Follicular nodules are the most common source of diagnostic difficulties in the practice of surgical pathology of the thyroid. This is due to a variety of factors, the most salient of which is the lack of well-defined criteria and evidence-based data for the diagnosis of these lesions.
To discuss some of the assumptions that have been accrued over the years regarding the criteria by which we evaluate such lesions.
The information presented herein is based on review of the literature and the author's personal experience.
Thyroid nodules with a predominant follicular growth pattern span the range from benign lesions (hyperplastic nodules, adenomatoid nodules, follicular adenomas) to malignant neoplasms (follicular carcinoma, follicular variant of papillary carcinoma) with a host of intermediate or indeterminate lesions found in between. Advances in immunohistochemistry and molecular pathology have not yet provided a reliable way of separating the borderline or intermediate cases. Low-grade and intermediate or borderline follicular-patterned thyroid lesions are those most often prone to difficulties for interpretation. Newer and potential future approaches for the evaluation of these lesions are discussed.
Follicular-patterned lesions are the most common type of surgical specimen from the thyroid, reviewed routinely by surgical pathologists. Most such cases represent benign conventional hyperplastic nodules. Problems arise when such nodules contain a fibrous capsule or are composed of a cell population or architectural growth pattern that is different from the surrounding normal thyroid. It has traditionally been accepted that benign-appearing follicular tumors of the thyroid that are completely surrounded by a fibrous capsule correspond to follicular adenomas, whereas well-circumscribed but unencapsulated tumors correspond to hyperplastic nodules. The problem is that hyperplastic nodules can quite often undergo regressive changes, including perinodular fibrosis, which can closely resemble a fibrous capsule. Distinction between an adenoma and a hyperplastic nodule with regressive changes can be very difficult under such circumstances. The problem is compounded when the regressive changes causing the fibrosis are not uniform and the deposition of fibrous tissue within and around the nodule is uneven creating irregularities of the “capsule,” often with entrapment of follicles at the periphery that can be easily interpreted as invasion of the capsule (Figure 1). Another type of follicular-patterned thyroid nodule that has been a source of controversy is one in which the follicles are lined by cells displaying nuclear features that are traditionally associated with papillary thyroid carcinoma. For many years, all such cases were regarded as examples of the follicular variant of papillary thyroid carcinoma (FVPTC). A recent study, however, demonstrated that tumors with such features that are noninvasive and completely encapsulated or sharply circumscribed behave in an indolent fashion and have been redesignated as noninvasive follicular tumors with papillary-like features.1 Finally, criteria for the diagnosis of well-differentiated, minimally invasive follicular carcinomas and their distinction from benign follicular nodules that may simulate an invasive malignancy have not yet been adequately delineated, leading to a proposal by the Chernobyl Group of Pathologists for applying the designation of “follicular tumor of undetermined malignant potential” for cases with borderline or equivocal features.2 This review will focus on 2 main areas of clinical significance: noninvasive follicular tumors with nuclear features suggestive of papillary carcinoma, and follicular tumors with borderline or equivocal features of malignancy.
Hyperplastic nodule with extensive regressive changes, including stromal hyalinization with pericapsular fibrosis and calcification. Note the focus of follicular structures within the capsule that can simulate capsular and/or vascular invasion (hematoxylin-eosin, original magnification ×20).
Figure 2 Scanning magnification of noninvasive follicular tumors with papillary-like features showing a well-circumscribed and sharply demarcated follicular nodule completely surrounded by a fibrous capsule (hematoxylin-eosin, original magnification ×10).
Figure 3 Higher magnification of the nodule in Figure 2 showing follicular structures lined by cells displaying the nuclear features of papillary thyroid carcinoma, corresponding to a grade 2 nuclear feature in the Nikiforov schema (hematoxylin-eosin, original magnification ×600).
Figure 4 Example of an encapsulated, noninvasive follicular nodule that on higher magnification showed a few scattered small foci with the nuclear features of papillary thyroid carcinoma, but most of the nodule was composed of otherwise normal-appearing follicular cells (hematoxylin-eosin, original magnification ×400).
Figure 5 Encapsulated follicular nodule showing portion of the tumor partially infiltrating the capsule but without complete, through-and-through penetration or infiltration of surrounding tissues (hematoxylin-eosin, original magnification ×200).
Figure 6 Another example of pseudocapsular invasion in an encapsulated adenomatoid nodule shows islands of follicular structures embedded within the capsule but without complete penetration (hematoxylin-eosin, original magnification ×200).
Hyperplastic nodule with extensive regressive changes, including stromal hyalinization with pericapsular fibrosis and calcification. Note the focus of follicular structures within the capsule that can simulate capsular and/or vascular invasion (hematoxylin-eosin, original magnification ×20).
Figure 2 Scanning magnification of noninvasive follicular tumors with papillary-like features showing a well-circumscribed and sharply demarcated follicular nodule completely surrounded by a fibrous capsule (hematoxylin-eosin, original magnification ×10).
Figure 3 Higher magnification of the nodule in Figure 2 showing follicular structures lined by cells displaying the nuclear features of papillary thyroid carcinoma, corresponding to a grade 2 nuclear feature in the Nikiforov schema (hematoxylin-eosin, original magnification ×600).
Figure 4 Example of an encapsulated, noninvasive follicular nodule that on higher magnification showed a few scattered small foci with the nuclear features of papillary thyroid carcinoma, but most of the nodule was composed of otherwise normal-appearing follicular cells (hematoxylin-eosin, original magnification ×400).
Figure 5 Encapsulated follicular nodule showing portion of the tumor partially infiltrating the capsule but without complete, through-and-through penetration or infiltration of surrounding tissues (hematoxylin-eosin, original magnification ×200).
Figure 6 Another example of pseudocapsular invasion in an encapsulated adenomatoid nodule shows islands of follicular structures embedded within the capsule but without complete penetration (hematoxylin-eosin, original magnification ×200).
NONINVASIVE FOLLICULAR THYROID NEOPLASM WITH PAPILLARY-LIKE NUCLEAR FEATURES
The encapsulated form of the FVPTC has long been suspected of having a different nature than the tumors showing features of FVPTC that were invasive. Chan3 in 2002 proposed a set of stringent criteria by which to make the diagnosis of FVPTC and pointed out in his review of the literature that the encapsulated types behaved as indolent tumors, with only 1 case recorded that was associated with death of the patient due to tumor.4 A subsequent large study from Italy showed that encapsulated well-differentiated follicular-patterned thyroid carcinomas were not associated with any fatality rates on long-term follow-up.5 This and other findings led to a growing suspicion that such tumors need not be approached as a malignant neoplasm and stimulated an international group of endocrine pathologists to undertake a formal study of this problem, with the conclusion that such tumors observe an indolent behavior.1 From their observations, they proposed a change in nomenclature for the encapsulated variant of FVPTC to noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP).1 This iconoclastic approach has generated much interest in the general public and the lay press6 and has now been widely embraced by the pathology community and clinicians. The criteria for making this diagnosis as spelled out in the original study by Nikiforov et al1 included the following: (1) encapsulation or clear demarcation; (2) follicular growth pattern with less than 1% papillae, no psammoma bodies, and no more than 30% solid/trabecular/insular growth pattern; (3) nuclear score 2 to 3; (4) no vascular or capsular invasion; (5) no tumor necrosis; and (6) no high mitotic activity (defined as at least 3 mitoses per 10 high-power fields) (Figures 2 and 3). Since the publication of this seminal study, a wealth of literature has appeared in recent years related to this topic, proposing minor changes to the definition of these tumors and addressing the potential impact of this change in cytopathology interpretation of thyroid fine-needle aspirates.7–9 Most authors are in agreement that cytologic features cannot reliably distinguish NIFTP from conventional papillary thyroid carcinoma; the diagnosis thus requires complete excision and evaluation of the capsule for invasion.10–12 For this reason, the diagnosis of papillary thyroid carcinoma, based on cytology, needs to be approached with more caution since the introduction of NIFTP. Most cases of NIFTP are classified into one of the indeterminate categories in the Bethesda System for Reporting Thyroid Cytopathology, since assignment of these tumors to a malignant category would now constitute overdiagnosis. The real impact for patient management, however, may be minimal given that lobectomy represents the generally accepted treatment for tumors diagnosed from aspirates in the Bethesda indeterminate categories, which is also the recommended treatment for NIFTP.8 More significant has been the realization that some reported cases that qualified for a diagnosis of NIFTP have been associated with regional lymph node metastases, as well as BRAFV600E mutations or high-risk TERT promoter mutations.13,14 This has led to the recent proposal by Nikiforov et al15 to change one of the diagnostic criteria for NIFTP from “less than 1% papillae” to “no well-formed papillae,” given that the cases that metastasized displayed focally papillary architecture. Moreover, the recommendation was made that the entire tumor should be processed for histology to rule out the presence of any papillary architecture. In the original publication, Nikiforov and associates1 also introduced a novel nuclear score for evaluating the nuclear features of NIFTP; a score of 2 to 3 was required to make the diagnosis.1 In a more recent publication, it was acknowledged that NIFTP more often falls within a nuclear score of 2/3, making tumors that show a nuclear score of 3 more highly suspicious for being a papillary thyroid carcinoma.15 An issue that has not been addressed yet, however, is the occasional occurrence of tumors that display only focal nuclear features of papillary thyroid carcinoma. This issue has been addressed previously for well-circumscribed follicular tumors of the thyroid but has not been assessed in tumors otherwise qualifying for the diagnosis of NIFTP.16 The question thus arises as to whether a tumor that shows a few microscopic foci containing follicles with the nuclear features of papillary carcinoma, but which is otherwise well-circumscribed and/or encapsulated, should be regarded as NIFTP (Figure 4). Another issue that remains somewhat arbitrary is the issue of what constitutes “clear demarcation” in these tumors. Irregularities at the periphery of thyroid nodules, with small islands of entrapped follicles distributed parallel to the outer surface of the tumor, can sometimes be observed in a variety of follicular neoplasms. In the absence of active penetration of the capsule by tumor, these entrapped islands of thyroid elements can be difficult to evaluate and may trigger a diagnosis of malignancy. The question of what constitutes capsular invasion in NIFTP thus remains a subject of debate. The criteria for diagnosing capsular invasion in follicular neoplasms in general are still not well defined and subject to variable interpretation based on the experience of the observer. While many require through-and-through infiltration of the capsule with obvious penetration by tumor into the surrounding tissues, others consider any type of infiltration into the capsule as diagnostic of capsular invasion. As in the case of follicular neoplasms (see below), what constitutes evidence of capsular invasion may vary among observers. Despite these gaps in our understanding of these lesions, the introduction of the NIFTP terminology has been a strong step in the right direction for the interpretation of follicular thyroid nodules. The introduction of this category is likely responsible for sparing countless patients the fear and stigma of a diagnosis of cancer, reducing the morbidity of unnecessary aggressive cancer treatment, and saving the system significant amounts of money. Future refinements in diagnostic criteria and increased experience with these lesions will hopefully lead to better stratification and improved outcomes for these patients.
MINIMALLY INVASIVE FOLLICULAR CARCINOMA
Perhaps the most significant challenge today for the diagnosis of thyroid follicular neoplasms is the separation of the so-called minimally invasive follicular carcinoma from benign follicular tumors. Fear of missing a diagnosis of cancer has led to the widespread tendency for overdiagnosing follicular lesions of the thyroid as minimally invasive follicular carcinoma by both practicing general pathologists and expert consultants.
The current World Health Organization (WHO) classification of thyroid neoplasms acknowledges the existence of 3 variants of follicular thyroid carcinoma: minimally invasive, widely invasive, and encapsulated angioinvasive.17 The minimally invasive tumors are defined as those displaying limited capsular and/or vascular invasion, while widely invasive tumors are defined as having widespread infiltration of adjacent thyroid tissue and/or blood vessels. A precise definition of what constitutes capsular invasion, however, is still lacking. This is acknowledged in the WHO monogram, where it is stated that lesions that show as their only finding irregular contours of the inner capsular border, capsular pushing by the tumor cells, or tumor cell nests embedded in the capsule are considered by many authors as insufficient for a diagnosis of malignancy (Figures 5 and 6). They also state that when neoplastic cells penetrate the adjacent normal thyroid, a stromal reaction may be elicited forming a secondary limiting fibrous band, but no mention is made on how to distinguish such fibrous stromal reaction from incomplete penetration of the capsule by tumor, or how to distinguish small nodules with similar cell composition outside the lesion from separate and independent hyperplastic or adenomatoid nodules rather than invasion. Partial but incomplete penetration of the capsule by tumor has been widely regarded by some as being sufficient to qualify for invasion, and therefore for a diagnosis of carcinoma18,19 ; however, a consensus regarding the significance of such finding is not yet available. Other authors20–22 consider nests of tumor cells in the capsule as representing entrapped follicles and require complete penetration of the capsule by the tumor for a diagnosis of carcinoma. The assessment of capsular invasion thus remains a contentious area in the evaluation of these tumors.
One of the general assumptions that has been made regarding the capsule in follicular carcinoma is that it is often thicker than in adenomas and heavily calcified. This observation stems back to a study by Evans23 in 1984 in which he noticed that 3 of the cases regarded by him as examples of “encapsulated follicular carcinoma” that had thick and heavily calcified capsules presented with bone metastases at the time of the initial diagnosis. Such cases did not show vascular invasion or invasion outside the capsule. The rarity of this occurrence begs the question of adequacy of the sampling and whether incomplete sampling may have been responsible for missing areas of vascular invasion in those tumors; after all, the modern guidelines for complete sampling of the capsule in suspicious follicular lesions was not yet in place in 1984. Yet, the presumption still prevails that thyroid follicular nodules with thick calcified capsules showing infiltration of tumor cells into the capsule, but not beyond the capsule, are malignant. How to distinguish such cases from adenomatoid or hyperplastic nodules with severe regressive changes that develop thick capsules with entrapment of follicles and heavy calcification has not yet been properly addressed.
Another significant issue for the definition and evaluation of minimally invasive follicular thyroid lesions is whether capsular invasion, by itself, is sufficient to qualify a lesion for the diagnosis of carcinoma. It has been generally acknowledged that the minimally invasive tumors observe a very indolent behavior, while the widely and angioinvasive tumors usually show a propensity for distant metastases. This was first highlighted in a study by Van Heerden et al24 from the Mayo Clinic, in which 65 cases that had been diagnosed as minimally invasive follicular thyroid carcinoma were followed up for a median period of 11 years. The authors found that the cases that had been diagnosed as follicular carcinoma on the basis of capsular invasion alone did not result in either distant metastases or cancer-related deaths, while those cases in which vascular invasion was demonstrated showed distant metastases in 19% and tumor-related patient deaths in 28% of cases. Given that this contradicted prevailing dogma, they cautiously titled their article “Follicular Thyroid Carcinoma With Capsular Invasion Alone: A Nonthreatening Malignancy.” This issue has not been seriously studied and the 2 largest studies published since then addressing this topic have yielded conflicting results. Thompson et al18 in 2001 studied 27 patients with tumors showing only capsular invasion from a series of 95 patients diagnosed with minimally invasive follicular thyroid carcinoma and found no evidence of recurrence or metastases or patient deaths in this group during a mean follow-up period of 16.5 years. Surprisingly, despite these findings they stated the opinion that “capsular invasion alone is sufficient for a diagnosis of minimally invasive follicular carcinoma.” A subsequent study by D'Avanzo et al25 in 2004 in a series of 132 patients with follicular thyroid carcinoma found that among 45 patients who had tumors with capsular invasion, only 6 patients developed recurrences and 7 patients died. In their study, patients who had angioinvasion with or without capsular invasion had a less favorable prognosis than patients who showed capsular invasion only. The 5-year survival for their patients was 98% for minimally invasive follicular carcinoma with capsular invasion only, 80% for minimally invasive follicular carcinomas with vascular invasion, and 38% for widely invasive tumors. The authors proposed reclassifying these tumors as minimally invasive when there was only capsular invasion, moderately invasive for tumors with angioinvasion with or without capsular invasion, and widely invasive for tumors that invade directly into adjacent muscle and or thyroid parenchyma. Although the source of the patient population and the dates of accession for the various cases in their study was not stated, given the retrospective nature of the study it is safe to assume that the sampling of the tumors was not uniform and likely incomplete in some cases, casting doubt on whether the cases designated as having capsular invasion only may have not also contained foci of vascular invasion that were not sampled, thus accounting for their behavior. The most recent study on follicular-patterned thyroid carcinomas by Piana et al,5 which included all types of follicular carcinomas, papillary thyroid carcinomas with follicular pattern of growth, and other thyroid tumors, contained 29 cases of minimally invasive follicular carcinoma of which 23 showed capsular invasion only and 6 showed vascular invasion; none of the patients with either capsular invasion only or capsular and vascular invasion died of their tumors. It is hazardous to speculate on this topic given that we do not really have all the information at hand, but the fact remains that we do not have ample data on which to base our assumptions and very few studies have been conducted on this topic, using stringent criteria for diagnosis and using material that has been properly sampled with adequate modern grossing protocols. A multicenter study of adequately sampled and diagnosed minimally invasive follicular thyroid carcinoma is urgently needed to reevaluate this issue and may well lead to a similar conclusion derived from the study by Nikiforov et al1 on the encapsulated variant of FVPTC, namely, that we have been overdiagnosing some of these tumors for many years with the consequent adverse effects for patients and its attendant negative economic impact on our health system. It could well be that a properly conducted study may lead to reclassification of minimally invasive follicular thyroid carcinoma with only capsular invasion to “invasive follicular tumor with indolent behavior.”
THYROID FOLLICULAR TUMOR OF UNCERTAIN MALIGNANT POTENTIAL
Which brings us to the problem of thyroid follicular tumors for which it is difficult to determine with certainty their malignant potential. In 2000, an international group of pathologists met to review several hundred cases of thyroid tumors from patients who had survived the Chernobyl nuclear accident and they reported some of the difficulties they had encountered for proper classification of those lesions in an editorial titled: “Two Proposals Regarding the Terminology of Thyroid Tumors.”2 They indicated that their greatest diagnostic difficulties were related to encapsulated tumors with a follicular architecture. They delineated 2 common scenarios: (1) deciding whether minor nuclear changes of the type seen in papillary thyroid carcinoma justify a diagnosis of follicular variant of papillary thyroid carcinoma, and (2) deciding whether minor degrees of capsular penetration justify a diagnosis of malignancy. Another issue they addressed was that of encapsulated tumors lacking papillary carcinoma-like nuclear features or vascular invasion for which genuine doubt existed about capsular penetration. The authors proposed the term follicular tumor of uncertain malignant potential for such cases. In their definitions they proposed 2 diagnostic categories: “well-differentiated tumor of uncertain malignant potential” for tumors that were encapsulated and showed questionable papillary carcinoma-type nuclei, no vascular invasion, and absent or questionable capsular invasion; and the previously mentioned “follicular tumor of uncertain malignant potential” for tumors with similar features but that were unquestionably of follicular type without any suggestion of papillary carcinoma-like nuclear changes. The former category can show some overlap with NIFTP, particularly for the tumors with low nuclear scores of 1 to 2 in the proposal of Nikiforov et al.1 The latter category overlaps with the group of tumors that falls within the minimally invasive follicular carcinoma without angioinvasion category, in which capsular invasion is questionable or incomplete (ie, without through-and-through penetration of the capsule into surrounding tissues), or for tumors with minimal but convincing capsular invasion that lack evidence of vascular invasion. I have found these categories helpful in my practice and use them routinely for borderline tumors or for cases with equivocal findings. The authors of the editorial made the disclaimer that these terms should not be used as a substitute for adequate capsular sampling or to avoid the intellectual commitment for making a clear distinction between benignity versus malignancy. Although in the past, surgeons have objected to pathologists rendering equivocal diagnoses, the fact remains that there are still areas in pathology where we do not yet have definitive answers. Intellectual and professional honesty demand that we acknowledge such instances and indicate in our report our level of uncertainty regarding the diagnosis. Borderline or indeterminate lesions are a fact of life in all areas in surgical pathology and across all organ systems; there is no good rationale or justification for not acknowledging them in thyroid pathology. Until we have better studies or more reliable tools for unequivocally separating benign from malignant thyroid follicular nodules, adoption of this system of terminology may avoid unnecessary overtreatment and the attendant psychological and financial burden associated with a diagnosis of cancer. Given that the past literature and experience appear to indicate that lesions falling within these categories are unlikely to behave aggressively, adopting this terminology appears to be the lesser of the evils.