This retrospective cohort study assessed 757 patients diagnosed with AF who underwent first-time cryoablation at the Cardiology Department of Nanjing First Hospital between January 2017 and December 2023. The diagnosis of AF was established on the basis of the following criteria: (1) documentation of AF on standard 12-lead ECG showing irregular RR intervals and no distinct P waves for at least 30 s; (2) 24-hour Holter monitoring confirming AF episodes; and (3) classification of the AF type (paroxysmal or persistent) on the basis of the 2020 ESC Guidelines for the diagnosis and management of AF. The eligibility criteria included a definitive diagnosis of AF via medical history, electrocardiogram (ECG), or Holter monitoring, coupled with a successfully executed cryoablation procedure using a second-generation (Arctic Front Advance, Medtronic, Inc., Minneapolis, MN, USA) cryoballoon catheter. The ablation strategy was standardized according to the type of atrial fibrillation: patients with paroxysmal AF underwent pulmonary vein isolation (PVI) only, whereas patients with persistent AF underwent PVI plus posterior wall isolation (PWI). The exclusion criteria were as follows: (1) previous AF ablation; (2) end-stage renal disease; (3) thrombus in the left atrium or its appendage; (4) valvular AF; (5) refusal or nonadherence to postoperative oral anticoagulant (OAC) therapy; (6) absence of blood cell count data; (7) acute inflammatory response, characterized by a white blood cell (WBC) count >10 × 10⁹/L or C-reactive protein (CRP) > 10 mg/L, or a diminished WBC count (<4 × 10⁹/L); and (8) autoimmune connective tissue disease or use of oral steroid medication. These criteria aimed to ensure a consistent study sample and reduce potential confounders that might compromise the reliability of the results. The study adhered to the principles outlined in the Declaration of Helsinki, and its protocol was approved by the Ethics Committee of Nanjing First Hospital. All patients provided written informed consent. Patients over the age of 75 were asked to sign a dual-signature informed consent form as an additional safeguard of their rights and to ensure that their family was involved in the decision-making process. Therefore, a patient and one relative (first degree) signed the informed consent form. This measure was not a substitute for patient consent, and no legal guardians signed on a patient's behalf. A flowchart of the participants is shown in Figure 1. Only cryoablation cases were included; no RF or PFA procedures were part of this cohort. Accordingly, generalizability to other ablation modalities cannot be assumed.

Data collection was guided by biological considerations and the literature, ensuring that all potential prognostic variables were collected before ablation surgery. These variables included demographic and clinical characteristics such as age, sex, body mass index (BMI), type of atrial fibrillation (paroxysmal or persistent), duration of atrial fibrillation, and comorbidities such as smoking history, alcohol consumption history, hypertension history, cardiovascular disease (CVD) history, diabetes history, and prior stroke or transient ischemic attack (TIA). Risk stratification of patients was conducted via the CHA2DS2-VASc score. Blood samples collected within 24 h preceding the procedure were analyzed to determine the laboratory parameters. These included a complete blood count (comprising white blood cell count, neutrophil count, monocyte count, lymphocyte count, platelet count, red blood cell count, and hemoglobin level) and B-type natriuretic peptide (BNP) levels. BMI was calculated as weight (kg) divided by height (m) squared. Cardiac evaluation included recording standard 12-lead ECG parameters, such as the heart rate, PR interval, and QT interval, on the first postoperative day. Transesophageal echocardiography was conducted before ablation to determine the left atrial diameter, left ventricular ejection fraction (LVEF), and left atrial appendage flow velocity (LAAFV) while confirming the absence of intracardiac thrombi. The compiled data were meticulously reviewed by two clinicians, with any discrepancies subjected to in-depth evaluation by an electrophysiological medicine specialist within the cardiology department.

All blood samples were collected in the morning (between 7:00 AM and 9:00 AM) after an overnight fast and analyzed within 2 h of collection. The routine blood tests included white blood cell count (normal range: 4–10 × 10⁹/L), neutrophil count (2–7 × 10⁹/L), lymphocyte count (0.8–4 × 10⁹/L), monocyte count (0.12–0.8 × 10⁹/L), platelet count (100–300 × 10⁹/L), and hemoglobin level (130–175 g/L for males, 115–150 g/L for females). Several inflammatory markers are derived from these parameters: the NLR, which is calculated by dividing the absolute neutrophil count by the absolute lymphocyte count; the SII, which is calculated as the platelet count × neutrophil count/lymphocyte count; the SIRI, which is calculated as the neutrophil count × monocyte count/lymphocyte count; and the PIV, which is calculated as (neutrophil count × monocyte count × platelet count)/lymphocyte count. All indices were log-transformed (natural log) for modeling.

All patients were followed up at the outpatient clinic 1, 3, 6, and 12 months after ablation and every 6 months thereafter. The median follow-up duration was 24.5 months (interquartile range: 18.3–36.7 months). During each visit, the patients underwent 12-lead electrocardiography (ECG) and 24-hour Holter monitoring. Additionally, the patients were instructed to use a handheld ECG monitor (provided at discharge) to record their cardiac rhythm when experiencing symptoms suggestive of arrhythmia. AF recurrence was defined as any documented symptomatic or asymptomatic atrial tachyarrhythmia lasting ≥30 s after a 3-month blanking period. Time-to-event was measured from the end of the blanking period to the first documented recurrence or censoring at the last follow-up. Asymptomatic episodes were detected through routine Holter monitoring or occasional ECG examinations during the follow-up visits. For symptomatic episodes, documentation was obtained through handheld ECG monitoring, 12-lead ECG during unscheduled hospital visits, or 24-hour Holter monitoring. All documented arrhythmia episodes were independently reviewed by two electrophysiologists to ensure accurate classification, with any diagnostic discrepancies resolved by consensus.

Statistical analyses were conducted via SPSS 24.0 and R 4.0.2, with a significance threshold of p < 0.05. Variables with more than 30% missing values were removed, whereas those with fewer missing values were handled via multiple imputation by chained equations (MICE). Statistical tests were selected on the basis of data type and distribution: t-tests for normally distributed continuous variables, Mann‒Whitney U-tests for nonnormally distributed continuous variables, and chi‒square tests for categorical variables. Continuous variables are presented as the mean ± standard deviation or median (interquartile range) as appropriate.

The cutoff values for the NLR, SII, SIRI, and PIV were determined via ROC curve analysis to optimize sensitivity and specificity. These cutoffs were subsequently used to classify the markers into tertiles for further analysis. Multivariable Cox regression models were employed to assess associations with AF recurrence, adjusting for potential confounders, such as sex, age, BMI, and AF type. Comprehensive adjustments incorporated additional clinical and biochemical factors, including left atrial diameter, to ensure robust model performance. The results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs).

Restricted cubic spline (RCS) analysis was used to explore the potential nonlinear relationships between AF recurrence and four systemic inflammatory markers. If the RCS analysis identified a distinct inflection point in the curve, indicating either a U-shaped or L-shaped pattern, the data were segmented into two independent parts on the basis of this inflection point for separate regression analysis. Additionally, the prognostic value of these inflammatory markers for 1-year and 2-year AF recurrence was evaluated via the area under the receiver operating characteristic (ROC) curve (AUC). For all the statistical tests, p < 0.05 (two-sided) was considered statistically significant.

A cohort of 757 patients with atrial fibrillation who underwent cryoablation was analyzed, comprising 448 patients with recurrence and 309 who remained recurrence-free (Figure 1). Table 1 presents the baseline characteristics, which showed significant between-group differences. Patients in the recurrence group presented a greater proportion of females (46% vs. 33%, p < 0.001), a greater prevalence of persistent atrial fibrillation (38% vs. 22%, p < 0.001), and elevated systolic blood pressure (133.7 ± 13.9 mmHg vs. 130.1 ± 13.0 mmHg, p < 0.001). No significant differences were observed between the two groups in terms of age, BMI, AF duration, CHA2DS2-VASc score, or underlying comorbidities (all p > 0.05). Patients without recurrence presented higher prescription rates of rivaroxaban (75% vs. 64%, p < 0.01), statins (41% vs. 25%, p < 0.001), and ARNIs (8% vs. 4%, p = 0.04). Significant variations were also observed in inflammatory markers and indices. Neutrophil levels were elevated in the recurrence group (4.7 ± 4.7 × 10⁹/L vs. 3.3 ± 1.6 × 10⁹/L, p < 0.001), whereas lymphocyte (1.8 ± 0.7 × 10⁹/L vs. 1.6 ± 0.7 × 10⁹/L, p < 0.001) and monocyte levels (0.6 ± 0.2 × 10⁹/L vs. 0.4 ± 0.1 × 10⁹/L, p < 0.001) were reduced. Inflammatory indices, including the NLR (3.7 ± 6.3 vs. 2.1 ± 1.4, p < 0.001), SII (645.5 ± 904.7 vs. 386.9 ± 294.2, p < 0.001), SIRI (2.0 ± 2.6 vs. 0.9 ± 0.7, p < 0.001) and PIV (349.6 ± 411.1 vs. 159.2 ± 137.4, p < 0.001), were significantly greater in the recurrence group. For log-transformed indices, the recurrence group presented significantly greater values for the logNLR (1.0 ± 0.6 vs. 0.6 ± 0.5, p < 0.001), logSII (6.2 ± 0.6, vs. 5.8 ± 0.6, p < 0.001), logSIRI (0.4 ± 0.7 vs. −0.4 ± 0.6, p < 0.001), and logPIV (5.6 ± 0.7 vs. 4.8 ± 0.7, p < 0.001). These comparisons support the prognostic relevance of systemic inflammatory activity for post-cryoablation AF recurrence.

To further elucidate the relationship between systemic inflammation and the AF recurrence rate, we performed subgroup analyses by stratifying participants into three tertiles according to their log-transformed inflammatory indices (Figure 2). We analyzed the rate of AF recurrence in relation to the tertiles of the logNLR, logSII, logSIRI, and logPIV. For the logNLR, the recurrence rates in tertiles 1–3 were 91, 160, and 196 patients, representing 35.83%, 63.49%, and 78.40%, respectively. In the case of logSII, the recurrence rates for tertiles 1–3 were 95, 156, and 196 patients, with corresponding proportions of 37.70%, 61.90%, and 77.78%, respectively. For the logSIRI, the recurrence counts across tertiles 1–3 were 66, 169, and 212 patients, leading to percentages of 26.19%, 66.54%, and 85.14%, respectively. With respect to logPIV, tertiles 1–3 presented recurrence rates of 69, 162, and 217 patients, constituting proportions of 27.38%, 64.29%, and 86.11%, respectively. Taken together, these findings demonstrate a monotonic increase in recurrence across tertiles for all four indices.

Univariable Cox regression showed that higher neutrophil counts (HR = 1.083; 95% CI, 1.069–1.098; p < 0.001), NLR (HR = 1.056; 95% CI, 1.046–1.067; p < 0.001), SIRI (HR = 1.174; 95% CI, 1.148–1.200; p < 0.001), and PIV (per-unit HR = 1.001; 95% CI, 1.001–1.001; p < 0.001) were each associated with increased risk of atrial fibrillation recurrence (Table 2). Log-transformed indices exhibited larger effect sizes: logNLR (HR = 3.019; 95% CI, 2.588–3.522; p < 0.001) and logSII (HR = 3.225; 95% CI, 2.751–3.780; p < 0.001).

Multivariable Cox regression analysis revealed a significant association between higher tertiles of systemic inflammatory markers and an elevated risk of AF recurrence (Table 3). The associations for logNLR, logSII, logSIRI, and logPIV were statistically significant in the unadjusted, partially adjusted, and fully adjusted models. According to the fully adjusted model (Model 3), compared with individuals in the lowest tertile, those in the highest tertiles of the logNLR, logSII, logSIRI, and logPIV had 3.96-fold, 4.13-fold, 5.81-fold, and 6.39-fold higher adjusted hazards of AF recurrence, respectively (Table 3 and Figure 3).

To assess the uniformity of the relationships between the four inflammatory indices (logNLR, logSII, logSIRI, and logPIV) and the recurrence of AF across diverse subgroups, we conducted a detailed subgroup analysis (Figure 4). This analysis included subgroups categorized on the basis of sex, age, BMI, and the nature of AF (paroxysmal or persistent). Relationships between all four inflammatory markers and AF recurrence were consistently observed across the subgroups, with no significant interactions for any variable (all p for interaction > 0.05). These findings indicate that the associations are consistent across demographic and clinical subgroups.

RCS analysis revealed nonlinear associations of the logNLR, logSII, logSIRI, and logPIV with AF recurrence risk after adjustment for confounding factors (Figure 5). The inflection points were logNLR = 1.0, logSII = 6.0, logSIRI = 0.0, and logPIV = 5.0. According to the inflection points, we divided the data into two groups for each indicator (below or above the inflection point) and performed segmented regression analysis on the two groups (Table 4). When the logNLR was <1.0, a one-unit increase was associated with a 2.139-fold greater risk of AF recurrence (HR = 2.139, 95% CI: 1.372–3.336; p < 0.001). When the logNLR was greater than 1.0, a one-unit increase was associated with a 3.161-fold greater risk of AF recurrence (HR = 3.161, 95% CI: 2.362–4.229, p < 0.001). When the LogSII was lower than 6.0, a one-unit increase was associated with a 2.476-fold greater risk of AF recurrence (HR = 2.476, 95% CI, 1.470–4.172; p < 0.001). When the LogSII was greater than 6.0, a one-unit increase was associated with a 3.961-fold greater risk of AF recurrence (HR = 3.961, 95% CI: 3.138–4.999; p < 0.001). When the LogSIRI was lower than 0.0, a one-unit increase was associated with a 3.349-fold greater risk of AF recurrence (HR = 3.349, 95% CI, 1.693–6.627; p < 0.001). When the LogSIRI was equal to or greater than 0.0, a one-unit increase was associated with a 2.903-fold greater risk of AF recurrence (HR = 2.903, 95% CI: 2.320–3.633, p < 0.001). When logPIV was lower than 5.0, a one-unit increase was associated with a 4.366-fold greater risk of AF recurrence (HR = 4.366, 95% CI: 1.884–10.118, p < 0.001). When logPIV was ≥5.0, a one-unit increase was associated with a 3.487-fold greater risk of AF recurrence (HR = 3.487, 95% CI: 2.846–4.271; p < 0.001).

The time-dependent receiver operating characteristic (ROC) curves for evaluating the predictive efficacy of the logNLR, logSII, logSIRI, and logPIV for AF recurrence at 12 and 24 months are presented in Figures 6A,B. Tables 5, 6 show the corresponding areas under the curve (AUC) values and prognostic cutoff points. At the 12-month assessment, the AUC values were as follows: logNLR: 0.715 (95% CI: 0.669–0.761, p < 0.001), logSII: 0.740 (95% CI: 0.696–0.784, p < 0.001), logSIRI: 0.742 (95% CI: 0.699–0.784, p < 0.001), and logPIV: 0.764 (95% CI: 0.723–0.805, p < 0.001). The optimal prognostic thresholds for these markers were 0.644, 5.792, −0.005, and 5.132, respectively. The logPIV exhibited the highest sensitivity (79.2%) and specificity (68.3%), suggesting its superior predictive capacity during the initial year of follow-up. At the 24-month evaluation, the AUC values for the four indicators were as follows: logNLR: 0.719 (95% CI: 0.674–0.764, p < 0.001), logSII: 0.732 (95% CI: 0.688–0.776, p < 0.001), logSIRI: 0.724 (95% CI: 0.679–0.769, p < 0.001), and logPIV: 0.741 (95% CI: 0.698–0.785, p < 0.001). The cutoff values for these biomarkers in predicting AF recurrence remained consistent with those observed at 12 months. Among the four biomarkers, logPIV maintained the highest sensitivity (79.2%) and specificity (68.3%), as indicated in Table 5. In summary, the analysis revealed that the logPIV consistently demonstrated superior predictive capability for AF recurrence, whereas the logSII, logNLR, and logSIRI exhibited moderate prognostic efficacy throughout both study phases.