Chest Imaging
CT Findings and Progression of Small Peripheral Lung Neoplasms Having a Replacement Growth Pattern
Audio Available | 
Share
OBJECTIVE. We investigated the imaging findings and progression of replacement lung neoplasms that were revealed on thin-section CT and serial CT.
MATERIALS AND METHODS. We evaluated the age of patients and thin-section CT findings (lesion size; percentage of ground-glass opacity areas; and presence or absence of solid portions, lobulation, coarse spiculation, air bronchogram, cavity, multiplicity, and pleural tags) in 73 lesions (11 atypical adenomatous hyperplasias, 17 type A [Noguchi's classification], 18 type B, and 27 type C small peripheral adenocarcinomas). We compared the serial findings of 48 of 73 lesions on low-dose screening CT (n = 21) or thin-section CT (n = 27) obtained at a mean interval of 450 days (range, 85-951 days). Progression from atypical adenomatous hyperplasia through type A to type B and then to type C tumor was studied using trend tests.
RESULTS. A significant linear trend was seen for lesion size (r = 0.55; p < 0.001), percentages of ground-glass opacity areas (r = 0.75; p < 0.001), and the prevalence of lobulation (p < 0.001), spiculation (p = 0.001), air bronchogram (p = 0.023), cavity (p = 0.046), pleural tag (p < 0.001), and solid portions (p < 0.001). In general from serial CT assessment, lesions were recognized first as a ground-glass opacity nodule (56% of 48 lesions) with subsequent increase in size (75%), then solid portions appeared in the nodule (17%), and finally solid portions increased (23%) with occasional augmentation of tissue contraction (6%).
CONCLUSIONS. CT analysis revealed stepwise progression of replacement-type lung neoplasms.
| Introduction |   |
|---|
Noguchi et al. [1] classified small peripheral adenocarcinoma of the lung (≤20 mm) into six types, A—F. Among these, types A, B, and C showed a replacement tumor growth of alveolar lining cells, whereas the other types showed a solid and expansive tumor growth. Prognosis of patients with a localized type A or B tumor and patients with a type C tumor differed significantly. Patients with type A or B tumor showed no lymph node metastasis and had an excellent prognosis with a 5-year survival rate of 100%, whereas patients with type C tumor showed a high incidence (28%) of lymph node metastasis with poorer prognosis (5-year survival rate, 75%). As a result, the investigators regarded types A and B as localized in situ bronchioloalveolar carcinoma and type C (adenocarcinoma with a replacement growth pattern) as an advanced form of types A and B tumors.
Atypical adenomatous hyperplasia also exhibits a replacement growth with atypical cells and is frequently found in cancer-bearing lungs, especially in those with adenocarcinoma [2,3,4]. According to the latest World Health Organization lung tumor classification, this entity is classified as a preinvasive neoplasia of the lung [5]. On the basis of the morphologic and molecular biology findings, many investigators believe that this entity may be a forerunner of bronchioloalveolar carcinoma of the lung [2,3,4, 6,7,8,9]. Thus, a concept of multistep progression from atypical adenomatous hyperplasia through localized bronchioloalveolar carcinoma (types A and B) to advanced adenocarcinoma with a replacement growth pattern (type C) has developed [1, 8, 10].
Several reports characterize subtypes of lung adenocarcinoma by means of groundglass opacity areas on thin-section CT [11, 12]. However, to our knowledge, few articles report systematic assessment on CT of small peripheral lung lesions having a replacement growth pattern. The purpose of our study was to investigate the imaging findings and progression of replacement-type neoplasms of the lung from atypical adenomatous hyperplasia through type A to type B, and then to type C small peripheral adenocarcinomas that were revealed by thin-section CT and serial CT.
| Materials and Methods | |
|---|
Between July 1996 and June 2000, 212 patients with adenocarcinoma and 11 patients with atypical adenomatous hyperplasia underwent surgery in our hospital. Among the 212 patients, 64 patients (30%) had a tumor equal to or smaller than 20 mm in the greatest diameter, and the other 148 patients (70%) had a tumor of less than 20 mm. These 64 adenocarcinomas equal to or smaller than 20 mm were classified into six subtypes as defined by Noguchi et al. [1] (Table 1). When the patient had multiple lesions with different histologic diagnoses, the histologically most advanced lesion was chosen for diagnosis and evaluation. When a patient had multiple lesions with the same histologic diagnosis, the largest lesion was used for evaluation. Histologic diagnosis was made by consensus of two pathologists.
Thus, this retrospective study consisted of the 62 patients with type A (n = 17), type B (n = 18), and type C adenocarcinoma (n = 27) of equal to or less than 20 mm and 11 patients with atypical adenomatous hyperplasia. There were 44 women and 29 men whose ages ranged from 33 to 76 years (mean age, 64 years). We obtained informed consent from each patient, and this study was approved by our institutional ethics committee. All 73 primary lesions were surgically resected within 27 days (range, 1-27 days; mean, 12 days) after CT examination. Fifty-three patients had standard thoracotomy either by lobectomy (n = 48) or segmentectomy (n = 5); the other 20 patients underwent lobectomy (n = 3), segmentectomy (n = 10), or wedge resection (n = 7) under video-assisted thoracic surgery. A special marking device (VATS Marker; Hakko Medical, Tokyo, Japan) had been placed adjacent to the lesion under CT guidance before surgery in seven patients who underwent wedge resection.
Surgical specimens were fixed in an inflated state by transbronchial infusion of formalin and serially sectioned transversely. All sections of lesion were stained with H and E and elastica—van Gieson. A slice with the largest tumor analogous to the plane of the CT scans was used for microscopic analysis.
After obtaining conventional CT scans through the chest with contiguous 10-mm-thick sections, we obtained helical CT scans through the lesion on a CT scanner (HiSpeed Advantage; General Electric Medical Systems, Milwaukee, WI) with sequential 1-mm-thick sections during one breathhold with a pitch of 1. High-resolution CT images were reconstructed at 0.5-mm intervals, using a bone algorithm, with a 20-cm field of view and a 512 × 512 matrix (pixel size, 0.4 mm). The images were photographed using a window level of -550 H and a window width of 1500 H for lung windows and a level of 35 H and a width of 250 H for mediastinal windows.
Without knowledge of the histologic diagnosis, two radiologists, by consensus, evaluated thin-section CT findings: presence or absence of solid portions, lobulation, coarse spiculation, air bronchogram, cavity, multiplicity, or pleural tags of the lesion. Lobulation was defined as an abrupt bulge of lesion contour. Coarse spiculation was defined as the presence of 2-mm or thicker strands extending from the nodule margin into the lung parenchyma without reaching the pleural surface [13].
Both radiologists independently measured the greatest diameter of the lesion and calculated the percentage of ground-glass opacity areas relative to the lesion, using CT scans at the level of the center of the lesion. The mean of the two results obtained by the two radiologists was used for analysis. Ground-glass opacity was defined as a hazy increase in lung opacity on CT without obscuration of underlying vessels [14], whereas a solid portion of the lesion was defined as an area of increased homogeneous lung opacity that obscures underlying vessels. To quantify the percentage of ground-glass opacity areas, we used a dotted grid with equivalent intervals between horizontal and vertical dots. Without knowledge of histologic diagnoses, the two observers counted the number of dots corresponding to the areas of ground-glass opacity and used them as ground-glass opacity areas, whereas the number of dots corresponding to the lesion were used as an entire lesion. Then the percentage of ground-glass opacity areas to an entire lesion was calculated for each lesion.
We correlated the age of patients and thin-section CT findings in the four lesion categories. One pathologist compared CT findings with pathologic data in concert with one radiologist.
Among 73 patients, 48 (66%; nine with atypical adenomatous hyperplasias, 13 with type A, nine with type B, and 17 with type C tumor) underwent low-dose helical CT for screening lung cancer at least twice. Screening CT was performed using a helical CT scanner (CT-W905SR; Hitachi, Tokyo, Japan) with scanning parameters of 120 kV, 50 mA, 10-mm collimation, and a pitch of 2. CT scans were constructed with a 30-cm field of view and a 512 × 512 matrix (pixel size, 0.6 mm). Among the 48 patients, 27 had a set of thin-section CT scans.
The two observers assessed the serial CT scans by consensus, paying special attention to the change in size, number of solid portions in the lesion, and the augmentation of spiculation, pleural tag, air bronchogram, cavity, or vascular convergence. This comparison was performed by displaying two image sets side by side on a cathode-ray tube monitor in the 21 patients in whom only lowdose helical CT scans were available. In the remaining 27 patients who had a set of thin-section CT scans, photographed CT scans were compared. A threshold size for the growth of a lesion was 0.5 mm for high-resolution CT and 1 mm for low-dose CT; these dimensions were larger than the pixel size (0.4 mm for high-resolution CT and 0.6 mm for low-dose CT). The interval between the initial and last CT examination for the 48 patients ranged from 85 to 951 days (mean ± SD, 450 ± 230 days).
We performed a trend test for all CT findings to evaluate the linear progression from atypical adenomatous hyperplasia through type A to type B, and then to type C. The trend was calculated using the linear regression model for three continuous data (age of patients, lesion size, and percentage of ground-glass opacity), and the chi-square test for trend in proportions was used for seven category data (solid portions, lobulation, coarse spiculation, air bronchogram, cavity, multiplicity, and pleural tag of lesion) [15]. We obtained the linear regression model using the three continuous data as dependent variables and four lesion categories as independent variables after ranking these four lesion categories according to the histologic grades at regular intervals. The Mann-Whitney U test was used to compare differences in the three continuous data between each pair of the four lesion categories. The Fisher's exact test was used to compare the prevalence of seven category data between each pair of four categories. A p value of less than 0.05 indicated a statistically significant difference. All the statistical calculations were performed using SPSS software (Statistical Package for the Social Sciences, Chicago, IL).
| Results | |
|---|
As shown in Table 2, a statistically significant trend was seen in lesion size (r = 0.55; p < 0.001); ground-glass opacity areas (r = 0.75; p < 0.001); and the prevalence of solid portions (p < 0.001), lobulation (p < 0.001), coarse spiculation (p = 0.001), air bronchogram (p = 0.023), cavity (p = 0.046), and pleural tag (p < 0.001).
No statistical difference in any of the parameters was seen between atypical adenomatous hyperplasias and type A tumors. However, there was a trend toward large size (p = 0.176) and less ground-glass opacity areas (p = 0.058) for type A tumors than for atypical adenomatous hyperplasias. Ground-glass opacity areas were significantly greater (p = 0.004) in type A tumors than in type B tumors, whereas the prevalence of solid portions (p = 0.003) and lobulation (p = 0.007) was significantly greater in type B tumors than in type A tumors.
Lesion size was significantly greater (p = 0.040) in type C tumors than in type B tumors, whereas ground-glass opacity areas were significantly smaller (p < 0.001) in type C tumors than in type B tumors. The prevalence of solid portions (p = 0.026), coarse spiculation (p = 0.014), air bronchogram (p = 0.018), and pleural tag (p = 0.010) was significantly greater in type C tumors than in type B tumors.
Multiple lesions were detected on CT in six (55%) of 11 atypical adenomatous hyperplasias and in 12 (19%) of 62 adenocarcinomas. Of the 18 lesions, coexistent lesions were surgically resected in nine (two atypical adenomatous hyperplasias and seven adenocarcinomas). The two atypical adenomatous hyperplasias had other multiple lesions of atypical adenomatous hyperplasia at pathologic examination. The other seven lesions (three type A, one type B, and three type C tumors) had other coexistent lesions of varied histologic diagnoses: three type C tumors had a coexistent type C tumor, type A tumor, and a focal fibrosis each; one type B tumor had one atypical adenomatous hyperplasia; two type A tumors had another type A tumor; and one type A tumor had multiple atypical adenomatous hyperplasias. In another three patients with adenocarcinoma who had a single lesion seen on CT, one or more co-existent atypical adenomatous hyperplasias of ≤ 3 mm were found pathologically. In pathologic studies, foci of atypical adenomatous hyperplasia were identified in a cancer lesion in nine (15%: three type A, five type B, and one type C) of 62 adenocarcinomas.
The mean size at initial detection was 8.9 mm (range, 5.5-16 mm) for nine atypical adenomatous hyperplasias, 8.2 mm (range, 6-13.5 mm) for 13 type A, 9.4 mm (range, 6-18.5 mm) for nine type B, and 11.1 mm (range, 4-18 mm) for 17 type C tumors. Of the 48 lesions, eight (17%; one atypical adenomatous hyperplasia, three type A, one type B, and three type C) were detected first on the most recent CT scans (Table 3). The mean size of these eight lesions at initial detection and the mean interval of the serial CT scans were 7.7 mm (range, 4-13 mm) and 637 days (range, 185-951 days), respectively. Of the 48 lesions, 12 (25%; four atypical adenomatous hyperplasias, five type A, two type B, and one type C) remained the same size over a mean period of 310 days (range, 85-772 days).
Of the 48 lesions, 28 (58%; four atypical adenomatous hyperplasias, five type A, six type B, and 13 type C tumors) increased in size. The mean value of increase and the mean period were 0.9 mm (range, 0.5-1 mm) and 352 days (range, 103-659 days), respectively, for atypical adenomatous hyperplasias; 2.2 mm (range, 1-4 mm) and 454 days (range, 375-724 days), respectively, for type A; 2.8 mm (range, 1-5 mm) and 508 days (range, 101-765 days), respectively, for type B; and 3.6 mm (range, 1-13 mm) and 467 days (range, 140-761 days), respectively, for type C tumors. The smallest nodule that grew in this series was a 4-mm nodule of type C tumor that grew to 16 mm over 726 days.
Of the eight lesions detected first on the most recent CT scans, five (one atypical adenomatous hyperplasia, three type A, and one type C) showed ground-glass opacity only, and three (one type B and two type C tumors) showed ground-glass and solid opacity (Fig. 1A,1B,1C,1D). Of the 48 lesions, 27 (56%; all nine atypical adenomatous hyperplasias, 12 type A, four type B, and two type C tumors) showed a ground-glass opacity nodule on the most recent CT scans; a mean interval of serial CT was 404 days (range, 103-772 days) for atypical adenomatous hyperplasias, 325 days (range, 85-724 days) for type A, 460 days (range, 101-765 days) for type B, and 290 days for one type C tumor (Fig. 2A,2B,2C). Solid portions were detected first on the most recent CT scans in eight lesions (17%; four type B and four type C); a mean interval was 473 days (range, 303-720 days) for type B tumors and 684 days (range, 615-726 days) for type C tumors (Figs. 3A,3B,3C and 4A,4B,4C,4D). Increase in solid portions on serial CT was seen in 11 (23%; one type A and 10 type C) tumors; the interval was 375 days for the type A tumor and from 140 to 603 days (mean, 362 days) for type C tumors (Fig. 5A,5B). Vascular convergence, spiculations, or cavities became more prominent in three type C tumors (6%) over a mean period of 667 days (range, 603-726 days).
 View larger version (163K) | Fig. 1A. —66-year-old man with atypical adenomatous hyperplasia first detected on serial low-dose CT. Lesion is barely identifiable on initial transverse low-dose CT scan. |
 View larger version (145K) | Fig. 1B. —66-year-old man with atypical adenomatous hyperplasia first detected on serial low-dose CT. Serial transverse low-dose CT scan obtained 951 days after A shows 9.5-mm nodule of ground-glass opacity (arrow). |
 View larger version (169K) | Fig. 1C. —66-year-old man with atypical adenomatous hyperplasia first detected on serial low-dose CT. Transverse high-resolution CT scan shows ground-glass opacity nodule (arrowhead) in which air bronchograms are seen. |
 View larger version (250K) | Fig. 1D. —66-year-old man with atypical adenomatous hyperplasia first detected on serial low-dose CT. Photomicrograph of histopathologic specimen shows growth of atypical cells along alveolar lining. (elastica—van Gieson,×12.5) |
 View larger version (151K) | Fig. 2A. —71-year-old woman with type A tumor that showed growth on serial high-resolution CT. Initial transverse high-resolution CT scan shows 5-mm nodule of ground-glass opacity (arrowhead). |
 View larger version (159K) | Fig. 2B. —71-year-old woman with type A tumor that showed growth on serial high-resolution CT. Serial transverse high-resolution CT scan obtained 503 days after A shows growth of nodule to 6-mm with ground-glass opacity (arrowhead). |
 View larger version (243K) | Fig. 2C. —71-year-old woman with type A tumor that showed growth on serial high-resolution CT. Photomicrograph of histopathologic specimen shows growth of tumor cells replacing alveolar lining cells without alveolar collapse. (H and E,×12.5) |
 View larger version (171K) | Fig. 3A. —51-year-old woman with type B tumor that has developed solid portions in ground-glass opacity nodule visualized on serial high-resolution CT. Initial transverse high-resolution CT scan shows slightly lobulated 10-mm nodule of ground-glass opacity (arrow). |
 View larger version (148K) | Fig. 3B. —51-year-old woman with type B tumor that has developed solid portions in ground-glass opacity nodule visualized on serial high-resolution CT. Serial transverse high-resolution CT obtained 395 days after A shows growth to 12-mm nodule of ground-glass opacity in which solid component (arrowhead) is seen. |
 View larger version (237K) | Fig. 3C. —51-year-old woman with type B tumor that has developed solid portions in ground-glass opacity nodule visualized on serial high-resolution CT. Photomicrograph of histopathologic specimen shows replacement-type tumor growth with foci of alveolar collapse (AC). (H and E, ×12.5) |
 View larger version (144K) | Fig. 4A. —56-year-old woman with type C tumor that has developed solid portions visualized on serial low-dose CT. Initial transverse low-dose CT scan shows 8-mm nodule of ground-glass opacity (arrow). (Reprinted with permission from [30]) |
 View larger version (158K) | Fig. 4B. —56-year-old woman with type C tumor that has developed solid portions visualized on serial low-dose CT. Serial transverse low-dose CT scan obtained 615 days after A shows growth to 14-mm nodule of ground-glass opacity in which solid component (arrowhead) is now visible. (Reprinted with permission from [30]) |
 View larger version (163K) | Fig. 4C. —56-year-old woman with type C tumor that has developed solid portions visualized on serial low-dose CT. Transverse high-resolution CT scan shows lobulated nodule that consists of solid portion (asterisk) and ground-glass opacity area (arrowheads). (Reprinted with permission from [30]) |
 View larger version (198K) | Fig. 4D. —56-year-old woman with type C tumor that has developed solid portions visualized on serial low-dose CT. Photomicrograph of histopathologic specimen shows areas of fibroblastic proliferation (F) and replacement-type tumor growth along alveolar lining. Disrupted framework was verified in specimen stained by elastica—van Gieson (not shown). (H and E, ×12.5) |
 View larger version (185K) | Fig. 5A. —69-year-old woman with type C tumor that has increased in solid component seen on serial high-resolution CT. Initial transverse high-resolution CT scan shows lobulated solid nodule of 8 mm with small cavities (arrowhead). |
 View larger version (146K) | Fig. 5B. —69-year-old woman with type C tumor that has increased in solid component seen on serial high-resolution CT. Serial transverse high-resolution CT scan obtained 385 days after A shows growth of solid nodule to 13 mm. Small cavities (arrowheads) in nodule have become more prominent since A. |
When we compared the last low-dose helical CT with preoperative thin-section CT for the detectability of the solid component in 21 patients in whom low-dose helical CT scans were used for serial CT assessment, the solid portions were correctly detected with low-dose helical CT in nine cases of type C tumor. Absence of the solid component was correctly predicted in five cases (one atypical adenomatous hyperplasia, three type A, and one type B tumor). There were no false-positive findings. Solid portions were missed with low-dose helical CT in seven cases (one atypical adenomatous hyperplasia, one type A, three type B, and two type C tumors). Thus, when we used the thin-section CT findings as a gold standard, sensitivity, specificity, and accuracy of low-dose helical CT for detecting solid portions were 56% (9/16), 100% (5/5), and 67% (14/21), respectively.
Among 11 atypical adenomatous hyperplasia lesions, four (36%) showed lymphocytic infiltration and edema around the vessels in the lesion, which appeared as dilated branching structures of high-attenuation areas on CT. In three cases (27%), CT showed scattered dotlike solid portions in the lesions, which were attributed to areas of markedly reduced alveolar lumina associated with an increased amount of elastic fibers and lymphocytic infiltration within the alveolar septa (Fig. 6A,6B).
 View larger version (138K) | Fig. 6A. —64-year-old woman with atypical adenomatous hyperplasia with dotlike solid portions shown on high-resolution CT. Transverse high-resolution CT scan shows 9-mm nodule (arrow) of ground-glass opacity in which dotlike solid portions are seen. |
 View larger version (224K) | Fig. 6B. —64-year-old woman with atypical adenomatous hyperplasia with dotlike solid portions shown on high-resolution CT. Photomicrograph of histopathologic specimen corresponding to dotlike solid portions on high-resolution CT scan (A) shows areas of markedly reduced alveolar lumina associated with intense lymphocytic infiltration. Growth of atypical cells along alveolar lining is also seen. (H and E, ×400) |
As found in cases of atypical adenomatous hyperplasia, lymphocytic infiltration, edema, and increased elastic fibers were frequently seen in the interstitium of patients with adenocarcinomas. The portions that were replaced with cancer cells without marked reduction in alveolar lumina and without marked thickening of alveolar septa appeared as ground-glass opacity areas on CT. On the other hand, the portions of densely proliferated cancer cells with markedly reduced alveolar lumina associated with thickened alveolar septa appeared as solid portions on CT. Solid portions having these underlying pathologic features were seen in three (18%) of 17 type A, three (17%) of 18 type B, and four (15%) of 27 type C tumors.
Relatively large portions of collapsed alveoli and fibroblast proliferation were recognized as solid portions on CT. Foci of collapsed alveolar structures were recognized as solid portions on thin-section CT in 12 (67%) of 18 type B tumors; however, the foci were too small to be identified on CT in six tumors (33%). Portions of fibroblastic proliferation were shown as solid portions in 25 (93%) of 27 type C tumors; however, these foci were too small to be depicted on CT in the other two tumors (7%). However, areas of dense tumor proliferation, collapsed alveoli, and active fibroblastic proliferation were indistinguishable from each other on thin-section CT.
| Discussion | |
|---|
Many researchers have found that certain populations of atypical adenomatous hyperplasia cells share morphologic, biologic, and genetic properties with peripheral lung adenocarcinoma cells [2,3,4, 6,7,8,9]. Kitamura et al. [7] morphologically classified replacement lesions as low-grade atypical adenomatous hyperplasia, high-grade atypical adenomatous hyperplasia, and atypical adenomatous hyperplasialike carcinoma and suggested that these lesions represent a spectrum of bronchioloalveolar neoplasia, because the proliferative cell fraction and both p53 and expression of carcinoembryonic antigen increased as the cell atypia advanced.
Because most atypical adenomatous hyperplasia lesions were found incidentally in the lung resected for primary lung cancer, radiologic studies of this entity are rare [16, 17]. Kushihashi et al. [16] reported that all seven atypical adenomatous hyperplasias in their series showed a ground-glass opacity nodule of 10-32 mm on CT. Atypical adenomatous hyperplasias in our series showed a ground-glass opacity predominant nodule of 6-16 mm associated sometimes with lobulation or air bronchogram. However, on CT approximately one third of the lesions showed dotlike solid portions in the nodules that represented areas of markedly reduced alveolar lumina with thickened alveolar walls.
We think that type A tumor may be an advanced form of atypical adenomatous hyperplasia on the basis of the following facts: First, although no statistical difference between atypical adenomatous hyperplasias and type A tumors was seen in any of the parameters and both categories showed similar progression on serial CT, there was a trend toward a large size (p = 0.176) and smaller ground-glass opacity areas (p = 0.058) for type A tumors than for atypical adenomatous hyperplasias. We think that the lack of statistical difference may have been due to the small number of patients in this series. Second, foci of atypical adenomatous hyperplasia were identified in a cancer lesion in 15% of our 62 cancers, which may support a transformation of atypical adenomatous hyperplasia cells into bronchioloalveolar carcinoma cells. Such coexistence in a single lesion is also described in the literature [9, 16, 18].
In our transverse study using thin-section CT, areas of solid portions and the prevalence of spiculation and pleural tag significantly increased as histologic grades advanced from type A through type B to type C. The pathology literature reports that type A tumor shows a pure replacement tumor growth, and when the tumor progresses to type B, collapsed alveoli appear in the tumor [1, 8]. As the tumor advances to type C, central fibrosis due to desmoplastic reaction develops in the tumor, and the foci of fibrosis increase as the tumor progresses [1, 19, 20]. In our CT—pathologic comparison, areas of collapsed alveoli and fibroblastic proliferation showed solid attenuation, whereas areas of replacement tumor growth showed ground-glass opacity on Ct. Central fibrosis and resultant tissue contraction cause fibrotic strands and vascular convergence around the tumor, which are recognized on CT as pleural tags or coarse spiculations or vascular convergence [1, 13, 19]. These underlying pathologic processes depending on the tumor progression may explain the difference in the amount of solid portions and the prevalence of signs of tissue contraction on CT among subtypes of lung adenocarcinoma.
As we found in our study, Kuriyama et al. [11] also described ground-glass opacity areas that were significantly larger in types A and B than in type C. Yang et al. [12] evaluated the percentage of ground-glass opacity areas measured by manual tracing and distribution of solid components of lung adenocarcinomas. The researchers mentioned that the percentage of ground-glass opacity areas in type A or B tumors was significantly larger than that in type C or D tumors. Yang et al. also found that 94% of 17 type A tumors showed pure ground-glass opacities, 71% of 14 type B tumors showed heterogeneous low-attenuation nodules, and 50% of 24 type C tumors appeared as homogeneous nodules of soft-tissue density. We assessed only the percentage of ground-glass opacity areas of a lesion at one level using a dotted grid overlaid on CT scans. However, we think that a more accurate estimate of the percentage of ground-glass opacity areas of a lesion can be obtained by volume with a three-dimensional analysis.
In a study of lung adenocarcinomas that primarily included 20- to 30-mm tumors, Aoki et al. [21] found that air bronchogram, pleural tag, and spiculation were seen in 48%, 76%, and 48% of 21 type C tumors, respectively. The prevalence of pleural tag and spiculation in their series was greater than that (41% and 41%, respectively) in our cases of type C tumors, but the reverse was true for the prevalence of air bronchogram (59% of our type C tumors). Kuhlman et al. [22] and Zwirewich et al. [13] found a greater prevalence of spiculation, lobulation, air-containing space, and pleural tag in localized bronchioloalveolar carcinoma than we did in our study. However, the mean size of the tumors found by Kuhlman et al. (mean, 16 mm) and Zwirewich et al. (mean, 31 mm) exceeded the mean size in our series (12 mm). The prevalence of those CT features may depend on tumor size.
In our longitudinal study using serial CT, 56% of the 48 lesions were recognized first as ground-glass opacity nodules, and 75% showed growth. Appearance of solid portions in the lesion was seen in 17%, increase in solid portions was found in 23%, and augmentation of signs of tissue contraction was detected in 6%. In our transverse study using thin-section CT, lesion size, solid areas, and the prevalence of spiculation, lobulation, air bronchogram, and pleural tags significantly increased as the histologic grades advanced from atypical adenomatous hyperplasia through type A to type B and to type C tumor. Although similar observations in lung adenocarcinomas have been described by another group of researchers, the number of patients who underwent follow-up CT was limited (n = 10) [21]. Thus, our CT findings based on the longitudinal and transverse studies may support stepwise progression from atypical adenomatous hyperplasia through type A to type B and to type C. We also agree with the opinion that most of the central scars develop after the appearance of carcinoma [19].
Most small peripheral lung adenocarcinomas (< 20 mm) with a replacement growth pattern are invisible on chest radiographs [23, 24]. All patients with type A or B tumors are curable, whereas 25% of patients with type C tumors die of lung cancer [1]. Therefore, to characterize the CT appearance of each category and to clarify the evolution of the neoplasms are clinically important in managing and in providing a likely diagnosis of the lesions. Although there are some exceptions, the possibility of atypical adenomatous hyperplasia or type A tumor is high for a small nodule with pure ground-glass opacity, and the possibility of type B or C tumor is high for a larger nodule with ground-glass opacity and solid components. A nodule with a predominant solid component accompanied by CT features of tissue contraction raises the possibility of type C. Although ground-glass opacity reflects a wide variety of pathologic conditions, when a focal area of ground-glass opacity without decrease in size is associated with air bronchogram or lobulation, the likelihood of a replacement-type lung neoplasm is high [11, 25].
Final diagnosis must be made with pathologic studies. CT-guided biopsy is indicated for lung nodules invisible on chest radiography [26, 27]. However, the most important separation among replacement-type adenocarcinoma is the diagnosis of types A and B from type C tumors. Cells or tissues obtained by biopsy are insufficient to determine the presence or absence of fibroblastic foci that is a hall-mark of type C tumors. Thus, we recommend that a patient with a ground-glass opacity nodule with or without solid component that shows no decrease in size should undergo wedge resection under video-assisted thoracic surgery after CT-guided localization. In sectioning the specimens, we must inform the pathologist of the presence or absence and the location of solid portions in the nodule that are vital to excluding an invasive nature.
There are limitations in our study. First, our results can be applied only to the replacement-type neoplasms. However, replacement-type adenocarcinoma accounts for most of peripheral lung cancers detected on a population-based lung cancer screening CT [28]. Sone et al. [28] discovered 60 cancers and nine atypical adenomatous hyperplasias in a population-based lung cancer screening CT. Adenocarcinoma composed 85% of the 60 cancers, replacement-type adenocarcinoma occupied 82% of all adenocarcinomas, and replacement-type neoplasms composed 73% of all cancers and atypical adenomatous hyperplasias.
Second, we must acknowledge the limitation of our longitudinal studies caused by the use of low-dose helical CT. Unchanged tumor size was found in 38% of type A, 22% of type B, and 6% of type C tumors in this series. Yankelevitz et al. [29] reported that follow-up thin-section CT obtained less than 169 days after the initial CT showed tumor growth in all nine adenocarcinomas. However, most of these tumors had a tumor doubling time of less than 100 days. Aoki et al. [21] mentioned that types A and B had a tumor doubling time of more than 365 days, and type C had a broad doubling time ranging from 42 to 1346 days. The 10-mm helical slices used in this study result in an effective slice thickness of about 12.5-13 mm. This method is substantially less sensitive than thin-section CT for detecting subtle changes of lesion size. Low-dose helical CT and long tumor doubling time may be the reasons for the poor sensitivity of tumor growth in our study.
Third, solid components may have been underestimated with low-dose CT in this series. However, solid portions detected on low-dose CT invariably indicated solid portions on thin-section CT. Therefore, we think that at least the evaluation of the changes of solid portions was valid even with low-dose CT. Nonetheless, in following up ground-glass nodules, thin-section CT should be used because 10-mm-slice-thickness CT is suboptimal for assessment of the presence or absence and the amount of solid components in the nodule.
Fourth, this study was retrospective. A limited number of CT scans were obtained in a small number of patients, and serial CT scans were obtained at different time intervals in the various patients. We suggest that prospective studies with a large number of patients are necessary to confirm our results.
In conclusion, our CT analysis supports stepwise progression of replacement-type neoplasms. Characterizing the CT appearance of each category and clarifying the evolution of the lesions can offer valuable information in managing replacement-type lung neoplasms.
Address correspondence to S. Takashima.
| References | |
|---|
