Preoperative low tri-iodothyronine concentration is associated with worse health status and shorter five year survival of primary brain tumor patients

Background Low tri-iodothyronine syndrome is associated with worse prognosis of severely ill patients. We investigated the association of thyroid hormone levels with discharge outcomes and 5-year mortality in primary brain tumor patients. Methods From January, 2010 until September, 2011, 230 patients (70% women) before brain tumor surgery were evaluated for cognitive (Mini mental State Examination; MMSE) and functional (Barthel index; BI) status, and thyroid function profile. The Low triiodothyronine syndrome was defined as triiodothyronine concentration below the reference range. Unfavorable discharge outcomes were Glasgow outcome scale score of ≤3. Follow-up continued until November, 2015. Results Seventy-four percent of patients had Low triiodothyronine syndrome. Lower total tri-iodothyronine concentrations were associated with lower MMSE (p=.013) and BI (p=.023) scores independent of age, gender and histological diagnosis. Preoperative Low tri-iodothyronine syndrome increased risk for unfavorable discharge outcomes adjusting for age, gender and histological diagnosis (OR=2.944, 95%CI [1.314-6.597], p=.009). In all patients, lower tri-iodothyronine concentrations were associated with greater mortality risk (p≤.038) adjusting for age, gender, extent of resection, adjuvant treatment and histological diagnosis. The Low tri-iodothyronine syndrome was associated with greater 5-year mortality for glioma patients (HR=2.197; 95%CI [1.160-4.163], p=.016) and with shorter survival (249 [260] vs. 352 [399] days; p=.029) of high grade glioma patients independent of age, gender, extent of resection and adjuvant treatment. Conclusions The Low tri-iodothyronine syndrome is common in brain tumor patients and is associated with poor functional and cognitive status, and with worse discharge outcomes. The Low tri-iodothyronine syndrome is associated with shorter survival of glioma patients.


INTRODUCTION
The mortality associated with brain tumors remains high and largely depends on histological diagnosis, patient clinical status and extent of surgical removal [1,2]. Novel biomarkers are expected to aid in early diagnosis, assist in monitoring tumor recurrence and response to therapy, improve prognostication, and contribute towards developing novel intervention strategies of brain tumors [3,4]. The Low triiodothyronine syndrome is the most common pattern of abnormal thyroid hormone metabolism in critical illness that is characterized by decreased peripheral concretions of triiodothyronine, which is the most potent thyroid gland hormone [5,6]. The Low triiodothyronine syndrome is highly prevalent among patients suffering from neurological disorders and has been linked to more severe health status and to worse prognosis independently of disease severity [7][8][9]. Lower triiodothyronine concentrations were also linked to more aggressive behavior of cancer cells [10,11] and with shorter survival of lung cancer patients [12]. However, the clinical significance of the Low triiodothyronine syndrome in brain tumor patients remains understudied [13][14][15][16]. In 90 patients with various histological diagnoses of brain tumors we have documented that 54% of patients had the Low triiodothyronine syndrome that was associated with an 8-fold increase in risk for unfavorable discharge outcomes and with greater depressive and anxiety symptom severity [15]. However there are no studies investigating long-term prognostic value of the Low triiodothyronine syndrome in primary brain tumor patients.
The goal of the present study was to evaluate the association of the Low triiodothyronine syndrome and other HPT-axis hormone concentrations with disease severity, postoperative outcomes and 5-year prognosis of brain tumor patients.

Association with BT histological diagnosis and health status
Free and total triiodothyronine concentrations below the reference range were recorded in 74% and 50% of patients, respectively. High-grade glioma patients, relative to patients with other histological diagnoses, had significantly lower triiodothyronine (free and total) concentrations, greater rT3 concentrations, and greater prevalence of low free triiodothyronine concentrations (p≤.002) ( Table 2).
There was a positive correlation between total triiodothyronine concentrations and MMSE score (rho=.234, p=.023), but not BI score (p=.064). In univariate regression analyses, lower free and total triiodothyronine concentrations, and low total triiodothyronine concentration were associated with greater odds for unfavorable functional and cognitive status (Table 3). After adjusting for age, gender and histological diagnosis, lower free and total triiodothyronine concentrations remained associated with greater odds for unfavorable cognitive status (p=.039 and p=.013, respectively), and lower total triiodothyronine concentrations, with greater odds for unfavorable functional status (p=.023).
In univariate regression analyses, unfavorable discharge outcome was associated with lower total and free triiodothyronine concentrations, with greater rT3 concentrations, and with the Low triiodothyronine syndrome (p-values ≤.03).
In univariate Cox regression models for the total patient sample, low total and free triiodothyronine concentrations were associated with elevated 5-year overall mortality risk (HR=1 The Low triiodothyronine concentration below the reference range syndrome was associated with 2-fold increased risk of 1-year (p=.038), 2-year (p=.018) and 5-year (p=.008) mortality in high-grade and lowgrade patients and in high grade glioma patients only, independent of age, gender, extent of resection, adjuvant treatment and histological diagnosis (Table 4).

DISCUSSION
The Low triiodothyronine syndrome was frequently observed in brain tumor patients, with the greatest prevalence in high-grade glioma patients. Lower triiodothyronine concentrations were associated with worse functional and cognitive status. The Low triiodothyronine syndrome was associated with a 3-fold increased risk of unfavorable discharge outcomes independent of age, gender and histological brain tumor diagnosis. The Low triiodothyronine syndrome was associated with shorter survival of glioma patients.
Nearly two thirds of primary brain tumor patients had reduced triiodothyronine concentrations. The prevalence of the Low triiodothyronine syndrome was the highest in high-grade glioma patients. In a previous study of 90 patients with primary and metastatic brain tumors, we found that 38% of patients had reduced triiodothyronine concentrations before surgery, and that the prevalence of the Low triiodothyronine syndrome increased to 54% after surgery [15]. Greater prevalence rate of the Low triiodothyronine syndrome in the present study can be explained by the greater number of patients with malignant gliomas (21% vs. 13%) included in the present cohort. Two smaller studies reported lower triiodothyronine concentrations in patients with brain tumors relative to healthy controls [13,14]. Our findings suggest that greater reduction of triiodothyronine concentration might reflect more aggressive brain tumor biology and greater health status impairment imposed by brain tumor. High prevalence of the Low triiodothyronine syndrome was also reported in other critical illnesses, such as sepsis (65%) [17] and severe traumatic brain injury (60%) [8]. Impaired peripheral tissue 5'-deiodination and reduced hypothalamic and pituitary TRH and TSH secretion are the chief mechanisms responsible for development of the Low triiodothyronine syndrome [6,9]. We found that triiodothyronine/thyroxine ratio was lower and rT3/thyroxine ratio was greater in patients with high-grade gliomas relative to patients with other diagnoses, suggesting suppressed of 5' deiodinase activity in malignant glioma patients. Further studies should attempt to elucidate biological mechanisms responsible for impaired thyroid hormone metabolism in brain tumor patients.
The association of lower triiodothyronine concentrations with suboptimal functional and cognitive health status is new to the literature. In previous studies, lower triiodothyronine concentrations were associated with worse health-related quality of life [18], and with greater depressive and anxiety symptom severity [15] of brain tumor patients independently from histological diagnosis, patients' age, gender and functional status. These findings suggest that thyroid impairment should be suspected in brain tumor patients presenting with cognitive, mental and functional complaints. Thyroid hormone replacement therapy was shown to improve cognitive functioning in hypothyroid patients [19] and in aneurysmal subarachnoid hemorrhage survivors [20]. However, whether thyroid hormone replacement therapy can improve health status in brain tumor patients remains to be seen.
Low total triiodothyronine concentration before surgery was associated with a nearly 3-fold increased risk for unfavorable outcomes at discharge, and this association was independent of patients' age, gender and histological brain tumor diagnosis. These findings agree with previous data showing 8-fold greater risk for poor discharge outcomes in brain tumor patients with reduced free triiodothyronine concentrations [15]. An association of lower triiodothyronine concentrations with worse hospital discharge outcomes was previously reported in stroke [21] aneurysmal SAH [22] patients. Our findings imply that brain tumor patients with low preoperative triiodothyronine concentrations should be considered at increased risk for poor discharge outcomes.
For the first time in the literature we have documented that the Low T3 syndrome is associated with shorter survival in brain tumor patients. Reduced triiodothyronine concentrations were also associated with greater mortality, suggesting dose-response association between triiodothyronine concentrations even with the normal range and prognosis of brain tumor patients. In malignant glioma patients who had the lowest preoperative tri-iodothyronine concentrations relative to other brain tumor patients, preoperative Low triiodothyronine syndrome was associated with a 2-fold greater 5-year mortality risk, and this association was independent of patients' age, gender, extent of resection and adjuvant therapy. The latter findings suggest the Low T3 syndrome can help to identify malignant glioma patients who are at the greatest risk for poor prognosis. Previous studies also documented that the Low triiodothyronine predicts greater mortality risk of stroke [23] and lung cancer patients [12] independently of clinical disease severity. Reduced triiodothyronine concentration should be considered an independent prognostic biomarker of shorter survival in brain tumor patients.
There remains an ongoing debate in the literature as to whether the Low triiodothyronine syndrome should be treated in critically ill patients [6,24]. Two observational studies have reported improved survival of malignant glioma patients treated with large doses of triiodothyronine possibly due to increased brain tumor radiosensivity associated with triiodothyronine administration [25,26]. It was also documented that triiodothyronine promotes redifferentiation of glioma cells and suppresses proliferation of high-grade glioma cells [10]. We believe that there is an urgent need for clinical trials aiming to investigate whether the treatment of the Low triiodothyronine syndrome can improve the prognosis of patients suffering from these devastating disorders.
Our study has limitations. Heterogeneous sample in terms of histological diagnosis prevented from investigation of the prognostic value of the Low triiodothyronine syndrome in more homogenous subgroups of brain tumor patients. Due to absence of longterm cognitive function assessment we were not able to evaluate the association of thyroid hormones with patientoriented outcomes. On the other hand, large sample size, long follow-up period and in-depth assessment of HPTaxis functioning are the major advantages of the study.
In conclusion, the Low triiodothyronine syndrome is a common complication in brain tumor patients that is associated with poor functional and cognitive health status, and with increased risk for unfavorable discharge outcomes. Lower triiodothyronine concentrations are associated with greater 5-year mortality risk independent of age, gender, extent of resection, adjuvant treatment and histological diagnosis. In glioma patients, the Low T3 syndrome is associated with shorter survival independently of age, gender, extent of resection, adjuvant treatment and histological diagnosis. There is a need to evaluate whether treatment of the Low triiodothyronine syndrome can improve the prognosis of brain tumor patients.

MATERIALS AND METHODS
Institutional review board approval for the study was obtained. All participants provided informed consent. Consecutive adult patients admitted for brain tumor surgery at the Department of Neurosurgery of Lithuanian University of Health Science in a period from January, 2010 until September, 2011, were considered for this prospective observational cohort study. Patients included in the pilot study 12 were not included in this report. Twohundred and seventy eight patients were identified and invited to participate in the study. However, 48 (12%) patients were excluded because they had missing blood samples (n=30) or were diagnosed with metastatic brain tumors (n=11) and other intracranial lesions (n=7), leaving a final sample of 230 patients (70% women; median age 58 [21] years).
On admission, patients were evaluated for demographic characteristics and previous brain tumor treatments. In addition, 105 (46%) patients were evaluated for functional status (Barthel index; BI [27]) and 94 (41%) patients were evaluated for cognitive status (Mini Mental State Examination; MMSE) [28]. Patients were considered as having unfavorable functional status and unfavorable cognitive status if their BI and MMSE scores were <90 and <24, respectively.
Blood samples for evaluation of the HPT-axis hormone concentrations were obtained within two days of admission and before brain tumor surgery. Clinical outcomes at discharge were evaluated using the Glasgow outcome scale (GOS) [29]. Patients were considered to have unfavorable outcomes if their GOS score ranged between 1 (death) and 3 (severe disability). GOS scores of 4 (moderate disability) and 5 (low disability) were considered as favorable outcomes. Mortality data were collected from the national death registry. Deaths that occurred between the study entry date and November, 2015 were considered for the analyses. Causes of death coded as C70-71, D32, D35.2, D43 or D44.3 were classified as brain tumor associated deaths [30].
On the basis of free and total triiodothyronine laboratory reference ranges, patients were determined as having low free triiodothyronine concentrations (<2 pg/ ml) and low total triiodothyronine concentrations ( <70 pg/ml), respectively.

Statistical analysis
Data were analyzed with SPSS 17.0 for Windows. All thyroid hormone serum concentrations were not normally distributed therefore non-parametric tests were used.
First, we investigated thyroid hormone serum concentrations as a function of histological brain tumor diagnosis by performing Kruskal-Wallis and Chisquared tests. The association of MMSE and BI scores with HPT-axis hormone concentrations was evaluated by the Spearman correlation and binary regression analyses. Significant associations in univariate regression analyses were subsequently adjusted for age, gender and histological diagnosis (enter method).
Next, we investigated the association of discharge GOS score with HPT-axis hormone concentrations by performing Spearman correlation analyses and binary logistic regression analyses with poor discharge outcome (GOS score ≤3) as dependent variable and low triiodothyronine (free and total) concentrations as independent adjusting for age, gender and histological diagnosis (enter method).
Finally, we investigated the association of the Low triiodothyronine syndrome with 5-year overall and brain tumor survival in an all patients and in glioma patients by using Kaplan-Meier and Cox regression analyses adjusting for age, gender, extent of resection, adjuvant therapy and histological diagnosis. In regression analyses, age (in years) was entered as a continuous variable, and gender (men=1 or women=0), brain tumor histological diagnosis (high-grade glioma=1; low-grade glioma=2; meningioma=3; pituitary adenoma=4; acoustic neuroma=5; or other histological diagnosis=6), extent of resection (gross total resection=1 or subtotal resection / biopsy=0), and adjuvant chemotherapy or radiotherapy (yes=1 or no=0) were treated as categorical variables.