Abstract
KEY POINTS
In this large, retrospective cohort study from a global database, the authors examine the risk of long-term dysglycemia associated with the use of teprotumumab for the treatment of thyroid eye disease.
The results showed a significantly increased odds of prediabetes, diabetes, and initiation of antidiabetes medications after teprotumumab use as compared to controls, with greater risk in older and female patients.
The risk of incident hyperglycemia substantially exceeds clinical trial rates and persists for years after treatment, warranting long-term surveillance beyond treatment completion, especially in subgroups at greater risk.
SUMMARY
Background
Teprotumumab, an insulin-like growth factor 1 receptor (IGF-1R) inhibitor, has revolutionized the treatment of thyroid eye disease (TED). However, hyperglycemia has emerged as a significant adverse event, with a reported incidence of approximately 10%. Small real-world studies subsequently suggested a potentially much higher rate and a broader spectrum of severity, including life-threatening events.1–4 This study by Lishinsky-Fischer et al. aimed to definitively characterize the risk of hyperglycemia associated with teprotumumab by leveraging a vast, multinational electronic health records database, thereby providing a more generalizable estimate of this adverse effect in clinical practice. 5
Methods
The investigators conducted a retrospective cohort study using the TriNetX Global Collaborative Network, encompassing over 165 health care organizations worldwide. They identified TED patients who received at least four infusions of teprotumumab and used 1:1 propensity-score matching to create a rigorously balanced control group of TED patients not exposed to the drug. Matching accounted for demographics, key comorbidities affecting glucose metabolism, baseline HbA1c, BMI, and corticosteroid exposure. The primary outcomes were the 5-year cumulative incidence of incident prediabetes, diabetes (by laboratory criteria and diagnostic codes), severe hyperglycemia, and initiation of antidiabetes medications, analyzed using Kaplan–Meier survival and Cox proportional-hazards models.
Results
After matching, 792 patients were included in each cohort over a follow-up period of up to 5 years. Teprotumumab treatment was associated with a significantly increased 5-year risk for all glycemic outcomes. The hazard ratio (HR) was 2.03 (CI, 1.51–2.72) for incident prediabetes and 2.23 (CI, 1.46–3.43) for incident diabetes. The cumulative incidence of prediabetes reached 24% in the treated group versus 10% in controls. The need to initiate antidiabetes medication (HR, 1.68; CI, 1.27–2.22) was also significantly higher with teprotumumab. Subgroup analysis confirmed a heightened risk for females, patients ≥ 55 years of age, and White individuals. Notably, the risk for patients with a BMI ≥ 30 was not statistically significant, despite higher absolute event numbers, and mean HbA1c values over time did not differ substantially between groups.
Conclusions
In a large real-world cohort, teprotumumab treatment was associated with a sustained, twofold increased risk of prediabetes and diabetes over 5 years, confirming that hyperglycemia is a clinically significant and persistent, yet manageable, adverse effect.
COMMENTARY
The large-scale cohort study by Lishinsky-Fischer et al. provides the definitive real-world quantification of teprotumumab-associated hyperglycemia, affecting 24% of treated patients compared with 10% of controls (not exposed to the drug). It doubled the 5-year risk of incident prediabetes and diabetes and raised glucose-lowering medication use by 68%. 5 These findings not only validate the warning signals raised by previous small real-world studies, such as the Stanford cohort, 4 but also exceed the magnitude of risk observed in pivotal clinical trials.
Gigliotti et al. showed that TED and Graves’ disease confer a 60% higher baseline risk of hyperglycemic events as compared with the general population, likely reflecting a convergence of chronic autoimmune inflammation, hyperthyroidism-driven metabolic stress, and systemic insulin resistance. 6 Teprotumumab-associated hyperglycemia is primarily attributed to inhibition of the insulin-like growth factor 1 receptor (IGF-1R), which shares structural homology with the insulin receptor and forms hybrid receptors, thereby impairing insulin signaling and peripheral glucose uptake, while compensatory increases in growth hormone further promote insulin resistance and gluconeogenesis. Teprotumumab, therefore, delivers a “second hit” to an already vulnerable substrate, an effect the Lishinsky-Fischer study quantifies with a hazard ratio for independent diabetes at 5 years of 2.23 (CI, 1.46–3.43) adjusted for baseline diabetes and other covariates. Meanwhile, a meta-analysis of seven randomized controlled trials by Song et al. further confirmed the elevated risk of hyperglycemia, reporting a pooled risk ratio (RR) of 2.82 (95% CI, 1.08–7.37), 7 and the registration study of IBI311 (an IGF-1R inhibitor) in TED (RESTORE-1) randomized clinical trial reported an incidence of hyperglycemia of 13% in Asian patients. 8 This heterogeneity highlights that controlled trial settings cannot fully capture real-world complexity. The Lishinsky-Fischer study makes this clear, documenting risks possibly far exceeding those in most RCTs and demonstrating a value of real-world evidence: that it can reveal drug performance under broader criteria, nonstandardized monitoring, and routine care.
Crucial granularity comes from Zhao et al.’s pooled post hoc analysis of IBI311 trials, which identified a rapid pattern (13.2%) with early glycemic spikes (≤6 weeks) that respond to prompt intervention, and a delayed pattern (>6 weeks) (11.8%) with insidious HbA1c rise despite stable fasting glucose and minimal baseline metabolic abnormalities. 9 Among IBI311-treated participants, 25.0% (19 of 76) developed associated hyperglycemia during treatment, representing an absolute risk increase of 21.4% (95% CI, 9.5–33.3), and 31.1% (23 of 74) progressed to an elevated glycemic category by week 24 of 30 (absolute risk increase, 27.2%; 95% CI, 14.4–40.1). 9 Rapid glycemic deterioration appears to preferentially occur in metabolically vulnerable individuals. This early deterioration often requires oral glucose-lowering therapy, after which glycemic levels generally improve without insulin. In contrast, the delayed pattern may represent a subtler and less predictable metabolic phenotype that can evade routine glucose surveillance. This stratification explains the diverging trajectories and powerfully critiques the one-size-fits-all approach to risk translation.
These converging data may lead to a shift in clinical management from passive monitoring to proactive metabolic risk management. Pretreatment evaluation that assessed comprehensive metabolic health and not just glucose could lead to risk-adapted monitoring. For example, high-risk patients, characterized by older age, higher baseline glycemia, obesity, adverse lipid profiles, and greater metabolic syndrome burden, warrant early HbA1c measurement at week 6 and possible self-monitoring, while low-risk patients could be counseled about the potential for delayed deterioration and commit to long-term annual follow-up, anchored by the documented 5-year risk. Regardless, the stratified risk data along with temporal pattern information should be fully disclosed to TED patients before medication, with joint decision-making by clinicians and patients. Embracing residual uncertainty about long-term reversibility and practicing shared decision-making are essential.
Ultimately, this study integrates baseline vulnerability, drug risk, and temporal heterogeneity into a unified picture. As we harness the ocular benefits of IGF-1R inhibitors, we must elevate metabolic safety to the same strategic priority, practicing long-term, individualized, and temporally informed risk management.