Original Study

Semiquantitative Computed Tomography Characteristics for Lung Adenocarcinoma and Their Association With Lung Cancer Survival Hua Wang,1,2 Matthew B. Schabath,3 Ying Liu,1,2 Anders E. Berglund,4 Gregory C. Bloom,4 Jongphil Kim,5 Olya Stringfield,2 Edward A. Eikman,6 Donald L. Klippenstein,6 John J. Heine,2 Steven A. Eschrich,4 Zhaoxiang Ye,1 Robert J. Gillies2,6 Abstract In this study we developed 25 computed tomography descriptors among 117 patients with lung adenocarcinoma to semiquantitatively assess their association with overall survival. Pleural attachment was significantly associated with an increased risk of death and texture was most important for distinguishing histological subtypes. This approach has the potential to support automated analyses and develop decision-support clinical tools. Background: Computed tomography (CT) characteristics derived from noninvasive images that represent the entire tumor might have diagnostic and prognostic value. The purpose of this study was to assess the association of a standardized set of semiquantitative CT characteristics of lung adenocarcinoma with overall survival. Patients and Methods: An initial set of CT descriptors was developed to semiquantitatively assess lung adenocarcinoma in patients (n ¼ 117) who underwent resection. Survival analyses were used to determine the association between each characteristic and overall survival. Principle component analysis (PCA) was used to determine characteristics that might differentiate histological subtypes. Results: Characteristics significantly associated with overall survival included pleural attachment (P < .001), air bronchogram (P ¼ .03), and lymphadenopathy (P ¼ .02). Multivariate analyses revealed pleural attachment was significantly associated with an increased risk of death overall (hazard ratio [HR], 3.21; 95% confidence interval [CI], 1.53-6.70) and among patients with lepidic predominant adenocarcinomas (HR, 5.85; 95% CI, 1.75-19.59), and lymphadenopathy was significantly associated with an increased risk of death among patients with adenocarcinomas without a predominant lepidic component (HR, 3.07; 95% CI, 1.09-8.70). A PCA model showed that texture (ground-glass opacity component) was most important for separating the 2 subtypes. Conclusion: A subset of the semiquantitative characteristics described herein has prognostic importance and provides the ability to distinguish between different histological subtypes of lung adenocarcinoma. Clinical Lung Cancer, Vol. -, No. -, 1-11 ª 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Keywords: CT, Feature, Lepidic growth, Prognosis, Quantitative

Introduction Adenocarcinoma of the lung is a major cause of cancer-related morbidity and mortality worldwide,1 and its incidence has been 1 Department of Radiology, Tianjin Medical University Cancer Institute and Hospital, National Clinical Research Center of Cancer, Key Laboratory of Cancer Prevention and Therapy, Tianjin, PR China 2 Department of Cancer Imaging and Metabolism, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 3 Department of Cancer Epidemiology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 4 Department of Biomedical Informatics, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL 5 Department of Biostatistics, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL

1525-7304/ª 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). http://dx.doi.org/10.1016/j.cllc.2015.05.007

increasing over the past several decades.2,3 Histological characteristics obtained from relatively small portions of tumor might not be representative of the entire tumor, although histological 6

Department of Radiology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL Submitted: Jan 27, 2015; Revised: May 16, 2015; Accepted: May 19, 2015 Addresses for correspondence: Robert J. Gillies, PhD, H. Lee Moffitt Cancer Center and Research Institute, 12902 Magnolia Drive, Tampa, FL 33612 Fax: 813-745-7265; e-mail contact: [email protected]fitt.org Zhaoxiang Ye, MD, Tianjin Medical University Cancer Institute and Hospital, HuanHu-Xi Road, Ti-Yuan-Bei, He Xi District, Tianjin 300060, PR China Fax: þ86-22-23537796; e-mail contact: [email protected]

Clinical Lung Cancer Month 2015

-1

Lung Adenocarcinoma Characteristics on CT and Survival characteristics and staging are commonly used to determine treatment, intratumoral heterogeneity can be a limiting factor to predict prognosis and treatment response. Thus, characteristics derived from analyses of radiological images that represent the entire tumor might have diagnostic and prognostic value. Specifically, imaging features that are associated with the underlying tumor biology could have clinical translational implications. For example, in contrast to solid- or micropapillary-predominant adenocarcinoma, lepidicpredominant adenocarcinoma could eventually be treated with optimized tissue-sparing resection because of the much better outcome and the scarcity of nodal metastases in this subtype.4 Medical imaging can provide noninvasive measurements of tumor features. However, current radiological practice is generally qualitative and provides only limited quantitative information such as dimensional measurements of tumor size. Efforts have been made to develop a standardized lexicon for description of lung tumor features and a standard method for conversion of these descriptors into quantitative, mineable data with the intent of discovering their associations with patient survival.5-7 Computational technical development has permitted a high-throughput process in which a large number of shape, edge, and texture imaging features are extracted.8,9 However, computerized algorithms are more highly dependent on harmonized acquisition and reconstruction parameters than are human readers. The environment of the tumor, which includes important prognostic information, such as desmoplastic response, vascular supply, or localized infiltration of the tumor cannot be segmented effectively. Therefore, computational analysis is not yet able to replace the trained eyes of a radiologist. Nevertheless, computer-derived features can aid radiological diagnosis with extraction of quantitative and unbiased features. Computationally-derived imaging features have been developed to annotate radiological observations,10,11 and the expertise of the radiologist can provide guidance for automated approaches to develop the imaging features that have clinical relevance. Thus, we hypothesized that a standardized set of semiquantitative imaging features can predict prognosis of the patients and benefit the development of prognostic-relevant computerized features. The purpose of this study was to develop and test a standardized set of semiquantitative computed tomography (CT) descriptors of lung adenocarcinoma and assess their association with overall survival. This approach has the potential to support automated analyses by providing guidance and expert evaluation of necessary imaging characteristics, and it can ultimately be used to develop decisionsupport clinical tools to increase accuracy and efficiency of radiological diagnosis.

Patients and Methods

their growth pattern, we classified tumors into 2 subtypes as in previous studies12-14: (1) lepidic predominant adenocarcinomas (n ¼ 55); and (2) adenocarcinomas without a predominant lepidic growth (n ¼ 62); among these cases, 11 had a small proportion of a lepidic component. Lepidic growth pattern was defined as involving alveolar septa with a relative lack of acinar filling. In terms of the new multidisciplinary classification of lung adenocarcinoma sponsored by the International Association for the Study of Lung Cancer, American Thoracic Society, and European Respiratory Society in 2011,15 our subtype of lepidic predominant adenocarcinomas included adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant invasive adenocarcinoma.

Computed Tomography Imaging and Analyses All CT scans were performed before surgery. Ninety-five patients underwent contrast-enhanced CT scan and 22 patients had nonenhanced CT scan. A clinical radiologist with 7 years of experience in chest CT diagnosis developed 25 descriptors and subsequently reviewed all of the CT images. The goal was to develop an initial set of descriptors that would cover a broad area of characteristics with as much resolution as possible. As shown in Table 1, these descriptors were classified into 3 categories: (1) measures that describe the tumor (n ¼ 16); (2) measures that describe the surrounding tissue (n ¼ 5); and (3) measures that describe associated findings (n ¼ 4). Among these descriptors, 17 were rated using a 1 to 5 ordinal scale and 8 descriptors were binary categorical variables. Examples of CT images for each scale of characteristics are shown in Supplemental Figure 1 (in the online version). Our set of descriptors was adapted in part from the Breast Imaging Reporting and Data System (BI-RADS) of the American College of Radiology,16,17 although differences exist between degree (eg, we used 5 levels compared with 2) and organ-specific descriptions. We also adapted measures from the lexicon of the Fleischner Society18 that captured lung cancer features. Other descriptors were adapted from the literature.19,20 In particular, we combined “cavity” and “pseudocavity” used by the Fleischner Society into “air space” as did Matsuki et al20 because of the difficulty of differentiation of them on CT images. The tumor size was measured in the long axis and then classified according to new 7th lung cancer tumor, node, metastases classification and staging system, which has 5 size-based categories with cutoff points at 2, 3, 5, and 7 cm.21 Each tumor was rated by assessing all slices and reporting with a standardized scoring sheet. A second radiologist, with 5 years of experience in chest CT diagnosis, then independently rated the cases using the scoring sheet after training.

Study Population

2

-

The institutional review board approved this retrospective study and waived the informed consent requirement. Data were collected and handled in accordance with the Health Insurance Portability and Accountability Act. This study included 117 patients diagnosed with histologically confirmed adenocarcinoma of the lung who had surgery for primary lung cancer in our institution between January 2006 and June 2009. The mean age of the patients was 65.1  7.5 years, 93.3% self-reported race as white, 55.2% were female, 90.5% were ever-smokers, and 47% had stage I lung cancer. According to

Clinical Lung Cancer Month 2015

Statistical Analyses The agreement between the 2 readers was measured using Kappa for binary variables or Weighted Kappa index for ordinal variables. The k value was interpreted as follows: < 0: poor agreement; 0 to 0.2: slight agreement; 0.2 to 0.4: fair agreement; 0.4 to 0.6: moderate agreement; 0.6 to 0.8: substantial agreement; > 0.8: almost perfect agreement.22 KaplaneMeier survival curves with the log-rank test were performed using R version 2.14 (R Project for Statistical Computing;

Hua Wang et al Table 1 Distribution and Lexicon of the CT Imaging Characteristics CT Characteristic

n (%)

Definition

Scoring Definition

Tumor Space localization Location

Lobe location of the tumor

1

35 (29.9)

1-Right upper lobe

2

9 (7.7)

2-Right middle lobe

3

20 (17.1)

3-Right lower lobe

4

37 (31.6)

4-Left upper lobe

5

16 (13.7)

Distribution

5-Left lower lobe Central location: tumor located in the segmental or more proximal bronchi; peripheral location: tumor located in the subsegmental bronchi or more distal airway

0

9 (7.7)

1

108 (92.3)

0-Central 1-Peripheral Tumor abuts the fissure, tumor’s margin is obscured by the fissure

Fissure attachment 0

81 (69.2)

0-No

1

36 (30.8)

1-Yes

Pleural attachment 0

82 (70.1)

0-No

1

35 (29.9)

1-Yes

1

29 (24.8)a

1-long axis 2 cm

2

46 (39.3)

2-long axis >2 and 3 cm

Size

The greatest dimension in lung window

3

35 (29.9)

3-long axis >3 and 5 cm

4

7 (6.0)

4-long axis >5 and 7 cm

5

0

5-long axis >7 cm

Shape Sphericity

Roundness

1

7 (6.0)

1-Round

2

18 (15.4)

2, 3, 4, 5-Elongated with increasing degrees

3

65 (55.6)

4

24 (20.5)

5

3 (2.6)

Lobulation

Contours with undulations

1

3 (2.6)

1-None

2

34 (29.1)

2, 3, 4, 5-Lobulated with increasing degrees

3

48 (41.0)

4

28 (23.9)

5

4 (3.4)

Concavity

Concave cuts

1

6 (5.1)

1-None

2

37 (31.6)

2, 3, 4, 5-Concave with increasing degrees

3

38 (32.5)

4

28 (23.9)

5

8 (6.8)

Irregularity

Complex shape

1

0

1-Smooth

2

22 (18.8)

2, 3, 4, 5-Irregular with increasing degrees

3

44 (37.6)

4

34 (29.1)

5

17 (14.5)

Clinical Lung Cancer Month 2015

-3

Lung Adenocarcinoma Characteristics on CT and Survival Table 1 Continued CT Characteristic

n (%)

Definition

Scoring Definition

Margin Border definition

Well or ill-defined border

1

2 (1.7)

1-Well defined

2

50 (42.7)

2, 3, 4, 5-Poorly defined with increasing degrees

3

40 (34.2)

4

17 (14.5)

5

8 (6.8)

Spiculation

Lines radiating from the margins of the tumor

1

6 (5.1)

1-None

2

46 (39.3)

2, 3, 4, 5-Spiculated with increasing degrees

3

33 (28.2)

4

26 (22.2)

5

6 (5.1)

Density Texture

Solid or ground-glass opacity

1

3 (2.6)

1-Nonsolid (pure ground-glass opacity)

2

18 (15.4)

2-Partly solid, small extent of solid component

3

35 (29.9)

3-Partly solid, large extent of solid component

4

30 (25.6)

4-Solid, with relatively low density

5

31 (26.5)

Air Space

5-Solid Including cavity and pseudocavity because of the difficulty to differentiate on CT images

1

65 (55.6)

1-None

2

30 (25.6)

2, 3, 4, 5-From small to large extent of lucency

3

16 (13.7)

4

5 (4.3)

5

1 (0.9)

Air Bronchogram

Tube-like or branched air structure within the tumor

1

75 (64.1)

1-None

2

26 (22.2)

2, 3, 4, 5-From small to large extent of air bronchogram

3

8 (6.8)

4

6 (5.1)

5

2 (1.7)

Enhancement Heterogeneity

Heterogeneity of tumor on contrast-enhanced images

1

2 (1.7)

1-Homogeneous

2

10 (8.6)

2, 3, 4, 5-From mildly to highly heterogeneous

3

34 (29.1)

4

28 (23.9)

5

18 (15.4)

NA

25 (21.4)

Calcification

Any patterns of calcification in the tumor

0

111 (94.9)

0-No

1

6 (5.1)

1-Yes

Surrounding Tissues Pleural retraction

4

-

Retraction of the pleura toward the tumor

1

12 (10.3)

1-None

2

34 (29.1)

2, 3, 4, 5-From slight to obvious pleural retraction

3

42 (35.9)

Clinical Lung Cancer Month 2015

Hua Wang et al Table 1 Continued CT Characteristic

n (%)

4

26 (22.2)

5

3 (2.6)

Vascular convergence

Definition

Scoring Definition

Convergence of vessels to the tumor, only applied to the peripheral tumors

1

23 (19.7)

1-None

2

35 (29.9)

2, 3, 4, 5-From slight to obvious vascular convergence

3

22 (18.8)

4

24 (20.5)

5

6 (5.1)

NA

7 (6.0)

Thickened adjacent bronchovascular bundle

Widening of adjacent bronchovascular bundle

1

83 (70.9)

1-None

2

15 (12.8)

2, 3, 4, 5-From slightly to obviously thickened

3

17 (14.5)

4

1 (0.9)

5

1 (0.9)

Emphysema periphery

Peripheral emphysema caused by the tumor or preexisting emphysema

1

52 (44.4)

1-None

2

25 (21.4)

2, 3, 4, 5-From mild to severe emphysema

3

28 (23.9)

4

9 (7.7)

5

3 (2.6) Peripheral fibrosis caused by the tumor or preexisting fibrosis

Fibrosis periphery 1

7 (6.0)

1-None

2

27 (23.1)

2, 3, 4, 5-From mild to severe fibrosis

3

50 (42.7)

4

26 (22.2)

5

7 (6.0)

Associated Findings Nodules in primary tumor lobe

Any nodules suspected to be malignant or indeterminate

0

60 (51.3)

0-No

1

57 (48.7)

1-Yes

Nodules in nontumor lobes

Any nodules suspected malignant or indeterminate

0

45 (38.5)

0-No

1

72 (61.5)

1-Yes Thoracic lymph nodes with short axis >1 cm

Lymphadenopathy 0

83 (70.9)

1

34 (29.1)

Vascular involvement

0-No 1-Yes Vessels narrowed, occluded, or encased by the tumor, only applied to the contrast-enhanced images

0

33 (28.2)

0-No

1

61 (52.1)

1-Yes

NA

23 (19.7)

Abbreviation: CT ¼ computed tomography. a None of the long axis tumor sizes were smaller than 1.02 cm.

Clinical Lung Cancer Month 2015

-5

Lung Adenocarcinoma Characteristics on CT and Survival Table 2 Hazard Ratios for the Association of the CT Characteristics With Overall Survival Overall

Subtype 1a

Subtype 2a

mHR (95% CI)b

mHR (95% CI)b

mHR (95% CI)b

Main Effects Pleural attachment

3.21 (1.53-6.70)

5.85 (1.75-19.59)

1.83 (0.61-5.52)

Air bronchogram

1.21 (0.58-2.54)

0.99 (0.28-3.55)

1.18 (0.44-6.18)

Lymphadenopathy

1.89 (0.91-3.92)

0.48 (0.05-4.53)

3.07 (1.09-8.70)

Final Modelc Pleural attachment

3.40 (1.57-7.40)

5.86 (1.75-19.59)

NI

Air bronchogram

NI

NI

NI

Lymphadenopathy

1.97 (0.92-4.21)

NI

3.07 (1.09-8.71)

Values shown in bold font indicate a statistically significant result. Abbreviations: CT ¼ computed tomography; mHR ¼ multivariable hazard ratio; NI ¼ not included in final model. a Subtype 1 is lepidic predominant adenocarcinomas; subtype 2 is adenocarcinomas without a predominant lepidic component. b Adjusted for age, race, sex, smoking status, histological subtype, long axis tumor size (cm), and stage, where appropriate. c The final models were derived from reverse stepwise regression modeling with a 0.1 significance level for inclusion into the model. Adjustment factors were forced into the reverse stepwise selection model.

6

-

http://www.r-project.org) and multivariable Cox proportional hazard regression was performed using Stata/MP 12.1 (StataCorp LP, College Station, TX). Among characteristics that were found to be statistically significantly associated with overall survival in univariate analyses, we used reverse selection methods to model which sets of CT characteristics were associated with overall survival. Standard clinical risk factors, including age, race, sex, smoking status, histological subtype, long axis tumor size (cm), and clinical stage were incorporated into the modeling where appropriate. We also performed Classification And Regression Tree (CART) adapted for failure time data that used the Martingale residuals of a Cox model to approximate c2 values for all possible cut points for the characteristics (http://econpapers.repec.org/software/bocbocode/s456776. htm). The false discovery rate (FDR) was used to account for multiple testing. The previous value for a feature with an FDR  0.25 is regarded as modest confidence that the association is unlikely to represent a false-positive result and a feature with an FDR  0.05 is regarded as high confidence that the association is unlikely to represent a false positive result. Principle component analysis (PCA) was performed using Evince V2.5.5 (UmBio AB, Umeå, Sweden) to determine characteristics that might differentiate histological subtypes. PCA is a technique that reduces a high-dimensional data set to a lowdimensional data set while retaining most of the variation in the data.23 The new low-dimensional data set is created by the PCA-derived principal components also called scores. These are a linear combination of all variables, where the loadings describe the importance of the original variable for each principal component. The first principal component describes most of the variance and is often considered the most important principal component, and the following principal components show a decreasing amount of explained variance. The results of PCA models are frequently visualized in score and loading plots. The score plot is related to the samples and shows which samples are similar to each other, groupings between classes of samples, and also outliers. The loading plot shows which variables are important for the results seen in the score plot and also which variables are similar to

Clinical Lung Cancer Month 2015

each other. Each variable was normalized to unit variance before PCA.

Results Reader Reproducibility All cases (n ¼ 117) were independently read by 2 radiologists. The agreement of the 2 readers, measured according to the k value, ranged between 0.68 and 1.00. Lobulation, concavity, pleural retraction, fibrosis periphery, and nodules in nontumor lobes were in substantial agreement and all other characteristics had almost perfect agreement (see Supplemental Table 1 in the online version).

Semiquantitative CT Characteristics and Overall Survival Computed tomography imaging data were available for 117 patients (Table 1) but complete survival data were only available for 105 patients. The range of the tumor size was 1.02 cm to 6.72 cm (mean, 2.89  1.22 cm). Based on the distributions in Table 1, we analyzed the association of each of the 25 characteristics with overall survival (Figures 1 and 2, and see Supplemental Figure 2 in the online version). The characteristics that were statistically significantly associated with overall survival (Figure 1) were pleural attachment (P < .001), air bronchogram (P ¼ .03), and lymphadenopathy (P ¼ .02). Size, lobulation, and thickened adjacent bronchovascular bundle were also significantly associated with overall survival (P < .05), yet the associations were likely driven by small numbers in the distribution of the descriptors. This can be observed in Figure 2, where the extremes of the characteristic distributions represented the poorer survival groups. FDR revealed with high confidence that the association between pleural attachment and overall survival (FDR ¼ 0.022) was unlikely to represent a false-positive result. FDR revealed with modest confidence that the associations of air bronchogram (FDR ¼ 0.213) and lymphadenopathy (FDR ¼ 0.212) with overall survival were unlikely to represent a false-positive result. As shown in Table 2, for the multivariable Cox proportional hazard models we first determined the main effects for pleural

Hua Wang et al Figure 1 KaplaneMeier Survival Curves for Pleural Attachment (A), Air Bronchogram (B), and Lymphadenopathy (C) of All Adenocarcinomas, Which Are Statistically Significantly Associated With Overall Survival

attachment, air bronchogram, and lymphadenopathy, and then stratified the data according to histological subtype. Pleural attachment (hazard ratio [HR], 3.21; 95% confidence interval [CI], 1.53-6.70) was statistically significantly associated with an increased risk of death among all patients and among patients with lepidic predominant adenocarcinomas (HR, 5.85; 95% CI, 1.75-19.59). For patients with adenocarcinomas without a predominant lepidic growth, lymphadenopathy was associated with an increased risk of death (HR, 3.07; 95% CI, 1.09-8.70). A reverse stepwise selection approach revealed findings similar to our main effects analyses. When we performed a CART analysis (Figure 3) for pleural attachment, air bronchogram, and lymphadenopathy, we found that patients without pleural attachment and without lymphadenopathy

had significantly improved survival compared with patients with pleural attachment (P < .001).

Differences Between the Characteristics of Histological Subtypes Principle component analysis of the imaging characteristics identified 2 principal components that explained 14% and 11% (totally 25%) of the variance. These 2 principal components demonstrated that the 2 subtypes (lepidic predominant adenocarcinomas and adenocarcinomas without a predominant lepidic growth) were separable, as shown in the PCA score plot in Figure 4A. The separation of the 2 subtypes was mostly along the second principal component, shown on the y-axis. The PCA

Clinical Lung Cancer Month 2015

-7

Lung Adenocarcinoma Characteristics on CT and Survival Figure 2 KaplaneMeier Survival Curves for Size (A), Lobulation (B), and Thickened Adjacent Bronchovascular Bundle (C) of All Adenocarcinomas, Which Are Suggestive of Association With Overall Survival

loading plot in Figure 4B showed that texture was most important for separating the 2 subtypes, when lepidic predominant adenocarcinomas tended to have more of a ground-glass appearance (ie, lower value for the texture characteristic). It is also noteworthy that the surrounding tissues and associated findings were important to the PCA model (their loading values were not 0) and added important information to the tumor characteristics, as shown in Figure 4B. Interestingly, adenocarcinomas with only a minimal lepidic component also showed some extent of lepidic growth characteristics. As shown in Figure 4A, some of the adenocarcinomas without a predominant lepidic growth “misclassified” into the subtype of lepidic predominant adenocarcinomas were actually adenocarcinomas with a small proportion of a lepidic component.

8

-

Clinical Lung Cancer Month 2015

Discussion In this study we developed 25 CT descriptors among 117 patients with lung adenocarcinoma and found that, of these, pleural attachment was most significantly associated with an increased risk of death overall and among patients with lepidic predominant adenocarcinomas, and lymphadenopathy was significantly associated with an increased risk of death among patients with adenocarcinomas without a predominant lepidic component. We used the lexicon of BI-RADS and the Fleischner Society as the guiding principle to develop our descriptors for lung cancer. However, our goal was not limited to structured reporting; we aimed to develop a lexicon that could support automated analysis in the clinical setting by providing guidance and expert evaluation

Hua Wang et al Figure 3 Decision Tree (A) and KaplaneMeier Survival Curves (B) Based on CART Analyses for the 3 Statistically Significant Characteristics in Figure 1

of important imaging characteristics. In many instances, documenting the presence of a given characteristic might be insufficient. For example, previous work used spiculation as one possible margin rating. In contrast, we used spiculation as a variable unto itself with 5 degrees. We intentionally broadened the ordinal scale, which is important for developing quantitative measures that can distinguish prognostic groups with higher resolution and provide the opportunity for automated analytical techniques to be designed to detect features not detectable by the human eye. Many investigators23-27 have reported a correlation between histopathologic and CT findings in adenocarcinomas. Adenocarcinomas showing ground-glass opacity (GGO) on CT scans usually possess a lepidic growth pattern. In our study we analyzed the semiquantitative CT characteristics of adenocarcinomas using PCA modeling and found lepidic predominant

adenocarcinomas can be separated from adenocarcinomas without a predominant lepidic growth, and the most important characteristic that differentiated those 2 subtypes was texture (GGO component). We further analyzed the adenocarcinomas without a predominant lepidic growth and it was interesting that adenocarcinomas with only a minimal lepidic component also showed some extent of lepidic growth characteristics. These results suggest semiquantitative CT characteristics can be used to predict histological subtypes of adenocarcinoma based on the lepidic component. Some reports have shown prognostic factors of lung adenocarcinoma from CT findings.24,28,29 A smaller extent of GGO, lack of lobulation, lack of air bronchograms, presence of coarse spiculation, or thickening of bronchovascular bundles around the tumors have been correlated with poorer survival, which was similar to our results. However, because some characteristics of our

Clinical Lung Cancer Month 2015

-9

Lung Adenocarcinoma Characteristics on CT and Survival Figure 4 Principle Component Analysis Plot. The Score Plot (A) Shows a Separation Between the Different Subtypes: Lepidic Predominant Adenocarcinomas (Green Squares), Adenocarcinomas Without a Lepidic Component (Blue Circles), and Adenocarcinomas With a Small Proportion of a Lepidic Component (Orange Triangles). The Loading Plot (B) Shows How the Characteristics Are Related to Each Other and Which Characteristics Separate the Subtypes. Characteristics Related to Tumor Are Represented by Blue Circles, Surrounding Tissues by Yellow Squares, and Associated Findings by Dark Red Triangles

from those for adenocarcinoma without a predominant lepidic growth. This suggests that the different histological subtypes of adenocarcinoma based on the lepidic component should be analyzed separately when semiquantitative CT characteristics are assessed for their association with lung cancer outcomes. There are several limitations in the present study. First, the sample size of our study was 117 patients and only small numbers of cases were rated into the extreme scales of some descriptors, which could have an effect on the survival analysis. Second, the CT parameters were not consistent for all the cases, and some CT scans were not performed with thin slices. However, because this was a retrospective study with 5-year survival status of the patients, we aimed to keep the sample size as large as possible. Third, we classified the tumors according to their growth pattern; we did not further classify the histological subtype because of the relatively small sample size, and our study showed that CT characteristics and prognostic factors of lepidic predominant adenocarcinomas are different from other subtype of adenocarcinoma. In addition, we did not analyze the effect of surgical technique and other treatments after surgery on survival. Moreover, the information of positron emission tomography-CT scan is useful to predict survival; we will include that information in the future studies.

Conclusion The initial results of our study show that semiquantitative CT characteristics were associated with overall survival in a cohort of lung adenocarcinoma patients. Specifically, pleural attachment was associated with an increased risk of death, especially among lepidic predominant adenocarcinomas. Texture was most important for distinguishing lung adenocarcinomas with and without a predominant lepidic component.

Clinical Practice Points  Efforts have been made to develop a standardized lexicon for

10

-

study have only small numbers of individuals for some scales, we cannot make a definite conclusion because of the small sample size. We did not observe a relationship between extent of GGO with survival. Because this relationship was reported to be found in small (< 3 cm) tumors,24,28,29 it should be investigated further using quantitative GGO evaluation with computerized methods. This approach would be helpful to determine the prognostic value of GGO or GGO ratio. In particular, we found that pleural attachment was one of the most important characteristics correlated with overall survival, especially for lepidic predominant adenocarcinomas. Visceral pleural invasion in lung cancer increases T staging from T1 to T2 even if the tumor is < 3 cm in size. As we know, simple contact between the tumor and neighboring structures does not necessarily mean invasion; CT findings can be ambiguous concerning thoracic invasion.30 Although the CT findings of pleural attachment differed from pathological pleural tumor invasion, this simple characteristic appeared to have prognostic significance. In addition, we found that the prognostic factors for lepidic predominant adenocarcinomas were different

Clinical Lung Cancer Month 2015

describing lung tumor features and a standard method for converting these descriptors into quantitative, mineable data with the intent of discovering their associations with patient survival.  A subset of the semiquantitative characteristics described herein has prognostic importance and provides the ability to distinguish between different histological subtypes of lung adenocarcinoma.  The retrievable data elements in the semiquantitative CT characteristics can be used for data mining and development of automated objective features, which would benefit therapy planning and ultimately improve patient care.

Acknowledgments This research was supported by the National Cancer Institute (grants U01 CA143062 and P50 CA119997), Florida Biomedical Research Programs, King Team Science (grant 2KT01), and the Cancer Informatics Core Facility at the H. Lee Moffitt Cancer Center and Research Institute, a National Cancer Institute-designated Comprehensive Cancer Center (grant P30 CA076292).

Hua Wang et al The authors are thankful for the collaboration between Tianjin Medical University Cancer Institute and Hospital and H. Lee Moffitt Cancer Center and Research Institute.

Disclosure The authors have stated that they have no conflicts of interest.

Supplemental Data Supplemental figures and table accompanying this article can be found in the online version at http://dx.doi.org/10.1016/j.cllc. 2015.05.007.

References 1. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin 2010; 60:277-300. 2. Auerbach O, Garfinkel L. The changing pattern of lung carcinoma. Cancer 1991; 68:1973-7. 3. Barsky SH, Cameron R, Osann KE, Tomita D, Holmes EC. Rising incidence of bronchioloalveolar lung carcinoma and its unique clinicopathologic features. Cancer 1994; 73:1163-70. 4. Sica G, Yoshizawa A, Sima CS, et al. A grading system of lung adenocarcinomas based on histologic pattern is predictive of disease recurrence in stage I tumors. Am J Surg Pathol 2010; 34:1155-62. 5. Rubin DL. Creating and curating a terminology for radiology: ontology modeling and analysis. J Digit Imaging 2008; 21:355-62. 6. Opulencia P, Channin DS, Raicu DS, Furst JD. Mapping LIDC, RadLex, and lung nodule image features. J Digit Imaging 2011; 24:256-70. 7. Channin DS, Mongkolwat P, Kleper V, Rubin DL. The Annotation and Image Mark-up project. Radiology 2009; 253:590-2. 8. Kumar V, Gu Y, Basu S, et al. Radiomics: the process and the challenges. Magn Reson Imaging 2012; 30:1234-48. 9. Aerts HJ, Velazquez ER, Leijenaar RT, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun 2014; 5:4006. 10. Yanagawa M, Tanaka Y, Kusumoto M, et al. Automated assessment of malignant degree of small peripheral adenocarcinomas using volumetric CT data: correlation with pathologic prognostic factors. Lung Cancer 2010; 70:286-94. 11. Gimenez F, Xu J, Liu Y, et al. Automatic annotation of radiological observations in liver CT images. AMIA Annu Symp Proc 2012; 2012:257-63. 12. Lee HJ, Kim YT, Kang CH, et al. Epidermal growth factor receptor mutation in lung adenocarcinomas: relationship with CT characteristics and histologic subtypes. Radiology 2013; 268:254-64.

13. Yang Y, Zhou X, Song X, et al. EGFR L858R mutation is associated with lung adenocarcinoma patients with dominant ground-glass opacity. Lung Cancer 2015; 87:272-7. 14. Weichert W, Warth A. Early lung cancer with lepidic pattern: adenocarcinoma in situ, minimally invasive adenocarcinoma, and lepidic predominant adenocarcinoma. Curr Opin Pulm Med 2014; 20:309-16. 15. Travis WD, Brambilla E, Noguchi M, et al. International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society: international multidisciplinary classification of lung adenocarcinoma: executive summary. Proc Am Thorac Soc 2011; 8:381-5. 16. Molleran V, Mahoney MC. The BI-RADS breast magnetic resonance imaging lexicon. Magn Reson Imaging Clin N Am 2010; 18:171-85, vii. 17. American College of Radiology. Breast Imaging Reporting and Data System, Breast Imaging Atlas. 4th ed. Reston, VA: American College of Radiology; 2003. 18. Hansell DM, Bankier AA, MacMahon H, McLoud TC, Muller NL, Remy J. Fleischner Society: glossary of terms for thoracic imaging. Radiology 2008; 246: 697-722. 19. Gevaert O, Xu J, Hoang CD, et al. Nonesmall-cell lung cancer: identifying prognostic imaging biomarkers by leveraging public gene expression microarray dataemethods and preliminary results. Radiology 2012; 264:387-96. 20. Matsuki Y, Nakamura K, Watanabe H, et al. Usefulness of an artificial neural network for differentiating benign from malignant pulmonary nodules on highresolution CT: evaluation with receiver operating characteristic analysis. AJR Am J Roentgenol 2002; 178:657-63. 21. Mirsadraee S, Oswal D, Alizadeh Y, Caulo A, van Beek E Jr. The 7th lung cancer TNM classification and staging system: review of the changes and implications. World J Radiol 2012; 4:128-34. 22. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159-74. 23. Lee HY, Lee KS. Ground-glass opacity nodules: histopathology, imaging evaluation, and clinical implications. J Thorac Imaging 2011; 26:106-18. 24. Aoki T, Tomoda Y, Watanabe H, et al. Peripheral lung adenocarcinoma: correlation of thin-section CT findings with histologic prognostic factors and survival. Radiology 2001; 220:803-9. 25. Hashizume T, Yamada K, Okamoto N, et al. Prognostic significance of thin-section CT scan findings in small-sized lung adenocarcinoma. Chest 2008; 133:441-7. 26. Honda T, Kondo T, Murakami S, et al. Radiographic and pathological analysis of small lung adenocarcinoma using the new IASLC classification. Clin Radiol 2013; 68:e21-6. 27. Lederlin M, Puderbach M, Muley T, et al. Correlation of radio- and histomorphological pattern of pulmonary adenocarcinoma. Eur Respir J 2013; 41:943-51. 28. Takashima S, Maruyama Y, Hasegawa M, Saito A, Haniuda M, Kadoya M. Highresolution CT features: prognostic significance in peripheral lung adenocarcinoma with bronchioloalveolar carcinoma components. Respiration 2003; 70:36-42. 29. Takashima S, Maruyama Y, Hasegawa M, et al. Prognostic significance of highresolution CT findings in small peripheral adenocarcinoma of the lung: a retrospective study on 64 patients. Lung Cancer 2002; 36:289-95. 30. Imai K, Minamiya Y, Ishiyama K, et al. Use of CT to evaluate pleural invasion in nonesmall-cell lung cancer: measurement of the ratio of the interface between tumor and neighboring structures to maximum tumor diameter. Radiology 2013; 267:619-26.

Clinical Lung Cancer Month 2015

- 11

Lung Adenocarcinoma Characteristics on CT and Survival Supplemental Figure 1

11.e1

-

Examples of Computed Tomography Images for Each Scale of Characteristics (Size, Location, Nodules in Primary Tumor Lobe, and Nodules in Nontumor Lobes Are Not Included)

Clinical Lung Cancer Month 2015

Hua Wang et al Supplemental Figure 1

Continued

Clinical Lung Cancer Month 2015

- 11.e2

Lung Adenocarcinoma Characteristics on CT and Survival Supplemental Figure 1

11.e3

-

Continued

Clinical Lung Cancer Month 2015

Hua Wang et al Supplemental Figure 1

Continued

Clinical Lung Cancer Month 2015

- 11.e4

Lung Adenocarcinoma Characteristics on CT and Survival Supplemental Figure 1

11.e5

-

Continued

Clinical Lung Cancer Month 2015

Hua Wang et al Supplemental Figure 1

Continued

Clinical Lung Cancer Month 2015

- 11.e6

Lung Adenocarcinoma Characteristics on CT and Survival Supplemental Figure 2

11.e7

-

KaplaneMeier Survival Curves for Computed Tomography Characteristics That Were Not Statistically Significantly Associated With Overall Survival

Clinical Lung Cancer Month 2015

Hua Wang et al Supplemental Figure 2

Continued

Clinical Lung Cancer Month 2015

- 11.e8

Lung Adenocarcinoma Characteristics on CT and Survival Supplemental Figure 2

11.e9

-

Continued

Clinical Lung Cancer Month 2015

Hua Wang et al Supplemental Figure 2

Continued

Clinical Lung Cancer Month 2015

- 11.e10

Lung Adenocarcinoma Characteristics on CT and Survival Supplemental Figure 2

11.e11

-

Continued

Clinical Lung Cancer Month 2015

Hua Wang et al Supplemental Table 1 Agreement Between the 2 Readers Characteristic Location Distributiona Fissure Attachmenta Pleural Attachmenta Size Sphericity Lobulation Concavity Irregularity Border Definition Spiculation Texture Air Space Air Bronchogram Enhancement Heterogeneity Calcificationa Pleural Retraction Vascular Convergence Thickened Adjacent Bronchovascular Bundle Emphysema Periphery Fibrosis Periphery Nodules in Primary Tumor Lobea Nodules in Nontumor Lobesa Lymphadenopathya Vascular Involvementa a

(Weighted) k 0.969 0.866 0.896 0.840 1 0.884 0.782 0.714 0.883 0.896 0.889 0.866 0.887 0.897 0.935 1 0.788 0.893 0.943 0.847 0.676 0.897 0.776 0.958 0.840

Binary variable.

Clinical Lung Cancer Month 2015

- 11.e12

Semiquantitative Computed Tomography Characteristics for Lung Adenocarcinoma and Their Association With Lung Cancer Survival.

In this study we developed 25 computed tomography descriptors among 117 patients with lung adenocarcinoma to semiquantitatively assess their associati...
5MB Sizes 0 Downloads 6 Views