EXPERIMENTAL AND CLINICAL TRANSPLANTATION

Objectives: The neutrophil-to-lymphocyte ratio is an inexpensive and accessible inflammatory biomarker that reflects the balance between innate and adaptive immunity. This study evaluated whether perioperative or early posttransplant neutrophil-to-lymphocyte ratio predicts baseline lung allograft dysfunction or acute cellular rejection within the first year after lung transplant.
Materials and Methods: In this single-center retros-pective cohort study, we analyzed 80 adult bilateral lung transplant recipients (transplanted in 2021-2023). Patients had peripheral blood neutrophil-to-lymphocyte ratio measured pretransplant and at 3, 6, 9, and 12 months posttransplant; bronchoalveolar lavage neutrophil-to-lymphocyte ratio was calculated from differential counts when available. We assessed associations between neutrophil-to-lymphocyte ratio, baseline lung allograft dysfunction, and acute cellular rejection within the first year with the Mann-Whitney U test, χ² test, or Fisher exact test. We performed receiver operating characteristic analysis with Youden index-based cutoffs and multivariable logistic regression adjusted for baseline dysfunction, diagnosis group, and sex.
Results: Peripheral blood neutrophil-to-lymphocyte ratio at 3, 6, 9, and 12 months posttransplant was consistently higher in patients with baseline dysfunction than in those without, but differences were not significant. Bronchoalveolar lavage neutrophil-to-lymphocyte ratio showed no group differences and substantial variability. Receiver operating characteristic analyses showed no discriminatory ability of neutrophil-to-lymphocyte ratio to predict acute rejection (area under the curve ≈ 0.5 across all time points), and seemingly high sensitivities and specificities at certain cutoffs were artifacts accompanied by negligible Youden indices. In multi-variable models, neither blood nor bronchiolar lavage neutrophil-to-lymphocyte ratio was independently associated with rejection (all P > .05). Median follow-up was 12 months.
Conclusions: Neutrophil-to-lymphocyte ratio did not discriminate baseline lung allograft dysfunction or acute rejection, with regression analyses showing no significance. Neutrophil-to-lymphocyte ratio lacks clinical utility as a standalone marker and, if used at all, should be embedded within multimodal surveillance frameworks.

Key words : Biomarker validation, Bronchoalveolar lavage, Immune monitoring, Risk stratification, Systemic inflammation

Lung transplantation markedly improves survival and quality of life in selected patients with end stage respiratory disease, yet early allograft dysfunction events remain common and prognostically relevant. Both acute cellular rejection (ACR) in the first year posttransplant and the failure to achieve normal spirometric function, termed baseline lung allograft dysfunction (BLAD), are associated with subsequent chronic lung allograft dysfunction and mortality.
Baseline lung allograft dysfunction is an early phenotype characterized by persistently reduced lung function during the first year posttransplant.¹ When recently established spirometric criteria were applied in our center, nearly 60% of bilateral lung transplant recipients met BLAD criteria, a finding associated with longer hospital and intensive care unit (ICU) stays, greater perioperative burden, and impaired functional recovery.¹ Other data have suggested that BLAD may also be linked to decreased long-term survival, although its pathophysiology and timing of onset remain incompletely understood.²
Acute cellular rejection is a common immune-mediated injury classified histopathologically accor-ding to International Society for Heart and Lung Transplantation (ISHLT) criteria.³ Even minimal ACR (grade A1) may be clinically silent yet still has been independently associated with an increased risk of chronic lung allograft dysfunction.⁴ In patients with BLAD, even low-grade ACR may have clinical consequences, potentially warranting intensified sur-veillance and modification of immunosuppression.
Transbronchial lung biopsy is the diagnostic reference for ACR, but its sensitivity is imperfect and the procedure is invasive, with routine surveillance schedules varying widely across centers.^3,5 Therefore, there is a clinical need for noninvasive, inexpensive biomarkers that help risk stratify recipients and guide the intensity of surveillance.
Among candidate indices, the neutrophil to lymp-hocyte ratio (NLR) is attractive because it captures, in a single measure, opposing poles of the host response (Figure 1). Notwithstanding its biological appeal, NLR is inherently nonspecific and influenced by infection, perioperative stress, and immunosup-pressive therapy, leading to high interindividual variability and limiting its reliability as a standalone biomarker early after transplant. Neutrophils, core effectors of the innate immune system, are rapidly mobilized by ischemia-reperfusion injury, tissue damage, and infection and can contribute to early graft injury through release of proteases, reactive oxygen species, and proinflammatory mediators.^6-8 Lymphocytes are the principal effectors of the adaptive response; T cells drive ACR, and B cells/plasma cells contribute to antibody mediated mechanisms.⁹ Standard immunosuppression prefe-rentially reduces circulating lymphocytes in the first year posttransplant.¹⁰ An elevated NLR may thus reflect neutrophilia from sterile inflammation or infec-tion combined with therapy related lymphopenia, that is, the net balance between immune activation and suppression. Conceptually, NLR could function as a low cost surrogate of the “immune set point” after transplant.
Observational data outside lung transplantation support biological plausibility. Elevated NLR has been linked to adverse outcomes in cardiovascular disease, malignancy, infections,^11,12 interstitial lung disease (ILD),¹³ pulmonary hypertension (PH),¹⁴ and chronic obstructive pulmonary disease (COPD).¹⁵ Elevated NLR also has been associated with adverse outcomes across several solid organ transplant settings.^16-19
Within lung transplantation, small studies have suggested that perioperative or early posttransplant NLR is associated with subsequent rejection risk, whereas bronchoalveolar lavage (BAL) and blood compartment readouts can diverge and may be variably influenced by infection, sampling time, and immunosuppression.^18-20 In lung transplantation, findings are inconsistent. Kanou and colleagues²⁰ reported that elevated preoperative NLR predicted asymptomatic ACR, whereas Wiegand and colleagues,¹⁸ who measured NLR in the acute postoperative phase, reached different conclusions on NLR as a reliable inflammatory marker in posttransplant bacterial and fungal infections after lung or liver transplant. Discrepancies may reflect differences in study design, patient selection, and timing of NLR measurements.
The value of serial NLR measurements in peripheral blood and BAL fluid for detecting BLAD or predicting ACR within the first year posttransplant is unknown. Although biologically plausible, the prognostic role of NLR in lung transplant remains unproven, and current surveillance tools, such as donor-specific antibody testing, spirometry, and transbronchial biopsies, are limited by imperfect accuracy or feasibility issues. The level of NLR is also influenced by factors such as infection, perioperative inflammation, and corticosteroid therapy. Converging evidence from molecular assays such as donor-derived cell-free DNA and blood-based transcriptomic profiling suggests that composite approaches may outperform single cellular ratios by capturing complementary dimensions of allograft injury and alloimmune activation.^21,22
We hypothesized that elevated peripheral blood and BAL NLR at predefined time points (pretransplant and at 3, 6, 9, and 12 months posttransplant) would be independently associated with increased risk of BLAD and/or ACR and that tracking NLR trajectories could enhance current risk stratification strategies.

Materials and Methods

Study design and population
In this single-center retrospective cohort study, we included all adult patients (≥18 years) who underwent bilateral lung transplant at the University Hospital Zurich, Switzerland between January 1, 2021, and December 31, 2023. Patients were eligible if complete clinical and follow-up data were available for at least 12 months posttransplant. We excluded patients who underwent retransplant within the first year or who had incomplete follow-up data.
The Cantonal Ethics Committee of Zurich, Switzerland (BASEC-Nr. 2024-02487) approved the study before study initiation. All procedures were conducted in accordance with the ethical standards of the institutional research committee and in accordance with the 1975 Declaration of Helsinki, as revised in 2013.

Data collection
We obtained baseline demographic data, including age, sex, and primary pulmonary diagnosis (COPD, ILD, PH, cystic fibrosis [CF]). We also obtained transplant-related variables (donor age and cause of death, surgical duration, and lengths of stay in the ICU, transplant hospital ward, and rehabilitation facility).
We analyzed peripheral blood neutrophil and lymphocyte counts obtained pretransplant (within 24 h before surgery) and at 3, 6, 9, and 12 months posttransplant. We calculated NLR as the absolute neutrophil count divided by the absolute lymphocyte count.
Bronchoalveolar lavage cell differentials were collected during surveillance bronchoscopies at corresponding follow-up time points when available. We calculated BAL NLR as the percentage of neutrophils divided by the percentage of lymphocytes. Absolute BAL cell counts were not available for all procedures and were therefore not used in the NLR calculation. This approach was chosen to ensure consistency across procedures, as absolute counts were unavailable in a substantial subset. Although we did not perform a formal comparison between patients with and without BAL differentials, missingness primarily reflected procedural factors (eg, low sample recovery) rather than clinical condition. Surveillance bronchoscopies with transbronchial biopsies and BAL were scheduled at approximately 3, 6, 9 and 12 months posttransplant. Acute cellular rejection had been graded according to the 2006 ISHLT Working Formulation by thoracic pathologists blinded to NLR results.³ Patients received pulmonary function tests (FEV1 and FVC, in liters and % predicted) at every follow-up visit (including at 3, 6, 9, and 12 months).

Definitions
We defined BLAD as failure to achieve a peak FEV1 and/or FVC ≥80% predicted on at least 2 pulmonary function tests performed ≥90 days apart within the first year posttransplant, in accordance with previously published criteria.¹ We defined ACR histopathologically (on transbronchial lung biopsy) as ISHLT grade A1 to A4, with all grades grouped as “ACR” for analysis.
To minimize peri-event confounding, we excluded NLR measurements taken within 14 days of a clinically adjudicated infection or a high-dose corticosteroid pulse (≥250 mg methylprednisolone equivalent/day for ≥1 day).

Handling of missing data
We restricted our analyses to patients with at least 3 evaluable longitudinal NLR measurements. For each time point, only patients with both NLR data and outcome data were included (pairwise deletion). No imputation of missing values was performed.
We performed the BAL NLR analyses on the subset of patients with available BAL differential counts.

Outcomes
For study outcomes, we included presence of BLAD and occurrence of ACR within the first year posttransplant. Both BLAD and ACR were assessed independently and were not mutually exclusive.

Statistical analyses
We presented continuous variables as mean ± SD when normally distributed or as median (interquartile range) when skewed, which we compared using the appropriate statistical test. We performed between-group comparisons (BLAD vs no BLAD) for continuous variables with the Mann-Whitney U test. We compared categorical variables with the χ^² test or the Fisher exact test, as appropriate.
We used receiver operating characteristic (ROC) curve analysis to evaluate the ability of peripheral blood NLR at each time point to predict ACR. We calculated the area under the curve (AUC) with 95% CI by using the nonparametric DeLong method. We determined optimal cutoff values by maximizing the Youden index (sensitivity + specificity - 1) and reported corresponding sensitivities, specificities, and Youden indices.
We performed multivariable binary logistic regression to examine the association between month 3 NLR and ACR, adjusting for BLAD status, sex, and diagnosis group (COPD, ILD, PH, CF). Diagnosis groups were entered as categorical variables (reference: COPD) to account for potential disease-specific immune variation. We assessed model fit with the Hosmer-Lemeshow goodness-of-fit test and expressed explanatory power as Nagelkerke’s R².
Given the retrospective nature of this study and the absence of prior effect size estimates, no formal sample size or power calculation was performed. All analyses were considered exploratory. We considered 2-sided P < .05 as statistically significant. We used IBM SPSS Statistics for Mac version 30.0.0.0 (IBM Corp) for statistical analyses. We exported ROC coordinate tables via the Output Management System to extract cutoffs and corresponding diagnostic performance metrics. To reduce confounding by acute inflammatory episodes, we excluded NLR values obtained within 14 days of a clinically adjudicated infection or high-dose corticosteroid pulse (≥250 mg methylprednisolone equivalent per day) from analysis. Data on immunosuppressive drug levels and perioperative complications were heterogeneous and therefore not included as covariates. A post hoc power estimation indicated that, with 80 recipients and 19 acute rejection events, the study had approximately 40% power (α = 0.05, 2-sided) to detect an odds ratio of 2.5 for NLR. Analyses were therefore considered exploratory and primarily descriptive.

Study population
Eighty bilateral lung transplant recipients met the inclusion criteria. Baseline characteristics are sum-marized in Table 1. The mean age was 54.3 ± 12.7 years, and 63% were male. The most common underlying diagnoses were COPD (59.8%) and ILD (30.5%).
Among 80 patients, 47 (58.8%) presented with BLAD (BLAD group) and 33 patients (41.2%) had no baseline dysfunction (non-BLAD group). The BLAD group had a significantly longer total hospital stay compared with the non-BLAD group (48.5 ± 21.4 vs 32.7 ± 16.9 days; P < .001).
During the first year posttransplant, ACR occurred in 19 of 70 patients with complete data (27.1%), without significant difference between BLAD and non-BLAD patients (P = .391).

Peripheral blood neutrophil-to-lymphocyte ratio
Mean peripheral blood NLR values at each time point are presented in Table 2. Across all post-transplant intervals, BLAD patients exhibited nume-rically higher NLR values than non-BLAD patients. The largest difference was observed at 12 months (8.72 ± 15.48 vs 4.11 ± 2.99; P = .057). Pretransplant NLR values were similar between groups (5.50 ± 5.44 vs 4.32 ± 3.79; P = .360). Although mean NLR was numerically higher in BLAD patients at all time points, the differences were not significant, and interindividual variability was marked, exemp-lified by the very wide standard deviation at 9 months (37.25).

Bronchoalveolar lavage neutrophil to lymphocyte ratio
Bronchoalveolar lavage NLR data are shown in Table 3. No significant differences were observed between BLAD and non-BLAD patients at any follow-up time point. Considerable interindividual variability was present, particularly among BLAD patients. At 9 months, BLAD patients showed a higher mean BAL NLR (6.72 ± 12.64) compared with non-BLAD patients (0.85 ± 0.74), but the difference did not reach significance (P = .141). Considerable interindividual variability was present, especially among BLAD patients at earlier follow-up.

Receiver operating characteristic curve analysis for prediction of acute rejection
Receiver operating characteristic curve analysis demonstrated poor discriminatory performance of peripheral blood NLR for predicting ACR at any of the assessed time points within the first year posttransplant (AUC range, 0.381-0.516) (Table 4). Optimal cutoff values, defined by the highest Youden index, yielded sensitivities between 77.8% and 100% and specificities between 73.9% and 96.1%. However, in most cases, high sensitivity was accompanied by a low Youden index, indicating limited overall dis-criminatory ability. Subgroup analysis in BLAD patients showed similarly poor discrimination (AUC range, 0.38-0.60; all P > .05), with low Youden indices and no consistent pattern in optimal cutoffs.

Multivariable analyses
In multivariable logistic regression (Table 5) inclu-ding month 3 NLR, BLAD status, sex, and diagnosis group, none of the covariates were independently associated with ACR (all P > .05). The model demonstrated acceptable fit (Hosmer-Lemeshow P = .683) but explained only a small proportion of variance (Nagelkerke R2 = 0.134). In univariate analysis, COPD was more common in the non-BLAD group (70.6% vs 52.1%), although the result was not significant (P = .11).

In this single-center cohort of 80 bilateral lung transplant recipients, we found no evidence that NLR, measured in either peripheral blood or BAL fluid, predicted BLAD or ACR during the first year posttransplant. Although mean NLR values were consistently higher in BLAD patients, differences were small, highly variable, and not significant. Across all assessed time points, NLR performed at chance level for predicting year 1 ACR (AUC ≈ 0.5). Apparent high sensitivities or specificities at some thresholds were small-sample artifacts accompanied by very low Youden indices (<0.20), confirming the absence of meaningful discrimination.
Neither ROC curve analysis nor multivariable logistic regression showed that peripheral blood NLR could meaningfully predict ACR in the first year after lung transplant. Area under the curve values were uniformly low (range, 0.381-0.516), and regression coefficients were not significant, indicating that NLR (whether used alone or adjusted for clinical covariates) did not reliably distinguish between patients with and without rejection or allograft dysfunction. Some cutoff values reached high sensitivities, but values came with only moderate specificities and little improvement in overall classification. In BLAD-only subgroup analyses, discriminatory performance remained poor (AUC range, 0.38-0.60; all P > .05), with low Youden indices and no consistent optimal cutoff pattern, supporting the robustness of the primary results.
Although none of the associations between NLR and BLAD or ACR reached significance in 2-sided analyses, we noted that, in an exploratory 1-sided test at 12 months, an association between elevated NLR and BLAD was significant. Given the post hoc and directional nature of this analysis, these findings should be regarded as hypothesis-generating only and interpreted with caution.
Our findings are in line with previous solid-organ transplant studies showing that single, nonspecific inflammatory biomarkers often have limited value in the complex posttransplant immune environment, where infection, perioperative stress, and immuno-suppressive therapy can markedly alter leukocyte profiles. The discrepancy between these results and Kanou and colleagues, who reported that elevated preoperative NLR predicted asymptomatic ACR, is likely due to differences in study design, post-transplant immunosuppression protocols, and espe-cially timing of measurement.²⁰ Their pretransplant assessments captured existing systemic inflammation, whereas our repeated measurements up to 12 months posttransplant were influenced by infections, immunosuppression, and perioperative stress. Together, these findings strengthen the case for a multimodal risk-stratification strategy that integrates clinical data, histopathology, molecular diagnostics, and possibly absolute blood or BAL cell counts, rather than relying on single indices. Such composite models are more likely to capture the complex immune changes after transplant and to guide more precisely, individualized surveillance and possibly treatments.
Similar heterogeneity has also been observed in other solid-organ transplant settings. For example, in living-donor liver transplant, peri-event increases in NLR were associated with subsequent ACR,²³ whereas, in heart transplant, temporal trends in NLR, rather than isolated, static measurements, were more predictive of risk of rejection.²⁴ This differs from Krishnan and colleagues,¹⁹ who associated elevated baseline NLR with long-term survival rather than early rejection, highlighting that outcome definitions critically influence apparent predictive value.Together, these findings emphasize that both the timing of NLR assessment and the evaluation of its dynamic changes over time are critical when considering this parameter as a potential biomarker. This temporal dimension may partly explain why, in our cohort, NLR measured at predefined, fixed posttransplant intervals failed to demonstrate consistent predictive value.
The relevance of BLAD as a clinical endpoint has been well established. Contemporary multicenter data have confirmed that BLAD, defined by failure to achieve ≥80% predicted FEV1 or FVC on repeated testing, is common and associated with adverse outcomes.^1,25 Given its high prevalence, early iden-tification of BLAD patients who are also at increased risk for ACR would be clinically valuable. However, our findings suggested that NLR, measured in blood or BAL, is not sufficiently sensitive or specific to serve this purpose.
Bronchoalveolar lavage fluid offers a direct win-dow into the allograft immune environment. Previous studies have linked increased BAL neutrophils to bronchiolitis obliterans syndrome,²⁶ and immu-nophenotyping has shown the ability to distinguish infection from rejection.¹⁹ Systematic reviews advocate for composite biomarker frame-works that integrate cellular, cytokine, and transcriptomic data rather than reliance on single indices such as BAL NLR.²⁷ The poor performance of BAL NLR in our study likely reflects the limitations of percentage-based ratios, which may obscure biologically meaningful differences compared with absolute cell counts.²⁸
An additional consideration relates to the statistical power of our study. With 80 patients and 19 acute rejection events, our cohort was modest in size, limiting the ability to detect small-to-moderate effect sizes. Although our analyses were exploratory and not formally powered, the consistently low AUC values and nonsignificant regression results suggested that the absence of predictive value is unlikely to be explained solely by type II error. Nevertheless, larger multicenter cohorts will be required to exclude more subtle associations.
Missing BAL data represented another relevant limitation. Bronchoalveolar lavage cell differentials were incomplete at several time points, and analyses relied on percentage-based ratios rather than absolute counts. Both factors may have attenuated potentially meaningful signals. Standardized collection of absolute BAL counts across centers would enhance comparability and interpretability in future studies.
From a clinical perspective, our results do not support the routine use of NLR alone to guide surveillance intensity or immunosuppressive adjust-ments in the first year posttransplant, even in higher-risk subgroups such as BLAD patients. The variability and lack of discriminatory performance observed here reinforced that single, nonspecific inflammatory markers are unlikely to provide robust decision-making support in lung transplantation. In practice, single nonspecific inflammatory indices such as NLR are unlikely to provide robust decision support in isolation. Noninvasive molecular tools, including donor-derived cell-free DNA and blood-based transcriptomic signatures, offer complementary, mechanistic insights into allograft injury and alloim-mune activation. Embedding simple cellular ratios within such multimodal frameworks, alongside spiro-metry, histopathology, and serological assays, may enable more precise and individualized surveillance (Figure 2).
Several features of our design may have biased results toward the null. By excluding NLR measu-rement taken within 14 days of infection or corticosteroid pulses, we reduced, but did not eliminate, confounding from peri-event inflammation. Residual variability from unmeasured factors, together with the modest sample size, may have masked weak associations. Overall, any remaining bias is more likely to have diluted true effects than to have generated false-positive findings.
This study has several limitations that should be considered when interpreting the findings. First, the modest sample size (80 recipients with 19 acute rejection events) limited statistical power; post hoc estimation indicated approximately 40% power (α = 0.05, 2-sided) to detect a moderate effect (odds ratio 2.5). The single-center design, while it ensured standardized diagnostic and follow-up protocols, may limit generalizability to centers with different surveillance schedules, patient populations, or immunosuppressive regimens.
Second, BAL data were incomplete at several follow-up time points, reducing statistical power and potentially introducing selection bias if patients with missing BAL data differed systematically in clinical status or rejection risk. In addition, BAL NLR was derived from percentage-based differentials rather than absolute cell counts, a pragmatic approach chosen because absolute counts were unavailable for a substantial subset. Missingness was primarily procedural (eg, low sample recovery) rather than clinical. This may still dilute biologically relevant signals, as absolute cell concentrations can provide distinct and potentially more meaningful information.
Third, although NLR measurements taken within 14 days of infection or high-dose corticosteroid pulses were excluded to reduce confounding, residual variability due to unmeasured inflammatory events or differences in immunosuppressive exposure cannot be completely ruled out.
Fourth, potential detection bias related to the bronchoscopic surveillance schedule could have influenced the measured incidence of ACR. Although the protocol was standardized, variations in for-cause biopsy rates or timing may have affected the identification of asymptomatic or low-grade rejection episodes.
Finally, we did not formally assess the temporal relationship between NLR changes and subsequent ACR; we also did not explore potential interactions between NLR and specific immunosuppressive regimens. Future multicenter studies with larger cohorts, more rigidly standardized BAL quantification, longitudinal trajectory analyses, and integration of NLR into composite biomarker models are warranted to clarify its potential role in lung transplant immune monitoring and outcome prediction.

In this cohort of bilateral lung-transplant recipients, neither peripheral blood nor BAL NLR demonstrated meaningful predictive value for BLAD or ACR within the first year after lung transplant. Despite its biological plausibility and ease of measurement, NLR lacks discriminatory power and should not be used as a stand-alone surveillance tool. The role of NLR may lie within integrated, multimodal biomarker frameworks that combine functional, histological, and molecular parameters to improve posttransplant risk stratification.

Volume: 23 Issue: 11 November 2025

FULL TEXT