Abstract
Background:
A systematic analysis of the clinical and pathologic patterns of childhood “sporadic” thyroid carcinoma in Belarus, in the absence of the “Chernobyl radioactive iodine factor,” has never been performed. The aim of this study was to establish the essential features of “sporadic” papillary thyroid carcinoma (PTC) in Belarusian children and adolescents, and the relationship of tumor pathology to extrathyroidal extension (ETE) and lymph node metastases.
Methods:
This was a retrospective population-based study with assessment of histological samples of 119 cases of thyroid cancer in Belarusian children and adolescents of 0–18 years old registered during 2005–2008 years. Sporadic PTC was noted in 94 children who were not exposed to the Chernobyl radiation release. None of the 119 cases of thyroid were follicular thyroid cancer.
Results:
The incidence rate of PTC was 1.13 per 100,000 persons. The median age at diagnosis was 15.1 years with fourfold predominance of diagnosis in female patients. Relapse was detected in 2% of cases with median follow-up of 4.2 years. Median tumor size was 12 mm. Three percent of the cases of PTC had multifocal growth. The classical variant of PTC was registered in 46% of the patients with thyroid cancer, the follicular variant of PTC was noted in 20% of the cases. The percent of rare types of PTC (tall cell and diffuse sclerosing) were equal to that for solid PTCs (13%, 12%, and 10%, respectively). Adolescents had a pure papillary carcinoma more often compared to children who represented tumors with mixed papillary/follicular patterns more frequently (p<0.05). Two-thirds of the patients with PTC had regional lymph node metastases. ETE was established in 39 of 74 patients in whom ETE could be assessed by morphology. Multivariate analysis showed that lymphatic invasion was the strongest independent factor associated with both ETE (p<0.0001) and lymph node metastases (p<0.0001).
Conclusion:
In 2005–2008, sporadic thyroid cancer in children of Belarus was represented by high prevalence of PTC and absence of follicular thyroid cancer. Sporadic cases of PTC in Belarus were characterized by smaller tumor size, a small number of cases with multifocal growth, an equal number of rare types and solid PTCs, a relatively high prevalence of pure papillary variant of PTC in adolescents, and a low frequency of early relapses. A high frequency of ETE and lymph node metastases was detected. The strongest morphologic factor associated with both of them was lymphatic invasion.
Introduction
There are few well-documented facts concerning the clinical presentation and pathology of PTC in children exposed to radioactive contamination. Nevertheless, some common features were revealed, including a high frequency of extrathyroidal extension (ETE), lymph node metastases, and lung metastases (10,11). As a rule, radio-induced PTC tended to be the mixed pattern of papillary follicular and solid follicular structures, with no tall cell or columnar cell variants being identified (12). Less information is available for “sporadic” PTC (13).
A systematic analysis of clinical and pathologic patterns of childhood “sporadic” thyroid carcinoma in Belarus in the absence of the “Chernobyl radioactive factor” has never been performed. The aim of this study was to establish the essential features of “sporadic” PTC in Belarusian children and adolescents, and the relationship of tumor pathology to spread beyond the thyroid capsule and lymph node metastases.
Methods
This retrospective study was approved by the Belarusian Research Center for Pediatric Oncology and Hematology institutional board and recruited all cases of any form of thyroid cancer (119 patients) registered prospectively in the population-based Cancer Subregistry of Belarus between January 2005 and December 2008.
Patients
Ninety-four consecutive cases of sporadic PTC were included into the study. None of these had a history of previous irradiation or chemotherapy, or any apparent familial or congenital predilection. Excluded patients were as follows: 10 children and adolescents suffering from PTC had contact with external (4 patients received multimodal treatment due to epithelioid sarcoma, medulloblastoma, Hodgkin's disease, and acute lymphoblastic leukemia) or in utero (4 girls were born between May 1986 and January 1987) irradiation or other malignancies (PTC in 2 patients associated with familial adenomatous polyposis). Seven patients suffered from thyroid carcinoma but not PTC (three from well-differentiated “not otherwise specified,” and four from medullary carcinoma). In six children and adolescents the diagnosis of malignancy was not confirmed in the final histology, and no slides were available in last two cases.
Pathological diagnosis
The initial pTNM staging was derived from patients' records; reclassification of pT categories was determined according to the seventh edition of TNM classification (14). Removed parts of the thyroid gland were sectioned along cephalocaudal axis (from the upper to the lower pole in vertical plane). Equal fragments of 3–5-mm thickness were fixed and embedded in paraffin; histological slides of 5-micron thickness, stained with Mayer's hematoxylin and eosin, were made.
Two professionals in endocrine pathology (M.F. and K.W.S.) conducted the retrospective review of all histological slides—agreed on a diagnosis, subclassification, and any specific features. Thyroid gland size, tumor dimension on gross examination (information presented in histopathological reports), multifocality, degree of associated lymphocytic infiltration or Hashimoto thyroiditis, ETE, circumscribed or infiltrative growth, blood and lymphatic vessel invasion, and the character of connective tissue distribution were reevaluated. Finally, the 94 PTC cases of this series were subdivided into six distinguished variants (15).
Statistical methods
The statistical package R was used to perform statistical computing (16).
The difference between the frequency of each feature represented by categorical variable was compared using Fisher exact test for 2×2 tables, Fisher–Freeman–Halton exact test for R×C tables, or χ2 test when expected frequencies in the R×C tables were >5 for 80% of cells; some levels of the categorical variable were collapsed for analysis to prevent empty cells in the R×C table. The difference between values of each feature represented by continuous variable was compared using Mann–Whitney test. p-Values <0.05 were considered statistically significant without multiple comparison correction.
Multivariate analysis was used to define sets of variables associated with ETE and lymph node metastases. Only variables with p-values <0.2 in the bivariate analysis were selected for the multivariate models. We also included a patient's age as a well-known confounder although it was not significant in bivariate analysis. ETE (yes/no) and lymph node metastases (yes/no) were considered as two separate outcomes and the multivariate logistic regression was applied for both preliminary models. Since the complete separation occurred for each preliminary model, we implemented Firth's penalized likelihood logistic regression (17,18). The preliminary models were significant comparing with null models by penalized likelihood ratio test (PLRT). To find the smallest set of variables that significantly associated with outcome, the different subsets of variables in each preliminary model were explored (the null hypothesis of zero coefficients in the preliminary model for subset of variables was tested by PLRT). As soon as the largest subset of variables that do not significantly reduce the penalized likelihood of the preliminary model was found, the rest of variables were considered as the smallest set. The final models were fitted based on the smallest set of variables (for each model, respectively). Confidence intervals were estimated based on profile penalized likelihood. Odds ratio for predictors was calculated as exponential transformation of corresponding values of coefficients and interval estimations. The package “logistf” of R was used to perform Firth's penalized likelihood logistic regression (19).
Results
By our results, PTC was the dominant form of the thyroid malignancy in Belarusian children (93.6%; 104 of 111 patients in whom the diagnosis of thyroid cancer was confirmed). “Sporadic” PTC was registered in 84.7% (94 of 104 cases); its main clinical and pathological features are listed in Table 1. ETE was noted in 39 of 74 patients in whom ETE could be assessed by morphology (Table 2). We revealed a significant association between ETE and peritumoral/intraglandular distribution of psammoma bodies (p=0.0015), nodal status (p=0.0055), any volume of solid pattern (p=0.004), extensive fibrosis (p=0.005), and lymphatic invasion (p<0.001). Moreover, a significant association has been found between nodal disease (N1) and expansive or encapsulated growth (p<0.001), lymph vessel involvement (p<0.001), presence of peritumoral and/or intraglandular psammoma bodies (p<0.001), and heavy (nodular type) of mononuclear infiltration (p=0.002) (Table 3).
Data are presented as n (%), mean ± SD, or median (range minimum–maximum) unless otherwise indicated. [Subcategories expanded for analysis are presented in brackets.]
p-Values were determined using: *Fisher exact test; #Fisher–Freeman–Halton exact test; ‡Mann–Whitney test; ×χ2 test.
In multifocal growth, only the largest tumor was considered.
N/A, not applicable; LND, lymph node dissection.
Data are presented as n (%). [Subcategories expanded for analysis are presented in brackets.]
p-Values were determined using: *Fisher exact test; #Fisher–Freeman–Halton exact test; ×χ2 test.
ETE, extrathyroidal extension.
Data are presented as n (%). [Subcategories expanded for analysis are presented in brackets.]
p-Values were determined using: *Fisher exact test; #Fisher–Freeman–Halton exact test; ×χ2 test.
Multivariate analysis
The preliminary model for ETE included the following variables: (i) patient's age group, (ii) localization: inside lobe versus others, (iii) tumor size: ≤10 mm versus >10 mm, (iv) nodal status, (v) growth: expansive or encapsulated versus others, (vi) solid present in pattern, (vii) fibrosis, (viii) vascular invasion, (ix) lymphatic invasion, and (x) peritumoral/intraglandular psammoma bodies. Penalized log likelihood of the preliminary model was found at −24.88, compared with null model at −44.75 (χ2=39.73 [df=10], p<0.0001). PLRT for the null hypothesis of zero coefficients of subset of variables (patient age, localization, nodal status, growth, fibrosis, psammoma bodies, and vascular invasion) showed χ2 as 5.89 (df=7), p=0.552. Therefore, we excluded tested variables from the model. The final model contains variables presented in Table 4.
OR, odds ratio; CI, confidence interval; LI, lymphatic invasion; SP, solid present; TS, tumor size.
The preliminary model for lymph node metastases included the following variables: (i) patient's age group, (ii) localization: inside lobe versus others, (iii) growth: expansive or encapsulated versus others, (iv) pure follicular not present in pattern versus others, (v) lymphatic invasion, (vi) peritumoral/intraglandular psammoma bodies, and (vi) mononuclear infiltration: nodular versus others. Penalized log likelihood of the preliminary model was found at −23.95, compared with null model at −51.29 (χ2=54.68 [df=7], p<0.0001). PLRT for the null hypothesis of zero coefficients of subset of variables (patient age, localization, growth, psammoma bodies, and mononuclear infiltration) showed χ2 as 9.09 (df=5), p=0.106. As a result, we excluded tested variables from the model. The final model contains variables presented in Table 5.
FN, pure follicular not present.
Lymphatic invasion was the strongest independent factor associated with the ETE (Table 4) and lymph node metastases in children with PTC (Table 5; p<0.0001).
Discussion
This study represents a unique cohort of 94 patients with sporadic PTC patients younger than 18 years of age, whereas <100 cases of different morphological and etiological types of thyroid cancer combined together have been described in the majority of published articles (3,5,20 –23) with the age extension up to 19 and 20 years in some of them (3,20 –22). Thyroid cancer is uncommon in childhood. In Belarus, the incidence of radioactivity-induced childhood thyroid carcinoma was significantly increased during the first decade after Chernobyl accident (9 –11). However, the incidence rates of “sporadic” PTC registered in the years 2005–2008 in children and adolescents seem to be slightly higher in Belarus than in other countries (1 –5), probably, due to early detection.
In the cohort of sporadic PTC analyzed in our study, the median age at diagnosis was 15.1 years. There was a fourfold predominance of diagnosis in female patients, and we detected numerous aggressive patterns despite few instances (4%) of lung metastasis. Specifically, 75% of 94 tumors had an infiltrative growth pattern, and two-thirds of the cases had spread to regional lymph nodes. Lymphatic invasion was detected more often than vascular invasion (80% and 15%, respectively). The classical histological type of PTC was registered more frequently (46%), but other types (follicular, diffuse sclerosing, tall cell, and solid) were also found. Comorbidity was diagnosed in one-third of patients, and autoimmune thyroiditis was more common. Differences in clinical and pathological characteristics of sporadic PTC between children and adolescents were not revealed (p>0.05) except for morphology. Adolescents had a pure papillary carcinoma more often than children, who more frequently had tumors with mixed papillary/follicular patterns (p<0.05).
Compared to “sporadic” PTC in different countries, a similar median age and female/male ratio was found in our cohort of PTC (2,19). However, the following peculiarities that contrasted with typical features were observed. The tumor nodules were relatively small (median 12 mm), there was a total absence of follicular thyroid carcinoma cases during the study period, and there were a relatively small number of PTC cases with multifocal growth (only 3 of 94 cases). For example, the median for primary tumor size in Canadian and German children was 31 mm and 23 mm (5,23); the median frequency of follicular thyroid carcinoma was 15%–20% (20,24) and varied from 6.1% in Russia to 10%–11% in the United Kingdom, Canada, and the United States (3 –5,13). Multifocal growth, noted by other authors as typical for childhood thyroid cancer, was registered more often in other studies (20,23). Classical and follicular histological types of sporadic PTC were as predominant in our patients as in children in Europe and United States (4,13); however, the percentages of rare types of PTC (tall cell and diffuse sclerosing) were equal to solid PTC (13%, 12%, and 10%, respectively).
In our cohort, assessment of the relationship between tumor essentials and spread of sporadic PTC beyond the thyroid gland capsule showed that ETE was diagnosed significantly more often in patients with tumor size more than 10 mm, N1b nodal status, solid patterns of carcinoma, extensive fibrosis, lymphatic invasion, and peritumoral/intraglandular psammoma bodies (p<0.05). Moreover, lymph node metastases were revealed more often in cases of PTC with infiltrative growth, and predominance of papillary pattern. Pure follicular sporadic PTC was associated with N0 in the majority of cases (p<0.05). Multivariate analysis showed that lymphatic invasion was the strongest independent factor associated with both ETE and lymph node metastases (p<0.0001).
Such characteristics as ETE and lymph node involvement corresponded to those published in children with PTC (ETE ranges from 9.6% in Russia to 67.6% in South Korea; frequency of lymphatic nodes' involvement is at the high level from 40.7% in Canada up to 84.3% in Germany) (3,5,22,23). Interestingly, the age of our patients at the time of operation did not play significant role in such signs of aggressive behavior as ETE, nodal disease, or recurrence. However, all cases of an advanced process (lung metastases) were registered only in children <14 years old.
Radioactive–induced thyroid cancer in Belarus had a lower preponderance of females (1.6/1) compared with sporadic thyroid cancer in our study, but the vast majority of both etiological types of thyroid cancers were PTC with frequent ETE and lymph node metastases (12,24 –26).
PTC in children and adolescents in other countries tended to have a high relapse level up to 34% with numerous contributing factors, including the invasion of capsule and perithyroid soft tissues, presence of metastases in lymph nodes and lungs, tumor nodule size exceeding 10 mm, and focality (20,22,27). We have not analyzed these risk factors, as there were only two patients who had relapse of their PTC.
In conclusion, in 2005, 2006, 2007, and 2008, sporadic thyroid cancer in children of Belarus was characterized by high prevalence of PTC and absence of follicular thyroid cancer. Most of analyzed clinical and morphologic parameters of sporadic PTC in Belarusian children were similar to those described in other countries. However, sporadic PTC in Belarus displayed a lower tumor size, small number of cases with multifocal growth, prevalence of pure papillary variant in adolescents, and a low frequency of early relapses. A high frequency of ETE and lymph node metastases was detected, and lymphatic invasion was established as the strongest morphology factor associated with both of them.
Footnotes
Acknowledgments
The authors express our deep appreciation to Professor Olga V. Aleinikova, the Director of Belarusian Research Center for Pediatric Oncology and Hematology (Minsk, Belarus), for her collaboration.
This study was supported by the International Science and Technical Center (ISTC project B-1910).
Disclosure Statement
The authors declare that no competing financial interests exist.