Prognostic value of morphometry in papillary thyroid carcinoma

Prognostic Value of Morphometry in Papillary Thyroid Carcinoma ROBERT A AMBROS, MD, RONALD C. TROST, PHD, ALKIS Y. CAMPBELL, MD, AND W, CLARK IAMBERT, MD The value of morphometric analysis in addition to standard prognostic indicators was studied in 28 cases of papillary thyroid carcinoma. Standard features included age, sex, lymph node status, tumor sire, and encapsulation. The mean follow-up was 47 months (maximum, 140 months). Recurrences were documented in six patients at a mean time of 34 months; five patients recurred with distal metastases and one patient recurred with local disease. Univariate analysis most closely associated tumor recurrence with nuclear anisotropism (the standard deviation of the estimated nuclear area [ENASD]) and tumor size. With forward stepwise incremental analysis, the value of tumor size was lost and only the ENASD and the cellularity mean index (CMI), defined as the percentage of tumor volume composed of tumor cells, significantly correlated with recurrence. Fifty-five percent of patients with an ENASD > 17 pm* and a CM1 > 40% developed recurrence as compared with 5% of patients with lesser values (P = .OOOl). Morphometric analysis may significantly contribute to the role of histopathology in the evaluation of papillary thyroid carcinoma and may also provide information regarding prognosis not obtained by standard methods. HUM PATHOL 20:215-218. 0 1989 by W.B. Saunders Company.

Morphometric analysis has been demonstrated to be a useful technique in the objective quantification of histologic features and in the identification of prognostic indicators not obtained by standard methods.10-12 In a previous study of morphometric analysis of thyroid carcinoma, mean nuclear area was found to reliably differentiate between benign and malignant tumors. ls Similarly, a study of thyroid cell aspirates by Wright et all4 revealed statistically significant differences in mean nuclear area and circumference between benign and malignant thyroid lesions. The value of morphometric analysis in the prediction of recurrence for papillary carcinoma has not been reported. We studied the prognostic value of morphometric analysis in 28 cases of papillary carcinoma of the thyroid by the quantification of cytologic and architectural features commonly used in subjective grading systems.

Papillary carcinoma is the most common malignancy of the thyroid and is associated with the best the majority of patients with prognosis. *,* Although this disease can be adequately managed with lobectomy or subtotal thyroidectomy,3-6 there exists a subgroup of patients who develop recurrent disease following therapy. Reported indicators of prognosis have included age, sex, tumor size, multicentricity, and local tumor extension.‘,* The importance of histologic features, however, has not been clearly defined. Carcangiu et al’ reported no relationship between prognosis and a number of histologic features which included papillary formation, occurrence of ground glass nuclei, vascular invasion, and the occurrence of solid areas. In a review of 859 cases of papillary carcinoma, McConahey et ala found a significant correlation between survival and tumor grade. However, the method of grading was not described. Similarly, Tennvall et a1,g who found marked cellular atypia to be associated with poor prognosis in differentiated thyroid carcinoma, did not describe criteria for inclusion into this category.

MATERIALS AND METHODS

PatientSelection All cases of papillary carcinoma accessioned in the tumor registry of Beth Israel Medical Center (Newark, NJ) over the past 12 years were reviewed. The diagnosis of papillary carcinoma was based on the World Health Organization classification of thyroid tumors.15 Cases of the follicular variant of papillary carcinoma were included.16 Patients who originally presented with distal metastases were excluded from the study, as were those who were lost to follow-up. Twenty-eight of the 33 cases met these criteria. Surgical pathology reports were reviewed and each case was reexamined histologically.

Morphometric Techniques Tissue sections used for morphometric analysis were stained with hematoxylin-eosin and cut into 5 pm sections. One section from each case, subjectively considered to contain the most tumor, was studied. The initial field of study was selected at an edge of the tissue section. Successive fields were selected by alternate vertical and horizontal turns of the microscope stage until the prior field was just out of view. Evaluation was limited to fields entirely occupied by tumor. With the exception of the mitotic index, all morphometric parameters were analyzed by projection of the microscopic image onto a video monitor via a medical color camera (Hitachi, Tokyo; model DK5053U). Calibrations were made by using a slide micrometer. Mitotic index. The mitotic index was defined as the number of mitoses per ten high power fields (magnification x 400) as evaluated by routine light microscopy. Only readily recognizable figures in metaphase or anaphase were included in the count.

From the Department of Pathology, Beth Israel Medical Center. Newark, NT; and the Deoartment of Patholow, UMDNI-New _1erj;y Medical School, New&k. Accepted for pu&ation J;ly 19, .n

Kqr words: thyroid carcinoma, papillary carcinoma, morphometry, quantative microscopy. Address correspondence and reprint requests to Robert A. Ambros, MD, Department of Pathology, UMDNI-New [ersey Medical School, 185 S Orange Ave, Newark, NJ 07i03. 0 1989 bv W.B. Saunders Comnanv. 0046-817il89/2003-0004$5.00/d ’

215

HUMAN PATHOLOGY

Volume 20, No. 3 [March 1989)

sule at surgery. One multicentric lesion was identified and in four cases the tumor was encapsulated. Two patients with the follicular variant of papillary carcinoma were identified. Of the 28 patients, six had tumor recurrence, including five cases of distal metastases and one case of local recurrence. The mean time of recurrence after excision of the primary lesion was 34 months. The probability of tumor recurrence in relation to time is shown in Fig 1. There were two disease-unrelated deaths; there was one disease-related death in a patient who developed distal metastasis. Table 1 illustrates the association of the clinical and morphometric variables with tumor recurrence. By univariate analysis, only two variables significantly correlated with recurrence (Table 2). These included the ENASD and tumor size. Several other factors appeared to be associated with recurrence, but the degree of association did not achieve statistical significance. These included sex, lymph node status, cellularity mean index, papilla formation, and the mitotic index. Stepwise discriminate analysis most significantly associated recurrence with the nuclear area standard deviation. Once the value of this parameter was considered, tumor size yielded no additional information. The only factor that further contributed to prediction of recurrence was the cellularity mean index (P < .05). Using the ENASD and the CM1 as prognostic factors and values of 17 pm* and 40% as the respective cut-off points, five of nine patients (55%) with both values greater than the cut-off points developed recurrence. In contrast, only one of 19 patients (5%) developed recurrence when either of these values were below the cut-off point (P = .OOOl). Probabilities of no recurrence for patients with good and poor indicators are shown in Fig 2. No significant differences were found in nuclear values for formalin-fixed tissue and frozen sections (Table 3).

The cellularity mean index. The cellularity mean index (CMI) represented the percentage of tumor occupied by tumor cells in relation to stroma, cystic spaces, and glandular lumens. A grid containing subdivided arbitrary equidistant points was superimposed on the video screen (magnification x 400). The percentage of tumor volume occupied by tumor cells was then estimated by conventional point counting. l”,17 The CM1 represented the percentage of points superimposed by tumor cells. Nuclear parameters. Calculations of nuclear parameters were performed under oil immersion and at a magnification of x 1,000. By the selection technique described above, the largest and smallest axes of 25 nuclei were measured. The estimated nuclear area was then calculated by multiplying the greater axis by the lesser axis, as described by Bhattacharjee et al. I8 The mean areas of 25 nuclei and their standard deviations (estimated nuclear area standard deviation [ENASD]) were then recorded. Papilla formation and ground glass nuclei. Formalin fixation of papillary thyroid carcinomas has been associated with cytoplasmic invagination into the nuclear substance, forming the so called “ground glass nuclei.“lg In order to study the effect of this phenomenon on nuclear parameter values, frozen sections and formalin-fixed tissue sections were studied in the seven cases in which original frozen sections were available for evaluation. The point counting technique was then used to estimate the percentage of tumor occurring in papilla formation and the percentage of tumor cells with ground glass nuclei. The percentage of tumor occurring in papilla formation was obtained by calculating the percentage of nuclei superimposing points that comprised part of a papilla. Using the same technique, the percentage of tumor cells with ground glass nuclei was calculated.

Statistical Analysis The survival analysis procedure (SAS Institute, Cary, NC) was used to compute nonparametric estimates of sur-

vival distribution for censored observation (incomplete data due to withdrawal of experimental units or termination of the experiment), uncensored observations, rank tests for the association of the response variable with other variables, probability functions of recurrence in relation to time, and optimal subsets of independant variables correlating with recurrence by forward stepwise incremental analysis based on the largest increase in the joint test statistic. The product limit estimate was calculated by the method described by Kaplan and Meir.20 Differences between recurrence-free survival time in patients with good and bad indicators were evaluated by the log rank test and differences between mean nuclear parameters in formalinfixed tissue and frozen sections were evaluated by the paired Student’s t test. In all calculations, a P value of <.05

DISCUSSION This study reports the application of morphometric analysis in the prediction of recurrence of pap-

was considered statistically significant. RESULTS

Diagnosis was established in all 28 patients by either lobectomy or subtotal thyroidectomy. Twenty patients were women. At the time of diagnosis, the mean age was 48 years. The mean follow-up was 47.2 months (minimum, 6 months; maximum, 140 months). Eleven of the 28 patients had lymph node metastases at the time of original surgery. One patient had extension of tumor beyond the thyroid cap-

TIME (MONTHS) FIGURE 1. Distribution function for tumor recurrence in 28 patients with papillary thyroid carcinoma.

216

PAPILIARY TWROID CARCINOMA (Ambros et al]

TABLE 1.

Association of Clinical and Morphometric Parameters With Tumor Recurrence

Variable

Total

No.

---

PATIENTS WITH ENASD s

-PATIENTS

Recurrence

and CM=-40%

(n=9)

___---‘------‘-__‘---------,

I.0

(%)

17.0& and/orCMI< 40% (n=!9)

WITH ENASD>I~.O~~~'

I I

Age (~4 <50 >50 Sex Female Male Lymph node status Negative Positive Tumor size <3.0 cm >3.0 cm Encapsulation Yes No Estimated nuclear area ~60 pm* >60 pm ENASD <17 pm2 Z-17 Frn2 Mitoses l CM1 <40% >40% % Papillary <50% >50% % Ground glass nuclei <50% >50%

15 13

3 (20) 3 (23)

20 8

2 (10) 4 (50)

17 11

2 (12) 4 (36)

21 7

3 (14) 3 (43)

4 24

0 (0) 6 (25)

12 16

2 (17) 4 (25)

14 14

0 (0) 6 (43)

25 3

5 (20) 1 (33)

9 19

5 (26)

LOG RANKTEST' P=o.o001

7

5

CM1 Estimated nuclear area Encapsulation Ground glass nuclei Papillary formation Lymph node status Age Sex Tumor size Mitotic index

X2

P

17 11

2 (12)

15 13

4 (27) 2 (15)

4 (36)

Foward Stepwise Analysis

Value

X2 Increase

.03 .39 .38 .40 .74 .14 24 .81 .lO .04 .09

4.94 4.08 1.99 1.99 1.79 0.80 0.15 0.27 0.02 0.01 0.02

P

75

sented as the standard deviations of the nuclear area (anisotropism) and nuclear shape (true pleomorphism). By univariate analysis, we found only tumor size and the ENASD to correlate with recurrence. Forward stepwise incremental analysis revealed that the ENASD most closely correlated with recurrence. The value of tumor size was lost, implying redundant data, and only the CM1 further contributed to prediction. No other factor in our study was found to be of significance, including sex, age, encapsulation, mitotic rate, percentage of tumor occurring as papillae, percentage of tumor cells with ground glass nuclei, and lymph node status at surgery. Encapsulated papillary carcinomas have been associated with a very low rate of recurrence.*’ Although there was no recurrence in four cases of encapsulated tumors in our series, because of relatively short follow-up periods, this factor did not achieve statistical significance. The mitotic rate in our study was low in all cases. This corresponds to the findings of Lee et al,** who found a low rate of mitoses in cases of papillary carcinoma and suggested that the correlation of low mitotic index, the high degree of differentiation, and the low potential for invasion may explain high cure rates for treated papillary carcinoma. Although formalin fixation is associated with artefactual invagination of cytoplasmic processes into the nucleus, we found no difference in nuclear area between formalin-fixed and frozen tissue, suggesting that either are suitable for morphometric analysis. The biologic behavior of papillary carcinoma has been more closely linked to the occurrence of ground

2. Univariate and Foward Stepwise Incremental Analysis of Clinical and Morphometric Variables in Relation to Tumor Recurrence

4.94 0.74 0.79 0.69 0.11 2.22 1.38 0.05 2.72 4.40 2.82

-50

FIOURE 2. Distribution function for tumor recurrence according to ENASD and cellularily mean index.

TABLE

Variable

25

TIME (MONTHS)

illary thyroid carcinoma. In previous studies of papillary carcinoma, 4*8 the regional lymph nodes have been the most common sites of recurrence. Involvement of this structure, however, has not been associated with an unfavorable prognosis. In contrast, the most common type of recurrence in our review was distal metastasis which has been closely related with death due to tumor.8 Although pleomorphism literally means “many forms,” this term is frequently used to describe a variety of sizes and shapes. Quantification of nuclear pleomorphism by morphometric analysis is repre-

ENASD

t -0

l(ll)

Univariate Analysis

0.2

??

Value

TABLE

.03 .04 .16 .16 .18 .37 .70 .60 .88 .93 .88

3. Comparison of Frozen Sections With Formalin-Fixed Tissues in Seven Cases

Variable

Frozen Sections

Formalin-Fixed Tissue Sections

Estimated nuclear areat ENASDt

67.6 17.4

67.1 16.4

Abbreviation: NS, oat significant. * Paired Student’s t test. t In square micrometers.

217

P

Value* NS NS

HUMAN PATHOLOGY

Volume 20, No. 3 [March 1989)

glass nuclei than to actual papilla formation.3 As in a previous study,7 we found no correlation between recurrence and papillary formation or the percentage of tumor cells with ground glass nuclei. In summary, our study indicates that morphometric analysis may significantly contribute to the role of histopathology in the evaluation of papillary thyroid carcinoma and may also provide valuable information regarding prognoses not obtained by standard methods. Larger studies may establish the exact role of morphometry in evaluating the biologic behavior of this malignancy. Acknowledgment. The authors thank Fran Candies of the Tumor Registry, Newark Beth Israel Medical Center, for assistance.

REFERENCES 1. Tubiana M, Schlumberger M, Rougier P, et al: Long term results and prognostic factors in patients with differentiated thyroid carcinoma. Cancer 55:794-804, 1985 2. Torres J, Volpato RD, Power EC, et al: Thyroid cancer, survival in 148 cases followed for 10 years or more. Cancer 56:2298-2304, 1985 3. Rosai J, Carcangiu ML: Pathology of thyroid tumors: Some recent and old questions. HUM PATHOL 15:1008-1012, 1984 4. Vickery AL, Wang CA, Walker AM: Treatment of intrathyroid papillary carcinoma of the thyroid. Cancer 60:2587-2595, 1987 5. Black BM, Ya Deau RW, Wollner LB: Surgical treatment of thyroidal tumors. Arch Surg 88:610-618, 1964 6. Tollefsen HR, Shah JP, Huvos AC: Papillary carcinoma of the thyroid gland after surgical treatment. Am J Surg 124:468-472, 1972 7. Carcangiu ML, Zampi G, Rosai J: Papillary carcinoma of the thyroid. A clinicopathologic study of 241 cases treated at the University of Florence, Italy. Cancer 55:805-828, 1985

8. McConahey WM, Hay ID, Wollner LB, et al: Papillary thyroid cancer treated at the Mayo clinic, 1946 through 1970: Initial manifestations, pathologic findings, therapy, and outcome. Mayo Clin Proc 6 1:978-996, 1986 9. Tennvall J, Biorklund A, Moller T, et al: Is the EORTC prognostic index of thyroid cancer valid in differentiated thyroid carcinoma? Cancer 57:1405-1414, 1986 10. Weibel ER: Principles and methods for the morphometric study of the lungs and other organs. Lab Invest 12:131-154, 1963 11. Baak JPA, Van Dop H, Kurver PHJ, et al: The value of morphometry to classic prognosticators in breast cancer. Cancer 56:374-382, 1985 12. Van Der Linden, Baak JPA, Lindeman J, et al: Morphometry and breast cancer cells with high malignant potential in patients with spread to lymph nodes: Preliminary results. J Clin Path01 39:603-609, 1986 13. De Santis M, Sciarreta F, Sudan0 L, et al: Morphometric evaluation of histologic sections of the thyroid gland in benign and malignant follicular lesions. Diagn Cytopathol 3:60-67, 1987 14. Wright RG, Castles H, Mortimer RH: Morphometric analysis of thyroid cell aspirates. J Clin Path01 40:443-445, 1987 15. Hedinger LE, Sobin LH: Histological Typing of Thyroid Tumors, Vol II. International Histologic Classification of Tumors. Geneva, WHO, 1974 16. Chen KTK, Rosai J: Follicular variant of thyroid papillary carcinoma: A clinicopathologic study of six cases. Am J Surg Path01 1:123-130, 1970 17. Baak JPA, Oort J: A Manual of Morphometry in Diagnostic Pathology. Berlin, Springer-Verlag, 1983 18. Bhattacharjee DK, Harris M, Faragher EB: Nuclear morphometry of epitheliosis and intraduct carcinoma of the breast. Histopathology 9:511-516, 1985 19. Albores-Saavedra J, Altamivano-Dimos M, AlcortaAnguizola B, et al: Fine structure of human papillary thyroid carcinoma. Cancer 28:763-774, 1971 20. Kaplan EL, Meir P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1957 2 1. Evans HL: Encapsulated papillary neoplasms of the thyroid. A study of 14 cases followed for a minimum of 10 years. Am J Surg Path01 11:592-597, 1987 22. Lee TK, Myers RT, Marshall RB, et al: The significance of mitotic rate: A retrospective study of 127 thyroid carcinomas. HUM PATHOL 16:1042-1046, 1985