The role of cytology in endobronchial ultrasound‐guided transbronchial needle aspiration: A study of 813 cases focusing on diagnostic yield, an analysis of misdiagnosed cases and diagnostic accordance rate of cytological subtyping

Abstract Background Endobronchial ultrasound‐guided transbronchial needle aspiration (EBUS‐TBNA) is a minimally invasive technique for cytological and histological diagnosis. The objective of this study was to explore the role of cytological diagnosis in EBUS‐TBNAs. Methods Eight hundred and thirteen consecutive cases performed EBUS‐TBNA with both cytological and histological diagnoses were retrospectively reviewed. All patients were followed up for clinical data. Results Before immunohistochemical examination, the cytological sensitivity, specificity, and diagnostic accuracy of EBUS‐TBNAs were 92.9% (421/453), 98.9% (348/352), 95.5% (769/805), respectively. After immunohistochemical examination, the sensitivity, specificity, and diagnostic accuracy were 93.0% (423/455), 99.4% (348/350), 95.8% (771/805), respectively. The majority of false‐negative were cases whose cytological diagnosis was “atypical” or the cytological diagnosis suggested “inadequate.” “Neoplastic” were also prone to false‐negative cytology. The diagnostic accordance rate of cytological subtyping was 90.3% for squamous‐cell carcinoma, 99.2% for adenocarcinoma, and 98.1% for small‐cell carcinoma before immunohistochemical examination, and became 85.9%, 98.5%, and 98.2% after immunohistochemical examination, respectively. Conclusion Cytological diagnosis in EBUS‐TBNAs had a good sensitivity and high specificity. The sensitivity and specificity of cytological diagnosis were proved to be higher after the immunohistochemical examination. At the same time, cytology had high accordance rate in subtype diagnosis. False‐negative results occurred more commonly in cases whose cytological diagnosis was “atypical” or the cytological diagnosis suggested “inadequate” or the corresponding histological diagnosis was “Neoplastic.”


| Ethics approval and consent to participate
The Institutional Review Board of Chinese Academy of Medical Sciences and Peking Union Medical College approved this study and waived the requirement for informed consent.

| Procedures
EBUS-TBNA was carried out in patients under local anesthesia and performed with an echobronchoscope (BF-UC160F-OL8, Olympus, Tokyo, Japan). The locations, adjacent structures, and sizes of the lesions were assessed by color Doppler imaging (EU-C2000). A dedicated aspiration needle (22-gauge NA-201SX-4022) was then placed in the working channel and advanced into the lesion, the stylet was withdrawn, suction was applied to the needle, and the needle was then moved forward and backward within the lesion. A total of 10 to 15 passes per lesion was obtained. The aspirate material was placed onto a glass slide. First, the strip-shaped component was picked from the glass and fixed in formalin, delivered to the pathology department and stained with H&E for further histological diagnosis. Then, the remainder of the aspirates on glass were smeared and immediately fixed in 95% ethanol, while the remaining aspirates obtained from flushing the needle with a small amount of saline was injected into vials of Cytolyt solution (Cytyc Company Products). Next, the smears and material vials were delivered to the cytology room. The smears were stained with H&E, while the material vials were vibrated, centrifuged, and transferred into a vial containing a preservative fluid.
Then, the material vials were inserted into Thinprep 5000 for preparation, and staining was performed with Pap staining.
In some cases, immunohistochemistry (IHC) or immunocytochemistry (ICC) was performed to obtain additional information. Immunohistochemical or immunocytochemical analysis in our study was performed on an autostainer, which was a Ventana Benchmark XT (Ventana Medical Systems, Inc), according to the manufacturer's protocols. Our institution used a panel of thyroid transcription factor-1 (TTF-1), Napsin-A, p40, and p63 to subtype most N-SCLCs. Synaptophysin (Syn), Chromogranin A (CgA), CD56, and Ki-67 were used to identify small-cell carcinomas. Other antibodies were also used when needed for tumor origin identification and differentiation between benign and malignant status. In our series, IHC was used for 125 cases of the histological samples and ICC was used for 21 cases of the cytological samples, no patient had both immunohistochemical and immunocytochemical examination.
The cytological diagnosis was based on a combination of 2 to 4 conventional smears and a liquid-based preparation. The histological diagnosis was based on 1 to 3 H&E slides of biopsy tissue and/or surgical samples.
In the retrospective analysis, we used a six-tiered system to reclassify the samples as follows: "nondiagnostic"; "negative for malignancy/neoplasia"; "atypical"; "neoplastic"; "suspicious (of malignancy)"; and "malignant." "Nondiagnostic" specimen was defined as that with a cytological diagnosis of "negative for malignancy/neoplasia" and a nucleated cell quantity of less than 50 in every slide, regardless of conventional smear or liquid-based preparation. Specimens "negative for malignancy/neoplasia" contained normal, inflammatory, or granulomatous lesions. An "atypical" cytologic specimen demonstrates cytologic features of greater dysmorphology than those assigned to the "negative for malignancy/neoplasia" category but falling short of those assigned to the suspicious for malignancy or neoplasm. Specimens that were "neoplastic" meant that the specimens were benign neoplasm, or low-grade cancer, and the nuclear atypia were mild. In our study, these tumors included carcinoids, atypical carcinoids, and low-grade malignant tumors of the salivary gland. The classification of "suspicious" meant that the specimen was deemed suspicious of malignancy. A "malignant" specimen indicated a malignant tumor, and malignant tumors could be further divided into "squamous-cell carcinoma," "adenocarcinoma," "small-cell carcinoma," "non-small-cell carcinoma," "unclassified carcinoma," or "other special type." In addition to these six categories, when "negative for malignancy/neoplasia" specimens contained a nucleated cell quantity of more than 50 but less than 200 or "atypical" specimens contained a nucleated cell quantity of less than 200 in conventional smear or liquid-based preparation, the cytological diagnosis will suggest the specimens were "inadequate." The histological and cytological diagnoses of all cases were performed by senior pathologists and cytologists.

| Statistical analysis
All patients were followed for at least 3 years. Nondiagnostic cases were excluded from subsequent analysis, nondiagnostic rate is the number of "nondiagnostic" divided by the total number of specimens.
The gold standard for true diagnosis is based on a histological diagnosis or clinical diagnosis, and we compared the accuracy of cytological diagnosis with this gold standard.
A cytological or histological positive was defined as a diagnosis of "neoplastic," "suspicious," or "malignant." A clinical positive was defined by X-ray or CT scans showing clear lesions or metastases (including in the lungs, mediastinum, brain, liver, bone marrow, etc.) and the implementation of chemotherapy and/or radiotherapy. A cytological or histological negative was defined as a diagnosis of "negative for malignancy/neoplasia" or "atypical." A clinical negative was defined by no growth upon repeat imaging or testing negative for malignancy upon repeat biopsy or surgery during the follow-up period.
In our study of the cytological positive and histological negative patients, 33 had a subsequent clinical positive result. The 33 histological diagnoses were all based on biopsy tissues, and all reports noted that the tissue amounts present were minimal. The 33 cases were included as cytological true positives.
The diagnostic sensitivity, specificity, and accuracy were calculated using the following standard definitions: sensitivity-the probability of obtaining a positive test result for a subject with the disease; specificity-the probability of obtaining a negative test result for a subject without the disease; and diagnostic accuracy-the proportion of correctly classified subjects among all subjects.
In the diagnostic accordance rate of cytological subtyping for adenocarcinoma, squamous-cell carcinoma, and small-cell carcinoma, those cannot get a definitive subtyping were also included in the accordant cases. For example, the discordant cases of cytological adenocarcinomas were the cases whose cytological diagnosis was adenocarcinoma while the corresponding histological diagnosis was "negative for malignancy/neoplasia," "atypical," "other special type," "squamous-cell carcinoma," or "small-cell carcinoma"; the accordance number of cytological adenocarcinomas was the total number of cytological adenocarcinomas minus the discordant cases of cytological adenocarcinomas. The diagnostic accordance rate of cytological subtyping for adenocarcinoma was the accordance number of cytological adenocarcinomas divided by the total number of cytological adenocarcinomas.

| Nondiagnostic rate and inadequate rate
Eight out of 813 cases with few cells were categorized as "nondiagnostic," with the nondiagnostic rate of 0.98% (8/813), and the remaining 805 cases were included in the further analysis. Thirty cases suggested "inadequate," with the inadequate rate of 3.7% (30/813).

| Location
Of 805 EBUS-TBNA samples, those from the peritracheal lymph node/lump accounted for 94.0%, those from the mediastinal lump/ lymph nodes accounted for 2.8%, and those from the peritracheal lung lump accounted for 3.2%.

| The diagnostic results
The diagnostic details are shown in Table 1. The numbers in brackets are the results obtained before immunohistochemical examination.
IHC was carried out in 125 histological cases, and 30 cases corresponded to a changed diagnostic grade after IHC, with the specific changes shown in Table 2. ICC was carried out in 21 cytological specimens (no cases were examined by both IHC and ICC), all 21 cytological specimens were "malignant" before ICC and the purpose of the examination was to further subtype, these diagnoses grades did not change after immunocytochemical examination.

| The diagnostic yield of cytology
According to different diagnostic grades, we divided cytological diagnosis and histological/clinical diagnosis into negative and positive groups. After IHC, the groups of four cases (these four cases are shown in bold in Table 2) were changed; overall, the number of falsenegative cytology did not change (false-negative cytology increased and decreased by one case, respectively), while the number of falsepositive cytology was reduced by two cases; this change resulted in the total number has changed from 453 to 455 for the calculation of sensitivity, and from 352 to 350 for the calculation of specificity after IHC. The diagnostic grouping results are shown in Table 3

| The diagnostic results of cytology in falsepositive cases
There were two false-positive cases. One was diagnosed as "suspected adenocarcinoma" cytologically, while the histological diagnosis was "granulomatous nodule," and the other was diagnosed "suspected small-cell carcinoma," but the histological diagnosis was "degenerative cells, not enough to diagnose malignancy."

Cytologic (N) Histologic (N) Cytologic (N) Histologic (N)
Adenocarcinoma (1) Granuloma (1) Adenocarcinoma (2) Granuloma (1) Squamous-cell carcinoma (1) Squamous-cell carcinoma (6) Adenocarcinoma (6) Squamous-cell carcinoma (9) Adenocarcinoma (8) Neoplastic (1) Small-cell carcinoma (2) Atypical (2) Small-cell carcinoma (2) Atypical (1) Adenocarcinoma (1) Note: N is the number of cases. treatment and management. In this classification system, EUS-FNA specimens include six diagnostic categories: "Non-diagnostic," "Negative for malignancy/neoplasia," "Atypical," "Neoplastic (benign or other)," "Suspicious," and "Positive/malignant," 2 while the P.S.C also issued guidelines on standardized terminology and nomenclature for respiratory cytology, this system include the categories "nondiagnostic," "negative (for malignancy)," "atypical," "neoplasm, benign neoplasm, and low-grade malignancy," "suspicious for malignancy," and "malignant." 3 Referring to this classification systems, we used a six-tiered system to reclassify the EBUS-TBNA specimens as follows: "nondiagnostic," "negative for malignancy/neoplasia," "atypical," "neoplastic," "suspicious (of malignancy)," and "malignant." The nondiagnostic and adequacy criteria for EBUS-TBNAs were not well established, 4 and the criteria used by various groups in the literature ranged from simply noting the presence of lymphocytes/lymphoid tissue 5,6 to more quantitative measures, such as the existence of >40 lymphocytes per high-power field 7 or >5 low-power fields with >100 lymphocytes in each and <2 bronchial cell groups per low-power field. 8 In our study, specimens whose cytological diagnosis was "negative for malignancy/neoplasia" and which contained a nucleated cell quantity of less than 50 on the slides, regardless of conventional smear or liquid-based preparation, were considered nondiagnostic specimens and were excluded from subsequent analysis. Our nondiagnostic rate was 0.98%, which was lower than that reported in the literature. 9,10 This criterion was easier to apply in clinical diagnosis, but a minimum nucleated cell quantity of 50 as a diagnostic sample was still low, it will cause a certain false-negative rate.
EBUS-TBNA has been reported to have a cumulative sensitivity of 88.9% to 91.5%, 7,11-13 a cumulative specificity of 96.4% to 100%, 7,11-13 and a cumulative diagnostic accuracy of 93.0% to 93.2%, 7,11 but the sample sizes were all small. In this study, we analyzed 805 EBUS-TBNA samples, and the diagnostic details are shown in Table 1. We calculated the diagnostic yield of cytology before and after IHC. The sensitivity, specificity, and diagnostic accuracy of the EBUS-TBNA cytology were 93.0% (92.9%), 99.4% (98.9%), and 95.8% (95.5%) ( Table 3), respectively (the percentages in brackets are the sensitivity, specificity, and diagnostic accuracy before IHC). We found that the sensitivity, specificity, and diagnostic accuracy of EBUS-TBNA cytology all increased after IHC verification, this was because of some cytological results that were inconsistent with the histology before IHC but were confirmed to be correct after IHC verification (Table 2). To some extent, this finding can demonstrate that cytological diagnosis based on EBUS-TBNAs has certain advantages when there are no conditions for IHC.
It is worth noting that 33 cases had a "cytological positive" but "histological negative" diagnosis. In the subsequent clinical follow-up, these 33 cases were all verified to be clinically positive. This indicates that histological diagnoses based on tissue biopsies can give falsenegative results. Although the cell count of the 33 cytological specimens was lower, positive diagnoses were still able to be obtained, which shows that cytology has an advantage in the diagnosis of sparsely cellular specimens. Inadequate sampling would also lead to a diagnosis of "atypical" if the locations were at the edge of the malignant tumor. For stromal elements of the tumor or necrosis, an "atypical" diagnosis was easy to obtain. False negatives caused by "atypical" accounted for 31.3%
Of course, some "atypical" cases were due to interpretation errors, cases with a histological diagnosis of "neoplastic" are often  if these tumors are detected early, they are classified as more generic tumors to distinguish them from highly aggressive malignancies and to provide management flexibility for older patients with small tumors where the risk benefit of surgery is greater than that of conservative treatment.
Despite some difficulties, "neoplastic" is not undiagnosable in cytology. Carcinoid, for example, is generally identifiable in cytological specimens in previous literature reports. [16][17][18] Carcinoid tumor cells are discohesive, fragments of capillaries are a helpful clue if present.
Necrosis is absent and the background is clean. Individual cells are small and have round, oval, or spindle-shaped nuclei. Nuclear outlines are smooth. Chromatin is finely granular and nucleoli are not present.
In addition to the three types of tumors that were involved in our study, "neoplastic" also includes "pulmonary hamartoma," "sclerosing pneumocytoma," "myoepithelial neoplasms," and so on. The cytological characteristics of these tumors have been described in detail in the literature. 3 On the premise of understanding the cytological characteristics of these tumors, combined with clinical, imaging and immunocytochemical examination, it is possible to make a correct cytological diagnosis of these tumors to some extent, the key to making the right diagnosis is to be aware of the presence of such tumors.

| False-positive cases
There were two false-positive cases, including one with a cytological diagnosis of "suspicious malignancy" but the histological diagnosis of a granulomatous lesion. From the 805 EBUS-TBNAs, 119 cases of granulomatous lesions were diagnosed by histology, while the cytological diagnoses included 25 "granulomatous," 93 "negative for malignancy/ neoplasia," and 1 "suspicious malignancy." The sensitivity of a "granulomatous" diagnosis in cytology was 21.0%, which was lower than that reported in the literature. 19,20 The other false-positive case was a suspected cytologically small-cell carcinoma, but the histological diagnosis was "degenerative cells, not enough to diagnose malignancy." The follow-up examination of this patient was not carried out in our hospital, so we still consider this patient as a false-positive case. cancer. ICC is a powerful tool with which to improve the accordance rate of cytology for subtyping. 21 In the current study, the cases of misdiagnosed tumor subtypings did not implement ICC. For the 805 EBUS-TBNA specimens with cytology and histology, the implementation ratio of ICC and IHC was 1:6 (21-125). In recent years, with the continuous improvement of cell block technology, it is completely feasible to implement IHC on cell blocks. 22,23 It is believed that with continued clinical application, the role of cell blocks in subtyping can be further improved and that the diagnostic accordance rate of cytological subtyping will be further improved.

| The accordance rate of tumor subtyping in cytology
Our research included a deep discussion of the cytological value of EBUS-TBNA, but there were some limitations in our study. First, although the data collected in this research were large in number, this single-center study possibly included some bias. Second, ROSE was not used. Third, our study was a retrospective study, and immunohistochemical examination was not performed on all cases with undetermined subtypes, so the accordance rate of our tumor subtyping was inaccurate.

| CONCLUSIONS
In summary, in this study, we investigated a large series of EBUS-TBNA results and observed that this method has good sensitivity, high specificity, and a high accordance rate with tumor subtyping based on cytology. Cytological diagnoses of EBUS-TBNA samples have high value in clinical applications. Cytology has a certain diagnostic advantage when there are few puncture specimens, which can make up for the occurrence of false-negative histology results, but the existence of too few cells is also the main reason for a decline in the cytological diagnostic efficiency.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.