記住我
Abstract
Context: The Indian Academy of Cytologists published Guidelines and categories for Reporting Serous Effusions (IACGRSE) in 2020 to improve consistency and reproducibility of fluid cytology reporting and to guide patient management. Aims: To evaluate category-wise agreement while using IACGRSE 2020 categories. To analyze interobserver agreement among participants with different level of training and years of experience. Settings and Design: A retrospective interobserver variability study. Methods and Material: Study was done with four participants: an expert cytopathologist and three cytopathology fellows with varying experience. Fluid smears from 60 cases with clinical and/or radiological evidence of malignancy were categorized into one of the five IACGRSE 2020 categories. The interpretations of expert cytopathologist were taken as standard. Statistical Analysis Used: Interobserver agreement was analyzed using Kappa statistics. Results: Previous cases without definitive category got classified into “Atypical cells NOS” (3.33%) and “Atypical cells, Suspicious of Malignancy” (15%). Agreement analysis for IACGRSE 2020 categories showed better concordance for inadequate (I), malignant (V), and benign (II) categories. The range of Kappa for interobserver agreement of fellows was fair to substantial (range 0.1692–0.7249). The participant with substantial diagnostic agreement with expert (κ = 0.729, 88.3%) had the most experience. Causes of major discordance were pertaining to paucity and distribution of cells, and to misinterpretation of reactive mesothelial cells. Conclusions: IACGRSE 2020 categories and participants' experience were important determinants in classifying the effusion fluid cytology smears and interobserver agreement; emphasizing the need to use IACGRSE2020, and sufficient time and training required for accurate diagnosis of fluid specimens.
Keywords: Categorization, diagnostic categories, interobserver agreement, serous fluid
How to cite this article: Introduction 
The necessity of a common language and unified nomenclature for pathologists and clinicians usually gives rise to various classification systems in diagnostic cytopathology. This definitely helps in better patient management. The Bethesda System for Reporting of Cervical Cytology is thefirst and major success story, yet again emphasizing the need and actual utility of such systems.[1] Similar ones exist for urine, thyroid, salivary gland, and breast cytology.[2],[3],[4],[5] Although serous fluid is one of the most common specimens processed by cytopathology laboratories, a uniform reporting terminology and system was lacking until recently. This was the reason which led to the use of varied terms in interpretations and at times cases could not be allocated to any particular category. The Indian Academy of Cytologists (IAC) introduced guidelines and categories for reporting serous effusion cytology (IACGRSE) in 2020. This system provides detailed guidelines for the reporting of cells under each category in order to achieve consistency and reproducibility in the reporting of fluid specimens along with clear recommendations for the clinicians.
Efforts have been taken to study the effect of the reporting terminology systems using the reproducibility studies in gynecologic and urine cytopathology.[6],[7] However, there are no similar studies for serous effusion cytology as yet using IACGRSE 2020.
The level of specific training and exposure to a significant workload are the components of accurate cytology reporting. Also, expert review in cytopathology, including gynecological and non-gynecological, has been proven to have impact on the patient care with regards to the reporting atypical cells, high-grade lesions and cancer diagnosis.[8] This again emphasizes the impact of experience. Literature review has shown hardly any reference regarding effect of experience of cytopathologist on serous effusion cytology reporting.
Thus, the present study was carried out to fill these gaps in the current body of literature. The aims were to evaluate category-wise agreement while using IACGRSE 2020 and to find the interobserver variability among participants with different level of training and years of experience.
Subjects and Methods This was a retrospective interobserver variability study carried out at the Cytology Unit, Department of Pathology of a medical college and hospital. The participants included an expert cytopathologist with >34 years of experience post-MD and three cytopathology fellows: (A) 8 years of experience post-MD, (B and C) 1-year experience post-MD each. The interpretations of the expert cytopathologist were taken as standard. The material for this study comprised of cases of serous effusion smears with relevant clinical, radiological details, and histopathological follow-up. All of these cases had clinical and/or radiological evidence of malignancy and were not having any preanalytical issues related to fluid processing. In each case, two smears, one air-dried MGG stained and one wet fixed Papanicolaou stained smear were examined. The cytomorphological evaluation was carried out by light microscopy and interpretation was recorded by all the participants in a given format.
Categorization according to IACGRSE 2020 categories
Each case was categorized into one of the five recommended diagnostic IACGRSE 2020 categories that are as follows: I. Unsatisfactory. II. No Malignant cells/Benign cellular changes. III. Atypical cells, not otherwise specified (NOS). IV. Atypical cells—Suspicious of Malignancy. V. Malignant cells seen. Diagnosis of the cases included in the present study had representation of all the five categories.
Category-wise agreement analysis
The fellow participants' cytological interpretations classified by the IACGRSE 2020 categories were analyzed for interobserver agreement by category using Kappa statistics between the fellow participants. Also, the category-wise observations of each fellow participant compared to that of the standard were analyzed as well.
Agreement analysis of overall interpretations of participants
The interpretations of all the cases combined were assessed among all the fellow participants. Similarly, overall observations of each fellow participant were compared with standard to find whether experience or training is associated with overall agreement between the fellow participants and experts.
Results A total of 60 serous effusion cases were studied. Maximum cases were of ascitic fluid—32 (53.33%), followed by pleural 24 (40%) and pericardial effusion 4 (6.67%) [Figure 1]a.
Figure 1: (a) Types of fluids. (b) Category-wise distribution of cases according to IACGRSE 2020Categorization according to IACGRSE 2020 categories
Most cases belonged to category V—43, (71.67%) and least to category I—1, (1.67%) according to IACGRSE 2020 [Figure 1]b.
When the older terminology was used for these cases, 18.33% of cases could not be assigned any definite category. Whereas, with the use of IACGRSE 2020, all the cases were allocated to definite categories [Table 1].
Table 1: Comparison among distribution of cases in the categories of the IACGRSE 2020 versus older terminologiesCategory-wise agreement analysis
The agreement between category-wise observations of all participants versus standard when analyzed, showed maximum agreement in the Category I (unsatisfactory) with Kappa 0.5915, followed by Category II (Benign) with Kappa 0.4259 and Category V (Malignant)(Kappa 0.3729) [Table 2].
Table 2: Agreement Analysis for IACGRSE 2020 Categories: All participants' category wise impressions compared to standards' and between the participants' observations (Interobserver)Category I also showed maximum agreement between the three participants (A, B, and C) with Kappa value of 0.3829, followed by Category V (Malignant) with value of Kappa as 0.3345 [Table 2].
Agreement analysis of overall interpretations of participants
Agreement analysis between overall interpretations (i.e., the interpretation of all the cases) of each participant versus standard showed maximum agreement of 88.3% (53) cases with Kappa score of 0.7249 that is, good agreement of participant A with expert. In contrast this agreement with the expert for observers B and C was lower (Kappa 0.0875 and 0.1692) [Table 3].
Table 3: Agreement analysis between overall interpretations of each participant compared to that of standard36.7% (22 slides) of all participants' observations agreed with each other (Kappa score 0.2121) when the agreement analysis between overall observations of participants with each other (A vs. B vs. C) was performed [Table 2].
Discussion In recent times, a diagnostic consensus has become the goal in reporting the cytopathology of various systems, for example, Milan system for salivary lesions, the Bethesda system and so forth, so as to enhance professional communication leading to improved patient management and care. The effect of using recently published the IAC Guidelines and Categories for Reporting Serous Effusion Cytology 2020 (IACGRSE 2020) by means of the reproducibility studies was studied here.[9]
Categorization according to IACGRSE 2020 categories
The aim of IACGRSE 2020 to achieve consistency and reproducibility in fluid cytology seems to be achieved well as evident from the present study, where 18.33% of cases (11) unclassifiable with previous terminology could be categorized furthermore. With the use of this new system, of these 11 cases, two cases (3.33%) were placed in the Category III (Atypical cells NOS) and nine (15%) cases were categorized as Category IV (atypical cells, Suspicious of malignancy). Similar category-wise distribution was found in another work.[10] This standardized reporting also aided in rigorous activity of data entry (tabulation of interpretations) and comparison, and rest of the statistics of the present research. The clinical applicability was evident from the recommendations of “to repeat cytology” and “to correlate clinico-radiologically” for Category III cases and only “to repeat cytology” for Category IV cases [Table 1].
Similar to the interobserver studies for The Bethesda system of gynecologic cytology, Milan and Paris systems,[6],[7],[11] the performance of observers was directly associated with the specific diagnostic categories of IACGRSE 2020 and the experience of the reporting person.
Category-wise agreement analysis
Because of the simplicity of categories of IACGRSE 2020, observers could place each case in a proper category and this helped to achieve maximum interobserver agreement.
Out of the 22 cases (36.7%) in which all the fellows' opinions matched, Category I—(Inadequate) showed maximum agreement (Kappa 0.3829) followed by Category V—(Malignant) (Kappa 0.3345) [Table 2].
Similarly, of the 20 cases (33.3%) concordant between all the participants and the expert, maximum agreement in Category I (Inadequate) (Kappa 0.5915) was seen followed by Category II (Benign) (Kappa 0.4259) and Category V (Malignant) (Kappa 0.3729) [Table 2]. When the category-wise interpretations of the Participant A were compared with those of expert's interpretation, maximum agreement belonged to Category I, V and II in descending frequency. Thus, irrespective of the experience, inadequate (I), malignant (V), and benign (II) categories had a better concordance of all.
The value of interobserver agreement studies, using any reporting system, lies in the fact that a good to perfect agreement should be achieved for both benign and malignant categories between the observers.[12] Our study also supports this fact. The maximum interobserver variability was observed in the interpretation of smears showing atypical cells. The Kappa scores of these categories (III and IV) were least on comparing individual participant's category-wise interpretations with the expert [Table 4] as well as interobserver category-wise interpretations [Table 2].
Table 4: Agreement Analysis for IACGRSE 2020 Categories: each participant's category wise observations compared to that of standardAgreement analysis of overall interpretations of participants
This proved the effect of experience on the interobserver variability. It is well known that in addition to cytomorphology and sample processing, the experience of the person who interprets the cytology smear has an important role in the accurate diagnosis of the difficult cases, and can contribute to interobserver disagreement.[13] There are only few studies that have targeted the effect of experience of the observer on the accuracy of the diagnosis, and its impact on patient care.[8],[14] The present study also proves the value of the expertise and experience in the interpretation of fluid cytology smears. The overall interpretations of participant A (8 years of post-MD experience) mostly agreed with the standard (88.3%, Kappa—0.7249). The agreement decreased with the other two participants-with much less experience to a much lower level (55.0%, Kappa 0.1692 and 46.7%, Kappa 0.0875) [Table 3]. The role of experience also has increased importance in low-resource lab set ups due to the cost implications if ancillary techniques like cell blocks and immunohistochemistry are needed, which may be avoided with expert review in some cases.
When the causes of discordance between observers were analyzed, they seemed to be pertaining to paucity and distribution of cells and to misinterpretation of reactive mesothelial cells [Table 5]. Twelve cases showed major discordance that is, a difference of two or more categories.
Table 5: Major Discordant Cases: Categories allotted by the participants and expert, and most probable cause of discordancePaucity and distribution of few cells in a large smear area
This turned out to be the important cause of misdiagnosis by junior observers. When the malignant cells were very few and were scattered singly [Figure 2]a or not associated with cell balls or three dimensional clusters of malignant cells, they were missed by them (six cases) [Table 5]. This emphasizes the need of the cell concentration methods such as cytospin or liquid-based preparations again.[15],[16]
Figure 2: (a) Singly scattered malignant cells. (400x, Papanicolaou stain). (b) Reactive mesothelial cells. (400x, May-Grünwald-Giemsa stain)Few studies have recommended an optimal volume of 20–30 ml of fluid for the purpose of cell adequacy and this is the proposal of IACGRSE 2020 also. It is a well-known fact that cellularity does not matter as long as unequivocal atypical or malignant cells are seen.[9]
Reactive mesothelial cells
Two cases that were showing reactive mesothelial cells, and were reported as benign by the expert and senior observer (A), in contrast to malignant interpretation by junior observers [Figure 2]b.
Limitations of the present study
First, the sample size of the present work was modest. Second, the workload during the period of experience for different observers may have an additional role to play in the interobserver variability statistics.
Conclusion Application of IACGRSE 2020 categories allowed all cases to be reported with definitive impression as there were two specified categories allotted to the atypical cells. Interobserver agreement was good for benign and malignant categories with considerable poor agreement for the intermediate categories. The range of the Kappa statistics for the three cytopathology fellows was fair to substantial. The participant with substantial diagnostic agreement with expert had the most experience, emphasizing the need for sufficient time and training required for accurate diagnosis of fluid specimens.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References 
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/joc.joc_50_22
留言 (0)