A retrospective analysis of serous effusions based on the newly proposed international system for reporting serous fluid cytopathology: a report of 3633 cases in an oncological center

Effusions can be produced inside serous cavities in neoplastic and nonneoplastic lesions. Serous effusion cytology is a common clinical examination method to distinguish benign and malignant serous effusions due to its advantages of being minimally invasive, easily available, and cost-effective [1‐3]. In view of the importance of cytology in the evaluation of effusion specimens, its role in patient management has become increasingly important. An international group of cytopathology experts published “The International System for Reporting Serous Fluid Cytopathology” (TIS) to standardize the reporting terminology and criteria to establish diagnostic categories with high diagnostic value [4, 5].

The 5 proposed diagnostic categories are nondiagnostic (I, ND), negative for malignancy (II, NFM), atypia of undetermined significance (III, AUS), suspicious for malignancy (IV, SFM), and malignant (V, MAL). TIS has defined the AUS category as specimens that lack quantitative or qualitative cytologic features to be confidently diagnosed as either benign or malignant and that exhibit sufficiently clear morphologic features to exclude the possibility of classifying them as ND. The SFM category is defined as specimens showing cytologic features usually found in malignant lesions but insufficient either in quality or quantity for a definitive diagnosis of malignancy. Hou et al. [6] showed that the risk of malignancy (ROM) for SFM was significantly higher than that for AUS (P < 0.01), which supports the separate diagnostic categories of these two independent groups. A key question to ask is whether SFM category based on heterogeneous cytological features (quality or quantity) carries the same ROM and deserves the same clinical management.

To date, only a few publications have supported the use of a particular terminology for serous effusion cytology [6‐15]. In this study, we retrospectively applied TIS recommendations for reporting the cytological diagnosis of serous effusions. After sample reclassification, the ROM for each TIS category was calculated, and the performance evaluation was carried out between different sample preparations (conventional smears, liquid-based preparations and cell blocks). To the best of our knowledge, this is the first publication looking into the method of preparation employed in a retrospective cohort of serous effusions based on the TIS. In addition, we sought to review and subclassify the SFM category into 2 groups (IVa and IVb) by the cytological features (quality and quantity) to calculate the ROM of each subgroup and to evaluate the necessity of having a subclassification for the SFM category.

Serous effusion cytology is a minimally invasive and cost-effective diagnostic method used to investigate the etiologies of body cavity effusions and can guide clinical decision-making. To the best of our knowledge, this is the largest series to date of such analyses in serous effusions. We evaluated our application of the recently proposed TIS on reporting serous effusion cytopathology. A total of 3633 patients were included, among which 17 (0.5%), 1100 (30.3%), 101 (2.8%), 677 (18.6%) and 1738 (47.8%) cases were classified into ND, NFM, AUS, SFM and MAL groups, respectively.

The malignancy rate (47.8%) detected in our cohort was higher than that reported in the literature, which ranges between 4% and 22.4% [6‐9, 11‐13, 15], but was similar to other reports in the literature from oncological centers [6, 10, 14]. The percentage of our SFM cases was also slightly higher than that in other reported studies (range: SFM, 1.3%-6.3%) [6‐15]. These results are characteristic of an oncological center, where neoplastic conditions are the main cause of cytological examination of body fluids. If a case is diagnosed as SFM, combined with the tumor history and clinical symptoms, the clinical doctor will diagnose the case as positive and proceed directly to the next step of treatment, which makes some SFM lose the opportunity to be identified as MAL by immunocytochemistry and other auxiliary methods. In addition, our study included both outpatient and hospitalized patients, and the outpatients often failed to be identified as malignant by auxiliary examination. Moreover, we did not use a specific number of cells to determine whether a sample was suspicious or positive for malignancy. In patients with a clear history of disseminated malignancy, although there was no immunochemistry support, a few highly atypical cells might be sufficient to diagnose SFM, while in other clinical situations, the same number of cells might not be enough for a diagnosis of SFM and can only be diagnosed as AUS. If we follow the principles of clinical management, 96.6% of our serous effusions got a therapeutically meaningful diagnosis, including NFM, MAL and SFM, and only a small proportion of cases were diagnosed as ND and AUS (0.5%, 2.7%).

Our nondiagnostic rate was 0.5%. This is in line with other reports in the literature, which present nondiagnostic rates of 0% to 5.6% in serous effusion [6‐15]. The cases classified as ND in this series were all due to scant cellularity or excess degeneration, improper preservation and obscuring blood in the serous effusion specimens. Therefore, whole blood samples should be anticoagulated in a timely manner before sample preparation.

The minimal threshold of adequacy for fluid interpretation is still contentious and has not been described clearly in the TIS. Some earlier studies have suggested a minimum of 50–75 ml [17‐19], but the evidence is limited. Recently, Gokozan et al. [8] performed a root cause analysis of the diagnoses of atypia or suspicious for malignancy and showed that 50 mL and below were considered low volume samples, and were included as a root cause for indeterminate diagnoses. In the present study, the 17 nondiagnostic cases all had a specimen volume greater than or equal to 50 ml and many of our malignant cases had a very low volume submitted to our laboratory. Therefore, the volume threshold of adequacy should be regarded as a recommendation, with the final decision left to each practice.

In our cohort, the ROM values of serous effusions in the ND, NFM, AUS, SFM and MAL groups were 38.5%, 28.6%, 52.1%, 99.4% and 100%, respectively. It is worth noting that high ROMs can be seen in our study for SFM. One possible reason for this result could be due to the fact that our data come from a large oncological center, and the majority of the serous effusions often originate from tumors, providing a potential selection bias. Our high ROM for SFM supports the viewpoint that most clinicians will manage patients with SFM effusions similar to those with a malignant diagnosis. Besides, in our study, the ROMs for the nondiagnostic and negative categories were also high, which may also be attributed to the nature of the patient population in our cancer center, many of whom were referred to our hospital with an established malignant diagnosis and frequently at an advanced stage of disease. For the cases with high clinical suspicion, even though cytology was negative, patients usually underwent medical imaging examination and clinicians tended to pursue pleural biopsies; therefore, the ROMs for the nondiagnostic and negative categories will improve.

The TIS describes two indeterminate categories, AUS and SFM, created to encompass all of the fluid that could not be placed under the NFM or MAL categories. In our study, the ROM for SFM was significantly higher than that for AUS (P < 0.001), thus providing support for retaining the two indeterminate categories as independent ones. The difference between the ROMs for the IVa and IVb was insignificant with a P value of 0.124, which indicates that although different situations exist that can be diagnosed in the SFM category, there is no need to reclassify for the SFM category. From the Table 2, it is worth noting the wide range of ROM calculated for each diagnostic category, which is likely attributable to the variation in reporting among different institutions. Standardized reporting would also provide a meaningful language that clinicians can uniformly understand and utilize in their patient management. More research is needed to convey the ROM of each category to the corresponding clinical colleagues to optimize patient care.

Our performance analysis was in agreement with previous publications [10, 11, 13, 14, 16] (Table 4). By comparing the performance evaluation of different groups (considering cytological diagnosis of MAL + SFM + AUS as positive, considering cytological diagnosis of MAL + SFM as positive, and considering cytological diagnosis of MAL as positive), it is best to consider the categories of MAL and SFM as positive, while excluding AUS, to be beneficial to clinical management. This is supported by our data (Table 4) and agrees with the perspective of TIS [5]. Moreover, it is worth mentioning that when considering the categories of MAL and SFM as positive, there were only 4 false-positive cases, and the 4 cases were all diagnosed as IVa, and the total specificity and the PPV were as high as 99.5% and 99.8%, respectively.

Compared with conventional smears, liquid-based preparations require less skill, and more importantly, allow for the application of ancillary examinations [20]. Moreover, liquid-based preparations permit fa more even distribution of cells over the slide area, a reduction in obscuring background elements, and better preservation of nuclear detail and cytoplasm. Several studies have shown that slides prepared with liquid-based preparations have a lower nondiagnostic incidence and higher accuracy than conventional smears [21‐23]. In our study, the sensitivity, NPV and diagnostic accuracy of liquid-based preparations were all superior to conventional smears and cell blocks in detecting abnormalities. In addition, by comparing the performance evaluation among different methods of preparation, we found that applying two methods at the same time was superior to any single method and using three methods simultaneously had the highest sensitivity, NPV and diagnostic accuracy, while there was little difference between the two diverse methods of preparation. Therefore, in terms of the selection of preparation methods, it is best to use three different methods simultaneously for serous effusion cytologic examination.

There are 2 main limitations in our study. The first is attributed to the nature of the patient population in our cancer center, leading to a high rate of MAL and SFM, as well as a high ROM for each category. Another limitation of our study is that it was a single-center and retrospective study, and cytological diagnoses were interpreted by only two cytopathologists, who, although they reclassified cases according to TIS criteria, may still have individual biases.

In conclusion, this study supports the idea of retaining the two indeterminate categories (AUS and SFM) as independent ones, and there is no need to reclassify for the SFM category. By comparing the performance evaluation of different groups, it is best to consider the categories of MAL and SFM as positive, while excluding AUS. A total of 96.6% of our serous effusions received a directed diagnosis, including NFM, MAL and SFM, and only a small proportion of cases were diagnosed as ND and AUS (0.5%, 2.7%), demonstrating that TIS is a user-friendly reporting system. The sensitivity, NPV and diagnostic accuracy of liquid-based preparations were all superior to those of conventional smears and cell blocks in detecting abnormalities. It is best to use three different preparation methods simultaneously for serous effusion cytologic examination.