|Home | About | Journals | Submit | Contact Us | Français|
Although the relation between radiographic abnormalities and spirometric impairment in people with asbestosis has been studied extensively, the extent of spirometric impairment associated with milder radiographic abnormalities is not established.
To test associations between mild radiographic abnormalities and Lower Limit of Normal (LLN)-based spirometry interpretation.
Spirometry and CXRs were collected for 1,026 at low risk of exposure to pneumoconiotic agents participants in a medical screening program.
Individuals with each type of isolated or combined International Labour Organization (ILO) abnormalities had up to over sixfold statistically significant increase in odds of LLN-based restrictive pattern physiology (OR = 1.96, 95%CI 1.03–3.73 for parenchymal to OR = 6.09, 95%CI 1.94–19.10 for parenchymal and pleural) compared to those with normal films.
The findings from this study confirm the association of mild profusion abnormalities with clinically relevant, LLN-based lung function abnormalities.
The International Labour Organization’s International Classification System of Radiographs of Pneumoconioses (ILO system) is an epidemiologic tool for systematic radiologic evaluation of diffuse, exposure-related injury to lung parenchyma and/or pleura. Assessment of its validity has historically been based on correlations with various measures of lung function and exposure among workforces at high risk for pneumoconiosis.1–8 Cumulative exposure has been associated with lung function decline and spirometric impairment in asbestos workers with abnormal ILO profusion scores, severe fibrosis, focal and diffuse pleural abnormalities, and functional evidence of airways disease.9–11 The correlation of categorical characterization of lung function abnormality in relatively unexposed workforces with subtle radiographic abnormalities, corresponding to low grade ILO profusion scores has been less extensively studied.
Spirometric functions, such as forced vital capacity (FVC%Pred), correlate inversely with ILO profusion scores and pleural fibrosis in asbestos exposed workers.1–6,8 Characterizations of spirometry vary depending on the reference values and interpretation protocols used.12–15 Lower limit of normal (LLN) values have been proposed as an improvement to the previous spirometric gold-standard, %Pred protocol, to correct for “over-diagnosis” of obstruction in older subjects.16 Data are lacking on correlations between low grade fibrosis using the ILO scoring system and spirometric abnormalities defined by the LLN protocol.
This study was conducted to evaluate the physiologic and clinical significance of subtle abnormalities of parenchyma and pleura on screening chest x-rays (CXR) using LLN-based characterization of spirometry. Associations between ILO profusion scores and LLN spirometry results were compared with associations between ILO profusion scores and FVC%Pred values, in a cohort of former atomic weapons workers with minimal or bystander exposure to known fibrogenic agents.
Recruitment and methods for the screenings were previously described.17 Former nuclear weapons’ workers from an assembly plant and a federal research laboratory were invited for spirometry, posteroanterior chest X-ray, and non-fasting blood panels. Testing was offered on a three-to-five-year basis with most recent results used for analysis.
Written informed consent was collected from participants before screening. A modified American Thoracic Society (ATS) adult respiratory questionnaire18 was used to obtain basic demographic information, height, weight, and smoking history with body mass index (BMI) calculated according to published methods.19
Chest films were reviewed independently, by three experienced A-readers, independently, according to the most recent ILO guidelines20 without knowledge of individual clinical status or exposures. A-readers were employed to interpret the films given no access to B-readings by this study. Multiple readings were reconciled using median profusion scores for small opacities and two out of three consensus for pleural abnormalities. The profusion scores were collapsed into six categories for statistical analyses: 0 = 0/0; 1 = 0/1; 2 = 1/0; 3 = 1/1; 4 = 1/2; 5 = 2/1+, similar to the scheme proposed by Miller et al.6 Abnormal profusion was defined as ILO score ≥ 1/0. The ILO CXR categories used in the analysis in this study were mutually exclusive i.e. parenchymal vs. parenchymal with pleural vs. pleural.
Spirometry was done pre-bronchodilator in accordance with ATS guidelines.21,22 and calibration of testing equipment on a daily basis as recommended.21,22 An effort was made to obtain at least three acceptable and reproducible results, but no test was rejected based on the lack of three results.23
The Third National Health and Nutrition Examination Survey (NHANES III) equations with race corrections were used to calculate FVC%Pred and LLN reference values for FVC, FEV1, and FEV1/FVC%.24,25 Categorization of LLN results was based on a published algorithm26 with low FEV1/FVC% (< LLN), low FEV1 (< LLN), and normal FVC (≥ LLN) described as obstruction; normal FEV1/FVC%, low FVC and normal or low FEV1 interpreted as a restrictive pattern physiology; and low FEV1/FVC%, FVC, and FEV1 characterized as possible mixed obstructive airways and restrictive pattern physiology impairment. Individuals with low FEV1/FVC% but normal FEV1 were recommended for follow-up evaluation and their results were categorized as normal for the purposes of analyses.
Exposure to known fibrogenic agents such as asbestos, beryllium, and metal dusts was assessed qualitatively, by site and for the workers’ population at large, and used the previously described methods to inform this process.17,27 Industrial hygiene teams reviewed available industrial hygiene records, job titles, job descriptions, and area monitoring records to conclude that majority of screened workers were at low risk of exposure given their work in an assembly of pre-manufactured components and/or research and development. These worksites have been characterized as relatively clean by the workers with the greatest dust exposure among the weapons assembly workforce being to trinitrotoluene (TNT)-derived explosives dusts among a subset of production workers and to lanthanide metals dusts among a small subset of the research laboratory workforce. No individual exposure assessments were conducted for the purposes of this study.
SAS 9.2 statistical software was used to analyze de-identified data.28 Continuously distributed variables such as age, BMI, and FVC%Pred were described using mean, standard deviation, and ranges. Differences in medians of these covariates by smoking history, (never versus current and ex-smokers combined), were tested using the Wilcoxon rank-sum test. The Cochran–Armitage chi-square test was used to test the hypothesis of no difference in trend of ILO profusion score groupings between never and ever-smokers. Differences in the distribution of abnormal CXR ILO readings were tested using Pearson’s chi-square test, and Fisher’s exact chi-square test was used to test differences in distribution by gender.
Multivariable generalized logistic regression was used to model associations between radiographic abnormalities and spirometry. Separate models were fit using the LLN and %Pred protocols as dependent variables. The LLN spirometry results were characterized into normal (reference), restrictive, obstructive, and mixed results categories. The FVC%Pred values were stratified into normal (reference) 80%+, mild 79–70%, moderate 69–50%, and severe impairment < 50% categories. ILO readings were evaluated for associations with abnormality by both spirometry protocols in two fashions: as normal, isolated parenchymal or pleural fibrosis, or both, and separately by six profusion score categories. Smoking status and BMI were controlled for in every model.
A P-value of < 0.05 was considered statistically significant in all analyses.
The University of Iowa Institutional Review Board (ID 200008081 and 200509719) and the Department of Energy Central Beryllium Institutional Review Board (ID 209956) issued approvals for this study.
Spirometry results and ILO CXR readings for 1,026 individuals were analyzed. Characteristics of this population with never smokers compared to ever-smokers are given in Table Table11.
Gender was statistically significantly associated with smoking status with 85.1% of ever smokers being male compared to 59.1% never smokers. Current and ex-smokers were slightly older, on average, than never smokers. Positive smoking history was associated with lower mean FVC%Pred and FEV1/FVC%. Over 40% of ever-smokers had an abnormal LLN spirometry result with 4.7% having obstructive airways, when compared to 25% of overall abnormal LLN spirometry results and 1.2% of obstructive airways specifically among never smokers. Isolated parenchymal abnormalities were found in 5.2% (n = 53) subjects, 1.8% (n = 18) had concurrent parenchymal and pleural fibrosis, and 4.4% (n = 45) isolated pleural fibrosis.
The distributions of ILO abnormalities, profusion scores, and age by categorical spirometry results and mean FVC%Pred values are presented in Table Table22.
There were 253 (24.6%) individuals with restrictive pattern physiology using the LLN protocol, 33 (3.2%) with obstruction, and 75 (7.3%) with mixed impairment. LLN-based spirometric abnormalities were each statistically and significantly associated with abnormalities of parenchyma and/or pleura before adjusting for confounders. The majority (56%) of abnormal profusion scores were low grade 1/0, with 1/1 score being the second most common abnormal ILO profusion reading (20%).
Table Table33 summarizes the generalized logistic regression modeling of associations between spirometry and radiographic abnormalities.
Isolated and combined pleural and parenchymal abnormalities were each associated with LLN-based restrictive pattern physiology controlling for smoking and BMI. Concurrent parenchymal and pleural fibrosis was associated with a sixfold increase in odds of restrictive spirometry compared to normal radiography. Concurrent parenchymal and pleural fibrosis was associated with a sevenfold increase in odds of mixed spirometric impairment compared with normal radiography. Individuals with concurrent parenchymal and pleural fibrosis had a statistically significant nearly ten-fold increase in odds of FVC between 50 and 69% of predicted, while those with isolated pleural disease had a nearly fourfold increase in odds of FVC between 50 and 69% of predicted, compared to those with normal films.
Results of generalized logistic regression modeling of associations between spirometric abnormalities and fibrosis profusion scores are summarized in Table Table44.
LLN-based results were associated with fibrosis, with individuals with 0/1 profusion score having an over two times increase in odds and those with a 1/2 score an up to eight times increase in odds of restrictive pattern physiology, respectively, compared to those with a 0/0 reading; however there were only four (n = 4) subjects with restrictive pattern spirometry and 1/2 profusion score reading and the confidence interval was wide (95% CI 1.36–43.27). In addition, individuals with a 1/2 profusion score had a nearly fifteen-fold increase in odds of mixed impairment on spirometry compared to participants with normal films adjusting for smoking and BMI and all confidence intervals differing from one, but again there were only three subjects (n = 3) identified with these readings (95% CI 2.09–100.89). Low-grade fibrosis scores of 0/1 were associated with an over twofold increase in odds of mild and moderate restriction, (FVC%Pred 79–70% and 69–50%), scores of 1/0 associated with an over threefold increased odds of severe restriction (FVC%Pred < 50%), scores of 1/1 (n = 1) associated with a nearly fivefold increase in odds of moderate restriction (FVC%Pred 69–50%), and profusions scores of 1/2 (n = 3) associated with an over eighteen-fold increase in odds of moderate restriction (FVC%Pred 69–50%).
The association of radiographic evidence of pulmonary fibrosis with lung function abnormalities has been reported in populations highly exposed to asbestos.1–8 These studies found ILO abnormalities on CXR to be negatively correlated with spirometric parameters, specifically FVC%Pred, and this correlation was found for pleural disease alone and more pronounced for pleural coincident with interstitial changes. Our study confirmed these findings for all types of ILO abnormalities in a population with minimal to negligible exposures to asbestos, and other fibrogenic dusts and comparatively low-grade radiographic changes. Parenchymal, isolated pleural and parenchymal coincident with pleural fibrosis were each associated with lower FVC%Pred values and with restrictive physiology using the LLN protocol.
These results are significant for both clinicians and epidemiologists. The association of decrement in pulmonary function with pleural abnormalities could reflect occult interstitial changes too subtle to be detected with conventional chest films. The increase in the risk of parenchymal coincident with pleural abnormalities was up to three times as high compared to isolated parenchymal or pleural disease, but only for moderate to severe FVC%Pred impairment. The ILO profusion score is at present the most consistently agreed upon system for characterizing type and severity of pneumoconiosis, and may have relevance to pulmonary fibrosis in general. The study population was an elderly group of former weapons assembly and research and development workers who, for the most part, based on the qualitative IH assessment were not heavily exposed to beryllium, asbestos, or silica.17,27 The etiology of observed fibrotic changes is likely to represent a mix of age and undetermined exposures. These findings confirm ILO system as a valid tool in the detection of clinically relevant pulmonary abnormalities resulting from exposures at work.
Multiple authors reported finding no clear distinction in physiological measurements between ILO 0/1 and 1/0 profusion scores, the conventional cut-off for normal versus abnormal parenchymal findings.1,4,8,29 Our study did not find any significant difference in odds of abnormal lung function in individuals across these profusion scores; the odds of restrictive airways or lower FVC%Pred values in those with 0/1 and 1/0 readings were, however, consistently twice as high, though 1/0 and FVC 69–50%Pred not significant, as those with completely negative, (0/− and 0/0), readings but the numbers of individuals with these readings were low resulting possibly in inability to detect the difference. The significance of this finding may be, as other authors suggest, related to minimal and radiographically non-distinguishable abnormalities associated with fibrogenic dust exposures.6 This is also consistent with a lack of an objective or absolute cut-off point for discrimination of normal and abnormal lungs using conventional chest radiography. These findings imply the need for more rigorous radiographic evaluation of individuals with 0/1 and 1/0 scores, especially when associated with physiologic impairment.
The biggest increase in odds of abnormal lung function result in our study was noted for individuals with 1/2 ILO profusion scores, with a subsequent sharp decrease noted in scores of 2/1 and greater. The 1/2 score is considered by some to be the most clinically specific cut-score for normal vs. abnormal parenchymal readings.8 This finding is consistent with other studies.6 It should be noted that this increase in odds of association of lung function abnormality with abnormal vs. normal films (i.e. 0/− and 0/0) was found not only for lower FVC%Pred values but also for categorical spirometry results, specifically restrictive physiology and mixed obstructive airways and restrictive pattern physiology, using the LLN. To our knowledge, this is the first study examining correlations between abnormal ILO profusion score and LLN-based categorical spirometry results.
Previous reports have shown correlation between smoking and ILO profusion readings, considered the “dirty chest” issue.30–32 While questions have been raised regarding the methodology used in some of these studies33,34, our study showed a statistically significant increase in rates of abnormal ILO readings, both for parenchymal abnormalities and profusion scores with the exception of category 1/1, in former and current smokers compared to never smokers. Detailed smoking history was not available to further explore this association, but these findings certainly warrant the need for further studies and recommendations for ILO readers with regards to the significance of the smoking history. As described by other authors chest computed tomography (CT) scans likely provide both greater sensitivity and specificity for lung pathology, including fibrosis and airways disease, in subjects with lower profusion scores.35–37 At present, there is no accepted ordinal scale for degree of abnormality on CT scan, there are few large population studies assessing the significance of CT findings and ILO scores in these populations, and inter-reader variability in chest CT readings continues to be problematic.38 Further study with more advanced imaging techniques is warranted to explore the significance of smoking and clinically relevant pulmonary function abnormalities in subjects with low-grade ILO abnormalities on CXR.
Over one-third of spirometries in this study were characterized as abnormal using the NHANES III LLN-based reference values and spirometry interpretation protocol, with 3% of subjects demonstrating obstructive airways, 25% restrictive pattern physiology, and 7% mixed obstructive and restrictive impairment results. This finding is consistent with the 12.5% prevalence estimates of obstructive airways impairment in the NHANES III 2007–2008 population testing,39 based on the <FEV1/FVC LLN ratio as the only criterion to characterize obstruction and inclusive of mixed impairment.26 Using the same %Pred protocol to characterize abnormal spirometry showed differences in the estimates of restrictive pattern physiology between the two studies, the mean age of workers included in this study was, however, 68.7 (±11.4) years with 10% of individuals over the age of 80 compared to 44.8 (±0.4) years and 20 to 79 years old individuals included in the NHANES III study. As lung volumes were not available for either study, the correct interpretation of these results remains unknown.
This study employed three experienced ILO NIOSH A-readers with an inter-reader agreement in ILO categorical results ranging from moderate to substantial, as reported previously in the study of the subset of this population.27 The agreement in profusion scoring was also substantial between all three readers. Such high degree of agreement is certainly helpful in any type of validation studies involving ILO-guided interpretation of screening radiographs, it is however unknown, without the detailed exposure information or gold standard radiographic testing, whether this may have led to under- or overestimate of lung disease rates in this population.
The findings from our study confirm the utility of ILO scale, specifically in the range of minimal category 1 profusion abnormalities, in detection of clinically relevant abnormalities using the LLN-based spirometry interpretation.
No potential conflict of interest was reported by the authors.
Financial Disclosures: This work was supported by The US Department of Energy, Award No: DE-FC01-06EH06020. This study did not receive funding from any of following organizations: National Institutes of Health (NIH), Wellcome Trust, Howard Hughes Medical Institute (HHMI). This study did not receive any pharmaceutical or industry support.
This study would not have been possible without the participation of former workers from both DoE sites under study. We remain thankful to those workers as well as DoE staff, Mary Fields, Greg Lewis, Isaf Al-Nabulsi, Moriah Ferullo, Regina Cano, Libby White, and Dr. Patricia Worthington for their support toward the program. We would like to thank the University of Iowa team; Nicholas Hoeger, Christina Nichols, Jill Welch, Rick Paulos, and Dr Valentina Clottey for their ongoing contributions to the screening program.