Definition
The problem of the thyroid incidentaloma is the problem of the clinical significance of an impalpable small thyroid nodule (usually less than 10 mm diameter and not more than 15 mm) discovered incidentally during investigation for another disease process or during screening tests on a well subject. With minimal simplification, the point at issue is the risk of the nodule being papillary thyroid microcarcinoma and the significance of this diagnosis i.e. the potential morbidity and mortality. The term incidentaloma is a clinical one best used for the as yet undiagnosed nodule and if a malignancy is subsequently identified the term then to be used is microcarcinoma. It is impossible to avoid a degree of arbitrariness in definition, both in the criteria of size and impalpability. A nodule of 15 mm or even more in the postero-medial aspect of the gland may be impalpable whereas one of less than 10 mm in the isthmus might be readily apparent clinically. Nevertheless such categorization allows formal studies and is useful in management decisions.

Demographics
Palpable thyroid nodules have prevalence in iodine-replete populations of about 5% and an incidence of about 1 per 1000 per year (1). Ultrasonographically detectable thyroid nodules are even more common at 17-67% (2) but clinically important thyroid cancer is uncommon (20-80/million/year) and deaths from thyroid cancer are rare (5-8/million/year). Nevertheless it is clear that even impalpable thyroid nodules can be malignant and that malignancy in the multinodular thyroid has the same prevalence as in the solitary thyroid nodule (3).

It has been known for many years that there is a high prevalence of occult thyroid cancer detected at autopsy (almost entirely papillary carcinoma),) ranging from 2.7-28.4% in various surveys (4) with a median prevalence of about 6% (5) and it has been thought to pursue an indolent course (6, 7, 8). Recent publications reaffirm the generally indolent course and good prognosis of small papillary carcinomas under 10 mm (9) or under 15 mm (10).

The previously largely academic issue of occult thyroid carcinoma has now moved into the clinical domain as the problem of the thyroid incidentaloma, now commonly detected since the advent of widespread use of ultrasonography, although incidentalomas are now also being found by positron emission tomographic scanning (PET scanning) (11) which is (as yet) in much more restricted use. Kang et al (12) found a 2.4% prevalence of focal thyroid incidentalomas by PET scanning in a cancer screening program and a 1.6% prevalence in a combined group including subjects being evaluated for metastases from a non-thyroidal malignancy, indicating a high prevalence of incidentaloma although lower than that of both ultrasonographically-detected nodules and occult microcarcinoma.

TABLE 1: THYROID NODULE AND CANCER STATISTICS

These figures are approximate and have been rounded off for mnemonic value. In particular it should be noted the prevalence of nodules on ultrasonography increases with age.

The clinical problem
If an impalpable thyroid nodule is identified on ultrasonography performed for an unrelated reason the main questions that arise are:
Should we investigate further?
If we biopsy the nodule what does the cytological diagnosis of malignancy mean biologically? i.e. is the prognosis the same as for clinically apparent cancer and should our management protocols be the same?
Furthermore, if occult thyroid cancer is common should we actively seek it by population screening or case-finding in high-risk groups?
Although ultrasonography has created the problem, it at least offers a method of evaluation by allowing guided fine needle aspiration biopsy [USGFNA] (13) and ultrasonographic imaging characteristics have the potential to assist in defining the risk of malignancy.

Review of recent relevant studies
These studies have better defined the extent of the problem, in particular the prevalence of carcinoma and its likely anatomical extent, the ultrasonographic criteria of malignancy, and have provided some data assisting in prognostication and formulation of a management plan.

Leenhardt et al (14) in 450 impalpable nodules, found that malignancy was present in 21% of those operated [20/94] or in 4.4% of the total nodules, with a rate of 33% [8/24] in nodules under 10mm and 17% [12/70] in nodules 10mm or more. Solid hypoechoic nodules were significantly more likely to be malignant but neither blurred nodule margins nor calcification were significantly associated with malignancy.

Papini et al (15) found that of 402 impalpable nodules in which adequate cytology was obtained, carcinoma was present in 9.1% of nodules with a diameter more than 10 mm and in 7.0% of those with a diameter less than 10mm diameter. Extrathyroidal extension was present in 35.5% and regional lymph node extension in 19.4%. Ultrasonographic features of solid hypoechoic nodules, irregular nodule margins, intranodular vascular spots, and microcalcification were associated with malignancy.

Nam-Goong et al (16) performed USGFNA on 267 patients with 317 impalpable nodules ranging from 2-15 mm in a retrospective analysis of all patients referred to them over a two-year period. The histologically proven malignancy prevalence was 12%. Of particular importance is the finding that one or both of extrathyroidal extension of malignancy and regional lymph node metastases was present in 69% of the occult thyroid cancers. Extrathyroidal extension was present in 44%, regional lymph node metastases in 50%, and 39% of cancers were multifocal. No distant metastases were found after thyroidectomy and whole body [131]I scanning. There was no difference in the malignancy rate between thyroids with single or multiple nodules, or between nodules less than 10 mm diameter or 10-15 mm diameter. Ultrasonographic characteristics significantly associated with the subsequent confirmed diagnosis of malignancy were solidity, hypoechogenicity, and punctate calcification. Ill-defined nodule margins and increased intranodular vascularity on colour-flow Doppler imaging had an observed but non-significant association with malignancy.

Kang et al (17) in a retrospective analysis of 198 incidentalomas less than 15 mm diameter, in which 28.8% were malignant, identified ultrasonographic characteristics assisting in the diagnosis of malignancy. Malignancies were less likely to have a well-defined border, or to be cystic, and more likely to be solid, hypoechoic and to have calcification. Using these factors to derive an ultrasonographic index, a score of less than 2 had a diagnostic accuracy for a benign lesion of 85.5% and a score of more than 3 a diagnostic accuracy for malignancy of 89.9%. 12% had involved lymph nodes and 20% capsular invasion.

TABLE 2: ULTRASONOGRAPHIC CRITERIA OF THYROID MALIGNANCY

YES: significant association with malignancy NS: no significant association found
ND; not done or not reported

High quality ultrasonography of the incidentaloma itself may thus aid in diagnosis but is not informative about biological behaviour. A potentially helpful prognostic ultrasonographic feature, however, has recently been reported by Ito et al (18) of a higher rate of local lymph node recurrence when pre-operative ultrasonography has identified lateral compartment lymph node metastases (6.0% vs. 1.1% when absent) whereas medial compartment lymph node metastases had no such adverse correlation. The ultrasonographic diagnosis of lateral compartment lymph node metastases was likely to be confirmed at operation (positive predictive value 81%) but was insensitive (38%) although this insensitivity had no adverse effect in follow-up (19).

The malignancy rate in incidentalomas discovered by PET scanning appears to be similar. A high rate of clinically relevant malignancies has been suggested from a series of eight patient referred with suspicion of various malignancies (20) in whom two medullary thyroid carcinomas and three papillary carcinomas were found at surgery. In a retrospective series of 1330 subjects Kang et al (12) found 21 focal incidentalomas of which 15 were diagnosed histologically or cytologically diagnoses, and 4 (26.7%) were papillary carcinomas. The mean standard uptake value (SUV) of the malignancies was higher than that of the benign lesions.

Pellegriti et al (10) in a retrospective study of 299 patients operated for papillary carcinomas less than 15mm found 32% to be multifocal, 20% extrathyroidal, and 30% to have lymph node metastases. Tumours incidental to multinodular goitre, Graves’ disease, autonomous functioning nodules, and cysts (n=151) were less likely to be multifocal (25%), extrathyroidal (11%), or to have lymph node metastases (16%) but the rates of these adverse features are still of concern. Distant metastases were uncommon in the total group (8/299) but appeared to be even more uncommon in incidentalomas (1/151).

Discussion
These rates of invasive disease reported by Nam-Goong et al (16) Papini et al (15), Kang et al (17), and Pellegriti et al (10) are evidence against the view that such small thyroid cancers are mostly indolent and of little biological importance. It might seem axiomatic to conclude that invasive or metastatic carcinoma must have an adverse outcome unless it is managed aggressively but for thyroid carcinoma this view must be considered against the high prevalence of thyroid nodules, the much lower incidence of clinically apparent thyroid carcinoma, and the even lower mortality rate of thyroid carcinoma. For instance, the reported thyroid carcinoma incidence rate in Korea is 64/million/year and the death rate is 5/million/year similar to that of the USA for 1995-1999 of an incidence rate of 66/million/year and death rate of 5/million/year. In contrast, were the prevalence of thyroid carcinoma suggested by Nam-Goong et al (16) to be applicable to thyroid nodules in general, over 4% of the population would harbour thyroid cancer and 3% would have invasive disease.

How can we reconcile these disparate observations? First, it is probable that widespread use of thyroid ultrasonography, either in a screening program or as case finding in a higher–risk segment of the population, would indeed lead to a rise in incidence rates. Although retrospective studies and studies composed of subjects referred for investigation may result in some distortion of the true incidence in the general population it still seems likely that study findings are accurately showing in a clinical context what the past autopsy data has shown in a pathological context. Thus the critical issue is the natural history and prognosis of microcarcinomas presenting as incidentalomas not their incidence. Mortality rate data are sturdier than incidence rate data and are not directly challenged by any of the current studies of prevalence and invasiveness. If 3-5% of the population have invasive thyroid carcinoma with a presumptive adverse prognosis then we would have clearly seen this reflected in mortality data given we know the high prevalence of occult thyroid carcinoma is not a new phenomenon. Thus the benign assessments of an earlier era (6, 7) cannot be discounted however much we feel concerned about the data on invasion and regional lymph node metastases. Indeed the study by Pellegriti et al (10) confirmed the generally indolent course of thyroid microcarcinoma, with distant metastases uncommon in the total group (8/299) and even more uncommon in incidentalomas (1/151), and showed no mortality over 3.8 years mean follow-up, admittedly a period too short for conclusive assessment. Of note in this study was that incidentalomas had less persistent or relapsing disease (6/115, 5.2%) compared to non-incidental tumours (7/67, 10.4%) supporting the view that the clinical presentation should be taken into account in management strategy.

Although we still lack a conclusive understanding of the clinical significance of the high rate of differentiated thyroid malignancy in impalpable nodules, and of the prognosis even if locally invasive, any tendency to an aggressive strategy should be tempered by the knowledge that screening programs for early cancers (lung, breast, prostate and neuroblastoma) have not demonstrated a mortality difference between screened and unscreened populations despite detecting more disease at an earlier stage and demonstrating longer survival in screened groups (21).

Despite the difficulty of acquiring this full understanding without randomised prospective trials of observation versus standard thyroid cancer care, with all the attendant practical and ethical problems, Ito et al (9) have provided a partial answer. In a prospective study of 732 patients with papillary microcarcinoma less than 10 mm of whom 162 chose observation [now extended to 211, (19)] after surgery was offered to those with various features considered unfavourable (high grade cytology, primary near the trachea, possible invasion of the recurrent laryngeal nerve, possible lateral lymph node invasion on ultrasonography) more than 70% of observed tumours did not grow in over 4 years mean observation and in those followed for over 6 years stability or shrinkage in the size of the tumour was seen in 75%. Only 11% became greater than 10 mm, and only 1.2% came to surgery for metastases to lateral lymph nodes (although 60 of the extended observation group eventually underwent operation for a variety of reasons). There were no distant metastases or deaths in any of the 211 (151 unoperated). In the surgical group lymph node metastases were found in 51% and multiple foci in 43%, and it was considered that these preoperatively unidentified features would have been present in the observed group at a similar frequency, thus suggesting that many patients with lymph node metastases and multifocality nevertheless have an excellent prognosis.

Conclusions
Pearce and Braverman (22) interpret the data from Pellegriti et al (10) to support the previous recommendation to observe incidentalomas under 10mm and perform FNB only if more than 10mm (2), on the grounds that there is no evidence that more rigorous management improves survival and it would impose a high burden on health care providers. Certainly the opportunity cost of imposing such an investigative regimen might be prohibitive even in the health care systems of the wealthy nations of North America and Europe, and would be quite impracticable and of very low priority in countries with developing economies. It is suggested that in line with these previous recommendations (2, 22) the clinician can counsel observation of an incidentaloma of 10 mm diameter or less. It is suggested that adverse ultrasonographic characteristics so far identified relate to the possible diagnosis of malignancy per se and not to whether the course will prove to be indolent or aggressive, with the possible exception of lateral compartment lymph node enlargement (18), and thus should not lead to an alteration of a management strategy based on size alone. Growth of an incidentaloma or appearance of lateral lymph node compartment enlargement should lead to USGFNA provided the degree of imprecision of ultrasonographic reassessment is taken into account. The diagnosis of papillary carcinoma in such lesions should lead to total thyroidectomy (10, 22, 23). In general, non-incidental lesions should be investigated by USGFNA in view of the evident less indolent course (10). Conversely current evidence strongly indicates there is no basis for population screening despite the likely high occult papillary microcarcinoma prevalence. Nor, in non-irradiated patients, is there a basis for case finding of incidentalomas in specific patient groups. Finally, it will be evident to the clinician that some patients will select a more aggressive approach and others a less aggressive approach than the clinician has recommended depending on individual attitudes to the possible diagnosis of cancer, to proposed follow-up regimens, and to neck surgery. This is appropriate provided the discussion of the problem is balanced, with an accurate account of the uncertainties involved.