New Reference Values for Thyrotropin Hormone in Diabetic Patients

Diabetes is believed to affect the hypothalamo-pituitary thyroid axis as well as the peripheral conversion of tetraiodo to triiodothyronine. Thyroid hormones; namely Triiodothyronine (FT3) and Tetraiodothyronine (FT4) are considered as insulin antagonists which potentiate insulin action indirectly. Low thyroid hormones in diabetic patients may be explained by the decreased production of thyrotropin releasing hormone [6,7].


Introduction
Prevalence of thyroid diseases in the general population ranges from 6.6% to 13.4%. This prevalence is increasing, particularly in women. Both diabetes mellitus and thyroid dysfunction are related to each other through complex biochemical, genetic, and hormonal malfunctions [1][2][3][4][5].
Diabetes is believed to affect the hypothalamo-pituitary thyroid axis as well as the peripheral conversion of tetraiodo to triiodothyronine. Thyroid hormones; namely Triiodothyronine (FT3) and Tetraiodothyronine (FT4) are considered as insulin antagonists which potentiate insulin action indirectly. Low thyroid hormones in diabetic patients may be explained by the decreased production of thyrotropin releasing hormone [6,7].
Prevalence of abnormal thyroid function in T2DM is comparable to that in T1DM as per available reports although the genetic links are less clear. Despite the evident association between diabetes and thyroid dysfunction, there is no special recommendation for thyroid screening as part of diabetes health care. Guidelines are vague with a lot of variation ranging from ignoring the thyroid function test to yearly testing [8,9].
Recently, thyroid diseases including subclinical hypothyroidism, MPV and RDW showed strong relationship to cardiovascular and atherosclerotic heart diseases and are associated with higher morbidity and mortality [10][11][12].

Material and Methods
This is a retrospective cross-sectional study conducted at king Fahd Hospital, a tertiary care hospital, in Asir province, Saudi Arabia. A sample of 270 Saudi type 2 diabetic patients who had attended the endocrinology and diabetes clinics between July 2016 and July 2017 were randomly selected for the study. Their medical records were reviewed. The final number of patients after applying the inclusion and exclusion criteria was 186. Inclusion criteria were type 2 diabetes, of both sexes, whose age was above 18 years. The exclusion criteria included anemia, pregnancy, postpartum period and type 1 DM. Individuals with previous history of thyroid disease, co-existing hepatobiliary disease, heart failure, history of myocardial infarction or on systemic drug therapy such as, glucocorticoids or oral contraceptives were not included in the study. Patients were also excluded if they had recent acute condition such as acute respiratory or gastrointestinal infection within one month before data collection. Clinical data such as age, sex, weight, height, BMI and blood pressure were collected from medical records. BMI was calculated as per the equation: BMI = weight (kg)/ height (m 2 ). The Modification of Diet in Renal Disease (MDRD) study formula was used for calculation of eGFR [13].
Glucose oxidase method (Spinreact, Girona, Spain) was used to measure fasting plasma glucose. Total Hb was measured colorimetrically. HbA1c was determined immunoturbidimetrically and was expressed as a percentage of Hb by a conversion equation to match a HPLC reference method; HbA1c (%) = HbA1c/Hb x 175.8 + 1.73 [14].
Our laboratory reference ranges RBC 4.5-6.3 10 9 /L, WBC 4-11 10 9 /L, MCH 26-36 pg, MCHC 32-36 gm/dL, HCT 38-52%, Hb 14-18 gm/dL, Platelets 140-440 10 9 /L, RDW 11-14%, MPV 7-13 fL. TSH: 0.35-4.9 uIU/ml, FT4: 7.5-21.1 pmol/l, FT3: 3.8-7.8 pmol/l, TC: 0-5.2 mmol/l, LDL-C: 3-5.2 mmol/l, HDL-C: 1.04-1. 55 ) were compared using Chi-Square test (or Fisher's exact test). Quantitative data between two groups were compared using Independent-Samples t-test if data were normally distributed in both groups. Mann-Whitney U test was used if data are not normally distributed. Pearson correlation coefficient, denoted as r, and it is the coefficient that measures the strength and direction of a linear relationship between two continuous variables. Its value can range from -1 for a perfect negative linear relationship to +1 for a perfect positive linear relationship. A value of 0 (zero) indicates no relationship between two variables. The Spearman's rank-order correlation calculates a coefficient, r s , which is a measure of the strength and direction of the association/relationship between two continuous or ordinal variables. The diagnostic performance of a test, or the accuracy of a test to discriminate diseased cases from nondiseased cases is evaluated using Receiver Operating Characteristic (ROC) curve analysis (Metz, 1978;Zweig & Campbell, 1993). For any of the used tests, p value ≤ 0.05 indicated statistical significance. Appropriate charts were used to graphically present the results whenever needed including scatterplot [18].

Results
186 patients were valid for screening as they had TSH level as the screening test for thyroid function. Out of 186 patients, 18 patients (9.7% of the study sample) were on thyroxine replacement therapy indicating initial diagnosis of clinical or subclinical hypothyroidism. Of the 186 patients, 27 patients were excluded because of insufficient other data required for further statistical analysis. So, our statistical analysis was applied to 159 patients who were of a mean age 53.9±11.5 years, diabetes duration 9.2±7.4 years, HbA1c 8.6±4.7% and TSH level 3.05±2.23 uIU/ml. Descriptive analysis is shown in (Table 1). Comparison of various parameters in males and females showed higher TSH, BMI and heart rate in females (p = 0.004, < 0.0001, 0.001 respectively) while, serum creatinine and BUN were significantly higher in males (p < 0.0001, 0.011 respectively). No significant gender difference was noticed in the other study variables ( Table 2).
Thyroid dysfunction was found in 15.7 % of patients. High TSH above the reference range was found in 24 patients (15.1%) indicating This Figure shows that female gender predominates among those with thyroid dysfunction while male gender predominates among those without thyroid dysfunction. However, this difference did not achieve a statistical significance (p=0.119).
TSH significantly and directly correlated to RDW, MPV and BMI (p= 0.027, 0.026, 0.015 respectively). Neither FT4 nor FT3 showed such significant correlations. No significant association was found between TSH and duration of diabetes, FPG, HbA1c or lipid profile (Tables 3 and 4

& Figures 2-4).
At TSH value of 5 uIU/ml as a cut off for hypothyroidism, no significant difference between patients on either side of this level as regard to any of the measured clinical or laboratory variables ( Table  5).
Because of the direct linear relationship between TSH level and RDW, using the highest point of the reference range of RDW, defined a cut off value of TSH of 1.68 uIU/ml with sensitivity of 79% but with low specificity 38% ( Figure 5).
Comparing various clinical and laboratory variables on both sides of TSH 1.68 uIU/ml did showed significant difference in WBC and pulse rate (p= 0.004, 0.018 respectively), being higher in the higher TSH group (Table 6).

Discussion
Similar to our finding of higher BMI in diabetic Saudi females than males (<0.0001), Kamath showed that a larger percentage of type 2 diabetic Indian females had significantly higher central and general obesity in comparison to males (p= <0.001) [19].
The observed lower serum creatinine in female patients can be explained by lower muscle mass than males while their higher TSH in our study is reported by many other authors [2,[20][21][22].
Noteworthy is our observation of sex difference in heart rate being higher in females. This was also mentioned in other reports [23,24].
Third generation TSH level is of above 98% sensitivity and 92% specificity in detection of both hypo and hyperfunction. So, in our study TSH was the main hormone available as a screen for thyroid function [3,25].
Being a retrospective study, we could not trace the presence of goitre or family history of thyroid dysfunction and no data were available about thyroid auto antibodies. Hyperthyroidism is a less common thyroid dysfunction in both general and diabetic patients. This is very clear in our study where we found only one case of subclinical thyrotoxicosis that may indicate the presence of thyroiditis or subclinical hyperthyroidism [29,35].
Although thyroid hormones affect insulin sensitivity, absence of an association between thyroid disease and glycemic control may be explained by the other various factors that can play a role in the control of diabetes including patient lifestyle, compliance to treatment and other used medications.
In a Jordanian study, prevalence was 12.5 %. In other gulf countries like Oman, overall prevalence was 12.5 % and was higher in females (p= 0.01) [31,36].
In concordance with several reports our study showed a strong In other population studies, thyroid disorders affected 16% of diabetic patients meanwhile Celani reported a prevalence of 9.7% [29,30].
In Saudi Arabia, Akbar reported a prevalence of 16 % while in another retrospective study, Al-Geffari found thyroid disorders in 28.5%, of the study group, with 25.3% as hypothyroidism and 3.2% as hyperthyroidism. Al-Geffari's report is one of the highest prevalence reports of thyroid dysfunction in T2DM [7,31,32].
In our study of Saudi population, the prevalence of thyroid dysfunction was 15.7%, 23.8% in females, 13 % in males without gender preference (p-value= 0.119).
Al-Geffari explained his observation by the high prevalence of Latent Autoimmune Diabetes of Adult (LADA) in Saudi type 2 diabetic patients reaching 26%. This was evident in his study population with high prevalence of autoimmune thyroid diseases. He also mentioned several risk factors for thyroid dysfunction of which are family history of thyroid disease, female gender, and diabetes duration more than 10 years. In our study, gender and duration of diabetes and its control did not correlate to thyroid dysfunction. Female gender was a risk factor for developing thyroid dysfunction in several studies while age, duration and control of diabetes were not in others. In Afkhami-Ardekani's report, HbA1C was significantly higher patients who had  direct association between TSH level and BMI (p=0.015) [19,37,38].
Similar to Khurana, we did not find an association between TSH and serum lipids. However, most of our patients were on hypolipidemic agents mainly statins that may altered their lipid profile [39].
Presence of an association between thyrotropin and both RDW and MPV in the current study is in agreement with both Geetha and Yu's reports. Kim and Lippi also demonstrated the strong positive association between thyrotropin and MPV. In Lippi's study, high TSH even with normal FT4 was an independent predictor of high MPV. Both RDW and MPV are believed to be risk factors for cardiovascular diseases. So the association between TSH and both risk factors in diabetic patients indicates the significance of screening and defining an upper limit of TSH reference in this special category of people who are at higher risk for cardiovascular disease due to diabetes [35,[40][41][42][43][44].
Since the 2002 Consensus Conference on subclinical thyroid disease studies showed that even slight elevation of TSH level (3-10 mIU/liter) can increase the risk of atherosclerosis in some individuals who are susceptible, like diabetic patients [45][46][47][48].
In 2002, the American Thyroid Association and American Association of Clinical Endocrinologists recommended that each laboratory should define its TSH reference values. In iodine sufficient areas, TSH range of 0.45-4.12 mIU/L should be considered if the laboratory reference is not available patients having TSH level higher than 10 mIU/L are recommended for thyroxine replacement to avoid heart failure and cardiovascular mortality. When TSH is in the range of upper normal level to 10 mIU/L, the need for thyroxine intake should be individualized according to several factors such as the presence of symptoms of hypothyroidism, autoantibodies, heart failure, atherosclerosis or risk factors for these diseases [49].
Several reports confirmed the improvement of markers of atherosclerosis such as serum lipids and carotid intima media thickness with thyroxine treatment even with very mild elevation of   TSH [50]. Patients at high risk for ASCVD with TSH levels of 6.1-10 mIU/L or higher than 10 mIU/L and their age was less than 65 years who did not receive thyroid hormone treatment had higher allcause mortality when compared with those who started their thyroid replacement therapy [51].
In a UK study, individuals with TSH between 5.01 and 10 mIU/L and normal FT4 in the age group of 40 to 70 years and older were followed for about 8 years. The hazard ratio for ischemic heart disease events significantly decreased in patients who received thyroxine treatment when compared to those who did not take the medicine (0.61, CI 0.49-0.92). This reduction was not observed in patients older than 70 years [52].
The American Association of Clinical Endocrinologists recommended a TSH upper limit of 3.0 mIU/liter as this level was associated with lowest prevalence of thyroid autoimmunity in NHANES III study [53].
In our study considering TSH upper limit of 5 µiU/ml could not detect a significant difference in CVD risks between the groups of patients having a value below and above this limit. On the other hand, a TSH level of 1.68 µiu/ml as an upper limit, was defined retrospectively, as the figure at which there is increase in RDW above the upper normal value (14%). This showed significant differences in RDW and WBC count. At a value of 2.5 µiU/ml as an upper limit of reference range, people with TSH higher than this level showed higher BMI, Triglycerides and WBC in comparison to those having lower TSH values [54][55][56].
BMI is well known to be associated with cardiovascular and coronary heart diseases and their associated risk factors such as hypertension and hyperlipidemia [57,58].
Hypertriglyceridemia is one of the criteria of the metabolic syndrome and is strongly associated with atherosclerosis, hypercoagulable and proinflammatory states [59].
Breakdown of triglycerides leads to release of Free Fatty Acids (FFA) into the circulation. Long term exposure of beta-cell of the pancreas to FFA and deposition of triglycerides in the pancreatic islets leads to suppression of insulin secretion [60].
Atherosclerosis is associated with an underlying inflammatory process. WBC is considered as one of the inflammatory markers that can predict the future development of a cardiovascular insult. Higher WBC is associated with higher body weight, systolic blood pressure, and insulin level; all of which are cardiovascular risks. Several studies  found that higher WBC is an independent predictor of coronary heart disease in diabetic patients in our study, higher WBC was found in association with TSH value higher than 2.5 uIU/ml indicating a higher risk of cardiovascular disease [61,62].
Lower reference values of thyrotropin such as 3 or 2.5 uIU/ml were previously suggested, however, this suggestion was criticized. Lowering the upper limit of TSH reference was favored by the observation of the presence of thyroid autoantibodies in subjects with TSH level between 3 to 5 uIU/ml, and those were at the highest risk to develop clinical thyroid dysfunction [57,58,63,64].
In NHANES III, mean serum TSH was 1.5 uIU/ml in healthy persons who did not have goiter, thyroid autoantibodies or family history of thyroid abnormality and highest value was 2.5 uIU/mlin. Similarly, African-Americans, known with the lowest susceptibility to Hashimoto thyroiditis; had a mean TSH of 1.18 uIU/ml indicating the vulnerability of this figure as a mean for healthy people [69,70].
In our study, TSH, at different cut off values, is associated with different risks for atherosclerosis and CVD. So, it is reasonable to recommend a TSH of 2.5 µiu/ml as an upper limit for normal thyroid function in type 2 diabetic population and advice a TSH level ≤ 1.68 uIU/ml as a target for diabetic patients on thyroxine replacement. We agree with other authors (2,70) in that decreasing the upper limit of reference range of TSH to 2.5 uIU/ml will help early detection of patients at risk of thyroid hypofunction with follow up and early management to avoid any complication.

Conclusion
Our study indicates that a routine screening for thyroid dysfunction is justified in T2DM as it is associated with many CVDS risks. We recommend a TSH of > 2.5 uIU/ml as an upper limit to diagnose thyroid hypofunction in T2DM. We also advise that diabetic patients who are receiving thyroid replacement therapy to attain a TSH level ≤ 1.68 uIU/ml.