Frontline Science: Low regulatory T cells predict perioperative major adverse cardiovascular and cerebrovascular events after noncardiac surgery

Immune cells drive atherosclerotic lesion progression and plaque destabilization. Coronary heart disease patients undergoing noncardiac surgery are at risk for perioperative major adverse cardiac and cerebrovascular events (MACCE). It is unclear whether differential leukocyte subpopulations contribute to perioperative MACCE and thereby could aid identification of patients prone to perioperative cardiovascular events. First, we performed a hypothesis‐generating post hoc analysis of the LeukoCAPE‐1 study (n = 38). We analyzed preoperative counts of 6 leukocyte subpopulations in coronary heart disease patients for association with MACCE (composite of cardiac death, myocardial infarction, myocardial ischemia, myocardial injury after noncardiac surgery, thromboembolic stroke) within 30 d after surgery. Regulatory T cells (Tregs) were the only leukocyte subgroup associated with MACCE. We found reduced Tregs in patients experiencing MACCE versus no‐MACCE (0.02 [0.01; 0.03] vs. 0.04 [0.03; 0.05] Tregs nl−1, P = 0.002). Using Youden index, we derived the optimal threshold value for association with MACCE to be 0.027 Tregs nl−1. Subsequently, we recruited 233 coronary heart disease patients for the prospective, observational LeukoCAPE‐2 study and independently validated this Treg cutoff for prediction of MACCE within 30 d after noncardiac surgery. After multivariate logistic regression, Tregs < 0.027 cells nl−1 remained an independent predictor for MACCE (OR = 2.54 [1.22; 5.23], P = 0.012). Tregs improved risk discrimination of the revised cardiac risk index based on ΔAUC (area under the curve; ΔAUC = 0.09, P = 0.02), NRI (0.26), and IDI (0.06). Preoperative Treg levels below 0.027 cells nl−1 predicted perioperative MACCE and can be measured to increase accuracy of established preoperative cardiac risk stratification in coronary heart disease patients undergoing noncardiac surgery.


INTRODUCTION
Cardiovascular diseases are substantially driven by innate and adaptive immune effector mechanisms. 1 In long-term prospective clinical trials, leukocyte subpopulations including classical, 2 intermediate, 3 and nonclassical monocytes, 4 as well as natural killer 5 and regulatory T cells (Tregs), 6 were ascribed an association with cardiovascular disease and have been shown to predict cardiovascular events. Acute perioperative stress during major noncardiac surgery implies significant immunomodulatory and inflammatory changes 7,8 associated with patients' susceptibility to major adverse cardiovascular and cerebrovascular events (MACCE). 7,9 MACCE related to noncardiac surgery are among the leading causes of perioperative morbidity and mortality. 7 Mechanisms underlying perioperative MACCE are incompletely understood. As a consequence, tools to accurately identify vulnerable patients prone to develop cardiovascular complications are limited and uncertainty exists regarding the optimal risk stratification model.
Clinical practice guidelines on perioperative risk evaluation [10][11][12] advocate the use of clinical risk indices such as the revised cardiac risk index (RCRI). 13 However, the RCRI shows only low discriminatory power in high-risk and vascular surgery patients 14,15 as it is based on conditions omnipresent in high-risk cardiovascular patients.
Recent perioperative guidelines encourage additional preoperative biomarker measurements in high-risk patients to identify patients with underestimated severity of cardiovascular disease. 12,16 Still, evidence is scant and their clinical benefit is under debate. 17 Moreover, cardiac biomarkers such as troponins and natriuretic peptides rise as a result of myocardial injury and heart failure, respectively.
Thus, they certainly reflect severity of preexisting diseases, but are associated with perioperative MACCE. 18 However, it remains unknown whether a patient's individual immune status renders him prone to perioperative cardiovascular events. In particular, evidence for a potential relation between preoperative levels of leukocyte subsets with perioperative MACCE is scarce. 19 Therefore, we evaluated preoperative values of 6 predefined leukocyte subsets in a post hoc analysis of the LeukoCAPE-1 study for association with perioperative MACCE. Based on these results, we subsequently conducted the prospective LeukoCAPE-2 study to validate the predictive cutoff value derived for preoperative Treg findings in an independent cohort.

Study design and population
The leukocytes and cardiovascular perioperative events-1 and -2 were documented before surgery. Conventional risk evaluation was based on high-sensitive cardiac troponin T (hs-cTnT) and N-terminal pro-brain natriuretic peptide (NT-proBNP). Patients with hs-cTnT

Laboratory measurements
Blood samples were collected and processed as described before. 18 NT-proBNP was measured preoperatively (Immulite, Siemens Healthcare Diagnostics, Erlangen, Germany); hs-cTnT was determined preoperatively and daily on postoperative days 1 until 3 (POD1-3; Cobas E4111, Roche Diagnostics, Mannheim, Germany). Automated differential blood counts were performed in the central laboratory.  data. All disagreements in ECG interpretation between the two physicians were discussed with a third physician and were resolved in consensus.

Detailed definitions of primary outcome variables
Cardiac death was defined as any death presumably of cardiac origin.

Sample size calculation
The sample size (n = 40) of the LeukoCAPE-1 study was initially calculated for testing the association of leukocyte subpopulation counts with noncardiac surgery. 18

Study population
A detailed description of the LeukoCAPE-1 patient flow and baseline characteristics have been published before. 18  Proportion of ASA 2 status was lower in the MACCE group. The majority of patients underwent general anesthesia alone (59%) or in combination with regional anesthesia (25%). Regional anesthesia (8%) and analgo-sedation alone (8%) were conducted less frequently. Surgical risk according to the European Society of Cardiology/European Society of Anaesthesiology 11 did not differ between the two groups (Table 1). Overall, baseline characteristics in both studies were similar.

Derivation of preoperative Treg threshold value derived from LeukoCAPE-1 post hoc analysis
For the main analysis of the LeukCAPE-1 study, WBCs and selected leukocyte subpopulations known to be associated with an increased cardiovascular risk were quantified in elevated-risk patients before and at different time points after noncardiac surgery. 18 Post hoc, we   In multivariable logistic regression analysis, Treg levels were entered as dichotomized variable and remained an independent risk factor for perioperative MACCE after adjustment for age, gender, ASA physical status, history of PCI, and creatinine levels (

Additive risk predictive value of Tregs
We next aimed to elucidate whether the addition of Treg levels improved risk stratification of guideline-recommended perioperative risk predictors. Therefore, we performed three measures (AUC, NRI, and IDI) with all markers being analyzed as continuous variables.
Measures of diagnostic accuracy for preoperative Treg levels and conventional risk predictors are reported in Table 2. Even though the RCRI is recommended by current guidelines and is therefore commonly used, its predictive accuracy is limited especially in cardiovascular risk patients. 30 As recent North American and European guidelines recommend preoperative cardiac risk assessment using both, the RCRI and additional measurements of biomarkers, 11,12,16 we first quantified the additive predictive value of preoperative  Data are expressed as odds ratios (OR) and 95% CI.MACCE: Major adverse cardiovascular and cerebrovascular events; PPV/NPV: positive/negative predictive value; RCRI: revised cardiac risk index; hs-cTnT: high-sensitive cardiac Troponin T; and NT-proBNP: N-terminal pro-hormone brain natriuretic peptide.  Figure A).

F I G U R E 4 Kaplan-Meier curves. (A) Patients
We next considered a combination of RCRI, NT-proBNP, and hs-cTnT as the basic risk model. AUC for the basic risk model was Data are expressed as odds ratios (OR) with 95% CI. Multivariable logistic regression modeling included factors that revealed a P value below 0.1 in univariate analysis and was conducted by means of a forward stepwise (Wald) technique. To avoid redundancy only one criterion for renal impairment was included. We chose creatinine over KDIGO stage, as creatinine is a continuous variable. We did not choose eGFR because it depends on age and gender, two variables already included in the multivariable analysis. Tregs: regulatory T cells; BMI: body mass index; ASA: risk classification according to the American Society of Anesthesiologists; KDIGO: Kidney Disease: Improving Global Outcomes; PCI: percutaneous coronary intervention; CABG: coronary artery bypass grafting; ACE: angiotensin converting enzyme; and eGFR: estimated glomerular filtration rate calculated by Chronic Kidney Disease Epidemiology Collaboration.  Figure 6B).

DISCUSSION
Here we report the derivation and independent validation of a pre- We assessed perioperative levels of 6 different leukocyte populations, which are associated with long-term cardiovascular risk. In our cohort, Tregs were the only cell population that differed between patients with and without perioperative MACCE. Numerous experimental and clinical studies have proven a protective role of Tregs in cardiovascular disease. In one previous study, the authors report small preoperative differences for some lymphocyte populations including Tregs in a highly selected patient population, but did not assess the potential for risk prediction. 19 Tregs constitute 5-10% of all peripheral CD4 + T lymphocytes and are considered negative regulators of cellular immunity. 31 Tregs help maintaining self-tolerance, T cell homeostasis and they are specialized for the suppression of pathogenic immune responses against self-and foreign antigens. 32  international large-scale study using the hs-cTnT assay. 26 We chose this definition for the subsequent LeukoCAPE-2 study accepting the likely increase in the event rate compared to LeukoCAPE-1. However, in Figure S3 we provide an additional post hoc analysis of the  38,46 and is proposed by the Human Immunophenotyping Consortium. 47 In addition, CD127 inversely correlates with FoxP3. [48][49][50] Therefore, selecting CD127 low leukocytes yields a population highly enriched with FoxP3 + cells. However, we cannot fully exclude that the CD4 + CD25 high CD127 low marker combination identified FoxP3 + non-Treg cells or that we missed some FoxP3 + Tregs.
In summary, results from our two independent studies suggest that reduced preoperative Treg levels independently predicted 30 d