Abstract
Background:
Moderate-to-severe thyroid eye disease (TED) is a potentially vision-threatening inflammatory condition that requires timely, evidence-based medical management. Although intravenous glucocorticoids remain the mainstay of therapy, several biologic and immunosuppressive agents have emerged as potential alternatives, particularly in steroid-refractory disease.
Objectives:
To evaluate and compare the efficacy, safety, and therapeutic positioning of medical treatments for active moderate-to-severe thyroid eye disease.
Data Sources and Methods:
A comprehensive search of PubMed, Embase, and the Cochrane Library was conducted through February 2025. Eligible studies included randomized controlled trials, meta-analyses, systematic reviews, observational cohorts, and selected case series evaluating pharmacological interventions for moderate-to-severe TED. Outcomes of interest were proptosis reduction, Clinical Activity Score (CAS) improvement, diplopia response, and safety/tolerability. A semi-quantitative synthesis approach was used to integrate evidence across heterogeneous study designs.
Results:
Fifty-eight studies met the inclusion criteria. Intravenous glucocorticoids (IVGCs) demonstrated the most consistent efficacy in controlling inflammatory activity, with modest effects on proptosis and favorable tolerability at cumulative doses below 8 g. Among biologic therapies, teprotumumab showed the greatest magnitude of benefit across all efficacy domains but was limited by safety considerations and access constraints. Rituximab and tocilizumab demonstrated moderate efficacy, particularly in glucocorticoid-resistant cases. Mycophenolate mofetil emerged as the most reliable non-biologic immunosuppressive option. Oral glucocorticoids and several adjunctive therapies showed limited or inconsistent benefit.
Conclusion:
This systematic review provides an integrated framework to support therapeutic decision-making in moderate-to-severe TED. Intravenous glucocorticoids remain the most consistently supported first-line therapy, while IGF-1R–targeted biologic therapy offers the most comprehensive efficacy across key clinical domains in selected patients.
Registration:
Not registered.
Plain Language Summary
Thyroid eye disease (TED) is an eye condition linked to thyroid problems that can cause pain, swelling, bulging of the eyes, double vision, and in severe cases, vision loss. When the disease is moderate to severe, it usually requires strong medical treatment. Currently, the main treatment is high-dose steroids given through a vein (intravenous glucocorticoids). These medicines help reduce inflammation and improve eye bulging but can have side effects if used for a long time or at high doses. In recent years, new medicines have been tested. One of them, teprotumumab, showed the best results in reducing eye bulging, inflammation, and double vision, but it is not always available and may cause side effects. Other medicines, such as rituximab and tocilizumab, may help patients who do not respond well to steroids. Among non-biologic drugs, mycophenolate mofetil was the most reliable option. Oral steroids and some other medicines were less effective or gave mixed results. Overall, this review of 58 studies shows that while IV steroids remain the first choice, newer biologic and non-biologic medicines are promising alternatives for certain patients. More research is needed to directly compare treatments, track long-term safety, and help doctors choose the best therapy for each patient.
Keywords
Introduction
Thyroid eye disease (TED), also referred to as Graves’ orbitopathy or thyroid-associated ophthalmopathy, is the most frequent extrathyroidal manifestation of Graves’ disease (GD), affecting up to 40% of patients with GD during their illness. 1 TED is an autoimmune condition driven by shared antigenic targets in the thyroid and orbit, particularly the thyrotropin receptor (TSHR) and the insulin-like growth factor 1 receptor (IGF-1R), leading to orbital fibroblast activation, adipogenesis, and deposition of glycosaminoglycans. 2 This process results in expansion of orbital tissues, causing a spectrum of ocular manifestations such as proptosis, eyelid retraction, diplopia, ocular surface exposure, and, in severe cases, optic neuropathy. 3 Risk factors associated with TED include smoking, older age, high thyroid-stimulating immunoglobulin levels, and radioiodine therapy. 1
The clinical expression of TED varies in activity and severity, with approximately 5% to 10% of cases progressing to a moderate-to-severe form that significantly compromises quality of life and may threaten vision. 1 Moderate-to-severe TED is defined by active inflammation (Clinical Activity Score ⩾3) and sight-threatening features, including significant diplopia, restrictive myopathy, corneal exposure, or optic nerve compression. Management of these patients requires a tailored approach, with intravenous glucocorticoids recommended as first-line therapy for active, moderate-to-severe disease. 3 Nonetheless, a substantial proportion of patients exhibit suboptimal response or intolerance to steroids, prompting the investigation and use of alternative treatments such as biologic therapies and targeted immunomodulators.2,3
In managing moderate-to-severe TED, therapeutic decisions must often be made in the context of heterogeneous evidence, limited head-to-head comparisons, and variable treatment availability. 4 This systematic review provides a consolidated evidence base that supports the continued use of intravenous glucocorticoids as first-line therapy, while identifying biologics such as teprotumumab, tocilizumab, and rituximab, as well as non-biologic immunosuppressants like mycophenolate mofetil, as viable alternatives in specific clinical scenarios. This review offers a pragmatic framework to inform individualized treatment planning by synthesizing efficacy and safety data across diverse study designs. Future research efforts should focus on standardizing outcome reporting, expanding access to novel agents, and addressing persistent gaps in comparative and long-term data to improve the quality and equity of care for patients with TED.
Methods
Systematic Review
The primary objective of this systematic review was to synthesize and critically appraise the available evidence on medical treatments for moderate-to-severe TED. The review was performed using Cochrane methodology, following the Cochrane Handbook for Systematic Reviews of Interventions, 5 and is reported in line with the PRISMA statement for systematic reviews of studies evaluating healthcare interventions. 6 While designed as a standalone academic product, this review also serves as a foundational input for a RAND/UCLA consensus project commissioned by the Colombian Association of Endocrinology. The methodological decision to include not only randomized controlled trials and meta-analyses but also observational studies was motivated by the requirements of the consensus process, which necessitates a broad and inclusive evidence base to inform expert deliberation across diverse clinical scenarios. This approach enhances the applicability of the findings to real-world decision-making, particularly in settings where direct comparative evidence is limited. The protocol for this review was not registered.
Inclusion Criteria
Studies were eligible for inclusion if they enrolled patients with moderate-to-severe active TED and reported on any form of medical treatment and any of the following outcomes: proptosis response, CAS change, and diplopia response. Definitions of proptosis reduction and diplopia response were extracted as reported by the original study authors. Given the heterogeneity and temporal evolution of outcome definitions across studies, no uniform thresholds were imposed; detailed study-specific criteria are provided in the Supplemental Tables. Active thyroid eye disease was defined according to the criteria used by the original study authors, most commonly a Clinical Activity Score (CAS) ⩾3, reflecting active inflammatory disease. This approach was adopted to accommodate the historical evolution of CAS definitions and variations in activity thresholds across studies, rather than retrospectively applying a single contemporary guideline-based definition. Eligible study designs included randomized controlled trials, systematic reviews and meta-analyses, comparative observational studies, and selected case series. Case series were included to address evidence gaps arising from the rarity of moderate-to-severe thyroid eye disease and the limited availability of randomized data for certain interventions. Case reports were excluded.
Search Methods for the Identification of Studies
Electronic Searches
The following databases were searched from inception to February 2025: MEDLINE (https://pubmed.ncbi.nlm.nih.gov/), EMBASE (https://www.embase.com/), and the Cochrane Library (https://www.cochranelibrary.com/). Studies published in any language, from any country, and at any date were considered eligible. In addition, the reference lists of relevant publications were screened to identify additional studies. The complete search strategy is detailed in Supplemental Table 1. Given the dual role of this review as an academic synthesis and as a preparatory evidence base for a RAND/UCLA consensus process, key disease-related terms were intentionally restricted to the title field to prioritize specificity and ensure inclusion of studies in which moderate-to-severe thyroid eye disease was the primary focus.
Study Selection and Risk of Bias
Two independent reviewers assessed the eligibility of studies based on predefined inclusion and exclusion criteria. The screening process was conducted using Rayyan, a web-based platform that enables anonymous, independent review and facilitates duplicate removal, tagging, and conflict resolution. 7 All titles and abstracts were screened in duplicate, and full-text eligibility assessments were performed independently. Although a third reviewer was designated to adjudicate disagreements, consensus was reached in all cases, and arbitration was not required. The selection process adhered to the PRISMA guidelines, 6 and the flow of studies through the selection process is depicted in the PRISMA flowchart, according to the PRISMA statement. 6 The risk of bias of RCTs was evaluated using the Cochrane Risk of Bias 2.0 tool (RoB 2). 8 For observational epidemiological studies, the ROBINS-I V2 tool was applied. 9 The methodological quality of included systematic reviews was assessed using the AMSTAR 2 tool. 10 The protocol for this systematic review was not registered in PROSPERO. The review was originally commissioned to inform a RAND/UCLA consensus process, and protocol registration was therefore not undertaken at the time of study initiation.
Presentation of the Results
The findings are presented through a narrative synthesis that integrates the extracted data. The report is organized into sections corresponding to specific interventions, with some sections further divided into subsections based on the type of study design.
In addition to the narrative synthesis, we developed a qualitative summary table to facilitate comparative interpretation across interventions. This table integrates findings from randomized controlled trials (RCTs), systematic reviews, meta-analyses, and observational studies, and case series, organized by pharmacologic strategy. Each intervention was assessed regarding its impact on 3 clinically relevant outcomes: proptosis response, Clinical Activity Score (CAS) change, and diplopia improvement. A semi-quantitative rating system (ranging from “–” to “++++”) was used to represent the relative magnitude and consistency of therapeutic effect, as well as safety and tolerability. Higher ratings indicate more substantial or more consistent evidence of benefit or safety.
Results
The results of this systematic review are based on a screening and selection process, depicted in Figure 1. 1038 records were retrieved from database searches, with 13 identified through other sources. After removal of duplicates, 541 records were screened, of which 81 full-text articles were assessed for eligibility. Ultimately, 58 studies met the inclusion criteria and were incorporated into the qualitative and quantitative syntheses. The remaining 23 full-text articles were excluded for reasons including narrative design, lack of relevant data, or restricted disease severity (eg, mild TED).
PRISMA flow diagram.
Table 1 summarizes the comparative assessment of the included interventions. This qualitative synthesis overviews therapeutic efficacy and safety across all pharmacologic options evaluated, including glucocorticoids, biologics, immunosuppressive agents, and other therapies. For each treatment, a semi-quantitative rating (ranging from “–” to “++++”) was assigned for key clinical outcomes, proptosis response, CAS change, and diplopia improvement, as well as safety/tolerability, based on the strength and consistency of the supporting evidence.
Qualitative Summary of Efficacy and Safety of Therapeutic Options for Moderate-to-Severe Thyroid Eye Disease.
Abbreviations: MAIC, matching-adjusted indirect comparison; RCT, randomized controlled trial.
Efficacy ratings are based on reported outcomes in proptosis response, CAS reduction, and diplopia improvement. Safety/tolerability reflects adverse event profiles as reported in clinical trials or observational studies. Ratings are semi-quantitative (– to ++++) and derived from the body of supporting evidence. While the semi-quantitative ratings primarily reflect consistency and magnitude of clinical efficacy and safety outcomes, greater interpretative weight was given to evidence derived from randomized controlled trials and studies with lower risk of bias as assessed by RoB2 and ROBINS-I tools.
Glucocorticoids
Glucocorticoid therapy remains the primary medical intervention for patients with active, moderate-to-severe TED, supported by decades of clinical use and formal guideline endorsement. This section synthesizes the evidence on glucocorticoid efficacy, safety, and administration strategies. Particular emphasis is placed on comparing intravenous and oral regimens, evaluating dosing protocols, and assessing the quality of the underlying evidence to inform current therapeutic practice. In this review, the term intravenous glucocorticoids (IVGC) refer to the therapeutic class, whereas intravenous methylprednisolone (IVMP) denotes the specific agent used in the majority of randomized trials.
Taken together, these meta-analyses consistently support the role of IVGC as the first-line treatment for active, moderate-to-severe TED over OGC. IVGC regimens had higher clinical response rates and better tolerability than OGC, with fewer systemic adverse effects such as hypertension and Cushingoid features. Among IVGC protocols, both weekly and monthly regimens showed favorable outcomes, and current evidence does not support the routine use of higher cumulative doses over the standard-dose regimen recommended by clinical guidelines. While the effect of IVGC on proptosis is modest and may not always result in clinically meaningful improvement, its overall benefit in reducing disease activity and improving patient-reported outcomes is well established. Diplopia outcomes were variably reported across meta-analyses and trials and appeared generally modest compared with effects on inflammatory activity. These findings reinforce the use of structured IVGC protocols.
Methodological quality of the included reviews was assessed using the AMSTAR 2 tool (Supplemental Table S3). All 4 studies fulfilled most critical domains, including using an a priori protocol, comprehensive literature searches, and appropriate methods for risk-of-bias evaluation in primary studies. Zhao et al. and Jia et al. provided GRADE assessments or ranking probabilities to enhance the interpretation of certainty in the estimates. Overall, the methodological quality was rated as moderate for all reviews, with common limitations including imprecision in effect estimates, small numbers of included studies in subgroup analyses, and limited reporting of funding sources or conflicts of interest in primary trials.
For instance, 2 studies22,23 found that IVGC regimens not only achieved higher clinical response rates (77% and 77.8%, respectively) but were also associated with fewer systemic side effects than high-dose oral prednisone. Similar outcomes were observed in other trials,16,17,20 consistently demonstrating superior CAS reduction and proptosis improvement in the IVGC arms. Additionally, comparisons of different IVGC regimens, such as weekly versus monthly dosing 15 or standard versus high cumulative doses, showed no clear superiority of higher dosing regarding treatment response or adverse events.
The risk of bias was evaluated using the Cochrane Risk of Bias 2 (RoB-2) tool for randomized controlled trials. Most studies demonstrated low risk across the majority of domains. However, 2 studies18,19 had “some concerns” primarily due to limitations in reporting the randomization process or outcome measurement. None of the studies were judged to have a high overall risk of bias (see Supplemental Table S5).
These observational studies provide complementary insights into using glucocorticoid-based therapies in routine and exploratory clinical settings. Across retrospective cohorts and case series, intravenous methylprednisolone (IV MPD) using the EUGOGO-recommended 4.5 g regimen showed variable response rates ranging from 40% to over 70%, depending on baseline characteristics and treatment context. Predictive factors for better outcomes included younger age, lower TBII levels, and short timing to pulse therapy after radioactive iodine therapy.
Alternative administration strategies were also explored, such as localized orbital injections or novel formulations like liposome-encapsulated prednisolone. While local injections demonstrated short-term improvements in inflammatory activity with minimal systemic toxicity, novel delivery systems showed limited efficacy. Across observational studies, CAS was the most consistently reported outcome, whereas proptosis and diplopia were reported less uniformly and showed variable or limited improvement depending on the intervention and study design
Comparative studies of weekly versus modified monthly IVGC schedules suggested similar overall efficacy, with some differences in soft tissue outcomes and patient-reported quality of life. Long-term observational data indicate that most patients experience spontaneous or partial remission without systemic intervention.
The risk of bias in the non-randomized studies was evaluated using the ROBINS-I Version 2 tool (Supplemental Table S7). Across the studies, key sources of bias arose primarily from inadequate control for confounding, absence of randomization, and lack of pre-specified protocols. The overall risk of bias was rated as serious in 4 of the 5 studies, and critical in one historical cohort, 29 where baseline confounding, selection bias, and reporting issues were most evident. While interventions were consistently well defined and outcomes generally measured using accepted clinical tools (eg, Clinical Activity Score), most studies’ lack of comparator groups or control for prognostic covariates limited causal inference. Only one study 28 attempted a comparative design without random allocation or matching. Most studies lacked methods to address missing data and did not present sensitivity analyses.
In summary, glucocorticoid therapy, particularly intravenous administration following standardized regimens, remains the cornerstone of initial treatment for active, moderate-to-severe TED. The evidence synthesized across systematic reviews, randomized trials, and observational studies consistently supports its effectiveness in controlling inflammation and improving key clinical outcomes, while highlighting its favorable safety profile compared to oral glucocorticoids. Although heterogeneity exists in dosing schedules and response rates, especially in real-world settings, the cumulative data reinforce the benefit of structured IV protocols as first-line therapy.
Biological Therapies
Biologic therapies have emerged as a promising alternative to conventional glucocorticoid regimens for patients with active, moderate-to-severe TED. A recent systematic review and meta-analysis by Fatani et al 30 synthesized data from 6 randomized controlled trials (n = 571) evaluating monoclonal antibodies—including rituximab, teprotumumab, and tocilizumab—compared with glucocorticoids or placebo. Pooled estimates demonstrated significant benefits in reducing clinical activity score (SMD = –1.44, 95%CI: –1.91-–0.97), proptosis (SMD = –4.96, 95%CI: –8.02-–1.89), and improving diplopia based on the Gorman score (OR = 3.42, 95%CI: 1.62-7.22), alongside improvements in quality of life (SMD = 2.64, 95%CI: 0.50 to 4.79). However, biologic treatments were also associated with an increased risk of adverse events (OR = 2.91, 95%CI: 1.12-7.56). Detailed characteristics of the included trials are available in Supplemental Table S7, and the methodological appraisal using the AMSTAR 2 tool is summarized in Supplemental Table S8.
Rituximab
To explore the role of rituximab in treating moderate-to-severe TED, a focused meta-analysis by Shen et al 31 pooled data from 4 randomized controlled trials comprising 293 patients. Rituximab was compared with intravenous glucocorticoids or placebo across key efficacy and safety outcomes. The analysis showed a statistically significant reduction in clinical activity score (WMD = 0.57, 95%CI: 0.25-0.89) in favor of rituximab, while differences in proptosis were modest and not statistically significant. Diplopia outcomes were inconsistently reported across trials and did not demonstrate a clearly reproducible benefit. Quality of life outcomes were mixed, with improvement observed in one trial but not replicated in another. Adverse event rates were comparable between treatment arms overall, although serious adverse events were more frequent in the rituximab group in 1 included study. Detailed characteristics and AMSTAR 2 assessment of the included trials are provided in Supplemental Tables S8 and S9.
Two randomized controlled trials have directly compared rituximab with placebo or glucocorticoids. In a double-blind trial, Stan et al 32 randomized 25 patients with active, euthyroid TED and CAS ⩾4 to receive 2 rituximab infusions (1000 mg) or placebo. At 24 weeks, no significant difference was observed in the proportion of patients achieving a ⩾2-point CAS reduction (31% vs 25%, P = .75). Disease inactivation by week 52 was slightly more frequent in the rituximab group (60% vs 40%), but differences in proptosis and diplopia were minimal. While the overall number of adverse events was higher among rituximab recipients, most were mild to moderate. In contrast, Salvi et al 33 found superior efficacy of rituximab compared with intravenous methylprednisolone in a trial enrolling 31 patients. At 24 weeks, 100% of patients in the rituximab arm achieved disease inactivation versus 69% in the glucocorticoid group (P = .043), with lower relapse and surgery rates in the rituximab arm. However, both trials had methodological concerns that influence confidence in these findings. Further details and RoB 2.0 assessments are reported in Supplemental Tables S10 and S11.
Additional evidence on rituximab derives from observational studies and case series. Manousou et al 34 evaluated rituximab plus methotrexate in patients who failed to respond to IV glucocorticoids in a nonrandomized controlled study. Compared with continued glucocorticoid therapy, the rituximab-based regimen did not significantly improve CAS or quality of life, and response rates at 12 weeks were similarly low across all groups. The study was rated as having a serious risk of bias due to confounding and mixed study design (Supplemental Table S12).
Two retrospective case series further examined low-dose rituximab regimens. Insull et al 35 treated 12 patients early in the disease course using a single 100 mg dose combined with adjunctive therapies. Significant reductions in CAS and VISA severity scores were observed, with no serious adverse events reported. Another study 36 evaluated low-dose rituximab (100-400 mg cumulative) in 15 patients refractory to standard treatments. CAS decreased significantly at short-term (week 17) and long-term (week 64) follow-up, and TRAb levels declined. However, improvements in proptosis or ocular motility were not consistently observed. Both studies were rated as having a serious risk of bias due to retrospective design and lack of comparators (see Supplemental Tables S12 and S13).
Tocilizumab
Tocilizumab, an IL-6 receptor antagonist, has gained attention as a therapeutic option for patients with corticosteroid-resistant or relapsing TED. A recent systematic review by Duarte et al 37 synthesized data from 29 studies involving 208 patients, predominantly treated in the setting of glucocorticoid resistance. Although only one randomized controlled trial was identified, observational studies and case series findings consistently reported meaningful reductions in clinical activity. A ⩾3-point decrease in CAS was frequently achieved, while proptosis outcomes varied, being non-significant in the controlled trial but more favorable across uncontrolled reports. Diplopia outcomes were reported less consistently, with variable improvement across uncontrolled series. Relapse following discontinuation occurred in approximately 8.2% of cases. Both intravenous and subcutaneous formulations appeared effective, and adverse events were generally mild and manageable. While the review lacked meta-analytic pooling or formal bias appraisal, it supports the emerging role of tocilizumab in refractory TED. Study-level details are provided in Supplemental Table S8, and AMSTAR 2 quality assessment in Supplemental Table S9.
The clinical efficacy of tocilizumab was evaluated in a double-blind randomized controlled trial involving 32 patients with moderate-to-severe disease unresponsive to glucocorticoids. Participants received 4 monthly infusions of 8 mg/kg or placebo, with follow-up extending to 40 weeks. By week 16, 93.3% of the tocilizumab group achieved a ⩾2-point CAS reduction compared to 58.8% in the placebo group (P = .04), and disease inactivation (CAS <3) occurred in 86.7% versus 35.2% (P = .005). Significant improvements were also seen in the EUGOGO composite outcome (73.3% vs 29.4%, P = .03), and median proptosis reduction reached —1.5 mm (P = .01). Adverse events were more common in the active treatment group, but remained mild. This trial demonstrated robust clinical benefit and was judged to have low risk of bias (see Supplemental Tables S10 and S11). 38
Real-world observational studies provide additional insights. A retrospective case series involving 11 patients (22 eyes) reported substantial clinical improvements following monthly intravenous infusions of 8 mg/kg. CAS decreased from a median of 5 to 1 by week 16, with 75% of eyes achieving ⩾2-point reductions and 67% attaining inactivation (CAS ⩽1). Proptosis improved in most evaluable cases, and diplopia resolved or improved in several patients. Quality of life scores improved in 9 of 11 participants, and adverse events were limited to mild neutropenia in 4 cases. This study was rated as having a serious overall risk of bias due to its retrospective design and small sample size (Supplemental Tables S12 and S13). 39
Similarly, a prospective single-center study of 10 patients with corticosteroid-resistant TED—including 3 with compressive optic neuropathy—showed marked improvements after treatment with intravenous tocilizumab. CAS decreased from 4.7 to 0.2 (P = .003), and TSI levels declined significantly (P = .006). Visual acuity and field index improved substantially in those with optic neuropathy, and no adverse events or disease reactivation were observed. The prospective design and standardized outcome assessment led to a moderate risk of bias rating (see Supplemental Tables S12 and S13). 40
Teprotumumab
Teprotumumab, an IGF-1R antagonist, has emerged as a targeted therapy for active, moderate-to-severe TED. A comparative analysis using a matching-adjusted indirect comparison (MAIC) evaluated its effectiveness against intravenous glucocorticoids (IVMP) and placebo. 13 Pooling individual patient data and published trial results (n = 79 for teprotumumab, n = 419 for IVMP, n = 83 for placebo), the analysis showed that teprotumumab produced greater proptosis reduction (mean difference −2.31 mm, 95%CI: −3.45-−1.17) and higher diplopia response (OR 2.32, 95%CI: 1.07-5.03) compared to IVMP. By contrast, IVMP demonstrated only a marginal benefit over placebo. While the analysis applied valid adjustment techniques, limitations included a lack of protocol registration and formal risk of bias assessment. Summary characteristics are presented in Supplemental Table S8, with AMSTAR 2 appraisal in Supplemental Table S9.
Two randomized controlled trials have established the efficacy of teprotumumab. A phase 3 multicenter trial randomized 83 patients to receive 8 infusions of teprotumumab or placebo over 24 weeks. 41 Teprotumumab significantly outperformed placebo in proptosis response (83% vs 10%, P < .001), overall response (78% vs 7%), diplopia improvement (68% vs 29%), CAS reduction to 0 to 1 (59% vs 21%), and quality of life. Adverse events were primarily mild or moderate. A preceding phase 2 trial in 88 patients showed similar results: proptosis response occurred in 71% versus 20% with placebo, and diplopia improved in 53% versus 25%. Both trials demonstrated consistent clinical benefits across endpoints and were judged as low risk of bias (see Supplemental Tables S10 and S11). 42
Real-world data have further contextualized these findings. A retrospective study of 118 patients who completed the full treatment course reported a median MRD1 reduction from 5.25 to 4.66 mm (P < .001), with long-term stability or improvement in most eyes. 43 However, 15% required eyelid surgery post-treatment, and the response was not always associated with proptosis improvement. Another case series with 21 patients and over 20 months of follow-up found an initial response in 81%, but nearly half experienced disease reactivation after approximately 1 year.44,45 Diplopia often persisted, and a third of patients required orbital decompression. A multicenter cohort study of 66 patients with refractory TED found high response rates: 85.9% for proptosis, 93.8% for CAS, and 69.1% for diplopia. 46 Nevertheless, adverse events were common (76%), including severe hearing loss in 6%. These studies were rated as having a serious overall risk of bias due to retrospective designs and potential confounding (see Supplemental Tables S11 and S12).
A separate cohort of 119 patients was followed to examine proptosis regression post-treatment. 47 Among initially responsive eyes, 65.4% showed partial regression over time, and 25.6% worsened beyond baseline. Greater initial proptosis and male sex were predictors of regression. Finally, in the OPTIC-X extension study, teprotumumab was evaluated in patients previously treated with a placebo or those with flare or inadequate response. 48 Among those newly treated, 89.2% achieved a proptosis response, 65.6% reached CAS ⩽1, and over half resolved diplopia. Retreated patients also showed benefit, particularly in cases of relapse. The majority maintained their response at 48 weeks. This prospective study was rated as moderate risk of bias (see Supplemental Tables S12 and S13).
Adalimumab
Adalimumab, a tumor necrosis factor-alpha (TNF-α) inhibitor, has been explored as a potential steroid-sparing agent in thyroid eye disease (TED). In a retrospective case series, 10 patients with active inflammatory-stage TED received an initial subcutaneous dose of 80 mg followed by 40 mg every 2 weeks over 3 months. 49 A decrease in the inflammatory composite score (ICS) was observed in 6 patients, while it increased in 3 and remained stable in 1. Among patients with more severe inflammation at baseline (ICS >4), the mean score reduction was 5.2 ± 2.7 points (P < .01), and 4 of these individuals reported subjective clinical improvement. However, no meaningful changes in proptosis or extraocular motility were documented. One serious adverse event—sepsis—was reported, but overall the regimen was well tolerated. Given the small sample size, lack of a comparator, and absence of standardized CAS reporting, the study was rated as having a serious overall risk of bias (see Supplemental Tables S12 and S13).
In summary, biologic therapies represent an expanding frontier in managing active, moderate-to-severe thyroid eye disease, particularly in patients with inadequate response or intolerance to glucocorticoids. Among the agents evaluated, teprotumumab has demonstrated the most robust evidence of efficacy across multiple randomized trials and observational cohorts, with consistent improvements in proptosis, diplopia, disease activity, and quality of life. Tocilizumab offers a promising alternative in steroid-resistant cases, supported by a positive randomized trial and corroborative real-world data. Evidence for rituximab remains mixed, with heterogeneous results across trials and modest benefits in uncontrolled studies. Experience with TNF-α inhibitors such as adalimumab is limited and exploratory. While safety profiles have generally been acceptable, variability in study designs and risk of bias highlight the need for better head-to-head comparisons and longer-term follow-up to delineate these agents’ role in routine clinical practice.
Other Immunosuppressive Therapies
Mycophenolate Mofetil
Mycophenolate mofetil (MMF) use in moderate-to-severe TED has been primarily evaluated through a single systematic review and meta-analysis by Feng et al, 50 representing the most comprehensive synthesis of controlled clinical trials available to date. This review included 6 comparative studies investigating MMF either as monotherapy or in combination with GCs, providing critical insights into its efficacy and safety relative to standard GC-based regimens. Interventions included oral MMF at various doses, compared with oral or intravenous glucocorticoids such as methylprednisolone or prednisone. Outcomes assessed included overall response, changes in clinical activity measure using CAS, and incidence of adverse events (AEs). Risk of bias in the included studies was evaluated using the Cochrane RoB tool, and publication bias was assessed via funnel plot analysis.
The primary pooled analysis demonstrated that MMF, used alone or with GCs, was significantly more effective than GCs alone in achieving composite clinical response (OR 3.34, 95%CI 2.17-5.14; P < .00001). Subgroup analyses confirmed the benefit of MMF both as monotherapy (OR 4.46, 95%CI 2.52–7.87) and in combination with GCs. Three studies reported CAS as an outcome, showing a significant reduction with MMF (WMD −0.29, 95%CI −0.48-−0.10; P = .002; I2 = 0%). Regarding safety, MMF was associated with a lower overall incidence of AEs (OR 0.20, 95%CI 0.06-0.72; P = .01), although the analysis exhibited high heterogeneity (I2 = 79%). Notably, when combined with GCs, MMF did not consistently reduce the rate of AEs compared to GC monotherapy. Risk of bias assessment revealed that 2 studies were at high risk for selection and performance bias, and 3 additional studies presented high risk in at least one domain. Only one trial met low-risk criteria across all domains. Visual inspection of the funnel plot indicated a low likelihood of publication bias. Sensitivity analyses confirmed the robustness of the findings for treatment response.
According to the AMSTAR 2 tool, the methodological quality of the review was rated as moderate. Strengths included explicit inclusion criteria, comprehensive literature searches (including Chinese databases), dual reviewer processes, and appropriate meta-analytic methods. However, the review lacked a registered protocol, did not report funding sources, and did not perform a formal GRADE evaluation of evidence certainty or explore the impact of risk of bias on pooled estimates. In summary, current evidence suggests that MMF may offer superior efficacy and favorable tolerability compared to glucocorticoids in managing moderate-to-severe TED. Nonetheless, conclusions are limited by the moderate quality of the review and methodological concerns in the included trials.
Overall, evidence from a moderate-quality systematic review suggests that mycophenolate mofetil, either as monotherapy or in combination with glucocorticoids, provides superior clinical response and improved tolerability compared with glucocorticoids alone; however, the strength of these conclusions is tempered by methodological limitations and residual risk of bias in the underlying studies.
Methotrexate
The available evidence on the role of methotrexate in the treatment of TED remains limited but suggests a potential steroid-sparing effect in selected clinical contexts. The most robust data stem from a randomized controlled trial conducted by Shen et al, 51 who evaluated the efficacy and safety of adding methotrexate to either reduced or full-dose IVGC in patients with active, moderate-to-severe TED. A total of 138 patients were randomized into 3 arms: reduced-dose IVGC (3.0 g), reduced-dose IVGC plus methotrexate (3.0 g + MTX), and full-dose IVGC (4.5 g). The primary endpoint, defined as achieving a ⩾2-point reduction in CAS without worsening in other disease domains at week 12, was met in 75.6% of patients in the combination group, significantly higher than in the reduced-dose IVGC group (56.5%, P = .04), and non-inferior to the full-dose group (71.1%). CAS improvements were sustained through week 24, with mean reductions of −3.7, –2.9, and −3.4 in the combination, reduced, and full-dose groups, respectively. Rates of proptosis response and diplopia improvement were comparable among groups. Adverse event frequency was similar, although mild liver function test abnormalities occurred more frequently in the methotrexate arms. The study was judged to have low risk of bias across all domains using the Cochrane RoB 2.0 tool.
Observational evidence on methotrexate in TED is scarce and methodologically limited. Strianese et al. 52 reported findings from a retrospective comparative case series involving 36 patients with active TED who had discontinued corticosteroids due to adverse effects. Patients received weekly weight-adjusted oral methotrexate for 12 months, accompanied by folic acid supplementation. Significant CAS reductions were observed at 3, 6, and 12 months (P < .0001), along with improved ocular motility at later time points (P < .001). However, there were no significant changes in proptosis, upper eyelid position (MRD), or visual acuity, and no adverse events were reported. Risk of bias was assessed using the ROBINS-I tool and rated as serious, primarily due to the lack of a comparator group, absence of adjustment for key confounders (eg, baseline disease activity, smoking, prior therapies), and unblinded outcome assessments. The retrospective design further limits the internal validity of the findings, precluding strong causal inference.
Overall, available evidence suggests that methotrexate may serve as a steroid-sparing adjunct to intravenous glucocorticoids in selected patients with active moderate-to-severe TED, achieving comparable disease activity control with reduced cumulative steroid exposure; however, its impact on proptosis and diplopia appears limited, and conclusions are constrained by the scarcity of high-quality data beyond a single randomized trial.
Sirolimus
The available evidence on sirolimus for the treatment of TED is limited to 2 non-randomized observational studies conducted by the same research group in Italy. These studies explored sirolimus as a second-line option in patients with active, moderate-to-severe TED who had either failed or were ineligible for further glucocorticoid therapy.
The first study was a retrospective cohort., 53 compared sirolimus (initial dose 2 mg orally, followed by 0.5 mg/day for 12 weeks) to intravenous methylprednisolone in 30 patients (15 per group). The primary endpoint was overall TED response at 24 weeks, defined as meeting at least 2 improvement criteria (CAS, exophthalmos, eyelid aperture, ductions, or visual acuity) without worsening in any domain. The response rate was significantly higher in the sirolimus group (86.6% vs 26.6%, OR 17.8, 95%CI 2.7-116.8; P = .0026). Sirolimus also led to superior outcomes in secondary endpoints such as quality of life, proptosis, and CAS reduction. No serious adverse events were reported, and drug tolerability was favorable. Despite the observational design and potential for selection bias, the investigators employed consecutive sampling and baseline group comparability to minimize confounding. However, as treatment allocation was not randomized and blinding was absent, the study was judged to have a moderate overall risk of bias according to ROBINS-I, mainly due to confounding and selection bias.
The second study, published in 2024, 54 examined the long-term durability of response in 13 patients treated with sirolimus, followed for a median of 27 months. Most patients maintained clinical remission with low CAS and improved quality of life scores. Due to its small sample size, lack of control group, and absence of a comparative design, this study presents a serious risk of bias under ROBINS-I, particularly in confounding, missing data handling, and outcome assessment. Nevertheless, it offers complementary evidence on long-term tolerability and durability of benefit.
Overall, observational evidence suggests that sirolimus may offer clinically meaningful improvements in disease activity, proptosis, and quality of life in selected patients with active moderate-to-severe TED who are refractory to or intolerant of glucocorticoids; however, the strength of this signal is limited by non-randomized designs, small sample sizes, and residual confounding, precluding definitive conclusions regarding comparative efficacy.
Cyclophosphamide
The available evidence on cyclophosphamide (CYC) for TED has been systematically reviewed and meta-analyzed by Xiang et al. 55 This meta-analysis included 13 randomized controlled trials (RCTs), all conducted in China and published in Chinese, encompassing a total of 932 patients. The studies compared combined GC and CYC therapy versus GC alone or a negative control. The primary outcome was clinical response, uniformly defined across studies as a ⩾2-point reduction in CAS or the achievement of disease inactivation (CAS <3). The pooled analysis showed a significantly higher response rate in the combination group (RR 1.27, 95%CI 1.19-1.37), with consistent benefits across both subgroups: CYC/GCs versus GCs (RR 1.24, 95%CI 1.14-1.36) and CYC/GCs versus negative control (RR 1.43, 95%CI 1.16-1.76). Subgroup analyses by GC backbone (CYC/DEX, CYC/MPS, CYC/P) did not yield statistically significant differences, though point estimates varied. Publication bias was suggested by Egger’s test (P = .0023), but adjusted analysis using the trim-and-fill method yielded similar estimates (RR 1.26, 95%CI 1.18-1.36). Only one trial reported on proptosis reduction, and 5 reported adverse events, limiting the scope for conclusions beyond clinical response.
The quality of the included RCTs was assessed using the Cochrane Risk of Bias tool, with most studies showing moderate risk due to limited reporting of allocation concealment and blinding. Evaluation of the methodological quality of the meta-analysis using the AMSTAR 2 tool revealed moderate confidence in the results. The review adhered to key domains such as comprehensive search strategy, dual independent screening, detailed risk of bias evaluation, and appropriate synthesis methods. However, limitations were noted in the lack of protocol registration, restricted language inclusion, insufficient consideration of funding sources, and conflict of interest. Additionally, the GRADE framework was applied, but the certainty of evidence was rated as low to moderate, primarily due to limitations in safety data, poor follow-up reporting, and lack of ethnic diversity. Overall, this review provides preliminary support for CYC as an adjunct to GC therapy in TED, particularly in patients with poor response to steroids or those at high risk of relapse.
Taken together, evidence from a moderate-quality meta-analysis of randomized trials suggests that cyclophosphamide, when used as an adjunct to glucocorticoids, is associated with higher rates of disease activity improvement compared with glucocorticoids alone; however, limited reporting on proptosis, diplopia, and safety outcomes, along with methodological constraints and low-to-moderate certainty of evidence, restrict its clinical positioning.
Cyclosporine
Two randomized controlled trials have evaluated the role of cyclosporine in combination with glucocorticoids for the treatment of active, moderate-to-severe TED. In an early trial by Kahaly et al, 56 20 patients were randomized to receive either prednisone alone or prednisone plus cyclosporine over 12 weeks. Patients in the combination arm showed significantly greater improvement in inflammatory signs, ocular motility, and Clinical Activity Score (CAS), with a reported response rate of 80% versus 50% in the prednisone-alone group. However, nephrotoxicity was observed in 30% of patients receiving cyclosporine, raising safety concerns. The study lacked allocation concealment and blinding, and no pre-registered protocol was available, contributing to a “some concerns” rating in the risk of bias assessment using RoB 2.0.
Prummel et al 57 conducted a larger, double-masked, randomized controlled trial involving 84 patients with severe TED. Participants were assigned to receive either prednisone alone or prednisone combined with cyclosporine for a total of 16 weeks, followed by 16 weeks of cyclosporine monotherapy in responders. The primary outcome was a composite ophthalmic score assessing inflammation, proptosis, and diplopia. The combination therapy resulted in significantly higher response rates (86% vs 53%, P < .01), with sustained benefits in responders after prednisone withdrawal. Adverse events included mild to moderate nephrotoxicity in the cyclosporine arm. This study was rated as low risk of bias based on the RoB 2.0 tool, as it employed proper randomization, blinding, complete outcome data, and intention-to-treat analysis, and showed no evidence of selective reporting.
The available randomized controlled trials suggest that cyclosporine, when combined with glucocorticoids, may enhance treatment response in patients with severe or glucocorticoid-refractory TED compared to glucocorticoids alone. Improvements in inflammatory activity, motility, and composite ophthalmic outcomes were consistently observed across both studies. However, the clinical utility of cyclosporine is limited by its adverse effect profile, particularly nephrotoxicity, which occurred in a notable proportion of patients. Moreover, the emergence of alternative immunomodulatory agents with more favorable safety profiles may have contributed to a decline in clinical interest and research activity involving cyclosporine in this setting. As a result, evidence remains confined to early trials with no recent updates.
Intravenous Immunoglobulin
Baschieri et al 58 conducted a prospective, non-randomized comparative study to assess the efficacy and safety of high-dose intravenous immunoglobulin (IVIG) versus systemic corticosteroids in treating moderately to severely TED. The study included 65 patients, 35 receiving IVIG and 30 treated with corticosteroids. Although allocation was not randomized, the 2 groups’ baseline characteristics were comparable in age, sex, duration of TED, and thyroid status. The primary outcomes were evaluated using the NOSPECS classification, subjective eye score, and orbital computed tomography, with assessments performed in a blinded manner. At the end of follow-up (mean 21 months for IVIG), overall response, defined as a reduction in the highest NOSPECS class or grade, was observed in 76% of IVIG-treated patients and 66% of corticosteroid-treated patients, a difference that was not statistically significant. Both groups also showed Comparable improvements in soft tissue involvement, diplopia, and proptosis. Importantly, IVIG was associated with significantly fewer adverse events, while corticosteroid therapy led to serious complications, including hemorrhagic gastritis, psychosis, and reduced bone mineral content.
Using the ROBINS-I tool, the overall risk of bias was judged as serious. The primary source of bias was confounding, given the lack of control for baseline disease severity and other prognostic variables. Selection bias was also a concern due to non-random treatment allocation. However, the classification of interventions and outcome measurement was appropriate, and adherence to assigned treatment was adequate. Some problems were noted in reporting adverse events and missing data, particularly in the corticosteroid group.
In summary, the evidence supporting immunosuppressive therapies beyond conventional glucocorticoids and biological treatment in moderate-to-severe TED remains heterogeneous in quality and design. Randomized controlled trials (RCTs) and meta-analyses of RCTs, such as those evaluating mycophenolate mofetil, methotrexate, cyclophosphamide, and cyclosporine, offer the most reliable estimates of efficacy and safety. MMF and methotrexate appear promising as glucocorticoid-sparing strategies, with acceptable tolerability profiles and consistent improvements in clinical activity scores. Conversely, cyclosporine and cyclophosphamide have shown therapeutic potential but are limited by toxicity concerns and methodological shortcomings in the underlying studies. Observational studies assessing agents such as sirolimus and intravenous immunoglobulin provide preliminary insights into alternative mechanisms of action and long-term durability of effect. Still, their inherent susceptibility to confounding, selection bias, and lack of control groups precludes definitive conclusions. Overall, while emerging therapies may hold promise for specific subgroups, particularly those with glucocorticoid intolerance, refractory disease, or high relapse risk, the generalizability and strength of the current evidence are constrained by small sample sizes, limited follow-up, and variable methodological rigor. High-quality comparative trials with standardized outcome definitions and stratification by disease phenotype are needed to define their role in clinical practice further.
Other Therapies
A variety of immunomodulatory and symptomatic treatments have been investigated for TED beyond the conventional use of glucocorticoids, orbital radiation, and decompression surgery. These “other therapies” encompass pharmacologic agents with anti-inflammatory, anti-fibrotic, or immunosuppressive properties.
Somatostatin Receptor Analogs
Somatostatin receptor analogs (SRAs), such as octreotide and lanreotide, have been explored as potential immunomodulatory treatments for TED due to their anti-inflammatory and anti-angiogenic effects. Although interest in SRAs has declined with the advent of targeted biologics, several clinical trials and observational studies have evaluated their efficacy in patients with active moderate-to-severe TED.
A randomized, double-masked, placebo-controlled trial 59 assessed the efficacy of long-acting release (LAR) octreotide in 29 euthyroid patients with active TED (CAS ⩾3). Patients received 4 monthly injections of octreotide LAR (20 mg) or placebo. The median CAS improvement was significantly greater in the treatment group (−2.5 vs −1.0; P = .02), and 86% achieved a reduction of ⩾2 points in CAS. Improvements in eyelid fissure width were also significant. However, there were no differences in proptosis or extraocular motility. The treatment was well tolerated. Using RoB 2.0, the study was rated as having “some concerns,” primarily due to small sample size and minor imbalances at baseline.
Kung et al 60 conducted an open-label randomized trial comparing subcutaneous octreotide (n = 8) to oral prednisone (n = 10) in patients with moderately to severely TED. Both groups experienced significant CAS reduction and improvements in palpebral aperture, but only the prednisone group showed reductions in intraocular pressure and muscle size. No differences were found in proptosis or overall response rates. While octreotide was well tolerated, prednisone was associated with infections and hypertension. The study had “some concerns” for bias due to lack of blinding, absence of allocation concealment, and small sample size.
In a prospective observational study, 61 10 patients with active TED and positive somatostatin receptor scintigraphy received either octreotide or lanreotide for 3 months and were compared to a control group. Significant reductions in CAS and somatostatin tracer uptake were observed (CAS from 5.4 ± 1.4 to 1.75 ± 0.82; OR: SK ratio from 1.85 to 0.99; both P < .001). No benefit was seen in the untreated control group. The study was rated as having a serious risk of bias under ROBINS-I, primarily due to a lack of adjustment for confounders.
Botulinum Toxin
The role of botulinum toxin (BTX) in treating TED has been explored primarily in non-comparative studies and summarized in a systematic review and meta-analysis, 62 representing the most comprehensive synthesis of available evidence. The authors included 30 studies published between 1984 and 2022, with a combined sample of 299 patients treated for upper eyelid retraction and 205 for TED-associated strabismus. Most studies involved patients with inactive or mild-to-moderate TED, and only a few included patients with active or severe disease, typically at earlier phases of treatment or when systemic inflammation had subsided. The included studies were observational primarily (case series, cohort studies) and lacked consistent comparative groups.
BTX was delivered via transcutaneous, transconjunctival, or subconjunctival routes for eyelid retraction. The pooled success rate—defined as a ⩾2 mm reduction in marginal reflex distance 1 (MRD1) or eyelid position reaching or falling below the superior limbus—was 84% (95%CI: 0.69-0.92). The mean reduction in MRD1 was 2.42 mm (95%CI: –2.03-–2.80), with the effect typically lasting 1 to 6 months. Adverse events were infrequent and mostly mild, including transient ptosis (13%) and diplopia (2%).
In patients with TED-related strabismus, the results were more heterogeneous. Only 24% achieved resolution of diplopia following BTX injection, and 58% eventually required strabismus surgery, indicating limited efficacy as monotherapy. The rate of ptosis in this group was low (2%), and no severe complications were consistently reported, although a few isolated cases of optic neuropathy were noted in the broader literature.
The methodological quality of the meta-analysis was evaluated using the AMSTAR 2 tool and judged to be of moderate confidence. The review adhered to several key methodological standards: it employed a comprehensive search strategy across 4 major databases (PubMed, Embase, Web of Science, and Cochrane Library), used dual independent screening with high agreement (94.6%), and followed PRISMA reporting guidelines. The authors applied appropriate meta-analytic models, logit transformation for proportions, and inverse variance methods for continuous outcomes.
Nevertheless, the review exhibited essential limitations. No study protocol was registered or published, and the quality assessment relied on the NIH Quality Assessment Tool, which, while systematic, is not aligned with GRADE or Cochrane risk-of-bias frameworks. Risk of bias was not discussed in detail for individual studies, and no subgroup analyses were performed based on TED activity or severity. The substantial heterogeneity observed across pooled estimates, especially for outcomes such as MRD1 reduction (I2 = 71%), was not fully explored through meta-regression or sensitivity analyses. In addition, many included studies had small sample sizes and lacked standardized outcome definitions.
Colchicine
A prospective, randomized, parallel-group trial conducted at the Universidade Federal de São Paulo evaluated the efficacy and tolerability of colchicine compared to oral prednisone in treating the active inflammatory phase of TED. 63 Twenty-two euthyroid patients with moderate-to-severe TED (NOSPECS classes II–IV) were randomized into 2 arms: colchicine (1.5 mg/day for 1 month, then 1 mg/day for 2 months) and prednisone (0.75 mg/kg/day tapered over 3 months). Clinical Activity Score (CAS) and MRI signal intensity ratio (SIR) of the extraocular muscles were assessed before and after treatment. Blinded evaluation of outcomes was performed at baseline and after 3 months.
Significant improvement in CAS was observed in both groups (colchicine: median CAS from 5.0 to 3.0, P < .0001; prednisone: 4.0 to 1.0, P = .0003). The proportion of orbits achieving a reduction of at least 2 CAS points was equal (68%) in both arms. Similarly, SIR significantly decreased in both groups, with no statistical difference in the magnitude of change between them. Importantly, no side effects were reported in the colchicine group, while the prednisone group exhibited frequent steroid-related adverse events, including hypertension, edema, and gastrointestinal symptoms.
The trial was assessed to have some concerns for the overall risk of bias. While the study employed appropriate randomization (using permuted blocks), blinded assessment of outcomes, and pre-specified outcome criteria, limitations include the absence of allocation concealment and the lack of blinding of participants and treating physicians, which may have introduced performance bias. The small sample size and lack of trial registration or pre-specified protocol raise additional concerns. Nevertheless, the outcome measurement was objective and appropriately blinded, and the data were complete and analyzed using appropriate statistical methods.
Triamcinolone
Bordaberry et al. conducted a prospective, non-randomized pilot study evaluating the efficacy of repeated peribulbar injections of triamcinolone acetonide in patients with active, moderate-to-severe TED, including cases with optic neuropathy (ON). 64 Twenty-one patients (42 orbits) received four 20 mg injections at 2-week intervals and were followed for at least 14 months.
At baseline, the mean CAS was 6.38 ± 1.49, significantly decreasing to 1.8 ± 1.12 at 6 months (mean reduction 4.57 points; P = .01) among the 10 patients with TED-associated optic neuropathy. 66% experienced complete resolution of ON with triamcinolone alone, while others required decompression surgery. Visual acuity, motility impairment, and inflammation scores all improved significantly. Treatment was generally well tolerated; only 2 patients experienced transient increases in intraocular pressure, and no systemic adverse events were reported.
Based on the ROBINS-I tool, the risk of bias for the study by Bordaberry et al. was considered serious overall. The primary concern lies in confounding, as the study lacked a control group and did not account for potential prognostic variables such as baseline disease severity, smoking status, or prior treatments. Selection bias is also present due to non-random inclusion of participants based on clinical judgment, potentially favoring those more likely to respond. While the classification of the intervention and adherence to the protocol were appropriate, the measurement of outcomes was not blinded, which introduces detection bias, especially in subjective assessments like CAS. Missing data were minimal, and selective reporting was not evident. Overall, although the intervention was clearly described and the follow-up was complete, the non-comparative design and absence of randomization substantially limit the internal validity of the findings.
Plasma Filtration
Cap et al 65 conducted a randomized controlled trial to evaluate the efficacy and safety of plasma filtration as an adjunct to IVGC in patients with severe, active TED (CAS >3). Twenty patients received IV methylprednisolone pulses and were randomly assigned to receive either adjunctive plasma filtration (n = 10) or IVGC alone (n = 10). Plasma filtration was administered in 10 sessions over 10 weeks. The primary outcome was change in CAS; secondary outcomes included visual evoked potentials (VEP), proptosis, extraocular muscle width, and immunologic markers.
Patients in the plasma filtration group showed a more rapid reduction in disease activity, with median CAS decreasing from 3.6 to 0.6 at 1 month, compared to a decrease from 4.1 to 2.0 in the control group (P = .027). This group also experienced earlier improvements in VEP amplitude, while changes in proptosis and extraocular muscle width did not differ between groups. Immunological assessment revealed significant reductions in immunoglobulins (IgA, IgM, IgG) and autoantibodies (anti-TPO, anti-TG, TRAb) and a transient decline in inflammatory markers such as soluble CD30 and MCP-1. Plasma filtration was well tolerated, with only minor adverse events reported.
The study was rated as having “some concerns” overall. Although random allocation was applied and group characteristics were broadly balanced, the randomization process lacked detailed reporting on sequence generation and allocation concealment. Outcome adjudicators were blinded for objective measures (eg, CT, VEP). However, the absence of blinding for participants and treating physicians introduces potential performance bias.
Although the therapies summarized in this section represent a heterogeneous group of interventions with variable mechanisms and targets, several have demonstrated promising results in reducing inflammation and improving specific clinical outcomes in patients with active moderate-to-severe TED. Plasma filtration, supported by a well-conducted RCT, showed favorable effects on clinical outcomes and inflammatory markers. Somatostatin analogs and colchicine also yielded clinically relevant improvements in CAS in smaller trials or observational settings. While effective for eyelid retraction, Botulinum toxin has limited utility in strabismus management. Nevertheless, the overall strength of evidence remains low to moderate, with most studies constrained by small sample sizes, limited follow-up, and methodological concerns. Observational studies, particularly those lacking comparators or blinded assessments, cannot establish definitive efficacy. While some therapies may be valuable in select clinical scenarios, such as patients with contraindications to corticosteroids or localized orbital involvement, further high-quality RCTs are needed to confirm their role in routine clinical practice.
Comparative Effectiveness of Treatment Modalities
Recent systematic reviews and network meta-analyses have comprehensively and comparably evaluated available treatments’ relative efficacy and safety across different disease phases, from active inflammation to chronic sequelae.
Alves et al 66 conducted a broad systematic review and meta-analysis including 36 comparative studies (RCTs and non-RCTs) with 2276 participants. Interventions were evaluated separately for the active and chronic phases of TED. In the active phase, IVGCs were superior to OGCs for reducing CAS, while biologics such as tocilizumab and teprotumumab demonstrated the most pronounced improvements in inflammatory activity and proptosis, respectively. In the chronic phase, surgical decompression emerged as the most effective intervention for diplopia and proptosis. Across both phases, teprotumumab yielded the highest overall ophthalmic improvement. This review fulfilled all AMSTAR 2 critical domains and was rated highly methodological (see Supplemental Tables S14 and S15).
Li et al. 67 Conducted a frequentist network meta-analysis of 18 RCTs (n = 1208) comparing 8 immunosuppressive and biologic interventions for moderate-to-severe active TED. Teprotumumab was ranked highest for reducing proptosis and improving overall response. Mycophenolate, both as monotherapy and in combination with IVGCs, also showed superior efficacy compared to IVGCs alone. In contrast, oral glucocorticoids and placebo were consistently the least effective. The authors applied a structured risk-of-bias assessment and a GRADE-like framework, achieving a high AMSTAR 2 rating for methodological rigor.
Zhou et al 68 performed a comprehensive network meta-analysis including 45 RCTs and 1846 patients with active moderate-to-severe TED, comparing 21 treatment strategies. While IVGC + orbital radiotherapy (ORT), IVGC alone, and mycophenolate mofetil were ranked among the most effective for achieving overall clinical response (⩾2-point CAS reduction or CAS <3), biologics such as rituximab and infliximab showed more variable efficacy. Somatostatin analogs and IVGC demonstrated the most favorable tolerability profiles. Despite the breadth of comparisons, the study was limited by low-to-very-low certainty of evidence due to heterogeneity, indirectness, and lack of GRADE evaluation. AMSTAR 2 assessment rated the methodological quality as moderate (see Supplemental Tables S14 and S15).
Stiebel et al 69 synthesized data from 29 studies, including RCTs and observational designs, to compare the efficacy of glucocorticoids, radiotherapy, and surgical decompression. Their findings reinforced the superiority of IVGCs over OGCs in reducing disease activity and improving composite response. Radiotherapy had comparable efficacy to IVGCs for ocular motility, while surgical decompression was more effective for proptosis but associated with higher complication rates, including new-onset diplopia. Methodological limitations included the absence of protocol registration, inconsistent bias assessment, and lack of GRADE use, resulting in a moderate AMSTAR 2 rating.
In summary, high-quality comparative reviews and network meta-analyses confirm IVGCs as the first-line treatment for active moderate-to-severe TED, with emerging evidence supporting biologics, particularly teprotumumab and tocilizumab, for treatment-refractory or severe cases. Mycophenolate represents a promising immunosuppressive alternative with favorable safety.
Discussion
This systematic review offers a comprehensive and structured synthesis of the current evidence base on medical therapies for moderate-to-severe TED. Integrating data from randomized controlled trials, systematic reviews, meta-analyses, and observational studies provides a broad yet critical evaluation of pharmacological interventions targeting inflammatory activity, proptosis, and diplopia in TED.
The clinical management of TED remains challenging due to its heterogeneity in presentation, unpredictable disease course, and variable response to treatment. Although glucocorticoids continue to be recommended as first-line therapy,4,70 a growing body of evidence supports using alternative immunomodulatory agents, including biologics and immunosuppressants, particularly in steroid-refractory or intolerant cases. This review aimed to systematically assess and compare these interventions’ therapeutic efficacy and safety using a unified framework, facilitating interpretation across different study designs.
To facilitate clinical interpretation across pharmacologic strategies, this review employed a semi-quantitative synthesis framework that rated each intervention in terms of its effect on proptosis, Clinical Activity Score (CAS), and diplopia, as well as safety and tolerability. Unlike formal network meta-analyses, which offer greater statistical rigor in estimating relative treatment effects, our approach was designed to inform applied clinical decision-making by integrating data from randomized, observational, and real-world sources into a coherent comparative matrix. When contrasted with the methodologically robust network meta-analysis by Zhou et al, 68 which ranked combinations such as IVGC plus orbital radiotherapy, MMF, and teprotumumab among the most effective interventions, our synthesis is broadly concordant. Both syntheses support the central role of IVGC and recognize the emerging relevance of biologics and immunosuppressive agents. However, our review emphasizes practical therapeutic hierarchies based on outcome consistency, tolerability, and evidence breadth, thereby complementing the statistical ranking with an application-focused perspective more suitable for individualized treatment planning in moderate-to-severe TED.
Glucocorticoids remain the cornerstone of treatment for moderate-to-severe active TED, with IVGCs constituting the recommended first-line approach in both the 2021 European Group on Graves’ Orbitopathy (EUGOGO) guidelines 4 and the 2022 joint American Thyroid Association–European Thyroid Association (ATA/ETA) consensus statement. 70 These recommendations are grounded in randomized controlled trials demonstrating superior efficacy and a more favorable safety profile of IVGCs than oral glucocorticoids. Evidence from direct comparisons shows that IVGCs induce greater and more rapid disease activity control, with fewer cumulative adverse events, particularly hepatotoxicity and metabolic complications. This is further supported by network meta-analysis results, 68 where IVGCs demonstrated significantly greater odds of response versus OGCs (OR 4.96; 95%CI 1.96-12.55). Nevertheless, the strength of this evidence is tempered by heterogeneity in dosing protocols (eg, cumulative dose, pulse frequency), variability in response definitions, and limited long-term follow-up. Our review noted that most IVGC regimens followed the EUGOGO-proposed cumulative dose of 4.5 g methylprednisolone, but several observational series reported modified regimens with variable outcomes.
While less effective overall, oral glucocorticoids remain in use due to their accessibility, lower cost, and ease of administration. Their utility may be considered in cases where IV administration is not feasible. However, the increased frequency of systemic adverse effects and the requirement for longer treatment durations reduce their appeal. Our synthesis rated OGCs lower than IVGCs across all domains, particularly in proptosis and overall disease response, aligning with findings from prior systematic reviews. Future studies should aim to standardize IVGC protocols, investigate dose-response relationships, and evaluate predictors of treatment resistance to refine their role in stratified treatment algorithms.
Among the biologic agents evaluated for moderate-to-severe TED, teprotumumab, rituximab (RTX), and tocilizumab (TCZ) are the most studied immunomodulatory options for patients unresponsive or intolerant to glucocorticoids. In our synthesis, teprotumumab consistently ranked highest across clinical outcomes—proptosis reduction, CAS improvement, and diplopia response—corroborating results from pivotal trials and meta-analysis, 30 which reported significant superiority over placebo in CAS (SMD −1.46), proptosis (SMD −12.52), and diplopia (OR 4.31). These findings support its designation as a preferred second-line option in EUGOGO and ATA/ETA consensus guidelines.4,70 However, concerns around high cost, limited access, and adverse events, particularly hearing loss, limit its widespread applicability.
Rituximab, a CD20-directed monoclonal antibody, has shown more heterogeneous results. While open-label studies and meta-analyses 31 suggested benefit in CAS reduction and disease inactivation, results from 2 RCTs were discordant. One trial demonstrated non-inferiority to IVGCs with lower relapse rates, whereas another failed to show significant superiority over placebo.32,33 Differences in patient selection, disease duration, and outcome definitions may partly explain this variability. Our review categorized rituximab as having moderate efficacy and tolerability, with higher uncertainty around proptosis outcomes and a need for further head-to-head trials. The 2022 ATA/ETA consensus cautiously supports rituximab for steroid-resistant or relapsing TED, especially in early inflammatory phases.
Tocilizumab, an IL-6 receptor antagonist, has demonstrated promising efficacy in patients with glucocorticoid-resistant TED. Evidence from RCTs and observational cohorts consistently shows significant reductions in CAS (⩾3 points in most patients), improvement in proptosis, and low relapse rates. Tocilizumab was rated favorably in our synthesis across efficacy domains and exhibited a good safety profile in published data. Although formal guideline endorsement remains limited due to the paucity of large RCTs, EUGOGO and ATA/ETA documents acknowledge its potential role, especially in patients with partial response or relapse after IVGCs.
In summary, biologic therapies represent a significant advancement in TED management, particularly for patients without adequate glucocorticoid control. While teprotumumab offers the most robust evidence, access barriers and safety considerations limit its universal applicability. Rituximab and tocilizumab provide additional options supported by growing, but heterogeneous, evidence bases. Rituximab demonstrated more variable efficacy. While several trials and observational studies have reported improvements in CAS (SMD −1.51, the effects on proptosis and diplopia remain inconsistent. 30 The meta-analysis by Fatani et al indicated a significant reduction in but no meaningful impact on proptosis or diplopia, reflecting heterogeneity in study populations and trial design. In our review, rituximab was rated moderately effective with acceptable tolerability, but insufficient consistency to support broad use. Current guidelines recommend it selectively, particularly in early active disease or glucocorticoid-resistant patients.
Beyond glucocorticoids and biologics, several non-biologic immunosuppressants have been explored in TED, often as glucocorticoid-sparing agents or in refractory disease. Mycophenolate mofetil (MMF) has emerged as one of the most promising in this group, with consistent efficacy in CAS improvement and disease inactivation across RCTs and meta-analyses. In our synthesis, MMF ranked among the highest in tolerability and sustained clinical response, aligning with its favorable positioning in recent network meta-analyses. Methotrexate and cyclosporine showed modest benefit, primarily in inflammatory control, while azathioprine and cyclophosphamide remain less frequently studied and are limited by safety concerns. MMF offers the most compelling benefit-risk profile among traditional immunosuppressants, though access and long-term data remain scarce.
Adjunctive therapies such as somatostatin analogs, intravenous immunoglobulin, colchicine, plasma filtration, and botulinum toxin have shown isolated signals of benefit in selected outcomes or case series, but lack consistent supporting evidence. Their use remains restricted to specific clinical scenarios, including steroid-intolerant patients or localized symptomatic relief (eg, botulinum toxin for diplopia). While some interventions showed favorable safety profiles, none demonstrated superiority over established therapies in comparative trials.
Safety and tolerability profiles varied considerably across interventions, reflecting differences in pharmacologic class, route of administration, and study design. Intravenous glucocorticoids, despite their established efficacy, are associated with dose-dependent risks of hepatotoxicity, cardiovascular events, and metabolic disturbances, particularly at cumulative doses exceeding 8 g. Oral glucocorticoids were linked to a higher incidence of systemic adverse effects, including weight gain, glucose intolerance, and mood alterations, mainly due to prolonged exposure and slower tapering regimens.
Biologic therapies presented a mixed safety profile. Teprotumumab, while generally well tolerated, was consistently associated with hearing-related adverse events and hyperglycemia, warranting pre-treatment audiologic and metabolic screening. In most reports, Rituximab and tocilizumab showed acceptable tolerability, although infusion reactions and infections remain relevant concerns. Among non-biologic immunosuppressants, mycophenolate mofetil demonstrated a favorable safety profile, while agents like cyclophosphamide and azathioprine carry well-documented hematologic and gastrointestinal toxicity risks. Reporting adverse events was inconsistent across studies, and few trials employed standardized grading systems, limiting cross-comparison. These findings underscore the need to balance therapeutic efficacy with individual risk profiles, especially in long-term disease management.
Since completion of the literature search, additional randomized controlled trials evaluating IGF-1R inhibition in active thyroid eye disease have been published and merit consideration. The RESTORE-1 phase 3 trial evaluated IBI311, an IGF-1R inhibitor with an identical amino acid sequence to teprotumumab but a different formulation, in Chinese patients with active moderate-to-severe TED. 71 In this double-masked, placebo-controlled study, IBI311 demonstrated a markedly higher proptosis response rate at 24 weeks compared with placebo (85.8% vs 3.8%), alongside significant improvements in CAS and overall response, with no new safety signals. Similarly, a randomized, double-masked, placebo-controlled trial conducted in Japan confirmed the efficacy of teprotumumab in an Asian population, reporting a proptosis response rate of 89% versus 11% with placebo, as well as meaningful improvements in diplopia and disease activity. 72 Importantly, both trials reinforce the consistency of IGF-1R–targeted therapy across diverse ethnic populations and healthcare settings, strengthening the external validity of earlier pivotal trials and supporting the generalizability of IGF-1R inhibition as a highly effective therapeutic strategy in active TED.
This review has several limitations inherent to the current evidence on medical treatments for moderate-to-severe TED. First, studies have substantial heterogeneity regarding patient populations, disease severity, treatment protocols, outcome definitions, and follow-up duration. Most trials define clinical activity and response using different criteria, limiting comparability across interventions. Furthermore, many studies, particularly those evaluating non-biologic immunosuppressants and adjunctive therapies, are small, single-center, or observational, introducing the risk of bias and confounding. The scarcity of head-to-head randomized controlled trials also hampers direct comparisons between therapeutic classes, especially for newer biologics.
In addition, this review has limitations related to its design. Although we employed a comprehensive and systematic search strategy, including diverse study designs and semi-quantitative synthesis, it introduces subjectivity in evidence grading. While this approach enhances clinical applicability, it lacks the statistical precision of network meta-analysis models. Some data were derived from open-label or retrospective studies with limited control for bias, and adverse event reporting was inconsistent across sources. Finally, while our framework aimed to reflect real-world decision-making, it does not account for cost-effectiveness, access barriers, or patient preference, critical factors in clinical implementation. These limitations highlight the need for standardized outcome reporting and large-scale comparative trials to optimize treatment selection in TED.
Conclusion
This systematic review provides an integrated evaluation of the current evidence on medical treatments for moderate-to-severe, active thyroid eye disease. Intravenous glucocorticoids remain the most effective and well-supported first-line therapy, with consistent benefits in disease activity control and a favorable safety profile compared to oral regimens. Biologic agents, particularly teprotumumab, and non-biologic immunosuppressants such as mycophenolate mofetil, offer valuable alternatives for patients with glucocorticoid-refractory disease or contraindications. However, variability in study design, outcome reporting, and long-term follow-up limits the ability to establish definitive therapeutic hierarchies.
This synthesis reinforces the importance of an individualized treatment strategy for moderate-to-severe TED, taking into account, therapeutic response, and patient-specific safety considerations. The variability and limitations of the current evidence base underscore the need for standardized clinical endpoints and direct comparative trials to improve consistency and applicability. As new therapies continue to emerge, methodologically rigorous and clinically grounded evidence syntheses such as this one are essential to support nuanced decision-making in a condition marked by heterogeneity and substantial impact on patients’ quality of life.
Supplemental Material
sj-docx-1-end-10.1177_11795514261426446 – Supplemental material for Pharmacologic Therapies for Active Moderate-to-Severe TED: A Comprehensive Systematic Review
Supplemental material, sj-docx-1-end-10.1177_11795514261426446 for Pharmacologic Therapies for Active Moderate-to-Severe TED: A Comprehensive Systematic Review by Carlos E. Builes-Montaño, Alejandro Román-González, Henry M. Arenas-Quintero, Natalia Aristizábal-Henao, María Del S. Cabarcas-Solano, Jennifer Camargo González, Alejandro A. Castellanos-Pinedo, Marta L. Muñoz-Cardona, Katherine Restrepo-Erazo, Hernando Vargas-Uricoechea and María G. Mejía-López in Clinical Medicine Insights: Endocrinology and Diabetes
Footnotes
Acknowledgements
Festina Lente provided methodological support and medical writing and editorial assistance. The authors retained full independence in data interpretation, manuscript drafting, and decision to submit for publication, and assume full responsibility for the content and conclusions of this manuscript.
Ethical Considerations
Not applicable. This study is a systematic review of published literature and did not involve direct participation of human subjects, patients, or animals. Therefore, institutional review board approval was not required.
Consent to Participate
This study is a systematic review of published literature and did not involve direct participation of human subjects. Therefore, informed consent was not required.
Consent for Publication
Not applicable. No individual patient data, images, or identifiable information are included in this manuscript.
Author Contributions
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Festina Lente support was funded by the Asociación Colombiana de Endocrinología, Diabetes y Metabolismo. This work was funded by the Asociación Colombiana de Endocrinología, Diabetes y Metabolismo through an unrestricted educational grant.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The authors declare the following conflicts of interest: Natalia Aristizábal-Henao has received academic consulting fees from Amgen. Katherine Restrepo-Erazo has participated in conferences and advisory boards for AstraZeneca, Boehringer Ingelheim, Novo Nordisk, Merck, Sanofi, Recordati, Eli Lilly, Bayer, and Abbott. Hernando Vargas-Uricoechea has received speaking honoraria from Sanofi, Abbott, and Boehringer Ingelheim. Carlos E. Builes-Montaño has received consulting and speaking fees from Sanofi, Novo Nordisk, Novartis, Recordati Rare Diseases, Valentech, Janssen, Abbott, and Boehringer Ingelheim and is a shareholder of Festina Lente. Marta Lucía Muñoz-Cardona has received travel support from Amgen. María del Socorro Cabarcas-Solano, María G. Mejía-López, Alejandro Román-González, Jennifer Camargo González, and Alejandro Castellanos-Pinedo report no conflicts of interest.
Data Availability Statement
All data supporting the findings of this study are derived from publicly available published literature. No new datasets were generated.
Use of Artificial Intelligence
Artificial intelligence tools were employed to support specific phases of the manuscript development. The screening process used Rayyan, a web-based platform integrating AI-assisted functionalities to facilitate duplicate removal and blinded study selection. Grammarly was used to assist with grammar and style revision during the final editing phase. No generative AI tools were used to produce original scientific content, interpret results, or replace the authors’ critical analysis and writing responsibilities.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
