On-site evaluation of pancreatic lessions according to Papanicolaou Society of Pancreatobiliary Cytology, how can it contribute to diagnostic accuracy?

Background: The use of ROSE in EUS-FNA pancreatic lesions is still controversial in many centers. In this study, we aimed to demonstrate the diagnostic accuracy of ROSE and its contribution to the diagnostic accuracy, as well as its assistance to the pathologist/cytopathologist. Methods: 162 EUS-FNA biopsies were included in the study. EUS-FNA cytology results were reported according to the six-tiered system of Papanicolaou Cytopathology Society and compared to their nal diagnosis with histopathology and/or clinical follow-ups regarding malignancy. In ROSE and non-ROSE patients, diagnostic ability , the difference in diagnostic accuracy, and its contribution to providing the pathologist with sucient amount of tissue acquisition (number of slides and cell blocks) for later examination were compared. Results: In the non-ROSE group, the diagnostic accuracy according to the nal diagnoses was 96% and the sensitivity was 94.44%, specicity 100%, PPV 100%, NPV 87.50%; while diagnostic accuracy was 97.09%, sensitivity 97.47%, specicity 95.83%, PPV 98.77%, NPV 92% in patients with ROSE. There was no signicant difference in diagnostic accuracy between those with and without ROSE (p:0.078). In diagnostic cases, the number of passes, slides and cell blocks were signicantly higher in patients with ROSE than those without ROSE (p:0.003, p:0.007, p:0.012, respectively). ROSE was independently associated with diagnostic ability when evaluated by number of passes, slides, cell blocks in regression analysis (p:0.001, OR:5.07, condence interval: 1.89-13.5). If the lesion is solid or contains a solid component, cytological corcordance with nal diagnosis is higher than cystic ones (p<0.001). Conclusion: ROSE may increase the acquisition of sucient tissue, but it does not have an advantage in diagnostic accuracy. ROSE may raise the number of required slides, which may benet the pathologist in making the diagnosis. If the lesion is solid and/or contains a solid component, the success “VI- malignant”. Cytological diagnoses (malignant / non-malignant) were compared with nal histopathological and/or clinical diagnoses (malignant / non-malignant). “Diagnostic ability” was dened for cases which can be diagnosed other than category I according to PSC-PS. This study Non-diagnostic calculating Based on distribution characteristics, results were expressed as mean ± standard deviation (SD) or median with interquartile range (IQR). The comparison of qualitative variables was analyzed using Pearson Chi-square test or Fisher exact test; Quantitative independent variables were compared using t-test and Mann Whitney U test for parametric and non-parametric distributions respectively.


Background
Although pancreatic cancer is the 14th most common cancer in the world, it ranks 7th in cancer deaths [1]. The reported 5-year survival rate for pancreatic cancer patients is quite low, about 6%, ranging from 2% to 9% [2]. Therefore, prompt diagnosis and early initiation of treatment of pancreatic cancers is very critical for patients.
Today, endoscopic ultrasound guided ne needle aspiration (EUS-FNA) biopsy greatly contributes to the early diagnosis of pancreatic cancers. The correct interpretation of the cytological material and the communication between the pathologist and the clinician are very important, and the Papanicolaou Society of Cytopathology in Pancreatobiliary System (PSC-PS), revised in 2014, makes signi cant contributions to this goal [3]. However, tissue acquisition and diagnostic accuracy under EUS guidance are affected by various factors and their in uences on the diagnostic e cacy cannot be ignored. Needle type, number of needle passes, endoscopist's experience, and rapid on-site evaluation (ROSE) are some of these factors. In addition, the amount of tissue acquisition, the quality of the cytological material, and the presence of the cell block are among the other factors that help cytopathologists / pathologists in subsequent examinations [4,5].
Regarding EUS-FNA, it can be argued that especially rapid on-site evaluation (ROSE) can greatly contribute to the diagnostic e cacy, obtaining adequate tissue and to the diagnostic accuracy. As a matter of fact, the absence of ROSE has been shown to reduce the diagnostic accuracy by 10-15% [6]. In a meta-analysis, studies with ROSE showed a higher sensitivity with a slightly higher speci city as compared with studies without ROSE [7]. However, many high-volume centers do not practice ROSE, since this process requires extra time and workforce [8]. It is important to reveal the power of ROSE in clinical practice and to optimize the evaluation. The number of passes, especially the number of slides and cell blocks required for adequate tissue and high diagnostic accuracy with ROSE has not been adequately studied in the literature.
Our aim in this study is 1) to show the adequacy of tissue acquisition and diagnostic accuracy 2) to show if and how ROSE contributes to this 3) to investigate the relation of the number of passes, the number of slides and the availability of cell block with ROSE to the diagnostic e cacy in EUS-FNA biopsies in pancreatic lesions.

Methods
Patients older than 18, who applied to the gastroenterology unit between 2016-2021 and underwent EUS-FNA for pancreatic lesion were evaluated retrospectively. The clinical follow-up of the patients after EUS-FNA biopsy for at least 6 months, and the histopathological diagnoses obtained by surgery or tru-cut biopsy, if any, were analyzed. Patients whose follow-up were in other centers were evaluated using nation-wide medical electronic records. Patients' oncology admissions, treatments (chemotherapy-radiotherapy regimens for pancreatic malignancy) and pathology reports, radiological work-ups, surgical follow-ups were analyzed. Patients' histopathological diagnoses obtained by surgery or tru-cut biopsy were regarded as " nal histopathology". Patients who were diagnosed with pancreatic carcinoma clinically and / or histopathologically and received oncological treatment were regarded to have "malignant clinical diagnosis" A total of 162 EUS-FNA biopsies were included in the study. All EUS procedures were performed by the same experienced gastroenterologist. EUS-FNA cytology results were reported according to the six-tiered system of Papanicolaou Cytopathology Society (PSC-PC) [3]. Reported categories were as follows; "I-non-diagnostic"; "II-benign"; "III-atypical"; "IVa-neoplastic benign"; "IVb-neoplastic other"; "V-suspicion of malignancy"; "VI-malignant". Cytological diagnoses (malignant / non-malignant) were compared with nal histopathological and/or clinical diagnoses (malignant / non-malignant). "Diagnostic ability" was de ned for cases which can be diagnosed other than category I according to PSC-PS. This study was approved by the ethics committee of the University of Health Sciences Umraniye Training and Research Hospital and meets the requirements of the Declaration of Helsinki.

Technique:
All patients provided informed consent for endoscopy and EUS-FNA. The EUS-FNA was performed using 22 G needle.
(Fujinon Fuji lm, Tokyo, Japan, VP-4450 HD, EG 580 UT) ROSE was provided by pathology residents, pathologists or cytopathology technicians present in the endoscopy unit. One or more air-dried Diff-Quick stained smear was prepared from each pass and quali cation were granted.
Remaining tissues were spread on slides and xed in ethanol. The thick, white-yellowish colored biopsy material remaining at the needle tip after smears was xed in tubes lled with 10% formaldehyde. Then the slides, which were xed in ethanol, were stained with Papanicolaou stain in the pathology laboratory. The biopsy material was processed and embedded in para n blocks, 4.5-5-micron sections were taken and stained with Hematoxylin-eosin. Slides and cell block sections, were evaluated by two pathologists, immunohistochemical staining and molecular testing were performed when necessary.
Statistical Analysis: Sensitivity, speci city, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy of PSC-PS were calculated using cytological diagnosis as benign or malignant with the nal histopathological and/or clinical diagnosis whether they are malignant or non-malignant. Non-diagnostic (category-I) cases were not included when calculating diagnostic accuracy. Based on distribution characteristics, results were expressed as mean ± standard deviation (SD) or median with interquartile range (IQR). The comparison of qualitative variables was analyzed using the Pearson Chi-square test or Fisher exact test; Quantitative independent variables were compared using t-test and Mann Whitney U test for parametric and non-parametric distributions respectively.
The results were evaluated in 95% con dence interval and statistical signi cance level was de ned as p < 0.05. The analyzes were performed using IBM SPSS-21 (Statistical Package for Social Sciences, Chicago, IL, USA).

Results
162 cases with known nal diagnosis were evaluated. 102 (62%) of the patients were male and the mean age in the whole group was 60.54 (23-86). There was no age difference between women and men (59.7±13.3, 61.8±12.9 respectively, p:0.6). EUS characteristics of 22 lesions were purely cystic, 36 were cystic with solid components, and 104 were solid.
ROSE was available for 121 cases, while it was not performed in 41 cases. The distribution of the lesions according to the anatomical regions of the pancreas were examined; there was no signi cant difference between the localizations according to the cystic / solid characteristic, the size of the lesions, and whether they had ROSE. The distribution of the lesions was as follows; 33 in the uncinate, 71 in the head, 29 in the body, and 29 in the tail.
When pure solid lesions (n:104) and solid containing cystic lesions (n:36) were evaluated together and compared with pure cystic, ROSE was found to be performed signi cantly in favor of lesions with solid component (p:0.004). The number of passes was signi cantly higher in patients with ROSE compared to patients without ROSE (median: 2, mean:1.79; 1, 1.37 respectively; p:0.004). The median number of slides was also signi cantly higher in patients who underwent ROSE (Table 1). In addition, the median number of cell blocks obtained in patients with ROSE was 2, while it was 1 in patients without ROSE, and it was signi cantly higher in patients with ROSE (p<0.001) (

Discussion
One of our aims in this study was to assess the amount of tissue acquisition and diagnostic accuracy by EUS-FNA Studies have focused mainly on solid lesions of the pancreas and report con icting results about the effect of ROSE on diagnostic accuracy [11]. To a lesser extent, ROSE for cystic lesions of the pancreas has not been shown to have a signi cant effect on diagnostic accuracy, and there are even some studies emphasizing that ROSE should not be performed on cystic lesions [12]. In this study, we showed that the availability of ROSE in case of solid lesions is associated with higher diagnostic ability and diagnostic accuracy than cystic lesions. Also, if a cystic lesion contained a solid component, diagnostic ability was positively affected compared to pure cystic ones. However, evaluation of pure cystic lesions in our study showed that ROSE had no effect on diagnostic ability. There are some data in the literature that the demographic characteristics of the patients and the location of the lesions in the pancreas do not affect ROSE [13]. Also; in this study, patient's age, size of the lesions, and distribution of the lesion did not differ with availability of ROSE.
Another purpose of our study was to evaluate the necessity of using ROSE. Since, the need for ROSE for adequate tissue acquisition with EUS-FNA biopsy still remains as one of the most discussed issues in practice. In centers where it is possible, ROSE can provide the additional material needed in di cult situations and thus provide suitable material for histological evaluation that will combine the bene ts of cytology and histology. Iglesias-Garcia et al. showed that the presence of ROSE was associated with signi cantly fewer passes, fewer scant samples, higher diagnostic yield, and higher diagnostic accuracy for malignancy [13]. In this respect, the necessity of ROSE procedure is still controversial [4]. Some large centers do not use ROSE due to loss of time, human resources and workforce; but some studies continue to emphasize the superiority of ROSE [3,5]. The diagnostic accuracy of EUS-guided tissue acquisition under ROSE is reported to be higher than 90% in most studies; however, comparable results have also been reported from some trials without ROSE [13,17]. In one recent meta-analysis, authors found that there was no indication that the application of ROSE improved the diagnostic yield and without ROSE; performing with ROSE had a higher pooled sensitivity with similar pooled speci city; in particular, sensitivity and speci city were 83% (95% CI, 64%-93%) and 98% (95% CI, 80%-100%) when ROSE was present, respectively, as compared with 65% (95% CI) , 57%-73%) and 94% (95% CI, 31%-100%) when ROSE was not available [7]. However, Fabbri et al. in his study, when 333 pancreatic solid lesions were divided into ROSE and non-ROSE, no signi cant difference in sensitivity, speci city, positive and negative likelihood ratios, and diagnostic accuracy was found between the two study groups [11]. In addition, some studies have shown that ROSE's success in providing diagnostic results is insu cient even if it is applied in non-diagnostic cases [17]. In our study, in 34 non-diagnostic cases, no signi cant difference was observed in the success of acquiring nal diagnostic results between those who underwent ROSE and those who didn't.
Tissue acquisition with ROSE can also contribute to the pathologist in terms of the number of passes, slide smearing and cell blocks. In this study, the amount of tissue acquired with ROSE (number of slides, cell blocks) was found to be signi cantly superior than those without ROSE. The reason for this may be the intervention of the cytopathologist/ pathologist/ cytology technician who quali es the cytological material that is insu cient during ROSE, increasing the number of passes and the manipulation for obtaining tissue, also spreading the tissue to the slides and forming the appropriate cell block. The advantages of obtaining a cell block, especially with ne needle aspiration, are being able to see the lesion pattern, performing immunohistochemical staining to support the diagnosis, and acquirement of tissue for molecular analysis, even contributing to the chemotherapy regimen and molecular target therapy [19].
Studies report that the needle type used and macroscopic on-site evaluation by the endoscopist increase the quality and tissue acquisition and thus success in acquiring cell block [20]. Although white, yellowish-colored biopsy material of appropriate thickness helps to acquire more cells and to show the pattern and architecture of the lesion, the effect of making a slide smear in excluding differential diagnoses and reaching a de nitive diagnosis cannot be ignored [4,13,21]. Adequate slide number and presence of cell block may be su cient to make a de nitive diagnosis without the need for a second interventional procedure. In addition, it is very critical for the patient that the cell block can  [19]. Therefore, even a single cell block from the lesion can be very valuable, increasing the sensitivity and speci city for accurate diagnosis. In addition to all these, although there are studies on the optimization of the obtained cytological material with the slide smear technique, but, the optimum "number of slide smearing" requiring for a diagnostic ability has not been su ciently examined in the literature for both ROSE and non-ROSE applications.
In our study, the application of ROSE in diagnostic cases was signi cantly higher than those who didn't have ROSE in terms of the number of passes, slides, and cell blocks. However, in cases with ROSE, the diagnostic cases was not different than non-diagnostic ones regarding the number of passes, slides and cell blocks. ROSE was independently associated with diagnostic ability in regression analysis, which can be interpreted as follows: 1) Although there is a pathologist at the bedside, it may not have su ciently increased the number of passes and thus the tissue acquisition in non-diagnostic cases. In addition, 2) slides and cell blocks may have been prepared technically improperly, and it may have caused degeneration of cells and tissues with artefactual changes. This could have prevented giving diagnostic competence. 3) If the possibility of complications is high in patients who have undergone intervention, the number of passes may be reduced.

Conclusion
As a result, ROSE can increase the number of passes, slides and cell blocks, but it does not have an advantage in diagnostic accuracy. The optimum number of slides has not been examined in the literature, and ROSE may increase the number of required slide smears, which may bene t the pathologist in diagnosis. If the lesion is solid and/or contains a solid component, the success of obtaining a diagnostic ability is higher in patients with ROSE than in those without. Therefore, ROSE still maintains its applicability in terms of making the nal diagnosis to the patient and increasing the diagnostic e ciency.  Tables   Table 1 Comparison of patients with/without ROSE by using passes of needles, slides and cell blocks.