associated with other unfavorable prognostic features means that it should still be considered indirect evidence of advanced atherosclerosis.
↵∗ 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.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Howell N.J.,
- Head S.J.,
- Freemantle N.,
- et al.
- Topol E.J.,
- Nissen S.E.
- Drieghe B.,
- Madaric J.,
- Sarno G.,
- et al.
- Philip F.,
- Gornik H.L.,
- Rajeswaran J.,
- Blackstone E.H.,
- Shishehbor M.H.