Emerging role of free triiodothyronine in patients with anti-N-methyl-D-aspartate receptor encephalitis

We aimed to investigate the role of free triiodothyronine (FT3) in patients with anti-N-methyl-D-aspartate receptor (anti-NMDAR) encephalitis. 137 consecutive inpatients (2016–2019) were registered prospectively and followed up for 12 months. 96 eligible patients were included in the study. The modified Rankin scale (mRS) score was collected, and the score of 3–6 was defined as a poor outcome. The patients were equally classified into 3 subgroups based on their FT3 levels obtained within 24 h of admission, and the subgroup differences were analyzed by parametric or nonparametric tests as appropriate. Logistic regression analysis was performed. We found that there was no difference in the mRS scores upon admission among 3 subgroups, however, patients in the low-FT3 subgroup tended to have higher disease severity during hospitalization and worse outcome in follow-up visits, represented by higher chances of intense care unit (ICU) admission (P < 0.001), longer hospital stay (P < 0.001), greater maximum mRS scores during hospitalization (P = 0.011), lower rates of getting clinical improvement within 4 weeks of starting treatment (P = 0.006), and higher percentages of poor 1-year outcome (P = 0.002). The level of FT3 was an independent factor correlated with ICU admission (P = 0.002) and might be a potential predictor for 1-year outcome. Our preliminary results suggest that the FT3 may be a risk factor involved in the evolution and progression of anti-NMDAR encephalitis, whereas the underline mechanisms remain to be explored. Attention should be paid to these patients with relatively low FT3 upon admission, which might possibly aid clinical prediction and guide clinical decision-making.


Results
Descriptive characteristics of the study population. In total, 137 consecutive patients were screened, and 96 patients met the inclusion criteria (median 30 [21-40.5] years; 44.79% females). Figure 1 summarizes the cohort selection process, and the baseline clinical features of the study population is shown in Table 1. The most common main symptoms included psychiatric behaviors (79.17%), seizures (65.63%) and cognitive dysfunction (58.33%). More than half of the patients had abnormal MRI (57.29%) and pneumonia (56.25%), but only a few of them (5.21%) were detected with tumors, including 2 pulmonary cancers and 3 immature ovarian teratomas. A quarter of the patients had thyroid dysfunction, including 11 patients with subclinical hyperthyroidism (11.46%), 1 patient with clinical hyperthyroidism (1.04%), 2 patients with subclinical hypothyroidism (2.08%), 6 patients with low T3 syndrome (6.25%), 4 patients with other conditions (4.17%). A majority of the patients (94.79%) received first-line immunotherapies, but only a minority of them (12.5%) were treated with second-line immunotherapies additionally. All patients were followed for 12 months, and most of them (76.04%) achieved a favorable outcome. A small number of the patients (6.25%) died within 12 months due to serious complications, including 3 patients with severe infections and 3 patients with intractable status epileptics.

Comparisons of clinical features among subgroups.
Based on the serum FT3 levels, the patients were equally divided into 3 subgroups, including the low-FT3 subgroup (FT3 value < 4.13 pmol/L), the middle-FT3 subgroup (FT3 value 4.13-4.84 pmol/L) and the high-FT3 subgroup (FT3 value > 4.84 pmol/L) (Fig. 1). In view of clinical data obtained within 24 h of admission (Table 1), there were no differences in gender, clinical symptoms, the modified Rankin scale (mRS) score, ICU admission, FT4 levels, FT3-test delay and thyroid antibodies among subgroups. However, compared with those in the high-FT3 subgroup, patients in the low-FT3 subgroup had an older age at onset (34 [23.5-48.5] vs. 26 [17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33]  In Regard of the clinical data obtained during hospitalization (Table 1), there were no differences in tumor presence, abnormal MRI, CSF tests and immunotherapy among subgroups. However, compared with those in the high-FT3 subgroup, patients in the low-FT3 subgroup seemed to be more likely to deteriorate after admission, indicated by higher rates of consciousness declination (65.63% vs. 28.13%, P = 0.01) and central hypoventilation (34.38% vs. 6.25%, P = 0.036). Their chance of accompanying pneumonia was also higher (71.88% vs. 34.48%, Figure 1. Cohort selection process of the study population. Of 137 admissions for anti-NMDAR encephalitis reviewed, 12 patients were excluded because their IgG anti-GluN1 antibodies were only positive in serum but negative in CSF, which couldn't meet the diagnostic criteria of definite anti-NMDAR encephalitis. Another 29 patients were excluded for the lack of results of thyroid hormones within 24 h of admission. Ultimately, a total of 96 patients were included in this study. Based on the levels of serum FT3, the patients were equally divided into three subgroups, i.e. the low-FT3, middle-FT3 and high-FT3 subgroup. Since some patients' length of hospital stay was shorter than 28 days, the data of "lack of clinical improvement within 4 weeks of starting treatment" were fully completed during follow-up visits. Compared with those in the high-FT3 subgroup, patients in the low-FT3 subgroup were more likely to lack clinical improvement within 4 weeks (53.33% vs. 16.67%, P = 0.013), had higher NEOS scores (2.8 [0.96] vs. 1.57 [1.07], P = 0.001), and were prone to achieve poor 1-year outcome (46.88% vs. 9.38%, P = 0.007). The differences remained significant between the low-FT3 subgroup and the middle-FT3 subgroup for lack of clinical improvement within 4 weeks (53.33% vs. 22.58%, P = 0.047) and poor 1-year outcome (46.88% vs. 15.62%, P = 0.029).
Association between the FT3 and ICU admission. Considering that patients in the low-FT3 subgroup had significant higher rates of ICU admission, whereas the percentages of ICU admission upon admission were approximately identical among different subgroups, we performed the univariate and multivariate logistic regression analysis to assess the relationship between the FT3 and ICU admission ( Table 2). Given the phenomenon that the FT3 level in critical illness might change over time 15,16 , we only analyzed the data obtained within the same period as FT3, i.e. within 24 h of admission. The tumor presence, although discovered later during hospitalization, was also included as it could not change in a short time. The level of FT3 (P < 0.001) was significantly correlated with ICU admission in the univariate regression model. Other related factors included consciousness declination (P = 0.005), psychiatric behavior (P = 0.021), the WBC count (P = 0.003) and the TSH (P = 0.011). The multivariate regression model confirmed the FT3 as an independent factor (P = 0.002) after adjustments for the other 4 confounding factors.

Correlation between the FT3 and 1-year outcome.
In the univariate regression model assessing all clinical data obtained within 4 weeks of starting treatment (Table 3), the level of FT3 was significantly correlated with 1-year outcome (P = 0.004). Other related factors included gender (P = 0.008), age (P = 0.034), tumor presence (P = 0.018), consciousness declination (P = 0.047), central hypoventilation (P = 0.002), ICU admission (P = 0.003), lack of clinical improvement within 4 weeks (P < 0.001) and the NEOS score (P < 0.001). However, the prognostic value of the FT3 was attenuated (P = 0.594) after adjustment for other confounding factors. The NEOS score turned out to be the sole predictor independently related to poor 1-year outcome (P = 0.007).
Considering the fact that the level of FT3 might change over time in critical illness 15,16 , the clinical data obtained within the same period as the FT3 would be more comparable and suitable to assess the role of FT3. Therefore, we performed another univariate logistic regression analysis of the clinical data obtained within 24 h of admission and the tumor presence (Table 4), 5 factors turned out to be significant, including the FT3 (P = 0.004), gender (P = 0.008), age (P = 0.034), tumor presence (P = 0.018) and central hypoventilation upon admission (P = 0.01). The prognostic value of the FT3 remained to be significant (P = 0.030) after adjustment for the other 4 confounding factors.

Discussion
This study has demonstrated two major findings. First, the patients with relatively low FT3 within 24 h of admission tend to have higher disease severity during hospitalization and worse outcome in follow-up visits, represented by higher chances of ICU admission, longer hospital stay, greater maximum mRS scores during hospitalization, lower rates of getting clinical improvement within 4 weeks of starting treatment, and higher percentages of achieving poor one-year outcome. Second, the level of FT3 was an independent factor correlated Table 2. Univariate and multivariate logistic regression analysis of predictors obtained within 24 h of admission for ICU admission during hospitalization. In the univariate model, the eligible variables obtained within 24 h of admission and tumor presence were analyzed, and those identified as not predictors of ICU admission during hospitalization (P > 0.05) were not shown in this table. In the multivariate regression model, the predictive value of FT3 was adjusted by consciousness declination, psychiatric behavior, WBC count and TSH within 24 h of admission. *Indicates P < 0.05. OR, odds ratio; CI, confidence interval; FT3, free triiodothyronine; WBC, white blood cell; ICU, intense care unit; TSH, thyroid stimulating hormone. www.nature.com/scientificreports/ with ICU admission during hospitalization and might serve as a potential predictor for 1-year outcome. Taken altogether, our results reveal that the FT3 might be a risk factor involved in the evolution and progression of anti-NMDAR encephalitis. Thyroid hormones are able to cross the blood-brain barrier and affect neurogenesis, cell differentiation, and myelination 17,18 . A growing body of literature shows that the FT3 is also closely involved in regulating neutrosphere biology and immune system function (e.g., cell-mediated immunity) 19,20 . Anti-NMADR encephalitis is mediated by antibodies to neuronal surface antigens, with 25% suffer significant morbidity or mortality 21,22 . Given the critical role of the FT3 in the proliferation and differentiation of neuronal and glial progenitors during brain development and in the regulation of immune system function 17,23 , we hypothesize that the FT3 might be a risk factor in the progression and outcome in anti-NMDAR encephalitis. The preliminary results of the study bolstered this assumption, because: (1) Patients in the low-FT3 subgroup, although with the similar symptoms and mRS scores upon admission compared to those in the high-FT3 subgroup, seemed to be more likely to aggravate during hospitalization and exhibit higher disease severity, indicated by greater maximum mRS scores during hospitalization, longer hospital stay, and higher rates of ICU admission. After adjustment for relevant clinical variables, the level of FT3 turned out to be an independent prognostic factor related to ICU admission; (2) Patients in the low-FT3 subgroup had much higher chances of achieving poor one-year outcome, and the univariate regression analysis confirmed the level of FT3 as a significant prognostic factor, although its predictive value was attenuated after adjustment for other confounding factors obtained within 4 weeks of starting treatment. However, since the level of FT3 might change over time in critical illness 15,16 , the clinical data obtained within the same period as the FT3 would be more comparable and suitable to assess the role of FT3. Therefore, we performed another multivariate regression analysis of clinical data obtained within 24 h of admission. The tumor presence, although discovered later during hospitalization, was also analyzed as it could not change in a short time. As expected, the predictive value of the FT3 remained significant after adjustment for gender, age, the tumor presence and central hypoventilation obtained within 24 h of admission.
The mechanisms of how the FT3 influences the evolution and outcome of anti-NMDAR encephalitis remain elusive. First, we assume that patients with low FT3 might experience decreased neuroprotection and increased  www.nature.com/scientificreports/ secondary brain damage after anti-NMDAR encephalitis, leading to higher disease severity during hospitalization and poorer outcomes. In some vitro studies, the FT3 has been proven to exert a protective effect against glutamate toxicity in neurons and glial cells through both transcriptional and non-transcriptional mechanisms 24,25 . Second, patients with low FT3 might exhibit a suppression of endogenous brain repair systems in light of the crucial role of the FT3 in the generation and maturation of new neurons and axonal myelination 26 . Increasing evidence suggests that the FT3 is closely related to brain acetylcholine activity, cholinergic function, and the secretion of various neurotrophic factors such as nerve growth factor, which are pivotal to the brain repair system 27 . However, since several critical conditions may have decreased FT3 levels 6,7,10 , and the exact immunologic relevance of FT3 in anti-NMDAR encephalitis remains obscure 19,20 , the FT3 might as well be at first interpreted as a biological risk factor, but not as a causal factor directly altering outcome of patients with anti-NMDAR encephalitis.
In the multivariate regression model assessing all clinical data obtained within 4 weeks of starting treatment, the predictive value of the NEOS score was validated in this prospective cohort, which provided another Class III evidence of using the NEOS score as a tool to predict 1-year functional status in patients with anti-NMDAR encephalitis 3,5 . It is of interest to notice that patients in the low-FT3 subgroup had higher NEOS scores, which might provide a clue to predict the NEOS score in the early stage of the disease.
Several limitations merit consideration in the interpretation of this study. First, it is observational in nature and based on data from a single hospital, which might have led to unintentional bias. Second, the number of patients in this analysis is of only moderate size limiting the statistical power. Third, we only evaluated serum thyroid hormones at a single time point. The level of FT3 in critical illness changes over time, which is a dynamic process 15,16 . Future studies evaluating multiple time points are required for validating the predictive role of FT3 in anti-NMDAR encephalitis.
In conclusion, our analysis reveals that the FT3 might be a risk factor involved in the evolution and progression of anti-NMDAR encephalitis, whereas the underline mechanisms remain to be explored. Attention should be paid to these patients with relatively low FT3 upon admission, which might possibly aid clinical prediction and guide clinical decision-making. Further investigations in larger cohort of patients are definitely needed to confirm the prognostic value of FT3.

Methods
Subjects and evaluation. In this prospective study, we evaluated consecutive inpatients with anti-NMDAR encephalitis admitted in the Zhengzhou University First Affiliated Hospital between January 1, 2016 and December 31, 2019. Depending on the clinical status, informed consent was obtained from the patients or their relatives. Prior approval of the study protocol was obtained from the ethical committee. All patients' records and information were anonymized and de-identified prior to analysis. The inclusion criteria were (1) met the diagnostic criteria of definite anti-NMDAR encephalitis, i.e. in the presence of one or more of the six major groups of symptoms and positive IgG anti-GluN1 antibodies in the CSF, after reasonable exclusion of other disorders 1 ; (2) thyroid hormones obtained within 24 h of admission, including serum FT3, FT4 and TSH.
Follow-up visits started from the day of diagnosis, i.e. the day of obtaining positive IgG anti-GluN1 antibodies in the CSF. Each patient underwent a follow-up evaluation by telephone or outpatient interview for 12 months.
The mRS values during hospitalization were calculated by referring physicians and medical records. A favorable outcome was defined as a mRS score of 0-2 and a poor outcome was defined as a score of 3-5 or death (mRS score of 6).

Clinical information and thyroid function measurements. Clinical data and physical examination
were obtained in all patients upon admission. Blood samples were collected within 24 h of admission and, in addition to standard blood tests, serum levels of FT3, FT4 and TSH were determined using a direct chemiluminescence assay (ADVIA, Bayer Health Care LLC Tarrytown, NY, USA). The reference intervals of our laboratory were as follows: 3.28-6.47 pmol/L for FT3, 7.9-18.4 pmol/L for FT4, and 0.34-5.6 µIU/ml for TSH. A small number of patients also received tests of thyroid autoantibodies, including anti-thyroglobulin antibodies (TGAb), anti-thyroid peroxidase antibodies (TPOAb), and anti-thyroid receptor antibodies (TRAb) 13,14 . "Clinical hyperthyroidism" was defined as suppressed TSH and elevated FT4 levels with relevant clinical symptoms, and "subclinical hyperthyroidism" was defined as a suppressed TSH level, FT3 and FT4 levels within the normal ranges, and the absence of symptoms. In contrast, "clinical hypothyroidism" was defined as elevated TSH and suppressed FT4 levels with relevant symptoms, and "subclinical hypothyroidism" was defined as an elevated TSH level, a normal FT4 level, and the absence of symptoms. "Low T3 syndrome" was defined as a low serum FT3, normal-low FT4, and normal-low TSH levels 28 . "Other conditions" included slightly elevated FT3 or FT4 levels, normal TSH and the absence of symptoms.
During hospitalization, all patients underwent the lumbar puncture, and the IgG anti-GluN1 antibodies in the CSF were detected by cell-based assays (CBA), using Euroimmun IIFT kits: Autoimmune Encephalitis Mosaic 1 (FA 1121-1005-1), and/or NMDAR kits (FA112d-1005-51), according to the manufacturer's instructions 1 . All measurements were performed by laboratory staff who were blinded to patients' clinical information. All patients received the following examinations to screen tumors during hospitalization, including the brain MRI, computerized tomography (CT) scan of the thorax, ultrasound of the abdomen, pelvic and reproductive regions. None of the patients were given methylprednisolone or any other immunotherapies before admission, and the immunotherapies were not started until the GluN1 antibodies in the CSF were confirmed.
The "pneumonia" was diagnosed by respiratory physicians according to the relevant criteria. The "tumor presence" was defined as having ovarian teratoma or any other malignant tumors. The "abnormal MRI" was defined as having relevant brain lesions confirmed by radiologists. The "FT3-test delay" was defined as the interval between the onset of symptoms and the test of FT3 value upon admission. The "immunotherapy delay"