Original Article |
Corresponding author: Anastas Chapkanov ( achapkanov909@gmail.com ) © 2024 Anastas Chapkanov, Miroslava Todorova, Antoaneta Chirlova, Blagoi Marinov.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Chapkanov A, Todorova M, Chirlova A, Marinov B (2024) Factors affecting prediction accuracy of postoperative FEV1 and DL,CO in patients undergoing lung resection. Folia Medica 66(2): 171-178. https://doi.org/10.3897/folmed.66.e121799
|
Introduction: Despite significant development in systemic therapy and radiotherapy, surgery is still the cornerstone for curative lung cancer treatment. Although predicted postoperative function (ppo) somewhat exactly correlates with actual postoperative function bigger differences may be a cause of serious clinical outcome.
Aim: The aim of our study was to identify clinical factors affecting prediction accuracy of postoperative lung function for more careful selection of operable lung cancer patients.
Patients and methods: Seventy patients were studied prospectively. The preoperative lung function tests (FEV1 and DL,CO) were performed within a week before surgery, and the follow-up tests were performed 4 to 6 weeks after surgery. Calculation of predicted postoperative values were calculated by three methods: two segment formulas and vibration response imaging (VRI). The correlation between each clinical parameter and accuracy of prediction was screened on univariate analysis of Pearson’s correlation coefficient, and significant factors were confirmed by multivariate linear regression analysis applying backward stepwise elimination approach.
Results: Univariate linear regression analysis between the predicted and the actual postoperative values of FEV1% and DL,CO showed the highest prediction accuracy with acoustic mapping (VRI). Multivariate regression analysis showed that prediction accuracy of postoperative lung function is significantly affected by COPD (p<0.001) and volume of resection (p<0.001).
Conclusion: Vibration response imaging (VRI) is a more accurate method for predicting postoperative lung function than segment method formulas. Anatomical calculation significantly underestimates the postoperative values of FEV1% in patients with COPD. Prediction of FEV1% and DL,CO with segment counting is significantly influenced by the volume of resection.
lung cancer, postoperative lung function, prediction accuracy, vibration response imaging
Despite significant development in systemic therapy and radiotherapy, surgery is still the cornerstone for curative lung cancer treatment. Pulmonary lobectomy is the standard operative treatment for primary non-small cell lung cancer (NSCLC). Because of relatively high incidence of postoperative complications, the hospital mortality, as well as disappointing long-term survival after surgical resection of lung cancer, the appropriate selection of patients for pulmonary resection is a continuing challenge. Forced expiratory volume in one second (FEV1) and the diffusing capacity for carbon monoxide (DL,CO) are the most commonly used predictors of postoperative outcome.[
The aim of our study was to identify clinical factors affecting prediction accuracy of postoperative lung function for more careful selection of operable lung cancer patients.
We conducted a prospective cohort study of patients undergoing anatomical lung resection in the Department of Special Surgery at St George University Hospital in Plovdiv and the Department of Thoracic Surgery at Kaspela University Hospital in Plovdiv.
The preoperative lung function tests were performed within a week before surgery, and the follow-up tests were performed 4 to 6 weeks after surgery. Lung function tests were performed using the MasterScreen Body/Diffusion™ computerized spirometer (Jaeger, Wuerzburg, Germany) with real-time curve drawing and automatic correction (BTPS – body temperature pressure saturated). The postbronchodilator lung function values were used.
Predicted postoperative FEV1% and DL,CO%, were calculated by three methods.
The simple calculation method introduced by Juhl et al.[
SF1–(Juhl & Frost, 1975; Zeiher et al., 1995).[
Number of segments (S): ppoFEV1 = preoperative FEV1
Modified segment formula introduced by Bolliger that takes into account only functional (none obstructed) segments
SF2–(Bolliger et al. 2002)[
Number of functional segments (FS): ppoFEV1 = preoperative function
where y is the number of functional segments to be removed and z is the total number of functional segments.
The vibration response imaging system (VRIxp, Deep Breeze, Or-Akiva, Israel) quantifies breath sounds and displays the results as a dynamic image and numerical values. It measures the vibration energy of lung sounds generated during the respiratory cycle. Vibration response imaging (VRI) technology is harmless, non-invasive and does not require the addition of a tracer to the inhaled air or blood stream. The technology and the calculation were described in detail earlier.[
In the literature, we have identified clinical factors with potential impact on the accuracy of prediction of postoperative lung function: sex, body mass index, smoking, preoperative FEV1%, presence of COPD and type of operation.[
Patients were divided in groups according to the studied factor: male/female; smoker/ex-no smoker; patients with baseline FEV1% ≥ 80% and with FEV1% ˂ 80%; with COPD and without COPD as defined by GOLD 2007 criteria; patients with COPD index (COPDI) <1.5 and those with COPDI ≥1.5. To investigate the influence of operative intervention on the accuracy of prediction of the postoperative lung function, we performed a comparison between the results obtained in patients with upper and lower lobectomy, lobectomy and pulmonectomy and removal of >4 and ≤4 functioning lung segments. We compared the prediction accuracy of the three calculation methods. The influence of the clinical parameters on prediction was studied separately for each method.
The adjusted coefficient of determination (adjusted R2) was used to compare the prognostic accuracy of the three methods.
To assess the influence of studied clinical factors on the prediction made by the three different methods used (SF1, SF2 and VRI), the value of relative deviation in percentages was introduced:
where ppo – predicted postoperative values of FEV1 and DL,CO; and apo – actual postoperative values of FEV1 and DL,CO.
Metric variables were checked for normality of distribution by the Kolmogorov-Smirnov test. An independent sample t-test was used to compare means of continuous variables. The Mann-Whitney U-test was used for variables not normally distributed. The correlation between each clinical parameter and accuracy of prediction was screened on univariate analysis of Pearson’s correlation coefficient, and significant factors were confirmed by multivariate linear regression analysis applying backward stepwise elimination approach. A p-value less than 0.05 was considered to be significant in all statistical analyses.
The study design was approved by the Medical University of Plovdiv’s institutional review board and is in accordance with the Declaration of Helsinki ethical standards.
The baseline characteristics of the patients are summarized in Table
Patient characteristics | n | (%) | |
BMI | |||
˂ 18.5 | 2 | 2.9 | |
18.5-24.9 | 31 | 44.3 | |
25-29.9 | 19 | 27.1 | |
˃ 30 | 18 | 25.7 | |
COPD index | |||
˂ 1.5 | 25 | 35.7 | |
˃ 1.5 | 45 | 64.3 | |
GOLD | |||
Without COPD | 51 | 72.9 | |
COPD | 19 | 27.1 | |
Type of surgery | |||
Pneumonectomy | 17 | 24.3 | |
Upper lobectomy | 34 | 48.6 | |
Lower lobectomy | 19 | 27.1 | |
Smoking status | |||
smoker | 38 | 54.3 | |
no+ex smoker | 32 | 45.7 | |
Number of resected functional segments | |||
≤ 4 | 30 | 42.9 | |
> 4 | 40 | 57.1 | |
Preoperative FEV1% | |||
˃ 80% | 48 | 68.6 | |
˂ 80% | 22 | 31.4 |
Univariate linear regression analysis between the predicted and the actual postoperative values of FEV1% and DL,CO showed the highest prediction accuracy with acoustic mapping (VRI). The adjusted coefficient of determination for FEV1% was R2 adj=55.12% for VRI, R2 adj=46.23% for SF1, and R2 adj=31.29% for SF2. For DL,CO%, VRI showed R2 adj=64.00%. SF1-R2 adj=47.85% and SF2-R2 adj=45.11.
Univariate analysis of BMI and smoking status showed no statistically significant influence on prediction accuracy with all three methods.
When examining the influence of gender on the determination of ppo FEV1.0% and DL,CO, a difference was found. In men, the values were negative, indicating that the predicted values were lower than those measured postoperatively. For women, the values were higher than actually measured. Statistically significant difference was found only for the prediction of ppo FEV1% with acoustic mapping. A similar trend was observed for DL,CO(%), but the sex difference did not reach significance.
To investigate the influence of baseline preoperative FEV1 (%), patients were categorized into two groups: FEV1 (%) >80% and FEV1 (%) ≤80%. All three methods reported an overestimation of functional loss. When using the SF1 segment formula, we observed a significantly greater value of relative deviation (D%) of FEV1 (%) in the FEV1 (%) ≤80% group with a median of −25.58% compared to median −4.97% in patients with FEV1 (%) >80%, p=0.001. With the SF2 formula, a significantly higher D% value was also found in the FEV1 (%) ≤80% group with a median of −7.68% compared to a median of −3.03% in patients with FEV1 (%) >80%, p=0.012. D% showed no significant difference between the groups when calculated with VRI, p=0.256. No statistically significant difference was observed between the two groups in the calculation of DL,CO% with all the three methods.
In patients with COPD, all methods reported an underestimation of postoperative FEV1 (%). Significantly greater values were found in patients with COPD predicted with SF1 formula – median −20.68% compared to median −5.69% in patients without COPD, p=0.002. The difference was significant with SF2 formula, with a median of −12.4% in the COPD group and 4.00% in the non-COPD group, p=0.005. The relative deviation for diffusion capacity (D%) DL,CO% showed no significant differences between patients with and without COPD in all three prognostic methods.
In patients with COPDI ≥1.5 and COPDI <1.5, a significant difference was found in the prediction of FEV1 (%) with SF1 segment formula, with a significantly greater value in the group with COPDI <1.5 (-18.65±15.7%) compared to those with COPDI ≥1.5 (−2.81±18.8%) , p=0.001.
Lobectomy was performed in 53 patients, in 34 of whom with upper lobectomy and in 19 with lower lobectomy. Pulmonectomy was performed in 17 patients. Comparative analysis of the relative deviation between lobectomy and pulmonectomy patients was preceded by an intergroup comparison between upper and lower lobectomy patients. Based on the intergroup analysis, upper and lower lobectomy patients were pooled into one lobectomy group, due to lack of significant difference between upper and lower lobectomy patients in all indices and methods
When predicting FEV1 (%), the D% values, predicted with either of the three methods, were negative as a result of the greater underestimation of postoperative FEV1 (%) in the pneumonectomy patients compared to the lobectomy patients. The difference was significant in patients with pulmonectomy calculated with the first variant of the segment method SF1 (p<0.001). Significantly higher negative D% DL,CO% values were found in the pneumonectomy patients predicted with all three methods.
A significant difference was observed in the prediction of DL,CO% with both segment formulas when comparing patients with more than 4 segments removed to patients with 4 or less removed segments. A significantly higher relative deviation of DL,CO% was observed with SF1 method in the >4 segments removed group (−12.94±18.4%) compared to the ≤4 segments removed group (−2.14±16.3%), p=0.015. For SF2, the relative deviation D% DL,CO% was (6.94±14.3%) in the ≤4 segments removed group, while in those with >4 segments removed, D% DL,CO% had a negative value (−8.52±18.6), p<0.001. In SF2, the same dependence was observed in D% FEV1 (%), where in the group of patients with ≤4 segments removed, the relative deviation had a positive value (8.22±24.1%) and negative in the group with >4 segments removed (−6.31±17.8%), p<0.001.
Sex and operative intervention (pulmonectomy/lobectomy) were identified by univariate analysis as factors affecting the accuracy of VRI to predict postoperative FEV1 (%) and DL,CO%. The results of the multivariate regression analysis showed that they do not contribute significantly to improvement of the prediction accuracy
For the first variant of the segment method (SF1), multivariate regression analysis found out that two of the six variables included in the prognostic model have a significant prognostic role: COPD (p=0.019), in the prediction of postoperative FEV1% and the type of operation (lobectomy/pulmonectomy) in the prediction of DL,CO% (p<0.001).
For the second variant of the segment method (SF2), COPD (p=0.004) and number of removed segments (p=0.049) were found to be significant factors in prediction accuracy of FEV1%.
Medical operability of lung cancer has been frequently determined based on FEV1, DLCO, and VO2max. Accurate prediction of postoperative residual lung function is mandatory to minimize postoperative morbidity and mortality. Although predicted postoperative function (ppo) somewhat exactly correlates with actual postoperative function bigger differences may be critical in the patients with marginal lung function after lung resection. We should consider that the accuracy can be affected not only by the technique to measure the regional lung function, but also several clinical factors
In our study, we found that VRI-based prediction was a more accurate method than anatomical calculation irrespective of the extent of resection. This result is in consistency with Berreta et al.[
We observed a higher overestimation of functional loss in patients with FEV1%<80%, (postoperatively measured values were higher than predicted) in all three methods. Boushy et al.[
This smaller difference in the degree of functional loss in individuals with poor baseline lung function has also been observed by Baldi et al.[
COPD is defined in several ways, and differences in definitions and diagnosis affect estimates of disease severity. The Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2007 defined the disease in degrees of clinical severity based on FEV1 and FEV1/FVC (forced vital capacity) from post-bronchodilator spirometry. To define airflow limitation, the fixed ratio FEV1/FVC <0.7, measured after the bronchodilator, is recommended, despite the risk of overdiagnosis.[
Using this definition, we found a significantly less volume loss in COPD patients when comparing predicted versus actual postoperative FEV1 (%) values calculated with the segmental (SF1 and SF2) formulas, and no significant differences in DL,CO%. Sekine et al.[
A study by Pompili et al.[
Another way to define and evaluate chronic obstructive pulmonary disease in the practice of thoracic surgery is the so-called “COPD index”. To classify patients according to severity and purity of obstructive pulmonary disease, Korst et al.[
We found underestimation of predicted postoperative FEV1 (%) and DL,CO% in patients with COPD index <1.5. Baldi et al.[
When examining the effect of the volume of resection (lobectomy/pulmonectomy) on the accuracy of prediction of ppo FEV1, we found greater underestimation in pneumonectomy patients compared with lobectomy patients. The difference was statistically significant only with the SF1 segment formula. This underestimation is statistically significant for DL,CO% in all three methods. Similar to our results, Bolliger et al.[
Kim et al.[
We found an underestimation of the postoperative indicators in the group of patients with more than 4 segments resected. This difference was significant when predicting DL,CO(%) and FEV1 (%) with the SF2 formula and for DL,CO (%) calculated with the SF1 formula.
In our study, we found no influence of the type of lobectomy (upper/lower) on the accuracy of prediction of postoperative lung function. Accurate assessment of anatomic-functional loss after lung lobectomy is also complicated by the fact that damaged lung areas, especially emphysematous areas, are often distributed heterogeneously in the upper or lower lobe, changing the functional roles of these lobes. Data are conflicting in the literature. Kim et al. found no influence of the type of lobectomy on the prediction of ppoFEV1 (%) and DL,CO (%).[
The theory of volume reduction surgery may explain the minimal change of apoFEV1 in patients with COPD. For patients without COPD, the authors speculated that accidental anatomical repositioning after upper lobectomy, which causes narrowing of the opening of the lower or middle lobe of the bronchi, and different movement and elevation of the diaphragm between upper lobectomy and lower lobectomy may be the potential reason for the minimal change in the cases of resection of the lower part of the lung.[
Sengul et al. found that in lower lobectomy, volume recovery is mainly due to expansion of the contralateral lung along with increase in the volume of ipsilateral remaining lung, especially after right lower lobectomy.[
We should consider that the accuracy of prediction can be affected not only by the technique to measure the regional lung function, but also several clinical factors such as preoperative FEV1%, the presence of obstructive lung disease, extent of lung volume resection
Vibration response imaging (VRI) is a more accurate method for predicting postoperative lung function than the segment method formulas.
Anatomical calculation significantly underestimates the postoperative values of FEV1% in patients with COPD.
Prediction of FEV1% and DL,CO with segment counting is significantly influenced by the volume of resection and should not be used alone to determine the postoperative values of FEV1% and DL,CO% in patients scheduled for an operative intervention greater than lobectomy.
A.C.: design, collection of data, operative task completion and critical review of manuscript; M.T.: design, collection of lung function and VRI data and critical review of manuscript; N.C.: collection of data, statistical analysis, and critical review of manuscript; B.M.: design, collection of lung function and VRI data and critical review of manuscript. All authors have contributed to the tasks of the study and critically reviewed the final version of the manuscript.
The authors have no support to report.
The authors have no funding to report.
The authors have declared that no competing interests exist.