Ultrasonographic modeling of lung and diaphragm mechanics: clinical trial of a novel non-invasive method to evaluate pre-operative pulmonary function

Introduction

Pulmonary function testing (PFT) is an important part of surgical evaluation of patients, and helps to monitor impaired function (Ruppel & Enright, 2012). Pre-operative PFT parameters including forced expired volume in the first second as a fraction of forced vital capacity are important predictors of post-operative complications (Matsumi et al., 2017; Zhang et al., 2015). However, the efficacy of PFT is constrained by the patient’s ability to comprehend the test’s fundamentals and to cooperate with the examiner (Tojo, 2006). Therefore, there is a pressing need to identify alternative methods for assessing pulmonary function that can mitigate these limitations. The aim of this study was to develop objective criteria to identify patients at high risk for pulmonary dysfunction.

Ultrasound (US) is a viable and non-invasive assessment instrument that only requires minimal cooperation from a patient, making it a promising tool for clinical applications.

Lung ultrasonography (LUS) is a valuable diagnostic tool that complements physical examinations across a variety of pulmonary disease states and clinical settings (Volpicelli et al., 2012). Its capability to detect alterations in both the lung parenchyma and pleural cavities enhances its accessibility. Numerous studies have demonstrated that LUS offers significant benefits in the diagnosis of various respiratory diseases, including acute respiratory distress syndrome (Stefanidis et al., 2011), pulmonary edema (Gargani et al., 2008), interstitial lung disease (Pitsidianakis et al., 2022), pneumothorax (Laursen et al., 2021), atelectasis (Monastesse et al., 2017), pneumonia (Ma et al., 2023), and others. Furthermore, LUS proves beneficial for diagnosing and assessing lung diseases in specific populations, such as newborns (Ma et al., 2023), children (Amatya et al., 2023), and pregnant women (Piccolo et al., 2023). While LUS can serve as a key diagnostic and evaluation tool for a range of respiratory conditions, it is important to note that the comprehensive evaluation of pulmonary function necessitates both lung and respiratory muscle assessments. Consequently, LUS alone is insufficient for accurately evaluating pulmonary function.

LUS examines the lung fields and pleura, but does not assess respiratory muscle function. Diaphragm ultrasound (DUS) is widely utilized to assess diaphragmatic activity. Several studies have indicated a correlation between the ultrasound indices of the diaphragm and lung function parameters across various conditions, including osteoporosis, vertebral fractures, kyphosis, neuromuscular disease (DePalo & McCool, 2002), patients with stroke (Chen et al., 2023), chronic obstructive pulmonary disease (COPD) (Topcuoğlu et al., 2022), and healthy subjects (Cardenas et al., 2018). US measurements of diaphragm excursion (DE) and diaphragm thickening fraction (DTF) have been validated as effective means for assessing diaphragmatic function (Tralhão et al., 2020; Unal et al., 2000).

In recent decades, LUS and DUS have been utilized to evaluate the characteristics of the lung and diaphragm, respectively, and serve as diagnostic methods for thoracic diseases that require minimal patient cooperation. While neither LUS nor DUS can fully assess pulmonary function independently, the potential for their combination to quantify pulmonary function remains unexplored. The primary objective of our prospective cross-sectional study was to develop a novel approach for assessing pulmonary function using a mathematical model based on LUS combined with DUS patterns in a cohort of perioperative patients by exploring the correlations between the ultrasound indices and PFT, thereby providing a non-invasive alternative for evaluating lung function in perioperative patients who may not fully comprehend or cooperate with the examiner.

Materials and Methods

Results

Of all 208 patients included in the study, 35 patients underwent thoracic surgery, 58 patients underwent urologic surgery, 65 patients underwent laparoscopic surgery, and 50 patients underwent breast surgery. Among these, 174 patients exhibited normal pulmonary function testing, while 34 patients had abnormal pulmonary function (Fig. 1). As shown in Table 1, the abnormal pulmonary function group had a greater mean age (67.82 ± 8.89 vs. 62.25 ± 11.38, P < 0.001), a higher LUS score (7.06 ± 2.79 vs. 1.85 ± 2.10, P < 0.001), a lower DE (Q-DE:1.91 ± 0.44 vs. 2.15 ± 0.21; D-DE: 4.01 ± 1.05 vs. 5.20 ± 0.47; P < 0.001), and a lower DFT (Q-DFT: 26.41 ± 7.07 vs. 41.22 ± 10.23; D-DFT: 41.08 ± 7.39 vs. 72.84 ± 16.36; P < 0.001) compared to the normal pulmonary function group. There were no significant differences in sex or body mass index between the two groups.

The analysis of the relationships among the variables revealed a strong negative correlation between the LUS score and several pulmonary function parameters, including FEV1/FVC, maximum expiratory flow at 50% of lung capacity (MEF50), FVC, maximum voluntary ventilation (MVV), and DLCO (P < 0.001). Conversely, D-DE, Q-DTF, and D-DTF exhibited significant positive correlations with FEV1/FVC, MEF50, FVC, and MVV (Fig. 2). Notably, no other US variables correlated with the pulmonary function parameters. In the univariate logistic regression analysis, six variables were strongly associated with the risk of pulmonary dysfunction: age, Q-DE, D-DE, Q-DTF, D-DTF, and LUS score. The multivariable analysis indicated that D-DTF (odds ratio (OR), 0.88; 95% confidence interval (CI), [0.83–0.94]; P < 0.001) and LUS score (OR, 1.44; 95% CI, [1.16–1.80]; P < 0.001) were independently associated with pulmonary function in our study (Table 2).

Based on the transformation of independent risk factors and the logistics equation as described by Nashef et al. (2012), we developed an ultrasonographic modeling score (U-score) represented by the equation U-score = −0.126 × D-DTF + 0.368 × LUS score. The model was evaluated using the Hosmer-Lemeshow goodness-of-fit test, which yielded a result of P > 0.05, indicating that the model passed the test.

Furthermore, the U-score demonstrated a more significant negative correlation with FEV1/FVC (r = −0.605, P < 0.001) compared to the LUS or DUS index when assessed individually (Fig. 3). The AUC of the U-score was greater than that of the other indices in the assessment of patients’ pulmonary function (Fig. 4, Table 3).

Ultimately, two sets of outliers were identified in the course of our trial. Two patients exhibited normal PFT but LUS scores of 9 and 17, respectively, and D-DTF values of 63.64% and 77.27%. The U-scores for these two patients were −4.71 and −3.48, respectively, which were greater than the cut-off value (−6.086).

Discussion

Pre-operative PFT are essential assessment tools for predicting pulmonary complications (Wu et al., 2021; Matsumi et al., 2017; Liu et al., 2019; van Huisstede et al., 2013). An abnormal PFT value, specifically a FEV1/FVC ratio of less than 70%, remains the sole predictive covariable for postoperative complications (Yoshimi et al., 2016). Among elderly patients with increasingly complex health conditions, reduced mobility, inadequate communication skills, cognitive impairment, ventilator dependency, or poor cooperation may affect the ability to complete PFT. Although pre-operative PFT is important for predicting postoperative complications, but the effectiveness is low in the elderly patients described above (Tojo, 2006). These patients need an alternative method to assess pulmonary function. The common non-invasive alternative to pulmonary function testing is the breath-holding test (Sylvester et al., 2020), but the result of the breath-holding test can be influenced by cardiorespiratory abnormalities (Sharma & Shekh, 2017) and should be judged on a clinical basis. It is important to note that some patients, despite some degree of abnormality on routine pulmonary function tests, may have the duration of the breath-holding within the normal range due to training in breath-holding (e.g., buteyko breathing technique (Mavkar & Shukla, 2024)), which may not be consistent with the PFT. Our trial developed a novel non-invasive approach for assessing pulmonary function using a mathematical model based on LUS combined with DUS patterns which only requires a minimal cooperation from patients.

Ultrasound is a viable and non-invasive assessment instrument that is independent of the features controlled by the patient and can be used as an alternative method to assess lung function and overcome these limitations. In this study, we have devised statistical models using US measurement to predict pulmonary function during surgery and identify major risk factors, providing an additional objective criterion for identifying patients at high risk of pulmonary dysfunction. Furthermore, according to the univariate and multivariable analysis, D-DTF (OR, 0.88; 95% CI [0.83–0.94]; P < 0.001), and LUS score (OR, 1.44; 95% CI [1.16–1.80]; P < 0.001) were each independently associated with pulmonary function in our study. We used the independent risk factors to produce a mathematical model that estimates pulmonary function. The variables incorporated into this model can be easily obtained in clinical settings, facilitating their extension and application.

In accordance with previous studies (Cardenas et al., 2018; Cohen et al., 1994), we found positive correlations between DE and lung volume (FVC and FEV1), but sex influenced these correlations. DE more significantly correlated with female sex. Beyond the known differences in airway structure between sexes (Harms, 2006), it has already been reported that females also have smaller dimensions of the rib cage and, during QB, they have lower tidal volume, minute ventilation and abdominal contribution than males (Romei et al., 2010). In this study, DE was not an independent risk factor in multivariate regression, this can be attributed to the no difference in sex between the two groups. But it may be an area that requires further study.

Our study revealed a noteworthy correlation between FEV1/FVC (one of the most important variables of PFT) and D-DTF (r = 0.519, P < 0.001), LUS score (r = −0.592, P < 0.001). A similar correlation has been observed in a previous study (Cardenas et al., 2018; Zheng, Wu & Tan, 2016). The correlation between D-DTF, LUS and FEV1/FVC suggests that ultrasound measurements can provide an accurate pulmonary function assessment in patients who cannot cooperate with PFT. This is especially valuable for patients who cannot go anywhere or are confused. One of the key findings of the present study was that U-score (r = −0.605, P < 0.001) has a stronger correlation with FEV1/FVC than D-DTF and LUS, respectively. ROC curve analysis showed that both D-DTF and LUS score were accurate in predicting abnormal lung function results; however, the U-score had a larger AUC (0.975) and higher specificity and sensitivity for predicting pulmonary dysfunction when compared to both D-DTF and LUS score. This suggests that the combination of LUS and DUS offers enhanced predictability in preoperative lung function assessments. Multiple studies have confirmed that LUS is highly accurate for diagnosing many lung diseases (Stefanidis et al., 2011; Gargani et al., 2008; Pitsidianakis et al., 2022; Laursen et al., 2021; Monastesse et al., 2017), and DUS is a novel technique for assessing diaphragmatic movement (Cardenas et al., 2018; Haji et al., 2018; Boussuges, Gole & Blanc, 2009). However, the current investigation demonstrated that the use of LUS or DUS alone is limited in predicting preoperative lung function. The U-score combines LUS and DUS results to obtain a comprehensive assessment of lung function with high specificity and sensitivity for pulmonary dysfunction. This overcomes one of the limitations of these tools. Two patients had normal PFT but with LUS score >9 in the present study. All the patients had a history of bronchiectasis or infection. Their DUS parameters were normal, whereas LUS showed multiple solid signs (bronchial inflation, fragmentation, and tissue-like signs) in the lesion. The U-score of all two patients were greater than the cut-off value (−6.086), indicating that two patients had abnormal lung function and that they had a significantly higher probability of developing postoperative pulmonary complications. During the post-operative follow-up, they developed varying degrees of post-operative lung infection, resulting in prolonged hospitalization in line with our predictions. A potential explanation for these findings is that PFT assess the overall function of both lungs and are not capable of evaluating the functional indices of individual lung lobes. Furthermore, PFT may exhibit a time lag, as they typically only become abnormal when the lung tissue has sustained significant damage (Akira et al., 2009).

This model can be used to assess the pulmonary dysfunction in patients undergoing perioperative surgery. The higher the total number of points in the model, the greater is the likelihood of pulmonary dysfunction. The model’s indicators are conveniently repeatable in clinical settings, facilitating the evaluation of whether patients are susceptible to pulmonary dysfunction, enabling the early identification of at-risk patients, as well as early detection, intervention, and improvement of post-operative outcomes.

The present study had some limitations. First, LUS and DUS are operator-dependent methods, it is necessary to improve the accuracy of ultrasound procedures by implementing standardized protocols and training for sonographers. Second, the efficacy of this method of assessment is not evaluated in different surgeries and must be studied in the future. Third, this was a single-center study, and future multicenter studies may confirm our findings in larger patient populations. Fourth, the exclusion criteria included body mass index exceeding 35 kg/m² in our trial, but obesity оr physical changes may interfere with the accuracy of diaphragmatic and lung ultrasound, affecting the reliability of U-scores in such populations. Therefore, subgroup analysis should be carried out in the future to expand on the findings.

Conclusions

With validation, the U-score using both LUS and DUS may provide a option for estimating pulmonary function. This approach facilitates the assessment of pulmonary function in patients who may be unable to reliably participate in pulmonary function testing. This model can be easily integrated into existing clinical workflows, as ultrasound data can be collected during routine examinations. This allows patients at high risk of postoperative complications to be identified early, and interventions can be initiated in a timely manner.

Supplemental Information