Human papillomavirus genotype distribution in head and neck cancer: Informing developing strategies for cancer prevention, diagnosis, treatment and surveillance.

Current clinical practice algorithms for HPV testing make no effort to discern the impact of genotypes for patients with head and neck squamous cell carcinoma (HNSC). Data was collected for all patients with HNSCs that had undergone HPV testing at an academic hospital as part of clinical care (2012-2019). Screening was performed using real-time PCR targeting L1 of low and high-risk HPV types, followed by genotyping of positive cases. Genotype status was correlated with age, site and histologic parameters. Of the 964 patients tested, 68% had HPV-positive cancers. Most arose from the oropharynx (OP) (89%) and sinonasal tract (5%). The most frequent genotype was 16 (84.4%) followed by 35 (5.6%), 33 (4.1%), 18 (2.7%), 45 (1.1%), 69 (0.8%) and others (1.3%). There was an association between genotype (16 vs non-16) and tumor origin (OP vs non-OP) (p < 0.0001). HPV18 was associated with transformation to an aggressive small cell phenotype, but HPV16 was not (22% vs 0%, p < 0.0001). Patients with HPV-non-16 OP carcinomas were older than patients with HPV16 OP carcinomas, but the difference was not significant. HPV genotypes are variable and unevenly distributed across anatomic sites of the head and neck. The association of HPV18 with small cell transformation suggests that variants can track with certain phenotypes in ways that may account for differences in clinical behavior. This study challenges the prevailing assumption of HPV equivalency across all high-risk genotypes in ways that may inform preventive, diagnostic, therapeutic and surveillance strategies.

The Prevalence, Anatomic Distribution and Significance of HPV Genotypes in Head and Neck Squamous Papillomas as Detected by Real-Time PCR and Sanger Sequencing.

Squamous papillomas (SPs) of the head and neck are generally regarded as a human papillomavirus (HPV)-driven process, but reported rates of HPV detection vary dramatically. Moreover, they are generally considered a benign condition, but the detection of high risk HPV types is commonly reported. This latter finding is particularly disturbing to clinicians and their patients given the alarming rise of HPV-associated head and neck cancer. The capriciousness of HPV detection reflects in large part differences in methodologies. The purpose of this study was to review an institutional experience using a state of the art detection method to determine the presence, type and anatomic distribution of HPV in head and neck SPs. The surgical pathology files of the Mount Sinai Hospital were reviewed for all SPs that had undergone HPV testing between 2012 and 2018. HPV screening was performed on tissue blocks with real-time PCR using primers designed to target the L1 region of low and high-risk HPV types. Genotyping was performed on HPV positive cases. HPV detection was repeated for cases that were originally reported to be positive for high risk HPV. 134 cases had undergone HPV analysis. Of the 131 with sufficient cellular material, 2 were excluded because the HPV testing yielded inconclusive results. The remaining 129 cases were the basis of this study. Thirty-eight cases (29%) were HPV positive and 91 (71%) were negative. The most common genotype was HPV 6 (n = 27, 71%), followed by HPV 11 (n = 10, 26%). One case (1%) was HPV positive but the genotype could not be determined. Of the HPV negative cases, 3 were originally reported as HPV 16 positive but found to be HPV negative on re-review and repeat testing. SPs arising in the larynx were more likely to harbor HPV than those arising in the oral cavity and oropharynx (64% vs. 10%, p < 0.00001). Similarly, recurrent respiratory papillomatosis (RRP) were much more likely to be HPV positive than solitary SPs (71% vs. 10%, p < 0.00001). Almost a third of head and neck SPs harbor HPV, but incidence is highly dependent on anatomic site. Those arising in the larynx are more prone to be HPV-driven than those arising in the oral cavity and oropharynx, particularly when occurring in the setting of RRP. High risk HPV could not be confirmed in any of the cases. Routine HPV testing as a strategy to unmask potentially malignant lesions harboring high risk HPV is not likely to be useful.

Real-time PCR HPV genotyping in fine needle aspirations of metastatic head and neck squamous cell carcinoma: Exposing the limitations of conventional p16 immunostaining.

BACKGROUND: Given the propensity for HPV-positive head and neck squamous cell carcinoma (HPV-HNSCC) to metastasize to cervical lymph nodes, fine needle aspiration (FNA) plays an important diagnostic role in their initial detection. Indeed, there is now an unwavering commitment to HPV testing of FNAs even in the absence of clear methodologic guidelines and threshold criteria. A particular difficulty pertains to the interpretation of p16 staining. DESIGN: Data was collected for 210 patients with suspected regionally metastatic HNSCC that had undergone FNA as part of standard clinical care. Initial HPV screening was performed on cell blocks with real-time PCR using primers targeting L1 of high-risk HPV types. Additional genotyping was performed on HPV-positive cases. The results were compared to p16 staining and subsequent excisions when available. RESULTS: Of the 207 samples with sufficient DNA, 175 (85%) were HPV positive. HPV-16 was the most commonly detected genotype (90%). Of the HPV-positive cases, the primary site was the oropharynx (n = 154, 88.0%), supraglottic larynx (n = 2, 1.1%), nasal cavity (n = 1, 0.6%), hypopharynx (n = 1, 0.6%) or unknown (n = 17, 9.7%). On comparison with 31 paired surgical excisions, HPV status was concordant in all cases (100% correlation). Of 142 HPV-positive cases with matching p16 stains, p16 staining was reported as positive (n = 85, 60%), focal (n = 27, 19%), negative (n = 24, 17%) or non-contributory (n = 6, 4%); and only 33% reached the standard threshold limit (i.e. 70%) for HPV positivity. CONCLUSION: For patients with metastatic HNSCC, real-time PCR of FNAs reliably reflects HPV status, and is superior to conventional p16 immunostaining.

Human papillomavirus (HPV) status of non-tobacco related squamous cell carcinomas of the lateral tongue.

OBJECTIVES: The human papillomavirus (HPV) is an important cause of some head and neck squamous cell carcinomas (HNSCCs), but its role in cancer of the lateral tongue is debatable. Suspicion of HPV causation is heightened when these lateral tongue carcinomas arise in patients that are young and/or have never smoked. The purpose of this study was to determine the incidence of transcriptionally active high risk HPV in these tumors, with a particular emphasis on non-smoking patients who are often presumed to have HPV-positive tumors. METHODS: We evaluated 78 HNSCCs of the lateral tongue for the presence of HPV using p16 immunohistochemistry and an RNA in situ hybridization assay targeting HPV E6/E7 mRNA. The study population was enriched for patients without traditional risk factors such as smoking and drinking. RESULTS: P16 overexpression was detected in 9 (11.5%) of 78 cases, but HPV E6/E7 mRNA transcripts were detected in only 1 (1.3%) case (positive predictive value of p16 staining for the presence of transcriptionally active HPV=0.12). HPV mRNA transcripts were not detected in any patient under 40 (n=11), or in patients who had never smoked (n=44), had quit smoking (n=15), and/or were only light consumers of alcohol (n=57). CONCLUSIONS: HPV is not detected in the vast majority of lateral tongue carcinomas. In light of the observation that HPV plays little if any role in the development of these cancers, routine HPV testing is unwarranted , even for patients without traditional risk factors. P16 staining is not a reliable marker for the presence of transcriptionally active HPV at this particular anatomic site.

Metastatic squamous cell carcinoma to the brain: an unrecognized pattern of distant spread in patients with HPV-related head and neck cancer.

Head and neck squamous cell carcinoma (HNSCC) is notorious for local recurrence and metastatic spread to regional lymph nodes. Distant spread is uncommon, and brain involvement is rare. Over the past decade there has been a rising incidence of HPV-related HNSCC, but it is not known if this escalation has had any impact on trends relating to brain involvement. Cases of metastatic squamous cell carcinoma (SCC) to the brain were identified from a computerized search of the surgical pathology files of The Johns Hopkins Hospital between 1985 and 2012. The medical records were reviewed to document primary site of tumor origin, treatment, and patient outcome. P16 immunohistochemistry and HPV in situ hybridization were performed on those metastases arising from the head and neck. Of the 38 metastatic SCCs, 7 (18 %) originated in the head and neck. HPV-16 was detected in 4 (57 %) of the metastatic HNSCCs. All 4 HPV-positive metastases were from oropharyngeal primaries. The time from treatment of the primary to development of the brain metastasis ranged from 19 to 57 months (mean, 45). Following aggressive treatment (surgery and radiation), two patients died of disease progression (7 and 34 months), and two are alive with recurrent brain metastases (4 and 10 months). Although HPV positivity is regarded as a favorable prognostic indicator, it does not safeguard from spread to the brain. In our experience, just over half of the HNSCCs that metastasized to the brain were HPV-related. The potential for developing a brain metastasis long after curative therapy argues for extended patient follow-up. The development of a brain metastasis is an ominous finding signaling rapid clinical deterioration.

Detection of Merkel cell virus and correlation with histologic presence of Merkel cell carcinoma in sentinel lymph nodes.

BACKGROUND: Adjuvant treatment can dramatically improve the survival of patients with metastatic Merkel cell carcinoma (MCC), making early, accurate detection of nodal disease critical. The purpose of this study was to correlate Merkel cell virus (MCV) detection with histopathologic disease in sentinel lymph nodes (SLNs) of MCC. METHODS: Merkel cell carcinoma cases with SLN (n=25) were compared with negative controls (n=27). Viral load was obtained by quantitative polymerase chain reaction (PCR) for regions VP1 and LT3 of MCV. Histopathologic disease and viral load were correlated. RESULTS: Merkel cell virus was detected in 16 out of 17 (94%) of primary MCC (mean viral load (MVL)=1.44 copies per genome). Viral load in the negative controls was <0.01 copies per genome. Merkel cell carcinoma was present in 5 out of 25 (20%) SLN by histopathology, and MCV was detected in 11 out of 25 (44%) MCC SLN (MVL=1.68 copies per genome). In all, 15 out of 25 (60%) SLN showed correlation between histologic and MCV results. In all, 2 out of 25 (8%) samples were histopathologically positive and PCR negative. Of note, 8 out of 25 (32%) samples had detectable MCV without microscopic disease. CONCLUSION: Patients with positive SLN for MCV even if negative by histopathology were identified. The application of molecular techniques to detect subhistologic disease in SLN of MCC patients may identify a subset of patients who would benefit from adjuvant nodal treatment.

Human papillomavirus and head and neck squamous cell carcinoma: recent evidence and clinical implications.

Over the past 20 years, high-risk human papilloma-virus (HPV) infection has been established as a risk factor for developing head and neck squamous cell carcinoma, independent of tobacco and alcohol use. In particular, HPV is strongly associated with the development of oropharyngeal cancer and a small minority of oral cavity cancers. In this review, we summarize what is currently known about the biology of HPV, the mechanisms by which it effects malignant transformation, and the potential impact of HPV status on the clinical management of persons with head and neck cancer.

The N-methyl-D-aspartate receptor type 2A is frequently methylated in human colorectal carcinoma and suppresses cell growth.

N-methyl-D-aspartate receptors (NMDARs) are the predominant excitatory neurotransmitter receptors in the mammalian brain. We found that among the three NMDARs examined (NMDAR1, NMDAR2A, NMDAR2B), only NMDAR2A was silenced in colorectal carcinoma (CRC) cell lines at basal line and reactivated by the demethylating agent, 5-aza-2'-deoxycytidine. NMDAR2A was expressed in normal colon epithelium, while expression was hardly detectable in colon cancer tissues. Promoter methylation of NMDAR2A was confirmed by bisulfite sequencing and combined bisulfite restriction analysis in the CRC cell lines and primary tumors. Quantitative methylation-specific PCR demonstrated NMDAR2A promoter hypermethylation in 82 of 100 primary human CRC, 15 of 100 normal corresponding epithelial tissues and 1 of 11 (9%) normal colon mucosa samples obtained from patients without cancer. Moreover, forced expression of full-length NMDAR2A in CRC cell lines induced apoptosis and almost abolished the ability of the cells to form colonies in culture, while NMDAR2A knockdown increased cell growth. Thus, NMDAR2A is commonly hypermethylated in primary human CRC and possesses tumor-suppressive activity.

Imaging findings of a hibernoma of the neck.

Hibernoma is an uncommon benign fatty tumor that arises from the vestiges of fetal brown fat. We present a case report of a hibernoma of the neck in an asymptomatic 19-year-old girl and describe the important imaging findings. Computed tomography (CT) shows a well defined hypodense mass with septations. Magnetic resonance imaging (MRI) shows intermediate T1 and bright T2 signal of the mass and also demonstrates the characteristic marked contrast enhancement.

Giant cell reparative granuloma of the sphenoid bone.

We present 2 patients with giant cell reparative granuloma (GCRG) of the sphenoid bone. The first patient is an 8-year-old boy with involvement of the greater wing, and the second is a 53- year-old man with a lateral pterygoid plate mass. Both patients presented with rapid expansion of lytic bone lesions, which had solid and cystic components and lacked matrix calcification. Biopsies were indeterminate for definitive diagnoses. The radiologic appearance, location, and incidence of the lesions, and the patient's age and medical history are helpful aids in narrowing the differential diagnosis of sphenoid bone lesions. However, the imaging and, occasionally, even the histologic findings may not suggest the specific diagnosis of GCRG, which must be added into the differential diagnosis of rapidly enlarging cystic bone lesions of the sphenoid bone.

EU-USA pathology panel for uniform diagnosis in randomised controlled trials for HRCT screening in lung cancer.

Randomised controlled trials for lung cancer screening using high-resolution computed tomography are now underway. In order to allow effective future comparison of the different trials, as well as strengthening conclusions based upon the analysis of larger data sets, uniformity and consistency of pathology diagnosis are essential. The aim of the present study was to determine the effectiveness of the learning process in this difficult area of diagnostic pathology. Eight pathologists received two CD-ROMs, each with digital images of 30 cases. After diagnosing the first series, selected background reading was provided. Kappa (kappa) scores were calculated for each pathologist and category, and were compared to the consensus score. The readings of the first series showed a moderate agreement kappa score: mean+/-sd for category numbers 8 (all eight categories) and 2 were 0.53+/-0.05 and 0.65+/-0.04, respectively. The kappa 2 score distinguished between categories denoting benign and malignant lesions. The second series resulted in a good agreement kappa score: 0.65+/-0.06 for category number 8 and 0.81+/-0.02 for category number 2. In conclusion, this study demonstrates that screen-detected cases pose particular problems for pathologists and that a trained pathology panel serving randomised controlled trials is likely to lead to more consistent and accurate tissue diagnosis.

Quantitative assessment of promoter methylation profiles in thyroid neoplasms.

CONTEXT: Cancer-specific molecular markers are needed to supplement the cytopathological assessment of thyroid tumors, because a majority of patients with cytologically indeterminate nodules currently undergo thyroidectomy without a definitive diagnosis. OBJECTIVE: The aim of this study was the quantitative assessment of promoter hypermethylation and its relation to the BRAF mutation in thyroid tumors. DESIGN: Quantitative hypermethylation of Rassf1A, TSHR, RAR-beta2, DAPK, S100, p16, CDH1, CALCA, TIMP3, TGF-beta, and GSTpi was tested on a cohort of 82 benign and malignant thyroid tumors and five thyroid cancer cell lines. SETTING: The study was conducted at a tertiary research hospital. PATIENTS: Patients underwent surgical resection for a thyroid tumor from 2000 to 2003 at our institution. INTERVENTIONS: There were no interventions. MAIN OUTCOME MEASURE: Final surgical pathology diagnosis was the main outcome measure. RESULTS: Thyroid tumors showed hypermethylation for the following markers: Rassf1A, TSHR, RAR-beta2, DAPK, CDH1, TIMP3, and TGF-beta. A trend toward multiple hypermethylation was evident in cancer tissues, with hypermethylation of two or more markers detectable in 25% of hyperplasias, 38% of adenomas, 48% of thyroid cancers, and 100% of cell lines. A rank correlation analysis of marker hypermethylation suggests that a subset of these markers is epigenetically modified in concert, which may reflect an organ-specific regulation process. Furthermore, a positive correlation was found between the BRAF mutation and RAR-beta2, and a negative correlation was found between the BRAF mutation and Rassf1A. CONCLUSIONS: Methylation-induced gene silencing appears to affect multiple genes in thyroid tissue and increases with cancer progression. Additional markers with better discriminatory power between benign and malignant samples are needed for the diagnostic assessment of cytologically indeterminate thyroid nodules.

Metastatic renal cell carcinoma to the pituitary presenting with hyperprolactinemia.

Metastases to the pituitary gland from systemic cancers is a rare phenomenon and usually occurs in patients with disseminated disease. The neurohypophysis is the most commonly involved site, and diabetes insipidus is the most common presentation in these patients. Breast and lung cancer are the most common cancers metastasizing to the pituitary. Involvement of the pituitary by renal cell carcinoma (RCC) is exceedingly rare. Mild-to-moderate degree of hyperprolactinemia is a rare presentation of pituitary metastases. We report the case of a woman with metastatic RCC to the pituitary presenting an unusually high degree of hyperprolactinemia.

Immunohistochemical staining for thyroid transcription factor-1: a helpful aid in discerning primary site of tumor origin in patients with brain metastases.

Metastatic carcinoma of unknown primary origin is a perplexing but common problem, accounting for up to 10% to 15% of all solid tumors at presentation. Many of these metastases presumably arise from primary lung carcinomas, but the morphologic features and immunohistochemical profile of lung cancer is often too nonspecific to permit unequivocal confirmation. Thyroid transcription factor-1 (TTF-1) is expressed in lung adenocarcinomas and thyroid carcinomas but not in adenocarcinomas arising from other sites. For patients with adenocarcinomas in the lung, TTF-1 staining is now routinely used to distinguish a primary lung cancer from a lung metastasis. Along these same lines, TTF-1 staining might prove useful in localizing the tumor origin of adenocarcinomas encountered outside of the lung. The archival surgical pathology files of The Johns Hopkins Hospital were searched for cases of brain metastases biopsied between 1990 and 2000. Tissue blocks were obtained and immunoperoxidase staining was performed using the TTF-1 antibody. The medical records were reviewed independent of the staining results to determine site of tumor origin. Seventy-five patients underwent biopsies of carcinomas metastatic to the brain. At the time of brain biopsy, the primary site of tumor origin was known in 45 cases and unknown in 30 cases. Ultimately, the primary site was established on clinical and radiographic grounds in 71 cases (95%). These included 40 (56%) metastases from a primary lung carcinoma and 31 (44%) metastases from some nonpulmonary carcinoma. TTF-1 staining was present in 31 of the 40 (78%) metastatic lung carcinomas, but in only 1 of the 31 (3%) metastatic nonpulmonary carcinomas (a small-cell carcinoma of the sinonasal tract). When the metastatic lung carcinomas were subtyped, TTF-1 staining was noted in 11 of 11 (100%) adenocarcinomas, in 6 of 7 (86%) small-cell carcinomas, in 15 of 19 (79%) large-cell carcinomas, and in none of 3 (0%) squamous cell carcinomas. TTF-1 staining is very reliable in discerning whether a brain metastasis has arisen from a pulmonary or nonpulmonary site, particularly when dealing with adenocarcinomas and large-cell carcinomas. TTF-1 immunohistochemistry could focus the search for the primary tumor for patients presenting with brain metastasis as the initial manifestation.

Loss of chromosome arms 3p and 9p and inactivation of P16 (INK4a) in normal epithelium of patients with primary lung cancer.

The accumulation of genetic alterations in the respiratory epithelium may give rise to cancer and often is accompanied by a series of histologic alterations over a period of several years. Recent studies have identified some molecular alterations in histologically normal-appearing epithelium among patients with lung cancer. To extend these observations, we investigated clonal genetic alterations by using fluorescence in situ hybridization (FISH) analysis and immunohistochemistry in 69 biopsy samples of histologically normal-appearing bronchial epithelium from 22 patients with or without lung cancer. Thirty-seven biopsy specimens from 13 patients were examined for loss of 3p14, and 48 biopsy specimens from 18 patients were examined for loss at 9p21 by FISH. P16(INK4a) expression was analyzed in 54 biopsy samples from 19 patients. In at least one biopsy specimen from five of the 13 patients with primary lung cancer, FISH or immunohistochemistry detected loss of the 3p14 or 9p21 region. In contrast, no alterations were detected for the same regions in the nine patients without primary lung cancer. Our results support the concept that the normal epithelial surface of large bronchi of patients with lung cancer has molecular changes suggestive of the outgrowth of numerous clonal foci.

Langerhans cell histiocytosis of the thyroid: a report of two cases and review of the literature.

Langerhans cell histiocytosis (LCH) is a monoclonal disease of histiocytes that can involve many or very few organ systems. It is a relatively benign disorder with a 3% mortality in adults. LCH rarely involves the thyroid gland. We report two cases, both presenting in males with a goiter. Both patients were treated with subtotal thyroidectomy. The first patient also received radiotherapy to his thyroid bed and scalp. We summarize the prior reported cases of LCH involving the thyroid and review the current treatment modalities used for LCH.

Allelic losses in OraTest-directed biopsies of patients with prior upper aerodigestive tract malignancy.

Genetic alterations at critical chromosome loci have been shown to be predictors of the progression of oral premalignancy-to-invasive cancer. We obtained a unique group of oral biopsies, initially collected during a prospective study designed to test the ability of OraTest (toluidine blue), to identify recurrent oral neoplastic lesions in patients with definite therapy for head and neck or upper aerodigestive tract (UADT) cancer. A total of 46 cases, including 13 squamous cell carcinoma (SCC), 11 carcinoma-in situ or dysplasia, and 22 morphologically normal oral biopsies, were analyzed for loss of heterozygosity (LOH) at 9p21, 3p21, and 17p13(TP53) by microsatellite analysis. LOH at one or more tested markers in at least one biopsy was detected in 76% (35 of 46) cases. All of the SCC and carcinoma-in situ cases showed LOH, and, strikingly, more than one-half (69%, 13 of 22) of morphologically normal epithelia also harbored LOH in at least one tested marker. The most frequent LOH was found on chromosome 9p21 (69%, 31 of 45). LOH was observed at 3p21, 17p13(TP53), or in multiple chromosomal arms significantly more often in SCC than in normal epithelia. In the majority of cases, two oral biopsies, one from an OraTest-staining positive area and another from a negative area adjacent to the stain, were collected. Among 25 LOH positive cases with two biopsies, identical allelic losses were confirmed between stained and nonstained biopsies in 16 cases. In the remaining nine cases with discordant LOH patterns between two biopsies, eight cases showed LOH at more genetic loci in OraTest-stained areas. Our data confirm that clonal genetic alterations, especially 9p21 deletion, are often present in the oral epithelia of patients with previous UADT malignancy and, combined with previous studies, suggest that genetic analysis will help stratify patients at risk of developing a secondary oral cancer. In addition to detecting cancer, our study suggests that OraTest can detect clinically occult lesions in the progression pathway to oral cancer.

Bizarre epithelial atypia of the sinonasal tract after chemotherapy.

Certain chemotherapeutic agents can induce bizarre epithelial atypia. The lower respiratory tract is a frequently targeted site, but similar changes have not been described adequately in the sinonasal tract. Unfamiliarity with these changes could potentially cause confusion with an infectious or neoplastic process. All biopsies of the sinonasal tract at The Johns Hopkins Hospital were reviewed prospectively over a 54-month period. Eleven cases with bizarre atypia of the respiratory epithelium formed the basis of this study. The medical records of these patients were reviewed. The specimens were from 11 patients who had previously undergone chemotherapy and bone marrow transplantation for acute myelocytic leukemia (n = 5), multiple myeloma (n = 3), acute lymphocytic leukemia (n = 2), and chronic myelocytic leukemia (n = 1). Although the chemotherapy regimens were highly variable, all included one or more of the alkylating agents (cyclophosphamide, n = 11; busulfan, n = 5; melphalan, n = 1). In all 11 patients, biopsies were acquired to rule out invasive fungal sinusitis. The atypical epithelial changes included striking nuclear enlargement, hyperchromasia, and pleomorphism. Sometimes these changes were full thickness and were associated with squamous metaplasia. Two of eight cases evaluated by frozen section were misinterpreted initially as high-grade epithelial dysplasia. Certain chemotherapeutic agents can induce striking epithelial atypia in the sinonasal tract. These changes should not be interpreted as neoplastic in nature, a potential pitfall in the frozen section evaluation of a destructive nasal process in oncology patients.

Randomized prospective evaluation of frozen-section analysis for follicular neoplasms of the thyroid.

OBJECTIVE: To evaluate the clinical utility of frozen section in patients with follicular neoplasms of the thyroid in a randomized prospective trial. SUMMARY BACKGROUND DATA: The finding of a follicular neoplasm on fine-needle aspiration prompts many surgeons to perform intraoperative frozen section during thyroid lobectomy. However, the focal distribution of key diagnostic features of malignancy contributes to a high rate of noninformative frozen sections. METHODS: The series comprised 68 consecutive patients with a solitary thyroid nodule in whom fine-needle aspiration showed a follicular neoplasm. Patients were excluded for bilateral or nodal disease, extrathyroidal extension, or a definitive fine-needle aspiration diagnosis. Final pathologic findings were compared with frozen sections, and cost analyses were performed. RESULTS: Sixty-one patients met the inclusion criteria. Twenty-nine were randomized to the frozen-section group and 32 to the non-frozen-section group. In the non-frozen-section group, one patient was excluded when gross examination of the specimen was suggestive of malignancy and a directed frozen section was diagnostic of follicular carcinoma. Frozen-section analysis rendered a definitive diagnosis of malignancy in 1 of 29 (3.4%) patients, who then underwent a one-stage total thyroidectomy. In the remaining 28 patients, frozen section showed a "follicular or Hürthle cell neoplasm." Permanent histology demonstrated well-differentiated thyroid cancer in 6 of these 28 patients (21%). Of the 31 patients in the non-frozen-section group, 3 (10%) showed well-differentiated thyroid carcinoma on permanent histology. Complications were limited to one transient unilateral vocal cord dysfunction. All but one patient had a 1-day hospital stay. There were no significant differences between the groups in surgical time or total hospital charges; however, the charge per informative frozen section was approximately $12,470. CONCLUSIONS: For the vast majority of patients (96.4%) with follicular neoplasms of the thyroid, frozen section is neither informative nor cost-effective.