Analysis of microRNA Expression in Liquid-Based Cytology Samples may be Useful for Primary Lung Cancer Diagnosis

Yusuke Araki, MD; Koji Arihiro, MD, PhD; Kakuhiro Yamaguchi, MD, PhD; Shinjiro Sakamoto, MD, PhD; Yasushi Horimasu, MD, PhD; Takeshi Masuda, MD, PhD; Shintaro Miyamoto, MD, PhD; Taku Nakashima, MD, PhD; Hiroshi Iwamoto, MD, PhD; Kazunori Fujitaka, MD, PhD; Hironobu Hamada, MD, PhD; Noboru Hattori, MD, PhD


Am J Clin Pathol. 2021;156(4):644-652. 

In This Article


In this study, we showed that 4 miRs—miR-21, miR-31, miR-182, and miR-183—extracted from lung tissue were significantly upregulated in comparison with noncancerous lung tissue and that the expression of the same 4 miRs extracted from cytologic samples of lung cancer cases obtained by bronchial brushing was higher than that in non–lung cancer cases. These results were consistent with those reported in previous studies.[12,26,28] In particular, for the samples that were obtained preoperatively from lung cancer cases but judged cytologically as benign and that did not contain obvious carcinoma cells, expression of the 4 miRs was significantly higher than that than in samples obtained preoperatively from noncancerous cases that were cytologically judged as benign.

To the best of our knowledge, there have been no reports regarding the evaluation of miRs using LBC samples obtained by bronchoscopy. Interestingly, although the expression of miRs extracted from cancer cases was higher than that of miRs extracted from noncancer cases, there were almost no differences in expression of the investigated miRs among cytology samples judged as benign, indeterminate, suspicious for malignancy, or malignant, regardless of the presence of malignant cells. In general, when bronchoscopy is performed, we obtain biopsy samples and cytology samples by brushing, while checking the location of the tumor by chest X-ray imaging. Usually, we obtain not only tumor cells but also noncancerous epithelial, stromal, and blood cells. For this reason, the cytology samples contained various cells. Samples obtained from surgically resected lung cancer tissue contained higher proportions of tumor cells than cytology samples obtained from lung cancer cases by bronchoscopy because cytologic samples contain various cells, and the proportion of these cells varied between samples. In other words, among the cases judged as malignant by cytology, some samples contained only a few tumor cells, whereas other samples contained many normal epithelial and/or blood cells. It is probably difficult to identify differences in miR expression between samples containing few tumor cells or many normal cells and in samples obtained from nontumor cases. We speculate that total RNA extracted from samples containing many blood cells may also contain RNA derived from inflammatory cells or RBCs; consequently, these may affect the detection of miRs. Some circulating miRs can be affected by hemolysis.[30] Furthermore, the quantities and types of miRs vary among the blood components involved.[31]

Pavel et al[32] showed that some miR expression levels in normal bronchial epithelium of smoking patients were higher in patients with lung cancer than in patients without cancer. Although the 4 miRs examined in the present study were not included in Pavel and colleagues' earlier study, the results of our research suggested that the expression of certain miRs might be affected in noncancerous cells by the presence of lung cancer. In the present study, it was considered difficult to eliminate all of these effects in samples composed of various cells.

In addition, although miR-182 and miR-183 are known as oncogenes, miR-182 and miR-183 are also reported to be tumor suppressors.[33,34] Grofu et al[34] showed that miR-183 regulates migration and invasion via ezrin in lung cancer cells and identified miR-183 as a potential inhibitor of metastasis. Levels of other miRs may also be affected by various factors, including clinical ones, and there is a possibility that this may be why it was difficult to identify significant differences in miR expression. Furthermore, the noncancerous samples may also have affected these results. In the present study, normal controls included cytologic samples gained not only from benign nodules but also from bronchoalveolar lavage of interstitial pneumonia, including idiopathic pulmonary fibrosis, chronic hypersensitivity pneumonitis, sarcoidosis, and so on. In general, because bronchoalveolar lavage of these interstitial pneumonias contains various inflammatory cells, the detected expression levels of miRs can be influenced by the presence of inflammatory cells.

To our knowledge, no reports have explored the expression of miR extracted from cytologic samples fixed with Cellprep and using automated extraction by Magcore. Cellprep is one of the LBC systems developed by Roche Tissue Diagnostics; Cellprep liquid is a methanol- and formalin-free fixative. In a previous study, fixation time did not affect the detection sensitivity of miRs.[35] It has been suggested that Cellprep liquid does not preclude the feasibility of miRNA analysis, and this method can be easy to apply to routine samples.

This study has certain limitations. First, there were many samples in which miR expression was undetectable. We considered 2 reasons for this problem. First, the detected expression levels of miRs extracted by Magcore were lower than those detected by an alternative manual extraction kit (Supplemental Figure 1). It may be suggested that the extraction efficiency for small RNAs, such as miRs, using Magcore was lower than when a manual extraction kit was used; therefore, it was considered that use of the manual extraction kit would be preferable. Another reason is that no specific kit was available for LBC samples. Because we assumed that the RNA extraction kit for cultured cells could be used for LBC samples, we extracted miRs from LBC samples using an RNA extraction kit intended for cultured cells. Although it seems that the extraction method used in this study was not optimal, Magcore is a fully automated extraction system, and the inherent simplicity of Magcore is a strong point. Indeed, although the expression of certain miRs was undetectable by RT-qPCR for some samples, many samples were successfully examined in this study, which suggests that the Magcore Total RNA Cultured Cells Kit could be useful for cytologic samples. Second, the question arises of whether the choice of internal controls was reasonable. Although RNU6B is one of the reference genes most frequently used as an internal control, use of endogenous controls for miRs extracted from whole blood remains controversial.[36] In the present study, although we thought that epithelial cells had a stronger influence than blood cells, inappropriate choices of internal controls can introduce bias in the comparison of expression of miRs.

In summary, we demonstrated that 4 miRs could be extracted from LBC samples for classification of their expression levels. Furthermore, the 4 miRs studied exhibited significantly higher expression levels in patients with lung cancer compared with patients without cancer. There is an opportunity to improve lung cancer diagnostic rates by using bronchoscopy and cytology and by performing RT-qPCR to examine miR expression levels.