Article Text

Download PDFPDF
Highly differentiated follicular thyroid-type carcinoma of the ovary reconsidered
  1. Lawrence M Roth1,
  2. Bernard Czernobilsky2,
  3. David J Roth3,
  4. Liang Cheng1
  1. 1 Pathology & Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
  2. 2 Patho-Lab Diagnostics Ltd, Ness Ziona, Israel
  3. 3 Independent Researcher, Cambridge, Massachusetts, USA
  1. Correspondence to Dr Lawrence M Roth, Pathology and Laboratory Medicine, Indiana University School of Medicine, Indianapolis, Indiana 46202, USA; lroth{at}iupui.edu

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

More than a decade ago, Roth and Karseladze1 first described the entity of highly differentiated follicular thyroid carcinoma of ovarian origin (HDFCO) and reassessed the previously applied term peritoneal strumosis. Subsequently, similar ovarian cases were reported as ordinary struma or follicular adenoma-like lesions lacking a capsule.2 3 In 2018, the alternate term extraovarian struma was proposed based strictly on the molecular findings of their cases.4 In 2019, the term benign deportation was suggested for those HDFCOs in which the dissemination is confined to the peritoneum and peritoneal lymph nodes.5

Histologically, HDFCO can resemble ordinary thyroid tissue, colloid or adenomatous nodule, or an adenoma-like lesion; however, a capsule is never identified in ovarian thyroid-type lesions. Because of the latter findings, the terminology of neoplasms arising in ovarian struma differs from that applicable to the cervical thyroid gland. A perplexing feature of HDFCO is that the diagnosis cannot be established until extraovarian dissemination occurs. Thus, the diagnosis of HDFCO is often not ascertained until several decades after the excision of the original ovarian struma. HDFCO has been accepted as an unusual form of struma-derived thyroid-type carcinoma by authorities in the field.5 6 The term struma-derived thyroid-type carcinoma is applicable to all malignancies that arise in struma ovarii. To distinguish HDFCO from the more common forms, we apply the term typical struma-derived carcinoma for the latter. Cases of ovarian struma-derived carcinoma that resemble papillary or follicular cervical thyroid gland carcinoma and especially HDFCO have a favourable prognosis.1 Herein, we report three cases of HDFCO including the molecular findings of these neoplasms that previously were reported in abstract form and later in full text.7 8

Materials and methods

Case series and histological validation

Formalin fixed, paraffin-embedded sections of the neoplasms in this investigation were stained with H&E. In case 1, the index case, samples for molecular study were obtained from peripheral blood, normal thyroid tissue and mandibular and cardiac tumour tissue prior to radioactive iodine therapy. Samples from case 2 were obtained from an omental nodule and a peritoneal lymph node, and samples from case 3 were obtained from the original ovarian struma. Samples from six additional ovarian strumas were also obtained.

OncoScan microarray analysis

DNA was isolated from frozen tumour or from formalin-fixed paraffin embedded tissues from seven intraovarian strumas and and two extraovarian strumas using standard laboratory procedure. The isolated DNA was analysed using the OncoScan Microarray system (Affymetrix, Santa Clara, California, USA).9 This platform consists of 300 000 markers throughout the entire human genome. This test compares the patient sample to control samples from the HapMap set of 270 individuals. Single-nucleotide polymorphism (SNP) genotyping on this platform has the enhanced ability to identify long contiguous stretches of homozygosity and uniparental disomy as well as targeted genetic mutations (ie, BRAF V600E). Chromosome Analysis Suite software was used for analysis. All copy number changes were determined using the human genome build 19 (hg19/NCBI build 37).

Whole exome sequencing

Whole exome sequencing was performed on the tissue from the mandibular and cardiac tumours obtained from case 1 to comprehensively evaluate for the common thyroid cancer-associated gene alterations.

RNA sequencing

RNA was extracted from formalin fixed, paraffin embedded sections by standard protocols. RNASeq analysis was performed on extraovarian struma from cases 1 and 2 and from ovarian struma in case 3 and sequenced to a read depth of 130M reads per sample using paired-end sequencing with the Mid Output 150-cycle kit (Illumina) on the Illumina NextSeq 500. Gene fusion analysis was performed using FusionCatcher and three aligners: Bowtie, BLAT and STAR.

Results

Case 1: highly differentiated follicular carcinoma of ovarian origin (index case)

A Caucasian woman in her early 70s presented with a history of right temporomandibular joint discomfort recurring for 2 years and right facial swelling for several months. Imaging studies demonstrated a 4.1 cm mass that had destroyed the right temporomandibular joint. Fine-needle aspiration (FNA) biopsy was diagnosed as metastatic well-differentiated follicular thyroid carcinoma; however, an ultrasound examination demonstrated no suspicious thyroidal lesion or lymphadenopathy. Peripheral blood, normal thyroid tissue and mandibular and cardiac tumour tissue were obtained for molecular analysis.

An iodine 123 (I-123) whole body scan identified multifocal areas of metastatic tumour that involved the right face, multiple osseous structures, the porta hepatis, and the cardiac atrial septum. A cardiac imaging study demonstrated a 2.3 cm well-circumscribed mass involving the right atrium that was attached to the atrial septum near the tricuspid annulus. Subsequently, the mass was operatively removed. Postoperatively, the patient received 150.2 mCi of I-131 to treat the metastatic strumal tissue together with a high-dose prednisone taper to minimise tumourous swelling. The right facial pain and swelling slowly resolved. A whole body I-131 scan performed 9 months later showed minimal residual uptake in the mandibular tumour and a 60% decrease in the mandibular tumour volume. Subsequently, she was successfully treated with a second course of 200 mCi of I-131.

On histological examination, the nodule in the pericardium overlying the right atrium varied in appearance. In areas, it was composed of bland follicular thyroid tissue surrounded by a capsule composed of a band of fibrous tissue with overlying adipose tissue (figure 1A). In other areas, the follicles varied in size and shape (figure 1B). Both microfollicles and macrofollicles were observed (figure 1C,D).

Figure 1

Case 1. Highly differentiated follicular carcinoma of ovarian origin (HDFCO). (A) A band of fibrous tissue surrounds a nodule of HDFCO in the omental fat. The thyroid follicles resemble those from an unremarkable thyroid gland. (H&E, ×40). (B) The thyroid follicles are more irregular in shape and vary in size and colloid content (H&E, ×100). (C) Note the macrofollicles and microfollicles (H&E, ×200). (D) The nuclei are dense and show no changes reminiscent of the follicular variant of papillary carcinoma (H&E, ×400). The thyroid-type follicles are filled with colloid.

The medical history was significant for a pelvic ‘dermoid’ removed 48 years previously without reported malignancy.

Case 2: highly differentiated follicular carcinoma of ovarian origin

An African-American woman in her late 50s presented with right upper and lower-quadrant abdominal pain 26 years after excision of an ovarian struma. A CT scan showed acute cholecystitis, but also demonstrated a para-aortic lymph node measuring 5 cm in greatest dimension and multiple peritoneal nodules. At the time of her cholecystectomy, multiple peritoneal and omental nodules and an enlarged para-aortic lymph node were removed. The histological interpretation was consistent with innocuous appearing thyroid tissue (ie, extraovarian struma ovarii). The neoplasm destroyed the right mandibular condyle, infiltrated the parotid gland and masticator space, and also involved the right temporomandibular joint. FNA biopsy disclosed a well-differentiated follicular thyroid-type carcinoma. An ultrasound examination demonstrated no suspicious intrathyroidal lesion or lymphadenopathy. A whole body iodine 123 (I-123) scan identified multiple foci of metastatic struma ovarii that involved the right face, multiple osseous structures, the porta hepatis and the atrial septum. A cardiac imaging study demonstrated a 2.3 cm well-circumscribed mass in the right atrium attached to the atrial septum near the tricuspid annulus that was subsequently removed operatively and histologically had the appearance of bland thyroid follicular tissue. Total thyroidectomy was performed, and the histological examination was negative for thyroid malignancy. Postoperatively, the patient received 150.2 mCi of I-131 to treat the metastatic strumal tissue together with a high-dose prednisone taper to minimise swelling of the tumour. The right facial pain and swelling slowly resolved. A whole body scan using I-131 performed 9 months later showed minimal residual uptake in the mandibular tumour and a 60% decrease in the mandibular tumour volume. Successful repeat treatment was given with 200 mCi of I-131. She underwent total thyroidectomy followed by an ablative dose of I-131. 2.5 years later, a radioactive iodine scan showed multiple focal areas of involvement in the pelvis and right upper quadrant. Because of persistent peritoneal disease she was treated with a second course of I-131. Six years after operation, she had increased radioactive uptake in the right upper quadrant of the abdomen. The thyroglobulin level remained elevated 6 years after laparotomy; however, her disease remained stable. At last follow-up, she was living with disease 8 years after operation.

The medical history was significant for a pelvic ‘dermoid’ removed 48 years earlier without any reported malignancy.

On histological examination, the peritoneal and omental nodules consisted of foci of thyroid follicles of variable size resembling a colloid nodule of the thyroid gland (figure 2A). Peripheral vacuolisation of the colloid was observed in the macrofollicles; however, in a few small areas, microfollicles predominated (figure 2B). The nuclei were uniform and normochromatic. Nuclear features of thyroid-type papillary carcinoma were absent, and no mitotic figures were identified. In one section, a sharp demarcation existed between the macrofollicular and microfollicular areas (figure 2C). The appendicular fat and the region of the cystic duct contained similar nodules of tumour. A para-aortic lymph node was partially replaced by similar thyroid type tissue (figure 2D). The gallbladder contained a calculus. A liver biopsy was negative for neoplasm. The cervical thyroid gland showed only a colloid nodule. Sections of the primary left ovarian tumour removed 26 years previously were unavailable for review. This case was previously reported.1

Figure 2

Case 2. Highly differentiated follicular carcinoma of ovarian origin. (A) Nodule of thyroid type tissue resembling a colloid nodule is observed in omental fat. Note the absence of a capsule (H&E, ×40). (B) The follicles vary in size and often show peripheral vacuolisation of the colloid (H&E, ×200). (C) A sharp demarcation exists between an area to the left that resembles a colloid nodule and a more cellular area to the right that has the appearance of an adenomatous nodule (H&E, ×200). (D) In one nodule, residual lymphoid tissue containing germinal centres is observed suggesting that the nodule is replacing a lymph node (H&E, ×40).

Case 3: highly differentiated follicular carcinoma of ovarian origin

A woman in her early 20s underwent a right ovarian cystectomy for an adnexal mass that was diagnosed as struma ovarii. No mature cystic teratoma was identified. The clinical International Federation of Gynaecology and Obstetrics (FIGO) stage was 1A. She had a local recurrence of the neoplasm 3.5 years after the initial operation, and the neoplasm had a similar innocuous histological appearance. Her thyroid gland was removed and contained a small colloid nodule but was otherwise unremarkable. She subsequently had abdominal recurrences 6 and 9 years after the original procedure. She had a thyroidectomy and was subsequently treated with I-131. She was living with disease at last follow-up 12 years after the initial diagnosis.

The original ovarian struma measured 6 cm in greatest dimension. On histological examination, the original neoplasm resembled nodular thyroid tissue with moderate variation in follicular size (figure 3A,B). She had a local recurrence of neoplasm 3.5 years after initial diagnosis with a similar innocuous-appearing histology (figure 3C).

Figure 3

Case 3. Highly differentiated follicular carcinoma of ovarian origin. (A) Unencapsulated nodule from the original excision of ovarian struma consists of a mixture of small and large thyroid type follicles (H&E, ×40). (B) At higher magnification, the follicles are filled with colloid of variable density (H&E, ×100). (C) Nodule from the first recurrence resembles an adenomatous nodule of the thyroid gland (H&E, ×200). (D) Nodule from the second recurrence shows that the follicular lining cells have small dense nuclei. No changes suggestive of follicular variant of papillary carcinoma are noted (H&E, ×400).

The histological appearance of the second and third abdominal recurrences that occurred 6 and 9 years after the original operation, also had a similar appearance (figure 3D). She was living with disease at the time of her last follow-up 12 years after the initial diagnosis.

Discussion

Historical background and current concepts

In 2008, Roth and Karseladze1 first described HDFCO, a rare form of struma-derived thyroid-type carcinoma. Historically, such cases had been referred to as peritoneal strumosis because of their bland histological appearance and were considered to be peritoneal implants due to a lack of appreciation of their malignant potential. Surprisingly to us, the latter term is occasionally applied for this neoplasm at the present time.10 Although the vast majority of ovarian strumas remain confined to the ovary, a few spread beyond to extraovarian sites.

Recently. Henderson et al 4 reported that ovarian and extraovarian struma have the same genetic mutations, most commonly segmental and less commonly complete whole genome homozygosity. Both profiles stem from errors in meiosis 1 and/or 2. Furthermore, they lack the driver mutations or gene fusions observed in typical thyroid carcinoma of follicular and papillary types.11 12 The data of Henderson et al 8 suggest that extraovarian struma is molecularly identical to ovarian struma without invasion of adjacent tissue. Additionally, such neoplasms at a later time can undergo so-called second hits that are associated with an altered histological appearance and a more aggressive clinical behaviour. Whole genome homozygosity likely predisposes these neoplasms to significant genetic mutations, chromosomal gains or losses, and gene fusions that may contribute to tumour aggressiveness and malignant behaviour. A highly significant finding in their investigation established that struma in both ovarian and extraovarian locations lacks the driver mutations or gene fusions found in papillary and follicular carcinomas of the thyroid gland. It remains unclear why some tumours remain within the ovary, whereas others spread to distant metastatic sites. Chaubey et al 7 in a brief publication from the same group identified serendipitous cross-over events in 7 specimens of struma ovarii (intraovarian struma) by whole genome SNP microarray analysis.

Recently, an alternate term extraovarian struma ovarii was proposed based strictly on the molecular findings of our cases.8 Whole genome homozygosity likely predisposes these neoplasms to significant genetic mutations, chromosomal gains or losses, and gene fusions that may contribute to tumour aggressiveness and malignant behaviour.

In an important review article, Zhang and Axiotis13 found that in a series of 15 patients diagnosed as HDFCO, none died of disease; however, the length of follow-up was not stated. Furthermore, the natural history of HDFCO is unknown because clinical oncologists typically treat patients with this diagnosis aggressively with I-131 therapy. Since these tumours are well differentiated, they have a slow rate of progression. Histologically, HDFCO can resemble normal thyroid tissue, colloid or adenomatous nodule, or follicular adenoma; however, a capsule has not been identified in ovarian struma-derived lesions. Neoplasms histologically resembling HDFCO can rarely contain foci of papillary or follicular carcinoma, and similar foci can also be identified in late recurrences5; however, such neoplasms should not be diagnosed as HDFCO per se. HDFCO appears to have a more favourable prognosis than papillary or follicular carcinoma, two other forms of struma-derived carcinoma.1

Recently, the term benign deportation was introduced for those cases of ovarian struma in which extraovarian spread is confined to the peritoneum and peritoneal lymph nodes.5 Although use of this terminology deserves careful consideration, it is significant that struma ovarii has never been described on the surface of the ovary; thus, it is unclear as to what method the foci of thyroid tissue can implant in the peritoneum or in peritoneal lymph nodes. In actuality, since struma ovarii itself is a neoplasm, we do not fully understand the justification for using any term that possibly could imply a non-neoplastic condition for a derivative of struma ovarii. If one prefers to consider those cases of HDFCO that are limited to the peritoneum and peritoneal lymph nodes as an exception to the concept of HDFCO, we would prefer the less committal term extraovarian struma ovarii.4

Nomenclature plays an important role in scientific communication and should concisely and accurately reflect current knowledge.14 The applicable terminology should concisely and accurately describe the biologic behaviour of the lesion to both clinicians and pathologists. Nomenclature needs to be modified as new concepts develop. When nomenclature becomes controversial, it should be resolved by a consensus committee and should never become personal.

An even more important question that cannot be answered at the present time is whether to diagnose HDFCO confined to the peritoneum and peritoneal nodes as extraovarian struma ovarii to avoid Iodine 131 (I-131) therapy. Such a study would be difficult or impossible due to the rarity of the neoplasm and its slow progression, and therefore, might not be feasible. Absent evidence to the contrary, however, we believe that all such patients should be diagnosed as HDFCO and treated with Iodine I-131. Interestingly, a similar controversy exists to this day regarding the concept of benign thyroid inclusions in cervical lymph nodes.6

Summary of reported cases of HDFCO

Including the originally described cases, we are aware of 10 cases (table 1). The age at detection of the ovarian neoplasm ranged from 22 to 50 years, and the age at recurrence ranged from 31 to 72 years. One case was unique in that it had already spread at the time of the detection of the original ovarian neoplasm.1

Table 1

Reported cases of highly differentiated follicular thyroid carcinoma of ovarian origin (HDFCO)

Differential diagnosis

An important differential diagnosis of HDFCO is the follicular variant of papillary thyroid-type carcinoma. Liu et al 15 reported a unique case of a 72-year-old woman who developed metastatic neoplasm over 30 years after hysterectomy and oophorectomies performed at separate times. The neoplasm had both benign-appearing areas that resembled HDFCO and malignant ones similar to the follicular variant of papillary thyroid-type carcinoma. Although the authors made the diagnosis of proliferative HDFCO, we would have interpreted the neoplasm as the follicular variant of papillary tyroid-type carcinoma arising in HDFCO. From a molecular standpoint, the HDFCO underwent a so-called second hit and formed a more aggressive thyroid-type neoplasm.

Molecular and other significant findings

Schmidt et al 12 reported BRAF mutations in papillary thyroid-type carcinoma of ovarian origin. Coyne and Nikiforov11 reported RAS mutations in the follicular variant of papillary thyroid carcinoma arising in struma ovarii. Carey et al 16 reported a case of HDFCO that recurred 31 years after total abdominal hysterectomy and salpingo-oophorectomy. Ranade et al 17 reported an unusual case of HDFCO associated with hyperthyroidism. Riggs et al 18 reported a case of HDFCO that they managed with minimally invasive surgery in a neoplasm that recurred more than 10 years after the original oophorectomy. In a highly significant article, Tsukada et al 19 reported a case of HDFCO with molecular findings that they diagnosed as malignant struma ovarii. Histologically, the neoplasm consisted of numerous thyroid-type follicles without the nuclear features of papillary thyroid carcinoma Tumour samples were investigated for 50 cancer-related genes, including RAS, BRAF and p53, and PPARg-PAX8 gene fusion by targeted DNA sequencing and fluorescence in situ hybridisation, respectively. No major oncogenic gene alterations were detected. These negative findings suggest a different mechanism of tumourigenesis from that of typical follicular thyroid-type carcinoma. Thus, they confirmed and extended the earlier findings of Henderson et al.8

Summary

In summation, recent histological observations and molecular studies have increased our knowledge of HDFCO and other struma-derived carcinomas. Herein, we revisit the entity of HDFCO, review the literature on the subject of peritoneal dissemination of ovarian struma, and incorporate recent information regarding the molecular findings in struma ovarii. Ovarian and extraovarian struma have similar molecular profiles and lack the driver mutations or gene fusions found in other struma-derived carcinomas that arise from ovarian struma. Molecular analysis techniques demonstrated that struma ovarii, whether ovarian or extraovarian, lacks the driver genes or gene fusions characteristic of typical papillary or follicular carcinoma of the thyroid gland, thus, for the first time providing a molecular explanation for the benign histological appearance of HDFCO. The absence of typical thyroid cancer gene mutations or fusions suggests that whole genome homozygosity may be the primary genetic alteration responsible for metastasis in ovarian struma. HDFCO resembles non-neoplastic thyroid tissue or a benign thyroid-type neoplasm and has a favourable prognosis; however, those HDFCOs that undergo so-called second hits behave more aggressively. In the future, these findings may prove to be important in the treatment of these unusual neoplasms.

Ethics statements

Acknowledgments

We wish to acknowledge the important contributions of Fredrik Skarstadt, Graphic Design, Pathology Department, Indiana University School of Medicine, Indianapolis, Indiana, USA.

References

Footnotes

  • Handling editor Mona El-Bahrawy.

  • Contributors LMR conceived the study and wrote the manuscript. All other authors reviewed and contributed to the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.