Clinical study on primary screening of oral cancer and precancerous lesions by oral cytology

Head and neck cancer is one of common malignancies in the world, and the most common histopathological type is squamous cell carcinoma (SCC). Several patients die every year because of advanced oral SCC [1]. Conversely, studies have reported that early detection and treatment of oral SCC can reduce mortality and morbidity and increase the likelihood of complete recovery [2, 3]. Therefore, it is important to use the simplest and the most accurate method that can to detect early-stage abnormalities in oral mucosal cells. An example of such method is exfoliated mucosal cytology, which involves making a diagnosis from minimally invasive oral mucosal cells [4].

Exfoliative cytology is the microscopic examination of shed or desquamated cells from the mucus membrane, and it is a simple, safe, and reliable approach. Exfoliative cytology consists of the conventional method and the liquid-based cytology (LBC) method [5]. The conventional method involves chair side scraping of the oral mucosa and then smearing it directly onto a glass slide. This method requires a suitable technique because the morphology of the collected cells would change if handled improperly. Conversely, LBC is a technique in which cells are scattered in a fixative liquid to produce a thin layer of cells on the slide. Therefore, the LBC method has been widely used because of its advantage of not requiring complicated operations on the chair side.

However, the current gold standard for diagnosing oral epithelial dysplasia and cancer is not exfoliative excision cytology; rather, it is resection biopsy or histological examination of surgical specimens [6, 7]. Unfortunately, excision biopsy is an invasive diagnostic method, and scrape cytology is suitable for the screening of pathological conditions considering minimal invasiveness. However, the diagnostic capabilities of two types of oral exfoliative excision cytology, the conventional and LBC methods, remain unclear. Furthermore, the diagnostic accuracy of the oral scraping cytology compared with the histopathological diagnosis has not been examined in detail. Therefore, it is important to consider whether the two types of cytology are sufficient to be used as a standard method for the diagnoses of suspicious oral lesions.

This study was conducted to compare the histological diagnostic accuracy between the conventional method and the LBC method and to clarify the effectiveness of cytology.

The results of our study demonstrate that the LBC method is superior to the conventional method for cytodiagnosis. The LBC method was especially superior for OLSIL and OHSIL for the positive predictive value. Conversely, because of a certain number of false-positives and false-negatives in cytodiagnosis, it is important to make a comprehensive diagnosis considering the clinical findings.

In this study, there were three cases of inappropriate cytological specimens in the conventional method. Sekine et al. [7] reported that 22.7% of unsuitable specimens were produced by the conventional smearing technique. They reported that the unsuitable samples in the conventional method were those that were considered to be inadequate for cytodiagnosis due to strong cell or air-drying artifacts. Our inappropriate specimens were similar, but the number of inappropriate specimens was very small in our study. We considered that this small number of inappropriate specimens occurred as the slides were created on the chair side by a specialist cytology technician when preparing samples using the conventional method. Conversely, in the LBC method, there were no cases of insufficient sample processing, and stable sample preparation was possible. This is consistent with previous reports showing that the LBC method obviously results in fewer inappropriate specimens [11]. Thus, it is important to understand and standardize the features and equipment of cytology collection methods to improve the quality of cytology slides.

Regarding the diagnostic efficacy of the methods, the LBC method has been demonstrated to be more effective than the conventional method [12]. Although past study have reported higher sensitivity (conventional/ LBC;85.7%/ 95.1%) and specificity (conventional/ LBC;95.9%/ 99.0%)for the LBC method than for the conventional method [6], another study reported that sensitivity (conventional/LBC; 96.3%/97.5%) of the LBC method was particularly good, whereas its specificity (conventional/LBC; 90.6%/68.7%) was reduced [13]. Our results indicated an increase in both sensitivity (conventional/LBC; 79.3%/81.8%) and specificity (conventional/LBC; 69.8%/85.4%). As the results were judged using the same criteria at the same facility, they were valuable demonstrating the usefulness due to differences in cytodiagnosis preparation. Conversely, there are also some very interesting results in our study. The positive predictive value was not different for NILM and SCC between the conventional method and the LBC method. However, a large difference was observed in the positive predictive value between OLSIL and OHSIL. Especially for OHSIL, it was a large difference. We speculate that this is due to a decrease in the number of negative cases misclassified as OHSIL and an increase in the number of detections of CIS and severe dysplasia. The LBC method contributes to the detection without missing CIS and severe dysplasia. These diseases are susceptible to SCC and are eligible for treatment, and appropriate detection of these diseases is clinically useful.

Despite the improvement in efficacy with the LBC method, there were a certain number of false-positives and false-negatives in this study. There were 12 cases of false-negatives in the LBC method and five cases in the conventional method (false-negative rate; LBC method 18.2%, conventional method 20.7%). When we examined the histology of cases diagnosed as NILM or OLSIL but with SCC, all the histological diagnoses of SCC were found to be well-differentiated SCCs. In addition, all cases were accompanied with well-differentiated keratinocytes lacking strong atypia on the surface. In this study, whether a problem existed with the site of cytology collection cannot be examined. The current diagnostic criteria for the JSCC are limited for the evaluation of atypical superficial keratinocytes. However, even in keratinized epithelium, which is histologically atypical, cytology has the advantage of examining individual cells in detail. Therefore, it might be possible to detect highly differentiated tumor cells that exhibit a tendency to keratinize. The diagnostic criteria of the JSCC take this point into account, but further examination of the diagnostic criteria is probably necessary in the future [8]. Suzuki et al. reported that false-negative cytology was more likely to occur in cases where the exposed cell area for diagnosis was very small or where very limited growth was observed [14]. Because it was difficult to collect basal or parabasal-like atypical cells, cells useful for cytodiagnosis were not sampled, thus leading to false-negative results. Sekine et al. reported [7] a false-negative rate of 22.2%, and our result for oral scraping cytology was acceptable.

Unfortunately, we also detected some false-positives. Despite the diagnosis of SCC by the conventional method and the LBC method, the cytodiagnosis was incorrect in each case. Despite the cytological diagnosis of SCC, one negative case was found with the use of both the conventional and LBC methods. However, there were 14 and 15 negative cases of OHSIL with the use of the LBC and conventional methods, respectively (false-positive rate: 14.6% by the LBC method and 30.2% by the conventional method). Many atypical epithelium with large nuclei are often observed, and the presence of epithelium that is partly suspected to be regenerating epithelium is suspected to be a malignant tumor; however, this is often confirmed even in cases associated with inflammation such as ulcer margin or candida infection [15]. In this study as well, false-positives were observed because there were cases of OLSIL that had to be differentiated from SCCs.

According to Remmerbach [16], applying the LBC method instead of the conventional method slightly reduced the false-negative results but still left a significant number of false-negative results. This result was similar to our study results. The false-negative results of cytology may exacerbate an untreated carcinoma, which may not be further treated and followed up. The fact that the lesion may subsequently be fatally exacerbated from false-negative results implies that further improvement is required before oral cytology by itself becomes a completely reliable method. Reducing false-negatives in cytology diagnosis is more important. To that end, in cases that make the diagnosis difficult, “suspect” results tend to be considered as “positive” in terms of further action required. In the oral cavity where various conditions are mixed, there is a limit in oral cytology in which cells of only the surface layer are collected and diagnosed under a microscope. A diagnosis that comprehensively considers clinical information will be necessary.

This study being a single-institution cross-sectional research has some limitations, perhaps including a case bias. Conversely, cytodiagnostic technologists and doctors had established the diagnosis on the basis of the same diagnostic criteria; therefore, it would be better to compare the accuracy of the conventional method and the LBC method according to the technology in a future study. The second limitation of this study is the small number of cases. Although the results of histological examinations were correct, the number of histological examinations was small in cases with NILM diagnosis. Nevertheless, since only a few studies conducted till have compared histology and the number of cases, our study could be useful.

This study demonstrates that the LBC method is superior to the conventional methods in that histological diagnosis is less discrepant with cytological classification. In particular, the LBC method has the advantage of reducing the number of misdiagnosed CIS and severe dysplasia cases in oral cytology. Conversely, because of a certain number of false-positives and false-negatives in cytodiagnosis, it is important to make a comprehensive diagnosis considering the clinical findings.