Risk Factors for New Hypothyroidism During Tyrosine Kinase Inhibitor Therapy in Advanced Nonthyroidal Cancer Patients

Abstract

Background:

Thyroid dysfunction during tyrosine kinase inhibitor (TKI) cancer treatment is common, but predisposing risk factors have not been determined. Recommendations for monitoring patients treated with one or multiple TKI and in conjunction with other relevant cancer therapies could be improved. The study objective was to assess the risk factors for new thyroid dysfunction in TKI-treated previously euthyroid cancer patients.

Methods:

A retrospective cohort study of patients with advanced nonthyroidal cancer treated with TKI from 2000 to 2017, having available thyroid function tests showing initial euthyroid status, excluding patients with preexisting thyroid disease or lack of follow-up thyroid function tests. During TKI treatment, patients were classified as euthyroid (thyrotropin [TSH] normal), subclinical hypothyroidism (TSH 5–10 mIU/L, or higher TSH if free thyroxine normal), or overt hypothyroidism (TSH >10 mIU/L, low free thyroxine, or requiring thyroid hormone replacement). The timing of thyroid dysfunction and TKI used were assessed. Risk factors for incident hypothyroidism were evaluated using multivariate models.

Results:

In 538 adult patients included, subclinical hypothyroidism occurred in 71 (13.2%) and overt hypothyroidism occurred in 144 (26.8%) patients with TKI therapy, following a median cumulative TKI exposure of 196 days (interquartile range [IQR] 63.5–518.5 days). The odds of hypothyroidism were greatest during the first six months on a TKI. Median exposure time on the TKI concurrent with thyroid dysfunction in patients treated with only one TKI was 85 days (IQR 38–293.5 days) and was similar to the 74 days (IQR 38–133.3 days) in patients treated previously with other TKI (p = 0.41). Patients who developed hypothyroidism compared to those who remained euthyroid had greater odds of being female (odds ratio = 1.99 [confidence interval 1.35–2.93], p < 0.01), but greater cumulative TKI exposure and greater number of TKI received were not associated with thyroid dysfunction.

Conclusions:

Thyroid dysfunction occurred in 40% of euthyroid patients. Monitoring thyroid function in TKI-treated patients is recommended, with particular attention to female patients and within the first six months of exposure to a new TKI.

Introduction

Tyrosine kinase inhibitors (TKI) are small-molecule reagents used in cancer therapy that target numerous cell proliferation and survival pathways. With more widespread application of these drugs, thyroid dysfunction is an increasingly recognized side effect (1). Multiple mechanisms for thyroid dysfunction have been postulated, including direct toxicity of TKI on follicular cells, induction of destructive thyroiditis, increased hormone clearance, thyroid capillary regression by vascular endothelial growth factor (VEGF) inhibition, and impaired iodide uptake (2).

Thyroid abnormalities have been reported in 8–85% of patients treated with TKI, including commonly axitinib, pazopanib, sorafenib, and sunitinib (1,3 –9). While the prevalence of thyroid abnormalities has been reported previously, risk factors for TKI-related hypothyroidism have not been determined. Furthermore, there are no standard recommendations for thyroid function monitoring in TKI-treated patients, especially those who have received more than one TKI drug. Most existing data come from clinical trials of cancer therapy, which frequently include patients with preexisting thyroid disease and evaluate only a single TKI (1,2). In clinical practice, cancer patients are often treated with serial agents over the course of their illness due to disease progression or side effects.

Existing data on the risk associated with cumulative TKI exposure are conflicting. Wolter et al. (10) proposed measuring thyrotropin (TSH) on days 1 and 28 of the first four cycles of sunitinib because abnormalities of thyroid function tests (TFTs) were most often noted to occur early. In contrast, several studies with sunitinib have suggested that the risk of developing hypothyroidism increases with time and cycles of therapy (10 –12), though cumulative TKI exposure was not specifically evaluated. Additionally, many patients receive other anticancer drugs that may cause thyroid dysfunction, such as VEGFR and checkpoint inhibitors, the effect of which in conjunction with TKI treatment is unknown.

This study analyzed the pattern of hypothyroidism during TKI therapy in the largest cohort to date of initially euthyroid cancer patients. Specifically, it sought to investigate the influence of multiple TKI exposures, cumulative TKI treatment time, patient sex, and relevant non-TKI therapy to address current limitations in the understanding of this clinical entity. More information about risk factors for TKI-associated thyroid dysfunction and the timing of its occurrence will better inform clinicians caring for these patients and recommendations for monitoring.

Methods

Patients and data collection

This was a retrospective cohort study of adult patients with advanced nonthyroidal cancer treated with common TKI and available TFT at three affiliated academic hospitals (Brigham and Women's Hospital, Dana Farber Cancer Institute, and Massachusetts General Hospital, Boston, MA), receiving first TKI treatment between August 2000 and August 2013. Institutional Review Board approval was obtained (Partners Protocol 2014P001974). Patient electronic medical records were reviewed for all data and follow-up through May 2017. Data collection included patient demographics, cancer diagnosis, stage (American Joint Committee on Cancer criteria where applicable), Eastern Cooperative Oncology Group (ECOG) performance status, treatment with antibody inhibitors of VEGFR (bevacizumab or ramucirumab) and immune checkpoint inhibitors, and date of death or last follow-up. TKI treatment data recorded included each TKI name and dates received during the study period (approximated as the first of the month if not otherwise specified). Patients who underwent bone-marrow transplantation were excluded.

The thyroid function data collected included history of thyroid disease, treatment with thyroid hormone or antithyroid drugs, serum TSH and free thyroxine (fT4) levels, date of first thyroid abnormality, and highest serum TSH (before replacement) with TKI therapy. Thyroid peroxidase antibodies (TPOAb) or thyroid ultrasound results were collected but were only available in a small minority of patients. Patients with preexisting thyroid dysfunction, including abnormal TFT, a preexisting diagnosis of thyroid disease or use of thyroid medication, thyroid cancer, or prior thyroid resection were excluded. Thyroid status with TKI treatment was determined from TFT and initiation of thyroid medication.

Included subjects were euthyroid at baseline and were classified during TKI treatment as euthyroid (TSH normal), subclinical hypothyroidism (SCH; TSH >5 and <10 mIU/L, or higher TSH if fT4 normal), or overt hypothyroidism (OH; TSH >10 mIU/L and low fT4 if available, or requiring thyroid hormone replacement). Incident hypothyroidism was defined as SCH or OH while on TKI or within six months of TKI therapy because of the nonstandard timing of TFT in this retrospective setting. Patients without TFT before or during therapy, or in whom thyroid status before and during treatment could not reliably be determined, were excluded. Additional details are provided in the Supplementary Methods (Supplementary Data are available online at www.liebertpub.com/thy).

Statistical analysis

Baseline demographic and clinical data were stratified by thyroid status. Mean and standard deviation for continuous normally distributed variables, median and interquartile range (IQR) for non-normal continuous variables, and number and percentage for categorical variables were calculated. Statistical comparisons were made using the Student's t-test (parametric) or the Mann–Whitney U-test (nonparametric) for continuous variables and Pearson's chi-square test or Fisher's exact test for categorical variables. Total and individual cumulative TKI exposure while at risk for developing hypothyroidism was calculated. Unadjusted and adjusted logistic regression analysis was performed for patients with hypothyroidism compared to euthyroid patients, including demographic and clinical parameters, TKI exposure time, and total time at risk of developing hypothyroidism. Collinearity diagnostics (variance inflation factor, condition index and principal component analysis) for highly inter-correlated variables were performed. In secondary analyses, patients who received only one TKI during their total risk period were similarly evaluated.

To find the association between individual TKI and risk of hypothyroidism, each predictor was analyzed in the univariate logistic regression model and statistically significant variables were included in the multivariate logistic regression models. Cumulative TKI exposure and time at risk in the regression model were not combined because of multicollinearity between these variables (variance inflation factor: 8.97; condition index: 15.06; 89% variability in the principal component analysis due to these two parameters). To evaluate the relationship between cumulative TKI exposure and follow-up time at risk, the ratio of these was calculated in patients with hypothyroidism compared to those who were euthyroid. Odds ratios and confidence intervals (CI) were calculated using logistic regression model. A two-tailed p-value of <0.05 was used for statistical significance. Analyses were performed using SAS v9.3 (SAS Institute, Cary, NC) and GraphPad Prism v6.0 (GraphPad Software, Inc., La Jolla, CA).

Results

From 1120 initial patients, 538 remained after considering exclusion criteria and were included in the final analysis. Incident hypothyroidism occurred in 215 (40%) patients with TKI therapy, including 71 (13.2%) with SCH and 144 (26.8%) with OH. Two (0.3%) patients only had follow-up TFT showing hyperthyroidism without additional available testing and were not included in analyses of hypothyroid groups. Though the frequency of thyroid function monitoring was limited in this retrospective study, a biphasic pattern of hyperthyroidism followed by persistent hypothyroidism consistent with thyroiditis was seen in 18 (12.5%) patients with TKI-related hypothyroidism.

The comparison of demographic factors and tumor type among euthyroid, SCH, and OH groups is shown in Table 1. Women were more likely to become hypothyroid than were men (94/178 [52.8%] vs. 121/358 [33.7%]; p < 0.0002). The incidence of hypothyroidism varied among cancer types present in the study cohort (p < 0.0001), including a greater proportion of patients with renal-cell carcinomas and gastrointestinal stromal tumors (GIST) in the OH group. A sub-analysis of GIST tumors showed no significant difference in frequency of hypothyroidism compared to other cancers (Supplementary Data). There were no differences among euthyroid, SCH, and OH groups in patient race/ethnicity, cancer stage, or ECOG status at baseline. Median peak TSH in SCH patients was 6.5 mIU/L (IQR 5.9–7.7 mIU/L) versus 12.9 mIU/L (IQR 10–22.7 mIU/L) in OH patients prior to thyroid hormone replacement (p < 0.0001). Of OH patients, 132/144 (91.7%) received thyroid hormone replacement. Treatment was initiated at the discretion of the treating physician for TSH elevation most often accompanied by symptoms including fatigue, depression, and declining functional status.

Table 1.

Patient Demographic Data and Characteristics for Total Study Population, Stratified by Thyroid Status

Characteristic	Total study population (n = 536)	Euthyroid (n = 321)	SCH (n = 71)	OH (n = 144)	p-Value
Median age, years (IQR)	60.2 (52.7–68.3)	60.0 (52.4–68.6)	65.3 (58.3–71.8)	59.3 (50.8–66.8)	0.0017^a,b
Sex (n)					0.0002^a,c
Female	178 (33.2)	84 (26.2)	29 (40.8)	65 (45.1)
Male	358 (66.8)	237 (73.8)	42 (59.2)	79 (54.9)
Cancer type (n)					<0.0001^b,c
Renal cell	253 (47.2)	127 (39.6)	40 (56.3)	86 (59.7)
GIST	88 (16.4)	52 (16.2)	7 (9.9)	29 (20.1)
Hepatocellular	40 (7.5)	32 (10.0)	5 (7.0)	3 (2.1)
Neuroendocrine	48 (9.0)	37 (11.5)	6 (8.5)	5 (3.5)
Colorectal	5 (0.9)	4 (1.3)	1 (1.4)	0 (0.0)
Sarcoma	73 (13.6)	48 (15.0)	7 (9.9)	18 (12.5)
Other carcinoma	23 (4.3)	15 (4.7)	5 (7.0)	3 (2.1)
Primary CNS	6 (1.1)	6 (1.9)	0 (0.0)	0 (0.0)
Stage at start (n)					ns
I/II	5 (0.9)	3 (0.9)	1 (1.4)	1 (0.7)
III/IV	516 (96.3)	306 (95.4)	69 (97.2)	141 (97.9)
Not applicable	15 (2.8)	12 (3.7)	1 (1.4)	2 (1.4)
ECOG at start (n)					ns
0	268 (50.0)	146 (45.5)	44 (62.0)	78 (54.2)
1	186 (34.7)	123 (38.3)	21 (29.6)	42 (29.2)
2	35 (6.5)	25 (7.8)	4 (5.6)	6 (4.2)
3	4 (0.8)	4 (1.3)	0 (0.0)	0 (0.0)
ND	43 (8.0)	23 (7.2)	2 (2.8)	18 (12.4)
Race/ethnicity, (n)					ns
Asian	19 (3.5)	9 (2.8)	4 (5.6)	6 (4.1)
Black	14 (2.6)	10 (3.1)	2 (2.8)	2 (1.4)
Hispanic	11 (2.1)	8 (2.5)	0 (0.0)	3 (2.1)
White	485 (90.5)	289 (90.0)	65 (91.6)	131 (91.0)
Unknown	7 (1.3)	5 (1.6)	0 (0.0)	2 (1.4)
VEGFR inhibitor					0.0422
Yes	63 (11.7)	48 (15.0)	5 (7.0)	10 (6.9)
No	473 (88.3)	273 (85.0)	66 (93.0)	134 (93.1)
Checkpoint inhibitor					0.9602
Yes	18 (3.4)	10 (3.1)	3 (4.2)	5 (3.4)
No	518 (96.6)	311 (96.9)	68 (95.8)	139 (96.6)

p-Values for evaluation of significant difference among euthyroid, SCH, and OH are shown in the far right column, with significant differences for pairwise comparisons indicated by superscript: ^aeuthyroid vs. SCH; ^bSCH vs. OH; and ^ceuthyroid vs. OH.

SCH, subclinical hypothyroidism; OH, overt hypothyroidism; IQR, interquartile range; GIST, gastrointestinal stromal tumor; CNS, central nervous system; ECOG, Eastern Cooperative Oncology Group; VEGFR, vascular endothelial growth factor receptor.

TKI, cumulative exposure, and risk of hypothyroidism

Patients included in this study received 21 different TKI, with axitinib, cabozantinib, imatinib, nilotinib, pazopanib, regorafenib, sorafenib, and sunitinib being the most common. Table 2 shows the number of patients who ever received each TKI in the study period and how many of these patients became hypothyroid while on that drug (i.e., for whom it was the insult TKI). It is acknowledged that in this retrospective study, the frequency with which different TKI were used was influenced by historical clinical practice and available TKI, with fewer patients receiving newer TKI. While acknowledging this limitation, Table 2 shows variability among TKI in the frequency of new hypothyroidism observed in patients during treatment. Specifically, axitinib, cabozantinib, cediranib, pazopanib, and sunitinib demonstrated higher frequencies of hypothyroidism during treatment, while dasatinib, imatinib, and sorafenib showed lower frequencies. The small numbers of patients treated with some of the included TKI and the many unique combinations of TKI received precluded a regression analysis specifically evaluating each TKI and the development of hypothyroidism. However, for a subset of patients treated with only one of the more common TKI (i.e., pazopanib, sorafenib, or sunitinib), the relative frequency and timing of TKI-associated hypothyroidism were compared, as shown in Supplementary Table S1. The incidence of hypothyroidism was significantly greater for monotherapy with sunitinib (48.5%) or pazopanib (42.5%) than it was with sorafenib (15.4%) (p < 0.05).

Table 2.

Number of Patients who Received each TKI in the Study Period, and How Many Became Hypothyroid While on that Drug

		No. of patients who became hypothyroid while on this medication
TKI received while at risk	No. of patients who ever received drug (n = 536)	Euthyroid	SCH	OH
Axitinib	44 (8.2)	29 (65.9)	5 (11.4)	10 (22.7)
Cabozantinib	14 (2.6)	5 (35.7)	3 (21.4)	6 (42.9)
Cediranib	4 (0.7)	2 (50.0)	1 (25.0)	1 (25.0)
Dasatinib	8 (1.5)	8 (100)	0	0
Erlotinib	3 (0.6)	3 (100)	0	0
Foretinib	2 (0.4)	2 (100)	0	0
Imatinib	86 (16.0)	85 (98.8)	0	1 (1.2)
Lapatinib	1 (0.2)	1 (100)	0	0
Lenvatinib	1 (0.2)	1 (100)	0	0
Linifanib	2 (0.4)	2 (100)	0	0
MSC20151	1 (0.2)	1 (100)	0	0
Neratinib	1 (0.2)	1 (100)	0	0
Nilotinib	17 (3.2)	17 (100)	0	0
Pazopanib	125 (23.3)	88 (70.4)	15 (12.0)	22 (17.6)
Ponatinib	2 (0.4)	1 (50.0)	1 (50.0)	0
Regorafenib	26 (4.9)	20 (76.9)	2 (7.7)	4 (15.4)
Semaxanib	1 (0.2)	1 (100)	0	0
Sorafenib	122 (22.8)	110 (90.2)	9 (7.4)	3 (2.5)
Sunitinib	371 (69.2)	240 (64.7)	34 (9.2)	97 (26.1)
Tivozanib	2 (0.4)	1 (50.0)	1 (50.0)	0
XL-820	5 (0.9)	5 (100)	0	0

For the total population and each subgroup (euthyroid, SCH, and OH), the number of patients ever treated with each TKI (far left, with percentage of the total study number) and the number of patients receiving each drug who remained euthyroid or developed thyroid abnormalities while on that TKI (i.e., for whom it was the insult TKI) are shown.

As shown in Table 3, patients who developed thyroid dysfunction had a median time to abnormal TFT of 196 days (IQR 64–833 days) for SCH and 252 days (IQR 69.5–688.5 days) for OH compared to a median time at risk of 612 days (IQR 296–1376 days) for patients who remained euthyroid (p < 0.0001). There was no difference in the median cumulative TKI exposure days among patients who remained euthyroid or developed SCH or OH (median 237 days [IQR 95–658 days] vs. median 195 days before abnormal TFT [IQR 63–586 days] vs. median 202.5 days before abnormal TFT [IQR 61–518.5 days]; p = 0.46], nor in the median cumulative TKI drug number received (p = 0.16).

Table 3.

TKI Exposure in Time and Drug Number While at Risk of Hypothyroidism, Stratified by Thyroid Status

Characteristic	Total study population (n = 536)	Euthyroid (n = 321)	SCH (n = 71)	OH (n = 144)	p-Value
Median time at risk, days (IQR)	462.5 (154–1134)	612 (296–1376)	196 (64–833)	252 (69.5–688.5)	<0.0001^a,c
Median (IQR) cumulative TKI exposure
Days	215 (82–603)	237 (95–658)	195 (63–586)	202.5 (61–518.5)	0.4556
No. of drugs	1	1	1	1	0.1605
Cumulative TKI exposure (drug no.)					0.0712
1	334 (62.3)	194 (60.4)	49 (69.0)	91 (63.2)
2	138 (25.8)	80 (24.9)	14 (19.7)	44 (30.6)
3	36 (6.7)	25 (7.8)	4 (5.6)	7 (4.9)
4	20 (3.7)	15 (4.7)	3 (4.2)	2 (1.4)
5	6 (1.1)	5 (1.6)	1 (1.4)	0 (0.0)
6	2 (0.4)	2 (0.6)	0 (0.0)	0 (0.0)

Cumulative TKI exposure (defined as the sum of days during which patients received TKI treatment while at risk) was further evaluated, with exposure stratified by six-month increments and inclusive of all TKI. Patients who developed hypothyroidism did not have increased odds of greater TKI cumulative exposure days compared to euthyroid patients (Table 4). Time at risk for hypothyroidism (defined as the total amount of time from study start until development of abnormal TFT or study censor) was evaluated in six-month increments, and greater time at risk was associated with decreased odds of developing hypothyroidism (p < 0.01). Compared to euthyroid patients, those with hypothyroidism had decreased odds of being at risk for longer than 0–6 months (odds ratio [OR] = 0.21 [CI 0.12–0.38] for 6–12 months; OR = 0.20 [CI 0.11–0.38] for 12–18 months; OR = 0.33 [CI 0.16–0.68] for 18–24 months; and OR = 0.17 [CI 0.11–0.28] for >24 months; p < 0.01).

Table 4.

Development of Hypothyroidism (SCH + OH) Was Not Associated with More TKI Exposure

Variable	Category	OR [CI]	p-Value
Cumulative TKI exposure	0–6 months	1.00 (Ref)
	6–12 months	0.75 [0.46–1.23]	0.25
	12–18 months	1.02 [0.57–1.82]	0.95
	18–24 months	1.14 [0.49–2.64]	0.76
	>24 months	0.75 [0.47–1.18]	0.21
Cumulative time at risk of hypothyroidism	0–6 months	1.00 (Ref)
	6–12 months	0.21 [0.12–0.38]	<0.01
	12–18 months	0.20 [0.11–0.38]	<0.01
	18–24 months	0.33 [0.16–0.68]	<0.01
	>24 months	0.17 [0.11–0.28]	<0.01

Relative odds of cumulative TKI exposure and cumulative time at risk in patients with hypothyroidism (SCH + OH) compared to patients who remained euthyroid.

OR, odds ratio; CI, confidence interval.

While some cancer patients receive only one TKI, many are treated with serial different TKI during the course of therapy. In this study, thyroid dysfunction occurred after treatment with the initial TKI in some patients, while other patients developed TKI-related hypothyroidism after receiving previous TKI drugs. The TKI concurrent with or immediately preceding the detection of abnormal thyroid function was considered the “insult TKI.” In patients treated with single or multiple TKI, the frequency with which TKI were the “insult TKI” and the median exposure time on that insult TKI preceding abnormal TFT are shown in Table 5. The median exposure time (days) on the insult TKI was determined for patients who received only one TKI compared to those in whom the insult TKI was used subsequent to other TKI (Table 5). The median TKI exposure prior to thyroid dysfunction in patients treated with one TKI of 85 days (IQR 38–293.5 days) was similar to the median insult TKI exposure of 74 days (IQR 38–133.3) for patients treated with multiple TKI (p = 0.41).

Table 5.

TKI Exposure Days to Thyroid Dysfunction Was Similar Whether First or Subsequent TKI

Patients treated with one TKI	Incident hypothyroidism (SCH + OH)	Median (IQR) insult TKI exposure (days)	Patients treated with multiple TKI	Incident hypothyroidism (SCH + OH)	Median (IQR) insult TKI exposure (days)
Sunitinib	97/200 (48.5)	88 (35–260)	Sunitinib	34/173 (19.7)	103.5 (42–192.8)
Pazopanib	31/73 (42.5)	77 (37–420)	Pazopanib	6/52 (11.5)	76.5 (44.3–109.5)
Sorafenib	8/52 (15.4)	81 (47.3–519)	Sorafenib	4/72 (5.6)	43 (35.5–90.8)
Axitinib	1/1 (100)		Axitinib	14/43 (32.6)	69 (34–110.8)
Regorafenib	1/6 (16.7)		Regorafenib	5/22 (22.7)	106 (98–118)
Tivozanib	1/1 (100)		Cabozantinib	9/14 (64.3)	42 (42–64)
			Cediranib	2/4 (50.0)	35 (28–42)
Median exposure days 85 (38–293.5)			Median exposure days 74 (38–133.3)

Median insult TKI (TKI coincident with hypothyroidism) exposure preceding development of thyroid abnormalities in patients treated with one (left) or multiple (right) TKI drugs.

Effect of non-TKI cancer therapy on risk of hypothyroidism

Inhibition of VEGF signaling is a proposed mechanism of TKI-mediated thyroid dysfunction (13). In this study, 63/536 (11.7%) patients received non-TKI VEGFR inhibitor therapy while at risk, and patients who developed hypothyroidism were less likely to have received non-TKI VEGFR inhibitor therapy than patients who remained euthyroid (6.98% vs. 14.95% in euthyroid patients; p = 0.042; Table 1). In this study, checkpoint inhibitor treatment (which can also be associated with thyroid dysfunction in a subset of patients) was not associated with hypothyroidism (p = 0.96), though interpretation is limited by few patients receiving such treatment (18/536).

Multivariate analysis of risk

Univariate analysis for risk factors of hypothyroidism in cancer patients treated with TKI showed significant associations with age, sex, non-TKI VEGFR treatment, time at risk, and the percentage of total days at risk during which a patient received TKI therapy. Multivariate analysis (Table 6, left), showed increased odds of hypothyroidism with female sex (OR = 1.99 [CI 1.35–2.93]; p < 0.01) and decreased odds of hypothyroidism with increasing time at risk (OR = 0.98 [CI 0.96–0.99]; p < 0.01) or preceding non-TKI anti-VEGFR therapy (OR = 0.43 [CI 0.23–0.8]; p < 0.01). Due to the presence of colinearity between TKI exposure and time at risk variables, a separate multivariate analysis included the ratio of TKI exposure days to total risk days (Table 6, right). In this model, female sex, non-TKI anti-VEGFR therapy, and the percentage of total days on which TKI was received while at risk were all significant factors for the development of hypothyroidism. Patients with hypothyroidism had increasingly greater odds of receiving TKI therapy on 25–50% (OR = 3.45 [CI 1.04–11.39]; p = 0.01), 50–75% (OR = 5.74 [CI 1.85–17.8]; p < 0.01), or 75–100% (OR = 39.67 [CI 13.74–114.53]; p < 0.01) versus 0–25% of days compared to patients who remained euthyroid. Multivariate analyses of risk factors for hypothyroidism with stratification by SCH and OH are shown in Supplementary Tables S2 and S3.

Table 6.

Multivariate Analyses of Risk Factors for Hypothyroidism in Cancer Patients Treated with TKI

		Euthyroid vs. SCH + OH				Euthyroid vs. SCH + OH
Variable	Category	OR [CI]	p-Value	Variable	Category	OR [CI]	p-Value
Age		0.99 [0.98–1.01]	0.45	Age		0.99 [0.98–1.01]	0.45
Sex	Male	1.00 (Ref)	<0.01	Sex	Male	1.00 (Ref)	<0.01
	Female	1.99 [1.35–2.93]			Female	1.99 [1.35–2.93]
Checkpoint inhibitor	No	1.00 (Ref)		Checkpoint inhibitor	No	1.00 (Ref)
	Yes	1.95 [0.70–5.41]	0.20		Yes	1.95 [0.70–5.41]	0.2
VEGFR inhibitor	No	1.00 (Ref)		VEGFR inhibitor	No	1.00 (Ref)
	Yes	0.43 [0.23–0.80]	<0.01		Yes	0.43 [0.23–0.80]	<0.01
Time at Risk		0.98 [0.96–0.99]	<0.01	Cumulative TKI exposure over days at risk	0–25%	1.00 (Ref)
					25–50%	3.45 [1.04–11.39]	0.01
					50–75%	5.74 [1.85–17.82]	<0.01
					>75%	39.67 [13.74–114.53]	<0.01

Patients with hypothyroidism (SCH + OH) were compared to those who remained euthyroid. Univariate analyses for risk factors of hypothyroidism showed significant associations with age, sex, non-TKI treatments, time at risk, and the percentage of total days at risk during which a patient received TKI therapy. Two separate multivariate analyses were performed due to the reliance of percentage of time at risk TKI exposure occurred (right), on overall time at risk (left).

Discussion

TKI are small-molecule reagents increasingly used in the therapy of many cancers and associated with thyroid dysfunction. This study analyzed the pattern of incident hypothyroidism during TKI therapy in patients with nonthyroidal cancer to optimize recommendations for thyroid monitoring. To the authors' knowledge, this is the largest cohort to date reporting potential risk factors of thyroid function abnormalities in previously euthyroid cancer patients treated with TKI and examining the effects of multiple drugs.

In this study of 538 previously euthyroid patients with nonthyroidal cancer, nearly 40% developed hypothyroidism during or within six months of TKI treatment. In agreement with previous studies, these data suggest a need to monitor for thyroid dysfunction in TKI-treated patients. Further, the data suggest factors associated with greater risk for developing hypothyroidism during TKI therapy. Patients who developed hypothyroidism compared to those who remained euthyroid had greater odds of being female and decreased odds of longer time at risk, specifically beyond six months. Increased cumulative TKI exposure was not associated with thyroid dysfunction, nor was a greater number of TKI received. The median TKI exposure of the offending TKI in patients who became hypothyroid was no different between those who were treated with only one TKI (median 85 days [IQR 38–293.5 days]) compared to those who had received more than one TKI while at risk (median 74 days [IQR 38–133.3 days]). Therefore, when initiating a new TKI we recommend baseline and every six-week TSH monitoring until six months, and then every three to six months until 18 months, or sooner if dictated by symptoms concerning for thyroid disease. Additionally, since hypothyroidism was less frequent but nonetheless observed beyond 18 months after initiating a new TKI, intermittent screening for thyroid dysfunction is reasonable in patients on continued therapy. It is acknowledged that in this retrospective study, the frequency with which different TKI were used was influenced by clinical practice, and fewer patients received newer TKI during the study period. Hypothyroidism was more frequently observed in patients during treatment with axitinib, cabozantinib, cediranib, pazopanib, and sunitinib, and clinicians should maintain a high suspicion for thyroid dysfunction in patients on these drugs. While a significant association between immune checkpoint inhibitors and hypothyroidism was not observed, this is likely related to the small number of patients who received these novel drugs during the historical period studied. Given the risk of autoimmune-related adverse effects with immune checkpoint inhibitors, more frequent thyroid monitoring may be needed (14) when concurrently used with TKI.

Inhibition of VEGF in thyroid capillary vasculature has been suggested as a mechanism of TKI-mediated hypothyroidism, but treatment with specific VEGFR inhibitors (bezacizumab or ramucirumab) was not associated with hypothyroidism in this study. Data supporting this mechanism include demonstration of capillary attenuation with VEGFR-targeted TKI and VEGF blockade in murine studies (2,13). Those data show that serum TSH levels were increased with axitinib treatment but not with other mechanisms of VEGF blockade, suggesting that the observed effects might be due to non-VEGF effects of TKI (13). Further, non-TKI VEGFR inhibitors such as bevacizumab are not frequently associated with hypothyroidism in patients, despite apparent decreased thyroid perfusion, and may be protective through facilitating preferential platelet-derived growth factor signaling (15). Another possibility is that patients predisposed to VEGF-related hypothyroidism in this study were excluded due to preexisting thyroid dysfunction from prior non-TKI VEGFR therapy, but after review of patients excluded due to preexisting thyroid disease, none was found to have developed hypothyroidism related to bevacizumab or ramucirumab treatment. Direct toxicity of TKI drugs on follicular cells has also been postulated as a mechanism of hypothyroidism, but cumulative TKI dose or drug number was not associated with increased hypothyroidism in this study. Since increased risk of hypothyroidism was observed in patients treated with TKI during a higher percentage of their days at risk, the mechanism of thyroid dysfunction may be related to peak or average drug exposure.

The risk factors identified in this study also may support an alternative immune hypothesis explaining TKI-induced thyroid dysfunction. Female sex is a well-recognized risk factor for the development of autoimmune thyroid diseases (e.g., Graves' and Hashimoto's thyroiditis), and its association with TKI-related hypothyroidism suggests an immune mechanism. Additionally, autoimmune processes including TPOAb development are reported in patients treated with TKI (16,17). Further supporting this possibility, TKI treatment of thyroid cell lines increased expression of MCH Class I molecules, surface receptors that allow immune cell detection, the loss of which is a known mechanism of tumor-mediated immune suppression (18). This effect was more pronounced with sunitinib compared to sorafenib treatment, which is consistent with the greater incidence of hypothyroidism in cancer patients who received sunitinib versus sorafenib in this cohort. Thyroiditis occurring with immune-modulating cancer therapies has been associated with greater response to therapy and improved prognosis (19 –21), but at present, the significance of new thyroid dysfunction in previously euthyroid patients treated with TKI is the subject of ongoing investigation (22). Lastly, the differences in frequency of hypothyroidism among TKI noted in this study and other reports warrants further study. Many of the so-called TKI also have effects at other kinases and kinase receptors, and these additional signaling effects may contribute to thyroid dysfunction.

Several limitations to the present study are acknowledged. The retrospective nature of the study provides an uncontrolled cohort of TKI-treated patients, and similar to prior observational studies, the frequency with which different TKI were used was influenced by historical clinical practice and available TKI. Thyroid testing was not available in all patients, and the timing of TSH measurement and initiation of thyroid hormone was a clinician-specific decision, introducing the potential for sample bias. In the study design, collection of TPOAb status was pre-specified, but unfortunately this was measured in <1% of patients. Similarly, thyroid ultrasound was rarely performed, and family history of thyroid disease was not considered to have been reliably collected. Assessment of TPOAb and family history of autoimmune thyroid disease may be useful parameters in determining induction of autoimmune thyroid disease or risk of TKI-related hypothyroidism.

In conclusion, hypothyroidism is common in euthyroid cancer patients receiving TKI therapy and is more likely to occur early after treatment initiation and in women. Routine monitoring of these patients with thyroid testing is suggested to detect incident cases.

Footnotes

Author Disclosure Statement

None of the authors has any relevant disclosures.

References

Illouz

, Braun

, Briet

, Schweizer

, Rodien

. 2014. Endocrine side-effects of anti-cancer drugs: thyroid effects of tyrosine kinase inhibitors. Eur J Endocrinol, 171:R91–99.

Lodish

. 2013. Clinical review: kinase inhibitors: adverse effects related to the endocrine system. J Clin Endocrinol Metab, 98:1333–1342.

Rini

, Escudier

, Tomczak

, Kaprin

, Szczylik

, Hutson

, Michaelson

, Gorbunova

, Gore

, Rusakov

, Negrier

, Ou

, Castellano

, Lim

, Uemura

, Tarazi

, Cella

, Chen

, Rosbrook

, Kim

, Motzer

. 2011. Comparative effectiveness of axitinib versus sorafenib in advanced renal cell carcinoma (AXIS): a randomised Phase 3 trial. Lancet, 378:1931–1939.

Motzer

, Escudier

, Tomczak

, Hutson

, Michaelson

, Negrier

, Oudard

, Gore

, Tarazi

, Hariharan

, Chen

, Rosbrook

, Kim

, Rini

. 2013. Axitinib versus sorafenib as second-line treatment for advanced renal cell carcinoma: overall survival analysis and updated results from a randomised Phase 3 trial. Lancet Oncol, 14:552–562.

Smith

, Sweeney

, Corn

, Rathkopf

, Smith

, Hussain

, George

, Higano

, Harzstark

, Sartor

, Vogelzang

, Gordon

, de Bono

, Haas

, Logothetis

, Elfiky

, Scheffold

, Laird

, Schimmoller

, Basch

, Scher

. 2014. Cabozantinib in chemotherapy-pretreated metastatic castration-resistant prostate cancer: results of a Phase II nonrandomized expansion study. J Clin Oncol, 32:3391–339.

Motzer

, Hutson

, Cella

, Reeves

, Hawkins

, Guo

, Nathan

, Staehler

, de Souza

, Merchan

, Boleti

, Fife

, Jin

, Jones

, Uemura

, De Giorgi

, Harmenberg

, Wang

, Sternberg

, Deen

, McCann

, Hackshaw

, Crescenzo

, Pandite

, Choueiri

. 2013. Pazopanib versus sunitinib in metastatic renal-cell carcinoma. N Engl J Med, 369:722–731.

Sabatier

, Eymard

, Walz

, Deville

, Narbonne

, Boher

, Salem

, Marcy

, Brunelle

, Viens

, Bladou

, Gravis

. 2012. Could thyroid dysfunction influence outcome in sunitinib-treated metastatic renal cell carcinoma?. Ann Oncol, 23:714–721.

Motzer

, Hutson

, Tomczak

, Michaelson

, Bukowski

, Oudard

, Negrier

, Szczylik

, Pili

, Bjarnason

, Garcia-del-Muro

, Sosman

, Solska

, Wilding

, Thompson

, Kim

, Chen

, Huang

, Figlin

. 2009. Overall survival and updated results for sunitinib compared with interferon alfa in patients with metastatic renal cell carcinoma. J Clin Oncol, 27:3584–3590.

Torino

, Barnabei

, Paragliola

, Baldelli

, Appetecchia

, Corsello

. 2013. Thyroid dysfunction as an unintended side effect of anticancer drugs. Thyroid, 23:1345–1366.

10.

Stefan

, Decallonne

, Dumez

, Bex

, Carmeliet

, Schöffski

. 2008. The clinical implications of sunitinib-induced hypothyroidism: a prospective evaluation. Br J Cancer, 99:448–454.

11.

Desai

, Yassa

, Marqusee

, George

, Frates

, Chen

, Morgan

, Dychter

, Larsen

, Demetri

, Alexander

. 2006. Hypothyroidism after sunitinib treatment for patients with gastrointestinal stromal tumors. Ann Intern Med, 145:660–664.

12.

Mannavola

, Coco

, Vannucchi

, Bertuelli

, Carletto

, Casali

, Beck-Peccoz

, Fugazzola

. 2007. A novel tyrosine-kinase selective inhibitor, sunitinib, induces transient hypothyroidism by blocking iodine uptake. J Clin Endocrinol Metab, 92:3531–3534.

13.

Kamba

, Tam

, Hashizume

, Haskell

, Sennino

, Mancuso

, Norberg

, O'Brien

, Davis

, Gowen

, Anderson

, Thurston

, Joho

, Springer

, Kuo

, McDonald

. 2006. VEGF-dependent plasticity of fenestrated capillaries in the normal adult microvasculature. Am J Physiol Heart Circ Physiol, 290:H560–576.

14.

Villadolid

, Amin

. 2015. Immune checkpoint inhibitors in clinical practice: update on management of immune-related toxicities. Transl Lung Cancer Res, 4:560–575.

15.

van der Veldt

, Lammertsma

, Smit

. 2013. Reduction in thyroid perfusion after bevacizumab treatment. Thyroid, 23:1329–1330.

16.

Pani

, Atzori

, Baghino

, Boi

, Tanca

, Ionta

, Mariotti

. 2015. Thyroid dysfunction in patients with metastatic carcinoma treated with sunitinib: is thyroid autoimmunity involved?. Thyroid, 25:1255–1261.

17.

Eroukhmanoff

, Castinetti

, Penel

, Salas

. 2016. Autoimmune thyroid dysfunction induced by tyrosine kinase inhibitors in a patient with recurrent chordoma. BMC Cancer, 16:679.

18.

Angell

, Lechner

, Jang

, LoPresti

, Epstein

. 2014. MHC class I loss is a frequent mechanism of immune escape in papillary thyroid cancer that is reversed by interferon and selumetinib treatment in vitro . Clin Cancer Res, 20:6034–6044.

19.

Nelson

, Hercbergs

, Rybicki

, Strome

. 2006. Association between development of hypothyroidism and improved survival in patients with head and neck cancer. Arch Otolaryngol Head Neck Surg, 132:1041–1046.

20.

Weijl

, Van der Harst

, Brand

, Kooy

, Van Luxemburg

, Schroder

, Lentjes

, Van Rood

, Cleton

, Osanto

. 1993. Hypothyroidism during immunotherapy with interleukin-2 is associated with antithyroid antibodies and response to treatment. J Clin Oncol, 11:1376–1383.

21.

Bladou

, Gravis

, Sabatier

, Deville

, Walz

, Narbonne

, Marcy

, Salem

, Brunelle

, Eymard

. 2010. Hypothyroidism and survival during sunitinib therapy in metastatic renal cell carcinoma (mRCC): a prospective observational analysis. J Clin Oncol, 28:e15013.

22.

Bilen

, Patel

, Hess

, Munoz

, Busaidy

, Wheler

, Janku

, Falchook

, Hong

, Meric-Bernstam

, Habra

, Naing

. 2016. Association between new-onset hypothyroidism and clinical response in patients treated with tyrosine kinase inhibitor therapy in Phase I clinical trials. Cancer Chemother Pharmacol, 78:167–171.