Subtyping of pulmonary adenocarcinoma in cytologic specimens

The Next Challenge


  • Andre L. Moreira MD, PhD

    Corresponding author
    1. Department of Pathology, Memorial Sloan-Kettering Cancer Center, New York, New York
    • Corresponding author: Andre L. Moreira, MD, PhD, Department of Pathology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue. New York. New York, NY 10065; Fax: (212) 639-6218;

    Search for more papers by this author

  • See referenced original article on pages 629–37, this issue.


The last World Health Organization classification recognized that pulmonary adenocarcinomas are heterogeneous and hinted at the possibility that different patterns of adenocarcinoma were associated with certain prognostic value. This concept was expanded and is now the basis for the new International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification.

The recent therapeutic revolution in pulmonary carcinomas brought many challenges to the way pathologists diagnose and classify these tumors. Not long ago, many pathologists would classify lung cancers into small cell and nonsmall cell lung carcinomas (NSCLC). This simple classification was well received by clinicians and was well suited to the available therapeutic option, because all NSCLC were treated similarly, that is, surgery for early stages and a similar regimen of chemotherapeutic drugs for systemic use in advanced stages.

The first signs of changes to come emerged from clinical trials indicating that new drugs, such as pemetrexed, had a differential effect on NSCLC. The drug was only effective in patients with the histologic diagnosis of non-squamous cell carcinoma.[1, 2] Studies with yet another drug (bevacizumab) demonstrated severe drug toxicity in patients with the histologic diagnosis of squamous cell carcinoma.[3] This was the birth of a new era of histology-based therapy for patients with lung cancer, which brought pathologists to the center stage in the clinical management of these patients. Almost simultaneously, the concept of target therapy, which is based on the use of drugs that target specific molecular alterations in the tumors, was solidified by the discovery of sensitizing mutations in the epidermal growth factor receptor (EGFR) gene in a subset of patients with pulmonary adenocarcinoma. Patients with EGFR mutations responded very well to specific EGFR inhibitors (erlotinig/gefitinib).[4-7] Since then, other targetable molecules have been identified, mostly in adenocarcinomas such as EML4-ALK rearrangement[8] and BRAF[9] mutations, among others. Both mutations can be treated with specific drugs.[9-11]

Therefore, the subclassification of NSCLC into adenocarcinoma and squamous cell carcinoma is essential for patient management, because this will determine the drug regimen for systemic therapy as well as the need for molecular characterization of the tumor.[12] Considering that the vast majority of patients with lung cancer present with advance stages of the disease, for which surgery is not a therapeutic option, the focus of tissue-based therapy turned to cytology and small biopsies specimens. The subclassification of NSCLC in small biopsy specimens, including cytology, still is a tremendous challenge to pathologists world wide.[13] Although subclassification is possible in most tumors with clearly defined differentiation features, the challenge persists for poorly differentiated NSCLC.[14, 15] In these cases, the use of immunohistochemistry (IHC) has been evaluated to help sort out adenocarcinomas from squamous cell carcinomas. The use of IHC posed yet another challenge for pathologists, and especially for cytopathologists, on the need to balance an accurate diagnosis and to preserve tissue for molecular characterization in small biopsy specimens.[12] Many pathologists responded to this new challenge and produced several studies that formed the basis for a minimal panel of IHC markers that appears to have gained acceptance among most practitioners for the purpose of subclassifying NSCLC while preserving tissue for further studies.[16-20]

The role of small biopsy and cytology for the management of lung cancer patients has been highlighted in the new International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society (IALSC/ATS/ERS) classification of adenocarcinoma.[21] The 2004 World Health Organization (WHO) classification of lung cancer was based exclusively on resection specimens[22]; however, the new version is expected to address tumor classification in small biopsy specimens as well. Because most of the changes in therapy appear to involve patients with adenocarcinoma, there has been a renewed interest in the study of these tumors. The last WHO classification recognized that pulmonary adenocarcinomas are heterogeneous and hinted at the possibility that different patterns of adenocarcinoma were associated with certain prognostic value. This concept was expanded and is now the basis for the new IASLC/ATS/ERS classification.

The new classification, which is still based on resection specimens, proposes that pulmonary adenocarcinomas should be classified by the predominant pattern present on the histologic section. A lepidic-predominant tumor has an excellent prognosis compared with tumors with a predominant solid or micropapillary pattern.[23, 24] This concepts of histologic features and prognosis is relatively new in the field of lung cancer; therefore, it is still not incorporated into patients' management plans, contrary to what happens in tumors from other organs, such as the prostate and breast. There are very few studies that seem to suggest a differential response to adjuvant therapy in patients with high-grade tumors (predominant solid and micropapillary tumors).[25, 26] Tumor grading and patient management appear to be the next challenge for pathologists working in the field of pulmonary carcinoma. Tumor grading in small biopsy specimens has not been validated.

An article in the current issue of Cancer Cytopathology by Rodriguez et al asks a very interesting question.[27] Can we subclassify pulmonary adenocarcinoma in cytologic specimens according to the new IASLC/ATS/ERS classification? The authors suggest that, although it might be possible, it is not going to be without challenges. From their article, it appears that, between the cytology subclassification and the predominant patterns in the resection specimen, perfect agreement was observed in only 40% of cases. The correlation was best for acinar pattern, the most common pattern in adenocarcinomas, but was poor for the solid and micropapillary patterns. Similar observations were made in the pre-IALSC/ATS/ERS classification era by Rudomina et al using material from direct smears and liquid-based cytology.[28] In the latter report, the authors demonstrated that the predictive value for the presence of acinar clusters in cytologic specimens as an indication of acinar pattern in the resection specimen was of 94%; however, sensitivity was 65%, and specificity was 66%. In contrast, the predictive value for the presence of micropapillary clusters in cytologic specimens to indicate the presence of micropapillary in the resection specimen was 64%, with 63% sensitivity and 51% specificity. The low sensitivity and specificity for cytology cell clusters indicate that this is an unreliable method for prognostication based on defined patterns. In addition, Rudomina et al could not identify a cytologic correlate to the solid pattern. It is noteworthy that papillary clusters characterized by fibrovascular cores were highly sensitive (83%) for the presence of papillary pattern on the resection specimens, but suffered from specificity (15%). This suggested that the absence of papillary clusters in cytologic material cannot exclude the presence of a papillary pattern in the surgical material, thus indicating that sampling plays a big role in cytology, which also is confirmed in the article by Rodriguez et al in this issue.[27]

Rodriguez et al also evaluated the role of cell block material compared with direct smears and observed a similar correlation; that is, a cell block preparation was as reliable as a direct smear in predicting the predominant pattern in the resection specimen. They also evaluated the cellularity of smears and cell blocks and concluded that specificity for pattern recognition was related to the number of cell clusters present in the preparation. The role of cell block preparation in cytologic diagnosis is increasing! Cell blocks are a very useful way to preserve cells for molecular and IHC studies, especially in pulmonary pathology.[12]

One of the most import observations from the article by Rodriguez et al is that the rate of correlation between cytologic subtyping of adenocarcinoma and resection specimens was increased from 40% to 56% if either the primary or the secondary pattern (that is, the second most predominant pattern of adenocarcinoma observed in the resection specimen) was taken into consideration. His indicates that, as discussed by the authors, despite the classification in the resection specimen by the predominant pattern, the sampling nature of the cytology material does not guarantee that the predominant pattern will be sampled. Therefore, the predominant pattern cannot be used as a classifier for adenocarcinomas in small biopsy specimens.

This leads us to a very pertinent question: Can we apply the same concept of patterns observed in resection specimens as a prognostic indicator to cytologic material? If the defined prognostic value of the 5 recognized patterns of adenocarcinoma (namely, acinar, papillary, solid, micropapillary, and lepidic) is to be applied to cytology specimens, then how will the issue of sampling be resolved? Should all patterns of cell clusters present in the cytology material (smear and cell block) be reported in hopes of providing a prognostic indication to the treating physician?

The current IASLC/ATS/ERS classification uses the concepts of tumor type and tumor grade interchangeably; however, these concepts need to be evaluated better in cytology and small biopsy specimens because of the sampling issue described above. If pathologists are required to subclassify adenocarcinomas in cytologic material according to patterns as a prognostic indicator, then the corresponding patterns of cell cluster arrangements in the histologic counterpart must be defined first. Cytopathologists must agree on what constitutes the acinar, papillary, and lepidic patterns in cytologic material. It appears from the few publications on this subject that these 3 patterns are the best ones represented in cytologic specimens.[28, 29] What to do with the lack of a cytologic correlate for high-grade patterns, such as solid and micropapillary, as has been described previously[28] and is confirmed in the article by Rodriguez et al[27]

If cytologic patterns are not adequate to provide a prognostic correlate, then is there a role for nuclear grading in cytology material? Sigel et al have addressed this issue and have suggested that nuclear grade of pulmonary adenocarcinoma can be used to provide prognostic information in cytology specimens.[30] Those authors evaluated several nuclear features and proposed that nuclear size, contour, and chromatin pattern can be used to stratify stage I pulmonary adenocarcinomas into a 2-tier grading system that correlates with histologic grading and disease-free survival. Is an association of nuclear grading and pattern recognition the next step in the field?

These are just a few questions that emerge from the article by Rodriguez et al in the current issue of this journal. I truly hope that pathologists/cytopathologists will continue to respond to the challenges to come with the same purpose and determination that they have demonstrated with previous issues in lung cancer pathology. Hopefully, the current article by Rodriguez et al can serve as an inspiration for more work that needs to be done in the fight against lung cancer. It is indeed an amazing time to be a pathologist/cytopathologist!


No specific funding was disclosed.


The author made no disclosures.