Cell block from a fine-needle aspirate obtained from a left nasopharyngeal mass in a patient who has undergone lung transplantation showing a monomorphic population of lymphoid cells with large, irregular nuclei and prominent nucleoli. Immunohistochemical stains demonstrate a λ-restricted clonal population, consistent with lymphoma (hematoxylin-eosin, original magnification ×630).
Top, Direct smear of a fine-needle aspirate obtained from a mass in the anterior aspect of the neck in a patient who has undergone radiation therapy for squamous cell carcinoma of the larynx. Highly atypical epithelial cells adjacent to dense stroma were originally misinterpreted as recurrent squamous cell carcinoma (Diff-Quik, original magnification ×630). Bottom, Corresponding surgical resection specimen showing a benign thyroid cyst with a sclerotic wall lined by follicular cells with reactive atypia, reflecting radiation therapy changes (hematoxylin-eosin, original magnification ×400).
Martha J. Sack, Randal S. Weber, Gregory S. Weinstein, Ara A. Chalian, Harvey L. Nisenbaum, David M. Yousem. Image-Guided Fine-Needle Aspiration of the Head and NeckFive Years' Experience. Arch Otolaryngol Head Neck Surg. 1998;124(10):1155–1161. doi:10.1001/archotol.124.10.1155
To evaluate the diagnostic utility of image-guided fine-needle aspiration (FNA) in the head and neck.
All image-guided FNAs of the head and neck performed January 1992 through June 1997 were included. All cytohistopathologic data were reviewed and collated. A slide review was performed in all cases with cytohistologic discrepancies.
The Department of Radiology, University of Pennsylvania Medical Center, Philadelphia.
Patients with deep-seated or poorly localized masses in the head and neck, representing both primary or recurrent/metastatic lesions, were referred.
There were 111 computed tomography–guided FNAs performed in 109 patients. Sites sampled included parapharyngeal (n=20), parotid or submandibular (n=25), thyroid (34), and neck, paratracheal/paraesophageal, skull base, and paraspinal (n=32). Diagnostic samples were obtained in 93 cases (83.8%). The procedures were well tolerated, without long-term complications. Cytologic examination detected a total of 39 malignancies, 24 of which were confirmed histologically. Eleven of the remaining malignant FNA cases reflected recurrent tumor; there were 3 false-positive FNA cases (2.7%), 2 in the setting of previous surgery and/or radiation therapy. There were 2 false-negative aspirates from sites deep in the neck (1.8%) among 7 of the 35 patients with benign aspirates who underwent surgery. Twenty six patients underwent ultrasound-guided FNA (thyroid gland only), revealing 1 papillary carcinoma and 1 intrathyroidal parathyroid gland, both of which were confirmed histologically. The findings in the aspirates from the rest of the patients were benign (n=18), Hürthle cell neoplasm (n=1), and nondiagnostic (n=5).
(1) The cytologic findings were supported clinically and/or histologically in 86 (92%) of the 93 diagnostic computed tomography–guided FNA cases. (2) Unnecessary surgery was avoided in 37% of the patients with recurrent tumor or benign diagnoses by cytologic assessment. (3) Potential pitfalls include false-positive diagnoses after radiation therapy and procedural or sampling limitations for deep neck and paraspinal lesions.
FINE-NEEDLE aspiration (FNA) biopsy has become a well-established technique in the diagnosis, staging, and follow-up of patients with head and neck lesions. Endocrinologists and otorhinolaryngologists use FNA to provide rapid diagnostic information regarding palpable masses of the thyroid gland, salivary glands, and cervical lymph nodes. However, the clinical assessment of deep-seated or poorly localized masses in this region is highly challenging owing to the complex anatomy and wide range of lesions. A transmucosal, peroral FNA approach to lesions in the parapharyngeal space can yield useful results, with reported accuracy rates ranging from 77%1 to 88%.2 However, this approach is accompanied by risk to carotid and jugular vessels and the facial nerve. Also, there are procedural limitations regarding the stabilization of the lesion and the range of the angle of approach that can result in false-negative rates as high as 19%.3A transoral open biopsy is generally contraindicated for the evaluation of these lesions owing to unacceptable morbidity. To minimize these limitations, FNA has been paired with imaging-guided localization for the cytologic assessment of deep-seated or poorly localized lesions. The current study evaluates the diagnostic utility of computed tomography (CT)– and ultrasound-guided FNA of masses in the head and neck.
The radiology files at the University of Pennsylvania Medical Center, Philadelphia, were searched for all imaging-guided FNAs of the head and neck performed January 1992 through June 1997. Patients are referred for these services primarily by the head and neck surgeons, skull base surgeons, and endocrinologists at our institution. All corresponding cytology reports and follow-up surgical pathology reports were evaluated, and the slides from all cases with discrepant cytologic-histologic findings were reviewed. A total of 111 CT-guided FNAs were performed in 109 patients (59 women and 50 men; age range, 18-92 years; median age, 59 years). Previous palpation-guided FNAs had been performed in 15 (13.7%) of these patients, revealing normal salivary gland in 2 patients, a few atypical cells in 4 patients, tumor in 1 patient, and nondiagnostic results in 8 patients. Twenty-six patients (18 women and 8 men; age range, 24-81 years; median age, 49 years) underwent ultrasound-guided FNA. In 4 of these patients, a nondiagnostic aspirate had been obtained previously.
All aspirates were procured by radiologists, with specimen preparation and evaluation performed on site by the cytopathologists in all but 1 case. Lesions were localized on CT using anteroposterior and lateral scout tomograms to plan the best angle and puncture site for sampling the mass.4,5An 18-gauge injection needle was placed adjacent to, and along, the line of site of the lesion, which was verified by scan, and then a 22-gauge spinal needle was inserted coaxially through the 18-gauge needle to the edge of the lesion. After CT confirmation of needle placement, suction was applied to the 22-gauge needle using a 20-mL syringe, and the lesion was sampled with several vigorous 1- to 3-cm excursions. Suction was then released while the needle was still in the lesion, and the needle was withdrawn through the 18-gauge needle, leaving the latter in place for repeated passes into the lesion. For the thyroid lesions sampled under ultrasound guidance, high-frequency ultrasound probes were used to image the nodules and to guide needle placement. A 25-gauge needle attached to a 10-mL syringe was advanced into the mass; negative pressure was applied; and several small excursions were made before pressure was released and the needle was removed.
In all image-guided cases, the aspirated material was processed by the cytopathologist as air-dried direct smears for a Giemsa-type stain (Diff-Quik, EM Diagnostic Systems, Gibbstown, NJ) for immediate microscopic evaluation and as wet-fixed smears for Papanicolaou stain on return to the laboratory. The needle and syringe were rinsed with normal saline, which was later concentrated onto a membrane filter (Millipore Corp, Bedford, Mass) and, if adequate material remained, into a cell block preparation. The aspiration procedure was repeated until diagnostic material was obtained, patient tolerance was reached, or 5 inadequate specimens were accrued. An average of 2.6 passes were performed for CT-guided FNA, with 78% of the cases completed in 1 to 3 passes. The average number of passes for ultrasound-guided procedures was 2.5; 80% of the cases were completed in 1 to 3 passes. Histochemical stains (mucicarmine, periodic acid–Schiff, or trichrome) and/or immunohistochemical stains were performed on direct smear or cell block preparations in 10 cases. The panel of immunohistochemical stains included thyroglobulin, S100 protein, HMB45, cytokeratins (AE1/3 and Cam 5.2), B-cell markers (L26, κ , and λ), epithelial membrane antigen, HHF35, neuron-specific enolase, synaptophysin, factor VIII, and CD34. Additional special studies included electron microscopy and RNA in situ hybridization performed on 1 case each.
A definitive cytologic diagnosis was rendered in 93 (84%) of the 111 CT-guided FNA cases. A total of 39 malignant neoplasms (35%) were diagnosed cytologically; benign neoplasms were reported in 19 cases (17%); 35 aspirates (31.5%) were considered benign/nonneoplastic (including reactive/inflammatory lesions, cysts, and goiter). There were 18 nondiagnostic aspirates: 3 (3%) were considered inconclusive, and 15 (13.5%) were not representative or had insufficient tissue for diagnosis.The cytologic results were supported histologically and/or clinically in 86 (92%) of the 93 diagnostic aspirates.
Sites sampled by CT-guided FNA included parapharyngeal (20), parotid/submandibular (25), thyroid (34), and a miscellaneous group of paratracheal/esophageal, neck, skull base, and paraspinal lesions (32 cases). Subsequent surgical resection specimens were obtained in 55 of the 109 patients. Among the cytologically definitive cases, CT-guided FNA correctly classified 81% of lesions undergoing surgical resection. Surgery was avoided in 46 patients with cytologically confirmed recurrent tumor or benign disease (including granulomatous inflammation, reactive lymph node, Warthin tumor, and nodular goiter). Malignancy was found in 4 of the 7 patients with nondiagnostic aspirates who underwent surgical resection.
The use of ultrasound-guided FNA for head and neck lesions was limited to the thyroid gland during the 5.5-year study period. A definitive diagnosis was established in 22 (79%) of the 28 aspirates obtained under ultrasound guidance. Two of these cases, a papillary carcinoma and a parathyroid lesion, were confirmed histologically. Nineteen of the remaining diagnostic aspirates were benign, and there was 1 follicular neoplasm. The cytologic results with histologic correlation are presented for each anatomical site in Table 1, Table 2, Table 3, andTable 4.
Squamous cell carcinomas were encountered in 6 of the 9 malignancies diagnosed in the parapharyngeal space by CT-guided FNA. Four of these cases represented recurrent disease from primary tumors of larynx, tonsil, tongue, and soft palate. The second most common lesion sampled in the parapharyngeal space was pleomorphic adenoma involving the deep lobe of the parotid gland or the minor salivary glands (4 cases, including 1 recurrent case). Other malignancies diagnosed cytologically in this region included a recurrent poorly differentiated carcinoma of the larynx and a spindle cell neoplasm in an 18-year-old man that, after resection and ancillary studies (immunohistochemical, electron microscopic, and molecular studies), proved to be a rhabdomyosarcoma, embryonal type. The final malignancy in this group was a case of posttransplant lymphoproliferative disorder in a patient who had undergone lung transplantation. The aspirate in this case revealed a monomorphic population of large, predominantly single cells (Figure 1) that were λ restricted by immunohistochemical stains. The results of Epstein-Barr virus in situ hybridization were negative in the limited material available for study. Adequate material was obtained for all the parapharyngeal aspirates, and there were no false-negative or false-positive cytologic results.
This group included 4 patients who had undergone resection and radiation therapy of primary squamous cell carcinomas (tonsil, tongue, retromolar trigone, and maxilla), 4 patients with histories of primary salivary gland neoplasms, and 1 patient with a history of breast carcinoma. There were 2 false-positive cytologic diagnoses: both were cases of sialadenitis (1 in the setting of radiation therapy) in which atypical squamous metaplastic changes were interpreted as malignant. Overlapping cytologic features between pleomorphic adenomas and mucoepidermoid carcinomas led to misclassification of 1 salivary gland neoplasm preoperatively. Finally, 5 aspirates of parotid gland (20% of the aspirates obtained in this group) were insufficient for diagnosis. Surgical procedures were performed in 2 of these cases, one of which revealed adenocarcinoma arising in a pleomorphic adenoma, and the other a squamous cell carcinoma of the maxilla, with lymphoma in the accompanying neck nodes (the parotid gland was normal in the resection specimen).
Computed tomography–guided aspirates of paratracheal/esophageal, neck, paraspinal, skull base, and infratemporal masses make up this group. The majority of the malignant aspirates reflected primary (3 cases) and recurrent (3 cases) squamous cell carcinomas. A case of hemangiopericytoma was accurately diagnosed in the cytologic material using ancillary studies (immunocytochemistry and electron microscopy). There was 1 false-positive aspirate from a mass in the anterior aspect of the neck in a patient with a history of laryngeal carcinoma: a small number of atypical cells from a fibrotic thyroid cyst were misinterpreted as malignant (Figure 2). Two false-negative aspirates were also identified in this group. These 2 cases reflect inadequate sampling due to technical limitations encountered in FNA of the pericarotid sheath in one case and a previous operative site near the mandible in the other case. These technical/sampling limitations are also evident in the 7 nondiagnostic aspirates; 5 of the target lesions were resected, revealing 2 cases of carcinoma, 1 intraspinal ependymoma with accompanying paraspinal fibrosis, 1 schwannoma, and the rare entity of heterotopic brain tissue in the pterygopalatine fossa.
The majority of thyroid lesions sampled by both CT- and ultrasound-guided FNA were cytologically benign, reflecting goiters, thyroiditis, and parathyroid tissue. Nodule size, which was recorded for 22 of the 28 ultrasound-guided FNAs, ranged from 0.7 to 4.2 cm in greatest dimension, with a mean of 2.29 cm. All of the malignant aspirates obtained by both imaging techniques were confirmed histologically with the exception of 1 case of recurrent squamous cell carcinoma (unknown primary site). The cytologic material obtained by CT-guided FNA in 1 case of papillary carcinoma with extensive cystic degeneration was considered suggestive of papillary carcinoma, but the scant cellularity and cystic background features precluded a definitive diagnosis. Two additional samples obtained by CT-guided FNA were considered inconclusive for neoplasm; however, the patients involved were unavailable for follow-up. Of the 3 resected adenomas/adenomatous nodules, 2 were reported as neoplastic on the basis of the aspirate. An intrathyroidal parathyroid adenoma was identified preoperatively as parathyroid tissue by ultrasound-guided FNA. The nondiagnostic rate was higher in the ultrasound-guided thyroid aspirates (21.4%) than in those guided by CT imaging (8.8%).
The treatment of patients with deep-seated or poorly localized lesions of the head and neck requires a multimodal approach that combines clinical, radiographic, and pathologic data. Anatomical location, growth pattern, and anticipated histologic findings all play an important role in preoperative planning.6- 8 Modern imaging techniques are recognized for their ability to delineate deep tissue anatomy. Magnetic resonance imaging is considered superior for soft tissue contrast resolution, while CT is best suited for evaluating subtle bony changes.9 Yet these techniques can only rarely provide specific histopathologic diagnoses. Similarly, when used in the follow-up of patients with malignancies in the head and neck, these studies are of limited value in distinguishing among recurrent tumor, scar tissue, radiation edema, or infection.9 Palpation-guided FNA is an established diagnostic technique for the preoperative assessment of palpable lesions of the head and neck.10- 14More recently, experience with image-guided FNA of nonpalpable abdominal and thoracic lesions has been applied in the head and neck region.15Using a technique described in previous reports,4,5lesions arising in the pericarotid sheath, thyroid gland, skull base, and parapharyngeal, parotid/submandibular, paratracheal, paraesophageal, paraspinal, and infratemporal regions have been sampled under CT guidance at our institution.
The most common tumors diagnosed by aspiration in the parapharyngeal region were squamous cell carcinomas (several of which were recurrent) and pleomorphic adenomas. These 2 diagnoses account for 50% of parapharyngeal lesions sampled, with the remaining neoplasms arising from neural, mesenchymal, and hematopoeitic tissues. Benign processes specifically diagnosed by CT-guided FNA included granulomatous inflammation, granulation tissue, and branchial cleft cyst. While the numbers are small, the distribution of diagnoses established by FNA in this region are similar to those reported in the surgical literature.6 Also, the technique proved to be highly effective in terms of diagnostic yield and accuracy, documenting tumor recurrence in 6 cases, avoiding unnecessary surgery in 6 cases of benign disease and a lymphoproliferative lesion, and providing histologic data for surgical procedure planning in the remaining cases. These results are in contrast to those achieved by Shoss et al16 in an earlier study of the preoperative and intraoperative assessment of parapharyngeal masses. In their study, 10 of 42 patients underwent open biopsy via an external (ie, transcervical) approach for what proved to be unresectable malignant neoplasms.
Computed tomography–guided FNA of the parotid and submandibular regions documented recurrent benign (1 case) and malignant (3 cases) salivary gland tumors, as well recurrent squamous cell carcinoma (2 cases). An additional 11 primary neoplasms were diagnosed by FNA, 7 of which were confirmed histologically. Interpretive errors were made in 3 cases from this group. In 1 case, a low-grade mucoepidermoid carcinoma was misclassified cytologically as a pleomorphic adenoma owing to misinterpretation of the mucoid material in the background of the specimen as representing the myxoid stroma seen in benign mixed tumors. The second was a case of suppurative sialadenitis in which the inflammatory component was not appreciated and the atypical metaplastic squamoid cells in a background of proteinaceous and cellular debris led to a false-positive diagnosis of mucoepidermoid carcinoma. These cases typify the well-recognized diagnostic pitfalls characteristic of the cytologic evaluation of this low-grade salivary gland malignancy.17 The third diagnostic error was a misinterpretation of atypical squamous metaplasia in postradiation sialadenitis. Five aspirates in this group were considered insufficient for diagnosis; 2 of the patients involved underwent surgery, revealing malignant tumors in both cases. One of these cases was a recurrent squamous cell carcinoma involving the maxilla, underlying the parotid gland. Review of the aspirate slides revealed only normal salivary gland tissue, indicating that the bony lesion was not effectively sampled. The second case was an adenocarcinoma in situ arising within a pleomorphic adenoma of the parotid gland that was adjacent to a Warthin tumor, highlighting the potential complexity of lesions in this region. Similar problems with nondiagnostic specimens were encountered in aspirates from paraspinal, pericarotid, and anterior neck regions (Table 3). These results suggest that attempts to optimize the FNA sampling of lesions in these sites via image-guided needle placement and on-site cytologic assessment can not always circumvent the technical/procedural limitations of access to these deep lesions adjacent to vital anatomical structures. In a large series of 1022 aspirates of tumors and tumorlike conditions of the oral and maxillofacial region, Daskalopoulou et al10encountered 18 cases with false-negative cytologic results and stressed that negative FNA results should not be relied on when the clinical assessment indicates malignancy. The present study findings also serve to emphasize the critical importance of generous sampling and a conservative approach to diagnosing recurrent tumor after radiation therapy.
THE EXPERIENCE with image-guided FNA of the thyroid gland in this series was relatively limited, with the majority of aspirates indicative of benign processes. All but 1 of the primary thyroid malignancies, along with a metastatic squamous cell carcinoma, were definitively diagnosed by FNA. Sampling of the remaining case, a cystic papillary carcinoma, was insufficient for a definitive diagnosis (papillary carcinoma was suspected), despite CT guidance to ensure aspiration of solid areas. A number of recent reports have demonstrated the diagnostic accuracy of ultrasound-guided FNA in the management of thyroid nodules.18- 21In a study by Yokozawa et al,18 a group of 678 patients diagnosed as having benign thyroid nodules by palpation-guided FNA underwent ultrasound-guided FNA within 2 to 24 months of the conventional procedure. Malignancy was suspected in 107 of these patients based on the ultrasound-guided procedure and was confirmed histologically in 99 of the 107. The most common cause for the cancers missed by conventional FNA was the presence of nonpalpable lesions (those associated with benign nodules or autoimmune thyroid disease and solitary small cancers). Inadequate sampling (eg, cystic lesions) (29.3%) and poor technique (15.2%) accounted for the missed cancers among the palpable lesions sampled by conventional FNA. In contrast, there was no significant difference in sensitivity and specificity between palpation- and ultrasound-guided FNA of thyroid nodules reported by Takashima et al19; however, insufficiency rates were significantly lower with the image-guided technique (3.7% vs 19%). The recognized advantages of ultrasound-guided FNA of thyroid nodules in this study included the specific sampling of solid areas within cystic masses and the sampling of nonpalpable, relatively large nodules in glands with diffuse disease or following surgery or radiation therapy. However, the use of this technique as a screening procedure is controversial, as it may lead to the detection of small cancers that are of questionable biological significance.20- 22
Controversy also surrounds the role of ultrasound-guided FNA in the assessment of cervical lymph node status in patients with head and neck cancers. Currently, elective (prophylactic) treatment of the neck is performed when the risk of occult metastases is greater than or equal to 15% to 20%,23 resulting in overtreatment in the majority of patients. Comparative studies of modern imaging techniques paired with FNA suggest that this approach may be refined in certain settings. In a prospective evaluation of 132 patients with squamous cell carcinoma of the head and neck, van den Brekel et al23 found CT, ultrasonography, and magnetic resonance imaging to be superior to palpation for cervical lymph node staging. Ultrasound-guided FNA was significantly better than any other technique evaluated, with sensitivity and specificity rates of 90% and 100%, respectively. In this study, ultrasound-guided FNA detected 75% of the patients with histologically proved lymph node metastases and negative palpation findings. However, these and other investigators emphasize that the technique is highly operator dependent and best applied to patients with inconclusive CT and/or magnetic resonance imaging results or uncertain palpatory findings.23- 25
These studies, along with the present report, demonstrate the expanding role of image-guided FNA in the evaluation of the head and neck lesions. Improvements in the accuracy of preoperative diagnostic techniques have been realized, with significant impact on patient treatment. For parapharyngeal masses, image-guided FNA can define whether surgery is needed, along with the appropriate surgical approach. In the setting of an immunocompromised patient, a distinction between neoplastic (eg, lymphoma) and inflammatory/infectious processes can be made, reducing dependence on open biopsy, with the accompanying risks. However, limitations remain, particularly in the setting of previous surgery and radiation therapy. Some of these limitations will be reduced by increased experience, which, when combined with refined patient selection criteria, will lead to optimized diagnostic utility of these techniques.
Accepted for publication April 29, 1998.
Presented in part as a poster at the 87th Annual Meeting of the US and Canadian Academy of Pathology, Boston, Mass, March 2, 1998.
Reprints: Martha J. Sack, MD, Department of Pathology, Abington Memorial Hospital, 1200 Old York Rd, Abington, PA 19001.