The criteria for minimally invasive (low grade) follicular carcinoma of the thyroid (MI) remain controversial, often resulting in unnecessary treatment.
The criteria for minimally invasive (low grade) follicular carcinoma of the thyroid (MI) remain controversial, often resulting in unnecessary treatment.
The records of 130 patients with minimally invasive (MI) follicular thyroid carcinoma were retrieved from the files of the Endocrine Tumor Registry of the Armed Forces Institute of Pathology.
Ninety-five patients were confirmed to have MI based on the authors' criteria of small-to-medium vessel invasion, capsular invasion of up to full thickness, no parenchymal tumor extension, and no tumor necrosis (patients with oxyphilic tumors were excluded). The remaining 35 patients had tumors that were reclassified as “not low grade” based on large vessel invasion, extension into parenchyma, and tumor necrosis (oxyphilic cases excluded). The MI patients included 67 women and 28 men, ages 20–95 years (average, 42.0 years). Nearly all patients presented with a thyroid mass (n = 90 patients). The mean tumor size was 2.8 cm. Histologic features examined for tumor classification included cellularity, capsule nature, capsular invasion, vascular invasion, extension into parenchyma, cytoplasmic oxyphilia, mitotic activity, and necrosis. All patients were treated with surgical excision. Adjuvant radioactive iodine therapy was performed in 24 patients. Five patients developed recurrent disease: four were alive or had died without evidence of disease after additional treatment (mean, 18.1 years), and one patient died with disease (MI tumor) at 15.1 years. All of the remaining patients were disease free (mean follow-up, 16.5 years).
There are reproducible histologic criteria to diagnose patients with MI follicular carcinoma. The overall excellent long term prognosis and a good patient outcome suggests that no additional surgery is necessary. Cancer 2001;91:505–24. © 2001 American Cancer Society.
Follicular carcinoma of the thyroid gland accounts for between 4% and 39% of all malignant thyroid tumors. Follicular carcinoma is more common in women who are affected over a broad age range, peaking in the 5th and 6th decades.1–16 Follicular carcinoma has been divided further based on the criteria set forth by the World Health Organization (WHO)17 into encapsulated (low grade, minimally invasive [MI]), widely invasive (high grade, poorly differentiated), oxyphilic (oncocytic, Askanazy cell), and other.10, 18, 19 The histologic criteria for separation into these various grades are not well developed, and it is no wonder that the prognostic implications for using these terms are unreliable. We believe that the term minimally invasive follicular carcinoma (encapsulated, low grade follicular carcinoma) should be strictly applied to a specific tumor that meets specific histomorphologic criteria. Therefore, we undertook a study of 95 MI follicular carcinomas to establish definitive clinical and histopathologic characteristics or criteria for the diagnosis of patients with MI (low grade) follicular carcinoma in a single, comprehensive study. We believe that our review of patients with MI follicular carcinoma in a reproducible fashion by the use of clinical features, histomorphologic criteria, treatment modalities, and patient outcome analysis allows us to make reasonable recommendations concerning the management of patients with MI follicular carcinoma.
The records of 188 patients diagnosed with MI follicular thyroid carcinomas were reviewed from the files of the Endocrine Tumor Registry at the Armed Forces Institute of Pathology from 1974 to 1983. These 188 patients were identified in a review of 6501 patients (2.89%) with benign or malignant primary thyroid tumors who were seen in consultation during this time. Complete clinical follow-up was absolutely necessary for inclusion in this study, and follow-up was available for only 130 patients, who formed the basis of the study. Eighty-one patients were from civilian sources, including university medical centers and foreign contributors, 38 patients were from military hospitals, and 11 patients were from Veterans Administration medical centers.
Materials within the Institute's files were supplemented by a review of the patient demographics (gender, age, and ethnicity); symptoms at presentation (including duration); medical history; laboratory values; radiographic studies; physical findings; surgical pathology reports; operative reports; and cancer registry records; by specific questionnaires or by direct communication with the physician or the patient. This clinical investigation was conducted in accordance and compliance with all statutes, directives, and guidelines of the Code of Federal Regulations (Title 45, Part 46) and the Department of Defense (Directive 3216.2) relating to human subjects in research.
Hematoxylin and eosin-stained slides from all patients were reviewed. We used the histologic criteria for follicular carcinoma as defined by the WHO17 as a “guide” to classification, although the WHO classification criteria for separation into various grades of follicular carcinoma are vague and are not reproducible. Therefore, for a better understanding of these carcinomas, we used histologic criteria alone to separate the tumors into low grade/MI follicular carcinoma and “not low grade” (NLG) follicular carcinoma. MI follicular carcinoma was defined by capsular invasion up to the full extent of the capsule (single or multifocal), small or medium vessel invasion, parenchymal extension if accompanied by fibrous capsular tissue, and a lack of necrosis (in nonoxyphilic tumors). Although the term “NLG” may be cumbersome and imperfect, we use this term to specifically exclude “high grade” follicular carcinoma from consideration. Furthermore, we defined NLG as a follicular carcinoma that demonstrates complete and multifocal areas of capsular invasion, vascular invasion of any kind, but particularly of large vessel(s), “unaccompanied” parenchymal extension, and/or tumor necrosis in nonoxyphilic tumors. Unaccompanied parenchymal extension is defined as tumors cells in direct contact with uninvolved thyroid parenchymal cells without any intervening fibrous capsular tissue. It is well known that oxyphilic tumors can undergo ischemic tumor necrosis. However, we annotated tumor necrosis only when it was identified in nonoxyphilic tumors. None of the 130 patients who were included in this study had follicular variant of papillary carcinoma, and, although all cases of “low grade follicular carcinoma” between 1974 and 1983 at the Armed Forces Institute of Pathology were reviewed initially, 35 patients had their tumors reclassified into the “NLG” category upon application of the criteria set forth above.
The number of histologic observations was recorded for each tumor as follows: The tumors were surrounded by a variable fibrous capsule, defined by the following: 1) a thick and well formed, clearly visible capsule on macroscopic examination of the specimen or the glass slide; 2) thin but well circumscribed, in which there was an aggregation of fibrous connective tissue, easily recognized as such by microscopic examination, although thinned in areas; and 3) irregular, uneven, and poorly formed fibrous capsule demonstrating areas of attenuation and, in foci, complete loss, in addition to areas of reactive connective tissue deposition, frequently related to fine-needle aspiration (FNA). An elastic stain (Musto) frequently was helpful in delineating the capsule in tumors within the latter category.
Capsular invasion was classified by the degree of invasion (< 50% or > 50% of the thickness of the capsule) and by single or multiple foci (Fig. 1A,B). If nests of neoplastic cells were found in the fibrous capsule outside the confines of the tumor, then it was classified as evidence of invasion. Accompanied parenchymal extension was defined by a protrusion of tumor cells outside the capsule (contour of the tumor), with uninvolved thyroid parenchyma identified on either side of the protrusion but still with fibrous connective tissue present (Fig. 1C). Unaccompanied parenchymal extension was defined as tumor cells in direct contact with uninvolved thyroid parenchyma without any intervening fibrous capsular tissue. The distinction of accompanied versus unaccompanied parenchymal extension was used in our classification to separate between MI tumors and NLG tumors.
Vascular invasion was defined by the size of the vessel as follows: 1) small, 2) medium, or 3) large, and it was also separated into single or multiple foci (Fig. 2). Small vessels usually were located within the capsule and were of a limited caliber (Fig. 2A–C). Medium-sized vessels without a smooth muscle wall often were located immediately adjacent to the capsule (Fig. 2D,E). Large-caliber vessels with smooth muscle in the vessel wall, although they were present at the capsule, usually were distant from the capsule (Fig. 2F). No distinction was made between veins, arteries, or lymphatic channels for purposes of this study, and no histochemical or immunohistochemical studies were performed in an attempt to identify vascular structures. Vascular invasion was identified by direct extension into the vessel lumen, intravascular attached tumor thrombi, and/or tumor nests covered by endothelium (Fig. 2).10, 19 We did not require that the exact point of penetration into a vessel be recognized to qualify as evidence of vascular invasion. Tumor plugs in vascular spaces within the tumor mass did not qualify as vascular invasion and had no impact on diagnosis or, thus, patient prognosis.
Cellularity was divided into three groups that were defined as follows: 1) low cellularity, i.e., with abundant colloid and a pattern of growth simulating an adenomatoid nodule but still with evidence of invasion (Fig. 2C,E); 2) moderate cellularity, i.e., colloid was identified easily within the center of follicular structures (Fig. 1A,B); and 3) high cellularity, i.e., with limited colloid and compact cell growth, often requiring a periodic acid-Schiff (PAS) stain to accentuate the colloid (Fig. 2A,F). The pattern of growth also was noted and included insular (Fig. 3), trabecular, and solid patterns.
Cytoplasmic oxyphilia or eosinophilia (“Askanazy cell” change) was defined by a fine to slightly coarse granularity of the abundant cytoplasm surrounding typical to variably atypical nuclei with frequent prominent eosinophilic nucleoli (Fig. 4). Mitotic figures were sought by five pathologists independently, each of whom examined at least five high-power fields (magnification, ×40 with a ×10 objective lens), and were graded as present (n = 37 tumors) or absent (n = 93 tumors), without an attempt at quantitation. It is noteworthy that, when mitoses were present, they usually were identified easily.
Tumor necrosis was defined as areas of cell death not associated with areas of degeneration or previous FNA and usually was found in the more central regions of the tumor mass (Fig. 5). Tumor necrosis was not graded in oxyphilic tumors because they frequently have tumor degeneration not related to true tumoral necrosis. The oxyphilic cells are known to have a low oxygen tension and thus, are more susceptible to nutrient deprivation than other follicular tumors.
The presence of other thyroid lesions, including adenomatoid nodules, diffuse hyperplasia, lymphocytic thyroiditis, and other tumors was noted in the thyroid parenchyma both macroscopically and microscopically.
At least one section per centimeter of tumor was required, with additional tissue submitted for a more complete embedding of the tumor capsule to thyroid parenchyma interface. In a number of tumors, the diagnostic areas of invasion were identified in the additional sections.
Formalin fixed, paraffin embedded sections were stained with PAS (with and without diastase digestion) to accentuated the thyroid colloid in high cellularity cases. An elastic stain (Musto) was employed to further define the tumor capsule, with black fibers oriented parallel to the fibrous connective tissue capsule taken to represent a true capsule rather than a green to beige background connective tissue.
Our review of the literature was extensive. The review was based on a MEDLINE research from 1966 to 1999, with a few specific, earlier articles included for balance and background. However, for purposes of succinctness, we used only research papers that included at least 30 cases of follicular carcinoma in their analyses and that were written in English.
Categoric variables were analyzed using chi-square tests to compare observed and expected frequency distributions. Comparison of means between groups were made with unpaired t tests or one-way analysis of variance, depending on whether there were two groups or more than two groups, respectively. Multiple comparisons were analyzed using the Tukey method. Linear regression was used to investigate two measured variables, and Pearson correlation coefficients were generated to measure the strength of the association. Confidence intervals of 95% were generated for all positive findings. The α level was set at P < 0.05. All analyses were conducted using SPSS software (version 8.0 for PC; SPSS, Inc., Chicago, IL).
A summary of the clinical information on the patients in this series is provided in Table 1. The patients included 67 females and 28 males, with a female to male ratio of 2.4:1.0. Patient age ranged from 20 years to 95 years (Table 1), with an average age of 42.0 years at the time of presentation. Sixty patients were white, 4 patients were African American, 6 patients were Asian or Indian, and 25 patients were of unspecified race.
|Characteristic||Total (n = 95 patients)|
The patients experienced a mass in the thyroid gland (n = 60 patients), an enlarging mass (n = 22 patients), or a diffuse enlargement of the gland that clinically simulated a goiter (n = 8 patients) (Table 2). Five patients presented during routine physical examination and were considered to be asymptomatic. Two patients presented clinically with Graves disease. The duration of symptoms ranged from 14 days to as long as 10 years, with a mean duration of 19.8 months.
|Signs and symptoms|
|Duration of symptoms (months)|
Radiologic procedures were performed on 36 patients, all of whom demonstrated a cold nodule on radioactive iodine uptake studies. Although many patients may have had an uptake study for a discrete nodule clinically, information regarding radiologic studies was not available in the charts or we were unable to review the charts ourselves, and it is not included here.
Six patients had a history of previous irradiation or radiation exposure, most in childhood for thymic enlargement, although one patient had exposure to radiation from the atomic bomb. Three patients were treated with lobectomy and isthmusectomy, whereas three patients were treated with complete thyroidectomy. Four patients received postoperative radioiodine 131 (131I) treatment. All of these patients were either alive (n = 5 patients) or had died of unrelated causes (n = 1 patient) without evidence of disease at last follow-up (average, 17.7 years).
Most of the tumors (67%) were located in the right lobe of the thyroid, but this finding was not statistically significant (P = 0.145) (Table 3). Tumors were found in the isthmus in only two patients.
|Tumor size (cm)|
|Average no. of sections examined||18.0|
|Average no. of slides examined||19.0|
The tumors ranged in size from 0.7 cm to 9.0 cm, with an average size of 2.8 cm. It is significant to note that tumors measuring > 3.0 cm in greatest dimension were more likely to occur in older patients (P = 0.042).
The tumors were encapsulated by a variable fibrous capsule that was distinguished easily from the surrounding, uninvolved thyroid parenchyma. The tumors were described as solid to focally degenerated and contained tissue of a consistency similar to that of thyroid parenchyma; in general, the macroscopic findings were nonspecific. The surrounding thyroid parenchyma was described in the surgical pathology reports as atrophic, nodular, or unremarkable, depending on the lesions affecting the remaining thyroid parenchyma.
A summary of the microscopic characteristics and features is presented in Table 4. Tumors were surrounded by a fibrous capsule, ranging from thick and well formed (n = 59 tumors) to thin and circumscribed (n = 20 tumors), whereas the remaining tumors demonstrated an irregular, uneven, and poorly formed capsule (n = 16 tumors) (Figs. 1, 2).
|Characteristic||No. of tumors|
|Thick and well formed||59|
|Irregular, uneven, and poorly formed||16|
|Mitotic figures (present)||20|
|< 50% Thickness (single focus)||21|
|< 50% Thickness (multiple foci)||55|
|> 50% Thickness (single focus)||20|
|> 50% Thickness (multiple foci)||51|
|Small/medium vessels (single focus)||14|
|Small/medium vessels (multiple foci)||54|
|Hyperplasia, adenoma, medullary carcinoma||4|
|No other disease||31|
Capsular invasion was graded as < 50% (Fig. 1A) or > 50% (Fig. 1B) invasion of the thickness of the capsule and was separated into invasion with a single focus or with multiple foci (Table 4). When the capsule was thin or irregular, the Musto stain (elastic stain) accentuated the capsule, making a determination of invasion easier. In addition, vascular invasion was identified in either small vessels (Fig. 2A–C), medium vessels (Fig. 2D,E), or large vessels (Fig. 2F), as defined in Materials and Methods, and also was separated into invasion with a single focus or with multiple foci. All tumors demonstrated small-to-medium vessel invasion and had evidence of single or multiple areas of capsular invasion (< 50% or > 50%), with the following notable exceptions: 3 patients had small-to-medium vessel invasion with no areas of capsular invasion or parenchymal extension (all were alive without evidence of disease); 25 patients demonstrated multifocal, > 50% capsular invasion without vascular invasion (all patients were without evidence of disease); and 2 patients demonstrated only multifocal, < 50% capsular invasion without vascular invasion (both patients were alive without evidence of disease). The seven patients with extension into the parenchyma all had tumors accompanied by capsular fibrous tissue intervening between the tumor cells and the surrounding parenchyma.
The tumor cells were arranged in follicular structures of variable size. The degree of cellularity was divided into three groups, as defined in Materials and Methods section (see above). The 14 tumors that demonstrated low cellularity with abundant colloid simulated adenomatoid nodules (Fig. 2C,E). Areas of capsular or vascular invasion still were identified to include these low cellularity tumors as carcinomas. A similar number of tumors demonstrated moderate cellularity (Fig. 1A,B) and high cellularity (Figs. 2A,F, 3). A variety of patterns of growth were present in the tumors, including trabecular (Fig. 3), insular, and solid patterns. Colloid was accentuated in the high cellularity tumors with a pink-to-reddish PAS positive material in the lumen of the follicle.
Cytoplasmic oxyphilia or eosinophilia (“Askanazy cell” change) (Fig, 4) was identified in 37 tumors (39%) (Table 4). Mitotic figures were sought by five pathologists and were graded as present (n = 20 tumors) or absent (n = 75 tumors) (Table 4). Tumor necrosis was not identified in any case, except in two tumors with cytoplasmic oxyphilia. In these two tumors, it was focal and was identified in the central portion of the tumor mass (Fig. 5), possibly related to the low oxygen tension in oxyphilic tumors or to FNA. The remaining thyroid parenchyma revealed lymphocytic thyroiditis, adenomatoid nodules, and papillary thyroid carcinoma along with other disorders (Table 4).
Multiple sections were needed to demonstrate capsular or vascular invasion, with an average of 18.0 sections examined per tumor, irrespective of the tumor size. This equates to approximately 6.4 sections per centimeter of tumor. Although this is a generous sampling and is not intended to suggest the type of sampling required in these tumors, we wanted to decrease the chance of overlooking a focus of invasive tumor (Table 3).
Although complete surgical excision was the treatment of choice, we did not explore in great detail the various therapeutic alternatives used because that was not the purpose of this investigation (Table 5). However, most patients underwent lobectomy (n = 50 patients), 15 patients underwent lobectomy and isthmusectomy, 16 patients underwent subtotal thyroidectomy, and only 14 patients underwent total thyroidectomy. A total of 21 patients received postoperative 131I therapy. One patient received postoperative chemotherapy. The type of surgical intervention did not change the overall patient outcome (P = 0.160).
|Treatment||All patients (n = 130)||Minimally invasive (n = 95 patients)||“Not low grade” (n = 35 patients)|
|Followed by completion thyroidectomy||10||8||2|
|Lobectomy and isthmusectomy||25||15||10|
|Followed by completion thyroidectomy||3||3||0|
|Followed by completion thyroidectomy||1||1||0|
|131I therapy (with or without thyroidectomy)||26||21||5|
|Lobectomy followed by 131I||13||9||4|
|Thyroidectomy followed by 131I||13||12||1|
Survival was excellent overall (Table 6). Of the 95 patients with MI tumors, 94 patients were alive at last follow-up or had died without evidence of disease (mean, 16.8 years). Four patients developed recurrent disease or a second primary thyroid tumor during follow-up, one of whom died with disease. A detailed description of the patients (including the patient with NLG tumor) who developed recurrence or secondary thyroid tumors is presented in Table 7. It is interesting to note that the secondary thyroid tumor or recurrence developed in the contralateral lobe in three of the patients, which seems to suggest the development of a second primary rather than a true recurrence. A disease recurrence would be more likely to appear on the same side as the resection, as it did in two patients. The second tumor developed an average of 8.6 years after the initial presentation. Furthermore, none of the patients initially underwent total thyroidectomy; instead, all underwent either lobectomy (n = 4 patients) or lobectomy and isthmusectomy (n = 1 patient). Four of the patients received 131I treatment with the development of the second tumor. With the exception of the single patient in the study who died from her disease, all of the other patients were alive or dead without evidence of disease at the last follow-up (mean, 18.1 years). Therefore, tumor recurrence or metastatic disease must be used as a reference rather than just 5-year or 10-year survival results.
|Outcome||All patients (n = 130)||Minimally invasive (n = 95 patients)||“Not low grade” (n = 35 patients)|
|Alive, no evidence of disease|
|No. (%)||102 (78.5)||78 (82.1)||24 (68.5)|
|Average follow-ups (yrs)||18.2||17.9||18.9|
|Dead, no evidence of disease|
|No. (%)||27 (20.8)||16 (16.8)||11 (31.4)|
|Average survival (yrs)||10.8||11.2||10.6|
|Dead, recurrent or metastatic disease|
|No. (%)||1 (0.7)||1 (1.1)||0|
|Average survival (yrs)||15.1||15.1||—|
|Developed recurrent disease|
|No. (%)||5 (3.4)||4 (4.2)||1 (2.9)|
|Average survival (yrs)||17.5||18.1||15.1|
|Characteristic||Patient 1||Patient 2||Patient 3||Patient 4||Patient 5|
|Age at presentation (yrs)||59||18||49||60||52|
|Size (in cm)||3.0||1.5||7.0||2.5||4.0|
|Side of original tumor||R||R||R||L||L|
|Tumor type (MI or NLG)||MI||MI||MI||MI||NLG|
|Treatment at initial presentation||L, I||L||L||L||L|
|Time interval to recurrence (yrs)||7||10||8||10||7|
|Side of recurrence/secondary tumor||R||L||R||R||R|
|Treatment at recurrence||C, 131I||C, 131I||C, 131I, chemotherapy||C||C, 131I|
|Status at last follow-up||A, NED||A, NED||D, WD||A, NED||D, NED|
The literature on thyroid follicular carcinoma is voluminous and controversial and it does not easily lend itself to a thorough exposition and summary. Most articles attempt to identify sociodemographic and histomorphologic characteristics that will give an indication of the long term prognosis. These articles include different classifications, attempts at subclassification, and inclusion of new entities with confusing names. In addition, clinical parameters, radiologic findings, serum thyroglobulin values, histochemical studies, lectin histochemistry, immunohistochemical antibody expression, tumor proliferation marker studies, matrix metalloproteinase determination, morphometric analyses, flow cytometry, stereology, chromosome analysis, cytogenetic studies, oncogene activation, and DNA ploidy analysis have been investigated in an endeavor to define follicular carcinoma as an entity, most with unrewarding results.1, 20–56 Unfortunately, the studies have been diluted by a lack of the presentation of strict histologic criteria that would allow for a comparison of one study with another. In fact, a recent study specifically cited poor reproducibility of the histologic criteria using the WHO classification for follicular carcinoma, with even less reproducibility when analyzing the grades of follicular carcinoma.18 Furthermore, the present staging systems generally do not take into consideration sociodemographic features (age and gender), although a complex prognostic index for thyroid carcinoma has been proposed.20 In the past, the scores for the age, grade, extent, and size (AGES; Mayo Clinic),57age, metastasis, extent, and size (AMES; Lahey Clinic),58 or grade of tumor, age, metastasis, extent, and size (GAMES) Memorial Sloan-Kettering Cancer Center)59 scales have been used but have been developed for and usually applied to patients with papillary thyroid carcinoma, not follicular carcinoma. Therefore, we have presented our clinical and pathologic findings based on patients with pure follicular thyroid carcinoma and the histologic criteria for the diagnosis of minimally invasive (low grade) carcinomas.
The mean age at the time of presentation of 42.0 years is within the mean age of other patient reports (between 36.0 years and 65.0 years)4, 6, 11, 14–16, 22, 24, 60–97 and supports the findings in the English literature suggesting that MI tumors occur in a younger age group (Table 1).24, 40, 70, 76, 78–81, 87, 88, 91, 92 Whereas the older patients presented in the literature tended to have a worse outcome (irrespective of tumor type),4, 9, 11, 14, 16, 22, 24, 58, 60, 63, 68, 70, 77 especially when using a cut-off value of > 35–45 years at the time of diagnosis,9, 16, 22, 23, 38, 45, 58, 60, 63, 70, 76, 78, 79, 81, 83–85, 87, 88, 90, 98, 100–103 we were not able to reproduce this finding in our series. Therefore, we cannot make a definitive statement about older age at the time of presentation implying a worse outcome.4, 9, 16, 23, 38, 45, 58, 63, 68, 70, 70, 83, 87, 88, 98, 103
There is no debate that follicular carcinoma is more frequent in women than in men, both in our series (female to male ratio, 2.4:1.0) as well as in the literature.6, 14, 16, 22–25, 38, 45, 60, 65, 70–76, 78, 79, 81–84, 87, 88, 91–96, 99 However, in the above-cited literature, men tended to have a worse prognosis compared with their female counterparts, all other factors being equal. Again, we did not find a difference in outcome based on gender, perhaps because so few of our patients had recurrences or developed metastatic disease. A few other authors have found a similar result.97, 105
All patients presented with a mass lesion in the thyroid gland, often increasing in size. Five patients' tumors were discovered incidentally and were asymptomatic. Our findings, as would be expected with such nonspecific clinical findings, reflect the findings of other authors.6, 14, 64, 67, 71, 72, 74, 75, 79, 81, 91, 95, 96, 106 Two of our patients with Graves disease clinically did not develop a tumor recurrence. This result replicates the findings reported in the literature of a better prognosis for patients with follicular carcinoma who have Graves disease.107 Furthermore, follicular carcinoma tumors tend to be smaller (mean, 2.3 cm) in patients with Graves disease.6, 75, 78, 84, 91, 95, 107, 108
The remaining thyroid parenchyma was affected by a number of abnormalities, including lymphocytic thyroiditis (38.9%), adenomatoid nodules (36.8%), and other thyroid malignancies (papillary carcinoma, 21.0%; medullary carcinoma, 1.0%). The papillary carcinomas were usually of microscopic size and are considered to be of limited biologic significance. This limited biologic significance was supported by the lack of recurrent or metastatic disease developing in any of the patients during the entire follow-up period, similar to findings in the literature.6, 66, 78, 109
All of our patients' tumors occurred within a single lobe of the thyroid, permitting a complete surgical excision with a lobectomy alone. Our findings of an excellent prognosis associated with complete surgical excision confirm the findings of others.2, 11, 14, 16, 24, 58, 62, 65, 70–72, 75 Completion thyroidectomy was performed in a number of our patients (n = 14), although none developed any complications, such as recurrent laryngeal nerve damage or hypoparathyroidism. The complication rate for total thyroidectomy is small (< 14%), but it has serious associated morbidity.24, 58, 62, 66, 72, 80, 84, 91, 93–95, 106, 113–116, 118, 119 Because there is no difference in patient outcome based on the type of surgical treatment performed (P = 0.160), only the most conservative surgery should be implemented initially.2, 4, 11, 58, 70, 83, 84, 86, 87, 93, 94, 98, 106, 110, 111, 113–117 Despite these findings, other authors have advocated total thyroidectomy as initial management16, 79, 80, 82, 90, 119, 120 or a completion thyroidectomy within a short time of the diagnosis.97 It seems pointless to assume the unnecessary risk of complications when the procedural effect does not yield a difference in long term patient outcome.
More than one follicular carcinoma was found in only 2.3% of our patients (one patient with two minimally invasive follicular carcinomas in the same lobe and two patients with bilateral follicular carcinomas), a finding within the range of up to 12% presented in the literature.24, 66, 70, 76, 79, 94, 106, 114, 116, 118, 121, 122 This finding is in sharp contrast to the frequent multicentricity of papillary carcinoma. Perhaps the very low incidence of multifocal follicular carcinoma lends further support to performing the most conservative surgery (i.e., lobectomy) rather than a total thyroidectomy.
A number of patients in our series received postoperative 131I (n = 26 patients from the whole group). Surprisingly, 50% of these patients (n = 13) were given radioactive iodine after undergoing only a lobectomy (and isthmusectomy) rather than after undergoing a completion thyroidectomy (which is the usual form of therapy). Whereas postoperative treatment with 131I yielded a statistically better clinical outcome (P = 0.020) in our patients treated with total thyroidectomy, a finding supported by many authors,5, 14, 16, 22, 24, 62, 63, 67, 72, 90, 101, 116–118, 123, 124 there are nearly as many authors who found no difference in long term clinical outcome when treatment with radioactive iodine was implemented.11, 60, 71, 99, 103, 113, 125, 126 Significant complications after 131I are quite rare (usually isolated sterility). This information suggests that the use of postoperative 131I only in patients who have undergone a total thyroidectomy for follicular carcinoma may yield longer disease free survival.14, 16, 22, 24, 62, 63, 67, 80, 101, 116–118, 123, 124
In the literature, the overall survival rate was between 27% and 100% for patients with follicular carcinoma, not further specified, without any sequelae when they were followed from 4 years to 20 years after presentation.1, 3, 6, 9, 11, 13, 14, 16, 22–24, 29, 38, 40, 45, 58, 60–63, 65, 67, 68, 70–72, 74 Needless to say, this remarkable variability makes it difficult to give a reasonable summary. However, in our study and in a few reported in the literature,3, 13, 16, 24, 40, 40, 65, 68, 72, 82, 83, 85 there were almost no disease-related mortalities, few local recurrences, few to no distant metastases, and an excellent prognosis when patients were followed for 10–20 years. Of the five patients in this study series who developed recurrent disease, four were either alive or had died without evidence of disease at last follow-up: three patients in the MI category and one patient in the NLG group. The one patient who died with recurrent disease died after 15 years and had locally recurrent disease in the neck. This latter finding is mirrored in the above-mentioned literature. When recurrent disease did develop, it was usually after a long interval, and the recurrent disease or “metastatic” disease often was present for a prolonged number of years. The only exception to these findings is in patients with endemic goiter, in which the recurrence and metastatic disease rates are higher.131
A finding of distant metastases at the time of diagnosis lends a more ominous clinical prognosis to patients with follicular carcinoma,3, 9, 13–16, 22–24, 24, 29, 38, 40, 45, 61–64, 67, 70–77, 79–81, 83, 85–87, 90, 94, 96, 116, 118, 120 but none of our patients presented in such a fashion. Whereas “carcinoma” may be defined by metastasis, many different types of carcinoma in pathology have the “potential” to metastasize, even though each individual tumor may not do so. Although none of the patients in our series developed metastatic disease during the course of follow-up, this does not diminish the classification of a low grade carcinoma. In fact, it confirms the low grade or minimally invasive nature of the neoplasm. Therefore, although we cannot make a comment about statements in the literature about metastatic disease at the time of presentation based on the results of this study, we concur that the presence of metastatic disease at the time of initial work-up of a thyroid neoplasm is a more ominous finding.
It has been suggested in more recent literature that patients present at an earlier stage of disease. It is with interest that we note a number of studies that have suggested a change in prognosis based on the time frame of original therapy (before 1960, 1970, or 1980), with a better prognosis found in patients who were treated more recently.10, 58, 74, 77, 86, 88, 112 Our study was completed for patients who were diagnosed and treated originally between 1974 and 1983. Therefore, we cannot propose that there has been a change in patient outcome based on treatment since all of our patients were treated before 1985, but still had an excellent outcome.
The size of the tumor did not yield a difference in patient outcome (recurrence or death from disease) (Table 3). Other authors have suggested that there is a difference in outcome based on tumor size. Even though the tumor categories may be well defined, the overall size varies from report to report, ranging from 1.0 cm to > 6.0 cm.6, 23, 24, 38, 45, 58, 63, 70, 72, 74, 75, 81–83, 85, 87, 90–92, 96, 98, 103, 112, 118 Although it is unwise statistically, despite using a variety of different tumor size cut-off values, we were unable to elicit a difference in patient outcome based on the size of the tumor (Table 8). 129 Nearly 17.9% of our patients (n = 17) had tumors measuring 1.5 cm or smaller, a finding higher than what usually is reported in the literature,24, 40, 66, 70, 72, 78, 83, 85, 87, 96, 107, 117, 118, 122, 129 although it is reasonable given the “minimally invasive” category we were studying. It is possible that the smaller tumors cause more diagnostic quandaries and, thus, would be sent to a referral center such as ours. Furthermore, it has been suggested in the literature that MI tumors are smaller than widely invasive tumors.24, 40, 70, 78, 87, 90, 92, 118 However, based on our series (no statistically significant difference in size of the tumors) and on a number of series of adenomas presented in the literature, the adenomas had an average size of from 4.0 cm to 6.3 cm,75, 78, 112, 129, 132 well above our series' average size. Therefore, tumor size cannot be used reliably to differentiate between follicular adenoma and carcinoma or, specifically, MI carcinomas. Parenthetically, we noted that older patients (age > 50 years) in our series were more likely to have tumors > 3 cm in size (P = 0.042). We do not have an explanation for this finding, even though reports in the literature have suggested a similar finding.81 There was no difference in average years of symptoms to account for a longer duration of growth in older patients versus younger patients.
|Factor evaluated||Statistical result (P value)|
|Older age at presentation for cases with cytoplasmic oxyphilia||0.001|
|Postoperative treatment with 131I yielded a better patient outcome||0.020|
|Tumors > 3 cm more likely to occur in older patients||0.042|
|No difference between tumors with/without mitotic figures||0.092a|
|No difference in size of tumor||0.100a|
|No difference in outcome based on tumor location||0.145a|
|Presence of lymphocytic thyroiditis did not affect patient outcome||0.146a|
|Type of surgery did not affect patient outcome||0.160a|
|No difference in patient outcome based on cytoplasmic oxyphilia||0.450a|
|No age differences between males and females||0.462a|
|Evidence of papillary carcinoma did not affect patient outcome||0.725a|
Although there were statistically significant findings in the current series (Table 8), none of these characteristics had clinical relevance or prognostic application. There was no difference in outcome between patients with MI tumors and patients with NLG tumors based on the following series of examined histologic features (the appropriate literature supporting these findings is cited after the P value): tumor size (P = 0.100),40 presence of mitotic figures (P = 0.092),6, 10, 23, 38, 45, 81, 112 tumor location (P = 0.145), tumor cellularity (P = 0.091),1, 6, 10, 69, 81, 83, 133 cytoplasmic oxyphilia (P = 0.450),1, 6, 10, 23, 38, 45, 74, 81, 89, 98, 103, 110, 127, 134 necrosis (only in NLG tumors; P = 0.083),6, 127 cellular atypia,6 thickness of the tumor capsule (P = 0.192),6, 82, 106, 135 or frequency or type of capsular or vascular invasion (P = 0.844).4, 10, 71, 72, 72, 75, 103, 105, 127
Expansion on a few of these features is deemed prudent to clarify a few of the issues raised in the literature. Oncocytic change in the cytoplasm is considered to be a functional variant of thyroid follicular epithelium. Cytoplasmic oxyphilia (granularity, oncocytic change, Askanazy cell change) is described in a variety of different tumor types and, by itself, does not change the outcome, prognosis, or management of the patient, even though there is still controversy in the literature about this topic.10, 12, 73, 134 Based on our series and on a number of series in the literature, it would seem that any difference in outcome is explained by the use of different diagnostic pathologic criteria and patient selection bias. Tumors with oncocytic or oxyphilic changes in the cytoplasm exhibit a biologic behavior identical to follicular carcinoma of similar stage and clinical presentation. It is of interest to note that older patients in our series were more likely to demonstrate cytoplasmic oxyphilia (P = 0.001) than younger patients, but this may be related to oncocytic/oxyphilic change noted in the organs of older patients in general.136
Insular growth pattern has been suggested as a risk factor for a poorer prognosis.12, 69, 137 However, our results and others in the literature do not duplicate this finding.82 Most of the patients presented in the literature were presumed to have a number of other confounding factors, such as increased age, large tumor size, necrosis, and widespread invasive growth. However, stage for stage, patients with tumors that have an insular growth pattern do not have a different clinical outcome. An insular pattern of growth can be seen in papillary and follicular carcinoma as well as in other benign conditions. Therefore, the identification of the pattern of growth alone cannot be proposed to be an independent prognostic predictor.
The number of areas of vascular invasion, the size of the vessels involved (excluding large vessels), the number of foci of capsular invasion, or the degree of penetration of the capsule may not be important, because there is no difference in the outcome, whether using tumor recurrence, metastatic disease, or patient death as an endpoint. Capsular perforation is present when the tumor cells are seen penetrating through the whole capsule or, alternatively, when there is interruption of the capsule by tumor cells perpendicular to the direction of the capsule fibers. Although the present series is a large and detailed analysis of this topic, these points of view are not necessarily shared by other authors.1, 3, 6, 9, 10, 13, 29, 58, 65, 66, 68, 70, 72, 73, 75–79, 82, 85, 86, 91, 94, 96, 98–100, 110, 127, 128 Others have suggested specific criteria to define invasion based on the degree and extent of invasion, such as complete penetration of the tumor capsule, penetration of the thyroid capsule, and penetration of vessels distant from the tumor but still within the thyroid parenchyma.1, 3–6, 10, 13, 21, 23, 29, 38, 40, 45, 58, 65, 68, 70, 72, 75, 77, 79, 82, 91, 98, 100, 110, 128 Furthermore, it has been suggested that vascular invasion may be a better indication of malignancy than capsular penetration.1, 3–6, 9, 10, 13, 29, 40, 72, 73, 75, 78, 79, 91, 100, 110, 128, 132, 138 Although this finding may be correct teleologically, we were unable to document it in the current series. When capsular invasion is the only finding, metastatic disease can develop; therefore, we believe that a diagnosis of carcinoma is justified in this setting, although we realize that other authors have questioned calling a tumor without vascular invasion carcinoma. Because distant metastases and recurrences have been documented in tumors without vascular or capsular penetration, presumably after adequate tumor sampling,6, 75, 81, 132, 139 we believe that capsular invasion alone is sufficient for a diagnosis of MI follicular carcinoma. It would seem that any type of invasive growth qualifies the tumor as a carcinoma, whether into vessel(s) only, into the capsule only (partial or full thickness), or a combination of the two. Splitting of the tumor capsule is difficult to interpret, because it may be a degenerative or retrogressive change in which there is deposition of “new” collagen, which may not disclose an “invasive” quality of the neoplasm. Entrapment of the tumor cells within the capsule based on the plane of histologic sectioning may imitate invasion. These concerns are not resolved easily, because the focus of obvious carcinoma frequently is isolated and/or small, and the only way to accurately determine the true nature of the neoplasm hinges on the number of blocks submitted or sections examined. Moreover, the areas of diagnostic invasive carcinoma may have been interpreted incorrectly. One study demonstrated that 8 blocks correctly identified only 50% of tumors, whereas 10 blocks identified practically all tumors independent of size.129 Although complete sampling of the tumor is ideal,110, 129, 138 it may be difficult in routine practice (especially with large tumors). On average, as identified in our study, there were 18.0 sections of the tumor to parenchymal interface (i.e., capsule) examined per tumor with an even greater number of slides reviewed (mean, 19.0 slides), because we often obtained serial sections to clarify an area of suspicious involvement. It has been suggested that between 4 sections and 15 sections are sufficient to make the differentiation between adenoma and follicular carcinoma irrespective of the size of the tumor.1, 6, 10, 38, 40, 75, 82, 110, 129 Arguably, over-sampling may not yield any benefit.138 However, prudence dictates a compromise, with the suggestion of 2 sections per centimeter of tumor, up to 10 sections, but with the sections prepared from the periphery or capsule to parenchymal interface of the tumor. The histologic appearance of the tumor at its center is not nearly as important as the findings at the tumor to thyroid parenchymal interface (once papillary, medullary, and metastatic carcinomas have been excluded). The sections from the periphery can be done in two sequences: If capsular irregularities or “pseudoinvasive growth” are identified on the first five sections, then process additional sections to search for capsular or vascular invasion.
A few other risk factors have been assessed in our series as well as in the literature to determine a good treatment regimen to yield the best long term prognosis. These include the presence of distant metastases (other sites or lymph nodes) and a history of radiation exposure.3, 9, 10, 12–14, 16, 24, 29, 40, 65, 70–73, 75–77, 79–83, 85–87, 94, 96, 99, 101, 103, 111, 117, 118, 127 Lymph node metastasis is quite uncommon, ranging from 6% to 30%, with an average of < 10%,5, 6, 11, 12, 16, 29, 66, 69–72, 78, 79, 81, 82, 84, 94, 104, 120, 129, 130, 140 even though none of our patients presented with this finding. The majority of patients who develop metastases develop bone or lung metastases.12, 70, 94, 122 We did not duplicate this result, because none of our patients developed distant metastatic disease. None of the patients in our series who had radiation exposure prior to the development of thyroid disease (n = 6 patients) developed recurrent disease, and all were either alive (n = 5 patients) or had died (n = 1 patient) without evidence of disease and were followed for an average of 18.1 years after diagnosis. Therefore, radiation exposure in our series did not portend a worse long term patient prognosis.
The differential diagnosis of follicular tumors of the thyroid includes adenomatoid nodules, follicular adenoma, follicular carcinoma, papillary carcinoma, encapsulated medullary carcinoma, and metastatic tumors. Even though FNA can identify the follicular epithelial nature of the tumor, it cannot accurately separate tumors into benign or malignant categories.10, 19, 90, 110, 141–143 This is due principally to the limited sampling of the epithelium without any demonstration of the capsule and associated vessels or a tumor capsule to thyroid parenchymal interface. In addition, intraoperative pathologic diagnosis is of limited value unless an area of invasion just happens to be sampled.95, 144
The discrimination of a follicular adenoma from a carcinoma relies only on the presence of angioinvasion or capsular invasion, because their overall histology can be indistinguishable. The distinction between adenoma and carcinoma often is inexact and, as stated above, depends explicitly on the presence or absence of capsular or vascular invasion, not specifically on the morphologic features within the center of the tumor. It goes without saying that the clinical observation of distant metastases also can be used to distinguish between benign and malignant thyroid tumors.4, 6, 9, 10, 59, 65, 71, 73, 76, 78, 81, 82, 100, 103, 128, 129, 132, 145 Based on our findings, features such as increased cellularity, specific growth patterns, cellular atypia or pleomorphism, and mitotic activity are present variably and may be different topographically from one part of the tumor to another. Therefore, these features cannot be used by themselves to diagnose carcinoma in contradistinction from their value in diagnosing malignancies of other organs.
The difference between MI and NLG follicular carcinoma was based on the extent of disease (single or multifocal), the type of invasion (vascular or capsular), and the degree of invasion (< 50% or > 50% capsular invasion and small, medium, or large vessel invasion). In the literature, the designations of slight, moderate, or marked invasion have been used to define the extent of disease but usually not to specifically separate tumors into different grades.4–6, 9, 10, 29, 40, 59, 66, 68, 71, 76, 78, 81, 94, 100, 103, 129 Therefore, it is difficult to compare outcomes specifically with the literature when exact criteria for these designations are not presented. However, “moderate to marked” angioinvasion is associated with an adverse clinical outcome,4, 5, 15, 29, 40, 66, 76–78, 81, 99, 100, 103 as is “marked” capsular invasion.5, 6, 29, 40, 68, 76–78, 81, 82, 99, 103 Once the tumor has been identified as having extended into the capsule or within a vessel, recurrence and metastasis can result; thus, the differentiation between vascular or capsular invasion does not appear to make a difference. The incidence and significance of either capsular or vascular invasion is then called into question, because “invasion” of whatever form, and no matter how frequent, can be considered to be a histologic marker for biologically aggressive behavior, albeit limited in extent. The terms metastasizing colloid goiter, benign metastasizing goiter, malignant adenoma, metastasizing adenoma, angioinvasive adenoma, invasive adenoma, atypical follicular adenoma, proliferating adenoma, wuchernde struma, encapsulated angioinvasive carcinoma, adenocarcinoma in situ, adenocarcinoma in an adenoma, and localized carcinoma in an adenoma are confusing, misleading, and inaccurate terms, serving no useful purpose as far as the management of patients is concerned and should not be employed in day-to-day diagnosis.
In conclusion, based on our thorough, long term clinicopathologic study of 130 follicular carcinomas, we believe that a pathologic distinction between low grade (MI carcinoma) and follicular carcinoma is valid (as suggested by the WHO).17 The lack of improved long term survival with a total thyroidectomy directs the application of the most conservative surgical procedure initially, with additional surgery used only for patients with recurrent disease. It has been suggested that the ambiguity of trying to separate follicular adenoma and low grade carcinoma can be resolved by combining these tumor types into a single category.17 Even though patients with either type of tumor usually have a good prognosis, the lack of separation histologically may result in under-treatment of certain patients.
In light of the literature and our own experience, we believe that we can make a number of reasonable recommendations concerning the management of patients with follicular carcinoma. The limited development of disease recurrence in this series of patients with thyroid follicular tumors allows us to recommend the use of the term “minimally invasive follicular thyroid carcinoma” for all tumors that have single or multiple foci of small-to-medium vessel invasion, single or multiple foci of up to full thickness capsular invasion, and “accompanied” parenchymal extension without large vessel invasion and unaccompanied parenchymal extension. Irregular nests or fingers of tumor cells projecting beyond the contour (bulk) of the tumor, even if still within the capsule, qualify as capsular invasion. Although vascular invasion and capsular invasion were seen exclusive of one another in our series, it is most uncommon, and they usually are seen simultaneously in the same tumor. Without close examination of the specimen and examination of many sections from the specimen, the true nature of the tumor may be missed or understated.
No ultrastructural, morphometric, flow cytometry, immunohistochemical, or oncogene expression studies have proved to be reliably predictable in separating benign neoplasms from malignant neoplasms or between grades of malignant neoplasms. Therefore, a reliance on histomorphologic criteria alone, for the present, is the only way to separate between these categories. Patients with MI thyroid follicular carcinoma demonstrated an excellent survival in our series. It is our conclusion that additional surgical therapy should be used only in the setting of disease recurrence or secondary tumor development. Lobectomy (with or without isthmusectomy) is sufficient initial surgical management. If 131I is used at all, it should only be given to patients who have had a throidectomy and must be individualized for each patient, because there is still controversy over the benefit of 131I therapy. Of course, clinical follow-up is suggested, often for a prolonged period, because the tumors are slow growing and may recur after many years.
The authors thank Pamela A. Thompson for her conscientious research assistance.