Author reply

In our recently published series of 16 cases of columnar cell types of papillary carcinoma,1 we included a critique of an article by Pilotti et al.2 in which those authors found the so-called "poorly-differentiated types" of papillary carcinoma, including the columnar cell and tall cell types, to be distinct morphologic types with uniformly aggressive biologic behavior. The findings of Pilotti et al.2 then would indicate that all thyroid papillary carcinomas with the morphologic characteristics of columnar or tall cells irrespective of size or extent of invasion should be considered and treated as aggressive thyroid carcinomas. In a correspondence regarding our article on the columnar cell type of thyroid papillary carcinoma,1 Pilotti et al. state that the conclusions we drew relative to the biologic behavior of the columnar cell type of papillary carcinoma are untenable owing to the absence of a statistical comparison with other thyroid papillary carcinomas. These authors state that to make a valid comparison a statistical correlation is required similar to the one that they performed in their study.2 It is true that in our article we did not include a statistical comparison with other types of papillary carcinoma. However, we would argue that a statistical comparison, although preferable, is not necessary to arrive at the conclusions we made in our article and the absence of statistical correlation does not invalidate our findings. Pilotti et al.2 continue to hold to the concept that the columnar cell type of papillary carcinoma (and for that matter the tall cell type) is a poorly differentiated tumor simply on the basis of cell type. However, we would ask what are the histologic features that define these tumors as poorly differentiated? In their letter to us, these authors fail to indicate what these "poorly differentiated" features represent but cite three other articles to support their contentions. The first of these three cited articles is by Sakamoto et al.,3 who separated well differentiated from poorly differentiated papillary carcinoma on the basis of whether a tumor had gland formation. These authors indicate that "When nonglandular components were found in papillary and follicular carcinomas on histologic examination, the tumor was denominated poorly differentiated carcinoma."3 We cannot accept this definition for a poorly differentiated thyroid carcinoma because solid, nonglandular foci can be observed in nonneoplastic lesions of the thyroid (e.g., dyshormonogenetic goiter), in benign thyroid tumors (e.g., hyalinizing trabecular adenoma), and in thyroid malignant tumors that behave indolently (e.g., minimally invasive follicular carcinoma and papillary carcinoma). According to the Rosai et al.,4 there are several histologic parameters that define a poorly differentiated thyroid carcinoma, including (but not limited to) the presence of nuclear pleomorphism, mitotic activity, and necrosis. None of these features were used by Sakamoto et al.3 in their study on so-called poorly differentiated thyroid carcinomas. The next study cited by Pilotti et al. in their letter is the article by Carcangiu et al.5 regarding the thyroid "insular carcinoma." It has been shown clearly by other authors6, 7 that an "insular" growth pattern does not in and of itself equate a poorly differentiated thyroid carcinoma nor does the presence of an insular growth confer an aggressive behavior to any thyroid neoplasm. Therefore, insular carcinoma should not be included uniformly as a poorly differentiated carcinoma simply on the basis of having an insular growth pattern. This is not to indicate that there are not aggressive-behaving thyroid carcinomas with insular growth. The latter do exist but in addition to an insular pattern includes the presence of necrosis, increased mitotic activity, and invasive growth. Pilotti et al. ask "how [the authors] can question insular carcinoma when their study is restricted to 16 cases of CC (columnar carcinoma, our addition)." We included insular carcinoma in our study of columnar cell carcinomas to refute the concepts that the biologic behavior of thyroid tumors should be predicated solely on the growth pattern (i.e., insulae) or on the cell type (i.e., columnar or tall.) To this end, we indicate that thyroid lesions not considered within the scope of "insular carcinoma" nonetheless may have an insular growth and that these lesions do not behave aggressively. Our findings (sans statistics) more than adequately refute these erroneous concepts, including those cited in reference 5 of Pilotti et al.'s correspondence in which Sobrinho-Simões et al. indicate that columnar cell carcinomas are poorly differentiated tumors. In their letter, Pilotti et al. state that "papillary features are well known in insular carcinoma." We would agree that "insular carcinomas" are not restricted to the category of follicular carcinomas and that papillary carcinomas also may demonstrate an insular growth pattern.5 However, there remain some unresolved issues relative to thyroid tumors with insular growth, including classification. According to the World Health Organization Committee on the classification of thyroid tumors, insular carcinoma is considered a morphologic variant of follicular carcinoma.8 The clinicopathologic features of the columnar cell and tall cell types of thyroid papillary carcinoma have not been defined completely. Not the least of the issues revolving around these tumor types is what exactly is a "tall" cell. For unexplained reasons, many studies reporting on the columnar and tall cell types of papillary carcinoma demonstrate the fact that these tumor types tend to occur in older patients with extrathyroidal extension of their tumor. We agree with Pilotti et al. that pT4 thyroid tumors, defined as a tumor of any size extending beyond the thyroid capsule, represents a stronger factor in predicting prognosis than tumor size. We would extend this argument to state that this applies to all thyroid papillary carcinomas irrespective of cell type such that pT1 columnar cell types of papillary carcinoma behave no differently from pT1 usual papillary carcinomas, and so forth. This matching of different types of papillary carcinoma in each pT category was not done by Pilotti et al.2 Furthermore, nowhere in their letter did Pilotti et al. discuss the other recently published studies showing that columnar cell carcinomas are not uniformly aggressive tumors but will behave in an indolent manner if they are encapsulated, have limited invasion, and do not show extrathyroidal invasion.9, 10 These findings support the contentions of our study1 and validate the fact that the behavior of all types of thyroid papillary carcinoma are based on features other than cell type (i.e., columnar, tall, oxyphilic, etc) growth pattern (i.e., insular). Our findings as well as those of other authors9, 10 dispute statements such as those made by Pilotti et al. indicating that " ...the categorization of a carcinoma as poorly differentiated papillary carcinoma, or as one of the three variants that belonging to the group (including columnar cell, tall cell, and mixed type, our addition), clinically implies the presence of a high risk tumor."2 We feel completely justified in making the point that each patient's thyroid papillary carcinoma, irrespective of growth pattern or cell type, should be evaluated on its own and not lumped within a category of tumors that clearly do not uniformly follow an aggressive biologic course. Our findings validate this point and show that the columnar cell type of thyroid papillary carcinoma is not necessarily a distinct biologic entity separate from the usual morphologic types of papillary carcinoma. Bruce M. Wenig M.D.*, Clara S. Heffess M.D.*