Pancreaticobiliary Cytology Practice in 2021: Results of a College of American Pathologists Survey

In 2021, the Pancreaticobiliary Cytology Practices Supplemental Questionnaire was mailed to 2092 laboratories worldwide that participated in the 2021-D CAP Nongynecologic Cytopathology (NGC) Education Program. Of the 890 laboratories (45%) that responded to the supplemental questionnaire, laboratories that skipped (n = 12) or provided a “Not Applicable” response (n = 266) to Question 1 (Does your laboratory evaluate pancreaticobiliary cytology specimens?) were excluded from all summaries. A total of 612 laboratories responded to Question 1, with 567 laboratories indicating affirmatively that they routinely evaluated pancreatobiliary cytology specimens; these 567 were included in final analysis.

The supplemental questionnaire asked laboratories for additional specifics regarding their pancreaticobiliary practice, their volume, types of pancreaticobiliary cytology specimens evaluated, sampling equipments, types of cytologic preparations, performance of rapid on-site evaluation (ROSE)/rapid on-site assessment (ROSA), indications for ROSE/ROSA, designation of person who performed ROSE/ROSA and the person who signed out pancreaticobiliary cytology specimens, variety of ancillary studies used, and lastly how pancreaticobiliary cytology specimens were reported (Supplemental Table 1; see supplemental digital content at https://meridian.allenpress.com/aplm in the June 2024 table of contents). The respondents were asked to answer all questions, based on actual clinical practice in 2021; however, not every laboratory responded to every question. Results for combined “Other” and “Other, Specify” responses were reviewed and compiled. The data were reviewed and adjusted for skip sequence and response inconsistencies.

Institution type and location were 2 independent variables used to examine associations with practice characteristics. Institution type was populated by using a database of NGC demographic data from 2017–2021. Institution categories were grouped as follows: hospital/medical center–academic, hospital/medical center–nonacademic (voluntary, nonprofit hospital, proprietary hospital, and city/country/state hospital), national/corporate laboratory, nonhospital clinic or laboratory (regional/local independent laboratory and clinic, group or doctor-office laboratory, public health, nonhospital), and Veterans hospital/Department of Defense (DOD) (Veterans hospital and Army/Air Force/Navy hospital). Institution location was dichotomized as United States and international laboratories.

We examined associations between independent variables and practice characteristics by using multiple logistic regression models to control for both institution type and location. If the logistic regression model was not fit, a χ² test of independence or a Fisher exact test was performed. “Unsure,” “None,” and “Not Applicable” responses were excluded in all tabulations and statistical calculations. The threshold for statistical significance was set to α < .05. Pairwise differences were computed for each level of institution type, with a Bonferroni correction to adjust for multiplicity in testing. Analyses were performed with SAS 9.4 (SAS Institute, Cary, North Carolina).

Among 612 respondents, 567 laboratories (92.6%) routinely evaluated pancreaticobiliary cytology specimens. Most laboratories (79.0%, 424 of 537) evaluated fewer than 100 pancreaticobiliary cytology specimens per year.

Among 562 respondents, biliary brushing was the most common pancreaticobiliary cytology specimen evaluated (85.1%, 478), followed by pancreatic FNA (78.8%, 443), pancreatic duct brushing (62.8%, 353), and ampullary brushing (53.6%, 301). It should be noted that almost half of laboratories evaluated pancreatic FNB (45.4%, 255), and only 20.5% (115) of the laboratories evaluated pancreatic cystic wall biopsy (Table 1).

For sampling, a 22-gauge needle was most frequently used (62.1%, 146 of 235), followed by a 25-gauge needle (41.3%, 97), 19-gauge needle (32.3%, 76), SharkCore needle (27.2%, 64), ProCore needle (19.6%, 46), Moray micro forceps (17.4%, 41), and Acquire needle (17.0%, 40). However, more than half of the respondents (306) were unsure of their sampling equipment (Table 1).

Slide preparation methods for pancreaticobiliary cytology specimens included direct smear, CB, liquid-based cytology (LBC), and cytospin. CB was the most frequently reported preparation method (76.1%, 388 of 510), followed by LBC (59.0%, 301), direct smear (51.2%, 261), and cytospin (41.4%, 211). Most laboratories used a combination of slide preparation methods, with LBC plus CB as the most frequently reported combination (17.5%, 89 of 510), followed by LBC and smear plus CB (15.9%, 81) and cytospin and smear plus CB (13.1%, 67) (Table 1).

Among the 320 laboratories that responded, almost all indicated that they performed ROSE/ROSA on pancreatic solid lesions (94.7%, 303), while a little more than half of laboratories indicated that they also performed ROSE/ROSA for cystic lesions (56.3%, 180) and duct lesions (54.4%, 174). Responses indicated that general pathologists primarily performed ROSE/ROSA in 55.5% (178 of 321) of laboratories that responded, while specialized cytopathologists performed ROSE/ROSA in 43.0% (138 of 321) of laboratories. Pathology trainees performed ROSE/ROSA in a minority of laboratories (6.9%, 22 of 321) (Table 1).

General pathologists primarily signed out pancreaticobiliary cytology specimens in 67.6% (374 of 553) of the laboratories that responded, while subspecialized cytopathologists primarily signed out in 63.8% (353 of 553) of laboratories. Additionally, subspecialized gastrointestinal pathologists signed out pancreaticobiliary cytology specimens in 12.5% (69 of 553) of laboratories. The same pathologist signed out the concurrent pancreaticobiliary surgical biopsy and cytology specimen in most of the laboratories surveyed (71.0%, 363 of 511) (Table 1).

Ancillary studies for pancreatic cystic lesions included fluid analysis, immunohistochemical or histochemical analysis of mucin, genetic mutation analysis, DNA analysis for aneuploidy or polyploidy, and others. The most common ancillary study was fluid analysis (79.1%, 336 of 425), followed by immunohistochemical or histochemical analysis of mucin (58.8%, 250) and genetic mutation analysis (34.6%, 147). For pancreatic solid lesions, immunohistochemical stains were most frequently used (64.0%, 190 of 297), followed by mismatch repair protein or microsatellite instability (58.6%, 174) and genetic mutation analysis (44.4%, 132). For biliary lesions, genetic mutation analysis was the prevailing ancillary study (66.7%, 98 of 147), followed by fluorescence in situ hybridization (48.3%, 71 of 147).

Among the 523 laboratories that responded, 275 (52.6%) performed Ki-67/MIB1 immunostaining for neuroendocrine tumors on cytology specimens. In addition, among the 267 responding laboratories that reported performing Ki-67/MIB1 immunostaining for neuroendocrine tumors on cytology specimens, 208 (77.9%) used the Ki-67/MIB proliferation index to grade neuroendocrine tumors on cytology specimens (Table 1).

Only 36% (193 of 530) of laboratories indicated that they used the PSCPC.

Five hundred eleven participants responded “Yes” to Question 1 and provided information for their institution types in the comparison of pancreaticobiliary cytology practice among different institution types. The distribution is summarized in Figure 1. Survey results were analyzed to identify statistically significant differences (Table 2).

Most academic hospitals (98.2%, 55 of 56) reported evaluating pancreaticobiliary cytology specimens, as compared to nonacademic hospitals (94.1%, 367 of 390), nonhospital clinic/laboratories (85.5%, 53 of 62), national/corporate laboratories (86.4%, 19 of 22), and veterans hospitals/DOD (89.5%, 17 of 19) (P = .046) (Figure 2).

Academic hospitals tended to report a higher annual volume of pancreaticobiliary cytology specimens than other institution types: 48.9% (22 of 45) of academic hospitals reported evaluating more than 100 specimens per year, compared to only 18.8% (66 of 352) of nonacademic hospitals (P < .001) and 19.2% (10 of 52) of nonhospital clinic/laboratories (P = .02) that reported evaluating more than 100 specimens per year (Figure 3).

For the specimen types evaluated among different institution types, the only significant difference was pancreatic duct brushing. Academic hospitals reported evaluating pancreatic duct brushing more than other nonacademic hospitals (80.0% [44 of 55] versus 61.7% [224 of 363]; P = .03) and Veterans hospital/DOD laboratories (80.0% [44 of 55] versus 35.3% [6 of 17]; P = .01).

General surgical pathologists tended to perform ROSE/ROSA more in nonacademic hospitals than academic hospitals (60.1% versus 28.2%; P = .01), while a higher percentage of academic hospitals reported that cytopathologists perform ROSE, compared to nonacademic hospitals (61.5% [24 of 39] versus 41.9% [85 of 203]).

Similarly, general surgical pathologists tended to sign out pancreaticobiliary cytology cases more in nonacademic hospitals than academic hospitals (71.3% [253 of 355] versus 27.3% [15 of 55]; P < .001), while cytopathologists tended to sign out pancreaticobiliary cytology cases more in academic hospitals than in nonacademic hospitals (83.6% [46 of 55] versus 60.8% [216 of 355]; P = .003).

Concurrent surgical biopsy and cytology pancreaticobiliary specimens tended to be signed out by the same pathologist more commonly in nonacademic hospitals than in academic hospitals (74.2% [241 of 325] versus 37.0% [20 of 54]; P < .001), suggesting more subspecialized practice in academic hospitals.

Among the 567 responding laboratories, 482 (85.0%) were from the United States, while the other 85 (15.0%) were international. We also compared pancreaticobiliary cytology practices between domestic and international laboratories, and the following statistically significant differences were identified (Table 3).

For specimen types, domestic laboratories tended to report evaluating biliary brushing and pancreatic duct brushing more than international laboratories (biliary brushing: 87.9% [377 of 429] versus 68.8% [53 of 77], P < .001; and pancreatic duct brushing: 64.8% [278] versus 50.6% [39], P = .008).

CB and LBC preparations were more frequently reported in domestic laboratories than in international laboratories (CB: 77.7% [299 of 385] versus 69.4% [50 of 72], P = .04; LBC: 62.3% [240 of 385] versus 36.1% [26 of 72], P < .001), while smear and cytospin were more frequently reported in international laboratories than domestic laboratories (smear: 65.3% [47 of 72] versus 47.8% [184 of 385], P = .01; cytospin: 52.8% [38 of 72] versus 39.0% [150 of 385], P = .04).

More international laboratories performed ROSE for pancreatic cystic lesions than domestic laboratories (78.6% [22 of 28] versus 52.7% [135 of 256]; P = .04). However, there were no statistically significant differences in performing ROSE for pancreatic solid lesions (P = .44) or duct lesions (P = .11) between domestic and international laboratories.

Subspecialized cytopathologists tended to perform ROSE more in domestic laboratories than international laboratories (48.8% [123 of 252] versus 21.4% [6 of 28]; P = .003), suggesting more domestic laboratories had subspecialized practice. Similarly, subspecialized cytopathologists tended to sign out pancreaticobiliary cytology specimens more in domestic laboratories than in international laboratories (67.1% [284 of 423] versus 54.1% [40 of 74]; P = .001).

Pancreaticobiliary cytology has been established as an important diagnostic modality for pancreaticobiliary lesions. Pancreatic FNA/FNB has demonstrated high sensitivity and specificity, while biliary cytology (brushing/FNA) shows high specificity, but relatively low sensitivity mainly associated with sampling errors.^1–23  Our results revealed that up to 92.6% of 612 responding laboratories routinely evaluated pancreaticobiliary specimens in 2021, suggesting that pancreaticobiliary cytology is indeed a widespread diagnostic modality in routine pathology practice.

It is not surprising that biliary brushing (85.1%) and pancreatic FNA (78.8%) were the top 2 most evaluated pancreaticobiliary cytology specimens among all respondent laboratories from our survey, given their well-established diagnostic utility for pancreaticobiliary lesions. Furthermore, our survey results revealed that 45.4% of respondent laboratories evaluated pancreatic FNB specimens in 2021, suggesting the recognition and rapid adoption of this relatively new technique. A variety of pancreatic FNB methodologies including SharkCore, ProCore, Moray micro forceps, and others are currently available and have demonstrated comparable diagnostic sensitivity and specificity to pancreatic FNA.^13,14,28  Additionally, pancreatic FNB can provide more consistent CB yield with higher cellularity, which not only enables pathologists to perform ancillary studies (eg, immunohistochemical analysis) to make accurate diagnoses, but also provides materials for molecular analysis for precision therapy.^29–31  Indeed, our survey results revealed the most common slide preparation method was CB preparation, which was performed in about 76% of respondent laboratories, indicating the importance of ancillary studies in current pancreaticobiliary cytology practice. In addition, our survey revealed most laboratories used a combination of slide preparation methods to evaluate pancreaticobiliary specimens, with LBC plus CB as the most common combination (17.5%) in order to improve sensitivity or specificity.

ROSE/ROSA performed by cytopathologist/cytotechnologist allows for the confirmation of sample adequacy for pancreatic FNA/FNB. Studies have revealed that ROSE/ROSA could improve diagnostic accuracy, increase material yield, and decrease repeated procedures in patients with solid pancreatic lesions.^32–35  However, the utility of ROSE/ROSA for cystic pancreatic lesions is debatable since a study has shown no significant improvement of diagnostic accuracy or increase of material yield.³⁶  Our results revealed that 94.7% (303 of 320) of respondent laboratories performed ROSE/ROSA on pancreatic solid lesions, while more than half of the laboratories indicated that they also performed ROSE/ROSA for pancreatic cystic lesions (56.3%, 180) and ductal lesions (54.4%, 174), suggesting the utility of ROSE/ROSA in improving specimen adequacy.

Ancillary studies are important and sometimes necessary to make an accurate diagnosis of pancreatic lesions. For pancreatic cystic lesions, 79.1% of respondent laboratories indicated that they used fluid analysis to aid in the diagnosis. Other ancillary studies such as immunohistochemical or histochemical analysis for the presence of mucin, and genetic mutation analysis and DNA analysis for aneuploidy or polyploidy, were also used. For pancreatic solid lesions, 64% of respondent laboratories used immunohistochemical stains for gene mutations, while other studies such as mismatch repair protein immunohistochemical stains/microsatellite instability and genetic mutation analysis were also performed, though to a lesser extent. For biliary lesions, genetic mutation analysis was the most used ancillary study (66.7%, 98 of 147). Additionally, Ki-67 immunostaining was performed on cytology specimens in 52.6% of respondent laboratories to evaluate pancreatic neuroendocrine tumors on cytology specimens, and 77.9% of those laboratories used Ki-67 proliferation index to grade neuroendocrine tumors on cytology specimens.

In 2015, the Papanicolaou Society of Cytopathology System was proposed to standardize the terminology and nomenclature in pancreaticobiliary cytology, and studies have demonstrated its utility in stratifying risk of malignancy in different categories.^24–27  However, our results revealed only 36% (193 of 530) of respondent laboratories used this system in 2021, although this percentage increased from 16.1% in the 2020 NGC 2020-D Terminology Supplemental Questionnaire. Recently, the World Health Organization (WHO), the International Academy of Cytology, and the International Agency for Research on Cancer have revised the Papanicolaou Society of Cytopathology System to propose an updated international classification system for reporting pancreaticobiliary cytology with a 7-tiered system.^37–39  It will be interesting to see if this newly proposed WHO Reporting System for Pancreaticobiliary Cytopathology will gain any popularity in a future survey.

When laboratories were stratified by their institution types, some significant differences were identified in specimen volume (higher volumes in academic hospitals), specimen type (pancreatic duct brushing evaluated by more academic hospitals than other institutions), ROSE practice and case sign-out practice (general pathologists performed ROSE; signed out cases; and signed out concurrent surgical biopsy, together with cytology specimens, more frequently in nonacademic laboratories than academic laboratories, indicating more subspecialized practice in academic setting).

Some significant differences existed between US domestic and international laboratories regarding specimen types, with biliary brushings and pancreatic duct brushings being evaluated more frequently in domestic than international laboratories. Slide preparations using CB and LBC methods were more frequently used in domestic laboratories, whereas conventional smears and cytospins were more frequently used in international laboratories. Other differences existed in ROSE and case sign-out practice, indicating more subspecialized practice in domestic laboratories.

This is the first survey study from the CAP to investigate pancreaticobiliary cytology practice among participating domestic and international laboratories, and our results provide a baseline for future studies in a variety of practice settings. Additionally, some practice differences were identified among different types of institutions and between domestic and international laboratories.