Largely the result of early lung cancer screening trials coupled with the widespread availability of MDCT scanners capable of providing contiguous high-resolution images throughout the thorax in a single breath-hold period, it is now appreciated that noncalcified pulmonary nodules are best characterized as either subsolid or solid in appearance (55). Subsolid nodules may be further divided into pure ground-glass nodules (focal densities in which underlying lung morphology is preserved) and mixed solid/ground-glass nodules.

To date, numerous reports have shown that there is a close but imperfect correlation between pure ground-glass lesions and either atypical adenomatous hyperplasia (AAH)
or true bronchoalveolar cell carcinoma (BAC), corresponding to Noguchi classification types A and B (Fig. 6-4) (56, 57, 58, 59, 60, 61, 62, 63, 64). As discussed in greater detail in Chapter 7, Noguchi et al. (65) proposed a new classification of adenocarcinomas based on a review of pathologic features of 236 cases of adenocarcinomas less than 2 cm. Briefly, types A and B correspond to true BACs characterized by pure lepidic growth in the absence of active fibroblastic proliferation, whereas type C corresponds to a mixed-type adenocarcinoma [as defined by the new World Health Organization (WHO) classification] (66), defined as having bronchoalveolar features as well as active fibroblastic proliferation. As originally defined, types A, B, and C together were referred to as “replacement tumors” and ostensibly represented progression from truly localized tumors to a more invasive form. Types D to F, in distinction, represent progressively more invasive, poorly differentiated adenocarcinomas without BAC features and likely arose de novo (65,67). Using this classification, Noguchi found significant differences in prognosis, with types A and B having a 100% 5-year survival; type C tumors, with a higher percentage of nodal and distal metastases, having a 75% 5-year survival; and types D to F having a still worse prognosis, with 5-year survivals of approximately 50%.

To date, numerous studies have positively correlated CT findings with Noguchi’s classification. Kim et al. (67a), for example, in a study of 224 patients (132 with BAC, 92 with adenocarcinoma), showed that the extent of ground-glass opacity identified by CT was greater in BAC than in other adenocarcinomas (p <0.001). Small ground-glass lesions, in particular, are likely to represent either foci of AAH or pure BACs (Figs. 6-5 and 6-6). Nakata et al. (68) showed that only 7% of tumors less than 1 cm with a pure ground-glass pattern proved to be mixed adenocarcinomas; the remaining were all BACs.

Mirtcheva et al. (61), correlating CT with pathologic findings in 29 patients (15 with Noguchi type A and B tumors, 14 with type C tumors), also reported close correlation between the appearance of lesions on CT and their pathologic appearance. Specifically using a classification of ground-glass opacities suggested by Gaeta et al. (69), these authors noted the finding of lesions appearing as pure ground-glass opacities, with or without air bronchograms or with surrounding ground-glass “halos,” were predictive of type A and B BACs, whereas mixed solid/ground-glass lesions (Fig. 6-7) or those with superimposed reticulation (Fig. 6-8) were highly predictive of invasive (type C) adenocarcinomas (61).

Not surprisingly, these authors and others have also reported good correlation between the extent of ground-glass attenuation and prognosis (Figs. 6-6, 6-7, 6-8, 6-9, 6-10) (70, 71, 72). Takashima et al. (72), for example, in a study correlating lesion size, percentage of ground-glass opacity, and air bronchograms with clinical data and pathologic findings in 52 consecutive patients with adenocarcinoma with BAC components less than 3 cm, documented a significant difference in prognosis based on the extent of ground-glass attenuation (p = 0.043) as well as the presence of an air bronchogram within the tumor (p = 0.003).

Although more controversial, several reports to date have shown that serial CT studies are of value by confirming that there is progression of disease from pure ground-glass opacities to more complex mixed solid/ground-glass lesions (Fig. 6-9) (61,73). In one study evaluating serial CT findings in 73 patients with a spectrum of adenocarcinomas, a significant linear trend was noted confirming progression from AAH, through type A and B, to type C adenocarcinomas, with lesions first presenting as focal ground-glass opacities, subsequently increasing



in size and density with the appearance of solid components (74).

It should be emphasized that, although pure ground-glass opacities are likely to represent either AAH or BAC, not all such lesions prove malignant (Fig. 6-11). In one study of 10 pure ground-glass opacities followed by serial CT examinations for a median period of 32 months and shown to enlarge, 3 proved to be due to lymphoproliferative disease and 1 proved to be due to focal fibrosis (60). Similarly, the finding of a mixed part solid–part ground-glass or solid lesion, although more likely to represent a mixed adenocarcinoma, may in fact represent virtually any type of lung cancer, including small cell lung cancer.

It is well documented that the radiographic presence of calcification within a pulmonary lesion is generally a reliable sign that a lesion is benign (46,47). The best documentation of the nature of calcification within pulmonary nodules was carried out at the Mayo Clinic and reported in a series of articles in the 1950s (75, 76, 77, 78). In addition to confirming a strong correlation between radiographic and/or tomographic evidence of calcification and benign disease, these authors also defined characteristic patterns of calcification. Based on specimen radiography obtained on a total of 207 solitary pulmonary masses, four different patterns of benign calcification were identified (Fig. 6-12): (a) a laminated pattern with calcium deposited in concentric layers,
between rings of fibrous tissue; (b) a dense central calcified nidus; (c) diffuse and/or irregular or nodular (so-called popcorn) calcifications; and (d) eccentric, single or multiple punctuate calcifications. With minor variation, these guidelines for pattern recognition remain in effect. Although the four patterns cited adequately categorize the range of heterogeneous calcifications encountered, it should be emphasized that for small nodules, particularly those that are 1 cm or less, the commonest CT finding is diffuse homogeneous calcification.

The concept of using CT to identify otherwise radiographically occult calcifications within pulmonary nodules was first published by Siegelman et al., in 1980 (79). In this initial series, 91 patients without radiographic or
tomographic evidence of calcification were evaluated with thin-section CT and assigned a representative CT number based on an average of continuous pixels measured within the estimated center of the lesion. These authors reported that although all 45 primary neoplasms and 13 metastases had representative CT numbers less than 164 Hounsfield units (HU), 20 of 33 benign lesions proved to have representative CT numbers greater than 164 HU. If the correct diagnosis of a benign lesion is considered a true positive and the designation of a malignancy as benign is considered a false positive, then the overall sensitivity of CT in this initial study proved to be 59%, with a sensitivity of 73% for lesions 1 cm or less in diameter and a specificity of 100% (79).

The density of a lesion, as measured by the degree to which it attenuates a collimated beam of x-rays, is an innate property of its tissue content. However, the determination of a reliable measurement in HU is subject to extensive variation. Measurements vary as a function of partial volume averaging, the reconstruction algorithm, true slice thickness, the size of the lesion, kilovoltage, and beam-hardening artifacts (80). The use of CT to evaluate nodules is now well established, as problems related to standardization of tissue attenuation measurements have been notably lessened, in particular, by algorithms to correct beam-hardening artifacts (81). It is generally held that in the majority of cases, calcified lesions are readily recognized by “obvious” high-attenuation values without the use of specific HU thresholds or the need for a reference phantom. As a consequence, it is mandatory that studies be performed with close attention to scan technique (Fig. 6-13). Although thin-section CT suffices in the majority of cases, in questionable instances, an adequate study requires that contiguous images through nodules be obtained with slice collimation less than half the diameter of the nodule to eliminate partial volume averaging, as even with the use of MDCT only a few sections will truly be representative. For small lesions, attenuation values will be overestimated by an edge-enhancing (sharp) algorithm and underestimated by a smoothing algorithm (80). Selection of a high-spatial-frequency reconstruction algorithm, in particular, may artificially create the erroneous impression of calcification within lesions, especially along their edge (82). As discussed in the section on contrast enhancement, this same logic applies to the timed evaluation of nodules following administration of intravenous contrast media. This technical consideration should not obscure the fact that 1-mm sections frequently disclose the presence of calcification when thicker 7- to 10-mm sections do not.

Primary lung malignancies may contain areas of calcification or ossification (Fig. 6-14). Theros (48), for example, found 7 examples of radiologically detected calcification in 1,267 lung cancers, a prevalence of approximately 1%. O’Keefe et al. (78), using radiographs of surgical lung specimens, found calcification in 8 (15%) of 52 resected primary lung cancers. With the addition of CT, the likelihood of detecting calcification within tumors has increased (83,84). As documented in one study, 53 (10.6%) of 500 consecutive patients with provisional diagnoses of lung cancer were found to have calcification on CT (83). Calcification within tumors may result for a number of reasons. These include (a) tumor engulfing previous calcification, as may occur with a preexisting granuloma; (b) dystrophic calcification, occurring in areas of tumor necrosis; and (c) primary tumor calcification, as may be seen in particular within mucinous adenocarcinomas. Radiographically, calcification within lung neoplasms typically appears stippled or eccentric and usually occurs in large central lesions (78). In their study of 53 patients with histologically verified calcification compared to a control group of 15 patients, Grewal and Austin (84a) recorded a mean diameter of 6.2 ± 3 cm for calcified neoplasms versus 4.4 ± 2.3 cm for noncalcified tumors, with 33 (85%) of the calcified tumors measuring more than 3 cm. In a smaller study, Mahoney et al. (84) also noted a propensity for calcification to occur in larger lesions: of 20 lesions, 15 (75%) were 5 cm or greater, whereas only 3 were 2 cm or smaller. Although a number of different patterns of calcification have been noted in these studies, including peripheral, central, diffuse, and amorphous, to date no study has shown any consistent correlation between patterns of calcification and histologic subtype. Indeed, virtually all primary lung cancers, including small cell carcinoma, may have areas of calcification identified on CT scans (83, 84, 85, 86, 87).

In addition to bronchogenic carcinoma, calcification has been identified in other primary lung tumors (88). By far the commonest of these are central carcinoid tumors (Fig. 6-15) (83,87,89,90). Classified as part of the spectrum


of neuroendocrine lung neoplasms, carcinoids are further characterized as either typical or atypical (91). These tumors tend to exhibit variable growth patterns. Most arise as endobronchial obstructing lesions, resulting on occasion in an appearance of focal mucoid impaction; alternatively, they may extend extraluminally, resulting in so-called iceberg lesions (Fig. 6-16). Unlike typical carcinoids, atypical carcinoids show a marked propensity to metastasize early, especially to regional lymph nodes. A typical triad of features has been described for central carcinoid tumors and includes a well-defined round or slightly lobulated lesion that abuts or narrows a central airway in which eccentric calcifications can be identified (Fig. 6-16). In addition, these lesions are characteristically extremely vascular and therefore show marked contrast enhancement following intravenous contrast administration. Although accurate preoperative evaluation requires precise definition of both the intraluminal and the extraluminal components of these tumors, recognition of the vascular nature of these lesions is also of obvious value, especially prior to bronchoscopy, given the propensity of these lesions to hemorrhage, sometimes massively, when biopsy is performed.

In a retrospective study of 31 patients with documented carcinoids (27 typical, 4 atypical), Zweibel et al. (89) found that 18 (58%) were central, whereas 13 (42%) were peripheral in location. Calcification was identified in 7 (39%) central lesions but was seen in only 1 (8%) peripheral lesion. In a similar study of 12 pulmonary carcinoids, Magid et al. (92) also noted a propensity for central tumors to preferentially calcify. In our judgment, the triad of a well-defined central lesion abutting or narrowing an adjacent bronchus with eccentric areas of calcification is all but pathognomonic of this diagnosis (93). The one exception is the rare case of a central, endobronchial hamartoma (see Fig. 5-25).

Despite the fact that most calcified parenchymal neoplasms are identifiable as such with CT, it should be emphasized that malignant lesions may still mimic the appearance of benign calcified granulomas. In addition to the well-described finding of calcified metastases from osteogenic sarcomas and chondrosarcomas, adenocarcinomas, both primary and metastatic, not infrequently are calcified. Review of previous published studies confirms that primary and metastatic adenocarcinoma remains the lesion most likely to be misinterpreted as benign (85, 86, 87,94).

Although routine axial CT images are now the accepted gold standard for evaluating the presence of calcifications within nodules, it is worth noting that two alternative approaches have also been advocated. The first involves the use of dual kVp to analyze nodules as an alternate method for detecting calcium (95). Although the necessity for images to be obtained at precisely the same level for
comparison purposes has, to date, proved a major limitation to this technique (96), the recent introduction of a dual source CT scanner raises the distinct possibility that this approach may supersede traditional CT analysis. The second involves the use of dual-energy digital subtraction radiography (97). Briefly, dual-subtraction chest radiography makes use of the fact that structures containing calcium selectively attenuate lower energy photons in the x-ray beam, allowing identification of calcified nodules, otherwise indistinguishable from noncalcified nodules by routine radiographic techniques. Although the use of this technique is not currently widespread, commercially available digital radiographic systems are now in use, making this a potentially attractive alternative to CT.

The presence of fat within the parenchyma is indicative of one of two diagnoses: hamartoma (Figs. 6-17, 6-18, 6-19) or exogenous lipoid pneumonia (Figs. 6-20 and 6-21). Hamartomas, the third commonest cause of a SPN, are considered benign neoplasms that originate in fibrous connective tissue beneath the mucous membrane of the bronchial wall (98,99). These lesions contain mixtures of myxomatous connective tissue, cartilage, epithelial-lined clefts, and variable amounts of fat, smooth muscle, marrow, and bone (100). Fat is a key element; if one combines three reported series with data on histology, 37 of 68 (54%) of the lesions contain fat (101, 102, 103). Calcification is more common in large lesions, occurring in less than 10% of lesions smaller than 2 cm, one third of tumors 3 to 4 cm in diameter, and 75% of lesions larger than 5 cm (104). Cartilage usually predominates, but occasionally fat or myxomatous connective tissue may be most prominent.

Hamartomas may be diagnosed when they are manifest on CT as lesions that contain fat, or fat plus calcium, and have a smooth contour and sharp outline (87,98). In a report of 47 lesions studied by CT, 30 (63.8%) contained fat only (n = 18), calcium and fat (n = 10), or calcium alone (n = 2) (98). The remaining lesions were diagnosed as indeterminate and managed with surgery or lung biopsy (98). In our experience, lung cancers never manifest as smoothly marginated lesions containing fat. Use of CT to definitively diagnose pulmonary hamartomas requires the same attention to scan technique as discussed previously pertaining to CT identification of calcifications: namely, that contiguous (and/or overlapping) thin sections less than half the diameter of the nodule be acquired reconstructed with a soft tissue algorithm to obviate potential misdiagnoses due to partial volume averaging either from adjacent lung parenchyma or from subtle cavitation resulting in the spurious appearance of HU units in the range of fat.

It is worth emphasizing that, despite their benign nature, hamartomas grow, albeit slowly. Bateson and Abbott (104), for example, noted slow growth in 40 of 45 patients with available serial chest roentgenograms. In one study of 11 cases that exhibited enlargement, recognizable changes were present only in those studied for at least 3 years (105).

A number of focal parenchymal lesions are distinguished by the presence of fluid. Included in this group are lung abscesses (Fig. 6-22), mucoid impaction of the airways (Fig. 6-23), and, rarely, intrapulmonary bronchogenic cysts (106) (Table 6-1) (Fig. 6-24). Even when surrounded by extensive consolidation or within atelectatic lung, abscesses characteristically appear well defined and smooth walled prior to the development of bronchial or pleural communications following the administration of intravenous contrast media and consequently are generally easily differentiated from tumors with CT. Similarly, mucoid impaction of the airways is easily identifiable by the presence of characteristic focal fluid-filled, nonenhancing branching structures, frequently associated with peripheral areas of emphysema (Fig. 6-23). Less commonly, central endobronchial tumors may present with focal mucoid impaction. In these cases, the use of a bolus of intravenous contrast may be of help in differentiating the central enhancing tumor from peripheral fluid-filled airways. Care should be taken to ensure that tumor has not invaded the airways directly by measuring attenuation values within these areas both prior to and following the injection of contrast media. Alternatively, differentiation of tumor from fluid-filled airways may be obtained with T2-weighted MR images (Fig. 6-25) (107,108).

CT is exquisitely sensitive to the presence of air, and it is not surprising that subtle cavitation within nodules is easily identified when otherwise radiographically occult (Figs. 6-26, 6-27, 6-28). In addition to assessing the relationship between cavities and airways and surrounding parenchyma, CT allows precise evaluation of the thickness of cavity walls as well as the presence of intracavitary filling defects and/or air-fluid levels. As reported by Woodring et al. (109), the most important of these is

assessment of cavity wall thickness. Of 65 solitary lung cavities initially assessed by these authors using plain radiographs, all lesions in which the thickest part of the cavity wall was 1 mm were benign. In distinction, of those with greatest wall thickness measuring between 5 and 15 mm, 51% were benign; of those over 15 mm, 95% were malignant. In a follow-up prospective study of 61 additional patients reported by these same authors, 19 (95%) of 20 cavities with a maximum wall thickness of 4 mm again proved benign; whereas 16 (84.2%) of 19 cavities with maximum wall thickness of 16 mm proved malignant (110). In our experience, although thin-walled cavities are almost always benign, thick-walled cavities are usually of indeterminate etiology (Fig. 6-26).

As important as assessing the thickness of the wall of cavities is the finding of bubblelike lucencies within focal lesions. Gaeta et al. (111), in a review of thin-section CT in 11 patients with nodular BAC, reported 2 lesions with serpentine radiolucencies that subsequently were shown histologically to represent air-containing glandular spaces within the tumor (Fig. 6-8). Kuhlman et al. (112), in a study of 30 patients with documented focal BACs, similarly described a pattern they interpreted as suggestive of pseudocavitation in 18 cases. Defined as small oval areas of lucency appearing in or around pulmonary masses, this appearance was interpreted as consistent with residual dilated bronchioles and expanded air spaces, the result of the distinctive propensity of these tumors to extend along alveolar walls without distortion of the underlying architecture. As noted by Zwirewich et al. (113), bubblelike lucencies are seen more commonly in BAC (50%) than in acinar adenocarcinoma (31%) or other lung tumors (11%), making this sign highly suggestive of BAC (113).