Rate and Risk of Follicular Carcinoma in Patients with a Cytologic Diagnosis of Follicular Neoplasia: Experience from Tertiary Hospital in Ethiopia

Background

Cytological diagnosis of malignant follicular lesions is difficult because of its failure associated with an identification of either capsular or vascular growth in cytology.¹ This failure of demonstrating either vascular or capsular invasion by thyroid cytology resulted in the term follicular neoplasia to reflect the limitations in cytological diagnosis.² When abundant follicular epithelial cells are seen with macrophages and degenerated cells in the background of scanty colloid, diagnosis of follicular neoplasia is made.²

In 2009 the Bethesda system for reporting thyroid cytopathology was introduced to standardize terms used in thyroid cytology and it classified thyroid fine-needle aspiration cytology (FNAC) into six categories.³ Each category is linked to a certain risk of malignancy and diagnosis of follicular neoplasia or suspicious for follicular neoplasia is among the list of categories. Since FNAC does not differentiate benign from malignant lesions accurately, patients are subjected to surgical removal of nodules that are shown to have follicular neoplasia. Unfortunately, only 15–20% of these nodules postoperatively are proven histologically to have malignant lesions.^3–7

According to the recommendations from the American thyroid association, diagnostic hemithyroidectomy is recommended for nodules with follicular neoplasia depending on clinical risk factors.⁸ Several pieces of literature suggested several clinical characteristics such as male gender, tumour size, and gland pattern that help in predicting the presence of malignancy in those patients with diagnosis follicular neoplasia.^9,10

The aim of this study is therefore to evaluate the prevalence of follicular carcinoma and to assess risk factors associated with the risk of malignancy in patients with a preoperative diagnosis of follicular neoplasia.

Methods

A retrospective review of medical records of patients who were operated on for preoperative diagnosis of thyroid follicular neoplasm at St. Paul’s Hospital Millennium Medical college from January 2015 to December 2019 was made. FNAC report of suspicious for follicular neoplasm was also included in the study. A list of patients’ medical records was initially identified from operation theatre logbooks from which medical record numbers were obtained. The charts of the patients were then retrieved from the hospital’s archive. Relevant information including patient (demographic, clinical) and clinical tumour characteristics done by physical examination (size, consistency, and nodularity), imaging and final postoperative histologic findings were collected from charts using a structured checklist. Criteria used for tumour classification used were according to the recent WHO classification² and the diagnosis of FTC was made when there was evidence of either capsular or vascular invasion in the cytology. Relevant information such as the presence of associated symptoms of malignancy, for example shortness of breath, hoarseness of voice, and difficulty of swallowing was also collected from the records. Only patients with postoperative biopsy results were included in the study to compare the difference in both pre- and post-operative pathological diagnosis. A comparison of clinical and tumour factors was made between patients with benign and malignant histopathologic diagnosis after thyroidectomy to identify relevant risk factors associated with the risk of follicular adenocarcinoma using both univariable and multivariable analysis. The risk for thyroid follicular carcinoma was expressed by odds ratio and a 95% confidence interval was calculated for the odds ratio. P values of less than 0.05 were taken for statistical significance. Univariate association between the presence of follicular carcinoma and clinical factors was analyzed using chi-square test and those variables with statistical significance upon univariable were assessed using stepwise logistic regression to identify independent factors.

Result

A total of 135 patients underwent thyroidectomy for follicular neoplasm in the study period but only the records of 110 patients were retrieved for analysis from medical archives. Of these 110 patients, 98 (89.1%) were female and 12 (10.9%) were male (F: M = 9:1). Among the 110 patients, 53 (48.2%) presented with less than 4 cm size of the mass, and 57 (51.8%) of them presented with greater than 4 cm size. Most of the patients (88; 80%) had associated symptoms of malignancy such as shortness of breath, hoarseness of voice, etc. The majority of patients (102; 90.9%) had firm consistency of the mass and 94 (85.4%) of them had multi-nodular patterns upon clinical examination (Table 1). The majority of patients (90%; 29/32) with malignant histology had multi-nodular character. Seventy-eight percent of patients (25/32) with malignant histology had the size of tumour greater than 4 cm but only 18.7% (6/32) of these were found to have hard consistency during physical examination (Table 1).

Table 1 Socio-Demographic and Clinical Characteristics of Patients Who Were Operated on from January 2015 to December 2019, at St Paul’s Hospital Millennium Medical College (N=110)

Out of all histopathologies performed postoperatively, only 32 (29%) patients had malignancy, consisting of 21 (19.1%) follicular carcinoma and 11 (10%) papillary carcinoma. Otherwise, 78 (70.9%) had benign pathology with 48 (43.6%) being follicular adenoma (Table 2).

Table 2 Post-Operative Histopathologic Findings of the 110 Patients Who Were Operated on from January 2015 to December 2019, at St Paul’s Hospital Millennium Medical College

A comparison of socio-demographic and clinical characteristics was made between those with malignant and benign histopathology (Table 1). The mean age of patients with malignant and benign histopathology is 48.3 (42.9–53.7) and 41.7 (38.7–44.7), respectively (p <0.05). A malignant diagnosis of follicular carcinoma was significantly more likely to be found in patients older than 45 years (40% vs 21.5%, p <0.036), and in those with clinical tumour size greater than 4 cm than those with smaller nodules (43.9% vs 13%, p <0.001). Furthermore, nodules with hard consistency were more likely to be malignant than those with firm consistency (75% vs 25%, p <0.001), and also patients with associated symptoms of malignancy were significantly more likely to have malignancy than those without these symptoms (59.1% vs 21.6%, p <0.001) (Table 1).

Gender and nodular characteristics of the mass did not significantly differ between patients found to have malignant lesions and those with benign histopathology (Table 1).

Data were analyzed both by univariable and multivariable analysis to determine important predictors associated with the presence of follicular carcinoma and the results are described in Table 3. Tumour size, consistency of nodules, and presence of associated symptoms of malignancy were identified as important predictors upon stepwise logistic regression. The odds of cancer for nodules greater than 4 cm is 4 times higher than the odds for smaller nodules. Patients with hard consistency of mass are 6 times at higher risk of having malignancy than those with firm consistency and patients with associated symptoms of malignancy such as shortness of breath are 4.8 times at higher risk of malignancy than those without symptoms.

Table 3 Univariable and Multivariable Analysis of Clinical Factors Associated with Risk of Malignancy of Patients Who Were Operated on from January 2015 to December 2019, at St Paul’s Hospital Millennium Medical College

Discussion

Diagnosis of carcinoma in the follicular lesion is often made following examination of biopsy sections after surgical procedure. This is because FNAC is unable to distinguish between adenoma and carcinoma though it is among the most reliable and cost-effective methods in diagnosing thyroid nodules. Patients are therefore subjected to diagnostic surgery to distinguish between the benign and malignant nodules by biopsy. Though diagnostic surgery is commonly performed there are reports that suggest centers use several clinical criteria to clinically follow patients with benign findings rather than subjecting them for surgery.¹¹

We retrospectively studied 110 patients who had thyroid nodules which were read as follicular neoplasia by FNAC and evaluated postoperative biopsy results.

The reported rate of malignancy according to some literature in thyroid nodules with FNAC diagnosis of follicular neoplasia after surgery is between 20–30%.^4,5,12,13

This study has also revealed a similar rate of malignancy of 29.1% that is consistent with the above studies. This is relatively higher when compared with other studies which reported from 16–18%.^14,15 The reason for the higher rate can be attributed to the center being one of the tertiary care centers in the country that many referred patients visit.

Of the malignant lesions identified in our study, the majority (65%; 21/32) is follicular carcinoma and 35% (11/32) papillary. This is in line with reports by Chen et al. and Kim et al.^12,16 The reason for the higher frequency of follicular carcinoma can be attributed partly to patient selection which was concentrated on lesions that were follicular from the outset but the fact that Ethiopia is among the countries that are iodine-deficient¹⁷ and since historically follicular carcinoma has been linked with regions of the world with iodine deficiency¹⁸ this might explain the higher rate of prevalence in our study. FTC is also said to arise from longstanding goiters and the majority of thyroid nodules in Ethiopia are multinodal goiter (MNG) according to some works of literature from Ethiopia.^19,20

Several studies have tried to combine and integrate different combinations of factors that would help in predicting the risk of malignancy in patients with thyroid nodules with the diagnosis of follicular neoplasia.^{6,9,15,21–24} These authors have concluded that clinical features such as old age, male gender, tumour size greater than 4 cm, nodular characteristics, and consistency of tumour would be of help to define a sub-group of patients with high risk for malignancy.

The calculated risk of malignancy for women with greater than 4 cm solitary nodules was 45% according to Tuttle but our study revealed no association of gender with the risk of malignancy on univariable analysis.⁹

Our study has shown that older age greater than 45 years, clinical tumour size greater than 4 cm, hard consistency of tumour are all associated with significantly increased risk of malignancy. These findings correspond with similar reports from several pieces of research that noted an association of older age greater than 50 years⁶ and tumour size greater than 4 cm with an increased risk for malignancy.¹⁰

Although there are reports of association of risk of malignancy with a nodular character that noted a higher risk of malignancy with solitary nodule this was not seen with our study which showed no significant association of risk of malignancy with the nodular character of goiters in univariable analysis. The reason for this may be the use of only clinical examination in characterizing nodularity of thyroid which usually misses the presence of multiple nodules when compared with high-resolution ultrasound²⁵ which in our case was not used for examination.

Univariable analysis in our study demonstrated that associated symptoms of malignancy such as shortness of breath, dysphagia, and change in voice are significantly associated with risk of malignancy (p <0.001) and there is only a single report by Tuttle et al. that showed no significant association of clinically worrisome features such as dysphagia and hoarseness of voice with the risk of malignancy.

Multivariable analysis using standard stepwise logistic regression in our study has shown that clinical size of tumour greater than 4 cm, hard consistency of thyroid nodule, and associated symptoms of malignancy are significant independent risk factors predictive of malignancy in our model. There are several pieces of research that tried to explore the interaction of multiple clinical factors that estimate the probability of malignancy for patients but there are no similar combinations of factors so far. Schlinkert et al. reported advanced age, size of the tumour, and solitary nodularity as independent risk factors for malignancy¹⁰ but Tuttle et al. demonstrated male gender, size of tumour greater than 4 cm, and solitary character were independently associated with risk of malignancy.⁹ Part of the explanation for such dissimilar conclusions would be the need for many years to accumulate the experience to incorporate a variety of clinical features into an estimation of risk of malignancy.⁹

The limitation of this study is that the data are from a single institution and a large number of patients were excluded because of the incompleteness of data. Since the use of ultrasound in predicting the risk of malignancy is currently being ascertained, failure of including ultrasound features in examining the risk of malignancy might have masked some of the features such as nodular characters of the tumors that are best seen by ultrasound.

We, therefore, conclude that simple clinical features such as the size of the tumor, hard consistency of the nodule, and symptoms associated with malignancy such as dysphagia and hoarseness of voice will help in identifying patients with an increased risk of malignancy when pre-surgical FNAC is used with follicular neoplasia. Integration of these simple features with cytologic findings might help physicians to stratify patients to high and low risk and also help in better decision making. Therefore we recommend careful clinical evaluation of patients to identify the risk of malignancy when pre-surgical FNAC shows follicular neoplasia.