A nomogram model for predicting pneumothorax after CT-guided localization of pulmonary nodules using autologous blood and methylene blue

Abstract

Background: The widespread use of low-dose chest computed tomography (CT) has significantly increased the early detection rate of small nodules. Existing localization methods have certain limitations. Preoperative localization using autologous blood combined with methylene blue has garnered attention due to its dual advantages. This study aims to evaluate the safety and pneumothorax risk of CT-guided preoperative localization using this technique and explore the risk factors associated with pneumothorax occurrence.

Methods: This retrospective study included 112 patients who underwent CT-guided preoperative lung nodule localization using autologous blood and methylene blue between November 2019 and November 2024 at The First Hospital of Putian. Patient demographics, imaging characteristics, procedural details, and post-localization complications were collected. Logistic regression was used to analyze the independent risk factors for pneumothorax.

Results: The localization success rate was 90.2%, and the pneumothorax incidence was 16.1%. Multivariate analysis identified white blood cell (WBC) count [odds ratio (OR) 1.43, 95% confidence interval (CI): 1.06-1.96, P=0.02] and the needle-tip-to-visceral-pleura distance (OR 2.36, 95% CI: 1.07-5.44, P=0.04) as independent risk factors for pneumothorax. A predictive nomogram model with an area under the receiver operating characteristic (ROC) curve (AUC) of 0.762 was developed, demonstrating good predictive performance.

Conclusions: Autologous blood combined with methylene blue is a safe and effective localization method for lung nodules. WBC count and needle-tip-to-visceral-pleura distance are independent risk factors for pneumothorax. The nomogram model provides valuable assistance for preoperative risk assessment.

Keywords: Computed tomography-guided localization (CT-guided localization); autologous blood; methylene blue; pneumothorax risk; pulmonary nodules.

Figure 1

The markers appeared blue-purple under thoracoscopic observation, indicating successful localization of the pulmonary nodule.

Figure 2

Forest plot of univariate and multivariate logistic regression analyses identifying risk factors for pneumothorax. BMI, body mass index; CI, confidence interval; FEV1/FVC, forced expiratory volume in one second/forced vital capacity; FVC, forced vital capacity; MVV, maximum voluntary ventilation; NA, not applicable; OR, odds ratio; WBC, white blood cell.

Figure 3

Nomogram for predicting pneumothorax risk. The lung depth and WBC count are assigned points based on their respective values. The total score is calculated by summing the individual points, which corresponds to the predicted probability of pneumothorax (Pr). *, P<0.05; **, P<0.01. WBC, white blood cell.

Figure 4

ROC analysis and bootstrap validation for predicting pneumothorax risk. (A) The ROC curve for the nomogram model, showing an AUC of 0.762 (95% CI: 0.702–0.771). (B) Internal validation using the bootstrap method, illustrating the distribution of t and quantiles of standard normal, confirming the model’s stability. t* represents the statistic (in this case, the AUC) calculated using the bootstrap method. * indicates that this statistic is derived from bootstrap resampling results. AUC, area under the receiver operating characteristic curve; CI, confidence interval; ROC, receiver operating characteristic.

Figure 5

This figure illustrates the calibration and clinical utility of the nomogram prediction model for pneumothorax risk. (A) Calibration curve of the nomogram prediction model. The dashed line represents the ideal model, the solid line represents the bias-corrected performance, and the dotted line shows the apparent performance. (B) DCA comparing the net benefit of the model against the “intervention for all” (trade-all) and “intervention for none” (repeat-none) strategies. The nomogram demonstrated higher net benefit across the 0–0.8 threshold probability range, indicating good clinical utility. DCA, decision curve analysis.