Predictive value of preoperative hematologic inflammation indices (NLR, PLR, SII) and clinical risk factors in predicting postoperative pulmonary complications after elective isolated on-pump coronary artery bypass grafting: a retrospective cohort study of 1034 patients

Abstract

Background

Pulmonary complications after coronary artery bypass surgery continue to be a significant problem, affecting 5–20% of patients, prolonging hospital stays and increasing costs. In this study, we investigated whether simple blood tests that measure inflammation, such as the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII), could help identify patients at risk before surgery.

Methods

The records of 1034 patients who underwent elective coronary artery bypass grafting with heart-lung machine support between 2023 and 2024 were retrospectively reviewed. NLR, PLR, and SII values were calculated from routine blood tests performed the day before surgery. Patients who developed pulmonary complications were defined according to the European Perioperative Clinical Outcome (EPCO) definitions.

Results

114 patients (11%) developed PPC. Preoperative inflammatory markers were significantly higher in patients with complications (NLR: 3.55 ± 1.07 vs. 1.79 ± 0.54, PLR: 156 ± 47 vs. 120 ± 36, SII: 825 ± 248 vs. 548 ± 164; all p < 0.001). ROC analysis demonstrated excellent discrimination for NLR (AUC 0.939), good for SII (AUC 0.818), and fair-to-good for PLR (AUC 0.724). In multivariate analysis, NLR was by far the strongest independent predictor of PPC, together with COPD, diabetes, active smoking, CPB and ACC durations. PLR and SII also reached statistical significance, but with effect sizes very close to unity (adjusted OR 1.02 and 1.01 respectively), indicating that they offered minimal additional discriminatory value once NLR was taken into account.

Conclusions

In this single-center retrospective cohort, preoperative NLR, PLR and SII were associated with PPC, with NLR accounting for most of the predictive signal. Because they are measured at a single preoperative timepoint, these indices reflect baseline inflammatory tone and cannot capture the acute, surgery-induced inflammatory response that drives postoperative pulmonary complications. In view of the single-center retrospective design, the absence of external validation, and the lack of comparison with validated risk scores such as EuroSCORE II or the STS score, these indices are not yet suitable to guide clinical decision-making on their own. Given their simplicity and ready availability, however, they appear to be promising candidate markers that merit further evaluation in prospective, multicenter studies also incorporating specific inflammatory mediators measured dynamically throughout the perioperative period.

Background

Coronary artery bypass grafting (CABG) remains one of the most effective treatment options for patients with severe coronary artery disease. However, postoperative pulmonary complications (PPC) occur in 5–20% of patients and constitute a significant clinical problem by prolonging hospital stay, increasing morbidity, and increasing healthcare costs [1, 2]. These complications include pneumonia, atelectasis, hypoxemic respiratory failure, prolonged mechanical ventilation, reintubation, and serious pleural complications [3, 4].

On-pump CABG surgery using cardiopulmonary bypass (CPB) triggers an intense systemic inflammatory response. During CPB, neutrophil activation, cytokine release, endothelial dysfunction, and increased pulmonary vascular permeability occur, leading to lung damage [5–7]. Because they are central to the circulation, the lungs are among the organs most affected by this inflammatory storm.

In recent years, hematologic inflammation indices derived from routine complete blood counts [neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and systemic immune-inflammation index (SII)] have emerged as accessible markers of systemic inflammation [8–10]. These indices can be obtained through simple calculations from the complete blood count routinely performed on every patient undergoing a surgical procedure. Previous studies have shown that these markers can predict adverse outcomes after cardiac surgery [11–13].

Methods

Study design and patient population

This retrospective cohort study included adult patients who underwent elective isolated on-pump CABG surgery at our institution between January 2023 and December 2024. The study was approved by the institutional ethics committee and conducted in accordance with the principles of the Declaration of Helsinki. Due to the retrospective design and use of anonymized data, individual patient consent was not obtained.

Patient selection proceeded through a systematic screening process. Initially, 1355 consecutive patients who underwent CABG during the study period were identified through the hospital’s cardiac surgery database. To ensure a homogeneous study population and minimize confounding variables, strict inclusion and exclusion criteria were applied.

Inclusion criteria

• Age ≥ 18 years.

• Elective (non-emergency) isolated on-pump CABG surgery.

• Complete preoperative laboratory data including full blood count within 24 h before surgery.

• Complete perioperative and postoperative clinical documentation.

• Minimum hospital stay of 48 h postoperatively (to allow adequate observation for PPC development).

Exclusion criteria

• Emergency or urgent CABG procedures (n = 127) - excluded due to varying preoperative inflammatory states and insufficient time for standardized preoperative optimization.

• Concomitant cardiac procedures (valve surgery, aortic surgery, atrial fibrillation ablation, etc.) (n = 145) excluded to maintain procedural homogeneity.

• Off-pump CABG (n = 3) - excluded as CPB exposure is central to the inflammatory hypothesis.

• Active infection or sepsis at the time of surgery (n = 2) - excluded due to pre-existing inflammatory state.

• Active malignancy or receipt of chemotherapy within 3 months (n = 6) - excluded due to altered inflammatory and immune profiles.

• Chronic immunosuppressive therapy (corticosteroids, biologics, disease-modifying antirheumatic drugs) (n = 5) - excluded due to modulated inflammatory response.

• Hematologic disorders (leukemia, lymphoma, myelodysplastic syndrome) (n = 6) - excluded due to unreliable blood count indices.

• Incomplete preoperative laboratory data or missing outcome data (n = 27).

• After applying these criteria, 1034 patients constituted the final study cohort for analysis.

Definition of postoperative pulmonary complications

Postoperative pulmonary complications (PPCs) were rigorously defined according to the standardized criteria established by the European Perioperative Clinical Outcome (EPCO) definitions [3]. These internationally recognized criteria ensure consistency and comparability with other perioperative outcome studies. A patient was considered to have developed a PPC if they met any of the following criteria during the postoperative hospital stay.

All patients were systematically screened for PPCs through daily clinical assessment by the cardiac surgery team, routine chest X-rays (performed on postoperative days 1, 3, and 5, or more frequently if clinically indicated), arterial blood gas analyses when clinically warranted, and review of all mechanical ventilation parameters. PPC determination was made by two independent cardiac surgeons blinded to the preoperative inflammatory marker values. In cases of disagreement, a third reviewer adjudicated, and consensus was achieved through discussion.

Data collection and variables

Data were extracted from electronic medical records, anesthesia records, perfusion logs, intensive care unit charts, and laboratory information systems by trained research personnel using a standardized data collection form. The following categories of data were systematically collected:

Demographic and baseline characteristics

Age, sex, comorbidities including diabetes mellitus (defined as documented diagnosis with hemoglobin A1c ≥ 6.5% or use of antidiabetic medications), chronic obstructive pulmonary disease (COPD; defined by pulmonary function testing showing FEV₁/FVC < 0.7 or documented clinical diagnosis with inhaler use), active smoking status (defined as current smoking or cessation within 3 months of surgery) and preoperative laboratory values including complete blood count parameters were systematically recorded.

Calculation of inflammatory markers

The major advantage of these markers is their availability in routine testing. A complete blood count is performed on every patient within 24 h before surgery. Three ratios were calculated from this single test:

Neutrophil-to-Lymphocyte Ratio (NLR):

NLR = Absolute neutrophil count ÷ Absolute lymphocyte count.

Platelet-to-Lymphocyte Ratio (PLR):

PLR = Absolute platelet count ÷ Absolute lymphocyte count.

Systemic Immune-Inflammation Index (SII):

SII = (Absolute platelet count × Absolute neutrophil count) ÷ Absolute lymphocyte count.

These indices require no additional blood sampling, laboratory costs, or processing time beyond the routine preoperative CBC that is standard practice for all surgical patients. This makes them highly accessible and practical for clinical implementation.

Statistical analysis

Sample size was determined by the retrospective cohort design based on available patients meeting inclusion criteria during the study period. The adequacy of the sample size was supported by the number of outcome events and the narrow confidence intervals observed in the multivariable model estimates.

Statistical analyses were performed using SPSS 21.0 (IBM Corp., NY, USA). Continuous variables are presented as mean ± standard deviation, and categorical variables as numbers and percentages. Statistical significance was set at p < 0.05.

Comparisons between PPC(+) and PPC(-) groups were performed using independent samples t-test for continuous variables and Chi-square or Fisher’s exact test for categorical variables.

ROC curve analysis assessed the discriminatory ability of inflammatory indices. AUC values were calculated with 95% confidence intervals. Optimal cut-off values were determined using Youden’s index. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for each cut-off.

Multivariate logistic regression identified independent predictors of PPC. Variables with p < 0.10 in univariate analysis were included in the model. Multicollinearity was assessed using variance inflation factor. Model discrimination was assessed using the C-statistic and calibration using the Hosmer-Lemeshow test. Results are presented as adjusted odds ratios (OR) with 95% confidence intervals.

Results

Baseline characteristics

Of the 1034 patients included in the study, 114 (11.0%) developed PPC. This rate is consistent with the 5–20% range reported in the literature [1, 2], supporting the appropriate application of standardized PPC criteria. Baseline characteristics, comorbidities, preoperative laboratory values, and surgical parameters are summarized in Table 1.

Table 1.

Baseline characteristics and comparison between PPC(+) and PPC(-) groups

Characteristic	PPC(+) (n = 114)	PPC(-) (n = 920)	p-value
Age (years), mean ± SD	65.2 ± 8.5	63.5 ± 7.9	0.042
Male gender, n (%)	88 (77.2)	690 (75.0)	0.652
Female gender, n (%)	26 (22.8)	230 (25.0)	0.652
Comorbidities
Diabetes mellitus, n (%)	90 (78.9)	242 (26.3)	< 0.001
COPD, n (%)	62 (54.4)	26 (2.8)	< 0.001
Active smoking, n (%)	65 (57.0)	57 (6.2)	< 0.001
Preoperative laboratory values
WBC (×10³/µL), mean ± SD	8.9 ± 2.1	7.2 ± 1.6	< 0.001
Neutrophils (×10³/µL), mean ± SD	6.2 ± 1.8	4.1 ± 1.2	< 0.001
Lymphocytes (×10³/µL), mean ± SD	1.8 ± 0.5	2.3 ± 0.6	< 0.001
Platelets (×10³/µL), mean ± SD	242 ± 58	238 ± 52	0.488
Inflammation indices
NLR, mean ± SD	3.55 ± 1.07	1.79 ± 0.54	< 0.001
PLR, mean ± SD	156 ± 47	120 ± 36	< 0.001
SII, mean ± SD	825 ± 248	548 ± 164	< 0.001
Surgical characteristics
CPB time (min), mean ± SD	116 ± 25	102 ± 22	< 0.001
ACC time (min), mean ± SD	68 ± 18	62 ± 15	< 0.001

PPC, postoperative pulmonary complications; SD, standard deviation; COPD, chronic obstructive pulmonary disease; WBC, white blood cell count; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; CPB, cardiopulmonary bypass; ACC, aortic cross-clamp

Significant differences were observed in inflammatory markers. The mean preoperative NLR was 3.55 ± 1.07 in patients who developed PPC, compared to 1.79 ± 0.54 in those without complications (p < 0.001). Similar differences were observed for PLR (156 ± 47 vs. 120 ± 36, p < 0.001) and SII (825 ± 248 vs. 548 ± 164, p < 0.001).

The prevalence of COPD was markedly higher in patients who developed PPC (54.4% vs. 2.8%, p < 0.001). Active smoking was present in 57.0% of the PPC group vs. 6.2% without PPC (p < 0.001). Diabetes mellitus prevalence was 78.9% vs. 26.3% (p < 0.001).

Both CPB duration (116 ± 25 vs. 102 ± 22 min, p < 0.001) and ACC duration (68 ± 18 vs. 62 ± 15 min, p < 0.001) were significantly longer in patients who developed PPC.

ROC curve analysis

ROC curve analysis demonstrated the discriminatory ability of inflammatory indices in predicting PPC (Table 2; Fig. 1). NLR showed excellent performance (AUC 0.939, 95% CI: 0.908–0.970, p < 0.001). SII demonstrated good discrimination (AUC 0.818, 95% CI: 0.770–0.867, p < 0.001), while PLR showed fair-to-good performance (AUC 0.724, 95% CI: 0.669–0.778, p < 0.001). Optimal cut-off values determined by Youden’s index were:

Table 2.

ROC curve analysis for inflammation indices in predicting postoperative pulmonary complications

Variable	AUC (95% CI)	Cut-off	Sensitivity (%)	Specificit y (%)	PPV (%)	NPV (%)	LR+	LR−	Youden Index
NLR	0.939 (0.908–0.970)	2.48	89.5	90.1	52.8	98.6	9.04	0.12	0.796
SII	0.818 (0.770–0.867)	741.59	64.0	87.4	38.6	95.1	5.08	0.41	0.514
PLR	0.724 (0.669–0.778)	159.52	50.9	85.9	30.9	93.4	3.61	0.57	0.368

AUC, area under the ROC curve; CI, confidence interval; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR-, negative likelihood ratio

Fig. 1.

ROC curves for inflammatory indices in predicting postoperative pulmonary complications. ROC, receiver operating characteristic; AUC, area under the curve; CI, confidence interval; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index

• NLR ≥ 2.48: Sensitivity 89.5%, Specificity 90.1%, PPV 52.8%, NPV 98.6%.

• SII ≥ 741.59: Sensitivity 64.0%, Specificity 87.4%, PPV 38.6%, NPV 95.1%.

• PLR ≥ 159.52: Sensitivity 50.9%, Specificity 85.9%, PPV 30.9%, NPV 93.4%.

The high specificity of NLR (90.1%) indicates low false-positive rates, and the high negative predictive values (> 93% for all indices) suggest that low values are reassuring. It should be noted that positive and negative predictive values are influenced by the prevalence of PPC in the study population and may vary across different clinical settings.

Multivariate analysis

In multivariate logistic regression analysis, COPD, diabetes mellitus, active smoking, NLR, PLR, SII, CPB duration, and ACC duration all remained statistically significant independent predictors of PPC (Table 3; Fig. 2). NLR emerged as the strongest predictor (OR 65.479, 95% CI: 16.747–256.020, p < 0.001), followed by active smoking (OR 60.529, 95% CI: 11.028-332.222, p < 0.001), diabetes mellitus (OR 18.226, 95% CI: 4.505–73.727, p < 0.001), and COPD (OR 13.395, 95% CI: 2.639–67.909, p = 0.002). Both CPB duration (OR 1.031 per minute, 95% CI: 1.004–1.059, p = 0.024) and ACC duration (OR 1.042 per minute, 95% CI: 1.006–1.080, p = 0.023) demonstrated independent effects. PLR (OR 1.020, 95% CI: 1.002–1.037, p = 0.025) and SII (OR 1.010, 95% CI: 1.006–1.014, p < 0.001) also retained statistical significance. For clinical interpretability, OR values for PLR and SII are expressed per 10-unit and 100-unit increases, respectively, reflecting their typical variability ranges.

Table 3.

Univariate and multivariate logistic regression analysis for postoperative pulmonary complications

Variable	Univariate Analysis OR (95% CI)	P value	Multivariate Analysis OR (95% CI)	P value
Diabetes Mellitus	10.506 (6.543–16.870)	< 0.001	18.226 (4.505–73.727)	< 0.001
COPD	40.997 (23.974–70.108)	< 0.001	13.395 (2.639–67.909)	0.002
Active Smoking	20.084 (12.710-31.737)	< 0.001	60.529 (11.028-332.222)	< 0.001
NLR	38.182 (20.601–70.766)	< 0.001	65.479 (16.747–256.020)	< 0.001
SII	1.008 (1.006–1.009)	< 0.001	1.010 (1.006–1.014)	< 0.001
PLR	1.025 (1.020–1.031)	< 0.001	1.020 (1.002–1.037)	0.025
CPB Time (min)	1.028 (1.019–1.038)	< 0.001	1.031 (1.004–1.059)	0.024
ACC Time (min)	1.026 (1.013–1.039)	< 0.001	1.042 (1.006–1.080)	0.023

OR, odds ratio; CI, confidence interval; COPD, chronic obstructive pulmonary disease; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; CPB, cardiopulmonary bypass; ACC, aortic cross-clamp

Odds ratios for PLR and SII are expressed per 10-unit and 100-unit increases, respectively

Bold values in Table indicate statistically significant results (P < 0.05).

Fig. 2.

Forest plot of independent predictors of postoperative pulmonary complications. OR, odds ratio; CI, confidence interval; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; COPD, chronic obstructive pulmonary disease; CPB, cardiopulmonary bypass; ACC, aortic cross-clamp

Multicollinearity was assessed using the variance inflation factor (VIF) and a Spearman correlation matrix of the model variables (Figs. 3 and 4). Although VIF values below 10 are generally considered acceptable, cardiopulmonary bypass (CPB) time and aortic cross-clamp (ACC) time demonstrated higher VIF values (15.54 and 12.84, respectively), reflecting their expected strong physiological correlation. Given their clinical relevance, both variables were retained in the primary multivariable model. Sensitivity analyses excluding either CPB time or ACC time yielded consistent results, indicating model robustness. It should be emphasized that these multicollinearity metrics describe only the statistical independence of the predictors and were not intended to quantify whether PLR and SII improve risk prediction relative to NLR, which we did not formally evaluate.

Fig. 3.

Variance inflation factor (VIF) Analysis.Variance inflation factor (VIF) values for each predictor included in the multivariable model. The horizontal axis lists the model variables and the vertical axis shows the corresponding VIF value. The two vertical (or horizontal, depending on bar orientation) dashed reference lines mark the conventional interpretive thresholds of VIF = 5 (moderate collinearity) and VIF = 10 (high collinearity); values above 10 indicate problematic multicollinearity. CPB time and ACC time exceed the VIF = 10 line, consistent with their expected physiological correlation. Abbreviations: NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; COPD, chronic obstructive pulmonary disease; CPB, cardiopulmonary bypass; ACC, aortic cross-clamp

Fig. 4.

Correlation matrix heatmaps for multicollinearity assessment. Correlation heatmap of the key clinical and inflammatory variables included in the analysis. Both the horizontal and vertical axes list the same set of model variables, so that each cell represents the pairwise correlation between the variable in its row and the variable in its column. Cell values are pairwise Spearman correlation coefficients, ranging from − 1 (perfect negative correlation) through 0 (no correlation) to + 1 (perfect positive correlation); the color scale encodes the magnitude and direction of correlation as indicated in the accompanying color bar. As an interpretive guide, absolute coefficients below 0.3 denote weak correlation and values above 0.7 denote strong correlation. NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; COPD, chronic obstructive pulmonary disease; CPB, cardiopulmonary bypass; ACC, aortic cross-clamp

Discussion

Main findings

The main findings of this 1034-patient study are: (1) the preoperative neutrophil-to-lymphocyte ratio, in particular, was strongly associated with PPC and carried essentially all of the predictive signal among the three indices; (2) although PLR and SII reached statistical significance in the multivariable model, their adjusted odds ratios were very close to unity (approximately 1.02 and 1.01), suggesting that they contributed little additional prognostic information beyond what NLR already captured; (3) diabetes was strongly associated with PPC, with an approximately 18-fold increase in adjusted odds; and (4) both CPB duration and ACC duration were independent statistical predictors of PPC.

It is important to keep in mind that these are all preoperative associations; given the retrospective design, they cannot tell us whether these indices causally contribute to lung injury or simply reflect the patients’ underlying risk profile.

The excellent performance of NLR (AUC 0.939) demonstrates the potential value of this simple index. The fair-to-good performance of PLR (AUC 0.724), good performance of SII (AUC 0.818), and excellent performance of NLR (AUC 0.939) can be attributed to biological underpinnings. Neutrophils disrupt alveolar-capillary integrity by releasing proteases, reactive oxygen species, and pro-inflammatory mediators, playing a central role in acute lung injury [14]. Lymphopenia, in turn, reflects impaired immune regulation, resulting in a weakened ability to control and resolve inflammation [14, 15].

SII and PLR add a further dimension by incorporating platelets, which contribute to CPB-associated inflammation and microthrombotic lung injury [16]. However, the magnitude of the independent effects of PLR and SII must be interpreted with caution. In the multivariable model, their adjusted odds ratios were very close to unity (PLR OR 1.020 per 10-unit increase; SII OR 1.010 per 100-unit increase), whereas NLR carried by far the largest effect (OR 65.5). Effect sizes of this magnitude, particularly in a cohort of 1034 patients, are likely to reflect the statistical power conferred by a large sample size rather than clinically meaningful incremental discrimination. We did not perform net reclassification improvement (NRI) or integrated discrimination improvement (IDI) analyses, and we therefore cannot claim that adding PLR and SII to NLR improves risk classification in any clinically useful way. The most defensible interpretation of our data is that NLR carries essentially all of the predictive signal among the three indices and that PLR and SII offer negligible incremental benefit once NLR is included in the model. The multicollinearity metrics we report (VIF and correlation coefficients) describe only the statistical independence of these variables; they do not establish that PLR and SII provide added clinical value, which remains unproven and would require dedicated reclassification analyses in an independent cohort.

The 13-fold increased risk in COPD is striking. COPD patients are characterized by chronic neutrophilic airway inflammation, impaired mucociliary clearance, and diminished pulmonary reserve [17, 18]. Smoking is also well known to be strongly associated with perioperative complications [19, 20]. When these compromised lungs are exposed to the additive inflammatory effects of CPB, the system becomes overloaded. The presence of COPD requires careful preoperative evaluation and perioperative optimization.

The prevalence of diabetes mellitus in the PPC group at 78.9% and its independent predictor status with an odds ratio of 18.2 warrants emphasis. The impact of diabetes on PPC occurs through multiple mechanisms. Chronic hyperglycemia impairs neutrophil chemotaxis, phagocytosis, and bactericidal activity [21, 22], increasing pneumonia risk. Diabetic patients exist in a state of chronic low-grade inflammation [23, 24], consistent with higher preoperative NLR values. The combination of diabetes and high NLR may synergistically increase PPC risk.

In our study, both CPB duration (OR 1.03 per minute, p = 0.024) and ACC duration (OR 1.04 per minute, p = 0.023) were independent risk factors. These findings highlight that both total extracorporeal circulation time and specific myocardial ischemia time contribute independently to pulmonary injury. Every minute of CPB maintains complement activation, cytokine release (IL-6, IL-8, TNF-α), and neutrophil activation [6, 7], leading to cumulative inflammatory burden. Prolonged ACC duration intensifies myocardial ischemia-reperfusion injury, contributing additional systemic inflammatory stress. These findings emphasize the importance of surgical strategies that minimize both CPB and ACC duration.

Clinical applications

The applications discussed in this section are hypothetical and are intended to outline how a marker of this type might eventually be used if it were prospectively validated; they do not describe a use that our data currently support. The appeal of NLR lies in its derivation from a complete blood count that is already obtained in every patient, so that its calculation adds no separate test. However, accessibility is not the same as validated clinical utility. Because our study is single-center and retrospective, lacks external validation, and does not compare these indices against established risk scores, none of the indices examined here is presently suitable to guide individual patient management.

If such a tool were validated, one could hypothesize that, for patients with low inflammatory indices and no major risk factors, low values might prove reassuring, whereas for patients with a high NLR and additional risk factors such as COPD or active smoking, intensified perioperative measures (preoperative respiratory physiotherapy, intraoperative lung-protective ventilation, close postoperative monitoring, and early respiratory support) might be considered.

If a relationship between the aforementioned inflammatory indices and PPCs is proven, then the necessary preventive measures should also be investigated. Statins have pleiotropic anti-inflammatory properties independent of their lipid-lowering effect; randomized studies have shown that simvastatin administration before CPB increases neutrophil apoptosis and significantly reduces pro-inflammatory cytokine levels [25]. Regarding corticosteroids, a single dose of methylprednisolone before CPB is known to suppress pro-inflammatory cytokines [26]; however, the effects of both pharmacological approaches on clinical outcomes in cardiac surgery have been found to be inconsistent in large randomized studies, and their specific effects on PPC have not yet been investigated [27]. Current STS guidelines recommend standard surgical antimicrobial prophylaxis for CABG (administration of cefazolin 60 min prior to incision and continued for 24–48 h) [28], and there is no evidence for additional antimicrobial strategies based on preoperative inflammatory indices.

We stress that these are conjectures for future testing rather than evidence-based recommendations arising from the present study, and that elevated indices should not by themselves alter clinical management at this time.

Surgical planning to minimize CPB and ACC times is critical, especially in high-risk patients. Optimizing preoperative glycemic control in diabetic patients and considering preoperative optimization strategies in patients with elevated inflammatory indices may be beneficial.

Comparison with the literature

Our findings are broadly consistent with previous reports linking inflammatory markers to cardiac surgery outcomes [11–13], although our study differs in several respects. First, our sample size (n = 1034) is among the largest to focus specifically on PPCs after CABG. Second, our use of the standardized EPCO definitions facilitates comparison with future studies. Third, although the full model that combined clinical factors with the inflammatory indices showed high apparent discrimination (C-statistic 0.979), this value was obtained in the same single-center sample in which the model was developed, without external validation; such an in-sample statistic is prone to optimism and does not, by itself, demonstrate clinically useful risk stratification. Fourth, while all three indices reached statistical significance in the multivariable model, the dominant contribution came from NLR; the very small effect sizes of PLR and SII indicate that their contribution is marginal, which is consistent with, rather than contrary to, the substantial overlap expected among blood-count-derived ratios.

These findings are directionally consistent with prior reports but differ notably in magnitude. In a cohort of 1199 elderly hip-fracture patients, preoperative NLR predicted postoperative pneumonia with an AUC of only 0.648 and again outperformed PLR and SII [29], whereas in cardiac-surgical cohorts postoperative NLR has predicted off-pump CABG–associated acute kidney injury and mid-term mortality with AUCs of approximately 0.78 [30] and 0.72 [31], respectively. The excellent discrimination we observed for preoperative NLR (AUC 0.939) therefore lies at, or above, the upper end of previously reported values; this, together with the single-center design and absence of external validation, suggests that our estimate may be optimistic and requires confirmation in independent cohorts.

A recent study used comprehensive biomarker panels including IL-6, PCT, and CRP, measured at multiple time points postoperatively [32]. These markers provide mechanistic insight into the acute inflammatory surge that directly precipitates PPCs.

In contrast, our study relies exclusively on preoperative NLR, PLR, and SII—indices that reflect baseline inflammatory tone but cannot capture the dynamic, surgery-induced inflammatory response. We did not measure IL-6, PCT, CRP, or any cytokines perioperatively, nor did we perform serial measurements. This represents a significant conceptual and methodological gap.

Our findings should be interpreted within this bounded context: preoperative indices may serve as low-cost initial screening tools but do NOT replace detailed mechanistic investigation using specific inflammatory mediators measured dynamically throughout the perioperative period.

Limitations

First, this was a single-center retrospective study, which limits generalizability to institutions with different patient populations, surgical practices, and perioperative protocols. Relatedly, the optimal cut-off values we report (NLR ≥ 2.48, PLR ≥ 159.52, SII ≥ 741.59) were empirically derived from this single dataset using the Youden index and were not validated in an independent cohort; they are therefore prone to optimistic bias, and their transportability across different populations, surgical techniques, and laboratory platforms remain unknown. External validation in independent, preferably multicenter cohorts is required before any of these findings or thresholds can be applied clinically.

Second, and most fundamentally, inflammation was assessed at a single preoperative timepoint using simple hematologic indices. This static approach reflects only baseline inflammatory tone and cannot capture the dynamic, surgery-induced inflammatory surge (triggered largely by cardiopulmonary bypass) that directly precipitates pulmonary injury. We did not measure specific mediators such as interleukin-6, procalcitonin, or C-reactive protein, nor did we obtain serial perioperative samples, so the relationship between these preoperative indices and the actual postoperative inflammatory cascade remains unexplored.

Third, we did not assess whether these indices add predictive value beyond established, validated risk scores such as EuroSCORE II or the STS risk calculator, an essential step in determining their practical utility.

Fourth, perioperative myocardial injury biomarkers were unavailable: cardiac troponin T was not part of our standardized retrospective data collection. As troponin T is an established prognostic marker in cardiac surgery [33], its absence prevented us from examining how it relates to PPC or whether it provides information beyond the preoperative indices. Combined dynamic assessment of hematologic inflammation indices and myocardial injury biomarkers is an important aspect for future prospective work.

Finally, the natural correlation among neutrophil-, lymphocyte-, and platelet-derived measures means that statistical independence may not reflect true biological independence, and alternative approaches (e.g., principal component analysis or machine learning) may offer further insight. More broadly, the observational design precludes causal inference: we cannot determine whether these indices contribute directly to pulmonary injury or merely reflect underlying risk.

Conclusions

In this single-center retrospective cohort, preoperative NLR was associated with postoperative pulmonary complications after CABG and showed high apparent discrimination (AUC 0.939), whereas PLR and SII, despite reaching statistical significance, had effect sizes close to unity and conferred no meaningful prognostic advantage over NLR alone. Critically, all three indices were measured at a single preoperative timepoint and reflect only baseline inflammatory tone; they cannot capture the acute, surgery-induced inflammatory response that directly drives these complications and should not be interpreted as mechanistic markers of it.

These findings must be weighed against important limitations: a single-center retrospective design without external validation, no comparison against established risk scores (EuroSCORE II, STS), and no measurement of specific inflammatory mediators (interleukin-6, procalcitonin, C-reactive protein) or serial perioperative sampling. Before any clinical use can be considered, the cut-off values require external validation in diverse, multicenter populations, and prospective studies must incorporate dynamic perioperative measurements and, most importantly, demonstrate that biomarker-guided intervention improves outcomes. As they stand, these preoperative indices regarded as an investigative hypothesis rather than a usable clinical tool.

Author contributions

ÖV: Conceptualization, Data collection, Formal analysis, Writing - original draft, Writing - review & editing. EET: Supervision, Methodology, Validation, Writing - review & editing. All authors read and approved the final manuscript.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The datasets generated and analyzed during the current study are not publicly available due to institutional data privacy policies but are available from the corresponding author on reasonable request and with appropriate ethics committee approval.

Declarations

Ethics approval and consent to participate

This study was approved by the Republic of Turkey Mersin University Rectorate Clinical Research Ethics Committee Ethics Committee (Approval No: 2025/0137 Date: 05.02.2025). The study was conducted in accordance with the principles of the Declaration of Helsinki. Due to the retrospective design and use of anonymized data, individual patient consent was waived by the ethics committee.

Consent for publication

Not applicable. This manuscript does not contain any individual person’s data in any form (including individual details, images, or videos).

Competing interests

The authors declare no competing interests.