Application of artificial neural network for classification of thyroid follicular tumors

OBJECTIVE: To analyze smears of 197 thyroid follicular tumors (adenoma and carcinoma). STUDY DESIGN: Several types of artificial neural networks (ANN) of various designs were used for diagnosis of thyroid follicular tumors. The typical complex of cytologic features, some nuclear morphometric parameters (area, perimeter, shape factor) and density features of chromatin texture (mean value and SD of gray levels) were defined for each tumor. RESULTS: The ANN was trained by means of cytologic features characteristic for a thyroid follicular adenoma and a follicular carcinoma. At subsequent testing, the correct cytologic diagnosis was established in 93% (25 of 27) of cases. The morphometry increased the accuracy of diagnosis for follicular tumors in up to 97% (75 of 78) of cases. ANN correctly distinguished an adenoma or a carcinoma in 87% (73 of 84) of cases when using color microscopic images of tumors. CONCLUSION: The usage of ANN has raised sensitivity of cytologic diagnosis of follicular tumors to 90%, compared with a usual cytologic method (sensitivity of 56%). The automatic classification of thyroid follicular tumors by means of ANN is prospective.     

Deletions of the long arm of chromosome 10 in progression of follicular thyroid tumors

Previous studies of follicular thyroid tumors have shown loss of heterozygosity (LOH) on the short arm of chromosome 3 in carcinomas, and on chromosome 10 in atypical adenomas and carcinomas, but not in common adenomas. We studied LOH on these chromosomal arms in 15 follicular thyroid carcinomas, 19 atypical follicular adenomas and 6 anaplastic (undifferentiated) carcinomas. Deletion mapping of chromosome 10 using 15 polymorphic markers showed that 15 (37.5%) of the tumors displayed LOH somewhere along the long arm. Thirteen of these tumors showed deletions involving the telomeric part of chromosome 10q, distal to D1OS 187. LOH on chromosome 3p was found in 8 (20%) cases. Seven of these also showed LOH on chromosome 10q. In eight cases LOH was seen on chromosome 10q but not 3p. In comparison, the retinoblastoma gene locus at chromosome 13q showed LOH in 22% of the tumors. Most of these also had deletions on chromosome 10q. The results indicate that a region at the telomeric part of 10q may be involved in progression of follicular thyroid tumors.     

Pituitary metastasis of follicular thyroid carcinoma

OBJECTIVE: The case of a 60-year-old male patient with follicular thyroid cancer who developed a pituitary mass proved to be a metastasis from thyroid cancer. METHODS: Assessment with whole-body scan, ultrasound, computed tomography and thyroglobulin measurements. RESULTS: Despite surgery and repeated doses of radioiodine, the patient developed diplopia and ptosis of the right eyelid, along with increasing thyroglobulin levels. A pituitary mass was discovered, with no signs of pituitary deficiency. The mass was removed and found to consist of neoplastic cells immunohistochemically positive to thyroglobulin. CONCLUSIONS: Distant metastases may develop in cases of follicular thyroid carcinoma, even after repeated doses of (131)I. Metastatic follicular thyroid carcinoma to the pituitary is a rare entity.     

Morphometric analysis of the degree of aggregation of follicular cells in the diagnosis of malignant thyroid tumors

OBJECTIVE: To develop a system of decision criteria based on a set of features of the degree of thyrocyte aggregation with assessment of their weighting coefficient in the diagnosis of malignant and benign thyroid pathology. STUDY DESIGN: Autopsy, operative and aspiration biopsy material were the object of the study. Light microscopic morphometry was performed with semiautomated computer analysis of images. RESULTS: Of all the sets of features of the degree of thyrocyte aggregation, 10 were chosen that provided the least overlap between the groups of malignant and benign pathology. Boundary values of these features to discriminate between malignant and benign pathology were assembled into a set of diagnostic decision criteria for thyroid carcinoma diagnosis. Based on results from trials, an equation for the calculation of a weighting coefficient for each feature was derived. A diagnostic index was determined by the sum of values of weighting coefficients of the features within the range of values listed in the set of diagnostic decision criteria. CONCLUSION: A set of diagnostic decision criteria was developed for thyroid carcinoma diagnosis that is based on morphometric features reflecting the regularities of changes in the structure of thyrocyte aggregates with malignant and benign pathology. Weighting coefficients of features were assessed, making it possible to perform a probabilistic carcinoma diagnosis.     

Allelotype of human thyroid tumors: loss of chromosome 11q13 sequences in follicular neoplasms.

Two classes of genes are the targets of mutations involved in human tumorigenesis: oncogenes, the activation of which leads to growth stimulation, and tumor suppressor genes, which become tumorigenic through loss of function, often through allelic deletion. To obtain evidence for a role for tumor suppressor genes in thyroid tumorigenesis, we examined DNA from 80 thyroid neoplasms for loss of heterozygosity in multiple chromosomal loci using 19 polymorphic genomic probes. None of the informative thyroid tumors studied had allelic loss detected with probes for chromosome 2q (D2S44), 3p (D3F15S2, D3S32), 3q (D3S46), 4p (D4S125), 6p (D6S40), 8q (D8S39), 9q (D9S7), 12p (D12S14), 13q (D13S52), 17p (D17S30), or 18q (D18S10). One of eight of the follicular adenomas had a 10q deletion detected with marker D10S15, and one of 26 had a 10q deletion detected with D10S25. One of two of the follicular carcinomas had an 11p deletion in the H-ras locus. The most significant findings were on chromosome 11q13, the site containing the putative gene predisposing to multiple endocrine neoplasia type I. Four of 27 follicular adenomas had loss of heterozygosity for probes in this region. Allelic deletions were detected with the following probes: D11S149, PYGM, D11S146, and INT2. None of 13 informative papillary carcinomas and none of two follicular carcinomas had loss of heterozygosity detectable with these 11q13 markers. Allelic loss is a relatively infrequent event in human thyroid tumors. Deletions of chromosome 11q13 are present in about 14% of follicular, but not papillary, neoplasms.(ABSTRACT TRUNCATED AT 250 WORDS)     

MIB-1 immunohistometry of follicular adenoma and follicular carcinoma of the thyroid gland

OBJECTIVE: To analyze cellular proliferative activity, MIB-1 immunopositivity of normal tissue (n = 20), follicular adenoma (n = 30) and follicular carcinoma (n = 32) of the thyroid gland was analyzed by means of immunohistometry. STUDY DESIGN: Immunohistochemical reactions were performed on 3-micron sections from routinely formalin fixed and paraffin embedded surgical specimens using an indirect peroxidase method. The rate of immunostained cells was determined using the CM-2 TV image analysis system (Hund, Wetzlar, Federal Republic of Germany). Forty viewing fields (1.94 mm2) were measured with 20:1 objective magnification. An average of 5,965 cells were assessed in each case. RESULTS: Mean MIB-1 immunopositivity was higher in follicular carcinoma (average, 2.30%) and follicular adenoma (0.58%) than in normal thyroid tissue (0.14%). The distribution of single values differed significantly between groups (P < .001). To test the suitability of MIB-1 immunohistometry for the differential diagnosis of follicular adenoma and follicular carcinoma, different four-field tables with varying thresholds were calculated. Using a threshold of 0.9%, follicular carcinoma could be detected with a sensitivity of 75% (24/32) and a specificity of 83% (25/30). If a specificity of 90% is required (27/30), the sensitivity of the test decreases to 69% (22/32), based on a threshold of 1.1%. CONCLUSION: As some overlap of single values has to be considered, MIB-1 immunohistometry, although presenting new insights into the proliferative potential of thyroid lesions, is of only limited value for the differential diagnosis of follicular lesions in routine surgical pathology.     

Intranuclear rodlets in woodchuck thyroid follicular cells

Summary Intranuclear rodlets are seen in thyroid follicular cells of woodchucks fixed during summer and fall. They are not observed in winter and spring animals. A functional significance of these structures is suggested.Supported in part by PHS 05429-13-5 (P.P. Krupp, Department of Anatomy) and PHS R01 10254 (E.A. Sims, Metabolic Unit, Department of Medicine)     

Allelotyping of human prostatic adenocarcinoma.

Allelotyping (using at least one probe detecting a restriction fragment length polymorphism on each chromosomal arm, with the exception of the short arms of the acrocentric chromosomes), showed loss of genetic information in 11 of 18 prostate adenocarcinoma specimens analyzed (61%). Frequent allelic deletions were detected on the long arm of chromosome 16 (6 of 10 informative cases, 60%), on the short arm of chromosome 8 (3 of 6 informative cases, 50%), and on the short and/or the long arms of chromosome 10 (6 of 11 informative cases (10p), 55% and 4 of 13 informative cases (10q), 30%, respectively). No losses of alleles were detected in any case unless at least one of the chromosomes 8, 10, or 16 also showed deletions. The long arm of chromosome 18 also showed a high frequency of allelic deletions (3 of 7 informative cases, 43%). Allelic deletions on the following chromosomes were detected at lower frequencies: chromosomes 2, 3, 7, 12, 13, 17, 22, and XY. Tumors with allelic deletions on more than one chromosome had a higher histological malignancy grade. Tumors from patients with advanced disease all showed allelic deletions.     

Calcium-activated cation channel in rat thyroid follicular cells

Using the patch-clamp single-channel current recording technique, a cation channel in the contraluminal membrane of rat thyroid follicular cells has been characterized. The channel has a unit conductance of about 35 pS and is equally permeable to sodium and potassium. The pattern of channel opening and closing is independent of the membrane potential. The channel is only operational when the ionized calcium concentration in the fluid which is in contact with the inside of the membrane is at least 1 microM. This conductance pathway can be classified as a calcium dependent non-selective cation channel and could explain stimulant-evoked depolarizations in the thyroid follicular cells.     

Morphometric image analysis of follicular lesions of the thyroid

OBJECTIVE: To determine the role of image morphometry in distinguishing various follicular lesions of the thyroid in cytologic smears. STUDY DESIGN: Archival fine needle aspiration smears of 10 cases each of follicular hyperplasia, follicular adenoma, follicular carcinoma and follicular variant of papillary carcinoma were used for the study. All cases were histopathologically proven. At least 100 random nuclei from each case were subjected to analysis with an image cytometer. Area, convex area, length, width, perimeter, convex perimeter and roundness of nuclei were measured using a 40 x objective (1 pixel = 0.446 micron). RESULTS: ANOVA showed that all the nuclear variables studied were significantly different (P < .05) in follicular hyperplasia as compared to follicular carcinoma and papillary carcinoma. All nuclear variables except roundness were also significantly different (P < .05) between follicular hyperplasia and follicular adenoma. However, between follicular adenoma, follicular carcinoma and papillary carcinoma there was considerable overlap of nuclear morphometric parameters. CONCLUSION: Image morphometry may help to distinguish nonneoplastic follicular lesions (hyperplasia) from neoplastic lesions (adenomas and carcinomas). However, to distinguish benign from malignant follicular lesions, image morphometry might not improve the accuracy of standard cytologic examination.     

The electrophysiological properties of the resting thyroid follicular cell

Glass microelectrodes have been useful in the study of the electrical properties of the resting thyroid follicular cell membrane. The resting transmembrane potential (RMP) has probably been underestimated in earlier work, possible as a result of leak artefacts, and it is clear that in most species the RMP is certainly greater than -60 mV. The ratio of membrane Na+ permeability to K+ permeability (PNa/PK) is of the order of 0.07 to 0.08, and Cl- is possibly (although not definitely) distributed in a passive fashion across the cell membrane, indicating that the transmembrane K+ gradient is the most important factor in the generation of the RMP. The existence of an electrogenic sodium pump in the follicular cell membrane has been demonstrated: the pump contributes about -2 mV to the RMP under control conditions. Follicular cells are completely electrically coupled, the basic coupled cellular unit probably being equivalent to the individual thyroid follicle, and the specific membrane resistance and specific membrane capacitance have been calculated to be 5 k omega. cm2 and 3.6 microF/cm2 respectively.     

Benign follicular thyroid lesions versus follicular variant of papillary carcinoma: differentiation by architectural pattern

Differentiation between benign and malignant follicular lesions is one of the difficult diagnostic areas in thyroid fine needle aspiration (FNA). Nuclear criteria are usually used to distinguish between them. In this study the microarchitectural pattern of common benign follicular lesions, namely nodular hyperplasia (NH) and follicular adenoma (FA) were analysed in comparison with those of follicular variant of papillary carcinoma (FVPC) in order to aid in their differentiation. The FNA smears of histologically proven cases of FVPC, NH and FA were reviewed and compared. The microarchitectural features of FVPC, NH and FA were described. Three cytological features--multi-layered rosettes, branching monolayered sheets and balls of thick pinkish colloid--were exclusively observed in FVPC. Hyperplastic papillae with intact follicles and colloid were frequently seen in NH, 83% and 100%, respectively. Albeit less frequently, they were also noted in FVPC, 25% and 75% of cases, respectively. These overlapping features make the distinction between FVPC and NH sometimes difficult; however, assessing the smears for the specific features of FVPC may help in their differentiation. None of the aforementioned microscopic findings with the exception of the seldom presence of colloid were documented in FA. The crowded clusters of follicular cells were seen both in FA and FVPC; however, they were complex and branching in the latter and round to oval in the former. Finally, smears with good recovery of material are indispensable for the identification of these helpful microarchitectural patterns.     

Planimetric studies on fine needle aspirates from follicular adenoma and follicular carcinoma of the thyroid

A planimetric study was performed on fine needle aspiration biopsy smears from 21 follicular thyroid adenomas, 13 follicular thyroid carcinomas and 7 nontoxic goiters. The nuclear and the cytoplasmic projected areas were measured in each smear on 50 cells with intact nuclei. The nuclear-cytoplasmic area was calculated. Significant differences in mean nuclear area were found between benign and malignant follicular neoplasms and between neoplastic cells and cells from nontoxic goiter. Planimetry of cells aspirated from follicular neoplasms permitted differentiation between carcinomas and adenomas with a high degree of statistical probability.