Prospective Randomized Trial to Compare the Safety, Diagnostic Yield and Utility of 22-gauge and 19-gauge Endobronchial Ultrasound Transbronchial Needle Aspirates and Processing Technique by Cytology and Histopathology

Abstract

Introduction:

Endobronchial ultrasound (EBUS)-guided transbronchial needle aspirate (TBNA) is a widely used method of minimally invasive lymph node sampling. The benefit of processing samples by cytologic methods versus “core biopsy” is unclear. It is unknown if safety or diagnostic yield varies by needle gauge.

Materials and Methods:

Between June 2018 and July 2019, 40 patients (56 lesions) undergoing EBUS TBNA lymph node evaluation were enrolled in this single center prospective trial. Patients were chosen by permuted block randomization to undergo EBUS TBNA starting with 22-gauge or 19-gauge needles. Separate samples were sent for processing by cytologic methods and histopathology. Surgical pathologists and cytopathologists were blinded to needle size. The primary endpoint was diagnostic yield. Secondary endpoints compared specimen adequacy by rapid on-site evaluation (ROSE), sample adequacy for molecular testing, sample quality, and safety.

Results:

Diagnostic yield for histopathologic examination was 87.5% and 83.9% for 19g and 22g respectively (P-value 0.625). There was no significant difference in diagnostic yield by cytologic examination based on needle size. There was no significant difference in slide quality. Molecular adequacy for core-biopsy was 77% and 80% for 22-gauge and 19-gauge needles, respectively. Molecular adequacy for cytology cell block was 77% and 80% for 22-gauge and 19-gauge needles, respectively. There were no significant procedural complications.

Conclusion:

Both the 22-gauge and 19-gauge EBUS TBNA needles provided a similar diagnostic yield and clinical utility for ancillary testing. Processing techniques by cytologic methods or “core biopsy” showed no significant impact in diagnostic yield or utility of molecular testing.

Introduction:

Bronchoscopy has emerged as a minimally invasive technique for achieving diagnosis and staging of lymph nodes and lung lesions by acquiring TBNA without the need for mediastinoscopy or surgical lung resection. [1] [2] EBUS guided TBNA has become a central component of thoracic lymph node staging in both early stage disease and locally advanced disease. [3] Accurate high-quality lymph node biopsies are imperative in the staging lung cancer and determining treatment options. [4] Needle biopsies tend to be much smaller than excisional biopsies and the demand for tissue continues to increase as more information can be gleaned from tumor samples. Today, needle biopsy specimens are routinely used to evaluate cytomorphology, immunohistochemical markers, mutational analysis, and advanced genetic panels. [5] This ever increasing battery of tests can quickly exhaust a biopsy specimen, leaving pathologists and oncologists needing more. Bronchoscopists have met this challenge by using minimally invasive techniques to collect larger and larger tissue samples. [3]

There has been much debate over the optimum needle size. Needles of different sizes have proven to be safe for use in transbronchial needle aspirates. [6] Intuitively, a larger needle should provide a larger biopsy sample; this has proven to be true in a controlled laboratory setting and a pre-clinical laboratory animal study. [7] [8] Clinical studies comparing EBUS TBNA needles reported in the literature have shown more variability. In some studies, larger needles have been shown to garner larger tissue samples or improved diagnostic yield while other studies have shown no benefit in yield of molecular adequacy. [9] [10] [11] [12] [13] Linear EBUS TBNA has been shown to be an excellent modality to collect lymph node samples for tumor mutational analysis and genotyping. [14] [15]

Historically, “core biopsy” was defined by a large gauge needle which would capture tissue fragments for histologic interpretation by a surgical pathologist, whereas fine needle aspirations (FNA) were small clusters of cells evaluated by a cytopathologist. But, it is important to highlight that histologic and cytologic samples typically differ in both size and processing technique. We sought to investigate whether the needle size and processing by cytology or histopathology methods would change the yield or quality metrics, especially the feasibility of molecular testing.

Material and Methods:

Between June 2018 and July 2019, 40 patients with hilar and mediastinal lymphadenopathy detected by chest CT scheduled to undergo EBUS-guided TBNA were enrolled in a prospective randomized study to contrast the diagnostic yield and utility of 22g (Olympus ViziShot, Redmond, WA, USA) and 19g (Olympus ViziShot Flex, Redmond, WA, USA) EBUS TBNA. The study was approved by the Fox Chase Cancer Center Institutional Review Board and all patients were consented with approved informed consent process.

The study flowchart diagram is shown in Figure 1. EBUS - TBNA was performed on an outpatient basis under general anesthesia with continuous monitoring. Any patient undergoing bronchoscopy who was on anticoagulation or antiplatelet agents had those agents held for the appropriate interval prior to bronchoscopy. No procedures were performed with patients on therapeutic anticoagulation or antiplatelet agents. All procedures were performed using a Fuji EBUS bronchoscope (Fujifilm Medical Systems, Lexington, MA, USA). Lymph nodes and central lung masses were identified by EBUS and biopsied with both 22-gauge (22g) and 19-gauge (19g) needles to generate samples for cytologic evaluation and separate samples processed as core biopsies for histopathologic assessment. Only one or two lesions per patient were sampled for this study, even when multiple other lymph nodes were sampled for the staging purpose. The needle size to be used first (19g vs 22g) was chosen by a permuted block randomization scheme. Three separate randomization schema were used, namely lymph nodes 5–10 mm, lymph nodes >10 mm, and central lung lesions. All passes were done with the first needle and the biopsy regimen was repeated using the alternate size needle. All lesions were biopsied with both needle sizes to provide paired samples that serve as a control. Additionally, for any patients where two lesions were enrolled, the starting needle size in the second lesion was alternated to serve as an intra-patient control. All procedures were performed by pulmonologists with extensive experience in interventional bronchoscopy.

Figure 1:

Study flowchart diagram

For each needle size, two cytology slides were made with 1 drop of aspirated material and the rest of the material was rinsed into 50 ml centrifuge tube containing a balanced salt solution (BSS). One smear slide was air-dried, stained with Diff-Quik method, and examined in real time for rapid on-Site evaluation (ROSE). The other smear slide was directly fixed in 95% ethanol and subsequently stained with the Papanicolaou method in the cytology laboratory. Two additional passes were placed directly in BSS for the cell block. Three additional passes were directly placed in 10% buffered formalin and processed as “core biopsy” in the histology laboratory. This yielded a total of 12 passes per lesion. ROSE was not performed on core biopsies. Specimen adequacy was defined by the ability of the pathologist to make a definitive diagnosis of normal lymphoid tissue, a benign abnormality (e.g., granuloma) or a specific malignant neoplasm by morphology with or without using immunohistochemical (IHC) stains.

Cytologic evaluation:

The 50 ml tubes were centrifuged at 1700 rpm for 10 minutes. After discarding the supernatant, the sediment was fixed in 20 ml of 10% buffered formalin and centrifuged again at 1700 rpm for 10 minutes to collect the sediment. A small amount of warm HistoGel medium (Richard-Allan Scientific, San Diego, CA, USA) was added to well-drained sediment in the centrifuge tube, mixed, cooled, placed into Histo-Wrap (Obex Industries, Chagrin, OH, USA) and processed as cell block in the histology laboratory. Four-micron paraffin-embedded sections were made at three levels and stained with hematoxylin and eosin (H&E). The smear slides and cell block sections were examined by a cytopathologist blinded to the needle size. IHC stains were performed on cell block sections, when necessary for classification of the tumor. Flow cytometric analysis was performed on the FNA material only when there was a past history of lymphoma, clinical suspicion of lymphoma or presence of atypical lymphoid cells on ROSE. Final cytologic diagnosis was made and reported by a cytopathologist, independent of the result of “core biopsy”.

Histopathologic evaluation:

After formalin fixation, visible tissue fragments were collected and processed as core biopsies in a routine fashion, embedded in paraffin block and the 4 μm sections were stained with hematoxylin and eosin. The core biopsies were examined by a surgical pathologist blinded to the needle size and reported independent of cytology results in a separate surgical pathology report. In addition to rendering a diagnosis, the pathologist evaluated each sample for the proportion of interpretable tissue on each slide (1–25% tissue, 26–50% tissue, 51–75% tissue, 76–100% tissue). The remainder of the material was blood.

Molecular evaluation:

Cytology cell block and core biopsies were evaluated by a molecular pathologist to determine the adequacy of the samples for routine molecular tests by estimating the number and proportion of tumor cells in the sections. We defined adequacy for molecular testing as the presence of a minimum of 150–200 viable tumor cells in one section with at least 20% ratio of tumor cells to total nucleated cells in the section. In our experience this amount of cellularity in a section provided adequate material for tests ordered by our oncologist during the study period, which included single gene EGFR, BRAF and KRAS hot spot(s) mutation testing by Sanger sequencing and ALK, ROS1 and RET by fluorescence in situ hybridization (FISH).

Complications:

Bleeding was evaluated by a scale previously described by Folch et al. [16] All patients were called by clinic staff the day after the procedure to assess for complications and follow up was performed at day 15 by the treating physician.

Statistical Analyses

We characterized the data using descriptive statistics such as percentages. We used exact McNemar’s tests to investigate diagnostic yield and molecular adequacy differences between 19g and 22g needles. We used a Stuart-Maxwell test to investigate pathology slide grade differences between needle sizes. We used Cohen’s Kappa statistic to assess the molecular adequacy agreement between needle gauges. We used a logistic regression estimated by Generalized Estimating Equations to investigate whether the needle effect on diagnostic yield differed between “core biopsy” and cytology smear & cell block (i.e. interaction effect).

Results:

Forty patients with a clinical need for lymph node evaluation were enrolled in the study with 56 lesions biopsied. Men and women, regardless of race, ethnic group or sexual orientation, were eligible for this study. Final diagnoses based on FNA and/or “core biopsy” results are shown in Table 1 (by patient) and in Table 2 (by lesion). These included 30 patients with neoplasms (26 primary or metastatic carcinoma, 1 thymoma, 3 lymphoma), 4 with granulomatous inflammation and 6 with normal lymphoid tissue. Of the three lymphoma cases, the diagnosis was aided by flow cytometry in one case of mantle cell lymphoma and by IHC stains in one case of large B-cell lymphoma and one Hodgkin lymphoma.

Table 1:

Diagnosis by patient, based on final FNA and/or core biopsy report (n = 40)

Diagnosis by Patient	Number of Patients
Carcinoma	26 (65.0%)
Thymoma	1 (2.5%)
Lymphoma	3 (7.5%)
Granuloma	4 (10.0%)
Normal Lymphoid Tissue	6 (15%)
Total	40 (100%)

Table 2:

Diagnosis by lesion, based on final FNA and/or core biopsy report (n = 56)

Tissue Diagnosis	Lesions (lymph node or lung mass)
Carcinoma	34 (60.7%)
Non-small cell, poorly differentiated	4 (7.1%)
Adenocarcinoma	14 (25.0%)
Squamous cell	5 (8.9%)
Small Cell	6 (10.7%)
Large cell neuroendocrine	2 (3.6%)
Nasopharyngeal primary	2 (3.6%)
Breast primary	1 (1.8%)
Thymoma	1 (1.8%)
Lymphoma	5 (8.9%)
Mantle Cell	2 (3.6%)
Large B-Cell	2 (3.6%)
Hodgkin	1 (1.8%)
Granuloma	6 (10.7%)
Normal Lymphoid Tissue	9 (16.1%)
Non-diagnostic	1 (1.8%)
Total	56 (100%)

Diagnostic Yield:

Diagnostic yield by “core biopsy” technique was 87.5% and 83.9% for 19g and 22g respectively (p-value 0.625) (Table 3). There was no difference in diagnostic yield by examination of cytology smears: 76.4% of 19g and 76.4% of 22g slides showed adequate material for diagnosis. When cytology cell block was added to the smears, the diagnostic yield increased to 89.3% and 87.5% for 19g and 22g needles, respectively. As shown in Table 3, there was no significant difference in the diagnostic yield (our ability to make a definitive diagnosis and classification of the lesion) between cytology (smears and cell block) and “core biopsy” technique (Figure 2). There were no discrepancies between the final FNA and “core biopsy” diagnoses on 40 lesions with neoplasms and 9 normal-appearing lymph nodes. However, of the 6 lymph nodes with core biopsies showing non-necrotizing granulomas, only 2 corresponding FNA samples showed granuloma, with no difference between the needle sizes. The remaining 4 lymph nodes were reported as inadequate (1) or as normal lymphocytes (3) on cytology cell block, but showed non-necrotizing granulomas on “core biopsy”.

Table 3:

Diagnostic yield of EBUS samples processed as core biopsy and cytologic method with and without cell block

Method	Needle Size		P value
	19g	22g
Core biopsy	87.5%	83.9%	0.625
Cytology smear	76.4%	76.4%	1.00
Cytology smear and cell block	89.3%	87.5%	1.00

The interaction p-value comparing whether the needle effect differs between core biopsy and cytology smear & cell block is p=0.848. Test that the core biopsy rate is equal to the cytology smear and cell block rates are p=1.00 (19g) and p=0.727 (22g).

Figure 2,

A-D: Comparison of representative fields of FNA cell block and corresponding samples processed as “core biopsy” obtained by different needle sizes in the same case (hematoxylin and eosin stain, original magnification 100x). A, 22-gauge FNA cell block. B, 19-gauge FNA cell block. C, 22-gauge “core biopsy”. D, 19-gauge “core biopsy”.

Specimen Quality:

As shown in Table 4, there was no significant difference in the marginal distribution of percent tissue vs blood on slides based on needle gauge.

Table 4:

Core biopsy slides graded for proportion of tissue versus blood

Pathology Slide Grade	19-gauge needle: number of lesions (%)	22-gauge needle: number of lesions (%)	P=0.91
1–25% tissue	28 (50.0)	30 (53.6)
26–50% tissue	12 (21.4)	10 (17.9)
51–75% tissue	8 (14.3)	7 (12.5)
76–100% tissue	8 (14.3)	9 (16.1)

Inadequate Specimens:

Of the 8 cases which had at least one inadequate value on cytology cell block or “core biopsy”, 5 were performed with a 22g needle and 3 with a 19g needle. Needle size, and processing technique did not predict ROSE adequacy or final specimen diagnostic adequacy. Also, lymph node size did not predict an inadequate specimen: 4 lymph nodes were > 10 mm, 4 lymph nodes were ≤10 mm. ROSE slides were judged to be inadequate in 23 of 112 aspirates (12 with 22g needle and 11 with 19g needle).

Molecular Adequacy:

Thirty-five lesions positive for carcinoma or thymoma were eligible for molecular testing based on standard practice. Molecular was determined by a cytopathologist and a surgical pathologist, both blinded to the needle size. With 35 lesions, 2 needle sizes for each lesion, and 2 processing techniques for each needle size, 140 samples were evaluated for molecular adequacy. As shown in Table 5, the molecular adequacies of both needle gauges when separated by processing technique and needle size were similar. The processing techniques for each needle size were compared within the lesions in 70 paired samples, as summarized in Table 6. Each data point indicates agreement between cytology cell block and pathology “core biopsy” for a given needle gauge used to generate specimens within the same lesion. There were 8 biopsy samples (4 were performed with 22g needle and 4 were performed with 19g needle) where within a single lesion and for a single needle gauge, molecular adequacy of “core biopsy” and cytology cell block were not in agreement. There were 4 biopsies that were inadequate by cytology but adequate by “core biopsy”, and 4 biopsies that were inadequate by “core biopsy” but adequate by cytology. Of the 11 samples which were inadequate by both cytology cell block and “core biopsy”, all were from 6 lymph nodes. For five of the lymph nodes, all samples regardless of needle size or processing technique were adequate for diagnosis but inadequate for molecular testing. Of those 11 inadequate samples, 6 were performed with 22g needle and 5 with 19g needle. Overall Kappa agreement of processing technique for all paired samples was 0.66, which is considered a substantial degree of agreement.

Table 5:

Molecular adequacy for 35 lesions with carcinoma or thymoma, separated by processing technique and needle gauge

Core Biopsy		Cytology with Cell Block
22-gauge	19-gauge	22-gauge	19-gauge
27/35 (77%)	28/35 (80%)	27/35 (77%)	28/35 (80%)

p=1.00 for the comparison of 19-gauge versus 22-gauge in both cases.

Table 6:

Molecular adequacy agreement within lesion.

		Cytology Cell Block
		Inadequate	Adequate
Pathology Core Biopsy	Inadequate	11	4
	Adequate	4	51

Kappa 0.66

Complications:

In one case, grade 1 bleeding occurred after the 19g needle was used to perform TBNA on an 11L lymph node. Hemostasis was confirmed prior to switching to the 22g needle. After needle passes were performed with the 22g needle, blood was noted in the airways and use of a diagnostic scope was required to assess bleeding and clear blood clots from the airways. Once the clot was cleared from the biopsy site there was no active bleeding and no intervention was required.

One patient was admitted to an outside hospital with fevers and lethargy 8 days after bronchoscopy. They were admitted for 2 days and treated with IV antibiotics for presumed pneumonia.

Discussion:

Our study demonstrates that both 22g and 19g needle sizes are safe in EBUS TBNA for uses in biopsy of lymph nodes at least 5 mm in size and in central lung lesions. This study demonstrated a high diagnostic yield with both 22- and 19g EBUS needles which is in keeping with previously published studies which have shown similar diagnostic yield independent of needle size. [8] [11] [12] [13] [17] [18] However, our study reports also on quality of the biopsy specimens in a number of ways. First, we assessed adequacy of cytology slides (made during the procedure for ROSE) and cell block for diagnosis by morphology and immunohistochemistry. Second, quality of the samples processed as “core biopsy” was assessed by a blinded pathologist by measuring the percent of tissue to blood. Third, we compared the diagnostic yield of needle aspiration material processed by two different methods, as cytology cell block and as “core biopsy.” Fourth, the samples processed as cytology cell block and “core biopsy” were assessed for molecular testing adequacy.

Patients enrolled in this study had high suspicion for malignancy. No more than two lesions were enrolled for any one patient and therefore, study specimens were taken only from the most suspicious lesions. All of the lymph nodes were clinically abnormal either by size criteria of at least 10 mm or combination of size of 5–10 mm with other suspicious ultrasound characteristic (loss of central hilar sign, shape, heterogeneity, or vascularity), thus explaining the high incidence of malignant lymph nodes and low rate of benign findings. Interestingly, the histologic examination found four additional lymph nodes with granulomas which were interpreted as normal lymph nodes by cytologic examination. The small samples size prevents generalization, but this is in keeping with other studies evaluating the utility of 19g needle in granulomatous disease. [17]

A previous study comparing average cell area of slides made with centrifuged specimens obtained with 21g and 19g needles showed a significant difference in average tissue area of slides obtained with a 19g needle. [9] In our study, a potential weakness is our inability to quantify, either by mass or volume, the amount of tissue obtained with each needle. To overcome this, we combined multiple metrics to evaluate overall utility of both needle sizes and processing techniques. We used a set number of needle passes for both needle gauges. There were 6 needle passes of 15–20 agitations for each needle size (3 passes for cytology and 3 passes for pathology) for a total of 12 needle passes per lesion. Our protocol standardized the number of needle passes used so that we could compare tissue retrieval over the same conditions for each needle gauge. Thus our results cannot determine the number of needle passes that provide optimum sample collection for each needle gauge.

In the analysis of the quality of samples processed as “core biopsy”, we measured the amount of tissue vs blood on the slide. This analysis was performed using pathology slides made from material that had not been centrifuged and were interpreted by a pathologist blinded to the needle size. We found no difference in the amount of tissue vs blood collected by the two needles. When separating the slides based on which needle was used first, agreement improved slightly in lesions where the 22g was used first. However, the degree of concordance did not rise above the level of fair. Three passes of each needle was sufficient to provide a high diagnostic yield and meet cellular adequacy for molecular testing. We can conclude that each needle meets the general diagnostic needs for clinical practice but are unable to report whether the 19g needle generates more tissue (or more blood) per pass compared to the 22g needle.

When considering the impact of preparation technique as “core biopsy” versus cytology smears and cell block, there was no significant difference in diagnostic yield for cancer and for molecular adequacy between the two techniques. While the total volume of tissue collected was not compared, samples obtained with each needle reached a similar benchmark in tissue adequacy. This is not surprising, as the same needle and aspiration technique was used to collect the material for both processing techniques. In fact, the amount of tissue in cell block sections was sometimes more than that in the “core biopsy” slides, as seen in Figure 2. This can be explained by the way the blocks are prepared in these two methods. For “core biopsy” visible fragments of tissue and blood clot are pick from the container, whereas for cell block the tube is centrifuged and the entire sediment is collected.

Despite the difference in size, the bronchoscopist performing the procedure did not feel that there was a major technical difference in performing the biopsies. As seen in Figure 3, the needle puncture site of the 19g is larger than the 22g needle. We found that both intra-procedure and post-procedure complications were very low and this was in keeping with previously published studies. [10] [9] [19] Throughout the study, we saw no cases of bleeding which required balloon occlusion, transfusion, or other therapeutic intervention. There were no episodes of desaturation and no patients required admission to the hospital or prolonged mechanical ventilation. At 15 day follow up we found no patient-reported complications. Both pulmonologists performing bronchoscopies subjectively reported that there was more bleeding from puncture sites with the 19g needle compared to the 22g needle, but not to a degree that was clinically meaningful or quantifiable by our bleeding scale.

Figure 3:

Puncture holes with 22g and 19g needles as seen on EBUS.

Conclusion:

Overall, lymph node staging by EBUS is a safe and effective procedure. Based on our data, the diagnostic yield and overall clinical utility of 19g and 22g needles is the same with no significant difference in complication rates. Lymph node size did not impact any of the outcome variables. Based on our findings, the authors conclude there is no benefit in diagnosing malignancy or performing molecular testing on cancer specimens from employing the 19g needle in place of a 22g needle. Further studies are needed to determine if there is benefit in nonmalignant disease. Our study also showed that processing the samples by cytologic methods (smear and cell block) would produce similar diagnostic yield and molecular test adequacy as processing the samples as “core biopsy”. In fact, based on this finding when ROSE suggests malignancy, we have now stopped collecting separate samples for “core biopsy” during EBUS and the entire sample is sent to the cytology laboratory for processing.

Abbreviation list:

EBUS

Endobronchial ultrasound

TBNA

Transbronchial needle aspirate

19g

19-gauge needle

22g

22-gauge needle

ROSE

rapid on-site evaluation

BSS

balanced salt solution

H&E

hematoxylin and eosin