Teprotumumab-Related Hyperglycemia

Abstract

Context

Graves orbitopathy (GO) or thyroid eye disease is a potentially sight-threatening and disfiguring autoimmune disease. Teprotumumab is a monoclonal antibody against the insulin-like growth factor-I receptor that was recently approved for GO treatment. Hyperglycemia is a recognized adverse event of teprotumumab, occurring in 10% of patients in 2 recent randomized controlled trials.

Objective

Our study aimed to report the incidence, severity, management, and longitudinal glycemic changes in patients treated with teprotumumab in an academic practice cohort.

Methods

This longitudinal, observational study included all consecutive patients treated with teprotumumab between March 2020 and May 2022 at 1 institution. Hemoglobin A1c (HbA1c) was measured every 3 months.

Results

Forty-two patients with baseline normoglycemia (n = 22), prediabetes (n = 10), and diabetes (n = 10) were followed for a mean of 47.5 weeks. Overall, HbA1c increased by 0.5% at 3 months. Least-squares mean changes in HbA1c at 3 months were 1.3 (P < .001), 0.7 (P = .01), and 0.1 (P = .41) in patients with diabetes, prediabetes, and normoglycemia, respectively. Twenty-two patients (52%) had hyperglycemia, which was graded as mild, moderate, and life-threatening in 55% (12/22), 41% (9/22), and 5% (1/22) of cases, respectively. Age, pre-existing diabetes, and Hispanic and Asian race/ethnicity were significant risk factors for hyperglycemia. Among patients with hyperglycemia, 36.4% (8/22) returned to baseline glycemic status at last follow-up.

Conclusion

While effective, teprotumumab carries a significant risk of hyperglycemia, especially in patients with diabetes. Hyperglycemia may persist after stopping teprotumumab. These findings underscore the importance of guidelines for screening and management of teprotumumab-related hyperglycemia.

teprotumumab, hyperglycemia, thyroid eye disease, IGF-1, adverse event

Graves orbitopathy (GO) or thyroid eye disease is a debilitating and disfiguring autoimmune disease, that can lead to periorbital tissue swelling, orbital inflammation, and fibrosis. These changes are thought to be mediated by increased insulin-like growth factor 1 receptor (IGF-1R) expression and orbital fibroblast activity. The resultant changes to periorbital structures often result in proptosis, diplopia, and, in some cases, vision loss that can greatly decrease quality of life (1).

Until recently, glucocorticoids have been the primary treatment for GO. However, they present several safety and efficacy concerns (2). New targeted biologic therapies promise greater efficacy and a more favorable adverse event (AE) profile. In January 2020, teprotumumab, an IGF-1R monoclonal antibody, became the first medication to receive United States Food and Drug Administration approval for the treatment of GO following the results of 2 randomized controlled trials (RCTs) that reported significantly improved proptosis, diplopia, disease activity, and quality of life (3–5). Based on the mechanism of action, hyperglycemia and hearing loss were anticipated side effects of teprotumumab (6–11). Other notable AEs observed in the RCTs included muscle spasms, nausea, alopecia, diarrhea, and fatigue.

The published data regarding the prevalence of hyperglycemia and its resolution in teprotumumab-treated patients are limited. The initial RCTs documented a 10% (8/84) incidence of hyperglycemia (3–5). In the RCTs, 63% (5/8) of patients with hyperglycemia had a history of diabetes. All cases of hyperglycemia were graded as mild or moderate and none led to discontinuation of teprotumumab. Six hyperglycemic events resolved within the treatment window and 3 resolved after the last dose. Most cases were controlled with medication adjustment (12). While no cases of the hyperglycemic hyperosmolar state or diabetic ketoacidosis (DKA) were reported in the teprotumumab RCTs, there have been emerging reports of patients experiencing such life-threatening events (13, 14). A recent case report described hyperglycemic hyperosmolar state in a patient with a 3-year history of prediabetes and baseline hemoglobin A1c (HbA1c) of 6.1. The patient was hospitalized 3 weeks after the first teprotumumab infusion and was successfully managed with fluids and insulin (13). This case illustrates that severe hyperglycemia may also occur in patients with prediabetes.

While calls for better guidelines have been raised, currently no formal protocol exists for monitoring and managing teprotumumab-related hyperglycemia in patients with GO (15). Before such guidelines are developed, a greater understanding of teprotumumab-related hyperglycemia is needed, including more information on its incidence, severity, risk factors, management, and resolution pattern. Here we report the incidence, severity, management, and longitudinal changes in hyperglycemia in a real-world cohort of patients treated with teprotumumab.

Patients and Methods

Patient Enrollment

We conducted a single-center, observational longitudinal study on all consecutive patients with GO treated with teprotumumab at Stanford University Medical Center between March 2020 and May 2022. Patients with GO were included in the study if they were over 18 years of age and received 4 or more infusions of teprotumumab. We included patients with all grades and stages of GO, though most patients had active, moderate to severe GO. Patients being treated with a second course of teprotumumab were excluded from the analysis. HbA1c was obtained before initiation of treatment and every 3 months thereafter. Patients were excluded from the analysis if they did not have both a pretreatment HbA1c and at least 1 post-treatment HbA1c. This study was approved by the Stanford Institutional Review Board and adheres to the tenets of the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Informed consent was required prior to participation in the study.

Data Collection

Teprotumumab was administered intravenously at 10 mg/kg for the first infusion and given every 3 weeks at 20 mg/kg for a total of 8 infusions. Patients were seen by an endocrinologist and ophthalmologist after receiving 2, 4, 6, and 8 infusions. Patients were also seen 6 months after the final infusion. Blood for laboratory tests, including thyroid-stimulating hormone, free thyroxine (FT4), total triiodothyronine (T3), and HbA1c, was drawn at baseline, after 4 infusions (at 12 weeks), 8 infusions (at 24 weeks), and at 52 weeks. Baseline laboratory blood samples were drawn 4 weeks prior to starting teprotumumab. Blood glucose (fasting or nonfasting) was drawn at baseline and before every teprotumumab infusion and at 52 weeks after teprotumumab initiation. According to our therapy plan, if blood glucose was below 50 or over 300, teprotumumab was held and the infusion center notified the treating physician for further recommendations and antidiabetic medication management.

The following demographics and medical history were collected via patient interview and chart review: age, sex, race/ethnicity, smoking status (current, past, never), statin use, baseline antihyperglycemic medications, alteration in antihyperglycemic and other medications during treatment, glucocorticoid therapy within 3 months of starting teprotumumab, family history of diabetes, and past medical history of hypertension. The thyroid status of patients was classified according to FT4 and total T3 into 3 categories: normal (normal FT4 and T3), high (elevated FT4 or T3), or low (low FT4). Thyroid-stimulating hormone was not used as it can lag behind hyperthyroidism correction.

Diagnostic Criteria

Patients were categorized as having normoglycemia, prediabetes, and diabetes based on baseline HbA1c and history of diabetes according to the American Diabetes Association criteria (16). Thus, patients were classified as having diabetes if HbA1c ≥6.5% or known history, prediabetes if HbA1c 5.7 to 6.4 and no history of diabetes, and normoglycemia if HbA1c ≤5.6, with no history of diabetes or prediabetes.

Teprotumumab-related hyperglycemia was graded according to the Common Terminology Criteria for Adverse Events (CTCAE) v5.0. The CTCAE scale for drug-related hyperglycemia has 5 grades, where grade 1 or mild hyperglycemia is an abnormal glucose above baseline requiring no medical intervention. Here we defined an abnormal glucose as a change in glycemic category or HbA1c change ≥0.5% (17). Grade 2 or moderate hyperglycemia is hyperglycemia requiring a change in daily management of diabetes, initiation of an oral antihyperglycemic agent, or workup for diabetes. Grade 3 or severe hyperglycemia is hyperglycemia requiring insulin therapy or hospitalization. Grade 4 or life-threatening hyperglycemia results in life-threatening consequences or requires urgent intervention. Grade 5 is fatal hyperglycemia.

Study Outcomes

The primary outcome measure was change in HbA1c from baseline to 3 months in patients with baseline normoglycemia, prediabetes, and diabetes. Secondary outcomes included the change in HbA1c at other timepoints, number of patients with a change in HbA1c category, incidence of HbA1c elevation, duration of HbA1c elevation, rate of HbA1c recovery, and CTCAE grade.

Statistical Methods

Statistical analysis was done using SAS version 9.4 (SAS Institute, Cary, NC, USA). Descriptive statistics were used to describe demographic data, laboratory values, and time intervals, with the most appropriate statistical test being chosen based on the variable type. Changes in HbA1c between groups over time were analyzed with a mixed effects 2-factor repeated measures analysis of variance (ANOVA) test. Least-squares mean change was used to assess per group changes in HbA1c at 3 months, 6 months, and 12 months. One-way ANOVA was used to assess differences in baseline characteristics between the different groups. Multiple linear regression analysis was used to evaluate risk factors for the 3-month change in HbA1c. A stepwise approach was used to select independent variables for inclusion in the multivariate model. Age, sex, smoking status, hypertension, statin use, recent glucocorticoid use, baseline HbA1c, baseline body mass index (BMI), baseline systolic blood pressure (SBP), race/ethnicity, and baseline glycemic status were entered into the stepwise model as independent predictors. The race/ethnicity variable was dummy coded into non-Hispanic white (reference), Asian, and Hispanic categories. The glycemic status variable was coded into normoglycemia (reference), prediabetes, and diabetes. No patients were current smokers; therefore, the smoking status variable was binary including never-smokers (reference) and past smokers. The P value thresholds for entering and removing predictors were .05 and .1, respectively. A significance level of .05 was used for all other statistical tests.

Results

Fifty-five patients were enrolled in the study. Forty-eight had ≥4 infusions of teprotumumab. Forty-two had pretreatment and post-treatment HbA1c measurements and were thus included in the analysis. Most (90%) of the study participants were female. Mean age was 58.2 ± 14.3 years. At baseline, 52% (22/42) of the patients had normoglycemia, 24% (10/42) had prediabetes, and 24% (10/42) had diabetes, with mean baseline HbA1c of 5.4 ± 0.2, 5.9 ± 0.2, and 6.4 ± 1.1, respectively. Patients received a mean of 6.7 ± 1.6 teprotumumab infusions and were followed for a mean of 47.5 ± 17.6 weeks after the first infusion and 26.4 ± 17.7 weeks after the last infusion. At the last follow-up, 93% (39/42) of patients had completed all 8 teprotumumab infusions. HbA1c values were available at the 3-month, 6-month, and 12-month timepoints in 95.2% (40/42), 78.6% (33/42), and 64.3% (27/42) of patients, respectively. Baseline characteristics were similar among groups (Table 1), except patients with diabetes had significantly higher HbA1c and rate of statin use at baseline than patients without diabetes.

Table 1.

Baseline characteristics

	All patients (n = 42)	Normoglycemia (n = 22)	Prediabetes (n = 10)	Diabetes (n = 10)	P value
Age, years (SD)	58.2 (14.3)	54.5 (13.5)	61.7 (10.8)	62.8 (17.9)	.271
Women, n (%)	38 (90.5)	20 (90.9)	9 (90.0)	9 (90.0)	.985
Race/Ethnicity
ȃnon-Hispanic White, n (%)	15 (35.7)	8 (36.4)	3 (30.0)	4 (40.0)	.469
ȃHispanic, n (%)	7 (16.7)	3 (13.6)	3 (30.0)	1 (10.0)
ȃAsian, n (%)	19 (45.2)	11 (50.0)	4 (40.0)	4 (40.0)
ȃOther, n (%)	1 (2.4)	0 (0.0)	0 (0.0)	1 (10.0)
BMI, kg/m2 (SD)	26.0 (6.7)	24.7 (5.5)	26.4 (7.7)	28.2 (8.0)	.415
Hypertension, n (%)	17 (40.5)	4 (18.2)	7 (70.0)	6 (60.0)	.085
Family history of DM, n (%)	16 (38.1)	10 (45.5)	3 (30.0)	3 (30.0)	.894
Statin use, n (%)	17 (40.5)	4 (18.2)	4 (40.0)	9 (90.0)	.022
Smoking status
ȃCurrent smoker, n (%)	0 (0)	0 (0)	0 (0)	0 (0)	.190
ȃPast smoker, n (%)	8 (19.0)	2 (9.1)	3 (30.0)	3 (30.0)
ȃNever smoker, n (%)	34 (80.9)	20 (90.9)	7 (70.0)	7 (70.0)
Recent GC use, n (%)a	15 (35.7)	5 (22.7)	4 (40.0)	6 (60.0)	.564
ȃPO, n (%)	6 (40.0)	3 (60.0)	0 (0)	3 (50.0)	.217
ȃIV, n (%)	10 (66.7)	3 (60.0)	4 (100.0)	3 (50.0)	.378
SBP, mmHg (SD)	129.7 (19.7)	127.2 (19.1)	131.9 (19.2)	132.8 (22.5)	.706
DBP, mmHg (SD)	73.1 (11.9)	77.4 (11.6)	67.6 (9.7)	69.1 (11.7)	.057
Thyroid hormone levels
ȃNormal, n (%)	34 (80.9)	19 (86.4)	9 (90)	6 (60.0)	.197
ȃHigh, n (%)	4 (9.5)	2 (9.1)	0 (0)	2 (20.0)
ȃLow, n (%)	4 (9.5)	1 (4.5)	1 (10.0)	2 (20.0)
Mean HbA1c, % (SD)	5.7 (0.7)	5.4 (0.2)	5.9 (0.2)	6.4 (1.1)	<.001
Mean infusions, n (SD)	6.7 (1.6)	6.8 (1.7)	6.4 (1.6)	6.8 (1.5)	.659
Follow-up first infusion, weeks (SD)	47.5 (17.6)	45.9 (14.8)	47.9 (21.1)	50.0 (20.8)	.613
Follow-up last infusion, weeks (SD)	26.4 (17.7)	25.5 (12.0)	24.7 (25.5)	29.1 (20.7)	.813

P values based on 1-way ANOVA or a chi-square test comparing patients with normoglycemia, prediabetes, and diabetes.

Abbreviations: BMI, body mass index; DBP, diastolic blood pressure; DM, diabetes mellitus; GC, glucocorticoid; IV, intravenous; PO, oral; SBP, systolic blood pressure.

Steroid use within 3 months of starting teprotumumab; 1 patient had both recent IV and PO steroid use.

Following teprotumumab treatment, 83.3% (35/42) of the patients increased their HbA1c by ≥0.1% and 30.9% (13/42) patients experienced an HbA1c increase of ≥0.5% at some point during the study. Ninety-one percent (32/35) of increases occurred within 3 months of starting teprotumumab and were significantly greater in patients with baseline prediabetes and diabetes (Fig. 1). Least-squares mean changes in HbA1c at 3 months were 1.3 (P < .001), 0.7 (P = .014), and 0.1 (P = .415) in patients with diabetes, prediabetes, and normoglycemia, respectively (Fig. 2). Patients with baseline diabetes also had significantly elevated HbA1c at 6 months (P = .02) (Table 2). Eleven patients had random blood glucose ≥200 mg/dL at least once during treatment with teprotumumab. Of these 11 patients, 72.7% (8/11) had diabetes and 27.3% (3/11) had prediabetes at baseline.

Figure 1.

Individual HbA1c trajectories. Each line represents serial HbA1c values from the same patient in the year following their first infusion of teprotumumab. Colors represent the 3 glycemic categories: green (normoglycemia), blue (prediabetes), and red (diabetes).

Figure 2.

Least-squares mean change in HbA1c relative to baseline. Bars show the change in HbA1c by glycemic category at 3 months (blue), 6 months (orange), and 12 months (grey) after their first infusion of teprotumumab. *Signifies significant (P < .05) change from baseline.

Table 2.

Change in HbA1c over time

	Baseline	3 Months	6 Months	12 Months
All patients (n)a	42	40	33	27
ȃMean HbA1c (SD)	5.7 (0.7)	6.2 (1.3)	6.2 (1.0)	5.8 (0.8)
ȃLSM HbA1c changeb	—	0.7c	0.5c	0.3
Normoglycemia (n)a	22	21	17	15
ȃMean HbA1c (SD)	5.4 (0.2)	5.5 (0.3)	5.5 (0.2)	5.4 (0.4)
ȃLSM HbA1c change (95% CI)	—	0.1 (−0.2-0.4)	0.2 (−0.3-0.6)	0 (−0.5-0.5)
Prediabetes (n)a	10	9	7	5
ȃMean HbA1c (SD)	5.9 (0.2)	6.2 (1.0)	6.1 (0.6)	5.8 (0.4)
ȃLSM HbA1c change (95% CI)	—	0.7 (0.2-1.3)c	0.6 (−0.2-1.4)	0.4 (−0.5-1.4)
Diabetes (n)a	10	10	9	7
ȃMean HbA1c (SD)	6.4 (1.1)	7.6 (1.8)	7.4 (1.0)	6.7 (1.1)
ȃLSM HbA1c change (95% CI)	—	1.3 (0.8-1.7)c	0.7 (0.1-1.3)c	0.4 (−0.3-1.1)

LSM, least-squares mean; HbA1c, hemoglobin A1c.

Number of patients in each glycemic category with available HbA1c values at baseline, 3 months, 6 months, and 12 months.

Change in HbA1c relative to baseline.

LSM change with P value < .05.

Following teprotumumab treatment, 40.9% (9/22) of patients with normoglycemia developed prediabetes, 30% (3/10) with prediabetes progressed to diabetes, and 37.5% (3/8) with controlled diabetes (HbA1c ≤ 7.0) increased HbA1c to >7.0%. Recovery to baseline normoglycemia, prediabetes, and controlled diabetes occurred in 33% (3/9), 0% (0/3), and 33% (1/3) of cases, respectively. Mean time to recovery was 38.7 and 42.7 weeks for patients with normoglycemia and diabetes, respectively.

Twenty-two patients had hyperglycemia that was classified as CTCAE grade 1, 2, and 4 in 55% (12/22), 41% (9/22), 5% (1/22) of cases, respectively (Table 3). Eight patients discontinued teprotumumab, 50% (4/8) due to hearing loss, 12.5% (1/8) due to hepatic encephalopathy, 12.5% (1/8) due to poor efficacy, 12.5% (1/8) due to DKA, and 12.5% (1/8) due to several intolerable AEs. The several intolerable AEs included hyponatremia, fatigue, and exacerbation of psychiatric symptoms.

Table 3.

Hyperglycemia severity grading

	CTCAE = 1	CTCAE = 2	CTCAE = 3	CTCAE = 4	Number with hyperglycemia (% of total)
Normoglycemia (n = 22)	9	0	0	0	9 (41%)
Prediabetes (n = 10)	2	0	0	1	3 (30%)
Diabetes (n = 10)	1	9	0	0	10 (100%)
All patients (n = 42)	12	9	0	1	22 (52%)

Number of patients with hyperglycemia CTCAE grades 1-4 stratified by baseline glycemic control.

Abbreviation: CTCAE, common terminology criteria for adverse events. Grade 1 hyperglycemia is an abnormal glucose above baseline requiring no medical intervention. Here we defined an abnormal glucose as a change in glycemic category or HbA1c change ≥0.5. Grade 2 hyperglycemia is hyperglycemia requiring a change in daily management for a patient with diabetes, initiation of an oral antihyperglycemic agent, or workup for diabetes. Grade 3 or severe hyperglycemia is hyperglycemia requiring insulin therapy or hospitalization. Grade 4 or life-threatening hyperglycemia results in life-threatening consequences or requires urgent intervention.

Among patients who developed hyperglycemia after teprotumumab (CTCAE grade ≥1), 40.9% (9/22) were taking antihyperglycemic medications at baseline. Hyperglycemia was managed with the addition of new antihyperglycemic medications in 36.4% (8/22) of cases, including insulin (1/22), and a change in antihyperglycemic medication dosage or type in 31.8% (7/22) of cases, including insulin modification in 18.2% (4/22) of cases (Table S1 (18)). One patient was hospitalized for DKA. The patient had baseline prediabetes that was diagnosed at the time of starting teprotumumab. She was admitted to the hospital after her third infusion of teprotumumab when she complained of weakness, polyuria, dry mouth, decreased appetite, and weight loss. Admission laboratory tests indicated severe DKA with random glucose of 920 mg/dL, HbA1c of 12.5%, anion gap of 30 mmol/L, serum bicarbonate of 17 mmol/L, β-hydroxybutyrate of 12.1 mmol/L, 3+ ketonuria, and a venous blood gas pH 7.25. DKA resolved with insulin and intravenous (IV) fluids. She was discharged after 5 days on insulin glargine 35 units nightly, and insulin aspart 11 units with meals. Metformin was started after discharge. At last follow-up, 26 weeks after the first infusion, insulin was discontinued.

Among patients with hyperglycemia (CTCAE grade ≥1), 36.4% (8/22) of patients returned to baseline glycemic status at last follow-up, 50.4 weeks after the first infusion. Four of these patients had baseline normoglycemia and 4 had diabetes. All 4 patients with diabetes were managed with antihyperglycemic medication adjustment. The 14 patients with persistent hyperglycemia above baseline included 5 patients with normoglycemia, 4 patients with prediabetes, and 5 patients with diabetes. Of the 10 patients who required treatment for hyperglycemia during the study, 60% had hyperglycemia above baseline at last follow-up.

We carried out a multiple linear regression analysis to test for an association between the 3-month change in HbA1c and age, sex, smoking status, hypertension, statin use, recent glucocorticoid use, baseline HbA1c, baseline BMI, baseline SBP, ethnicity, and baseline glycemic status. We found that a higher 3-month change in HbA1c was significantly associated with increased age (P = .01), Asian race/ethnicity (P = .007), Hispanic race/ethnicity (P = .02), and baseline diabetes (P = .005). The R² change indicated that the most important predictor of 3-month HbA1c change was baseline diabetes (Table 4). Notably, patients with diabetes had the highest 3-month HbA1c change.

Table 4.

Independent predictors of the 3-month change in HbA1c identified by multiple linear regression analysis

	β-Coefficient	R2	P value
Age	0.020	0.070	.011
Glycemic Status
ȃNormoglycemia (reference)
ȃPrediabetes	−0.066	0.001	.785
ȃDiabetes	0.940	0.296	.005
Race/ethnicity
ȃNon-Hispanic White (reference)
ȃAsian	0.635	0.067	.007
ȃHispanic	0.684	0.090	.020

R² values indicate independent contributions to R² for each predictor. The dependent variable was the 3-month change in HbA1c after starting teprotumumab. Independent variables included: age, sex, smoking status, hypertension, statin use, steroid use within 3 months of starting teprotumumab, baseline HbA1c, baseline BMI, baseline SBP, ethnicity, and baseline glycemic status. Only baseline glycemic status, age, and ethnicity were retained in the model. The other variables were removed because they were not significant predictors of the dependent variable. Baseline glycemic status: normoglycemia (reference), prediabetes, diabetes. Ethnicity: Asian, Hispanic, non-Hispanic White (reference). P < .05 was considered significant.

Ten patients were treated with IV glucocorticoids within 3 months of starting teprotumumab, but HbA1c values at baseline, 3 months, and 6 months did not significantly (P = .27) differ between patients who were or were not treated with IV glucocorticoids before teprotumumab (Fig. 3). BMI decreased by a mean of 0.5 ± 1.2 at 6 months after the first infusion of teprotumumab, but this was not found to be statistically significant (P = .69). Thyroid status did not significantly change during treatment (P = .371), with most (30/42) patients maintaining normal FT4 and T3 throughout treatment. Finally, a simple linear regression showed a significant correlation between baseline HbA1c, and 3-month change in HbA1c in the total study population (β = 0.55, P < .001, Fig. 4).

Figure 3.

Mean HbA1c over time according to recent IV glucocorticoid use. Lines represent patients with (blue, n = 9) and without (orange, n = 31) IV glucocorticoid use within 3 months of starting teprotumumab. There was no significant difference in HbA1c by IV glucocorticoid use (P = .27), using a mixed effects 2-factor repeated measures ANOVA with HbA1c as the dependent variable and IVglucocorticoid use and time as independent variables.

Figure 4.

Correlation between baseline HbA1c and 3-month change in HbA1c by glycemic categories. There was a significant association between baseline HbA1c and 3-month changes in HbA1c (β = 0.55, P < .001). Green squares represent patients with baseline normoglycemia, blue squares represent patients with baseline prediabetes, and red squares represent patients with baseline diabetes.

Discussion

Teprotumumab, an inhibitory monoclonal antibody against the IGF-1R, is a novel therapy against GO that targets the disease pathophysiology. Teprotumumab-related hyperglycemia is likely due to the partial homology of the IGF-1R with the insulin receptor, the existence of hybrid heterodimeric IGF-1/insulin receptor complexes, and the effects of elevated growth hormone on insulin resistance and gluconeogenesis (6). In the recent RCTs of teprotumumab used for treatment of GO, approximately 10% of patients developed hyperglycemia (3). However, the exact incidence, degree, time course, and resolution of glycemic changes with this agent have not been described and there are no formal guidelines regarding screening and monitoring the glycemic status of teprotumumab recipients.

Our study is the first to characterize the incidence and time course of hyperglycemia in 42 teprotumumab-treated patients, in an academic clinical setting, with an average follow-up of 47.5 weeks after the first infusion. The mean increase in HbA1c at 12 weeks after initiation of therapy was 0.5%, a magnitude of change that is considered clinically meaningful (17). Fifty-two percent of patients developed hyperglycemia and 31% of patients increased their HbA1c by ≥0.5%, with larger increases in HbA1c in patients with pre-existing diabetes or prediabetes than in normoglycemic patients. While patients with baseline dysglycemia were at higher risk for hyperglycemia, about 41% of patients with normoglycemia developed prediabetes, with only a third recovering to normoglycemia by the last follow-up. The incidence of hyperglycemia using CTCAE criteria ranged from 30% to 100%, depending on the baseline glycemic status of the patient (Table 3). Of the 22 patients that developed hyperglycemia during teprotumumab treatment, 10 required a change in medication management (Table S1 (18)). Sixty-four percent (14/22) of patients had persistent hyperglycemia, at last follow-up. Additional findings were that age, pre-existing diabetes, and Asian and Hispanic race/ethnicity, increased the risk of HbA1c elevation at 3 months, while recent glucocorticoid use was not found to significantly predict the change in HbA1c.

Our findings show a higher incidence and severity of hyperglycemia than the existing data in the teprotumumab RCTs. In the phase 2/3 studies, 8 of 84 patients (10%) developed hyperglycemia, with 5 of 8 having underlying diabetes mellitus compared with 10 of 22 patients in our study. Average HbA1c increased by 0.2 in the RCTs by 24 weeks vs 0.5 in our study (12). Hyperglycemia was described as “non-serious,” not requiring teprotumumab discontinuation, and resolving in all but 1 out of 8 patients (3, 12). In our study, only 36% of patients returned to their baseline glycemic status. There are several differences between our study population and patients in the RCTs that could explain these findings. Our study population was older (mean age 58 years compared with 51 years in the RCTs), and more racially diverse with higher proportion of Asian patients (45% vs 4%) and Hispanic patients (16.7% vs ≤1%). Mean BMI was 26 in our study but not reported in the RCTs. Approximately one-quarter (23.8%, 10/42) of our patients had recent glucocorticoid use of >1 g within 3 months of starting teprotumumab, whereas patients with a history of significant glucocorticoid exposure were excluded in the RCTs. Baseline A1C was similar (5.7 in our study vs 5.6 in the RCTs (12)) between the 2 groups. Our study included 10 patients with prediabetes, and 10 with diabetes; thus, approximately half (20/42) of our study population had diabetes or prediabetes at baseline while 11.9% (10/84) of patients in the RCTs had pre-existing diabetes (12). The number of patients with prediabetes was not reported. The older age, different racial/ethnic breakdown, and the inclusion of patients with recent history of glucocorticoid use could explain the increased incidence and severity of hyperglycemia in our study. Interestingly, the risk of hyperglycemia did not differ in our patients according to their history of glucocorticoid exposure. Finally, in our real-world cohort, hyperglycemia was managed by different endocrinologists and not through a uniform protocol, which may have resulted in less stringent monitoring, compliance, and antihyperglycemic treatment than patients in the highly controlled RCT environment. Future prospective studies with more stringent follow-up and real-world cohorts are needed to confirm these findings.

Our study has important implications for GO patients. It illustrates the high incidence of HbA1c elevation with teprotumumab treatment, describes the time course of HbA1c elevation and resolution, shows that nearly two-thirds of patients did not return to their baseline glycemic status, and highlights the variability in HbA1c responses based on age, race/ethnicity, and baseline glycemic status.

Our findings invite further research in the field and the development of guidelines surrounding glycemic management following teprotumumab treatment. Until such guidelines are published, we recommend the following: (1) All patients should be screened for diabetes and prediabetes before initiation of teprotumumab with a fasting glucose and HbA1c (19). (2) In patients with preexisting diabetes, HbA1c should be <7% prior to initiation of teprotumumab. Teprotumumab treatment should be delayed if HbA1c is >9%. (3) All patients should be educated on the possibility of hyperglycemia and the signs and symptoms of diabetes should be reviewed. Those with preexisting diabetes should be instructed to self-monitor their glucose daily using either a glucometer or a continuous glucose monitor. (4) At a minimum, patients need a repeat HbA1c at 12 weeks following teprotumumab initiation, especially since most cases of teprotumumab-related hyperglycemia will occur within this timeframe. (5) Blood glucose should be checked prior to each teprotumumab infusion. While monitoring fasting glucose may facilitate comparisons from baseline, a random glucose test may be more practical and can also identify new patients with diabetes if glucose is 200 mg/dL or greater. Random glucose testing may also identify individuals with pre-existing diabetes with suboptimal glucose control. (6) Patients taking teprotumumab should be managed by a multidisciplinary team, including an ophthalmologist and endocrinologist, for optimal monitoring and management of the glycemic side effects of teprotumumab.

Our study has several limitations. Blood glucose was not fasting in many of our measurements; this would have allowed for more accurate characterization of the glycemic abnormalities. We relied on HbA1c for our glycemic status classification, which may have misclassified some patients, as HbA1c may vary according to age, race/ethnicity, and genetic factors. Our sample size was relatively small. We did not have HbA1c data at all timepoints in all patients. Finally, even though our population was more racially diverse than the published RCTs, we did not have representation of Black patients.

This is the first study to prospectively evaluate the time course of glycemic changes after teprotumumab treatment in an academic clinic, in a cohort of diverse patients, of different underlying glycemic status. While hyperglycemia was more pronounced in patients with underlying diabetes, patients with normoglycemia were also at risk of developing hyperglycemia. Finally, a substantial proportion of patients did not have resolution of hyperglycemia at an average follow-up of 47.5 weeks after teprotumumab initiation. Clinical guidelines are needed around monitoring and treatment of this complication.

Funding

This work was supported by the National Institute of Health under grant NIH P30 026877; and Research to Prevent Blindness under an unrestricted grant.

Disclosures

S.H.K. is a consultant for Aligos and advises GI Dynamics. A.L.K. is a consultant for Immunovant Inc.

Data Availability

Original data generated and analyzed during this study are included in this published article or in the data repositories listed in References.

Abbreviations

 
• AE

  adverse event

 
• ANOVA

  analysis of variance

 
• BMI

  body mass index

 
• DKA

  diabetic ketoacidosis

 
• FT4

  free thyroxine

 
• GO

  Graves orbitopathy

 
• HbA1c

  hemoglobin A1c

 
• IGF

  insulin-like growth factor

 
• IV

  intravenous

 
• SBP

  systolic blood pressure

 
• T3

  triiodothyronine