Diabetes Distilled: Tirzepatide treatment associated with increased proliferative retinopathy but possible reduced new retinopathy

It has been known since 1998 that intensive glucose lowering reduces the risk of diabetic retinopathy (UK Prospective Diabetes Study Group, 1998). However, rapid reductions in glycaemia, which may occur with insulin therapy, injectable incretin drugs for type 2 diabetes or after bariatric surgery, may cause early worsening of diabetic retinopathy (EWDR). In the SUSTAIN-6 trial of semaglutide for type 2 diabetes, EWDR occurred with significant glucose reductions, particularly in participants who were also receiving insulin or who had more than background retinopathy at baseline (Marso et al, 2016).

In a previous meta-analysis, risk factors for EWDR included higher baseline HbA_1c, greater HbA_1c reduction, longer duration of diabetes and greater than background retinopathy at baseline (Feldman-Billard et al, 2018). The authors recommended eye monitoring every 3 months for those at high risk of EWDR during intensive glucose-lowering, although this is not yet included in guidelines.

In the case of tirzepatide, a recent meta-analysis of the SURPASS clinical trial programme did not identify any increased risk of EWDR (Popovic et al, 2024); however, no additional retinal surveillance was undertaken during the studies and people with proliferative retinopathy, severe pre-proliferative retinopathy or maculopathy were excluded.

Since tirzepatide is known to have potent glucose-lowering effects, the present study, published in Diabetologia, sought to clarify the impact of the drug on EWDR across all retinopathy groups.

Study design

In this real-world, retrospective cohort study, Buckley and colleagues compared 3455 people who had received tirzepatide for at least 180 days with 3454 people who had not been exposed to tirzepatide, matched for sex, retinopathy status, retinal screening history, diabetes duration, HbA_1c and glucose-lowering treatment. All study participants already had tight glycaemic control, with a mean HbA_1c of 56.1 mmol/mol (7.3%) across the groups at baseline.

Results

Retinal outcomes are summarised in Table 1. New-onset proliferative diabetic retinopathy (PDR), with or without maculopathy, was more common with tirzepatide, occurring in 33 tirzepatide recipients and in 17 of those without exposure (1.1% vs 0.5%). After adjustment for established risk factors, this gave an odds ratio (OR) of 2.15. Most of the new PDR occurred in those with moderate-to-severe (R2M1) or mild (R1M1) non-proliferative retinopathy with maculopathy.

In people without retinopathy at baseline, tirzepatide treatment was associated with significantly reduced odds of developing new retinopathy (OR 0.73).

There was no significant association between tirzepatide and retinopathy progression in those who had mild, non-proliferative diabetic retinopathy (R1MO or R1M1) at baseline.

In those without maculopathy at baseline, tirzepatide treatment demonstrated no increased association with developing maculopathy.

Discussion

The mean interval to diagnosis of PDR was 325±143 days in those treated with tirzepatide, meaning that worsening can occur later in treatment, even on stable doses.

The SURPASS clinical trial programme identified a low rate of PDR: only 21 adverse retinal events in those treated with tirzepatide compared with 19 in controls, thus implying no increased risk with tirzepatide. In this real-world study, however, there was an increased incidence of PDR in those exposed to tirzepatide. New PDR occurred in 17 out of 113 people (15%) treated with tirzepatide who had R2M0 or R2M1 at baseline, and there was a higher risk of PDR in those with maculopathy at baseline. Both of these are groups who should already be receiving specialist ophthalmology follow-up in the UK.

In a post hoc analysis of the SUSTAIN-6 trial of semaglutide, the degree of glycaemic reduction at 16 weeks correlated with the risk of new retinopathy (Vilsbøll et al, 2018), whereas in this real-world study, HbA_1c reduction was not independently associated with increased odds of PDR, possibly due to the already low HbA_1c at baseline.

Study strengths and limitations

Limitations include those of all real-world cohort studies, including possible additional confounders. The tirzepatide cohort had mostly transitioned from a GLP-1 receptor agonist to tirzepatide and, therefore, already had a mean HbA_1c of 55.2 mmol/mol (lower than would be usual for new initiations of an injectable incretin drug, including in the SURPASS clinical trials), and this also resulted in low initial HbA_1c reductions. Thus, these results may underestimate changes associated with tirzepatide in current clinical practice.

Strengths include the large size of the cohort and good matching with controls. There was consistent annual retinopathy examination, rather than just the retinal adverse event reporting that occurred in many of the clinical trials, and this allowed optimised diagnosis of EWDR.

The authors recommend that in people with maculopathy and/or moderate-to-severe diabetic retinopathy, discussion with their ophthalmologist should be considered prior to initiating tirzepatide. However, in people with background retinopathy only, tirzepatide was not associated with significant worsening of retinopathy in this study, so they do not believe any specific action is required.

For those with no retinopathy, tirzepatide treatment appeared to be associated with a reduction in risk of new retinopathy, and further studies will be needed to confirm this potential added benefit of tirzepatide, presumably linked to effective glycaemic control.

Implications for practice

This paper provides important guidance to inform safe initiation of tirzepatide in people with type 2 diabetes who already have more than background retinopathy. Although tirzepatide appears to protect against the development of new diabetic retinopathy and was not associated with increased risk of maculopathy compared to those not receiving tirzepatide, in people with existing moderate-to-severe retinopathy, treatment was associated with a significantly increased risk of new PDR. In this study, the risk appeared to be independent of rapid HbA_1c reduction, a mechanism previously confirmed to increase PDR risk, suggesting that caution is required irrespective of HbA_1c when initiation is being considered.

Advice issued following analysis of SUSTAIN-6 recommended that, if HbA_1c was >91 mmol/mol (10.5%) in those with more than background retinopathy, glycaemia should be reduced slowly with other glucose-lowering agents before starting semaglutide (either injectable or oral, since these have comparable efficacy) (Bain, 2021). Since its launch, tirzepatide guidance has been similar, although without firm supporting evidence.

These new findings, particularly since they occurred even in people with well-controlled HbA_1c at baseline, are a reminder that, in those known to have more than background diabetic retinopathy, we should discuss use of tirzepatide with their ophthalmologist prior to initiation.

Worryingly, third parties are prescribing tirzepatide for weight loss in people with diabetes, some of whom may have significant retinopathy and/or maculopathy, without access to the person’s full medical record. It is important to code retinopathy results accurately, and to warn those who may be at risk not to buy tirzepatide privately for weight loss without discussion with their eye specialist.

With the UK retinopathy screening interval increased to 2 years for those with no retinopathy on two annual screens, concerns have been raised about the risk of delay in diagnosis of sight-threatening retinopathy in younger adults, and in Black and South Asian groups labelled as low-risk (Olvera-Barrios et al, 2023). It will be important to understand more about the risks of developing maculopathy and moderate-to-severe retinopathy in the interval between screenings, particularly in people with a persistently high HbA_1c, and how this should influence our assessment of PDR risk in those without a recent retinal screen who initiate tirzepatide.

Further data on the risks of diabetic retinopathy associated with incretin therapies will be published from the FOCUS study (semaglutide) and the ANGIOSAFE2 study (semaglutide and tirzepatide).

Considering the possible risks of EWDR, it is crucial that we review retinal screening reports prior to prescribing. Furthermore, in view of the ongoing uncertainty around whether incretin drugs increase the risk of non-arteritis anterior ischaemic optic neuropathy (NAION) (Etminanet al, 2025), we need to warn people to seek urgent medical advice if they suffer eye symptoms during treatment with these drugs. Such a warning may improve understanding of the importance of retinal screening.

We need to help people protect their vision whilst also supporting appropriate access to tirzepatide and its potential benefits. This paper helps us to achieve that.