M cer has been shown to have both high sensitivity

Transcription

1 Intraoperative Imprint Cytology for Evaluation of Mediastinal LvmDhadenoDathv J I I J Martha R. Clarke, MD, Rodney J. Landreneau, MD, and Dennis Borochovitz, MB, BCh Department of Pathology and Section of Thoracic Surgery, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvarda Frozen-section (FS) analysis of mediastinal lymph nodes is commonly used in the staging of lung cancer and the evaluation of diagnostic tissue at mediastinoscopy. This approach facilitates definitive surgical intervention in a single operation and reduces costs. However, FS analysis can be labor intensive for the pathology department and time-consuming while the patient is anesthetized. Imprint cytology is more rapid than the FS procedure (average, 2 minutes versus 11 minutes per node) and allows more extensive sampling of the specimen. In this prospective study, we compared the diagnostic accuracy of imprint cytology and permanent sections on 121 mediastinal lymph nodes from 38 patients. There were no false-positive results and one false-negative result, al- though that patient was correctly classified based on positive cytology from another node. The sensitivity was 96.6%, the specificity was loo%, and the predictive value of a positive result was loo%, as no false-positives results were observed. The predictive value of a negative result was 98.9%, and the overall efficiency was 99.2%. These results compare favorably with those in other studies comparing the diagnostic accuracy of imprint cytology with that of FS analysis and with reported accuracy rates of FS technique. Our findings confirm the usefulness of this technique as an adjunct or substitute for FS analysis in the intraoperative pathologic evaluation of mediastinal lymphadenopathy. (Ann Tkorac Surg 2994;57:220ti-20) ediastinoscopy in preoperative staging of lung can- M cer has been shown to have both high sensitivity and high specificity superior to chest roentgenography, computed tomography, or magnetic resonance imaging in detecting mediastinal tumor spread [l-31. When used in conjunction with computed tomographic scanning, it is thought to be the most cost-effective approach to preoperative staging of lung cancer to determine resectability [4, 51. Many physicians believe that its routine use is justified by the 7% to 10% chance of finding mediastinal spread in radiographically normal-sized lymph nodes to avoid thoracotomy in patients who would benefit little from resection [6]. The associated morbidity and mortality of mediastinoscopy are negligible [7, 81, and the procedure can be performed in an ambulatory setting [9]. Mediastinoscopic staging of lung cancer using intraoperative frozen-section (FS) diagnosis and immediate thoracotomy in candidates for resection is used at this institution. This approach, combining staging with definitive surgical intervention, is more efficient in that it reduces total operating and anesthesia times, lessens operative risks, shortens hospital stay, and reduces hospital costs. Gephardt and Rice [lo] reported in 1990 their results using FS analysis of lymph nodes at mediastinoscopic staging of lung cancer and thoracotomy. They reported a high sensitivity, high specificity, and high accuracy com- Accepted for publication Aug 20, Presented at the 1993 meeting of the Society of Surgical Oncology, Los Angeles, CA, March 1&21, Address reprint request to Dr Clarke, Department of Pathology, Montefiore, University of Pittsburgh Medical Center, 200 Lothrop St, Pittsburgh, PA 1521X2582. pared with staging by permanent section. The falsenegative rate was only 1.6%, and the predictive value of a negative FS analysis was 94.7%. In addition, they calculated a cost saving of approximately 15% of the total hospital bill when staging was performed by FS analysis and mediastinoscopy combined with thoracotomy during one procedure. The use of intraoperative FS analysis in lung cancer, however, has certain drawbacks when used for staging. The FS procedure is time-consuming and labor intensive, especially when multiple samples are submitted for analysis within a short period or simultaneously, as occurs in staging procedures. Most FS laboratories have only one or two cryostats available, and, as each specimen must be separately embedded and frozen and then sectioned and stained, substantial delays can occur when multiple nodal stations are sampled. In addition, although frozen sections provide architectural detail comparable to that of permanent sections, FS slides often show artifactual changes secondary to the process of freezing and sectioning, which limit the ability of the pathologist to inteirpret the specimen. The addition of intraoperative cytology using fineneedle aspiration biopsy to FS analysis has been suclcessfully applied to the diagnosis of intrathoracic lesions as well as the diagnosis of mediastinal masses and meldiastinal and hilar lymph nodes. Christ and Fry [ll] found fine-needle aspiration cytology to be diagnostically accurate in the pathologic evaluation of tissue at the time of thoracotomy. It was particularly helpful in the case of deep parenchymal pulmonary lesions and mediastinal masses from which it was difficult to obtain biopsy specimens, especially masses close to major blood vessels by The Society of Thoracic Surgeons /94/$7.00

2 Ann Thorac Surg 1994; CLARKE ET AL 1207 Table 1. Pathologic Diagnoses in Patients Having Mediastinoscopic Lymph Node Evaluation Group 1 (IC) Group 2 (FS) Final Diagnosis (n= 38) (n = 36) Adenocarcinoma 22 9 Squamous cell carcinoma 8 10 Small cell carcinoma 1 3 Large cell carcinoma 1 2 Carcinoid tumor 1 Lymphoma/Hodgkin s disease 2 3 Sarcoidosis 2 3 Fungal infection 2 Granulomas 2 Reactive lymphoid hyperplasia 1 2 FR = frozen section analysis; IC = imprint cytology. An alternative approach to diagnose lymph nodes at the time of mediastinoscopy, or imprint cytology (IC), was undertaken at this institution. In IC, the tissue is touched several times to a glass slide or gently scraped with a scalpel, and the material is spread on a glass slide. These slides are immediately stained and interpreted, thus bypassing the time and steps involved in embedding, freezing, and cutting a tissue section. This technique allows a more rapid turnaround time and reduces delays when multiple nodes are sampled. Imprint cytology differs from fine-needle aspiration biopsy in that the entire surface area of the specimen can be sampled and multiple imprints made from the specimen. This allows more rapid sampling of a larger specimen than is possible by FS analysis. Also in IC, the integrity of the tissue is maintained. This is important in the case of small specimens, which can be entirely used up in the FS procedure. Therefore, tissue is still available for permanent section, FS, or special studies. In addition, IC eliminates the risk of cryostat contamination in infectious cases, including patients with tuberculosis and patients with known human immunodeficiency virus infection. In this study, we evaluated the accuracy of IC versus FS analysis in diagnosing mediastinal adenopathy at mediastinoscopy. We also evaluated the accuracy of IC versus FS analysis in mediastinoscopic staging of lung cancer. Material and Methods Patients Patients undergoing mediastinoscopy at Montefiore University Hospital from September 1990 until October 1991 were randomly evaluated at the time of operation by either intraoperative IC (group 1) or FS analysis (group 2). There were 38 patients in group 1, 24 men and 14 women with a mean age of 62.4 years (range, 33 to 79 years). Thirty-three were undergoing staging of lung cancer, and 5 were being evaluated for mediastinal lymphadenopathy in the absence of a pulmonary lesion. Group 2 comprised 36 patients, 25 men and 11 women with a mean age of 59.7 years (range, 21 to 76 years). Twenty-four patients were- undergoing staging of lung cancer, and 12 were being evaluated for mediastinal lymphadenopathy. Final pathologic diagnoses are listed in Table 1. Operative Technique The cervical mediastinoscopy was performed in a standard fashion as described previously by Ginsberg [8]. During the course of the mediastinoscopic examination, the nodal stations routinely sampled included the subcarinal plane (level 7) and both right and left low paratracheal planes (level 4). Imprint Cytology Technique Lymph nodes or fragments thereof received in saline solution are blotted dry, and a fresh cut area or an area free from capsule is gently scraped with a clean blade (Fig 1A). This material is immediately spread on a glass slide and fixed in 95% ethanol (Fig 1B). Alternatively, small tissue fragments can be directly imprinted or A B Fig 1. Imprint cytology technique: (A) Lymph node free from capsule is gently scraped with a clean scalpel. (B) Material is then spread on a glass slide and immediately fixed.

3 1208 CLARKE ET AL Ann Thorac Surg 1994;57120&10 smeared on a glass slide and fixed. Most important in the process is rapid smearing and immediate fixation to prevent air-drying artifact. Slides are then stained with a rapid hematoxylin and eosin stain as follows: hematoxylin (Richard Allan Medical, Richland, MI) for 60 seconds followed by water rinse and differentiation in ammonium hydroxide; eosin (Richard Allan Medical) counterstain for 30 seconds followed by dehydration in 95% and 100% ethanol and clearing in xylene. Slides are mounted with Accu-Mount (Baxter Scientific Products, McGaw, IL) and coverslips placed. To complete smear preparation and staining averages 2 minutes in our laboratory. Frozen-Section Technique Lymph node fragments or sections no larger than 1.0 cm and 2 to 3 mm thick are frozen in Optimal Cutting Temperature Medium (Miles Laboratory, Elkhart, IN) on a brass chuck in a -23 C cryostat. Six-micron sections are cut, placed on glass slides, and fixed for 1 minute in 95% ethanol. They are then stained by the same rapid hematoxylin and eosin technique used for imprint slides. To complete preparation and staining of a frozen section averages 11 minutes in our laboratory. Permanent-Section Technique All remaining tissue after imprint preparation is placed in 10% neutral-buffered formalin for processing as permanent sections. Frozen tissue is thawed and placed in 10% neutral-buffered formalin for permanent sections. All tissue remaining from mediastinoscopic biopsy specimens is submitted for permanent sections, including all nodes from all nodal stations. Pathologic Analysis Four pathologists, all of whom had experience in cytopathologic diagnosis, participated in the initial handling and interpretation of the specimens. Slides from imprints, frozen sections, and permanent sections of lymph nodes obtained at mediastinoscopy as well as subsequent resec- Fig 2. Imprint slide from a benign anthracotic mediastinal lymph node. The larger cells are histiocytes with anthracotic pigment. Benign lymphocytes are present in the background. (Hematoxylin and eosin; ~150 before 50% reduction.) Fig 3. Imprint slide from mediastinal lymph node with metastatic adenocarcinoma. Cytoplasmic vacuoles characteristic of adenocarcinoma are evident in some cells. (Hematoxylin and eosin; xz50 before 50% reduction.) tions were independently reviewed by two pathologists. Diagnoses rendered on imprints or frozen sections were compared with those for permanent sections of the lymph nodes. Sensitivity, specificity, predictive value of a positive result, predictive value of a negative result, and efficiency of the test were calculated using standard nnethods based on the Bayes theorem [12]. In the patients with lung cancer, staging at mediastinoscopy either by IC or FS analysis was compared with permanent-section staging. The results of IC and FS analysis were also compared with the final pathologic stage determined after resection, if undertaken. Results The slides from IC preparation showed excellent moiphologic preservation and cytologic detail. Benign mediastinal lymph nodes characteristically contained large histiocytes with abundant eosinophilic cytoplasm with fine black anthracotic pigment in a background of benign lymphocytes (Fig 2). In imprints from lymph nodes with metastatic carcinoma, cells with cytologic features of malignancy were present singly and in small cohesive clusters (Figs :3, 4). Distingu@hing features of cell type were often evident on imprint slides, and a diagnosis of squamous cell carcinoma, adenocarcinoma, small cell carcinoma, and large cell carcinoma could often be determined. In lymphoproliferative disorders, malignant lymphoid cells usually looked distinctly different from benign lymphocytes. The presence of giant cells in a mixed benign cellular background characterized imprints from lymph nodes with sarcoidosis (Fig 5) and other granulomatous reactions. Group 1 One hundred ninety-two lymph nodes were sampled in the 38 patients having IC. Of these nodes, 121 were analyzed at the time of mediastinoscopy by IC (Table 2). There were 28 true-positive lymph nodes, no false-

4 Ann Thorac Surg 1994:57120MO CLARKE ET AL 1209 Fig 4. Imprint slide from mediastinal lymph node with metastatic poorly diferentiated squamous cell carcinoma. A cluster of cells with cytologic features of malignancy are shown. (Hematoxylin and eosin; ~300 before 50% reduction.) positive findings, 92 true-negative results, and 1 falsenegative node. The procedure showed a high sensitivity rate, 96.6%, and, as there were no false-positive findings, a specificity of 100%. The overall efficiency of the test was 99.2%. The one false-negative result occurred in a patient in whom other nodes obtained at the same mediastinoscopic examination were accurately classified as positive for metastatic tumor. A subgroup of 18 patients from group 1 also had 39 lymph nodes concurrently evaluated by FS analysis. In these specimens, there were 8 true-positive results, 30 true-negative results, no false-positive findings, and 1 false-negative finding. In the false-negative frozen section, the tumor represented extremely small foci of micrometastases, which were accurately identified by IC. This finding is similar to that observed by Gentry [13]: IC is sometimes more sensitive than FS analysis in detecting micrometastases. Of the 33 patients evaluated for lung cancer by IC, 14 were in stage I, 5 were in stage 11, and 14 were in stage I11 [14]. Thirty-two of the 33 patients were accurately staged at the time of mediastinoscopy using IC. At the time of mediastinoscopy, 20 patients were classified as stage I or I1 by the pathologic diagnosis from IC, and 19 underwent subsequent resection with pathologic confirmation of stage I or stage I1 disease. One of these 20 patients was found at exploratory thoracoscopy to have positive paraesophageal lymph nodes (stage 111), and therefore resection was not undertaken. The 5 patients evaluated for mediastinal lymphadenopathy were all accurately classified in terms of pathologic process at the time of mediastinoscopy using IC. In patients with lymphoproliferative disorders, a specific diagnosis was usually not given, but an indication of a lymphoproliferative disorder was conveyed. Group 2 A total of 168 lymph nodes from 36 patients were sampled during mediastinoscopy. Of these, 128 were evaluated by FS analysis (see Table 2). There were 16 true-positive lymph nodes, no false-positive nodes, 109 true-negative findings, and 1 false-negative finding. The one falsenegative lymph node was in a patient who was accurately classified based on another true-positive node diagnosed by FS at the time of mediastinoscopy. Diagnosis was deferred on two lymph nodes. The sensitivity rate for FS diagnosis was 94%, and, as there were no false-positive results, the specificity was 100%. The overall efficiency of FS analysis compared with permanent section was 99.2%. Of the 24 patients evaluated for lung cancer, 9 were in stage I, 3 were in stage 11, and 12 were in stage 111. Twenty-three of the 24 patients were accurately staged at mediastinoscopy with FS analysis. At the time of mediastinoscopy, 13 patients were classified as a stage I or I1 by FS pathologic diagnosis. One of them underwent a repeat mediastinoscopic examination at a later date, was found to have mediastinal tumor spread, and was classified as stage I11 with no further surgical intervention undertaken. The other 12 patients underwent pulmonary resection and intraoperative mediastinal nodal dissection, which confirmed the validity of the FS staging. The 12 patients evaluated for mediastinal lymphadenopathy were accurately classified in terms of pathologic process at the time of mediastinoscopy with FS analysis. Comment The use of intraoperative cytology in surgical pathology was reported by Dudgeon and Patrick [15] in Since then, many reports [ 16-19] have confirmed the usefulness and accuracy of the technique as well as its application in diagnosing pathology of the central nervous system, breast, lymph nodes, thyroid gland, lung, and gastrointestinal tract. A recent report by Mair and associates [20] highlights the superior quality of the cytologic preparation compared with FS analysis as well as the major time savings with no loss of accuracy in diagnosis. Gentry [13] studied its application in intraoperative staging of prostate cancer with excellent results and suuerioritv to the FS Fig 5. lmprint slide from a patient with sarcoidosis. Two small, cohesive granulomas are present. (Hematoxylin and eosin; ~ 300 before 50% reduction.)

5 1210 CLARKE ET AL Ann Thorac Surg 1994; Table 2. Results of Imprint Cvtolom and Frozen Section Analysis Compared With Permanent Section Total Total Total Test No. of Nodes Nodes Nodes True Fake True False Sensitivity Specificity Efficiency Group Patients Sampled by IC by FS Positive Positive Negative Negative (%) (%) PVPR PVNR (I%) 1 (IC) (FS) FS = frozen section analysis; IC = imprint cytology; PVNR = predic :tive value of negative result; PVPR = predictive value of positive result. technique in detecting micrometastases. Although a great deal has been reported on the use of IC in many other organs, there is little information regarding its validity in mediastinoscopic staging of lung cancer. The results of this study confirm the usefulness of IC in intraoperative evaluation of mediastinal lymphadenopathy and the staging of lung cancer. Our results with a test efficiency of 99.2% compare favorably with those reported in other studies [ comparing the diagnostic accuracy of IC with FS analysis, with accuracy rates of 90% and greater reported. Our results also compare favorably with reported accuracy rates for FS analysis of 97% to 99% [21, 221 and meet the currently recommended accuracy rate of 97% for intraoperative consultation in quality control in pathology [23]. Of the 33 lung cancer patients staged by IC at mediastinoscopy, 32, or 97%, were correctly staged. The technique also proved reliable in categorizing mediastinal lymph nodes with lymphoid malignancies and benign inflammatory conditions. The interpretation of imprint cytologic preparations does require experience in cytology by the surgical pathologist evaluating the specimen. Use of this technique may not be appropriate in a hospital without an active cytology service. The main technical limitation of the procedure is the need of rapid fixation to prevent drying artifact. The technique is simple, reliable, and accurate and overcomes a number of disadvantages of FS analysis. Imprint cytology is far less time-consuming and labor intensive for the surgical pathology laboratory than FS analysis and lessens the demand that intraoperative staging imposes on pathology personnel. Further, because the tissue remains intact, FS analysis can still be performed, in addition to IC, in difficult or equivocal cases. Finally, IC results in a more rapid turnaround time, which is important for the anesthetized patient in whom multiple nodal stations must undergo biopsy. References 1. Patterson GA, Ginsberg RJ, Poon PY, et al. A prospective evaluation of magnetic resonance imaging, computed tomography, and mediastinoscopy in the preoperative assessment of mediastinal node status in bronchogenic carcinoma. J Thorac Cardiovasc Surg 1987;94: Coulomb M, Escolano E, Rose-Pittet L, et al. Mediastinal lymph node extension in primary bronchial cancer: correlation of x-ray computed tomography, magnetic resonance imaging and mediastinoscopy: apropos of 50 cases. J Radiol 1987;68: Van Schil PE, Van Hee RH, Schoofs EL. The value of mediastinoscopy in preoperative staging of bronchogenic carcinoma. J Thorac Cardiovasc Surg 1989; Eddy RJ. Cost-effectiveness of CT scanning compared with mediastinoscopy in the preoperative staging of lung cancer. Can Assoc Radiol J 1989; Keller R, Hausmann M. Cost-benefit analysis of thoracic computerized tomography as exemplified in the N-stagjng of bronchial carcinoma. Schweiz Med Wochenschr 1988;118: Benfield JR. The lung cancer dilemma. Chest 1991;100: Luke WP, Pearson FG, Todd TR, Patterson GA, Cooper JD. Prospective evaluation of mediastinoscopy for assessment of carcinoma of the lung. Thorac Cardiovasc Surg 1986;91: Ginsberg RJ. Evaluation of the mediastinum by invasive techniques. Surg Clin North Am 1987; Vallieres E, Page A, Verdant A. Ambulatory mediasiinoscopy and anterior mediastinotomy. Ann Thorac Surg 1991; 52: Gephardt GN, Rice TW. Utility of frozen section evaluation of lymph nodes in the staging of bronchogenic carcinoma at mediastinoscopy and thoracotomy. J Thorac Cardiovasc Surg 1990;100: Christ ML, Fry WA. Intraoperative fine-needle aspiiation and rapid diagnosis of thoracic lesions. Appl Pathol 1986;4: Winkel P, Statlane 8. Interpreting laboratory results: reference values and decision making. In: Henry JB, ed. Clinical diagnosis and management by laboratory methods. Philadelphia: WB Saunders, 1991: Gentry JF. Pelvic lymph node metastases in prostatic carcinoma: the value of touch imprint cytology. Am J Surg Pathol 1986;10: Mountain CF. A new international staging system for lung cancer. Chest 1986;89:225S-33S. 15. Dudgeon LS, Patrick CV. A new method for the rapid microscopical diagnosis of tumors: with an account of 200 cases so examined. Br J Surg 1927;15: Owings RM. Rapid examination of surgical specimens: a valuable technique in the surgical pathology laboratory. Hum Pathol 1984;15: Kuntozoglou TE, Cramer HM. The advantages of intraoperative cytology: analysis of 215 smears and review of the literature. Acta Cytol 1991;35: Sakai Y, Lauslahti K. Comparison and analysis of the results of cvtodiaenosis and frozen sections during -- operation. Acta Cyth 196$13: Suen KC. Wood WS, Sved AA, Ouenville NF, Clement PB. Role of imprint cytology in inhaoperative diagnosis:,value and limitations. J Clin Pathol 1978;31:32% Mair S, Lash RH, Suskin D, Mendelsohn G. Intraoperative surgical specimen evaluation: frozen section analysis, cytologic examination or both? A comparative study of 206 cases. Am J Clin Pathol 1991;96:% Holaday WJ, Assur 0. Ten thousand consecutive frozen sections: a retrospective study focusing on accuracy and quality control. Am J Clin Pathol 1974;51: Onesun RH, Minke JA, Silverberg SG. Intraoperative pathologic consultation: an audit of 1,000 recent consecutive cases. Am J Surg Pathol 1989;13: Association of Directors of Anatomic and Surgical Pathology. Quality control in pathology. Am J Surg Pathol 1991;15: