Relationship between TSH and free thyroxine in outpatient cancer patient population

The inverse log-linear relationship between Thyroid-stimulating hormone (TSH) and free thyroxine (FT4) is well established and reliably used for evaluation of hypothalamus-pituitary-thyroid (HPT) axis function. However, there are limited data regarding oncologic states in the TSH-FT4 relationship. The purpose of this study was to evaluate thyroid pituitary hypothalamic feedback regulation by the inverse log TSH and FT4 relationship in the cancer patient population at the Ohio State University Comprehensive Cancer Center (OSUCCC-James). This retrospective study analyzed the correlation between TSH and FT4 results from 18846 outpatient subjects collected in August 2019-November 2021 at the Department of Family Medicine (OSU Wexner Medical Center), Department of Oncology (OSUCCC-James). Patients with diagnoses related to cancers were included in the oncology group. Patients with diagnoses not related to cancers were included in the non-oncology group. Patients of the Department of Endocrinology, Department of Cardiology, Department of Obstetrics & Gynecology and Department of Hematology were excluded from this study. Time of collection for TSH and FT4 was from 7am to 7 pm. Data were analyzed by morning (7am–12pm) and afternoon (12pm–7pm). Spearman correlation and non-linear fit were used for data analysis. Sex differences were analyzed as well in each group. Overall, an inverse correlation was observed between TSH and FT4 in both groups (non-oncology and oncology) regardless of sample collection time and sex differences. Further analysis by linear model in log TSH and FT4 showed a significant inverse fit in males compared with females in the group of oncology, both in the afternoon (p < 0.05). Data were further analyzed by ranges of FT4, as lower or higher (pathophysiology) or within (physiology) the reference interval of FT4. There was no statistical significance between the non-oncology and oncology groups, but relatively good correlation in non-oncology group in either physiologic or pathophysiologic FT4 levels and sample collection time. Interestingly, the best correlation between TSH and FT4 was found in the non-oncology group at pathophysiologic FT4 concentrations (abnormally high). In addition, at pathophysiologic FT4 concentrations (abnormally low), the oncology group demonstrated a significant TSH response in the morning than in the afternoon (p < 0.05). Though overall the TSH-FT4 curves showed an inverse relationship, there are variations of TSH-FT4 relationship for collection times when considering FT4 in physiologic or pathophysiologic states. The results advance understanding of TSH response, which is beneficial for the interpretation of thyroid disease. We recommend re-evaluation for interpretation of pituitary hypothalamic axis by TSH results when FT4 is abnormally high in oncology patients or low in non-oncology patients, due to poor predictability and the potential for misdiagnosis. A better understanding of the complex nature of the TSH-FT4 relationship may need further study with better defining subclinical states of cancer patients.


Introduction
Thyroid hormones are critical in energy homoeostasis and tightly controlled in a normal population.Regulation of thyrotropin-releasing hormone (TRH), thyroid-stimulating hormone (TSH) and thyroid hormones is mainly accomplished by a negative thyroid pituitary hypothalamic feedback loop.The inverse log-linear relationship between serum TSH and the circulating thyroid hormones is the essential aspect of the regulation of thyroid function regarding diagnosis of thyroid disorders [1].Therefore, a minor change of serum free thyroxine (FT4) level results in a significant magnitude of changes in serum TSH, which may lead to significant larger portion of patients to be determined as hyper-or hypothyroid in clinical or subclinical settings [2].The understanding of the regulation of the hypothalamic-pituitary-thyroid axis during the past years has spurred the development of extremely sensitive and accurate thyroid function tests (e.g., TSH) [3].Due to the high sensitivity and accuracy of the modern TSH assay, TSH has been determined to be the principal diagnostic marker of systemic thyroid status [3].
The negative thyroid pituitary hypothalamic feedback loop includes the response of the thyroid to TSH stimulation and the sensitivity of TRH neurons to thyroid hormone feedback.Therefore, each individual might have a unique set point for hypothalamus-pituitary-thyroid (HPT) axis function [4,5].The precise nature of the relationship between TSH and T4 remains controversial [6,7] because the relationship varies and is affected by other factors such as age, sex, medications or smoking among individuals [7][8][9].While the HPT axis is well understood under physiologic conditions, alterations in HPT axis function under pathophysiologic conditions is not well understood.Here we use cancer pathogenesis as a representative example of such states.Many patients with a cancer diagnosis present with new or chronic thyroid issues [10,11].A frequently observed cause lies in the chemotherapies used to treat cancer.Targeted therapies such as tyrosine kinase inhibitors are known to dysregulate thyroid function in both euthyroid and hypothyroid individuals [12].Other small molecule inhibitors, including sunitinib and sorafenib have also been shown to induce hypothyroidism in previously euthyroid patients [13,14].Further, the presence of several cancer types alone, including breast and lung cancer, has been linked to causing thyroid dysfunction independent of other variables [15,16].Correspondingly, clinical and subclinical hypothyroidism or hyperthyroidism has been implicated in various common cancers including prostate cancer, lung cancer, gastrointestinal cancers, and breast cancer [17][18][19]; however, some of these cases of thyroid malfunction might be incidentally identified as thyroid disorders due to a result of their cancer status or treatment [20].Nevertheless, underestimated hypothyroidism and hyperthyroidism may pose a problem to oncologists treating patients with symptoms related to cancer pathogenesis and treatment, which may result in reducing medication doses unnecessarily or putting treatment temporarily on hold.Diagnosis and treatment of underlying thyroid disorders is imperative to foster care for our patient whole condition contributing to symptoms and ultimately improve quality of life.The aim of this study is thus to explore the relationship between TSH and FT4 in pathophysiologic states using the population of patients with cancer as a paradigmatic example.Ultimately, we aim to advance our understanding of the treatment and challenges of laboratory diagnosis of thyroid disease in patients with cancer.

Patients and data collection
Data of TSH and FT4 testing were obtained from the laboratory database of 18846 outpatient subjects collected in August 2019-November 2021 at the Department of Family Medicine (OSU Wexner Medical Center), Department of Oncology [OSU Comprehensive Cancer Center (OSUCCC-James)].This study was approved by the Institutional Review Board of the Ohio State University (#2022C0039).Patients with diagnoses related to cancer were included in the oncology group.Patients with diagnoses not related to cancer were included in the nononcology group.Patient data likely having factors confusing the physiologic HPT axis response were excluded.Therefore, hospitalized patients, pediatric patients, outpatients of the Department of Endocrinology, Department of Cardiology, Department of Obstetrics & Gynecology and Department of Hematology are excluded from this study.The HPT axis is maintained in hypothyroid and euthyroid patients taking prolonged thyroid hormone supplementation, as demonstrated by the presence of a log-linear relationship between TSH and FT4 [4].On account of this, we did not stratify our patient population by thyroid hormone supplementation.Time of collection for TSH and FT4 was from 7am to 7 pm.Data were analyzed by morning (7am-12pm) and afternoon (12pm-7pm).Sex differences were analyzed as well in each group.

Laboratory methods
Laboratory evaluation included measurement of TSH and FT4 from the same specimen.TSH was measured with a sandwich chemiluminescent immunoassay on the Siemens Centaur/Atellica.FT4 was measured with a competitive chemiluminescent immunoassay on the Siemens Centaur/ Atellica.The analytical measurement range of TSH was 0.008-150.000µIU/mL.The reference interval of TSH in normal subjects was 0.550-4.780µIU/mL for adults >18 years old.The analytical measurement range of FT4 was 0.40-6.00ng/dL.The reference interval of FT4 in normal subjects was 0.89-1.76ng/dL for adults >18 years old.Standard laboratory quality evaluation procedures and regular participation at inter-laboratory tests were routinely employed, as part of the quality management.

Statistics
The log TSH-FT4 linear curve fitting model and Spearman correlation were performed using GraphPad Prism version 9.20 (GraphPad Software, San Diego, CA, USA).One-way ANOVA was performed for slopes comparison using GraphPad Prism version 9.20 (GraphPad Software, San Diego, CA, USA).

Results
In total, 18,846 (Male = 6897; Female = 11,949) patients were included in the study.Demographic data are listed in Table 1.
Correlation of TSH and FT4 values is shown in Fig. 1, stratified by group, time of collection, and sex.Overall, an inverse correlation was observed between TSH and FT4 in the morning and afternoon for both male and female populations in both oncology and non-oncology groups.Spearman correlation shows a more negative relationship between TSH and FT4 in males in the PM (afternoon) nononcology group (Rho = −0.44)compared to the AM (morning) non-oncology group (Rho = −0.36).This pattern is also reversed in females in the AM non-oncology group (Rho = −0.48)and the PM non-oncology group (Rho = −0.42).Similarly, the inverse patterns were observed in the oncology group (Rho ranged from −0.43 to −0.46) in each subgroup.Continued analysis of this data by linear model showed a statistically significant slope comparison (p = 0.003) in the inverse T4 and log TSH correlation between males and females in PM oncology groups (y = −1.21x+ 1.65 for males vs y = −1.03x+ 1.37 for females in oncology PM group).There were no other statistically significant differences in linear fit between other groups.
The relationship between TSH and FT4 was further analyzed with data split by time of blood collection and ranges of FT4 (Fig. 2).Ranges include values of FT4 lower, within, and higher than the reference interval (RI) in normal subjects.The lower and higher RI FT4 ranges represent pathophysiologic status of thyroid, while within RI FT4 range is for physiologic status.An inverse correlation was observed between log TSH and FT4 in all the groups.The inverse relationship between log TSH and FT4 showed a better correlation in the non-oncology group with abnormally high FT4 in the morning (Rho = −0.49)and afternoon (Rho = −0.43)than all other groups.Within the oncology group, the best correlations are seen with abnormally low FT4 in the morning (Rho = −0.34),and normal FT4 in the morning (Rho = −0.33)and afternoon (Rho = −0.35).In addition, the AM oncology group with abnormally low FT4 demonstrated the most inverse TSH-FT4 relationship compared with the same group in the afternoon (y = −2.55x+ 2.65 AM vs. y = −1.96x+ 2.12 PM) (p = 0.0369).Notably, the goodness-of-fit measure, R 2 , was lowest among the non-oncology groups with abnormally low FT4 and oncology with abnormally high FT4.There was otherwise no significant variation between the inverse relationship slopes in other groups.

Discussion
The relationship between TSH and free T4 is well established in healthy populations, however, is not as clearly understood during pathophysiologic states, such as malignancy.The population of patients with cancer in particular poses unique challenges to evaluating thyroid function and diagnosing thyroid disorders.Our results demonstrated variations of the TSH-FT4 relationship when FT4 was at physiologic or pathophysiologic concentrations.Overall, we did not observe a statistically significant difference in the TSH-FT4 relationship between oncology and non-oncology groups when comparing collection time and sex of patients.However, when comparing the two groups based on FT4 at physiologic or pathophysiologic concentrations, the best correlation between TSH and FT4 was found in the nononcology group at pathophysiologic FT4 concentrations (abnormally high).At pathophysiologic FT4 concentrations (abnormally low), the oncology group demonstrated a significant inverse TSH response in the morning than in the afternoon.
Sex steroids may also impact thyroid function.When serum thyroid binding globulin (TBG) increased resulting from elevated estrogen by contraceptives, hormone replacement or ovarian hyperstimulation, the equilibrium of T4 and could be changed.When more T4 bind with elevated TBG, FT4 levels decreased which might affect the HPT axis function [21].In transgender female-to-male patients treated with testosterone, serum TBG levels decreased [22], resulting in elevated FT4 levels.In a study in men with prostate cancer initiating androgen deprivation therapy as a model to assess the effect of androgen deprivation on thyroid function, HPT axis was changed.This change was found uncorrelated with serum testosterone or estradiol concentrations but strongly related to the changes of patients' weight, BMI, fat mass and leptin levels [23].Our data demonstrated that the variation between sexes is sustained in cancer states.The analysis by linear model in log TSH and FT4 showed a significant inverse fit in males compared with females in the group of Oncology both in the afternoon.This result indicates that it needs relatively lower FT4 levels in males than females to trigger the same response of TSH.The hypothesis is that reducing FT4 levels in females due to elevated TBG might suppress the response of TSH which results in that it needs more FT4 in females to trigger the same response of TSH like in males.However, we did not observe the same scenario in other groups in our study, and increased TBG in females might or might not be the explanation for that.In addition, TBG was elevated in cancer patients without clear mechanism [24], while TBG was also reported reduced in patients with systemic metastases [25].Further studies targeting TBG in specific cancer might be helpful to explain that.Fluctuations in TSH and FT4 in cancer populations with an abnormal circadian rhythm might also complicate the TSH-FT4 relationship.In humans, circulating TSH levels demonstrate a clear daily rhythm, increasing levels in the late afternoon or early evening that peaking in the early night [26].This normal TSH rhythm can be interrupted by cancers [27].In primary hyperthyroidism, circadian rhythmicity of circulating TSH levels was not expected as TSH is suppressed, thus preventing the physiologic diurnal rhythm of thyroid hormones production controlled by TSH [28].Respectively, the quotidian rhythm is most prevalent in groups of hypothyroidisms, or FT4 less than the reference interval --a trend that has been evidenced by prior studies regarding thyroid function and diurnal rhythm [26].In our study, when FT4 was at pathophysiologic concentrations (abnormally low), the oncology group demonstrated a significant inverse TSH response in the morning than in the afternoon.The TSH-FT4 relationship may require special attention on time of collection for clinical practice, though subclinical states of cancer patients need to be defined for further analysis.
In addition to the variables we examined, there is a multitude of known complications for managing thyroid disease in patients with cancer.For example, prominent Please note the equations representing the best fit linear curve for each respective panel cancer therapies such as tyrosine kinase inhibitors induce thyroid dysfunction by gene expression related to TSH secretion, complicating treatment in these individuals [20,29].Moreover, the constant inflammatory state of patients with cancer releases cytokines which interrupt thyroid hormone synthesis and expression [30] and can decrease sodium-iodine symporter activity [31].Thus, our study explored the TSH-FT4 relationship in patients with cancer in order to gain insight into this unique population, as well to better understand the effects of thyroid function in pathophysiologic states.Taking this into account, we recommend interpreting patient TSH and FT4 within specific contexts of sex, blood collection time, BMI, cancer therapy, cancer type, systemic metastases, etc. to improve the accuracy.Multiple TSH and FT4 testing might be needed to reduce some variations due to the above discussion.Ultimately, more research is needed to better understand the complex nature of the TSH and FT4 relationship in patients with cancer or cancer treatment and other pathophysiologic states.H Data of FT4 within reference interval in the oncology group, with blood collected in the afternoon (n = 5929); I Data of FT4 higher than reference interval in the non-oncology group, with blood collected in the morning (n = 115); J Data of FT4 higher than reference interval in the non-oncology group, with blood collected in the afternoon (n = 75); K Data of FT4 higher than reference interval in the oncology group, with blood collected in the morning (n = 626); L Data of FT4 higher than reference interval in the oncology group, with blood collected in the afternoon (n = 466).Please note the equations representing the best fit linear curve for each respective panel

Fig. 1
Fig. 1 Relationship between log TSH and FT4 levels in the various groups.The respective lines represent the linear curve fitted to the data points.A Male subjects in the non-oncology group with blood collected in the morning (n = 410); B Male subjects in the non-oncology group with blood collected in the afternoon (n = 320); C Female subjects in the non-oncology group with blood collected in the morning (n = 1273); D Female subjects in the non-oncology group with blood collected in the afternoon (n = 1094); E Male subjects in

Fig. 2
Fig. 2 Relationship between log TSH and FT4 levels across three distinct segments of FT4 values in the various groups.Data were split to three distinct segments by the reference interval (RI) of FT4.The respective lines represent the linear curve fitted to the data points.A Data of FT4 lower than reference interval in the non-oncology group, with blood collected in the morning (n = 89); B Data of FT4 lower than reference interval in the non-oncology group, with blood collected in the afternoon (n = 100); C Data of FT4 lower than reference interval in the oncology group, with blood collected in the morning (n = 682); D Data of FT4 lower than reference interval in the oncology group, with blood collected in the afternoon (n = 531); E Data of FT4 within reference interval in the non-oncology group, with blood collected in the morning (n = 1479); F Data of FT4 within

Table 1
Demographic data of subjects enrolled