We retrospectively identified all Stanford type-A aortic dissection patients undergoing surgery between April 2017 and April 2019 at Xiangya Hospital and the Second Xiangya Hospital of Central South University, China. The dissection is considered as AAD based on the onset of symptoms <14 days prior to admission (12). Inclusion criteria: All Stanford type-A aortic dissection patients undergoing surgery. A total of 317 patients with AAD were identified. Exclusion criteria: Patients were excluded because of pregnancy, as were those with infection, immunodeficiency syndrome, cancer, sub-acute or chronic dissection and those with missing data of preoperative blood cell count. The remaining 295 patients were included in the study population (Supplementary Figure 1).

A cohort of 40 consecutive AAD patients receiving total arch replacement were prospectively identified and recruited at the time of admission between June 2019 and January 2020. Inclusion criteria: All Stanford type-A aortic dissection patients undergoing surgery. Exclusion criteria: The study subjects with immunodeficiency syndrome, cancer, sub-acute or chronic dissection, coronary heart disease, diabetes, heart failure and cerebral vascular disease were excluded. And those who recently had a surgery or infectious diseases were also excluded (Supplementary Figure 2). Fresh blood samples for all experiments were collected within an hour before the induction of anesthesia. A control group consisted of 20 healthy age- and gender-matched subjects, each providing a single morning blood sample. Only healthy control subjects were recruited (Supplementary Table 1). All prospective participants provided written informed consent. Ethical approval for the study was granted by the institutional medical ethics review board of the Second Xiangya Hospital. This study was registered in Chinese Clinical Trial Registry (ChiCTR), with registration number: ChiCTR1900023815.

All of the methods were in accordance with the Declaration of Helsinki. All baseline data were done by one member in research team, and outcome documents were collected by a different team member blinded to baseline data. All documentation was analyzed by a third member in group.

Peri-operative surgical management and clinical practices at the two centers were similar and followed the procedure previously described (13–15). In brief, arterial cannulation was done through the right axillary artery. Femoral artery cannulation was occasionally chosen in the case of dissection in the right axillary artery or high pump pressure. Antegrade cerebral perfusion was started after the arrival of target cooling temperature. After completing the anastomosis, perfusion in the lower body was resumed, the CPB flow was gradually returned to 2.0–2.4 L/m²/min, and rewarming was initiated. During the rewarming phase, the branches of the aortic arch were reconstructed. After operation, the patients were transported to the intensive care unit (ICU).

The primary end point was the incidence of MAEs during hospitalization. Postoperative complications included acute kidney injury (AKI), infection, arrhythmia, myocardial infarction, cerebrovascular accident, spinal cord injury, re-intubation, re-operation. Mortality was defined as in-hospital mortality. Patients meeting at least one criterion, were classified as suffering from postoperative MAEs.

The database included pre-operative demographic data, medical history, laboratory results, intraoperative surgical related factors and postoperative complications. Malperfusion was defined as occlusion of the vessels observed by contrast-enhanced CT or the symptom of ischemia or infarction. Arrhythmia is defined as any clinically apparent heart rhythm disturbance, including atrial fibrillation, supraventricular tachycardia, and sudden cardiac arrest. Infection includes one or more of the following: pneumonia, deep sternal wound infection, urinary tract infection, and septicemia. AKI is defined as a two-fold increase in baseline creatinine, or the need for renal replacement therapy (16). Cerebrovascular accident include transient ischaemic attack, stroke, or cerebral haemorrhagic events during postoperative period (17). Mechanical ventilation time is defined as the period between patient admission into ICU and extubation. Re-intubation was defined as re-intubation for any reason during hospitalization after extubation. Re-operation was defined as the need for re-operation for any reason during hospitalization after the initial cardiac procedure (18).

Leukocyte analysis was done on blood collected in ethylene diaminetetra acetic acid (EDTA)-treated tubes using automatic analyzers under standard operating procedures approved for clinical use. The preoperative blood results (complete blood counts) of each patient were identified, and those closest to the time of surgery were recorded. Lymphopenia was indicated by a total lymphocyte count of <1,000/μl (19).

Blood was collected in 4 ml EDTA-treated tubes (BD Biosciences, California, USA). Flow cytometry was performed on whole blood within 4 h after blood collection. Absolute numbers of T and B lymphocytes were quantified using TruCount tubes (BD Biosciences, California, USA). Absolute counts of CD45⁺ cells, CD3⁺, CD4⁺, CD8⁺ T lymphocytes and CD19⁺ B lymphocytes were analyzed on a 5-color BD TruCount flow cytometric assay, as described previously (20).

To measure apoptosis by flow cytometry, we stained the cells with Annexin V and propidium iodide (PI) (BD Biosciences) in order to estimate the rate of early-phase apoptosis, late-phase apoptosis or necrosis in each sample. PI indicates late apoptosis or necrotic cells. Cells that are positive for Annexin V, but not for PI are considered to be in early-phase apoptosis (21). To this end, cells were washed twice with cold PBS (Gibco) and resuspended at a concentration of 1 × 10⁶ cells/ml in Annexin V-binding buffer (BD Biosciences, San Jose, California, USA). They were then incubated for 15 min at room temperature with 5 μl Annexin V and 5 μl PI. 400 μl of Annexin V-binding buffer was then added and samples analyzed on a FACS Calibur cytometer (Cytek) using FlowJo v10 (Tree Star Corp) software. Every measurement includes 10⁴ cells.

To detect apoptosis, PBMCs were isolated and immediately resuspended at 1 × 10⁶ cells/ml. Multicolor cytofluorimetric analysis was then done using CD3-PE/cy7, CD4-percp-cy5.5, CD8-BV510, CD19-APC (all from BioLegend) antibodies. This analysis was done by automatic compensation for minimized fluorescence spillover and by using fluorescence minus one (FMO) control to establish positive/negative boundaries.

To detect pyroptosis, PBMCs were stained with CD3-PE/cy7, CD4-percp-cy5.5, CD8-BV510 and CD19-APC (all from BioLegend) antibodies. They were then incubated with fluorochrome-labeled Caspase-1 Inhibitors (Immunohistochemistry Technologies), which irreversibly bind to activated caspase-1.

Patient clinical characteristics and postoperative outcomes were presented as frequencies and percentages for categorical variables and compared using chi-square test or Fisher exact test. Normally distributed continuous variables were presented as mean and standard deviations (SD) and compared using Student t test while non-normally distributed variables were presented as medians and interquartile 25th and 75th percentiles (IQRs) and compared using the non-parametric Mann-Whitney U tests. Survival curves within time in ventilation, length in ICU stay and hospital stay were plotted by the Kaplan Meier (KM) method and compared by log-rank test. For the retrospective analysis, multivariable logistic regression was used to evaluate the independent predictive value of pre-absolute lymphocyte count (ALC) in primary study endpoints and to adjust for possible confounding factors. Variables with p < 0.10 from univariable analyses results were considered confounders in the multivariable logistic regression analysis. Results of the logistic regression model are given as odds ratio (OR) and 95% confidence interval (CI). For the prospective cohort, CD4⁺ T cells counts as a predictor for postoperative outcomes were estimated by receiver operating characteristic (ROC) curve analysis. Youden's index was defined for all the points along the ROC curve, and the maximum value of the index was used as a criterion for selecting the optimum cut-off point. The ability of the cut-off value for CD4⁺ T cells counts to predict postoperative outcomes were further evaluated by using multivariable logistic regression analysis. We hypothesized that the area under the curve of the MAE of CD4⁺ T cells counts would be 0.8. We calculated the required sample size for the ROC analysis. Considering the α error of 0.05, 90% power, and sample size ratio in the negative/positive group of 1, among MAEs and No MAEs, 34 patients were needed. Considering the attrition rate of 10%, at least a total of 38 AAD patients were included in the study. Actually, 40 AAD patients were included in our study. Linear regression was used to analyze the influence of apoptosis rate of lymphocyte (CD3⁺, CD4⁺ and CD8⁺ T cells) on their absolute counts. Data analysis was done using SPSS 23.0 software (SPSS, Chicago, IL, USA). All tests were two-sided and considered statistically significant at p < 0.05.

Clinical baseline characteristics in AAD are shown in Table 1. Demographic characteristics of lymphopenia and non-lymphopenia populations were similar in terms of pre-operative ejection fraction (EF) value, serum creatinine levels, hypothermic circulatory arrest (HCA) time, HCA temperature, aortic cross clamp time (ACCT), CPB time and comorbidities. Male gender and smoking status were more frequent among the group with non-lymphopenia (p < 0.05). Relative to non-lymphopenia patients, lymphopenia correlated with more advanced age, longer time of operation (p < 0.05) and more percentage of coronary heart disease (13 vs. 6%; p = 0.028). The percentage of lymphopenia patients undergoing coronary artery bypass grafting (CABG) surgery (13 vs. 4%; p = 0.011) and total arch replacement (98 vs. 91%; p = 0.014) was higher than non-lymphopenia patients. Other significant differences included shorter time of symptom onset to hospital presentation (12 vs. 21 h; p < 0.001) and lower pre-operative monocyte count (0.6 vs. 0.8 × 10⁹/L; p < 0.001).

Analysis of correlation between lymphopenia and postoperative MAEs revealed significantly higher rates of AKI, infection, arrhythmia, myocardial infarction, mortality, and overall MAEs in lymphopenia patients relative to non-lymphopenia group (p < 0.05, Table 2). 37% of the lymphopenia patients developed AKI after surgery, compared to17% in controls (p < 0.001). Mortality rate was higher in the lymphopenia relative to non-lymphopenia group (10 vs. 5%; p = 0.042). No difference was observed in lymphopenia patients relative to non-lymphopenia group in terms of cerebrovascular accident, spinal cord injury, re-intubation, and re-operation (Table 2).

Multivariable analysis showed that pre-operative lymphopenia independently correlated with increased risk of MAE after surgery (OR, 4.152; 95% CI 2.434–7.081; p < 0.001) (Tables 3, 4). Other predictors include organ malperfusion (OR, 2.481; 95% CI 1.432–4.298; p = 0.001), and CPB time (OR, 1.285; 95% CI, 1.063–1.552; p = 0.010). As shown in Supplementary Figure 3, KM curves indicated patients with lymphopenia had longer time in ventilation during a 72-h followed-up periods (p = 0.001), more length of ICU stay (p < 0.001) and hospitalizations (p = 0.005).

Next, we sought to identify the lymphocyte subgroups that correlate with clinically significant outcomes. We found that lymphopenia in AAD patients is primarily due to the reduction of T cells (Figure 1). CD3⁺ T cells counts, but not B cells, were significantly reduced in the AAD group compared to healthy donors (HDs) (CD3⁺ T cells: 615.6 ± 327.3 vs. 1,175.0 ± 264.3 cells/μl, p < 0.0001; CD19⁺ B cells: 177.8 ± 118.3 vs. 228.3 ± 94.5 cells/μl, p = 0.1048). Among CD3⁺ T cells, reduced CD4⁺ and CD8⁺ T cells were observed in the AAD group as compared with HD group (CD4: 346.7 ± 183.6 vs. 659.0 ± 214.6 cells/μl, p < 0.0001; CD8: 219.5 ± 178.4 vs. 354.4 ± 121.8 cells/μl, p = 0.0036) (Figure 1).

When stratified according to MAEs incidence, lymphocyte subset analysis showed that patients who developed complications had lower preoperative CD3⁺ and CD4⁺ T cells levels relative to those with an uneventful recovery (CD3⁺ T cells: 768.9 ± 302.4 vs. 476.9 ± 289.9 cells/μl, p = 0.0035; CD4⁺ T cells: 447.6 ± 179.0 vs. 255.4 ± 135.8 cells/μl, p = 0.004, No MAE group vs. MAE group) (Figures 2A,B). In contrast, CD8⁺ T cells and B cells did not differ between no MAE group and MAE group (CD8⁺ T cells: 264.3 ± 187.3 vs. 178.9 ± 163.9 cells/μl, p = 0.1326; CD19⁺ B cells: 188.3 ± 122.1 vs. 171 ± 116.2 cells/μl, p = 0.6505) (Figures 2C,D). Similar observations show CD4⁺ T cells levels relative to those with AKI and infection (Figures 2E–L). In addition, we found that the pre-operative CD4⁺ T cells counts at a cut-off value of 357.96 cells/μl is an effective and reliable predictor of MAEs (area under ROC curve 0.817; 95% CI, 0.684-0.950; p < 0.005). This cut-off achieves a 74% sensitivity and 81% specificity, supporting the hypothesis that CD4⁺ T lymphopenia may predispose to poor AAD surgical outcomes (Supplementary Figure 4). After adjustment for age, organ malperfusion, symptom onset to surgery, multivariate analysis revealed CD4⁺ T cells count as being independently correlated with elevated MAEs (HR, 9.384; 95% CI 1.85–47.59; p = 0.009) (Table 5; Supplementary Table 2).

We assessed the spontaneous cell death in AAD patients. Pyroptosis and apoptosis are two major types of active cell death. This analysis did not reveal differences between AAD patients and healthy subjects in terms of lymphocyte pyroptosis (Figure 3A). Of note, lymphocytes undergoing apoptosis were markedly elevated in AAD patients relative to those in healthy subjects (Figure 3B). The apoptosis rate of AAD T lymphocytes was significantly higher relative to that of healthy subjects (8.100 ± 3.958 vs. 22.12 ± 9.512%, p < 0.0001). Separate analysis of changes in T cells subsets, revealed that AAD patients exhibited significantly higher CD4⁺ (6.243 ± 3.168 vs. 20.010 ± 9.054%, p < 0.0001) and CD8⁺ T cells (8.003 ± 5.963 vs. 23.720 ± 12.920%, p < 0.0001) apoptosis relative to healthy subjects. Correlation analysis showed that CD4⁺ and CD8⁺ T lymphocytes apoptotic rates were inversely correlated to their absolute counts (Figure 3C).