Adequacy of pleural fluid cytology for comprehensive molecular analysis of lung adenocarcinoma: Experience of a large health-care system

INTRODUCTION

About 15% patients with lung cancer present with pleural effusion,^[1] which is classified as stage IV disease. The 5-year survival of patients with malignant pleural effusion (MPE) is very poor.^[2] However, achievements in molecular diagnostics and precision medicine have brought about a significant impact in improving overall disease-free survival rates. Pleural fluid examination is a minimally invasive procedure in the diagnosis and work up of patients with malignant effusions.^[3,4] Pleural effusion can be the primary manifestation of an occult or metastatic tumor in a small proportion of patients, or can be an early sign of disease recurrence.^[5] The performance and diagnostic utility of this procedure correlates with the tumor volume in the lung and with the histologic subtype. Also compared with needle biopsy, pleural fluid has a more robust diagnostic yield as the cell population is representative of a larger surface area.^[6] Diagnosis is based on cytomorphology, in conjunction with ancillary studies including immunohistochemistry and molecular studies, the results of which have direct therapeutic implications for the patient.^[7]

Pleural fluid evaluation is also an effective modality for identifying actionable genetic mutations to guide therapy. Up to 20% of lung adenocarcinomas contain activating mutations in epidermal growth factor receptor (EGFR), which can be effectively targeted by specific tyrosine kinase inhibitors like erlotinib. Similarly, up to 5% of lung adenocarcinomas harbor chromosomal rearrangements of the anaplastic lymphoma kinase (ALK) gene, with resultant overexpression of the ALK protein that can be effectively managed with ALK inhibitors like crizotinib. Although antibodies for immunohistochemical detection of these genetic alterations are available, they suffer from lack of sensitivity and specificity. Hence, the guidelines from National Comprehensive Cancer Network, the College of American Pathologists (CAP) and the International Association for the Study of Lung Cancer strongly recommend molecular detection of actionable EGFR mutations and ALK rearrangements in patients with non-small cell lung carcinomas (NSCLC).^[8,9] This detection not only leads to initiation of specific therapy but also predicts an inferior response to immune checkpoint inhibitor therapies. The first case of EGFR mutation in NSCLC was detected by polymerase chain reaction (PCR) in a pleural effusion specimen by Huang et al. in 2005.^[10] This was followed by other reports of therapy-responsive EGFR mutations.^[11-13] In addition, the molecular detection of BRAF V600E mutation, MET amplification, NTRK1 gene fusions, and ERBB2 exon 20 insertion mutations has also been shown to be actionable genetic alterations, and their detection in pleural effusion fluid can also lead to specific and effective therapies. Thus, pleural fluid is an ideal specimen for formulating targeted therapies for individual patients.

We wanted to establish the characteristics of an adequate specimen that would yield reliable results with current molecular testing platforms. The main variables studied for this purpose were the specimen volume, and total as well as relative tumor cellularity, as assessed on a cell block preparation. We also wanted to determine a cutoff for a minimum volume of pleural fluid that would be diagnostically meaningful, and to study if increasing the volume of the specimen over 100 ml increases the diagnostic yield significantly.

MATERIAL AND METHODS

This retrospective study (slide and chart review) was conducted after approval by the Institutional Review Board. A review of 500 MPEs, from pulmonary and non-pulmonary sources, was undertaken over a 4-year period from January 2015 to December 2018.

RESULTS

Five hundred cases of MPEs were reviewed that included predominantly metastatic carcinomas from various primaries, and few cases of sarcoma, lymphoma, melanoma, and mesothelioma. Of these 500 positive cases, 44 cases (from 42 patients) were positive for primary lung adenocarcinoma. Cell block slides were available for 41 of these 44 cases. The tissue of origin was determined using an appropriate immunocytochemistry panel that included CK7, CK20, p40, TTF-1, and Napsin A to confirm pulmonary origin. For cellular effusions with mixed morphotypes of cells, additional panels of immunostains were performed on serial cell block sections using a modified version of the SCIP approach (subtractive coordinate immunoreactivity pattern).^[15,16]

Molecular analysis was performed on all 44 cases; however, only 41 of these cases that had a concurrent cell block were included in the study. A complete NGS was performed on 35 out of 41 specimens. Two of these 35 specimens were deemed inadequate due to low cellularity and did not yield conclusive results. Thus, a complete NGS result was available for 33 of 35 specimens (94.3%). Individual molecular testing was performed on six specimens from 2015 before NGS testing was widely available or requested by the oncologists. The individual testing included PCR for detection of point mutations in EGFR and KRAS genes, and FISH for detection of rearrangements of ALK and ROS1 genes.

The minimum volume submitted was 10 ml and the maximum volume was 120 ml. The submitted volumes of the two inadequate specimens were 10 ml and 20 ml. The minimum volume of fluid that was diagnostically adequate, irrespective of the proportion of tumor cells, was 30 ml.

The minimum total cellularity, counted as number of tumor and non-tumor cells in the entire cell block, was 55 cells, and the maximum total cellularity was estimated to be over 10,000 cells. The total cellularity of the two inadequate specimens was 55 cells and 300 cells. We divided our specimens into four groups based on the total cellularity: 1–100 cells, 101–500 cells, 501–1000 cells, and more than 1000 cells.

The proportion of tumor cells in the cell block versus other benign non tumor nucleated cells was assessed as tumor to non-tumor ratio (T: NT). The specimens were divided into three groups for analysis based on the proportion of tumor cells being <20%, between 20% and 50%, and more than 50% of total nucleated cells.

Of the two inadequate specimens, the one with a total cellularity of 55 cells had a T: NT proportion of more than 50%, while the specimen with a total cellularity of 300 cells had <20% tumor cells.

The minimum number of malignant cells recommended for molecular testing, irrespective of the tumor cell proportion, is 500 cells. However, a significant proportion of our specimens (n = 18) had a total cellularity between 101 and 500 cells, but these were subjected to molecular analyses, anyway. Of these 18 specimens, only one specimen was inadequate that had a proportion of tumor cells of <20%. Although our specimen numbers were small, the proportion of tumor cells, rather than the total cellularity, seems to be more important for specimen adequacy. The distribution of total cellularity and tumor cell proportion across the specimens is shown in Table 1. Correlation of volume with cellularity is shown in Table 2.

DISCUSSION

The role of pleural fluid cytology in the diagnosis and staging of primary lung adenocarcinoma cannot be underestimated. It is a relatively non-invasive procedure which reduces healthcare costs, guides tumor staging and allows for early and targeted therapeutic interventions for individual patients, which is the sine qua non of “personalized” or “precision” medicine. Our study represents an institutional experience of studying the adequacy and other characteristics of pleural effusion fluids for the molecular diagnostics in lung adenocarcinoma.

A few studies have been performed to investigate the adequate volume of pleural fluid for a morphologic diagnosis of malignancy. Thomas et al. determined that a minimum volume of 25 ml is required for a positive diagnosis, and that volumes over 50 ml do not significantly increase the diagnostic yield.^[17]

This volume was utilized for morphologic, histochemical, and immunohistochemical assessment. However, molecular studies were not performed on these specimens.

Similar studies by Ong et al.^[18] revealed that the diagnostic yield of pleural fluid increased with examination of multiple specimens, and with the concurrent examination of a pleural biopsy specimen. However, this study did not specifically address the ideal minimum volume for examination. Similar to the study by Thomas et al., molecular studies were not performed.

In the most recent collaborative guidelines issued by the CAP, Roy-Chowdhuri et al.^[19] performed an analysis of seven studies that attempted to report on the ideal volume of pleural fluid to avoid a false negative diagnosis. However, similar to the studies by Thomas et al., and Ong et al., these studies emphasized on the morphologic diagnosis of malignancy, without taking into consideration ancillary molecular testing. In one of these studies, a volume as low as 10 ml was considered adequate for a solely morphologic documentation of malignant cells.^[20] Two of the studies in their analysis concluded that the positive yield of a fluid did not significantly increase with collection volumes exceeding 50–75 ml.^[21,22] Thus, though the recent CAP guidelines opine that “the proceduralist should send as much fluid volume as reasonably attainable,” they could not definitively recommend a maximum volume for examination. A minimum volume of 50–100 ml was suggested, but not recommended.

MPEs have been used in several studies to assess oncogenic driver mutations in pulmonary adenocarcinoma. Complete NGS was performed on 33 specimens. Multiple studies have performed oncogene analysis in pleural effusions; however, they used PCR based platforms for somatic mutation profiling.^[1,2] However, these studies did not determine the adequacy criteria of the submitted pleural fluid specimen for molecular testing. We performed molecular testing for detection of point mutations in EGFR and KRAS genes, and of rearrangements of ALK and ROS1 genes in six cases from 2015. However, targeted NGS testing works as a unifying platform for the detection of multiple genetic alterations at the same time and reveals several more potentially actionable mutations for therapeutic targeting.^[23,24]

The volume of pleural fluid submitted by clinicians ranged from a minimum of 10 ml to a maximum of 120 ml. The minimum cellularity (referring to one histologic section of cellblock) was 55 cells, and the maximum cellularity was over 10,000. All our specimens with a volume of over 30 ml, a cellularity of over 100 cells per cell block section, and a tumor cell proportion of over 20%, were adequate.

MPEs tend to comprise varying amounts of benign reactive cells, namely, mesothelial cells, histiocytes, and leukocytes. If the proportion of these cells is significantly more than that of malignant cells, molecular methods are likely to fail or give erroneous results as a result of intermixing of benign and malignant DNA. Thus, we propose that counting the tumor and non-tumor cells in one section of the cellblock and estimating a proportion is of importance to avoid a false negative result on NGS platforms. Cell blocks with a tumor cell proportion of <20% should not be forwarded for molecular studies and a repeat specimen can be requested.

We examined the major possible determinants influencing the specimen adequacy for molecular studies, including volume, overall cellularity, and tumor percentage. MPE volume does not have a direct correlation with total cellularity or tumor cell content. Furthermore, the effusion volume of cases that were deemed inadequate for molecular studies was not significantly different from the volume of those cases that were considered adequate and underwent full molecular profiling.

CONCLUSION

We conclude that of the three major determinants, effusion volume is the least important. In paucicellular specimens lacking adequate absolute numbers of tumor cells, or in cellular specimens with a predominance of non-tumor cells, increasing the submitted volume of effusion fluid to more than 100 ml will not lead to an increase in molecular diagnostic yield.