The fact that routinely effective treatments for disseminated lung cancer are not available has prompted the search for effective early detection systems. It is important to identify lung cancer while it is still confined to the bronchial epithelium and is potentially curable with local modalities. We have previously reported on an immunologically based assay to identify antigens expressed on shed bronchial epithelial cells. This assay resulted in a statistically significant correlation of immunostaining with the eventual development of lung cancer 2-4 years prior to routine clinical detection. Attempts to further improve this approach require an understanding of the basis for its success. Based on the work of Hakomori and coworkers, this difucosylated Lewis X structure would be a likely marker of carcinogenic transformation of the bronchial epithelium. In fact, an antibody to this structure was useful for sputum immunocytochemistry analysis for early lung cancer detection. Other carbohydrate structures would also be reasonable markers to evaluate for early detection application, based on the known pattern of expression of these structures in fetal, dysplastic, and neoplastic lung tissue. Another antibody used for sputum immunostaining recognizes a 31-kd protein structure; the antibody is not a known member of a likely class of early detection targets. The reported cases of lung cancer missed by the immunostaining approach included principally adenocarcinoma of the lung, suggesting that the addition of a marker(s) of that type of morphologic differentiation should be considered. Markers to dissect the various forms of lung adenocarcinoma are being characterized and are available for evaluation in early detection applications.(ABSTRACT TRUNCATED AT 250 WORDS)