We initiated our Early Lung Cancer Action Project (ELCAP) in 1992 to assess the usefulness of a regimen of CT screening for lung cancer, and we have reported the frequency of a positive result of the screening test and of the diagnosis of malignancy at both baseline and annual repeat screenings [1, 2]. A positive finding on the screening test was defined as detection of between one and six noncalcified nodules at baseline screening; and at repeat screening, detection of between one and six new noncalcified nodules showing growth since the previous screening.

All focal nonlinear densities were counted as representing nodules, and we classified a nodule as noncalcified if it failed to meet our criteria for a benign noncalcified nodule [1, 2]. Among these noncalcified nodules, we found solid nodules, which we defined as nodules that completely obscure the lung parenchyma within them. We also found nodules that did not completely obscure the entire lung parenchyma within them, commonly called “ground-glass opacities” [3,4,5,6,7,8]. We wanted to use the data from a CT screening program to address the frequency of occurrence of these different types of nodule and the frequency with which they represent malignancy.

We also wanted to introduce a different terminology for ground-glass opacities because we believe that this term is not appropriate in the context of CT screening for lung cancer. The term “ground-glass” was introduced to describe—and is still applied to—not only focal abnormalities but diffuse ones as well, so that, for our purposes, the adjective “focal” needs to be added. We also prefer to use the anatomic term “nodule” rather than “opacity” because the former is in line with other anatomic terms generally used in chest radiology, such as “bronchi” and “blood vessels.” Accordingly, in this article, we distinguish between solid and subsolid nodules. The former was defined as a nodule that completely obscures the entire lung parenchyma within it whereas the latter does not. A subsolid nodule can be further classified as either part-solid (in the case of a nodule with patches of parenchyma that are completely obscured) or nonsolid (in the case of a nodule with no such areas).

This study was based on the ELCAP experience with persons undergoing baseline and annual repeat CT screening for lung cancer [1, 2]. Among the 233 instances of positive findings on baseline CT scans, 159 of the findings involved only a single noncalcified nodule, 43 of the positive findings involved two noncalcified nodules, 16 positive findings involved three such nodules, seven positive findings involved four nodules, six positive findings involved five nodules, and two involved six nodules. Among the 30 instances of a positive finding on annual repeat CT scans, 28 involved a solitary nodule of the relevant type, one involved two nodules, and the remaining finding involved three nodules.

We reviewed all these instances of positive finding, classifying each nodule as to its type—solid, part-solid, or nonsolid—on the basis of the nodule's appearance on the high-resolution CT scans obtained after the identification of the nodule on low-dose CT. The high-resolution CT scans of the nodules were obtained with a targeted field of view (9.6 cm), 1-mm collimation with overlapping 0.5-mm reconstructions, and high-spatial-frequency algorithm. A nodule was classified as part-solid if it contained patches that completely obscured the lung parenchyma (Figs. 1 and 2) or had spiculations (Fig. 3). A nodule was classified as nonsolid if none of the lung parenchyma within it was totally obscured (Figs. 4 and 5). In making the distinction between a part-solid and a nonsolid nodule, we did not regard blood vessels as solid components despite their solid appearance (Fig. 5). Figure 6 shows a solid nodule for comparison.

Chi-square statistics with one degree of freedom were used for comparisons of the frequency of malignancy among the three types of nodule. The size-adjusted frequencies were obtained by deriving mutually standardized rates with the weighting proportional to comparative information.

Among the 233 positive findings on the baseline CT scans, 189 (81%) involved only solid nodules, 40 (17%) involved only part-solid or nonsolid nodules, and four (2%) involved a combination of both part-solid and nonsolid nodules.

The distribution of the positive findings by size of the (solitary or otherwise) largest nodule is shown in Table 1, grouped by nodule type. The median diameter of the largest nodule was 3 mm for the solid nodules, 10 mm for the part-solid nodules, and 7 mm for the nonsolid nodules. As this table indicates, part-solid nodules were less common than nonsolid ones. The largest nodule was part-solid in 16 instances and nonsolid in 28 instances.

The prevalence of malignancy by the size and type of the largest nodule is shown in Table 2. The prevalence of malignancy increased markedly as the size of the largest nodule increased and was distinctly higher among the part-solid and nonsolid nodules combined (15/44, 34%) than among solid nodules (14/189, 7%; p = 0.000001). After adjustment for the differing size distributions, the prevalence still was somewhat higher for the part-solid and nonsolid nodules combined than for solid ones (p = 0.04). Among the mutually standardized rates for all three types of nodules, the malignancy rate for the part-solid nodules was significantly higher than the rates for either solid nodules (p = 0.004) or nonsolid ones (p = 0.03). When adjusting the malignancy rate of solid nodules to the distribution of the nonsolid ones, the results were identical.

Table 3 gives the distribution of malignancy type by solid and subsolid nodule type. Among the malignancies associated with subsolid nodules, we found a preponderance of bronchioloalveolar carcinoma or adenocarcinoma with bronchioloalveolar features, which are quite different entities from the malignancies associated with the solid nodules (p = 0.0001).

Among the 30 positive findings on annual repeat CT scans were no instances of part-solid nodules, seven findings of nonsolid nodules only, and 23 findings of solid ones only. The prevalence of diagnosed malignancy in these cases was seven of 23 solid nodules and one of seven nonsolid nodules. These numbers are too small for further scrutiny.

Nodules that are not considered solid because they do not completely obscure the entire parenchyma we called subsolid nodules. These nodules can be further subdivided into part-solid and nonsolid nodules. They are a topic peculiar to CT of the chest, and their importance in CT-based screening for lung cancer has not been fully appreciated. Nevertheless, every effort was made in the ELCAP to identify them, and they indeed were fully included in the classification of the initial test finding as positive or negative. The present results indicate that the frequency of these subsolid nodules (in relation to that of solid nodules) in CT-based screening for lung cancer is appreciable (detected on approximately 20% of baseline screenings), a topic we believe has not been discussed in previous literature. With the finer CT collimation made possible by advancements in technology, the relative frequency of detection of the part-solid and, especially, the nonsolid nodule will be even higher.

A more notable finding in our study, however, has to do with the significance of these nodules: overall, the frequency of malignancy in them was much higher than in solid nodules, in the part-solid nodules in particular. To the best of our knowledge, no previous literature exists on malignancy rates in subsolid nodules. In addition, the distribution by malignancy type was strikingly different in the subsolid nodules as compared with the solid ones. The malignancies in subsolid nodules were typically bronchioloalveolar carcinomas or adenocarcinomas with bronchioloalveolar features, whereas in solid ones the malignancies were typically other subtypes of adenocarcinoma. This comparison also has not been addressed in the literature.

Although there may be suspicion that bronchioloalveolar carcinoma represents overdiagnosis of lung cancer—that is, that it progresses so slowly as to be inconsequential—the natural history of three of our patients who were clinically diagnosed as being in stage IA at the time of detection suggests otherwise. Two patients progressed to stage IV adenocarcinoma during the 4 years that elapsed before attempted surgical resection, whereas the third patient remained in stage IA for 2 years before undergoing resection. In the most notable report on the course of bronchioloalveolar carcinoma, Hill [9] noted that in all 46 patients with unresected clinically stage I bronchioloalveolar carcinomas, the disease progressed to a more advanced stage, typically within 2 years. Once the disease was beyond stage I, death typically occurred within 3 years. The progression was slow in only five of these 46 patients, but the disease in these patients, too, progressed to an advanced stage. Results of more recent studies have also suggested this progression on the basis of cytologic and histologic evidence and studies of monoclonality and molecular markers [10,11,12,13,14].

Therefore, because subsolid nodules are a topic of greater importance than previously presumed in the context of CT screening for lung cancer, these nodules need to be studied in greater detail. Work is needed on the quantification of the extent of the solid component and of the rates of growth of the solid and nonsolid components, and then on the significance of these determinations. The radiologic—pathologic correlation of these subsolid nodules and information on potential biomarkers need further elucidation. On the basis of our experience and that of others [3,4,5,6,7,8,9,10,11,12,13,14], we think that the solid component is caused by lack of aeration in the alveoli, perhaps because the alveoli are being compressed or are otherwise being inadequately oxygenated. To the extent that a nodule is not solid, it presumably owes its cloudy appearance to partially aerated alveoli or thickened alveolar walls. For further research, pooling of data is necessary to adequately determine the natural history of malignancies diagnosed in these subsolid nodules and in their more detailed subtypes.

Address correspondence to C. I. Henschke.