Brain Metastases from Thyroid Carcinoma: Prognostic Factors and Outcome

1. Introduction

Thyroid carcinoma (TC) is the most coparkmmon malignancy of the endocrine system with an average age of 40 years at diagnosis [1,2,3]. Women are more frequently affected than men with a male/female ratio of 1/3 [4,5]. Well differentiated TC including papillary thyroid carcinoma (PTC) and follicular thyroid carcinoma (FTC) are the most common histologic types of TC (85% to 90% of all newly diagnosed TC), both having a favorable prognosis after appropriate treatment [6,7]. Distant metastases from TC may involve the lung and bone [8,9]. Brain metastases (BM) from TC are rare, occurring approximately in only 1% of all patients with TC [10,11]. However, because of the implementation of systematic cerebral imaging before the start of treatment with tyrosine kinase inhibitors (TKIs), a higher frequency of asymptomatic BM has been documented more recently [12,13,14]. Surgical resection has been favored for treatment of BM either combined with stereotactic radiosurgery (SRS) or whole-brain radio therapy (WBRT) resulting in preservation of quality of life and prolonged survival [12,15,16,17,18,19]. Nevertheless, the presence of intracranial metastases from TC has been considered to be associated with poorer outcome and a lower survival rate [20,21,22]. Management of TC with BM is challenging because of limited experience and many questions remain unclear. Furthermore, there is still a lack of studies comparing the clinical characteristics and outcomes of patients with BM from different types of TC.

The aim of the present study was to review the experience with the management of BM from TC over a 30-year period at Hannover Medical School providing a maximum care referral center. For this purpose, we analyzed the clinical characteristics, the therapeutic approaches and corresponding outcomes. Additionally, we evaluated the prognostic factors affecting cancer-specific survival in patients with BM secondary to TC.

2. Materials and Methods

To identify patients with a diagnosis of BM from TC, a systematic review of the databank of the Department of Neurosurgery and the Department of Nuclear Medicine at Hannover Medical School was performed retrospectively over a 30-year period. Among 4320 patients with TC, 20 patients with BM were identified between 1989 and 2019. Histological confirmation of the thyroid origin of BM was present for 15 patients. The charts with demographic and clinical data (age, sex, mode of presentation and radiologic features of BM, treatment modalities including surgery, SRS, WBRT with or without boost, patient outcome, preoperative signs and symptoms and concomitant diseases, and histopathological findings were evaluated. The progress of treatment over the years was taken into account. Neurosurgical procedures were performed according to departmental standard techniques as described elsewhere [23,24]. Patients were followed up until December 2022. Overall survival (OS) after diagnosis of BM was calculated, and survival time was measured from the diagnosis of BM to the date of death or the last follow-up.

2.1. Statistical Analysis

Cox proportional hazards regression analysis was used for metric predictor variables and categories. In addition, the Cox regression model was applied for survival analysis including the simultaneous effects of multiple risk factors on survival. The model's coefficients formed the computational basis for hazard ratios (HR) to determine the effect size of predictors on survival (ranging in an open-right interval from 0 to infinity).

As predictors, a list of variables with a possible impact on survival time was selected such as patient age, the time interval between initial diagnosis and detection of BM, number of metastases found, presence of neurologic deficits, number of intracranial lesions, and gender. All predictors were categorized (dichotomized) using a two-level nominal scale into set thresholds (Table 1). Survival times were given as months; censoring data excluded four patients with extreme survival time data. Thus, 80% of data were found between 6 and 77 months. Two predictors, metastases and neurological deficits showed one missing value each.

Statistical analyses were performed using JMP®, Version 16.2.0 (SAS Institute Inc., Cary, NC, 1989–2021) and R Core Team (ver. 4.2.3, R Foundation for Statistical Computing, Vienna, Austria, 2023). A p-value of less than 0.05 was considered statistically significant.

3. Results

3.1. Patient Characteristics

Of the 20 patients with BM form TC, 14 were women and 6 were men (female/male ratio = 2.33:1). Age at the time of diagnosis of TC ranged between 26 and 79 years, with a mean age of 59.3 ± 14.1 years. The mean age at BM diagnosis was 64.8 ± 14.9 years.

3.2. Histological Findings of the Primary Tumor

Ten patients had PTC, seven had FTC, and three had a poorly differentiated carcinoma. According to the TMN classification, 2 patients were in stage I-II, 15 in stage III-IV, and in 3 patients the stage could not be determined. Nineteen patients presented with other previous and/or synchronous distant metastases: lung (n=17), bone (n=7), liver (n=1), kidney (n=2), spine (n=2), and 1 unknown. Total surgical thyroidectomy was performed in 17 patients, while 3 patients had subtotal thyroidectomy. Lymph node dissection was added in 15 patients, and five patients did not undergo lymph node dissection.

3.3. Clinical Features of BM from TC

Table 2 summarizes the demographic and clinical data of the patients with BM. The mean interval between the diagnosis of TC and BM was 63.4 ± 58.4 months (range 0-180 months), and from BM diagnosis to death 41.2 ± 60.7 months (range 1-252 months). Ten patients had a single BM, 8 patients had 2 or 3, and 2 patients had 4 or more BM. BM were revealed by neurologic symptoms in 11 patients (headache, nausea, motor or sensory deficits, ataxia, aphasia, confusion, and epileptic seizures). The other 9 patients had no symptoms and the BM were diagnosed incidentally. The diagnosis was made by magnetic resonance imaging in eight patients, by computed tomography in 6 patients, by iodine scans in three, and by PET imaging in three. The most common site of brain metastasis was the parietal lobe (10, 32%), followed by the frontal lobe (7, 22%), cerebellum (5, 17%), occipital lobe (5, 17%), temporal lobe (2, 6%) and brainstem (2, 6%). The Karnofsky Performance Score (KPS) in 17 patients (85%) was > 70, and in 3 patients (15%) it was ≤ 70.

3.4. Treatment

The primary treatment for BM was surgical resection in 15 patients, and surgery was the only treatment for BM in 4 instances. SRS and isotretinoin therapy was only used in one patient, respectively. WBRT alone was accomplished in three patients. The 11 other patients underwent a combined treatment algorithm. Surgical resection plus WBRT was performed in 8 patients. In one patient, radioactive iodine was applied after surgical resection, another patient underwent WBRT and tyrosinkinase inhibitor therapy after surgical resection. In another instance, WBRT was performed following SRS.

3.5. Survival

The overall median survival time was 15 months after diagnosis of BM from TC, and it ranged from 1 to 252 months. At the time of the last follow-up, 16 patients were dead, and 4 patients were still alive. Information regarding the cause of death was available for 14 patients. Death was presumed to be related to progression of multiple BM, recurrence of BM and progression of extracranial metastases in 12 patients, while two patients died from septic shock, and in 2 patients the cause of death was unknown.

The regression model’s significance level indicated a difference in survival time when at least one of the covariates was included in the model, i.e., the omnibus null hypothesis of multivariate Cox analysis could soundly be rejected (chi2(6) = 17.0149, p = 0.0092).

In detail, the hazard ratio for age, HR = e^Coef = 6.19, indicated a highly increased risk of death for patients over 60 years of age. Contrarily, the covariate duration, i.e., an interval longer than 24 months between the initial diagnosis and BM detection, markedly decreased the risk of death (HR = 0.21). Both test results reached statistical significance. The remaining analyses of the covariates metastases (HR = 1.62) and gender (HR = 1.07) showed opposite trends compared to neurological deficits (HR =0.79) and lesions (HR = 0.97) in their statistical relationship to survival times. The four latter effects, however, failed to be statistically significant (Table 3). Figure 1 shows the Kaplan-Meier survival analyses with corresponding 95% confidence intervals of predictors in the Cox regression analyses.

4. Discussion

The prevalence of BM in patients with TC was rare in our study (less than 1%) which was in keeping with published previous studies [2,13]. BM from TC, however, were reported more frequently in the last few years [25,26,27]. This increase is likely due to the more frequent use of cranial imaging and also to a longer survival in patients with advanced disease. BM were detected in several of our patients without neurological symptoms incidentally after a systemic image survey before targeted therapy. There was no difference between OS in patients with PTC or FTC. Patients with a single BM had a better outcome than those with multiple metastases. Our study demonstrated also that OS in patients younger than 60 years of age at the time of diagnosis of BM was significantly better in comparison to patients older than 60 years. Additionally, OS was better in patients with a period longer than 24 months between TC and BM diagnosis.

The interval between the first diagnosis of TC and BM in our study was relatively long, similar to previously published studies [3,16,28,29]. The most common location of BM was in the telencephalon, while 3 patients had cerebellar lesions and 2 patients had both infra- and supratentorial BM. Remarkably, Kwon et al reported that BMs from lung adenocarcinoma might preferentially involve the distal MCA territory and the cerebellum [30]. Further, Kyeong et al showed that different breast cancer subtypes might show different distributions of BMs [31]. Moreover, Kim et al. postulated that a “seed and soil theory” with an affinity of the tumor to a microenvironment would be partially responsible for the involvement of specific brain regions [3].

Our study is in line with previous studies concerning the coexistence of other organ metastases and BM [2,32,33,34]. In our series, 90% of patients had metastases involving two or more organs. The most frequent coexisting metastatic sites were lung (85%) and bone (35%). Further, our study showed that BM develops generally after the onset of other metastases [2,27].

Treatment decisions for BM in TC are influenced by various clinical factors including the extent of systemic tumor spread, the specific histopathology type, the tumor's responsiveness to radioiodine, and its location within the brain [33,34]. As recommended by the National Comprehensive Cancer Network (NCCN) guideline in 2022, surgical removal followed by WBRT or SRS combined with WBRT is recommended for patients with stable systemic disease or those newly diagnosed. Furthermore, either neurosurgical resection or stereotactic SRS is preferred over WBRT for patients who have solitary BM. Also, survival was significantly improved by surgical resection of one or more tumor foci. On the other hand, WBRT or SRS alone is suggested for patients with multiple (>3) metastatic lesions. According to published studies, radioiodine might not prove effective in treating cranial metastases from thyroid cancer due to the inability of most cranial metastatic lesions to concentrate radioiodine [35,36].

Various retrospective studies have shown that resection of BM improves OS in patients with BM from TC [12,37,38]. Previous studies [11,39,40] showed also, that patients with a KPS score of 70 or more have a better prognosis with longer survival. Chiu et al [41] reported that patients with a single BM have a better outcome than patients with multiple BM, while Wu et al [13] reported no difference in patients with single metastases in comparison to multiple metastases.

The results of the present study should be interpreted with caution due to the retrospective design of the study and the inclusion of a sample of patients from a single center. Moreover, the prognostic subgroups defined in the study were not validated in a prospective form, which, however, is not likely to be expected because of the rare incidence of BM from TC.

In summary, giving their rarity, the exact incidence of BM from TC has been difficult to estimate. Although patients with thyroid cancer generally have a favorable prognosis, those with BM typically experience a poorer prognosis, characterized by a shorter survival period. In our study the number of BM does not play a significant role in determining patient survival. Furthermore, in the case of a longer interval (more than 24 months) between TC diagnosis and BM manifestation, patients have a significantly better OS. Both, surgical resection and multidisciplinary adjuvant treatment are critical for patients with BM from TC.