Pulmonary volumes and signs of chronic airflow limitation in quantitative computed tomography

Computed tomography (CT) offers pulmonary volumetric quantification but is not commonly used in healthy individuals due to radiation concerns. Chronic airflow limitation (CAL) is one of the diagnostic criteria for chronic obstructive pulmonary disease (COPD), where early diagnosis is important. Our aim was to present reference values for chest CT volumetric and radiodensity measurements and explore their potential in detecting early signs of CAL.

Four-dimensional (4D) chest computed tomography (CT) has the potential to follow pulmonary regional and global volume changes, that is, airflow pattern, during the respiratory cycle, but the radiation doses limit its present clinical use (Guerrero et al., 2006;Matsuzaki et al., 2013;Nair et al., 2021;Pan et al., 2004;Qiu et al., 2023).Chest CT images from 3D CT, acquired at the end of inspiration and expiration, is currently the most attractive alternative, awaiting a 4D implementation (Murphy et al., 2012;Simon, 2000).In contrast to pulmonary function tests (PFTs) performed sitting or standing, 3D chest CT is performed in the supine position where the air exchange, as well as the intrathoracic pressure gradient, decreases and ventilatory inhomogeneity increases (Mcmichael & Mcgibbon, 1939).
Quantitative automated chest CT volumetric measures from these images are evolving (Crossley et al., 2018;Herth et al., 2018;Mascalchi et al., 2017) and automatized quantification of, for example, volumes and radiodensity, measured in Hounsfield units (HU) is developing (San José Estépar, 2021).Moreover, the degree of attenuation in chest CT has the potential to visualize pressure gradients (Yamada et al., 2020), degree of inhomogeneity and has shown valuable in the diagnosis of diseases with airflow limitation, including chronic obstructive pulmonary disease (COPD) and emphysema (Arakawa & Webb, 1998;Heussel et al., 2009;Karimi et al., 2017;Mets et al., 2011;Schroeder et al., 2013).Previous studies have mainly focused on inspiratory imaging in smokers and patients with COPD.Mainly due to radiation considerations, inspiratory and expiratory chest CT studies in healthy nonsmoking subjects without respiratory symptoms, are scarce (Gawlitza et al., 2018) and reference values from never-smokers are lacking.
COPD ranks as the third most common cause of death globally and early diagnosis is regarded the most important factor for successful treatment (WHO, 2021).According to the Global initiative for Chronic Obstructive lung disease (GOLD) criteria, diagnosing COPD necessitates not only the presence of respiratory symptoms but also the identification of chronic airflow limitation (CAL) (Agustí et al., 2023).Chest CT has the potential to raise suspicion of COPD also when performed for other diagnostic purposes unrelated to COPD, for example, cancer or atheromatous screening (Bakker et al., 2021), this capability can be pivotal as it may prompt preventive measures to mitigate the further development of the disease or reduce the need for additional diagnostic investigations.
Currently, the ability to detect patients at risk is limited and studies of normality and early disease are highly important (Yip et al., 2021).
Our objective was to present pulmonary volumetric and radiodensity reference values from chest CT, at inspiration and expiration, in a population-based cohort of mainly healthy middle-aged individuals with the aim of uncovering potential early signs of CAL, even among asymptomatic individuals.Through identifying participants with CAL, COPD and emphysema, we aimed to analyze the potential for quantitative chest CT to identify early signs of obstructive lung disease.We used volumetric measurements, mean lung density (MLD) measures, percentage of low attenuation volume (LAV%), LAV cluster volume and effects from the gravitational pressure gradient to find a robust measure of disease.

| Research persons
From a local cohort of the Swedish CArdioPulmonary bioImage Study (SCAPIS), a cross-sectional prospective population-based multicentre study on men and women (Bergström et al., 2015), we randomly selected 150 men and 150 women, 50-64 years old at inclusion (Supporting Information: Figure S1).The SCAPIS study adheres to the ethical standards of the 1975 declaration of Helsinki and all research persons provided written informed consent to the data handling after ethical and radiation safety committee approvals.The local cohort included inspiratory and expiratory chest CT exams.Pseudonymized query reports, demographic and dynamic spirometry data were accessed through the main study.Based on results from dynamic spirometry after bronchodilation, subjects with forced expiratory volume during 1 s (FEV 1 ) divided by forced vital capacity (FVC) with results ≥0.7 (FEV%) were categorized into a non-CAL group (n = 258) and those with FEV% <0.7 (n = 36) were categorized into a CAL group.Subjects with CAL and respiratory symptoms (either dyspnoea, cough, mucus or wheezing) were classified as having COPD (n = 15) as defined by the GOLD criteria (Agustí et al., 2023).Furthermore, we implemented and compared the classification of CAL based on the lower limit of normal (LLN) (Pellegrino et al., 2005) which resulted in a CAL group of 24 research persons (non-CAL, n = 270).Reference values were drawn from spirometric data in NHANES III (Hankinson et al., 1999) and were adjusted for age, sex, and height, it was assumed that all research participants were of Caucasian ethnicity.A subset of eight participants displayed a restrictive pattern, characterized by a FEV 1 /FVC ratio exceeding 0.7 and an FVC measurement below the LLN, as established in NHANES III.Seven cases were classified as mild (FVC 70%-89% of expected) and one case was classified as moderate (FVC 60%-69% of expected).These cases were not initially excluded from the analysis; however, subsequent investigations were conducted to assess the impact of this subgroup on the study's findings.Importantly, none of the participants reported that they suffered from a severe pulmonary disease, neither obstructive nor restrictive.Current and previous smokers were distinguished from never-smokers through their smoking history, known in all but seven participants.Neversmoking non-CAL research persons (n = 134) were selected and used as reference group.

| Computed tomography of the lung (chest CT)
Chest CT scans were performed according to the SCAPIS protocol (Bergström et al. 2015), such that images were acquired at breath hold after maximal inspiration (INCT) and expiration (EXCT), with a BÄCKLIN ET AL.
| 341 dual-source CT scanner (Somatom Definition Flash; Siemens Healthineers), equipped with a Stellar detector and radiation optimization (4D-care dose and SAFIRE; Siemens Healthineers).Scanning was performed from apex to base in supine position, no contrast medium or bronchodilation was used.Participants received coaching and practiced achieving and maintaining full inspiration and expiration before the actual scans.The field of view was selected to include the whole lung without any cut-off.The calculated median effective radiation dose was 1.7 mSv for inspiration, and 1.5 mSv for expiration.
Experienced radiologists conducted a visual assessment of emphysema in connection to the examination, following a standardized protocol.In brief, emphysema was assessed in inspiratory chest CT images by dividing the left and right lung into three zones; upper (above carina), middle (below carina and above inferior pulmonary vein) and lower (below inferior pulmonary vein).Presence and severity of emphysema (mild, 1%-24%; moderate, 25%-50%; or severe >50%) were reported for each zone, as well as the types of pulmonary emphysema.

| PFTs
Dynamic spirometry (Jaeger Spirometer™; Vyaire Medical) was performed according to clinical routine in seated position, using a nose clip, 1-2 weeks before the chest CT examination and 15 min after inhalation of 400 µg salbutamol (Ventoline ® Evohaler with spacer; GlaxoSmithKline).FEV 1 and FVC were measured, both according to ATS/ERS standards (Macintyre et al., 2005;Miller et al., 2005) as described further in the SCAPIS study protocol.No static PFT was performed in SCAPIS, why sex-adjusted mean reference values for total lung capacity (TLC) and residual volume (RV) from literature (Hedenström et al., 1985(Hedenström et al., , 1986) ) were used for comparison.

| Quantitative analysis of chest CT
INCT and EXCT images were semiautomatically analysed with a commercially available software (Syngo.CT Pulmo 3D beta version VB10A; Siemens Healthineers) (Blechschmidt et al., 2001).Lung proportion of low attenuation volume in the lung (LAV%), the proportion of voxels with HU < −950 (Crossley et al., 2018) and (3) the total volume of clusters of low attenuation (LAV cluster volume).
Clusters were defined as connected image voxels (six-neighbour method) with a volume of at least 2 µL (volume corresponding to normal lung tissue), indicating pure air (Blechschmidt et al., 2001).
Additionally, chest CT's potential to noninvasively detect effects from the gravitational pressure gradient (ventral to dorsal in the supine position) was analysed at INCT and EXCT to evaluate whether differences in lung elasticity could be detected.Manually defined volumes of interest (VOI, size 5 ± 0.5 cm 3 ) were here placed at three different heights at the hilar level of each lung (ventral, middle and dorsal -Figure 1).For each height the mean of the minimum radiodensity HU values was calculated.

| Research persons
Demography and results from dynamic spirometry and emphysema diagnosis are presented in Table 1 where the GOLD criteria for CAL was used (flow chart in Supporting Information: Figure S1).Twelve percent (n = 36) of all participants had CAL, and 5% (n = 15) had COPD.Ten percent (n = 30) of the participants had emphysema but only 30% of those (n = 9) had signs of CAL.Most of the emphysema cases (83%, n = 25) were graded as mild.In the non-CAL group, 46% (n = 118) were previous/current smokers, compared to 58% (n = 10) in the CAL group.Using the LLN approach 24 of the 36 participants with CAL were classified as having CAL, and 5 of whom were classified as having COPD.Calculation with this approach as well as exclusion of eight participants with restrictive pattern did not significantly affect the reference ranges calculated, or the results from ROC analysis (data not shown).

| Quantitative analysis of chest CT
MLD was significantly decreased, while both LAV% and LAV cluster volume was increased in the CAL group, both at inspiration and expiration.INCT and EXCT lung volumes were larger in the CAL group while ventilated volume (INCT-EXCT), was similar (Table 2).
When values of 134 non-CAL never-smokers were compared to 118 non-CAL smokers, no significant differences were found (Supporting Information: Table S1).Lung volumes, ventilated volume and LAV all exhibited significantly larger numbers in men than in women, while MLD did not significantly differ between men and women (Supporting Information: Table S2).Correlation analysis for measured parameters to age and height showed moderate correlations only between height and inspiratory and expiratory lung volumes (men 0.59 and 0.58, women 0.57 and 0.49, respectively) (Supporting Information: Table S3).
When comparing to literature mean reference values for TLC, RV and FVC (Table 3), INCT volumes were 20% less than TLC for both men and women, EXCT volumes were 30% higher than RV for men and 40% for women, and ventilated volume by chest CT (INCT-EXCT) 40% less than FVC for men and 50% for women.
Correlation between individual FVC measured by dynamic spirometry and chest CT was 0.41 for men and 0.42 for women.
In the ROC analysis, the best results for identifying CAL by radiodensity were obtained when performed on expiratory images (Figure 2).LAV% and LAV cluster volume were both significantly higher in the CAL group (Table 2) and both gave an area under the curve (AUC) of 0.82 (Supporting Information: Table S4).With a cut-off value of 0.6% for expiratory LAV sensitivity was 72%, specificity 85%, PPV 40% and NPV 96% (true positives 26/36; true negatives 219/ 258) in this material where the pretest probability for CAL was 12%.
For predicting emphysema, the highest AUC value, 0.63, was obtained for expiratory MLD.ROC analysis with the LLN approach showed similar results (AUC differed ≤ 3%) for the measured parameters.
Analysing the pressure gradient using VOIs at different heights and their minimum radiodensity (Figure 3, interobserver variability T A B L E 3 Chest CT lung volumes (L) in non-CAL individuals divided by sex, compared to reference material from PFT (Hedenström et al., 1985(Hedenström et al., , 1986) ) for TLC, RV and FVC.1% for INCT and 3% for EXCT) with linear regression showed in the non-CAL group a significant dorsal to ventral slope, but only at expiration (p < 0.01).This indicates visualization of the gradient with lower HU levels (more air) in the ventral than dorsal part.No significant slope/gradient was registered neither at inspiration nor expiration in the CAL group, where instead an overall lower HU was found (p < 0.05).Differences in EXCT values tended to be larger for cases with higher radiodensity (Supporting Information: Figure S2).

| DISCUSSION
The The frequency of CAL in this population-based study, 12%, is similar to that expected from literature 10% (Backman et al., 2018).
As LLN-based categorization of CAL and non-CAL showed only minor deviations in AUC from ROC analysis, we chose the GOLD classification method.

| Reference volumetric and radiodensity values
Reference values for quantitative chest CT in non-CAL neversmokers are scarce, and we present values from a group of 134 research persons.Our results were not significantly different between smokers and never-smokers, which is in contrast to previous studies where non-and never-smoking adults were reported with lower inspiratory LAV% (1.1% [Hoffman et al., 2014] -2% [Zach et al., 2012]); however, our smoker group had a relatively low tobacco exposure.
The comparison of lung volumes between men and women revealed, as expected, that men had greater inspiratory and expiratory lung volumes, as well as ventilated volumes.However, it is noteworthy that none of these measurements reached the values obtained through seated spirometry.Radiodensity measures of low attenuation volumes (LAV%, LAV cluster volume) follow the same pattern with higher values in men, while MLD are similar in both groups (Supporting Information: Table S2).Correlations for age and height to the radiodensity measures were low, while lung volumes resulted in low correlation to age.The low correlation to age is explained by the limited age range in our cohort; however, the mean age is similar to the mean age of seated reference values of Hedenström et al.A moderate correlation to height was found for both inspiratory and expiratory lung volumes, with moderate correlation also to reference values with 55% significant lower estimated FVC.The supine position where a more cranial position of the diaphragm, especially in overweight subjects, may add to the volumetric differences (Allen et al., 1985;Mcmichael & Mcgibbon, 1939), as can the breath hold static manoeuvre in the CT, problems with contribution from research persons during CT scanning and sex differences, where females produce even less ventilatory volumes than men in the CT (Supporting Information: Table S3).None of the participants suffered from a severe pulmonary disease that may highly affect RV (Lutfi, 2017).

| Findings of CAL
The greater lung volumes observed during both inspiration and expiration in the CAL-group are consistent with the well-documented phenomenon of increased lung volumes associated with expiratory airflow limitation, hyperinflation and reduced elasticity (O'Donnell & Laveneziana, 2006).The radiodensity findings with larger LAV and lower MDL in the CAL group could only partly be related to the larger lung volumes detected and ROC analysis show that the parameters contain additional information about signs of disease.
The observed increased MLD in chest CT scans is associated with smoking-induced inflammation (Karimi et al., 2014).However, in our study, we were unable to replicate this finding in both inspiratory and expiratory scans when comparing non-CAL smokers and non-CAL never-smokers.It is important to note that our smoking group (n = 118) had a lower smoking burden (average of 17 pack years) compared to the study by Karimi et al.'s study (n = 40) with an average of 35 pack years.This discrepancy in smoking history might have contributed to our inability to detect a similar relationship, and it is possible that the statistical power of our study was insufficient to capture the effect.
Increased LAV% has been connected to emphysema (Gevenois et al., 1995;Hoffman et al., 2014;Vikgren et al., 2019;Zach et al., 2012), and moreover, we found significantly larger values of LAV% in the CAL-group, both at inspiration (6.5% against 4.2%) and expiration, even though the CAL-group (n = 36) comprised only nine cases of emphysema.Surprisingly, in a previous publication from the SCAPIS study (Vikgren et al., 2019) no LAV% differences were detected between subjects with (8%) and without mild emphysema.
Mild emphysema dominated the few cases in our study and it has been found that the percentage of emphysema-like tissue may vary substantially with demographic factors and body size also among healthy never-smokers, which may partly explain the differences (Hoffman et al., 2014).Already in 1999, Mishima et al. (1999) studied clusters of low attenuation areas and found that cluster analysis could help in diagnosis of early emphysema.Still, we are not aware of any recent study describing LAV cluster volumes in healthy individuals or in more severe stages of emphysema when a more regional/lobular focus is applied.

| Effects of gravity
The minimum radiodensity at the three different heights in the lung displayed a significant influence from gravitational effects within the non-CAL group during expiration but not at inspiration.However, this effect was not observed in the non-CAL group, neither at inspiration or expiration.Whether the observed effects are due to lower elasticity in emphysematous lung or an effect of increased air content associated with larger lung volumes remains to be determined.
Although the manual positioning of VOIs showed high interobserver reproducibility, a more regional and automatized non-biased analysis of the whole lung and its lobes is warranted for further studies of pressure gradients and regionality.

| Limitations of the study
Spirometry and chest CT scans were not performed on the same day and no bronchodilators were administered before chest CT scans.
The calculated lung volume differences between inspiration and expiration in chest CT predominantly corresponds to changes in air volumes, but the volumes do not solely represent the air within the lungs or belong to the same breath as determined by PFT.Without static PFT in the SCAPIS protocol, and without measurement of exhaled air during CT, for comparison we instead relied on reference values for TLC/RV/FVC and individual FVC results from spirometry.

| Concluding discussion
A maximal day-to-day variability of 10% in pulmonary function testing for vital capacity and 20% for testing FEV 1 was reported for healthy controls (Hruby & Butler, 1975).Moreover only 3% of individuals with normal spirometry responded to a bronchodilator (Hegewald et al., 2012), why the reference values presented should not be highly affected.In COPD, however, an 18% increase in mean FVC is reported after bronchodilation (Hegewald et al., 2012).Future studies are needed to evaluate if bronchodilatation before CT-scan in COPD may further strengthen the obtained differences between CAL and non-CAL groups.Additionally, supine position is known to yield about 10% lower values than seated FVC (Allen et al., 1985) why the poor correlation to seated spirometric results implies that supine breathing manoeuvres with arms up, may be suboptimal despite prior training.This has been shown before (Garfield et al., 2012) which may motivate 4D-acquisition during a whole breath with reduced radiation doses, as well as a erect CT (Yamada et al., 2020).
Eight participants with spirometric suspicion of restrictive disease (7 mild, 1 moderate) were included in the non-CAL group for reference value calculations and radiodensity measures.Excluding them did not change the results from the non-CAL group, why we considered their inclusion important for future identification of CAL/ COPD from any chest CT.However, given their limited number, these individuals were not subjected to further analysis in this study.
Except for additional expiratory CT-scans the SCAPIS protocol remained unchanged.As a consequence, presented reference values are applicable only for a narrow age range, and findings relevant to procedures performed.
The choice of CT equipment, software and parameters utilized during the scanning, acquisition, and reconstruction phases of a CT scan, such as the reconstruction kernel, do have a substantial impact on the characteristics of the resulting image (Midya et al., 2018).
Broader clinical studies and local validation of each setting are needed before clinical implementation.Mitigating solutions for standardizing lung density measures across equipment and parameters have been proposed (Chen-Mayer et al., 2017) and recently a method for standardizing lung density measures from CT was described (Abadi et al., 2023).

| Conclusion and summary
From the ROC analysis we concluded that the capability of quantitative radiodensity to identify persons with CAL performs best on expiratory images.Data from expiratory chest CT are however scarce and needs to be further studied.It seems easier to identify CAL than emphysema, but power was too low for definitive conclusions (few and mild emphysema cases).The high negative predictive values suggests that CAL and emphysema can be excluded with more than 90% certainty.Further investigations are warranted, not only for volumetric and radiodensity findings but also for establishing reference values in never-smoking individuals spanning various age groups and in larger cohorts.Additionally, there is a need for continued validation, particularly for expiratory images, where segmentation can sometimes be challenging.It is worth noting that successful segmentation of all five lung lobes was achieved in only 78% of the research participants in this study.Consistent with our findings, the potential of expiratory chest CT image analysis has been previously acknowledged (Gawlitza et al., 2018), yet data from expiratory CT scans are scarce.
In summary, we present volumetric and densitometric measures from inspiratory and expiratory chest CT examinations in a local cohort of 294 men and women, collected from the Swedish prospective cross-sectional population cohort SCAPIS.We found statistically significant differences between groups with or without CAL, while never-smokers and smokers presented similar results in the group without signs of CAL.Further, our findings suggest that automatic quantification of radiodensity from expiratory chest CT images holds promise in mitigating suspicions of CAL.Modern image processing tools capable of analysing the lung at the regional and lobular level could potentially enhance the utility of chest CT screening for the preclinical diagnosis of COPD.
volumes were automatically determined by delineation of the outer contour of the lung, excluding trachea, hilus and larger bronchi, and were visually checked, revised and accepted by an experienced radiologist.Ventilated volume from CT was calculated as INCT − EXCT and compared to FVC from spirometry, INCT to TLC and EXCT to RV reference values.By measuring radiodensity in HU the following quantitative parameters were collected from Syngo.CT and further analysed: (1) MLD, the mean radiodensity of all voxels; (2) Descriptive statistics were expressed as mean (SD) for normally distributed variables, otherwise median (interquartile range, X-Y) was used.Normal distribution was assessed based on visual evaluation of histograms.Comparison between groups was performed by independent sample t-test (Welch's t-test) for normally distributed variables and nonparametric Mann-Whitney U-test otherwise.Proportions were compared between groups using Fischer's exact test.Correlation was calculated for age and height, to lung volumes and radiodensity measures, for men and women separately using the Pearson's correlation coefficient.Reference values for lung volumes from chest CT were defined by calculating mean ±1.96 SD, that is, an interval expected to comprise 95% of the observations for a normally distributed variable.Receiver operating characteristic (ROC) curves were used to compare measured parameter's ability to identify CAL and emphysema.Pretest probability of CAL and emphysema was F I G U R E 1 A chest computed tomography image at inspiration (horizontal view from feet) with manually positioned spherical 5 cm 3 volumes of interest (VOIs; light grey circles) at three heights at each lungs hilar level, avoiding larger airways.Lobe segmentation is indicated by white lines and gravitational pressure with an arrow.calculated as the proportion of disease in the cohort, and positive and negative posttest probability was calculated as positive (PPV) and negative predicted value (NPV).Linear regression analysis was used for comparison of mean radiodensity in VOIs at different heights in the lung.Reproducibility was checked for VOI measurements, between the two observers, and presented as interobserver variability (standard deviation of the differences divided by the mean) and evaluated for bias in a Bland-Altman plot.The Spearman correlation coefficient was calculated for FVC from SCAPIS and ventilated volume (INCT-EXCT) from CT to describe the relationship between the two variables.Statistical analysis was performed using SPSS (IBM Corp. Released 2017; IBM SPSS Statistics for Windows, version 25.0.) and R (Version 4.0.2,R Core Team 2020).A p-value <0.05 was considered significant.
T A B L E 1 Demography and results from PFT in all non-CAL and CAL research persons; grouped for those with known smoking habits, n = 294.
Note: Values are mean (SD) or frequency (%).CAL, FEV 1 /FVC < 0.7; non-CAL, FEV 1 /FVC ≥ 0.7.Seven of the 294 research persons in the study did not answer query on smoking (6 non-CAL (3 male) and 1 female in CAL), out of these two research persons in non-CAL had emphysema.
development of chest CT imaging and analytic tools enables semiautomatic quantification of pulmonary volumes, volume change with ventilation, and how the pure air content is distributed in each lung.We present results from inspiratory and expiratory chest CT examinations in a cohort of 294 men and women, collected from a local cohort of the cross-sectional population-based cohort SCAPIS.
Radiation exposure in healthy never-smokers is scarce in literature and to our knowledge no previous chest CT reference levels have been described.With available chest CT images from healthy never-smoking individuals, we evaluated if early signs of disease could be identified using the new semiautomatic tools applied.From healthy middle-aged never-smokers, we calculated pulmonary volumetric, and densitometric reference values from chest CT and their potential to detect changes in pulmonary elasticity.