JOURNAL OF VASCULAR SURGERY July 2014

270 Letters to the Editor

peripheral arterial disease, and other well-known risk factors. In addition, the procedure as well as follow-up investigations would have to be standardized with a robust end point, like verification of patency with duplex ultrasound. With all this taken into consideration, it is obvious that the efforts required to perform such a study would be substantial. This is with regard to logistical and financial issues but also with regard to the number of patients that would be needed for meaningful results to be obtained. In the age of evidence-based medicine, increasing health care costs, and need for standardization to increase efficacy, we will have to tackle this task sooner or later. The fact that young female patients have a higher life expectancy compared with men adds to the need of an optimization of treatment strategies in a patient cohort that has previously been identified as problematic. Retrospective studies like ours serve to facilitate decision-making as to whether more evidence is needed and what the best way might be to obtain it. Thus, whether gender does have an impact on the outcome of patients with peripheral arterial disease requiring iliac artery intervention will be answered only in a collaborative effort, an effort that is worthwhile because it could lead to practicechanging results. Beate Bechter-Hugl, MD Department of Vascular Surgery Medical University Innsbruck Innsbruck, Austria REFERENCES 1. Bechter-Hugl B, Falkensammer J, Gorny O, Greiner A, Chemelli A, Fraedrich G. The influence of gender on patency rates after iliac artery stenting. J Vasc Surg 2014;59:1588-96. 2. Timaran CH, Stevens SL, Freeman MB, Goldman MH. Predictors for adverse outcome after iliac angioplasty and stenting for limb-threatening ischemia. J Vasc Surg 2002;36:507-13. 3. Klein WM, Van der Graaf Y, Seegers J, Moll FL, Mali WP. Long-term cardiovascular morbidity, mortality, and reintervention after endovascular treatment in patients with iliac artery disease: the Dutch Iliac Stent Trial Study. Radiology 2004;232:491-8. 4. Goode SD, Keltie K, Burn J, Patrick H, Cleveland TJ, Campbell B, et al. Effect of procedure volume on outcomes after iliac artery angioplasty and stenting. Br J Surg 2013;100:1189-96.

Second, although the study was constructed on the basis of whole blood count, further details about the features and performance characteristics of the whole blood analyzer were not specified. It is well known that the reference ranges of whole blood count subtypes may vary with many factors, such as the population studied, individuality of laboratory and instruments (eg, types of collection tubes), and measurement methods used (eg, waiting period before analysis).5,6 Third, instead of using cutoffs determined by previous studies, it would be more appropriate to detect the optimum NLR cutoff point, with highest sensitivity and specificity, using the receiver operating characteristic curve analysis. This improper preference of cutoffs may have led to a bias in patient selection. Because, as seen in Table 2 of the original study, there are 360 patients with NLR .05 is not possible to reach the statistical significance in multivariate logistic regression analysis. This is another important statistical issue that should be taken care of by the authors. Instrumental features must be specified as they are easily affected by analytical and preanalytical variables in studies based on laboratory results. Unsettled parameters like NLR should be corroborated by known inflammation markers. Besides, determining a new cutoff value for these subjects will make the study more valuable because cutoff points may vary according to the patient groups in which the study was carried out. We are particularly grateful for the assistance given by Assoc. Prof. Serkan Tapan. Huseyin Kayadibi, MD, MSc Department of Medical Biochemistry Adana Military Hospital Adana, Turkey Metin Uyanik, MD

http://dx.doi.org/10.1016/j.jvs.2014.03.276

Department of Medical Biochemistry Gulhane School of Medicine Ankara, Turkey

Regarding “Neutrophil-lymphocyte ratio as a predictor of cognitive dysfunction in carotid endarterectomy patients”

Erdim Sertoglu, MD

We read with great interest the recently published article by Halazun et al1 that aimed to investigate the neutrophillymphocyte ratio (NLR) as a predictor of cognitive dysfunction in patients with carotid endarterectomy. They concluded that preoperative NLR is associated with cognitive dysfunction 1 day after carotid endarterectomy. However, there are some points that we would like to discuss about this study. First, NLR, which integrates the detrimental effects of neutrophilia (an indicator of inflammation) and lymphopenia (an indicator of physiologic stress), has emerged as a useful prognostic marker in many studies.2,3 In the original study, neutrophil and lymphocyte counts were not specified in detail according to the cognitive state of patients. In addition, unlike in other studies evaluating the same parameters, there were no exclusion or inclusion criteria that may affect the statistical analysis because of the probable outlier values derived from unidentified sepsis, weight loss, massive hemorrhage, and instrumental error, which make the NLR inaccurate.4 Moreover, the lack of data for other inflammatory markers has led to the insufficiency of evidence confirming the presence of inflammation in these patients.

Biochemistry Laboratory Ankara Mevki Military Hospital Anittepe Dispensary Ankara, Turkey REFERENCES 1. Halazun HJ, Mergeche JL, Mallon KA, Connolly ES, Heyer EJ. Neutrophil-lymphocyte ratio as a predictor of cognitive dysfunction in carotid endarterectomy patients. J Vasc Surg 2014;59:768-73. 2. Halazun KJ, Hardy MA, Rana AA, Woodland DC 4th, Luyten EJ, Mahadev S, et al. Negative impact of neutrophil-lymphocyte ratio on outcome after liver transplantation for hepatocellular carcinoma. Ann Surg 2009;250:141-51. 3. Gibson PH, Cuthbertson BH, Croal BL, Rae D, El-Shafei H, Gibson G, et al. Usefulness of neutrophil/lymphocyte ratio as predictor of newonset atrial fibrillation after coronary artery bypass grafting. Am J Cardiol 2010;105:186-91. 4. Demir M, Demir C. Neutrophil/lymphocyte ratio in patients with atrial septal aneurysm. Vasc Health Risk Manag 2013;9:365-8.

JOURNAL OF VASCULAR SURGERY Volume 60, Number 1

5. Sertoglu E, Uyanik M. Accurate use of neutrophil/lymphocyte ratio from the perspective of laboratory experts. Vasc Health Risk Manag 2014;10:13-4. 6. Buttarello M, Plebani M. Automated blood cell counts: state of the art. Am J Clin Pathol 2008;130:104-16. http://dx.doi.org/10.1016/j.jvs.2014.03.276

Reply Kayadibi et al raise a number of points that require clarification. First, the purpose of this report was to see if an elevated neutrophil/lymphocyte ratio (NLR) was associated with increased probability of developing early cognitive dysfunction. Second, Kayadibi et al question why “neutrophil and lymphocyte counts were not specified in detail according to the cognitive state of patients.” Previous publications emphasize the importance of the NLR ratio instead of using the individual neutrophil or lymphocyte values.1-3 We followed suit in this study in only calculating the ratio. Previously, we found an association between early cognitive dysfunction in asymptomatic patients undergoing carotid endarterectomy (CEA), and monocytes (P ¼ .01).4 Third, we studied all patients undergoing CEA who met criteria for inclusion or exclusion defined in our previous publications.5,6 However, their suggestion that outliers “may affect the statistical analysis.” is wrong. Approximately 90% of all patients were admitted from home, and the rest were recently admitted for a workup of new neurologic deficits. None of these patients were “septic, [with] weight loss, massive hemorrhage and instrumental error..” Fourth, the whole blood analyzer (WBA) used at this institution is the Sysmex XE-5000 (Sysmex, Kobe, Japan). Blood was collected in standard tripotassium ethylenediaminetetraacetic acid (K3EDTA) plastic tubes, and all samples were analyzed #8 hours of collection. Because these results included patients enrolled during a 19-year period, we could not determine which WBA was used for each sample. Even if there were differences over time in neutrophil and lymphocyte counts based on the type of WBA, there is no reason to think that the ratios would be different over time for the same patient. Fifth, we used cutoffs previously used in other studies to place our results in perspective with previous work. Kayadibi et al are incorrect in asserting that by using a receiver operating characteristic curve (ROC) analysis “the number. [of] the patients in each group would have been equally distributed.” Cutoff optimization via ROC analysis is based on optimization of sensitivity and specificity and does not distribute groups equally. If we had wanted to obtain equally distributed groups, we would have dichotomized at the median value; however, that would not require ROC analysis and would not have necessarily resulted in a clinically meaningful variable. Sixth, Kayadibi et al are mistaken in stating that a P value > .05 in univariate logistic regression automatically translates to a P value > .05 in multiple logistic regression. A variable may be associated with a nonsignificant P value in univariate logistic regression but still have a significant P value in multiple regression due to confounding or modifying effects of other covariates. It is standard to use parameters with univariate P values # .2 in a multivariable statistical model to avoid missing relevant statistical contributors to outcomes.7 Eric J. Heyer, MD, PhD Anesthesiology Columbia University New York, NY REFERENCES 1. Bhutta H, Agha R, Wong J, Tang TY, Wilson YG, Walsh SR. Neutrophil-lymphocyte ratio predicts medium-term survival following elective major vascular surgery: a cross-sectional study. Vasc Endovascular Surg 2011;45:227-31. 2. Gokhan S, Ozhasenekler A, Mansur Durgun H, Akil E, Ustundag M, Orak M. Neutrophil lymphocyte ratios in stroke subtypes and transient ischemic attack. Eur Rev Med Pharmacol Sci 2013;17:653-7.

Letters to the Editor 271

3. Shimada H, Takiguchi N, Kainuma O, Soda H, Ikeda A, Cho A, et al. High preoperative neutrophil-lymphocyte ratio predicts poor survival in patients with gastric cancer. Gastric Cancer 2010;13:170-6. 4. Mocco J, Wilson DA, Ducruet AF, Komotar RJ, Mack WJ, Zurica J, et al. Elevations in preoperative monocyte count predispose to acute neurocognitive decline after carotid endarterectomy for asymptomatic carotid artery stenosis. Stroke 2006;37:240-2. 5. Heyer E, Adams D, Todd G, Solomon R, Quest D, Steneck S, et al. Neuropsychometric changes in patients after carotid endarterectomy. Stroke 1998;29:1110-5. 6. Heyer EJ, Sharma R, Rampersad A, Winfree CJ, Mack WJ, Solomon RA, et al. A controlled prospective study of neuropsychological dysfunction following carotid endarterectomy. Arch Neurol 2002;59:217-22. 7. Hosmer DW, Lemeshow S. Applied logistic regression. 2nd ed. New York: John Wiley & Sons; 2000. http://dx.doi.org/10.1016/j.jvs.2014.03.275

Regarding “Presentation, treatment, and outcome differences between men and women undergoing revascularization or amputation for lower extremity peripheral arterial disease” We read with interest the article by Lo et al1 describing the recent trends in revascularization procedures for claudication and critical limb ischemia (CLI) in the United States. Taken together with the article by Wallace et al2 in the same issue of the Journal of Vascular Surgery, these contributions give a broad overview of the current landscape of interventions being performed for advanced peripheral arterial disease (PAD) in the United States. However, we believe that Lo et al1 have not fully considered two key factors influencing the utilization of revascularization procedures in their analysisdthe growing impact of restenosis and the influence of provider specialty/training on treatment choices. As noted by the authors, the administrative data sets used render them unable to link data longitudinally and, thus, to untangle the key relationships between procedures, unique patients, and unique limbs. Recent observations using linked Medicare or registry data provide important context for this report. First, restenosisd wherein having longitudinal data in individual patients is criticald represents a unique type of “disease progression” in PAD. The burden of restenosis in PAD is growing as well as its effect on the outcomes of secondary revascularizations.3 The data suggest that the rise in secondary procedures may be a major factor underlying overall volume trends as well as the types of interventions being used. The risk factors for restenosis are poorly understood and appear to vary between open and endovascular interventions. Further, atherosclerotic occlusive disease and restenotic lesions are approached differently. The preference to perform a certain type of intervention in a given subgroup (eg, women) could be linked to the relative prevalence of restenotic vs primary disease presentations. Although we have precious little in the way of level 1 evidence to support primary treatment choices in PAD, even less data are available to define the optimal approach for most scenarios of restenosis. For example, it seems plausible that provider specialty/training may influence the decision to repeat an endovascular intervention or move to an open bypass. We further note recent data on the wide variation in utilization and costs of invasive treatments for CLI in the Medicare population, suggesting a major disconnect between resources, procedures, and outcomes.4 Specifically, regions with the greatest spending and highest proportion of endovascular procedures also had some of the highest rates of amputation. So how do we interpret the volume trends reported by Lo et al1 in this context? That utilization rates of open and endovascular interventions per se are not directly associated with clinical effectiveness or value of care has become abundantly clear. Restenosisdand the reinterventions and outcomes that followdmust be considered in every