If surgical specimens of multinodular goiters are examined carefully, 4-17% are found to harbor a carcinoma.43,50-52 These carcinomas vary widely in size and are typically of the papillary variety. Similar tumors are occasionally found in thyroid glands affected by Hashimoto's thyroiditis and in otherwise normal glands. Stoffer et al.58 reported that 13% of the glands resected in thyroid operations for any reason contained papillary adenocarcinoma. In Japan, routine autopsies of patients who were not suspected of having thyroid disease and who had no known irradiation experience, 17% were found to have small carcinomas when careful serial sections of the thyroid glands were done.59 If the figures of Stoffer et al, that were recently confirmed 50, truly represent the prevalence of invasive carcinoma, one would certainly be forced to conclude that all multinodular goiters should be resected in order to prevent dissemination of malignant disease. However, it seems quite unlikely that all lesions that appear to satisfy the histological criteria for malignant neoplasia are potentially lethal. This view is strongly supported by the final report of the study on the significance of nodular goiter carried out by Vander et al.3 in Framingham, Massachusetts. They followed for 15 years all 218 nontoxic thyroid nodules previously detected in a total population of approximately 5,000 persons. None of these lesions showed any clinical evidence of malignancy at the end of that time.

A strong case can be made for the view that there is only minimal risk from carcinoma in multinodular goiter. Sokal 60,61 has presented this argument in detail, and we can do no better than to borrow directly from his published analysis. The prevalence of clinical nodularity of the thyroid is at least 4%, or 40,000 per 1,000,000 population.2 Use of a much higher figure can be justified by the autopsy studies described above. Despite the high frequency of nodular goiter, only 36-60 thyroid tumors appear per 1,000,000 persons each year.62,63 or by analysis of reported statistics on thyroid surgical specimens.61 A recent national cancer survey in the United States found an incidence of 40 per 1,000,000.63 Riccabona 64 published an overview of the incidence of thyroid cancer in 40 countries, both with and free of endemic goiter. The range of incidence varied between 7.5 and 56 per 1,000,000 persons each year. There is no increased goiter rate in endemic goiter areas. The prevalence of significant thyroid carcinoma at routine autopsy is less than 0.1% 61,65,66 and persons with this type of tumor are probably examined as frequently as are those with other forms of neoplasia. The United States mortality figures for thyroid carcinoma are constant at about 6 per 10-6 population each year. Riccabona also summarized death rates from thyroid cancer in non-endemic and in endemic countries.64 For Austria this was 16 per 10-6 per year in 1952 and 10 per 10-6 per year in 1983. For Switzerland this was in 1952, 18 per 10-6 per year and in 1979, 9 per 10-6 per year. The death rate per year for the United States in 1979 was 3 per 10-6, for Israel in 1952 1 per 10-6 per year and for the UK 7 per 10-6 in 1963. Death rates from thyroid cancer in endemic goiter areas from regions in Austria, Yugoslavia, Finland and Israel were between 10 and 16 per 10-6 per year between 1980 and 1984.

Lastly, it should be recognized that meticulous examination of autopsy specimens from persons dying of nonthyroid disease may show small (less than 0.5 cm) papillary lesions in 4-24% of human thyroid glands.67-69 A recent report of 1020 sequential autopsies revealed the presence of microscopic papillary carcinoma in 6%.66 Although the prevalence of this type of lesion increases with age, there is no question that such lesions may be present even in younger persons. The proportion of these lesions that even become clinically apparent is unknown, but their presence in otherwise normal thyroid glands should be kept in mind when evaluating reports of similar prevalences of thyroid carcinoma in multinodular thyroid glands.

If 4% of patients with nodular goiter actually have thyroid carcinoma, the prevalence of tumor in the general population would be 1,600 per 1,000,000. It is remarkable that only about 25 of these 1,600 hypothetical tumors would become apparent each year, or that only about 10 would prove fatal. Thus, there appears to be a gross discrepancy between the mortality from thyroid carcinoma and its reported frequency in surgical specimens of multinodular goiters. Reasonable arguments can be mustered in an effort to reconcile the information. Perhaps the most important single factor is selection. Persons with nodular goiter who come to operation are not representative of the general population but are patients with clinically significant thyroid disease who have been selected by their physicians for thyroid surgery. One of the factors controlling the selection process is the suspicion of malignant tumor. In fact, the selection process is especially good, as reflected by the high recovery of malignant thyroid tumors in patients operated on with this presumptive diagnosis.70 A second factor is that the histologic diagnosis of thyroid carcinoma may not correlate well with true invasiveness. It is impossible to prove this thesis, but pathologists agree that the criteria for judging malignancy are variable and that it is exceedingly difficult to predict with any degree of certainty the growth potential of a particular thyroid lesion.

Other arguments may be used to defend a conservative therapeutic position. In the first place, the tumors that are usually found in multinodular goiters are papillary tumors, and their degree of invasiveness is low. Indeed, the survival rate for intrathyroid papillary carcinoma is only slightly less than that for normal persons of the same age and sex.71 Furthermore, prophylactic subtotal thyroidectomy is not a guarantee of protection from cancer arising in a nodular goiter, since the process is usually diffuse, and it may be assumed that abnormal tissue is left in the neck after operation. In fact, unless replacement therapy is given, partial thyroidectomy might be expected to induce a tremendous growth stimulus in the remaining gland. A further point is that thyroidectomy, even in the best of hands, carries its own risk and its own morbidity, with dimensions comparable to those of missing a small papillary carcinoma within a multinodular goiter. Obviously this last possibility does not apply when a focus of unusual induration or rapid growth rate is detected clinically.