Highlights
In a propensity score-matched multinational cohort, teprotumumab treatment for thyroid eye disease was associated with significantly higher risks of prediabetes and diabetes over follow-up extending to 5 years.
At 5 years, 24% of teprotumumab-treated patients developed prediabetes versus 10% of matched controls, corresponding to a hazard ratio of 2.03 (95% CI, 1.51-2.72).
Teprotumumab exposure was also associated with increased incident diabetes risk, with a hazard ratio of 2.23 (95% CI, 1.46-3.43), and greater likelihood of starting antidiabetes medication, with a hazard ratio of 1.68 (95% CI, 1.27-2.22).
Although the relative risk increase was clinically relevant, the authors emphasized that the absolute excess risk remained modest, supporting risk-stratified monitoring rather than broad avoidance of treatment.
Background
Thyroid eye disease (TED), also called thyroid-associated ophthalmopathy, is a potentially sight-threatening and quality-of-life-limiting autoimmune disorder characterized by orbital inflammation, extraocular muscle enlargement, soft tissue expansion, pain, diplopia, lid retraction, and proptosis. The disease has historically been difficult to treat, particularly in patients with active, moderate-to-severe manifestations in whom corticosteroids, orbital radiation, and surgery each have important limitations.
Teprotumumab, a monoclonal antibody directed against the insulin-like growth factor-1 receptor (IGF-1R), changed the TED treatment landscape as the first US Food and Drug Administration-approved medical therapy specifically indicated for this condition. Its approval was driven by randomized trial data showing clinically meaningful improvements in proptosis, inflammation, and composite disease activity outcomes. Yet the same pathway that makes IGF-1R inhibition therapeutically useful in TED also raises metabolic concerns. IGF-1 signaling is intertwined with insulin signaling, and receptor blockade may impair glucose homeostasis, particularly in patients with pre-existing insulin resistance, prediabetes, or diabetes.
Hyperglycemia has been recognized in trials, case reports, and smaller retrospective series, but the longer-term frequency and clinical consequences of dysglycemia in routine practice have remained uncertain. That gap is clinically important because TED patients often have overlapping endocrine risk factors, including obesity, thyroid dysfunction, glucocorticoid exposure, smoking-related inflammation, and baseline metabolic disease. The present study addresses this unmet need by using a large electronic health record network to estimate the real-world glycemic risk associated with teprotumumab.
Study Design
Data source and design
Lishinsky-Fischer and colleagues conducted a retrospective cohort study using the TriNetX Global Collaborative Network, an electronic health records platform incorporating data from more than 165 health care organizations worldwide. This type of data source enables analysis at larger scale than most single-center TED cohorts, but it also inherits the usual limitations of routine-care databases, including dependence on coding accuracy and incomplete capture of laboratory testing.
Population and comparator
The investigators identified patients with TED who were treated with teprotumumab and matched them 1:1 to patients with TED who did not receive teprotumumab. After propensity score matching, the analytic cohort included 792 teprotumumab-treated patients and 792 controls. Propensity score matching is intended to reduce confounding by balancing measured baseline characteristics between groups, though it cannot eliminate bias from unmeasured differences such as disease severity, clinician prescribing preference, baseline nutritional status, or local monitoring practices.
Outcomes
The primary glycemic outcomes were incident prediabetes, incident diabetes, hyperglycemia defined by random blood glucose level at least 200 mg/dL, and initiation of antidiabetes medications. Follow-up ranged from 6 months to 5 years, and time-to-event analyses were performed using Kaplan-Meier methods to estimate cumulative incidence over time. Hazard ratios with confidence intervals were used to compare risk between groups.
Clinical rationale for endpoints
The selected endpoints are clinically meaningful because they capture not only coded diagnoses but also biochemical dysregulation and treatment escalation. In particular, new antidiabetes medication use offers a pragmatic marker of glycemic worsening with therapeutic consequences. However, the use of diagnosis codes and routine random glucose measurements may miss milder or transient disturbances, and the absence of standardized hemoglobin A1c collection limits mechanistic granularity.
Key Results
Prediabetes risk
The clearest signal in the study was for incident prediabetes. At 5 years, 24% of teprotumumab-treated patients had developed prediabetes compared with 10% of matched controls. The hazard ratio was 2.03 (95% CI, 1.51-2.72), indicating roughly a doubling of risk. This is a clinically relevant finding because prediabetes may represent the earliest detectable manifestation of treatment-associated impairment in glucose regulation and may also identify patients likely to progress further without surveillance or intervention.
From a practical standpoint, the absolute risk difference of about 14 percentage points over 5 years is not trivial. For clinicians counseling patients before treatment initiation, this figure may be easier to interpret than the hazard ratio alone. It suggests that dysglycemia is not merely an isolated post-infusion laboratory anomaly but may translate into sustained metabolic consequences in a meaningful minority of patients.
Incident diabetes
Teprotumumab was also associated with a significantly increased risk of incident diabetes, with a hazard ratio of 2.23 (95% CI, 1.46-3.43). Although the abstract does not provide the exact cumulative incidence percentages for diabetes in each group, the effect estimate indicates a substantial relative increase. This matters because progression from normal glucose tolerance or prediabetes to overt diabetes may change long-term cardiovascular risk, medication burden, perioperative planning, and endocrine follow-up requirements.
The association is biologically plausible. IGF-1R inhibition can interfere with insulin signaling crosstalk, potentially worsening insulin resistance and reducing glucose disposal. In predisposed patients, especially those with obesity or latent beta-cell dysfunction, this may unmask diabetes that would otherwise have remained subclinical for longer.
Need for antidiabetes therapy
Beyond coded diagnoses, teprotumumab-treated patients were more likely to require addition of antidiabetes medications during follow-up. The hazard ratio at 5 years was 1.68 (95% CI, 1.27-2.22). This endpoint strengthens the clinical relevance of the findings because it indicates that glycemic abnormalities were not purely nominal or incidental. They were sufficiently consequential to prompt pharmacologic management.
Medication initiation also serves as a proxy for care intensity and clinician concern. In an observational study, this is useful because some cases of hyperglycemia may never be assigned a formal diagnostic code, yet they still alter treatment plans. Even so, one must remember that prescribing behavior can vary across health systems and regions represented in TriNetX.
Hyperglycemia by random glucose
The study also evaluated hyperglycemia defined by random blood glucose at least 200 mg/dL. While the abstract does not provide the corresponding effect size in detail, inclusion of this endpoint is important because it captures acute biochemical abnormalities that may emerge during or soon after treatment. Such events may precede durable diagnoses of prediabetes or diabetes and could be particularly relevant for infusion-center monitoring.
Clinical Interpretation
The main clinical message is not that teprotumumab should be avoided broadly, but that its metabolic safety profile warrants systematic attention. The study supports a shift from anecdotal awareness of hyperglycemia to routine, protocolized risk assessment. The observed risks were consistent across several glycemic endpoints, and the longer-term follow-up suggests that dysglycemia may persist or evolve after treatment exposure rather than resolve uniformly after therapy completion.
At the same time, the authors appropriately note that the absolute risk increase was modest. This nuance is critical. For many patients with active moderate-to-severe TED, teprotumumab may prevent visual morbidity, reduce proptosis, improve diplopia, and decrease the need for rehabilitative surgery. Those benefits can be substantial. Therefore, the appropriate framework is individualized risk-benefit assessment, not reflex therapeutic avoidance.
Clinicians may find it useful to stratify patients into at least three practical categories before treatment: those with normal glycemia and low metabolic risk, those with prediabetes or metabolic syndrome, and those with established diabetes. The latter two groups likely merit closer monitoring, more frequent laboratory assessment, and lower thresholds for involving primary care or endocrinology.
Mechanistic Context
Teprotumumab blocks IGF-1R, a receptor implicated in TED pathobiology through effects on orbital fibroblasts, inflammatory signaling, and tissue remodeling. However, IGF-1R and insulin receptor pathways share signaling intermediates and can form hybrid receptors. Pharmacologic interruption of IGF-1R signaling may therefore reduce insulin sensitivity or alter glucose uptake. This mechanism fits the study findings and aligns with prior concerns raised during drug development and early postmarketing experience.
Importantly, treatment-emergent hyperglycemia may reflect both direct drug effect and patient susceptibility. People with obesity, family history of diabetes, pre-existing dysglycemia, glucocorticoid exposure, sedentary lifestyle, or high inflammatory burden may have less metabolic reserve. In those patients, teprotumumab may function less as a sole cause and more as a trigger that reveals an already fragile glycemic state.
Strengths and Limitations
Strengths
The study’s major strength is scale. A matched cohort of 1,584 patients with TED is sizable for this relatively uncommon disease, and the multinational EHR environment improves external relevance compared with tightly selected trial populations. The use of propensity score matching is also a methodological strength because it attempts to address baseline differences between treated and untreated patients. Finally, evaluation of several related glycemic endpoints provides a more complete picture than reliance on one laboratory threshold alone.
Limitations
As with any retrospective EHR study, residual confounding remains likely. Patients selected for teprotumumab may differ systematically from controls in ways not fully captured by recorded variables. Disease severity is a particularly important concern: patients with more severe TED may have had greater prior corticosteroid exposure or more intensive medical surveillance, both of which could influence glycemic detection.
Outcome ascertainment may also be incomplete or uneven. Diagnosis codes depend on clinician documentation, random glucose values are not standardized screening tools, and hemoglobin A1c data are not emphasized in the abstract. Surveillance bias is another possibility, because teprotumumab-treated patients may have had more frequent laboratory testing than controls, increasing the chance of detecting asymptomatic dysglycemia.
Finally, the abstract does not provide detailed subgroup analyses by baseline diabetes status, body mass index, race and ethnicity, corticosteroid use, or timing of glycemic events relative to infusion schedule. These details would be highly valuable for translating the findings into risk prediction at the bedside.
Implications for Practice
Pending more granular prospective data, a reasonable clinical approach is to monitor glucose proactively in all patients receiving teprotumumab and intensify surveillance in those at elevated metabolic risk. A practical framework would include baseline fasting glucose and hemoglobin A1c, review of diabetes history and current medications, and counseling regarding symptoms of hyperglycemia such as polyuria, polydipsia, fatigue, and blurred vision.
During treatment, interval glucose checks may be considered before infusions or at scheduled points during the treatment course, with more frequent assessment in patients with known diabetes or baseline prediabetes. Coordination between ophthalmology, endocrinology, and primary care is especially important when dysglycemia emerges, because early medication adjustment may prevent treatment interruption and reduce acute complications.
For patients with established diabetes, treatment is still possible, but shared decision-making is essential. Clinicians should discuss the known efficacy of teprotumumab in TED alongside the realistic possibility of worsening glycemic control. In some cases, optimizing diabetes therapy before the first infusion may be prudent. These measures are consistent with an individualized approach that preserves access to an effective TED therapy while acknowledging its metabolic tradeoffs.
How This Study Fits With Existing Evidence
Randomized trials of teprotumumab established efficacy and identified hyperglycemia as a notable adverse event, especially among patients with pre-existing carbohydrate intolerance. Smaller observational studies and case reports subsequently suggested that some patients experience marked or prolonged glycemic worsening. The present multinational cohort extends that literature by quantifying longer-term risk in a broader real-world population and by showing that the signal is not limited to transient laboratory changes.
The findings also reinforce current expert caution around metabolic monitoring. They do not negate the drug’s therapeutic value; rather, they sharpen the definition of good clinical use. In this respect, the study is best seen as practice-informing rather than practice-reversing.
Conclusion
This large-scale multinational cohort study indicates that teprotumumab treatment for thyroid eye disease is associated with increased risks of prediabetes, diabetes, and initiation of antidiabetes medications over follow-up of up to 5 years. The signal is biologically plausible and clinically coherent across multiple endpoints. Although the absolute excess risk appears modest, it is sufficiently consistent to justify routine glucose surveillance and early management planning.
For clinicians, the key takeaway is balance: teprotumumab remains an important therapeutic advance for TED, but its use should be accompanied by structured metabolic monitoring, especially in patients with baseline risk factors for dysglycemia. Future prospective studies should define the timing, reversibility, and predictors of glycemic deterioration more precisely, ideally incorporating standardized hemoglobin A1c and fasting glucose measurements as well as subgroup analyses by baseline metabolic status.
Funding and Trial Registration
The abstract provided does not report funding information or a ClinicalTrials.gov registration number for this retrospective database study.
References
Lishinsky-Fischer N, Pollack R, Gur Z. Teprotumumab-Associated Hyperglycemia: A Large-Scale Multinational Cohort Study. Thyroid. 2026 Apr 9:10507256261442501. PMID: 41954033.
Smith TJ, Kahaly GJ, Ezra DG, et al. Teprotumumab for thyroid-associated ophthalmopathy. N Engl J Med. 2017;376(18):1748-1761.
Douglas RS, Kahaly GJ, Patel A, et al. Teprotumumab for the treatment of active thyroid eye disease. N Engl J Med. 2020;382(4):341-352.
U.S. Food and Drug Administration. TEPEZZA (teprotumumab-trbw) prescribing information. FDA-approved labeling.
Douglas RS, Kahaly GJ, Ugradar S, et al. Management of thyroid eye disease: a Consensus Statement by the American Thyroid Association and the European Thyroid Association. Thyroid. 2022;32(12):1439-1470.
