Am. J. Respir. Crit. Care Med., Vol 154, No. 3, 09 1996, 741-748.
Radiographic and pathologic correlation of coal workers' pneumoconiosis
V Vallyathan, PS Brower, FH Green and MD Attfield
Division of Respiratory Disease Studies, National Institute for Occupational Safety and Health, Morgantown, West Virginia, USA.
The relationships between chest radiographs (CXR) and corresponding
pathology were investigated in 430 autopsied coal miners from West
Virginia. Whole-lung sections were reviewed and graded on four-point
severity scales for the following lesions of coal workers' pneumoconiosis
(CWP): macules, micro- and macronodules (small and large fibrotic nodules),
and progressive massive fibrosis (PMF). Antemortem CXR were classified by
three B readers using the 1971 International Labor Office (ILO) U/C
classification (6). On pathologic examination, 96% of miners had macules,
70% micronodules, 45% macronodules, 15% silicosis, and 28% PMF. By CXR, 69%
of the miners had small, rounded opacity profusions of category > or =
0/1. Data analysis revealed increasing odds that small opacities of
category > or = 0/1 would be detected with increasing grade of nodules.
Profusion category 0/0 was often reported for cases with macules of mild to
moderate grade and mild levels of micronodules. Overall, q-type opacities
were associated with macules and micronodules, whereas the large r-type
opacities were associated with macronodules. By CXR, large opacities showed
good correlation with pathologic PMF. However, about one-third of cases
identified as having large opacities by CXR were not substantiated as PMF
by pathology. One-fourth of these cases could be explained by lung lesions
such as Caplan's nodules, tuberculosis scars, and tumors. Similarly, 22% of
cases classified as PMF on pathology had no large opacities by CXR. In half
of these cases, the radiologists had noted other abnormalities (cancer,
tuberculosis) by CXR as large opacities. Overall, the study showed good
agreement (Somer's d = 0.64) between the predicted probabilities and
observed responses of a profusion category > or = 0/1 for pathologic CWP
lesions. However, the study also showed that CXR were insensitive for
detecting minimal CWP lesions, and were unreliable indicators in the
presence of concomitant pulmonary pathology.
