Visceral Pleural Invasion in Pulmonary Adenocarcinoma: Differences in CT Patterns between Solid and Subsolid Cancers

Abstract

Purpose

To analyze the incidence and CT patterns of visceral pleural invasion (VPI) in adenocarcinomas on the basis of their CT presentation as solid or subsolid nodules.

Materials and Methods

A total of 286 adenocarcinomas in direct contact with a pleural surface, resected at an institution between 2005 and 2016, were included in this retrospective, institutional review board–approved study. CT size and longest contact length with a pleural surface were measured and their ratios computed. Pleural deviation, pleural thickening, spiculations, different pleural tag types, pleural effusion, and the CT appearance of transgression into an adjacent lobe or infiltration of surrounding tissue were evaluated. Fisher exact tests and simple and multiple logistic regression models were used.

Results

Of the 286 nodules, 179 of 286 (62.6%) were solid and 107 of 286 (37.4%) were subsolid. VPI was present in 49 of 286 (17.1%) nodules and was significantly more frequent in solid (44 of 179; 24.6%) than in subsolid nodules (five of 107; 4.7%; P < .001). In solid nodules, multiple regression analysis showed an association of higher contact length–to-size ratio (adjusted odds ratio [OR], 1.02; P = .007) and the presence of multiple pleural tag types (adjusted OR, 5.88; P = .002) with VPI. In subsolid nodules, longer pleural contact length of the solid nodular component (adjusted OR, 1.27; P = .017) and the CT appearance of transgression or infiltration (adjusted OR, 10.75; P = .037) were associated with VPI.

Conclusion

During preoperative evaluation of adenocarcinomas for the likelihood of VPI, whether a tumor manifests as a solid or a subsolid nodule is important to consider because the incidence of VPI is significantly higher in solid than in subsolid nodules. In addition, this study showed that the CT patterns associated with VPI differ between solid and subsolid nodules.

© RSNA, 2019

Supplemental material is available for this article.

See also the commentary by Elicker in this issue.

Summary

When evaluating adenocarcinomas preoperatively for the likelihood of visceral pleural invasion (VPI), the solid or subsolid nodule presentation at CT is important, as the frequency of VPI is significantly higher in solid than in subsolid nodules, and CT features associated with VPI differ between solid and subsolid nodules.

Key Points

• ■ In lung adenocarcinomas in direct contact with a pleural surface, visceral pleural invasion (VPI) is significantly more frequent in solid than in subsolid nodules (24.6% vs 4.7%; P < .001).

• ■ In lung adenocarcinomas manifesting as solid nodules, higher contact length–to-size ratio (adjusted odds ratio [OR], 1.02; P = .007) and the presence of multiple pleural tag types (adjusted OR, 5.88; P = .002) are associated with VPI.

• ■ In lung adenocarcinomas manifesting as subsolid nodules, longer pleural contact length of the solid nodular component (adjusted OR, 1.27; P = .017) and the CT appearance of transgression or infiltration (adjusted OR, 10.75; P = .037) are associated with VPI.

Introduction

Visceral pleural invasion (VPI) in pulmonary adenocarcinoma has been associated with a higher cancer recurrence rate and higher mortality (1–4). According to the eighth edition of the American Joint Committee on Cancer Staging Manual, histologic evidence of VPI results in the upstaging of a pulmonary adenocarcinoma with an invasive component less than 3.0 cm to pT2a (5). Studies have also suggested that more radical surgery, such as lobectomy rather than sublobar resection, as well as more extensive lymph node sampling may be appropriate in the presence of VPI (6–9). Although postoperative histologic finding is the reference standard with which to confirm the presence or absence of VPI, preoperative CT can help to assess the likelihood of VPI (10–19) and aid surgeons in choosing their surgical approach. Previous studies have indeed shown that nodule size at CT, the length of pleural contact, and adjacent pleural thickening are associated with histologic VPI (10–16).

No previous study on VPI, however, has included and differentiated between adenocarcinomas manifesting as solid or subsolid nodules at CT. These CT presentations reflect different levels of histologic aggressiveness within the adenocarcinoma spectrum (5). As a result, it is conceivable that they also have different incidences and preoperative CT patterns that affect VPI. Therefore, we sought to analyze the incidence and CT patterns of VPI in pulmonary adenocarcinomas on the basis of their CT presentation as solid or subsolid nodules.

Materials and Methods

Study Material

Our institutional review board approved the study protocol and waived the need for informed consent (protocol 17–110). The researchers did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The study was Health Insurance Portability and Accountability Act compliant. The study is based on our hospital’s radiology and pathology data repository, which contains all pathologically diagnosed primary pulmonary adenocarcinomas surgically resected at our institution from January 2005 to July 2016. This data repository has been used for other studies on different topics in the past, with a partial overlap of lesions and patients (20–26; de Margerie-Mellon, unpublished data, 2019); however, no prior study specifically examined VPI. Inclusion and exclusion criteria of the data repository have been previously published (20). For the current study, all 286 adenocarcinomas in direct contact with a pleural and/or fissural surface at CT, with a nodule size of 30 mm or less at CT, were included. Of all adenocarcinomas, 148 of 286 (51.7%) were resected by lobectomy, 106 of 286 (37.1%) by wedge resection, 31 of 286 (10.8%) by segmentectomy, and one of 286 (0.3%) by pneumonectomy.

Study Population

A total of 286 adenocarcinomas were detected in 270 patients; 255 of 270 (94.4%) patients had one adenocarcinoma resected, 14 of 270 (5.2%) patients had two, and one of 270 (0.4%) patient had three. In the 15 patients with more than one adenocarcinoma, pathologic evaluation confirmed synchronous primary adenocarcinomas in all cases. The mean age of the 270 patients (standard deviation) was 68 years ± 9 (age range, 44–89 years). One hundred sixty-seven of 270 patients (61.9%) were women (mean age, 68 years ± 10; age range, 44–89 years), and 103 of 270 (38.1%) were men (mean age, 67 years ± 9; age range, 49–86 years). Nodules were located as follows: right upper lobe (101 of 286; 35.3%), left upper lobe (68 of 286; 23.8%), right lower lobe (53 of 286; 18.5%), left lower lobe (42 of 286; 14.7%), and right middle lobe (22 of 286; 7.7%). Of the nodules, 103 of 286 (36.0%) had contact with a fissure, 34 of 286 (11.9%) had contact with the pleura along the mediastinum, and 175 of 286 (61.2%) had contact with the chest wall or diaphragm. Of the perifissural nodules, 50 of 103 (48.5%) abutted the right major fissure, 39 of 103 (37.9%) abutted the left major fissure, nine of 103 (8.7%) abutted the right minor fissure, and five of 103 (4.9%) abutted both the right major and minor fissures.

CT Scan Acquisition

Various scanner units and acquisition protocols were used, all of which were considered state-of-the-art at the time of acquisition. The following scanners were most frequently used: Aquilion ONE (320–detector row unit; Toshiba, Otawara, Japan), Discovery CT750 HD (64–detector row unit; GE Medical Systems, Milwaukee, Wis), and LightSpeed VCT (64–detector row unit; GE Medical Systems). All examinations were performed over the entire thorax with the patient in the supine body position and at full inspiration. All images were reconstructed in transverse, sagittal, and coronal planes by using soft tissue and lung kernel reconstruction algorithms, with a section thickness of 0.625 to 2.5 mm and lung window settings (mean, –500 HU; width, 1500 HU).

More detailed information on CT acquisition parameters has been described in a previous publication (20).

Nodule Assessment

CT assessment.—Images of all CT examinations were anonymized and presented in a random order on a picture archiving and communication system to two independent observers (U.S.N. and B.H.H., with 5 and 3 years of experience, respectively). Measurements were performed in the lung window setting (mean, –500 HU; width, 1500 HU) on the transverse, sagittal, or coronal CT image that displayed the largest long-axis diameter and on the transverse, sagittal, or coronal CT image that displayed the longest continuous pleural contact. All measurements were rounded to the closest millimeter (27).

First, nodules were classified as solid or subsolid at CT. Subsolid nodules included both partially solid and pure ground-glass nodules. For our study, subsolid nodules were defined according to the 2017 Fleischner guidelines as nodules that are partially or completely rendered invisible when viewed with soft-tissue window settings (28). Any nodule components other than bronchial structures or normal vasculature that remains visible or obscures margins of vessels or airway walls were considered to be solid (28).

Second, the long-axis diameter of the nodule and the length of the longest continuous pleural contact were measured for all nodules by drawing a straight line (Fig E1 [supplement]). In subsolid nodules, the long-axis diameter and the length of the longest continuous pleural contact with the solid nodule component were also measured in a similar fashion (Fig E2 [supplement]). In pure ground-glass nodules and partially solid nodules without pleural contact of the solid nodule component, the measurements of the solid nodule component were set at zero.

Third, to assess the percentage of the longest pleural contact length relative to the long-axis diameter, we calculated the contact length–to–total size ratio, as previously described by Imai et al (18), as follows:

Of note, our contact length was based on a straight line, as is commonly performed in routine CT assessment by using the click-and-drag tool, whereas Imai et al used a curved line. In subsolid nodules, we also calculated the solid contact length–to–total size ratio and solid contact length–to–solid size ratio. In subsolid nodules that had a solid component and no contact with the pleura, both ratios were set as zero.

Fourth, the presence or absence of pleural deviation, pleural thickening, spiculations, pleural effusion, and types of pleural tags were recorded for all nodules. Pleural deviation was defined as the visible deviation of the pleura from its normal course. Pleural thickening was defined as a visible increase in pleural thickness beyond the “hairline” appearance of the pleura commonly considered normal (29). Spiculations were defined as linear strands that extended from the nodule surface into the lung parenchyma without reaching a pleural surface (30). Pleural tags were defined as linear strands that extended from the nodule surface to a pleural surface and were classified into three types according to Hsu et al (31): I, one or more linear tags on lung window settings; II, one or more linear tags with a soft-tissue component in the mediastinal window setting (mean, 40 HU; width, 400 HU); and III, one or more soft-tissue cordlike tags on the mediastinal window images.

Finally, we qualitatively assessed the presence or absence of transgression into an adjacent lobe or infiltration into the surrounding tissue at CT as a single binary variable. The presence of transgression was defined as ground-glass or soft-tissue opacity in direct contact with the resected nodule in an adjacent lobe. The presence of infiltration was defined visually as a change in opacity of the extrapleural soft tissue in direct contact with the resected nodule compared with the soft tissue adjacent to the lung elsewhere, as done in daily clinical practice (32).

Pathologic assessment.—All resected lung nodules were serially sectioned after formalin fixation by a pathology resident or pathology assistant according to standard department pathologic grossing protocols. Histopathologic examination using traditional hematoxylin-eosin–stained slides and subsequent surgical pathology report generation was performed by staff surgical pathologists. VPI as part of synoptic specimen reporting was assessed at the time of specimen signout; it was defined by histologic evidence of tumor cells that had transgressed beyond the thick elastic layer closest to the pleural surface (33). An elastic special stain was used at the discretion of the staff surgical pathologist for cases in which VPI was histologically indeterminate (33). In cases where VPI was initially reported as indeterminate by the original surgical pathologist, the slides were re-reviewed (K.R.A., a 5th-year senior pathology resident, followed by P.A.V., a pulmonary pathology subspecialty-trained attending surgical pathologist with 6 years of experience), with elastic stains performed as needed to definitively characterize the presence of VPI.

Statistical Analysis

First, we determined the CT features for all included nodules, as well as solid and subsolid nodules separately. We compared the frequency of VPI between solid and subsolid nodules with a Fisher exact test. Second, we evaluated the CT features of solid and subsolid nodules with respect to the presence and absence of VPI by using simple logistic regression analyses; pathologically proven VPI was the outcome and CT features were the predictor variables. Third, to identify CT features jointly associated with VPI, we used a multiple logistic regression model with backward stepwise selection (34). Significance levels for removal from and addition to the models were set at .10 and .05, respectively (34). For continuous variables identified in the multiple regression models, we defined multiple threshold levels and calculated their respective sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV). Additionally, we calculated the sensitivity, specificity, NPV, and PPV for categorical variables identified in the regression models.

All statistical analyses were performed by using Stata software, version 14.1 (Stata, College Station, Tex). Distributions were tested for normality by using a Shapiro-Wilk test. Normally distributed data were expressed as mean ± standard deviation. Nonnormally distributed data were expressed as median and 25%, 75% quartiles. For all tests, the level of statistical significance was set at P ≤ .05.

Information and data about observer agreement are detailed in Appendix E1 (supplement).

Results

The CT features of all 286 nodules are shown in Table 1. More nodules were solid (179 of 286; 62.6%) than subsolid (107 of 286; 37.4%). Of the 286 nodules, 49 of 286 (17.1%) showed histologically confirmed VPI. VPI was significantly more frequent in solid (44 of 179; 24.6%) than subsolid (five of 107; 4.7%) nodules (P < .001).

Table 1:

CT Parameters of All Included Nodules

Note.—Continuous parameters are displayed as mean ± standard deviation; categorical variables are expressed as numbers of patients, with percentages in parentheses.

Given the strong relation and the small absolute differences between two observers with respect to the measurements of nodule size and pleural contact length (Appendix E1, observer agreement [supplement]), the mean size and contact length as measured by the two observers were calculated and used for all subsequent analyses.

Solid Nodules

CT features of the 179 solid nodules with respect to the presence or absence of VPI are shown in Table 2. Imaging examples of nodules with and without VPI are shown in Figures 1 and 2. Pleural contact length was significantly greater in nodules with VPI (16 mm ± 6) than in those without VPI (13 mm ± 7) (P = .019). The contact length–to-size ratio was significantly higher in nodules with VPI (81.6% ± 21.4) than in those without (71.1% ± 24.5) (P = .014). Multiple tag types were significantly more frequent in nodules with VPI (nine of 44; 20.5%) than in those without VPI (seven of 135; 5.2%) (P = .004). Size, pleural deviation, pleural thickening, spiculations, and the CT appearance of transgression or infiltration did not significantly differ between nodules with and without VPI (P = .108 to .854).

Table 2:

CT Parameters of Solid Nodules with Respect to Presence or Absence of Histologically Confirmed Visceral Pleural Invasion

Note.—Unless otherwise indicated, continuous parameters are displayed as mean ± standard deviation; categorical variables are expressed as numbers of patients, with percentages in parentheses. OR = odds ratio.

*Data in parentheses are 95% confidence intervals.

Figure 1:

Solid lung nodule with visceral pleural invasion on the basis of pathologic examination in a 64-year-old woman. A, B, Transverse CT images show the solid lung nodule (open arrow) in direct contact with the pleura in the right upper lobe, with measurements, B, of pleural contact length (black arrow, 11 mm) and nodule size (white arrow, 11 mm) that resulted in a contact length–to-size ratio of 100%. C, Low-power histologic image with hematoxylin-eosin stain shows tumor cells advancing toward the pleural surface with a clear disruption of the elastic laminae (arrows), as highlighted by, D, an elastic stain (original magnification, ×100). E, Higher-power image with an elastic stain demonstrates the disruption of the elastic laminae (arrows) by tumor cells invading toward the pleural surface (original magnification, ×400).

Figure 2:

Solid lung nodule without visceral pleural invasion on the basis of pathologic examination in a 71-year-old woman. A, B, Transverse CT images show the solid lung nodule (open arrow) in the left upper lobe in direct contact with the left major fissure (solid arrows), with measurements, B, of pleural contact length (black arrow, 9 mm) and nodule size (white arrow, 18 mm) that resulted in a contact length–to-size ratio of 50%. Low-power histologic images with, C, hematoxylin-eosin stain and, D, elastic stain show thickening and fibrosis of the pleura but a fully intact elastic laminae (arrows) present between the tumor cells (bottom) and the pleural surface (top) (original magnification, ×100).

On multiple regression analysis, the contact length–to-size ratio and the presence of multiple tag types were jointly associated with the presence of VPI (P = .007 and P = .002, respectively). As shown in Table 3, the adjusted odds for VPI were increased 1.02-fold for each percentage increase in the contact length–to-size ratio. Simultaneously, the adjusted odds for VPI increased nearly sixfold if multiple pleural tags were present. Sensitivity, specificity, PPV, and NPV for potential thresholds for overall contact length–to–overall size ratio and for the presence of multiple pleural tag types are shown in Table 4.

Table 3:

Multiple Logistic Regression Analysis of Solid and Subsolid Nodules

Note.—OR = odds ratio.

*Data in parentheses are 95% confidence intervals.

^†Continuous variable with the OR representing an increase in odds per unit.

Table 4:

Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value for Multiple Thresholds of Contact Length–to-Size Ratio and Multiple Tag Types in Solid Nodules

Note.—Values are expressed as percentages, with numbers in parentheses and 95% confidence intervals in brackets.

Subsolid Nodules

For subsolid nodules, the CT features with respect to the presence and absence of VPI are shown in Table 5. CT and histologic images of nodules with and without VPI are shown in Figures 3 and 4.

Table 5:

CT Parameters of Subsolid Nodules with Respect to Presence or Absence of Histologically Confirmed Visceral Pleural Invasion

Note.—Continuous parameters are displayed as mean ± standard deviation; categorical variables are expressed as numbers of patients, with percentages in parentheses. NA, not applicable; OR = odds ratio.

*Data in parentheses are 95% confidence intervals.

Figure 3:

Subsolid lung nodule with visceral pleural invasion on the basis of pathologic examination in a 59-year-old man. A, B, Transverse CT images show the subsolid lung nodule (open arrow) with a ground-glass and solid nodule component in the left lower lobe in direct contact with the left major fissure (solid arrows) and the mediastinal pleura, with measurement, B, of the longest pleural contact length of the solid nodule component (double-headed arrow, 19 mm). Low-power histologic images with, C, hematoxylin-eosin stain and, D, elastic stain show fibrosis and thickening of the pleura; however, tumor cells (white arrow) are seen infiltrating beyond the thick elastic layer (black arrows) toward the pleural surface (original magnification, ×40).

Figure 4:

Subsolid lung nodule without visceral pleural invasion on the basis of pathologic examination in an 83-year-old woman. A, B, Transverse CT images show the subsolid lung nodule (open arrow) with a ground-glass and solid nodule component in direct contact with the right major fissure (solid arrows) in the right upper lobe, with measurement, B, of the longest pleural contact length of the solid nodule component (double-headed arrow, 5 mm). C, Low-power histologic image with hematoxylin-eosin stain shows predominantly lepidic adenocarcinoma (_*) approaching, but not extending to, the pleura (black arrows) (original magnification, ×40).

Pleural contact length of the solid nodule component was significantly greater in nodules with VPI (12 mm ± 5) than in those without VPI (4 mm ± 5) (P = .006). The solid contact length–to–total size ratio was significantly larger in nodules with VPI (41.6% ± 18.5) than in those without VPI (15.4% ± 20.4) (P = .017). In addition, the solid contact length–to–solid size ratio was significantly larger in nodules with VPI (78.9% ± 23.1) than in those without VPI (32.3% ± 41.4) (P = .042). Transgression through a fissure into an adjacent lobe or infiltration of surrounding tissue was significantly more frequent with VPI (two of five; 40.0%) than with no VPI (three of 102; 2.9%) (P = .004). Size, pleural deviation, pleural thickening, spiculations, and tag types did not differ significantly between nodules with and without VPI (P = .051 to .856).

On multiple regression analysis, solid contact length and the CT appearance of transgression or infiltration tissue were jointly associated with VPI (P = .017 and P = .037, respectively). As shown in Table 3, the adjusted odds for VPI were increased 1.27-fold for each millimeter increase in contact length. Simultaneously, the adjusted odds for VPI increased by nearly 11-fold if transgression or infiltration was present. Sensitivity, specificity, PPV, and NPV for potential thresholds for contact length of the solid nodule component and for the presence of transgression or infiltration at CT are shown in Table 6.

Table 6:

Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value for Multiple Thresholds of Pleural Contact Length of Solid Nodule Component and for CT Appearance of Transgression or Infiltration in Subsolid Nodules

Note.—Values are expressed as percentages, with numbers in parentheses and 95% confidence intervals in brackets.

Discussion

Our study of surgically resected lung adenocarcinomas shows that VPI is more common in solid than in subsolid nodules and that the CT features associated with VPI are different in solid than in subsolid nodules. Our findings indicate that for preoperative evaluation of adenocarcinomas for potential VPI, different morphologic aspects of solid and subsolid nodules require different perspectives of attention. In solid nodules, attention should be given to larger contact length–to-size ratio. In addition, the presence of multiple tag types should be confirmed or excluded. In subsolid nodules, attention should be given to the solid nodule component and its contact length to the pleura, as well as to the presence of transgression or infiltration of the fissure. Our results show that this information helps to improve preoperative likelihood estimation for VPI.

Such improved preoperative likelihood estimation for VPI is important because VPI has been associated with increased overall mortality and decreased disease-free survival (8,35–46). This has been controversially discussed in the literature; some studies have suggested that VPI is not a poor prognostic parameter, particularly in early-stage cancers (47–53). Meta-analyses, however, confirmed its association with increased overall mortality and decreased disease-free survival (1–4), with size-independent effects in cancers 30 mm or smaller (1–3). Consequently, the current eighth edition of the American Joint Committee on Cancer Staging Manual upstages adenocarcinomas with an invasive size of 30 mm or smaller to pT2a if VPI is present (5). Furthermore, surgical strategies may vary according to the presence of VPI. For example, in their study investigating the risk factors of poor disease-specific survival, Koike et al (6) concluded that only “patients with no suspicion of VPI would be suitable candidates for sublobar resection.” Similarly, Schuchert et al (8) showed that segmentectomy in the presence of VPI was associated with an increased recurrence risk in stage I lung cancers. In a study among patients with adenocarcinomas 30 mm or smaller who first underwent wedge resection, Wang et al (9) showed that completion lobectomy as a second step was indicated if VPI was present. Because of an increased rate of lymph node metastases in VPI (7,46,54), Kudo et al (7) proposed that if preoperative or intraoperative evaluation suggested VPI in tumors 30 mm or smaller, more extensive lymph node dissection may be warranted rather than only lymph node sampling.

Our study cohort included 286 nodules in contact with a pleural surface; 179 of 286 (62.6%) were solid and 107 of 286 (37.4%) were subsolid. With this, our sample size was larger than those of previous studies that ranged from 21 to 208 lesions (10–19). Results from those previous studies with regard to VPI varied. By using different multiple regression models, those studies showed an association between VPI and longer pleural contact length (13–16), pleural thickening (14,16), a larger ratio of solid nodule component size to total nodule size (16), larger size (15,16), pleural patterns (10), relation of the nodule to the pleura (14–16), and pleural deviation (14). Most previous studies included lesions without considering their presentation as solid or subsolid at CT (10–14,17,18). Only two studies included partially solid nodules (15,16), and one study included only pure ground-glass nodules (19). Further, previous studies also included histologic heterogeneous tumors (10,12,14,16–18) or lesions larger than 30 mm at CT (12–14,17,18). Therefore, to the best of our knowledge, our study is the only current investigation of differences in incidence and patterns of VPI that manifested as solid and subsolid nodules. The comparatively large sample size, the focus on distinct morphologic CT categories, and the focus on lesions of 30 mm or less add to the existing evidence on VPI.

As expected, our study confirmed that VPI was significantly more frequent in solid than in subsolid nodules. This finding is in line with the fact that pulmonary adenocarcinomas manifesting as solid nodules are considered to represent histologically more advanced and aggressive tumors than subsolid nodules (5,53,55).

For our solid nodules, the contact length–to-size ratio was significantly greater in nodules with VPI than in those without VPI. In contrast to our study, Ebara et al (10) did not find an association of the ratio with the presence of VPI in their regression model, which included histologically heterogeneous tumors without differentiating between solid and subsolid nodules. Imai et al (18) showed that the ratio was a noninvasive marker for the different depths of VPI in T3-T4 lung cancer. Similar to Ebara et al, those authors also included more heterogeneous histologic tumors than our study did, but they included advanced tumors that are likely to manifest as solid lesions at CT. Therefore, our study complements the previous study of Imai et al (18) by showing a relation between contact length–to-size ratio, measured in slightly different fashions, and the presence of VPI in clinical T1 pulmonary adenocarcinomas that manifest as solid nodules with a size of 30 mm or smaller. Of note, when we chose a cutoff of >50%, the sensitivity was 86.4% and the NPV was 85.7%. Therefore, the ratio may be used for screening for VPI or ruling out VPI.

For our solid nodules, multiple tag types were significantly more frequent in nodules with VPI than in those without VPI. The pleural tags, in particular type II, have been previously associated with VPI in non–small cell lung cancers without direct contact with a pleural surface (31). However, the role of pleural tags in nodules in direct contact with a pleural surface has not been investigated. Thus, to our knowledge, our study is the first to show a relation for pleural tags in those nodules. Indeed, if multiple tag types were present, the odds of VPI being present were five times higher. This suggests that pleural tags have a role in nodules with direct contact to the pleura and that they are an additional CT feature for preoperative evaluation. Overall, these findings indicate that the combination of direct contact with a pleural surface and multiple pleural tags should raise the suspicion for VPI in solid nodules.

For our subsolid nodules, features for assessing the likelihood of VPI were different from those for our solid nodules. For subsolid nodules, the solid contact length was significantly greater in nodules with VPI than those without VPI. This finding suggests that the solid component and not the overall tumor size is of major interest for likelihood estimations of VPI in subsolid nodules. This finding is consistent with the fact that the solid nodule component histologically tends to reflect the invasive (nonlepidic) adenocarcinoma pattern, which is also used for pathologic tumor staging (5).

Moreover, if the solid contact length was less than 5 mm, no VPI was present (100% sensitivity and 100% NPV). Thus, VPI is unlikely in pure ground-glass nodules and partially solid nodules with no or less than 5 mm of solid contact length. This finding is in line with results from a previous study showing that in adenocarcinomas without a solid nodule component, VPI was generally not present (19). Furthermore, our cutoff was similar to the 4.1 mm described by Ebara et al (10), also with 100% sensitivity and 100% NPV. Although we determined the likelihood of VPI based on the solid nodule component in subsolid nodules only, Ebara et al did not differentiate between solid and subsolid nodules (10). Furthermore, those authors determined contact length by drawing a curved line along the contact, whereas we did so by drawing a straight line. From a practical perspective, this shows that the simpler method of drawing a straight line, such as is routinely used in clinical practice, is similar to manually retracing the curved pleural contact. Overall, our findings indicate that, in contrast to solid nodules, where the contact length–to-size ratio should be calculated, using the solid contact length for likelihood estimations of VPI in subsolid nodules is justified.

Another CT feature in subsolid nodules associated with VPI was the appearance of transgression through a fissure into an adjacent lobe or infiltration of surrounding tissue. In fact, the odds of VPI increased by 11-fold if transgression or infiltration was present. From a histologic perspective, these radiologic features can represent tumor invasion or, conversely, other processes in the adjacent tissue, such as peritumoral organizing pneumonia, granulomatous inflammation, postobstructive pneumonia, or even parenchymal scarring or fibrosis (Fig 5).

Figure 5:

Adenocarcinoma without visceral pleural invasion on the basis of pathologic examination in a 61-year-old man. A, Transverse CT image shows solid nodule (open arrow) in direct contact with the right major fissure (small white arrows) in the right middle lobe. On the CT image, the nodule appears to be transgressing through the fissure into the right lower lobe (black arrow). B, High-power histologic image with hematoxylin-eosin stain reveals this presumed transgressing tumor to be a peritumoral inflammatory response composed of organizing pneumonia and interstitial lymphoplasmacytic chronic inflammation (original magnification, ×200).

Our study had several limitations. First, verification bias is inherent to our study design because only surgically resected and histologically proven adenocarcinomas were included. For this reason, our sample might have been biased toward larger, morphologically more conspicuous or aggressive nodules. Nevertheless, this inclusion criterion warranted a pathologically homogeneous sample of nodules. Furthermore, this bias might be counterbalanced by including only surgically resectable adenocarcinomas smaller than or equal to 30 mm at CT. Second, the technical protocols for the CT examinations analyzed were not ideally homogeneous, given that the examinations were performed during a period of more than 10 years with different scanner units from different manufacturers and installed in different hospitals. At the time of acquisition, however, each individual CT unit was considered to be state-of-the-art. We therefore believe that the variability over time in our CT protocols realistically reflects the clinical practice of a large academic medical center with multiple affiliated radiology satellite sites. Third, the partial overlap of some our parameters likely reflects the clinical context in which our nodules were collected. This overlap also may highlight that diameter measurements, which are the current standard, might be too simplistic for the often irregular and complex morphology of nodules such as those included in this study. Fourth, we included patients with more than one adenocarcinoma resected. However, because the number of multiple nodules per patient was small, this correlation was not thought to have a large effect on inference and P values in this report. Fifth, no information was available on patient outcome. Sixth, our results are limited by the small sample size of adenocarcinomas with VPI in those manifesting as subsolid nodules. Thus, further research is needed to validate our results in larger cohorts of this rare subcategory and to determine whether and how our observations will affect patient prognosis and outcome. Finally, it remains to be seen whether similar or different CT parameters can help predict VPI for other lung cancer types, specifically squamous cell carcinoma.

In conclusion, for evaluation of adenocarcinomas preoperatively for the likelihood of VPI, the solid or subsolid nodule presentation at CT is important because the frequency of VPI is significantly higher in solid than in subsolid nodules, and CT features associated with VPI differ between solid and subsolid nodules. Although in solid nodules contact length–to-size ratio and the presence of multiple tag types at CT were associated with VPI in subsolid nodules, attention should be given to the solid nodule component and its contact length with the pleura, as well as to the presence of transgression or infiltration. These observations could contribute to an improved preoperative assessment of lung nodules in direct contact with the pleura and strengthen the role of CT in the preoperative assessment of these lesions.