Cytologic subtyping of lung adenocarcinoma by using the proposed International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IASLC/ATS/ERS) adenocarcinoma classification
Erika F. Rodriguez MD, PhD,
Department of Pathology, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania
Presented in part at the 102nd Annual Meeting of the United States and Canadian Academy of Pathology (USCAP); March 2-8, 2013; Baltimore, Maryland.
The significance of histologic subtyping of surgically resected lung adenocarcinoma (ADC) was recently proposed by the International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IASLC/ATS/ERS) classification. Approximately 70% of lung cancer patients present with advanced disease, and small biopsies or cytology specimens are frequently the only available diagnostic material. It is uncertain whether proposed morphologic subtyping of ADC can be applied to small specimens. The objective of this study was to assess the applicability of morphologic subtyping of ADC on cytologic specimens.
Consecutive, newly diagnosed primary lung ADC specimens from patients with matched surgical resection and cytology specimens (n = 66) were selected for the study. The dominant morphologic pattern was determined according to the IASLC/ATS/ERS classification. The number and percentage of malignant cells in cytology specimens were also evaluated.
Concordant subtyping of ADC between the dominant pattern on resection and cytology specimens was observed in 26 cases (40%), and was discordant in 32 cases (48%). Concordance increased in specimens that had >200 cells and when correlating with the primary or secondary histologic pattern. The acinar pattern was the most common in concordant cases, whereas discordant cases had a predominantly solid pattern.
Adenocarcinoma (ADC) is the most common histologic subtype of lung cancer that encompasses a heterogeneous group of malignancies with different morphologic features, etiology, and molecular changes. The most recent World Health Organization (WHO) classification of lung carcinoma includes a list of various morphologic variants of lung ADC but does not take into consideration their clinical, prognostic, or molecular characteristics. A large proportion of lung ADCs are composed of several histologic patterns and would be classified as a mixed subtype according to WHO classification. For many years, pathologists have been able to recognize different histologic features of ADCs and their prognostic significance. However, these subtypes were not incorporated into the WHO classification, and the focus of the WHO was primarily on resection specimens as opposed to small biopsies and cytology specimens.[2-7]
Recent experience with targeted therapies in lung ADC suggest that various histologic subtypes have specific molecular profiles and exhibit different responses to therapy. All of the shortcomings of the WHO classification were recently recognized, and a new proposal for a multidisciplinary classification of lung ADC was put forward by the International Association for the Study of Lung Cancer (IASLC)/American Thoracic Society (ATS)/European Respiratory Society (ERS). This classification highlighted the significance of histologic subtyping of surgically resected lung ADC in routine pathology practice and provided standardized terminology for lung cancer diagnoses in small specimens.[8, 9] Hence, semiquantitative assessment of histologic patterns in 5% increments on surgical resection specimens was recommended. It was also suggested that ADC should be classified according to the predominant pattern in surgical specimens. Unfortunately, in a substantial number of patients with lung carcinoma, cytology specimens represent the only tissue material available for diagnostic interpretation, and the majority of these patients will not undergo surgical resection. The classification recommended precise subclassification of non–small cell carcinoma (NSCLC) into specific subtypes, such as ADC and squamous cell carcinoma, on small biopsies and cytology specimens. However, it is uncertain whether the morphologic subtyping of ADC proposed for surgical resection specimens can be applied to cytology specimens. The objectives of this study were to assess the applicability of proposed morphologic subtyping of lung ADC on cytologic specimens and to correlate the findings with histologic follow-up and cellularity.
MATERIALS AND METHODS
In total, 133 consecutive surgical resection specimens of a primary lung ADC with preoperative cytology specimens, including computed tomography-guided fine-needle aspiration specimens, bronchial brushings, and/or bronchial washings, were identified from the electronic records of the University of Pittsburgh Medical Center Department of Pathology from 2003 to 2007. Sixty-six patients had 68 diagnostic cytology specimens available for review. One patient had 3 concurrent cytology specimens.
All hematoxylin and eosin-stained sections from surgical resections and all available smears and cell blocks of cytologic specimens were reviewed. The diagnosis of ADC was based on 2004 WHO classification criteria and was subclassified further according to the 2011 IASLC/ATS/ERS classification, Tumors with more than 1 pattern were given a dominant pattern classification and a secondary pattern classification.[1, 8, 10]
The cytology specimens also were reviewed and classified into different subtypes based on the cytologic features that correlated with the 2011 IASLC/ATS/ERS classification to the extent possible. The cytology specimens were subclassified according to the predominant pattern as acinar, solid, papillary, micropapillary, and lepidic. Mucinous variants, and clear cell and signet ring cytologic changes also were documented. The cytologic criteria for classification included the following: the acinar pattern consisted of glandular structures with central lumens, the solid pattern had cohesive cell clusters in a nest-like pattern without identifiable lumens or a papillary configuration, the papillary pattern had papillary structures with fibrovascular cores and occasional intranuclear inclusions, the micropapillary pattern was defined as micropapillary tuffs of small papillae without central fibrovascular cores, mucinous ADC had columnar cells with eccentric nuclei and intracytoplasmic or extracellular mucin, and the lepidic pattern was defined as strips of orderly arranged cells with uniform nuclei and intranuclear inclusions. Clear cell changes were defined as large cells with vacuolated cytoplasm, and signet ring changes were defined as eccentric nuclei with intracytoplasmic mucin.
The tumor cellularity of the specimen on the smears was assessed on the most representative slide. The specimens were divided into 3 groups according to the approximate number of neoplastic cells (<50 cells, 50-200 cells, and >200 cells). Specimens with cell blocks also were classified using the same classification criteria described for the smears. Concordance between the cytologic and histologic specimens was evaluated by comparing the interpretation in the cytologic specimens with the primary pattern in the resection specimens. Additional concordance was evaluated between the cytologic results and the primary or secondary pattern in the resection specimen. The concordance also was compared with tumor cellularity.
Of 66 study patients, 23 (35%) were men, and 43 (65%) were women. The median age was 70 years (range, 44-85 years). There were 28 current smokers, 26 former smokers, 3 never smokers, and, in 9 patients, the smoking status was unknown. The tumor median size was 3.3 cm (range, 0.3-5.0 cm). Surgically resected tumors that grossly measured ≤3.0 cm in greatest dimension were submitted in their entirety. Surgical specimens included wedge resections (n = 15), segmental resections (n = 11), lobectomies (n = 36), bilobectomies (n = 1), and pneumonectomies (n = 2).
Histologic-Cytologic Classification of Lung Adenocarcinoma
Table 1 summarizes the histologic patterns of the 66 surgical resection specimens that included 47 mixed ADCs (71%) and 19 single-pattern ADCs (29%) (Table 1). Overall, the most common single or dominant pattern was acinar (n = 24; 36%), followed by solid (n = 17; 26%), lepidic (n = 9;14%), and papillary (n = 7; 11%). A mucinous variant was observed in 7 ADCs (11%), and clear cell changes were observed in 1 ADC (1%). Similarly, only 1 micropapillary ADC (1%) was identified. The most common combination observed in the mixed ADCs was acinar and solid (n = 11), followed by acinar and lepidic (n = 8), and acinar and papillary (n = 6). The micropapillary pattern was observed as a secondary pattern in papillary, acinar, and solid dominant ADCs.
Table 1. Histologic Patterns of the 66 Surgically Resected Adenocarcinomas
There was sufficient cellularity on cytologic smears in 58 of 66 cases (88%) for subtyping, whereas 8 cases (12%) could not be subclassified on cytology because of scant cellularity or indefinite patterns. Concordant subclassification of ADC between the dominant/single pattern on resection specimens and matched cytology specimens was observed in 26 cases (40%), whereas discordance was observed in 32 cases (48%). (Table 2) The most common patterns in concordant cases included acinar (n = 18; 69%) and solid (n = 6; 23%) followed by 1 papillary ADC (4%) and 1 mucinous ADC (4%) (Figs. 1A-C, 2A-C). Of 26 concordant cases, 14 (54%) had >200 tumor cells, and only 4 cases (12%) had <50 cells. When comparing the concordance between the cytology specimen and the primary or secondary pattern on the resection specimen, the concordance increased to 56%. In the acinar pattern, which had the largest number of cases, 18 of 24 cases (75%) were concordant with the histologically dominant pattern. The solid pattern was concordant in 6 of 14 cases (43%).
Table 2. Summary of Concordant and Discordant Adenocarcinoma-Dominant Patterns Between Surgical Resection Specimens and Cytology Specimens
Cytologic Classification in Discordant Cases [No.]
Concordant, N = 26
Discordant, N = 32
Acinar , papillary 
Solid , papillary 
Acinar , mucinous 
Acinar , solid , papillary 
Acinar , solid , lepidic 
Discordant cases were observed more often in mixed ADCs (n = 23; 72%) than in single-pattern ADCs (n = 9; 28%). The solid pattern (n = 8; 25%) and the lepidic pattern (n = 7; 22%) were most frequently discordant between surgical resection specimens and cytology specimens. Six acinar ADCs (19%) and 6 papillary ADCs (19%) also were misclassified. Discordant cases also included 4 mucinous variants (12%) and 1 ADC with clear cell changes (3%). Solid pattern ADCs were cytologically classified as acinar (n = 7) and papillary (n = 1). None of the lepidic-predominant invasive ADCs were recognized in cytologic smears, and these were classified as acinar (5 cases), solid (2 cases), and papillary (1 case) (Table 2). Similarly, none of micropapillary-dominant ADCs were recognized on cytology. It is noteworthy that 10 of 32 (31%) cytology cases that were discordant with the dominant pattern of invasive ADC were actually concordant with the secondary pattern. These cases were most often observed in ADCs with an acinar pattern (Fig. 3A-C). In total, 21 of 32 discordant cases (66%) had <200 cells in the cytologic smears for evaluation, including 17 cases (53%) with between 50 and 200 cells and 4 cases (13%) with <50 cells. However, there was no statistically significant difference in tumor cellularity between concordant cases and discordant cases.
We also compared cell blocks and surgical specimens, because 54 of 66 cases (82%) had cell blocks available. However, 15 cell blocks (28%) had no or very few tumor cells and were suboptimal for any further morphologic evaluation. Of the remaining 39 cell blocks (72%), 17 (44%) were concordant, and 22 (56%) were discordant with the surgical resection specimens. Similar to cytologic smears, ADCs with an acinar pattern had the highest concordance between cytology and surgical specimens. In contrast to cytologic smears, 2 lepidic-dominant ADCs were identified in cell block sections; however, not all lepidic-dominant ADCs were recognized. Similarly, the mucinous variant of ADC was frequently discordant. Only 12 cell blocks (22%) contained >200 tumor cells. Although there was a trend toward less tumor cellularity in cell blocks than in cytologic smears, that difference was not statistically significant.
The observations that have linked the histology of NSCLC with the efficacy of different therapies, including epidermal growth factor receptor-tyrosine kinase inhibitors (EGFR-TKIs), bevacizumab, and pemetrexed, have resulted in the need to classify NSCLC into specific subtypes, such as ADC and squamous cell carcinoma.[11-13] This is particularly the case with small biopsies and cytology specimens, which are the most common specimens and are often the only tissue specimens available in patients with lung carcinoma. The recent IASLC/ATS/ERS lung ADC classification has standardized terminology for the diagnosis of NSCLC subtypes in cytology and small biopsies and stressed the importance of histologic subtyping of ADC in histologic specimens because of the clinical and prognostic importance. Since the publication of that classification, many studies have investigated the feasibility of subtyping NSCLC into major histologic types on small specimens.[14-16] It has been observed that the diagnostic accuracy of NSCLC subtyping is influenced not only by pathologists' skills and experience but also by the procedure type and the size and tumor cellularity of the sample.[17, 18] Although the focus in the cytologic literature has been on the subtyping of NSCLC into broad categories of adenocarcinoma or squamous cell carcinoma, the IASLC/ATS/ERS lung ADC classification goes even further to subclassify ADCs into different morphologic patterns because of the association with prognosis, and this has been investigated predominantly in histologic material.[19-21] However, in the cytology literature, only a few studies have attempted to describe cytomorphologic features of adenocarcinoma subtypes predominantly using the 2004 WHO classification.[1, 22, 23] To our knowledge, this study is the first to address the cytologic-histologic correlation of the proposed IASLC/ATS/ERS lung ADC classification.
Although it has been reported that the subclassification of ADC on surgical specimens has poor reproducibility among pulmonary experts when subclassifying challenging cases, educational tools reportedly have increased reproducibility among surgical pathologists. Our study suggests that the subclassification of pulmonary ADC on cytology specimens may be feasible; however, because of the lack of well established guidelines for the subtypes and limited available data, there are some limitations. The concordance in this series was 40% when the subtyping performed on cytologic material was compared with the primary histologic pattern on matched resection specimens (either the pure ADC subtype or the dominant pattern in mixed ADC), and concordance increased to 56% when compared to the primary or secondary patterns on matched histologic specimens. In the acinar pattern, which was the most common in our study, 75% of the cases were concordant with the primary histologic pattern. These findings illustrate that, because of sampling, the subtyping based on cytologic material may not always represent the dominant pattern, particularly in mixed ADCs, and can be challenging because of the lack of well established criteria and the cytomorphologic overlap in some subtypes. In addition, some patterns, such as the acinar pattern, appear to have higher concordance than other patterns. Furthermore, the concordant cases were more likely to have >200 cells on the cytologic material evaluated, in contrast to the discordant cases, in which 66% had <200 cells for evaluation, which argues that optimizing sampling also may improve the ability to accurately determine the subtype.
The most common subtype of ADC reflected in the concordant cases was acinar. This may be explained in part by the finding that the acinar pattern is relatively easy to recognize, given the similarity to the findings classically described on histology with distinct lumens and glandular formation. The second reason may be that the acinar subtype of ADC was the most common pattern in our patient population, as demonstrated previously, and is 1 of the most common subtypes reported overall in mixed ADC. Patterns, such as lepidic and micropapillary, and cytologic changes, such as clear cell and signet ring, were not accurately or consistently recognized in the cytologic smears; however, there also were limited numbers of cases, and the extent of changes varied in the primary tumor, limiting interpretation. In addition, the cytomorphologic overlap between some patterns of ADC may limit the recognition of some subtypes, such as the lepidic pattern, in cytology smears, as reported previously.[26, 27] The lepidic subtype has been described as having uniform cells with bland nuclei; abundant, lacy cytoplasm distributed in single cells or forming papillary clusters; and intranuclear inclusions.[26, 27] However, some of these features can be identified in the other patterns, particularly in the acinar and papillary subtypes. Also, both the lepidic and papillary subtypes of ADC can have strips of cells on the cell block, which can appear similar and can overlap with the acinar pattern. This observation further supports the IASLC/ATS/ERS recommendation that the diagnosis of ADC in situ should be avoided in cytology specimens.
There was an almost equal distribution of all subtypes in the discordant cases, with the solid type being the most common. This may be because of the difficulty in distinguishing the architectural features in cytologic smears and the cytomorphologic overlap some cases. Furthermore, the percentage of the individual ADC patterns identified on histology of the primary tumor, in addition to the limited sampling by cytology, makes it uncertain whether the high discordance rate reflects that the observed cytomorphologic features are truly discordant from the dominant histologic pattern or simply that a less predominant pattern was sampled. Given that 31% of the discordant cases were actually concordant with the secondary pattern, sampling may be responsible for some of the discrepancies. It is interesting to note that the micropapillary pattern was not well recognized in our cytology specimens, which may reflect either a sampling issue or the difficulty distinguishing it from more common counterparts, like acinar, solid, and papillary patterns, when present in a specimen with limited cellularity or when it comprises a smaller proportion of the tumor. In our 4 cases with a micropapillary pattern, it was observed primarily as a secondary pattern comprising a minority of the specimen, and these specimens had low cellularity. However, micropapillary tufts have been reported in cytology specimens of other patterns of ADC (eg acinar, papillary, solid) and reportedly were not specific for the diagnosis of ADC with a micropapillary pattern. Thus, additional studies with a larger number of these cases and a higher proportion of the micropapillary component may be helpful to increase the ability of cytopathologists to recognize this prognostically important subtype of ADC.
The current study also indicates that the subclassification of ADC may differ in cytologic smears and cell blocks. Our study demonstrates that more smears had adequate cellularity for subclassification than cell blocks (86% vs 59%) and that some subtypes were recognized better in cell blocks, such as the lepidic pattern. Previous studies have reported that the enhanced morphologic clues provided by cell block specimens help classify ADC. Cell block specimens with optimal cellularity may correlate better with the described histologic features given the presence of tissue fragments, similar fixation, and similar staining with hematoxylin and eosin; however, sufficient material must be acquired. For this study, we used archived specimens prepared when optimization of the cell block was not as important as it is in today's modern era; thus, future studies may be needed to determine whether enhanced cellularity in cell block specimens increases subtyping concordance with matched histology specimens.
In summary, the morphologic subclassification of lung ADC on cytology specimens is challenging and largely depends on the sufficient cellularity of cytologic preparations. Some patterns, such as the common acinar pattern, appear to be more accurately subclassified on cytology. However, the identification of other patterns in cytologic specimens, such as the micropapillary and solid patterns, which are important because of their prognostic significance, can be extremely difficult and unreliable. Further studies are necessary to better define the cytologic criteria for the subtypes of ADC in different cytologic preparations and to help delineate the significance of cytology subtyping using these recently developed classification schemes for lung cancer.