Context.—The proposed International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification of lung adenocarcinomas has been published.
Objective.—To evaluate the correlation between epidermal growth factor receptor mutations and histologic subtypes of lung adenocarcinomas according to the upcoming new classification of lung adenocarcinomas.
Design.—Medical records and pathologic slides were reviewed for a total of 107 surgically resected lung adenocarcinomas. All tumors were reclassified according to the predominant histologic subtype, and comprehensive histologic subtyping with semiquantitative assessment of each of the histologic subtypes in increments of 5% was performed. Correlations with epidermal growth factor receptor status were then evaluated.
Results.—Epidermal growth factor receptor mutations were found in 54 cases (50.5%). Epidermal growth factor receptor mutations were significantly associated with the micropapillary-predominant subtype (P = .02) and with the presence (any amount) of the lepidic component (P = .02).
Conclusion.—The upcoming International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification of lung adenocarcinoma is relevant in that it shows a phenotype-genotype correlation. Comprehensive histologic subtyping should be recommended because of the association of particular subtypes with epidermal growth factor receptor mutations.
Non–small cell lung cancer (NSCLC) is a major cause of cancer deaths.1,2 Among histologic types of NSCLCs, rates of adenocarcinoma are increasing in most countries.3 Lung adenocarcinomas are well known for their clinical, radiologic, pathologic, and molecular heterogeneity. This heterogeneity has brought changes in the classification of lung adenocarcinomas during the past several decades, as reflected in the World Health Organization (WHO) classification of lung tumors. According to the 2004 WHO classification, however, more than 80% of lung adenocarcinomas are classified as mixed subtypes.4 This has resulted in pathologic classification not sufficiently reflecting histologic heterogeneity. Not only that, but since epidermal growth factor receptor (EGFR) mutations were discovered in 2004,5,6 numerous studies have been performed with the main focus on molecular pathogenesis and predictive factors of lung adenocarcinomas associated with targeted therapy.7–11 For these reasons the International Association for the Study of Lung Cancer, the American Thoracic Society, and the European Respiratory Society have been developing a new classification of lung adenocarcinoma during the past few years. A proposed classification was recently published.7 The major changes are as follows: (1) the cessation of the use of the term bronchioloalveolar carcinoma, (2) the addition of a category of minimally invasive adenocarcinoma, and (3) the classification of invasive adenocarcinoma according to its predominant subtype.7 Although the detailed revision and diagnostic criteria of the new lung adenocarcinoma classification are not yet officially approved, it is necessary to investigate its relevance correlated with EGFR mutations because EGFR mutation is established as one of the most important predictive markers for EGFR tyrosine kinase inhibitors in NSCLC.8–11 In this study, we classified lung adenocarcinomas according to their predominant subtype, and comprehensive histologic subtyping with semiquantitative assessment of each histologic component was performed as previously described.7 Then, the subtypes' correlations with EGFR mutations were examined.
For editorial comments p 1242.
MATERIALS AND METHODS
We analyzed medical records and archival slides from the collection of surgically resected NSCLCs from 2005 to 2009 at our institution. The cases were selected on the basis of availability of archival slides and tissues rather than consecutive surgeries. Of the selected 193 NSCLC cases, 86 cases with nonadenocarcinoma histology were excluded. Large cell carcinoma was differentiated from solid type adenocarcinoma by negative mucin staining. A total of 107 adenocarcinoma cases were selected. All of the cases were chemotherapy naïve. Clinical data (including age, sex, smoking history, tumor size, and staging) were obtained from each patient's medical records. The staging was in accordance with the standards of the American Joint Committee on Cancer's AJCC Cancer Staging Manual, 7th edition.12
Two pathologists (H.S.S. and S.H.K.) reviewed all available hematoxylin-eosin–stained slides, which included an average of 4.2 slides per case (range, 1–12). At least 3 representative sections were usually submitted for microscopic examination. If a tumor was less than 2 cm in size with ground-glass opacity in radiology, the tumor was totally sampled for microscopic examination and not saved for other tissue processing. Only 1 slide could be reviewed in 1 case. The tumor was 1.1 cm in the largest diameter and a portion of the specimen was submitted for the frozen section diagnosis. For more accurate histologic typing, we did not include the frozen section slide. Histologic classification was done according to the proposed International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society international multidisciplinary classification of lung adenocarcinoma.7 In the classification of invasive adenocarcinomas, 5 histologic patterns were defined as follows: (1) lepidic: growth of neoplastic cells along preexisting alveolar structures (Figure, A); (2) acinar: neoplastic glands that are round to oval-shaped with a central luminal space surrounded by tumor cells (Figure, B); (3) papillary: growth of glandular cells along central fibrovascular cores (Figure, C); (4) micropapillary: growth of tumor cells in papillary tufts that lack fibrovascular cores (Figure, D); and (5) solid: polygonal tumor cells forming sheets that lack recognizable patterns of adenocarcinoma (Figure, E).7 Comprehensive histologic subtyping was performed; that is, the percentage (in 5% increments) of each histologic pattern was recorded. The tumors were then classified according to the predominant histologic pattern. Detailed criteria with pitfalls for comprehensive histologic subtyping have previously been described.13,14 Among the tumors formerly termed mucinous bronchioloalveolar carcinoma, the majority of cases are now recognized to have invasive components and are classified as mucinous adenocarcinoma in the proposed new classification. The invasive components may show the same heterogeneous mixture of acinar, papillary, micropapillary, and solid growth as seen in nonmucinous tumors. If primary adenocarcinomas were morphologically similar to colorectal adenocarcinoma and immunohistochemically cytokeratin 20 or CDX2 positive, they were classified as enteric adenocarcinoma. Of the 107 cases, there were 10 cases of mucinous or enteric variants of invasive adenocarcinomas. Thus, they were also classified by comprehensive histologic subtyping.
EGFR Mutation Analysis
We selected the representative formalin-fixed, paraffin-embedded block with at least 75% of viable tumor that reflected the predominant histologic subtype. In each case, the tumor was dissected from five 4-µm-thick unstained histologic sections by microscopic examination. DNA was extracted by proteinase K digestion and DNeasy DNA isolation kit (Qiagen, Valencia, California) according to the manufacturer's instructions. Direct DNA sequencing of exons 18 through 21 of the EGFR gene was performed as previously described.15 Each case was classified as positive or negative for the EGFR mutation based on the comparison of the sequence with the wild-type sequence.
The relationships between EGFR mutation and clinicopathologic variables were evaluated using the χ2 test. Logistic regression models, with EGFR gene mutation status as the outcome, were used to examine the effects of sex, smoking status, and histologic subtype, with adjustment for each factor. P values less than .05 were considered statistically significant. All statistical data were analyzed using SPSS (version 17, SPSS, Chicago, Illinois).
Patient characteristics are shown in Table 1. There were 52 men and 55 women aged 42 to 80 years (mean, 61.3 years). All patients were Korean. There were 65 never smokers, 22 former smokers, and 20 current smokers. Tumor sizes ranged from 1.1 to 11.0 cm (mean, 3.5 cm). Pathologic stages were stage I in 42 patients, stage II in 21 patients, and stage III in 44 patients.
Correlation With EGFR Mutation and Clinical Features
Epidermal growth factor receptor mutations were detected in 54 of a total of 107 cases (50.5%). Exons with mutations were as follows: of a total of 54 mutated carcinomas, 33 (61.1%) were in exon 19, 4 (7.4%) in exon 20, and 17 (31.5%) in exon 21. The most common mutation was in-frame deletions in exon 19 and the second most common mutation was point mutations in exon 21. Epidermal growth factor receptor mutations were detected more frequently in women (32 of 55; 58.2%) than in men (22 of 52; 42.5%), but the difference was not statistically significant (P = .10). Age, tumor size, and pathologic stage also did not show significant associations with frequency of EGFR mutations (P = .29, .49, and .10, respectively). However, EGFR mutations were more frequent in never smokers (56.9%) than in former smokers (54.5%) and current smokers (25.0%; P = .04).
Correlation With EGFR Mutations and Pathologic Features
All cases were invasive adenocarcinomas and classified into 8 (7.5%) lepidic predominant type, 34 (31.8%) acinar predominant, 32 (29.9%) papillary predominant, 12 (11.2%) micropapillary predominant, and 21 (19.6%) solid predominant. The results of the comprehensive histologic subtyping are shown in Table 2. The most frequent subtype was acinar predominant.
Epidermal growth factor receptor mutations were significantly more frequent in the micropapillary predominant subtype than in the other subtypes (10 of 12 [83.3%] versus 44 of 95 [46.3%]; P = .02). In contrast, EGFR mutations were less frequent in the solid predominant subtype than in the other subtypes (6 of 21 [28.6%], versus 48 of 86 [55.8%]; P = .03) as indicated by the χ2 test (Table 3). There was a significant association between the existence of lepidic, papillary, or micropapillary component and EGFR mutations (P = .005, .006, and .03, respectively; Table 4). In addition, the logistic regression model revealed that the micropapillary-predominant subtype (P = .02) and the presence of lepidic component (P = .02) were significantly correlated with EGFR mutations.
In the proposed new classification, the mucinous and enteric adenocarcinomas are in separate variants and generally known as EGFR mutation negative. All mucinous and enteric adenocarcinomas in our study also were EGFR mutation negative. Thus, we analyzed 97 adenocarcinomas, excluding 10 mucinous and enteric variants. The results also showed that the micropapillary predominant subtype (P = .02) and the presence of lepidic component (P = .03) were significantly correlated with EGFR mutations.
A perplexing aspect of lung cancer is histologic heterogeneity. According to the 2004 WHO classification, more than 80% of adenocarcinomas were mixed type.4 This undermined the significance of adenocarcinoma subclassification in that tumors diagnosed as mixed-type adenocarcinoma show variable clinical prognoses. Since EGFR mutation has emerged as an important predictor for the responsiveness of gefitinib,5,6,16 it has been the subject of numerous studies. Adenocarcinomas with bronchioloalveolar pattern have been reported to be associated with EGFR mutations and the responsiveness of gefitinib.5,16 However, the category called “adenocarcinoma with bronchioloalveolar pattern” is vague and might encompass a broad spectrum of tumors ranging from pure bronchioloalveolar carcinoma to mixed-type adenocarcinoma with minor components of the bronchioloalveolar pattern. Consequently, prognoses ranged from 100% 5-year survival to 10% 3- to 5-year survival.7,17–20 This is one reason why the International Association for the Study of Lung Cancer, the American Thoracic Society, and the European Respiratory Society have been developing a new adenocarcinoma classification.7 In this study, we evaluated the correlation between EGFR mutations and the predominant histologic subtypes according to the new classification. Before the upcoming classification was preliminarily reported,7 there were many studies investigating the correlation between EGFR mutations and histologic subtypes. Epidermal growth factor receptor mutations correlated with adenocarcinoma with bronchioloalveolar pattern,5,16 terminal respiratory unit adenocarcinoma,21 papillary and micropapillary adenocarcinoma subtypes,13 micropapillary pattern and hobnail cell type of adenocarcinoma,22 and absence of solid growth pattern.23 Among these, Motoi et al13 reported that modification of the 2004 WHO classification including the major histologic subtype in mixed type adenocarcinoma suggested correlations between papillary and micropapillary subtypes, EGFR mutations, and gene expression analysis. The subclassification according to the predominant histologic subtype in this study demonstrates that EGFR mutations are associated with the micropapillary predominant subtype and the nonsolid predominant subtype.
In former studies, the histologic subtype most commonly discussed in relationship to EGFR mutations was bronchioloalveolar carcinoma. However, recent studies have revealed that papillary or micropapillary adenocarcinoma is more strongly related to EGFR mutations than bronchioloalveolar carcionoma.13,22,24 These conflicting data may result from the broad spectrum of bronchioloalveolar carcinoma in the literature and varied diagnostic criteria. The papillary and micropapillary pattern also did not receive attention compared with the bronchioloalveolar pattern at the time of the discovery of EGFR mutation.
The micropapillary pattern is characterized by small papillae without a fibrovascular core. It was not included in the subtypes of the 2004 WHO classification.4 However, micropapillary adenocarcinoma was first described by Amin et al25 in 2002, and it has recently been known as a distinct histopathologic marker associated with poor prognosis.25–29 It has also been reported that micropapillary adenocarcinomas are frequently associated with mutations. In one study, 11 of the 15 micropapillary adenocarcinomas (73%) harbored mutually exclusive mutations.24 In our study, 10 of the 12 micropapillary predominant adenocarcinomas (83%) harbored EGFR mutations. In previous studies, the extent of micropapillary patterns to define micropapillary adenocarcinoma has been variable and has ranged from more than 5% to 75%. In our study, the micropapillary predominant subtype contained at least 40% micropapillary patterns. It is thought to be simple and reproducible to define micropapillary adenocarcinoma as its predominant subtype.
In this study, comprehensive histologic subtyping recording the percentage of each histologic component (lepidic, acinar, papillary, micropapillary, and solid) was performed. To the best of our knowledge, comprehensive histologic subtyping was recently introduced by Motoi et al.13 Although it was mentioned that the pathologic diagnosis of mixed type adenocarcinoma should include each histologic pattern identified in the 2004 WHO classification, the comprehensive histologic subtyping is recommended in that it can more accurately reflect the heterogeneous status of each invasive adenocarcinoma by recording the percentage of each component. In our study, whether each component was present or not was also correlated with EGFR mutations. Although the lepidic-predominant subtype was not significantly correlated with EGFR mutations, the existence of the lepidic component was correlated with EGFR mutations. In addition to the lepidic component, the existence of papillary or micropapillary components also correlated with EGFR mutations. Thus, it is recommended to record not only the predominant subtype but also the percentage of each component when lung adenocarcinomas are pathologically diagnosed. On the other hand, we also think that detailed description for gross examination is necessary in the proposed new classification, because it emphasizes intratumoral heterogeneity and a major change is the classification of invasive adenocarcinomas according to their predominant pattern. We hope that more objective and detailed description for gross sampling of lung cancer specimens will be published.
We used a χ2 test to assess the relationship between EGFR mutations and each factor. We also used the logistic regression model, with EGFR mutation status as the outcome, to estimate the interaction of the clinicopathologic variables. The logistic regression model confirmed that the micropapillary predominant subtype and the presence of lepidic component are significantly associated with EGFR mutations. It indicates that histologic characteristics such as micropapillary predominance or presence of lepidic component can be better predictors for EGFR mutations than clinical parameters. Thus, patients with these histologic characteristics according to the new classification can be good candidates for EGFR tyrosine kinase inhibitors in a situation in which EGFR mutation analysis is not possible.
A limitation of this study is that no survival data were provided. We think that the follow-up period is insufficient because the cases were collected from 2005 to 2009. However, it is expected that future research will reveal the prognostic significance of the proposed new classification. In terms of survival, it is especially interesting that micropapillary-predominant adenocarcinoma is significantly related to EGFR mutation, because EGFR mutation is generally associated with good prognosis. Further study is necessary to reveal the mechanism of poor prognosis in subtypes of adenocarcinoma with EGFR mutation.
In conclusion, the micropapillary predominant subtype and the presence of the lepidic pattern were associated with a high incidence of EGFR mutations. The proposed International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification of lung adenocarcinoma is relevant in that it shows a phenotype-genotype correlation. The comprehensive histologic subtyping should also be mentioned because of the association of particular subtypes with EGFR mutations.
This study was supported by a faculty research grant of Yonsei University College of Medicine for 2009 (6-2009-0132).
From the Department of Pathology, Yonsei University College of Medicine, Seoul, Korea. Dr Shim is now with the Department of Pathology, Ewha Womans University School of Medicine, Seoul, Korea.
The authors have no relevant financial interest in the products or companies described in this article.