Dear Editor,

We read with interest McGroder et al’s study on the radiographic findings of patients four months after severe COVID-19 and the associated risk factors. Hürsoy and colleagues’ comment (1) on the paper was equally thought-provoking. We would like to further this discussion by contributing some of our observations from the pulmonology clinic at a major academic medical center in South East Asia.

It has been tremendously challenging globally to achieve precision in the diagnosis of Interstitial Lung Disease (ILD) post-COVID as invasive testing such as lung biopsies are performed sparingly. Histopathological pulmonary findings have largely remained inaccessible since COVID survivors are hypoxic so biopsies pose a high risk for the patient, and healthcare personnels are reluctant to perform such high-risk procedures. Hence, we are left to derive our diagnosis from radiological data and pulmonary function tests (PFTs) of the patient.

We propose that a consensus definition be reached for the diagnosis of post-COVID ILD, one that incorporates well-accepted radiological terms (used to represent any interstitial lung disease). We recommend that lung fibrosis only be classified as ILD if the lung parenchymal abnormalities persist for a minimum of six months after the COVID infection has resolved. Post-COVID ILD should then be further subclassified based on distinct radiological patterns. In our retrospective cohort study, four patterns of post-COVID lung parenchymal changes were exhibited by patients with no preexisting lung disease: persistent ground-glass opacities; interlobular septal thickening; reticulation and honeycombing; interlobular septal thickening and reticulations; and patchy consolidation with or without ground-glass opacity (2).

In terms of outcome, within our cohort of severe to critically ill patients, most developed persistent ground-glass opacities or patchy consolidation (with or without ground-glass opacity); a majority of these participants significantly improved (clinically and radiologically) upon administration of corticosteroids (2). Radiological improvement was defined as clearance of at least 50% of lung infiltrates. Reiterating our findings, Han et al. reported ground-glass opacities as the predominant pattern observed in follow-up CT scans of patients with fibrotic-like changes within six months of disease onset (3). Three of the five patients in our study who died had progressive disease with reticulation and honeycombing. To achieve greater accuracy in disease severity assessment and risk stratification, we suggest employing PFTs, particularly to obtain diffusion capacity of the lungs for carbon monoxide (DLCO) and forced vital capacity (FVC) measurements. For the eight out of thirty patients in our cohort who had significant HRCT findings, PFTs were ordered to assess physiological function; three presented with low FVCs (2). In Han et al’s study, abnormal DLCO (less than 80%) at 6-month follow up was a common occurrence in those with fibrotic-like changes (3).

Similar to McGroder et al’s findings, we concluded that the male gender was a significant predisposing factor for post-COVID pulmonary fibrosis. Moreover, like Li et al. and Han et al., we found diabetes mellitus and hypertension to be prevalent comorbidities in patients who developed fibrotic changes (2-4). Interestingly, in our cohort, disease severity did not significantly influence the development of any particular pattern of lung parenchymal abnormality. In light of these observations, we can conclude that the reported lung microstructure changes are not only a ramification of post-ARDS fibrosis or ventilator-induced lung damage but also a consequence of direct virus attack and aberrant local immune response. This is further evidenced by the finding that the Coronavirus induced fibrosis even in moderately ill patients who did not require invasive mechanical ventilation or ICU stay (2).

We hope that prospective studies will further enrich and broaden the global dialogue on post-COVID lung fibrosis.

References:

1. Hürsoy et al. 2021. e-letter https://thorax.bmj.com/eletters
2. Zubairi ABS, Shaikh A, Zubair SM, Ali AS, Awan S, Irfan M. Persistence of post-COVID lung parenchymal abnormalities during the three-month follow-up. Adv Respir Med. 2021;89(5):477–83. Available from: https://pubmed.ncbi.nlm.nih.gov/34612504/
3. Han X, Fan Y, Alwalid O, Li N, Jia X, Yuan M, et al. Six-month Follow-up Chest CT Findings after Severe COVID-19 Pneumonia. Radiology. 2021;299(1):E177-E186.
4. Li Y, Wu J, Wang S, Li X, Zhou J, Huang B, et al. Progression to fibrosing diffuse alveolar damage in a series of 30 minimally invasive autopsies with COVID-19 pneumonia in Wuhan, China. Histopathology. 2021;78(4):542-55.

We appreciate the thoughtful letter from Drs. Kanarek and Anderson. Our study does not address the well-established fact that asbestos exposure is the main causal factor of mesothelioma. The objective of our study was to investigate the risk of mesothelioma (and other asbestos related diseases) in motor vehicle mechanics. The key finding is that Danish motor vehicle mechanics do not on average have an elevated risk of mesothelioma during the studied up to 45 years of follow-up. This does not exclude the possibility that some subpopulations of motor vehicle mechanics with more extreme exposure/latency time are at increased risk – but this occupation as a group is not.

We agree that exposure misclassification is a potential problem in epidemiology studies based on occupation and industry titles. We also agree that lifetime asbestos exposure histories, if they could be obtained, might reduce exposure misclassification. However, asbestos exposure is often not recognized or recalled by workers, and workers often do not recall jobs in the distant past. Also, experts may misclassify self-reported jobs regarding asbestos exposure, particularly with respect to asbestos fiber type. Thus, while Drs. Kanarek and Anderson claim “obtaining an individual lifetime occupational and environmental exposure history is crucial to understanding individual work-related causes of disease” they offer no practical advice on how reliable asbestos exposure histories can be obtained. They also provide no evidence that such histories can differentiate between chrysotile and amphibole fibers, and provide no evidence that exposure misclassification from such histories would be less than from the methods we used.

We note that Dr. Anderson and colleagues recently published a study of mesothelioma incidence and mortality in Wisconsin (1) that identified asbestos-exposed occupations based on the self-reported “longest held job reported at the time of diagnosis” or the “usual industry and occupation” recorded on death certificates for mesothelioma cases. It was noted in that paper, that the authors were unable “to determine occupational exposure to asbestos that may have occurred from other work periods” and could not “identify nonoccupational exposure to asbestos.” Dr. Anderson acknowledged that “we would expect any bias introduced to be toward the null and lead to more conservative estimates of effect.” We believe that occupational histories in our study, which were based on historic, written employment records gathered in independent, population-based registers, were far more reliable than the self-reported usual occupations in Dr. Anderson et al.’s study.

There are now over 30 published studies of mesothelioma risks among automobile mechanics, only one of which found increased risk of mesothelioma in which the 95% confidence interval excluded the null, and which otherwise show no increased risk of mesothelioma among automobile mechanics. The consistency of these results observed by different persons, in different places, circumstances, and times (2) (some of which had lifetime occupational and environmental asbestos exposure histories) is a strong argument against the premise that exposure misclassification substantially biased the results.

Dr. Kanarek recently published a review paper in which he reported “There have not been definitive epidemiology studies of brake mechanics because of the nature of the workforce. It is generally nonunionized and spread out in car repair shops” and “Exposure to asbestos from brakes can occur to automobile or truck mechanics anywhere in a vehicle repair shop and the workers are highly transient and not documented.” Dr. Kanarek also noted there have not been “any prospective cohort studies conducted on a group of automobile mechanics.” (3). We believe our study, which is a prospective cohort study based on documented work in vehicle repair shops, answers many of his concerns.

Drs. Kanarek and Anderson point out that over 70% of our cohort of vehicle mechanics were first enrolled after 1986 when asbestos use in Danish vehicle brakes was disappearing. While this statement is correct, it ignores that 21,102 of our subjects were enrolled as mechanics in the 1970 nationwide census and that they were followed through 2015. Moreover, many of them were mechanics prior to the 1970 census. This is substantial number of people followed for a long time, regardless of whether they made up a small proportion of our entire study population. Selikoff’s landmark study of insulators (4) included only 17,800 workers, the majority of whom had yet to achieve 20 years from first exposure at the start of the study, and who were followed for only 10 years. Our population exceeded all these metrics: it was adequate to detect a substantial risk increase of mesothelioma, had there been one. Kanarek and Anderson recognize that our study was adequate to report a significant excess of asbestosis, even though there were fewer asbestosis deaths (19) than there were mesothelioma deaths (48) and the same numbers of incident cases of asbestosis cases (47) as mesotheliomas (47). This is again evidence that our study was adequate to find excess risks of both asbestosis and mesothelioma when they existed.

Drs. Kanarek and Anderson suggest that observing 313 cases of asbestosis in comparators supports the case for exposure misclassification in the comparison group. While it is possible that a small proportion of the 845,480 comparators (who functioned as 1,385,590 comparators after matching with replacement) may have had asbestos exposure at some point in their lives, the more appropriate conclusion is that the comparators were extremely unlikely to be diagnosed with asbestosis (incidence rate 1.19 per 100,000 person-years) or die of asbestosis (mortality rate 0.42 per 100,000 person years), and that not all cases of asbestosis (or mesothelioma) are caused by known exposure to asbestos. We believe a strength of our study was the choice of comparator occupations: we chose those that would be unlikely to have occupational asbestos exposure and that would have no overlapping skills with typical asbestos exposed jobs (such as in shipyards, construction, insulators, plumbing, etc.)

We acknowledge that asbestos is still used in several countries and that that these exposures should be eliminated. The findings in our paper cannot and should not be taken as an argument not to pursue this goal. We thank Drs. Anderson and Kanarek for facilitating a more thorough discussion of these issues.

References cited

1) Tomasallo CD, Christensen KY, Raymond M, Creswell PD, Anderson HA, Meiman JG. An Occupational Legacy: Malignant Mesothelioma Incidence and Mortality in Wisconsin. J Occup Environ Med. 2018;60(12):1143-9.
2) Hill AB. The environment and disease: association or causation. Proceedings of the Royal Society of Medicine. 1965;58:295-300.
3) Kanarek MS, Anderson HA. Mesothelioma from Asbestos Exposure in Brake Mechanics: Epidemiology in Context. Epidemiology (Sunnyvale). 2018;8(1):12.
4) Selikoff IJ, Hammond EC, Seidman H. Latency of asbestos disease among insulation workers in the United States and Canada. Cancer. 1980;46:2736-40.

Letter to the editor:
We appreciate the opportunity to comment on the article by Thomsen RW et al. Risk of asbestos, mesothelioma, other lung disease or death among motor vehicle mechanics: a 45-year Danish cohort study. We believe there are many problems in methodology and we disagree with author’s interpretations and conclusions especially in relation to asbestos and mesothelioma in vehicle mechanics in this article.

The epidemiology analysis described by Thomsen et al lacks asbestos exposure data and uses cross-sectional occupation data as surrogates for longitudinal use. Occupational categories are not equal to exposure. Especially for asbestos it has been clear that obtaining an individual lifetime occupational and environmental exposure history is crucial to understanding individual work-related causes of disease. Without longitudinal individual exposure histories in the Thomson et al study, there is undoubtably significant misclassification of exposure in both the motor vehicle mechanic group (unexposed considered exposed) and even more problematic in the control group (exposed classified as unexposed). This double likelihood of exposure misclassification creates unreliable analytics which result in an epidemiologic bias towards the null. 1

Thomsen et al used cross-sectional data at variable dates to place workers in their two study cohorts based on reported current occupation and industry. The occupation on the 1970 census or when first mentioned for individuals on later registrations was used to assign vehicle mechanic occupation with the assumption that asbestos exposure had occurred. The selected occupation categories for the “comparison” cohort assumed non-asbestos exposed occupations throughout their working lifetime. Other than looking for the vehicle mechanic category, apparently no search was made in subsequent registries to see if they had switched to another occupation with asbestos exposure likelihood. The authors do mention that about one half of the 21,102 1970 census vehicle mechanic individuals were in the same category in the 1996-1999 registry. However, was the work they performed the same? The cohort was very young at entrance at any of the time intervals (median age 25) and the median follow-up was 20 years. The age at which mesothelioma occurs is usually in the 70+ age group. Latency for mesothelioma can be as high as 50 years or more. Over 70% of the cohort were first enrolled after 1986 when asbestos use in Danish vehicle brakes was disappearing, exposure standards in place and the likelihood of high exposure for those just entering the cohort reduced. Only 23% of the cohort was deceased and the vast majority had not reached the age group where mesotheliomas or asbestosis would occur, given their long latent period.
The overall finding of Thomsen et al was that asbestosis was elevated in motor vehicle mechanics, but not mesothelioma. A clinical diagnosis of asbestosis requires a history of asbestos exposure while at autopsy a tissue diagnosis may be made when no exposure history is found. What is striking is that there were 313 comparator cases of asbestosis in the morbidity data and 108 in mortality. This information supports the case for exposure misclassification in the comparison group. Similarly, the occurrence of 553 mesothelioma/pleural cancer mortality cases among the comparators and 628 cases in the morbidity files also suggests misclassification among the “non-asbestos exposed” as an explanation for the lack of statistical differences between the cohorts.

The high number of asbestosis cases and deaths among the comparison group and the statistical excess among the vehicle mechanics but not an excess of mesothelioma/pleural cancer is puzzling as a higher dose of asbestos exposure is needed for asbestosis than for mesothelioma.2 An explanation may be that the years of follow-up in the Thomsen et al study may not have been long enough for the peak mesothelioma risk but sufficient for clinical detection of asbestosis/pleural abnormalities. Supporting this is that Thomsen et al found the highest association with mesothelioma in the 1970 census cohort, which had the longest follow-up.

Thomsen et al. states that asbestos use in brakes is now phased out in most countries. Unfortunately, this is not true. Chrysotile asbestos is currently being widely used in Brazil, Russia, China, India, Thailand, Malaysia and elsewhere and is being used in the manufacture of brakes. (2-5) Since there is potential exposure to toxic chrysotile asbestos fibers of various sizes in the dusty environment of motor vehicle mechanics, which is continuing in several countries around the world, precaution should point to policies that concentrate on industrial hygiene measures that limit worker dust exposure.

Marty S Kanarek, PhD1, Henry A Anderson, MD1

1 Department of Population Health Sciences and in the Nelson Institute for Environmental Studies, University of Wisconsin-Madison, Madison, Wisconsin, USA

Correspondence to Professor Marty Kanarek, Department of Population Health Sciences, School of Medicine and Public Health, 610 N. Walnut Street, University of Wisconsin-Madison, 53726; mkanarek@wisc.edu

Contributors MSK was the lead author. HAA was co-author, reviewing the original response and contributing original content in addition to editing. Final version was approved by both authors.

Funding None

Competing interests Both Kanarek and Anderson have served as consultants to government and international agencies on asbestos health effects, and have been consultants and witnesses on plaintiff’s litigation concerning asbestos and disease.

References

1. Checkoway H, Pearce N, Kriebel D. Research methods in occupational epidemiology. Monographs in Epidemiology, 2nd Ed. 2004, Oxford U Press.

2. Kanarek, MS, Anderson HA. Mesothelioma from asbestos exposure in brake mechanics: epidemiology in context. Epidemiology: Open Access 2018;8:2. DOI. 10.4172/2161-1165.1000340.

3. Kunpeuk W, Sataporn J, Mathudara P1, Jeerapa S1, et al
A scoping review on occupational exposure of silica and asbestos among
industrial workers in Thailand. Outbreak, Surveillance Investigation and Response OSIR Journal , 2021, Volume 14, Issue 2:.41-51.

4. Chen T, Xiao-Ming S, Wu L. High time for complete ban on asbestos use
in developing countries. JAMA Oncology 2019; May 23 E1-E2.

5. Omar A, Lamin F , Mohamed N. Comparative study of brake pads in Malaysian
automotive aftermarket. International Journal of Crashworthiness 2016;
http://dx.doi.org/10.1080/13588265.2016.1221372.

To the editor,

We thank N. Hürsoy and colleagues for their interest in our study of patients four months after severe COVID-19 [1]. We agree that there needs to be continued development of terms describing the radiographic appearance of post-COVID fibrotic-like patterns. We acknowledge that without the benefit of histopathology or serial imaging, our ability to define pulmonary fibrosis is limited.

The authors posit that parenchymal bands, irregular densities, and ground glass opacities, may be considered fibrotic-like patterns. We have included irregular densities, characterized as reticulations or traction bronchiectasis, as fibrotic-like changes. We did not include parenchymal bands [2], as these can be associated with atelectasis, which is common in COVID and can disappear over time [3]. Similarly, we did not include isolated ground glass opacities as fibrotic-like changes, as these have been found to decrease over time in CT lung cancer screening cohorts [4] and in other post COVID-19 cohorts [5, 6].

A priori, we evaluated for both previously established interstitial lung abnormality categories [7], as well as categories of radiographic abnormalities reported in Acute Respiratory Distress Syndrome (ARDS) survivors using an established scoring system [8]. This inclusive approach should facilitate meta-analyses and comparisons with future studies of COVID-19 survivors, interstitial lung disease studies, and studies of non-COVID ARDS survivors. Furthermore, it allows for future post-hoc analyses if alternate definitions of fibrotic-like patterns in COVID-19 survivors are established. Additionally, we showed that objective quantitative analyses closely agreed with visual assessments (Figure S2). These types of quantitative imaging analyses may facilitate the convergence of data from multiple centers if imaging protocols become more standardized [9].

Efforts are underway to characterize pulmonary impairments and radiographic abnormalities in our cohort over time in order to assess longitudinal evolution. We acknowledge that our findings do not exclude the possibility of pre-existing lung disease and we therefore look forward to reviewing independent studies, such as the Collaborative Cohort of Cohorts for COVID-19 Research (C4R) [10] project, which will provide better understanding of radiographic changes by comparing chest imaging studies before and after SARS-CoV-2 infection.

1. McGroder, C.F., et al., Pulmonary fibrosis 4 months after COVID-19 is associated with severity of illness and blood leucocyte telomere length. Thorax, 2021.
2. Pulmonary Parenchymal Band. Available from: https://www.ncbi.nlm.nih.gov/medgen/978776.
3. Kong, M., et al., Evolution of chest CT manifestations of COVID-19: a longitudinal study. J Thorac Dis, 2020. 12(9): p. 4892-4907.
4. Jin, G.Y., et al., Interstitial lung abnormalities in a CT lung cancer screening population: prevalence and progression rate. Radiology, 2013. 268(2): p. 563-71.
5. Nagpal, P., et al., Case Studies in Physiology: Temporal Variations of the Lung Parenchyma and Vasculature in Asymptomatic COVID-19 Pneumonia: A Multi-Spectral CT Assessment. J Appl Physiol (1985), 2021.
6. Liu, D., et al., The pulmonary sequalae in discharged patients with COVID-19: a short-term observational study. Respir Res, 2020. 21(1): p. 125.
7. Hatabu, H., et al., Interstitial lung abnormalities detected incidentally on CT: a Position Paper from the Fleischner Society. Lancet Respir Med, 2020. 8(7): p. 726-737.
8. Burnham, E.L., et al., Chest CT features are associated with poorer quality of life in acute lung injury survivors. Crit Care Med, 2013. 41(2): p. 445-56.
9. Nagpal, P., et al., Quantitative CT imaging and advanced visualization methods: potential application in novel coronavirus disease 2019 (COVID-19) pneumonia. BJR Open, 2021. 3(1): p. 20200043.
10. Collaborative Cohort of Cohorts for COVID-19 Research. Available from: https://c4r-nih.org/content/overview.