Review Anti Diabetic Drugs

Graefe's Archive for Clinical and Experimental Ophthalmology
https://doi.org/10.1007/s00417-023-06236-5
REVIEW ARTICLE
Effects of newer‑generation anti‑diabetics on diabetic retinopathy:

a critical review
Dimitrios P. Ntentakis1 · Victor San Martin Carvalho Correa1 · Anastasia Maria Ntentaki1 · Eleni Delavogia2 ·
Toshio Narimatsu1 · Nikolaos E. Efstathiou1 · Demetrios G. Vavvas1
Received: 20 June 2023 / Revised: 30 August 2023 / Accepted: 7 September 2023

© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2023
Abstract
Diabetic retinopathy (DR) is the leading etiology of blindness in the working population of the USA. Its long-term manage-
ment relies on effective glycemic control. Seven anti-diabetic classes have been introduced for patients with type 2 diabetes
(T2D) in the past two decades, with different glucose-lowering and cardiovascular benefits. Yet, their effects specifically on
DR have not been studied in detail. A systematic review of the literature was conducted to investigate this topic, focusing
on the available clinical data for T2D. Published studies were evaluated based on their level of statistical evidence, as long
as they incorporated at least one endpoint or adverse event pertaining to retinal health. Fifty nine articles met our inclu-
sion criteria and were grouped per anti-diabetic class as follows: alpha-glucosidase inhibitors (1), peroxisome proliferator-
activated receptor gamma (PPAR-γ) agonists (8), amylin analogs (1), glucagon-like peptide-1 (GLP-1) receptor agonists
(28), dipeptidyl peptidase 4 (DPP-4) inhibitors (9), and sodium glucose co-transporter-2 (SGLT-2) inhibitors (9), plus one
retrospective study and two meta-analyses evaluating more than one of the aforementioned anti-diabetic categories. We also
reviewed publicly-announced results of trials for the recently-introduced class of twincretins. The available data indicates
that most drugs in the newer anti-diabetic classes are neutral to DR progression; however, there are subclasses differences in
specific drugs and T2D populations. In particular, there is evidence suggesting there may be worse diabetic macular edema
with PPAR-gamma agonists, potential slight DR worsening with semaglutide (GLP-1 receptor agonist), and potential slight
increase in the incidence of retinal vein occlusion in elderly and patients with advanced kidney disease receiving SGLT-2
inhibitors. All these warrant further investigation. Longer follow-up and systematic assessment of at least one DR-related
endpoint are highly recommended for all future trials in the T2D field, to ultimately address this topic.
Keywords Newer-generation anti-diabetics · Diabetic retinopathy · Diabetic retinopathy complications · Type 2 diabetes
Dimitrios P. Ntentakis and Victor San Martin Carvalho Correa

contributed equally to this work.
* Demetrios G. Vavvas
[email protected]
1
Ines and Fredrick Yeatts Retina Research Laboratory,
Angiogenesis Laboratory, Retina Service, Department
of Ophthalmology, Massachusetts Eye and Ear Main
Campus, 243 Charles Street, Harvard Medical School,
Boston, MA 02114, USA
2
Department of Pediatrics, Harvard Medical School, Boston,
MA, USA
13
Vol.:(0123456789)
Key messages
What is known:
The DCCT and UKPDS trials demonstrated that intensive glycemic control with either insulin or sulfonylureas is
associated with a significant reduction in diabetic retinopathy (DR) progression, over long-term clinical monitoring.
Seven new anti-diabetic classes with documented glycemic and cardiovascular benefits have been introduced for
type 2 diabetes (T2D) within the past two decades, however their particular effects on DR remain poorly understood.
What is new:
The available data indicates that most drugs in these new anti-diabetic classes are neutral to DR progression,
however there are subclasses differences in specific drugs and T2D populations.
Randomized prospective data are currently scant on dedicated DR endpoints, and in the cases of GLP-1 receptor
agonists and DPP-4 inhibitors, the evidence from systematic DR assessment is almost contradictory to the findings
from adverse event reporting, retrospective studies, and meta-analyses.
There is evidence suggesting potentially worse diabetic macular edema with PPAR-gamma agonists, potential slight
DR worsening with semaglutide (GLP-1 receptor agonist), and potential slight increase in the incidence of retinal
vein occlusion in elderly and patients with advanced kidney disease receiving SGLT-2 inhibitors; all of which
warrant further investigation.
Introduction demonstrated by the ACCORD study, in 2856 T2D partici-

pants monitored over 4 years [5].
Diabetes mellitus, and type 2 diabetes (T2D) in particular,
Despite the long-term benefit of reduced HbA1c, a rapid
comprise an escalating epidemiological issue. Global preva-
decrease in HbA1c has been correlated with transient DR
lence is estimated to exceed 642 million cases between 20
worsening. In 1998, the DCCT study evaluated patients
and 79 years of age by 2040 [1], raising healthcare aware-
with type 1 diabetes and mild/moderate DR at baseline,
ness towards its life-threatening cardiovascular and micro-
receiving either intensive or conventional insulin treatment
vascular complications; the latter mainly refer to periph-
[6]. Early worsening of DR, defined by progression of 3
eral neuropathy, diabetic nephropathy (DN), and diabetic
steps or more on the Early Treatment Diabetic Retinopathy
retinopathy (DR). In the aforementioned age group, DR is
Study (ETDRS) chart, new exudates, intra-retinal micro-
the leading etiology of blindness in the USA [2]. Effective
vascular abnormalities, or any clinically-relevant evidence
glycemic control entails the cornerstone in the management
of DR, was significantly more frequent in the intensively-
of diabetic complications, and DR has been no exception.
treated group over the first 12 months. In the same group,
In the UKPDS 33, a prospective study of 3867 new T2D
DR progression at 18 months was found significantly more
patients, intense glycemic control with either insulin or
advanced compared to the conventionally treated controls.
sulfonylureas (SUs) achieved an 11% decrease in glycated
Predictors of early worsening were the magnitude of HbA1c
hemoglobin (HbA1c) over 10 years of follow-up [3]. This, in
drop from baseline, as well as DR severity at baseline [6].
turn, was associated with a 25% reduction in microvascular
Yet, after 25 years of follow-up over the same cohort, inten-
complications [3]. In the UKPDS 35 that followed, for each
sive glycemic control appears to have an overall protective
1% reduction in mean HbA1c, a 37% decrease was observed
effect on DR progression [7–9], as expected.
in the microvascular complications risk, defined as reduced
Following the hallmark findings of the UKPDS and
need for photocoagulation [4]. An analogous relationship
DCCT trials, seven new anti-diabetic classes have been
between intensive glycemic control and DR progression was
introduced within the past two decades. The majority of
13
them are orally administered and have a documented ben- Study design (PICOS) framework [14]. We included patient
efit in terms of glycemic control, cardiovascular morbidity, data from T2D studies with at least one outcome pertain-
and overall mortality in T2D patients [10]. Based on the ing to retinal complications or the retina in general. Animal
guidelines of the American Association of Diabetes, some data, letters to the editor, congress abstracts, publications
of these medications are currently prescribed as second- or in languages other than English, case reports, and reviews/
even first-line agents in T2D patients [10]. However, not systematic reviews without meta-analysis were excluded. To
much is known about their effects on retinal microvascular enhance the clinical relevance of our work, studies with a
complications at this point. sample size not exceeding 100 patients were omitted. Simi-
Hence, this article presents a critical and systematic larly, we excluded trials with follow-up duration of 6 months
analysis of the effects of newer anti-diabetics on the retina. or shorter; such a timeframe was deemed insufficient to cap-
Unlike previous approaches to this topic [11, 12], the clini- ture the insidious progression of T2D itself, as well as the
cal relevance of all evidence-based data is co-evaluated with onset of DR complications [15–17]. Collectively, 59 arti-
their corresponding strength of statistical evidence in our cles met our inclusion criteria, as illustrated on the PRISMA
work. We emphasize on the randomized controlled trials flow diagram (Fig. 1). We also included two Phase 3 trials
(RCTs) that included at least one systematically investigated completed in April 2022 for the recently introduced class of
endpoint pertaining to DR. Collectively, we aim to pinpoint twincretins and four ongoing trials from the clinicaltrials.
the specific anti-diabetics, as well as any patient subgroups gov website.
at higher risk of retinal complications. We also discuss the
limitations of the available literature, as well as the existing Data extraction and organization
clinical trial guidelines, with the intent of helping future
research in the field. Articles were organized per anti-diabetic class and further
categorized based on their level of evidence (Table 1). For
the latter, we applied a slightly-modified version of the
Methods Oxford Centre for Evidence-Based Medicine guidelines[21],
adjusted to the purpose of our analysis. Instead of addressing
Literature search all RCT data on DR as equivalent, we considered the ones
deriving from systematically evaluated endpoints to have the
A PubMed literature search was conducted between October highest clinical significance. For every article, the following
20, 2022, and January 31, 2023. We started by combining methodological parameters were recorded: first author, year
each of the terms “anti-diabetics,” “hypoglycemics,” and of publication, sample size (n), T2D duration, diabetic eye
“anti-hyperglycemics” with the keywords referring to retinal disease (DED) at baseline, intervention being tested, con-
complications in most T2D studies: “retinopathy,” “retina,” trol interventions, follow-up duration/study duration, DED-
and “microvascular.” The same strategy was then repeated, related outcomes. We also studied data on cardiovascular
this time using the specific names of the anti-diabetic classes morbidity, mortality, and microvascular complications (DR,
introduced within the past two decades, based on the 2021 DN, peripheral neuropathy) per anti-diabetic class (organ-
“Standards of Medical Care in Diabetes” by the American ized in a qualitative manner, available upon request) to
Association of Diabetes [10]. Our search was further nar- enhance the interpretation of the DR outcomes we reviewed.
rowed to the particular anti-diabetics in each of those catego-
ries, using the same approach and guidelines [10]. We also
searched the clinicaltrials.gov website for relevant ongoing Results
studies, on January 28th, 2023; we used “diabetes” for con-
dition, and “diabetic retinopathy” as the outcome measure Publications meeting our inclusion criteria are organized in 6
in the corresponding search fields. Our literature search was anti-diabetic categories as follows (Table 1): alpha-glucosidase
independently reviewed by three authors (DPN, VSMCC, inhibitors (1), peroxisome proliferator-activated receptor gamma
AMN) and the principal investigator (DGV). (PPAR-γ) agonists (8), amylin analogs (1), glucagon-like pep-
tide-1 (GLP-1) receptor agonists (28), dipeptidyl peptidase 4
Data selection (DPP-4) inhibitors (9), and sodium glucose co-transporter-2
(SGLT-2) inhibitors (9). All outcomes referring to DR or other
A total of 432 articles were identified via the aforemen- pertinent retinal complications are presented per anti-diabetic
tioned process and then carefully reviewed. We followed the class, preceded by a brief paragraph with introductory informa-
guidelines by the Preferred Reporting Items for Systematic tion for each drug group. Two (2) meta-analyses and one retro-
Reviews and Meta-analyses (PRISMA) statement [13] and spective study evaluating more than one anti-diabetic catego-
the Population, Intervention, Comparison, Outcomes and ries are also discussed within the aforementioned format, when
13
Fig. 1 PRISMA flow diagram summarizing our literature search, screening, and review process (13)
relevant [18–20]. Key methodological parameters and retinal pharmacology. Also, the design and status of four (4) ongo-
outcomes from those publications are shown in Tables 2–5, per ing trials pertinent to our topic are reviewed separately in the
anti-diabetic class. “Ongoing trials” subsection. Figure 2 illustrates a qualitative
Our layout follows the chronological order of each class’s synopsis of the key findings from all studies we reviewed.
introduction to the market. The only exception is the novel Statistical comparisons were impeded by the insufficient
(seventh in our analysis) class of “twincretins,” whose number of retina-specific data and the methodological limi-
clinical testing and approval were completed in April 2022 tations described in the the “Discussion” section.
[23]; publicly announced results of two trials studying the
first member of this group, tirzepatide, are discussed with Alpha‑glucosidase inhibitors
GLP-1 receptor agonists in the “GLP-1 receptor agonists
(—glutides—enatides) and twincretins (tirzepatide)” subsec- Membrane bound on the intestinal brush border, alpha glu-
tion, due to their similar mechanism of action and overall cosidase coverts starch and disaccharide carbohydrates to
13
Table 1 Overview of the clinical studies that met our inclusion criteria, categorized per anti-diabetic class and study design
Study design Alpha-glucosidase inhibi- PPAR-γ agonists Amylin analogs GLP-1 receptor agonists DPP-4 inhibitors SGLT-2 inhibitors
tors
RCTs with systematic N/A N/A N/A 3 1 1

assessment of at least one Marso et al. (2016) (a); Rosenstock et al. (2019) Wanner et al. (2016)
retinal endpoint Marso et al. (2016) (b);
Gerstein et al. (2019)
RCTs reporting retinal/eye 1 N/A 1 5 1 N/A
AEs Segal et al. (2005) Ratner et al. (2002) Kaku et al. (2011); Rosen- Green et al. (2015)
stock et al. (2014); Pfeffer
et al. (2015); Seino et al.
(2016); Husain et al.
(2019)
Post-hoc analyses N/A 2 N/A 2 N/A 1
Ambrosius et al. (2010); Vilsbøll et al. (2018); Bethel Inzucchi et al. (2019)
Gower et al. (2018) et al. (2020)
Retrospective studies N/A 6 N/A 8 8 4
Shen et al. (2008); Fong Varadhan et al. (2011, Kolaczynski et al. (2016); Dziuba et al. (2014); Chung
et al. (2009); Shani et al. 2014); Douros et al. Kim et al. (2018); Wang et al. (2019); Lee et al.
(2011); Idris et al. (2012); (2018); Fadini et al. et al. (2018); Chung (2021); Su et al. (2021)
Motola et al. (2012); (2018); Wang et al. (2018, et al. (2019); Kang et al.
Vallabhajosyula et al. 2019); Ueda et al. (2019); (2021); Ueki et al. (2021);
(2016) Gaborit et al. (2020) Deng et al. (2022); Sim-
mons et al. (2022)
Meta-analyses N/A N/A N/A 13 1 5
Dicembrini et al. (2017); Tang et al. (2018) Tang et al. (2018); Tsapas
Andreadis et al. (2018); et al. (2020); Li et al.
Avgerinos et al. (2018, (2021); Ma et al. (2022);
2020); Tang et al. (2018); Zhou et al. (2022)
Zhang et al. (2018); Tsa-
pas et al. (2020); Bethel
et al. (2021); Alexander
et al. (2022); Wang et al.
(2022); Wei et al. (2022);
Yoshida et al. (2022);
Albert et al. (2023)
The numbers underline the scarcity in randomized prospective data evaluating the potential association between newer anti-diabetics and diabetic retinopathy. Only a third of the eligible RCTs
have investigated the aforementioned relationship via a systematic approach. Two of the meta-analyses included [18, 19], and one retrospective study [20] evaluated more than one classes of
newer anti-diabetics
Abbreviations: RCT, randomized controlled trial; N/A, not applicable/available; GLP-1RAs, glucagon-like peptide-1 receptor agonists; DPP-4Is, dipeptidyl peptidase 4 inhibitors; SGLT-2Is,
sodium glucose co-transporter-2 inhibitors; AE, adverse events
13
Table 2 Key methodological parameters and retinal outcomes in association with TZDs use
First author, year Sample size (n) T2D duration (years) Baseline DR/DED Anti-diabetic(s) tested Control intervention(s) Follow-up/duration DR/DED outcome rates
(years) (treatment group versus
13
control group)
Post-hoc analyses
Ambrosius et al. 3473 with fundus pho- 10.1 ± 7.1 42.2%; TZDs (rosiglitazone/ Anti-diabetics other 2.4 No association between
(2010) tographs; ACCORD 4.1% with baseline pioglitazone) ± insu- than TZDs or no TZD use and DME
Eye Sub-study DME; 0.0% with lin for >3 months treatment upon unadjusted and
participants history of laser or adjusted analyses;
vitrectomy for DR no significant benefit
in VA with TZD use
upon unadjusted and
adjusted analyses—>
in the ACCORD Eye
cohort; ↑ DME associ-
ated with baseline DR
severity (p < 0.001)
and age (p = 0.03)
Gower et al. (2018) 2856; ACCORD Eye 10.0 ± 7.1 52.0%; TZDs Anti-diabetics other 4.0 No association between
Sub-study partici- 7.8% with DME in at than TZDs or no TZD use and DME
pants least 1 eye; 0.0% with treatment incidence upon
history of PDR/laser/ analysis of any use and
vitrectomy for DR duration of use
No significant differ-
ences in DR incidence/
progression or ultimate
VA outcome with TZD
use
Retrospective studies
Shen et al. (2008) 282; treatment group: Treatment group: 15.4 Treatment group: Rosiglitazone Any anti-diabetic other 2.8 (mean) ↓ Progression to PDR
124; control group: ± 7.8 NPDR 112/219 eyes than TZDs over 3 years, for eyes
158 Control group: 16.9 Control group: NPDR with severe NPDR at
± 8.8 146/259 eyes baseline: 19.2% versus
47.4% (p = 0.045; log-
rank, p = 0.059)
↓ Eyes with VA loss
of 3 lines or more,
regardless of baseline
DR status (p = 0.03;
log-rank, p = 0.03); no
significant difference
in DME incidence or
treatment-requiring
DME
Table 2 (continued)
control group)
Fong et al. (2009) 172,006; 59,013 with N/A N/A Glitazones (98% piogl- Anti-diabetics other 4.0 ↑ DME risk in treatment
at least one eye exam; itazone) than TZDs (SUs met- group (OR, 2.6 [95%
17,078 on TZDs formin meglitinide CI, 2.4–3.0]); ↑ DME
acarbose) ± insulin risk persisted (OR,
1.6 [95% CI, 1.4–1.8])
after adjusting for age
and HbA1c levels and
after excluding patients
with no benefit, no eye
exam, or HgA1c < 7.0
Cumulative dosing did
not affect DME rate
over time
Shani et al. (2011) 6689; treatment group: Treatment group: 3.2 N/A Rosiglitazone (for > 1 Anti-diabetics other 3.6 (median) ↑ Rate of retinal inter-
1304; control group: ± 1.7 year, > 80% adher- than rosiglitazone vention (photoco-
5385 Control group: 3.2 ence) and insulin agulation/vitrectomy):

± 2.0 115/1304 (8.8%) ver-
sus 379/5385 (7.0%), p
= 0.027
↑ HR for ophthalmo-
logic visit/retinal
intervention: 1.33
[95% CI: 1.08–1.64],
p = 0.008; before and
after adjustment for
demographics, HbA1c,
and ophthalmologic
visits
13
Table 2 (continued)
13
control group)
Idris et al. (2012) 103,368; treatment N/A 0.0 TZDs (rosiglitazone/ Any anti-diabetic other 10.0 ↑ DME incidence after
group: 3227; control pioglitazone) ± insu- than TZDs ± insulin 1 year of TZD use:
group: 100,141 lin for >6 months 41/3227 (1.3%) versus
227/100,141 (0.2%);
OR, 5.7 [95% CI,
4.1–7.9], p < 0.001;
OR remained ↑ after
adjustment for con-
founders and selection
bias
↑ DME incidence after
10 years of TZD
use (HR, 5.2 [95%
CI, 4.3–6.3], p <
0.001); HR remained
↑ after adjustment for
confounders and selec-
tion bias; ↑ HR with
TZDs + insulin (HR,
3.0 [95% CI, 1.5–5.9])
after propensity score
adjustment
Motola et al. (2012) 301,950 FAERS N/A N/A TZDs (56% rosiglita- Anti-diabetics 4.0 (FAERS reports) ↑ DME risk with TZD
reports; treatment zone; 44% pioglita- other than TZDs use: OR, 3.88 [95% CI,
group: 49,589; con- zone) (biguanides, SUs, 2.79–5.39], p < 0.001
trol group: 252,361 exenatide, insulin) ↑ DME risk particularly
attributed to rosigli-
tazone use: OR, 5.55
[95% CI, 3.94–7.79], p
< 0.001
Vallabhajosyula 520; treatment group: Treatment group: 9.0 Treatment group: Pioglitazone (± SUs, Anti-diabetics other 5.0 ↓ DR with TZD use:
et al. (2016) 260; control group: ± 6.9 87/260 patients metformin, and/or than TZDs RR, 0.7 [95% CI,
260 Control group: 10.5 Control group: N/A insulin) 0.5–0.8], p = 0.0003
± 8.2 (unadjusted)
Published articles are grouped per study design and presented in order of decreasing level of evidence, as described in the “Data extraction and organization” subsection (data extraction and
organization). The outcomes in bold font were reported statistically significant (p < 0.05)
Abbreviations: DR, diabetic retinopathy; DED, diabetic eye disease; RCT, randomized controlled trial; N/A, not applicable/available; ACCORD, Action to Control Cardiovascular Risk in Diabe-
tes trial; TZD, thiazolidinediones; VA, visual acuity; DME, diabetic macular edema; OR, odds ratio; PDR, proliferative diabetic retinopathy; aOR, adjusted odds ratio; NPDR, non-proliferative
diabetic retinopathy; HR, hazard ratio; FAERS, FDA Adverse Event Reporting System; RR, relative risk
glucose. Alpha-glucosidase inhibitors act by delaying glu- examination at the end of the 4-year evaluation period; of
cose absorption, thus blunting postprandial hyperglycemia these, 17,078 were receiving TZDs, mainly pioglitazone
[24]. Two such agents, acarbose and miglitol, were approved [39]. DME development was significantly more likely
by the Food and Drug Administration (FDA) for both mono- in the TZD group, both before (OR, 2.6; 95% CI, 2.4 to
therapy and combination formulas in 1995 and 1996, respec- 3.0) and after (OR, 1.6; 95% CI, 1.4 to 1.8) adjustment for
tively. They have been particularly recommended for gly- potential confounders [39]. Similarly, the risk of new DME
cemic control in T2D patients post-bariatric surgery, yet, was found significantly higher at 1 year (OR, 2.3 [95%
their use has been limited by the high occurrence of gas- CI, 1.5–3.6]) and at 10 years (HR, 2.3 [95% CI, 1.7–3.0])
trointestinal adverse events (AEs), including flatulence and following TZD use in a UK-based T2D cohort [40]. This
diarrhea [25–27]. In populations with glucose intolerance, association appeared to be strengthened when TZDs were
acarbose has been associated with a significant reduction combined with insulin therapy (HR, 3.0 [95% CI, 1.5–5.9])
in the risk of new T2D by as much as 25% [26, 28]. RCT [40]. Analysis of 49,589 records from the FDA Adverse
data on the risk of macro- and microvascular outcomes with Event Reporting System associated with TZD use between
alpha-glucosidase inhibitors suggest only a slight benefit in 2005 and 2008 also indicated a significantly higher DME
cardiovascular events and diabetic nephropathy [26, 29]. risk (OR, 3.88 [95% CI, 2.79–5.39]), particularly attributed
To date, no large RCT in the field of alpha-glucosidase to rosiglitazone (OR, 5.55 [95% CI, 3.94–7.79]) [41].
inhibitors has systematically evaluated a DR-specific out- Yet, in a smaller retrospective review of 124 diabetics
come (Table 1). In a small, singe-center RCT of 139 T2D on rosiglitazone versus 158 matched controls not receiving
participants, acarbose addition was no different to placebo any TZD, no difference was detected in DME incidence over
in terms of kidney function or fundoscopic findings over a 2.8 years of mean follow-up [42]. Meanwhile, progression
period of 18 months [30]. However, no details were pro- from non-proliferative (NPDR) to proliferative DR (PDR)
vided regarding DR grading or visual acuity (VA) measure- was reported significantly lower in the rosiglitazone group
ments [30]. In this setting, any particular effects of alpha- [42]. An analogous single-center cohort of 520 patients from
glucosidase inhibitors on the retina remain unknown. Yet, an India, retrospectively followed for 5 years, also showed a
overall protective trend towards T2D complications, such as significantly lower DR risk with pioglitazone versus non-
DR, could be extrapolated from the documented benefits of TZD users (RR, 0.7 [95% CI, 0.5–0.8]) [43]. Contradictory
acarbose in the prevention of new T2D and cardiovascular were the findings from a larger community-based study of
events [26, 28]. 6689 T2D patient records, with 1304 of them taking rosigli-
tazone for at least 1 year [44]. After 3.6 years of median
PPAR‑γ agonists (thiazolidinediones, glitazones) follow-up, the need for vitreoretinal intervention (photo-
coagulation or vitrectomy) was significantly higher in the
Thiazolidinediones (TZDs) are nuclear PPAR-γ agonists. rosiglitazone group; although the underlying DR pathology
PPAR-γ activation regulates the transcription of genes impli- was not systematically recorded [44].
cated in sugar and lipid metabolism, ultimately improving No connection between TZD use and retinal sequelae
insulin sensitivity [31]. Ciglitazone, the first TZD, was inci- could be detected via post hoc analyses of the ACCORD-
dentally discovered by researchers working with lipid-low- EYE trial dataset [45, 46]; the latter entailed ophthalmologic
ering fibrates in 1982 [32]. Early-phase TZDs such as cigli- records, color fundus photographs, and VA measurements
tazone and troglitazone have been withdrawn, as they were with the ETDRS logarithmic chart from T2D patients [47].
linked to significant hepatotoxicity [33, 34]. The unfavorable In a cross-sectional study, DME rates were found unaffected
SE profile remains an issue among the newer members of by recent TZD exposure of minimum 3-month duration [45].
this class, pioglitazone and rosiglitazone; with a broader tox- A longitudinal analysis of baseline and 4-year follow-up data
icity spectrum, which entails fluid retention, congestive heart also failed to correlate TZDs with VA decline or changes in
failure, adipogenic weight gain, and higher risk of bladder DR development and progression; VA measurements were
cancer with pioglitazone [35–37]. The available RCT data obtained from all 6245 participants versus 2856 participants
on the risk of cardiovascular outcomes with TZDs have not with gradable fundus photographs at both timepoints [46].
demonstrated a significant benefit, either [35, 38].
To this point, the potential connection between TZDs Amylin analogs
and retinal complications has not been addressed in RCTs
(Table 2). Nevertheless, evidence from three large retro- Amylin is a polypeptide hormone synthesized and co-
spective studies demonstrates a significantly higher DR rate secreted with insulin by pancreatic beta cells. It participates
with TZDs, as measured by diabetic macular edema (DME) in glucose homeostasis by delaying gastric emptying, pro-
incidence [39–41]. In 2009, Fong and Contreras studied moting satiety, and suppressing glucagon secretion; in this
59,013 diabetic patients with at least one ophthalmologic setting, it became a promising target for T2D management
13
[48]. Yet, as its nomenclature implies, human amylin is amy- CI, 1.11–2.78]) and was 91% more likely to require photo-
loidogenic; in fact, it might be involved in T2D pathogenesis coagulation therapy (HR 1.91 [95% CI, 1.11–3.28]). Inter-
itself [49, 50]. Thus, modifications were indispensable for estingly, semaglutide use was accompanied by significantly
the development of the less amyloidogenic and more soluble lower DN rates. Vitreous hemorrhage, intravitreal therapy,
analog, pramlintide [49]. Subcutaneous pramlintide received and diabetes-related blindness each trended higher in the
FDA approval in 2005, to be used in conjunction with meal- semaglutide group, yet without reaching statistical signifi-
time insulin [25]. cance [61]. Using a similar study design and epidemiologi-
So far, only one pre-marketing RCT has addressed reti- cal profile, the LEADER trial randomized 9340 T2D par-
nal outcomes in 538 T2D participants receiving pramlintide ticipants to either subcutaneous liraglutide daily or placebo
versus placebo, over 52 weeks [51]. Pramlintide raised con- [62]. Although higher, the incidence of complications attrib-
cerns for retinal toxicity only at daily doses higher than the uted to DED was not statistically significant in the liraglutide
maximum approved 120 ug. However, no systematic docu- group (HR 1.15 [95% CI 0.87–1.52]). Meanwhile, liraglutide
mentation of DR or other retinal AEs was applied; instead, appeared to have a significantly favorable effect in terms of
they were grouped under the term “retinal disorders” [51]. DN (0.78 [95% CI 0.67–0.92]). We should point out that all
Therefore, additional randomized data are required. DR and DN assessments were part of a composite microvas-
cular outcome [62]. Furthermore, the REWIND trial rand-
GLP‑1 receptor agonists (glutides, enatides) omized 9901 T2D patients with high cardiovascular risk to
and twincretins (tirzepatide) either weekly dulaglutide or placebo [57]. After a median
follow-up of 5.4 years, patients on dulaglutide had a higher
GLP-1 belongs to incretins, a group of intestinal hormones incidence of DR outcomes, but the risk was not statistically
released from enteroendocrine L-cells after oral food intake significant (HR 1.24 [95% CI, 0.92–1.68]) [57]. In a post hoc
[52]. Incretins are a major stimulant for insulin secretion, analysis of the SUSTAIN-6 dataset, the risk of DR compli-
accounting for the higher plasma insulin levels following cations was independently strengthened by pre-existing DR
glucose digestion versus intravenous administration [53]. and a rapid decline in HbA1c within the first 16 weeks, both
Apart from insulin secretion, GLP-1 reduces glucose absorp- at a statistically significant level [68].
tion by delaying gastric emptying and promotes satiety [54]. In terms of retinal complications reported as AEs, RCT
Glucose-dependent insulinotropic polypeptide (GIP) is the data remain inconclusive. No differences were detected with
latest member from the incretin family to attract pharmaco- semaglutide in SUSTAIN trials 1 to 5; though we should
logic attention in the T2D field [55]. point out that different comparators were used in each of
In 2005, the GLP-1 receptor agonist (GLP-1RA) exena- those studies, fundoscopic evaluations were inconsistently
tide [25] was approved by the FDA, the first of six medi- performed, and patients treated for PDR or DME were
cations currently available from this class. Over the years, excluded [68] (Table 3). The PIONEER-6 trial showed a
GLP-1RAs were found beneficial for weight loss and protec- higher incidence of retinal events with semaglutide (7.1%
tion against life-threatening cardiovascular events [56–65]. vs. 6.3%), over 1.3 years of median follow-up [60]. In three
Nowadays, the combination of GLP-1RAs with metformin RCTs studying GLP-1RAs other than semaglutide [69–71],
is recommended as first-line treatment for T2D patients with the rates of DR AEs trended lower in the GLP-1RA groups,
high cardiovascular risk or obesity [27]. The GLP-1RAs although not statistically analyzed. A similar observation
liraglutide and semaglutide are also prescribed for obesity was reported with lixisenatide [64], this time referring to
in non-diabetic patients [66, 67]. Tirzepatide, the first dual a more generic “serious eye AE” rate. Post hoc analysis of
agonist on both GLP-1 and GIP receptors, has demonstrated AE reports from the EXSCEL trial found no differences in
similar benefits for glycemic control and weight loss among DR incidence or severity between patients receiving exena-
obese patients [55]. In April 2022, it received FDA approval tide once weekly versus placebo in addition to standard of
as the first member of a new anti-diabetic class, referred to care, over a median period of 3.2 years [72]. Among the 9
as “twincretins.” RCTs with publicly available results on tirzepatide, only two
To date, three multi-center double-blind RCTs have have reported DR AEs (NCT03861039, NCT03861052).
included at least one endpoint that systematically evalu- Tirzepatide doses higher than 10 mg appeared to be associ-
ated retinal complications with GLP-1RA use [57, 61, 62] ated with higher DR rates when compared to lower doses
(Table 3); meanwhile, no such study exists for tirzepatide. (NCT03861039, NCT03861052), as well as to controls
In SUSTAIN-6, 3297 T2D participants with high cardio- receiving 0.75 mg dulaglutide (NCT03861052), but again,
vascular risk or chronic kidney disease were assigned to no statistical analysis was performed.
weekly subcutaneous semaglutide or placebo [61]. After Retrospective data have also failed to clarify the rela-
104 weeks of treatment, the semaglutide group had 76% tionship between GLP-1RAs and retinal complications.
higher risk of developing DR complications (HR 1.76 [95% Of note, semaglutide was not evaluated by the majority of
13
Table 3 Key methodological parameters and retinal outcomes in association with GLP-1RAs use
First author, year Sample size (treat- T2D duration (years); Baseline DR/DED Anti-diabetic tested Control Follow-up/duration DR/DED outcome rates
ment group; control intervention(s) (years) (treatment group versus
group) control group)
RCTs systematically assessing at least one retinal endpoint

Marso et al. (2016) 3297 13.9 ± 8.1 Treatment group: Semaglutide + SOC Placebo + SOC 2.1 ↑ DED total rate in
(a) (1648; 1649) 84.0%; Control treatment group: 50
group: 82.8% (3.0%) versus 29
(1.8%); HR, 1.76
[95% CI, 1.11–2.78],
p = 0.02; ↑ retinal
photocoagulation
rate in treatment
group: 38 (2.3%)
versus 20 (1.2%);
HR, 1.91 [95% CI,
1.11–3.28], p = 0.02;
rates of VH, intra-
vitreal therapy and

DED-related blind-
ness trended higher in
treatment group
Marso et al. (2016) 9340 Treatment group: N/A Liraglutide + SOC Placebo + SOC 3.8 (median) ↓ Incidence of com-
(b) (4668; 4672) 12.8 ± 8.0; Control posite microvascular
group: 12.9 ± 8.1 outcome (DN, DR)
in treatment group:
355 (7.6%) versus
416 (8.9%); HR, 0.84
[95% CI, 0.73–0.97],
p = 0.02; DR out-
come (need for retinal
photocoagulation/
intravitreal agents,
VH, DED-related
blindness) trended
higher in treatment
group (HR, 1.15 [95%
CI, 0.87–1.52])
Gerstein et al. 9901 Treatment group: Treatment group: Dulaglutide (SQ) Placebo 5.4 Risk of DR outcomes
(2019) (4949; 4952) 10.5 ± 7.3; Control 9.1%; Control trended higher in
group: 10.6 ± 7.2 group: 8.9% treatment group:
HR, 1.24 [95% CI,
0.92–1.68]
RCTs reporting retinal/eye AEs
13
Table 3 (continued)
13
Kaku et al. (2011) 400 8.3 ±6.7 N/A Liraglutide Glibenclamide 1.0 ↓ Rate of overall eye
(268; 132) disorders (17.5% ver-
sus 21.2%) and DR
(6.0% versus 6.8%)
with treatment; not
statistically analyzed
Rosenstock et al. 566 Treatment group: N/A Albiglutide + glargine Lispro + glargine + 1.0; 0.5 for DR ↓ Rate of DR AEs in
(2014) (285; 281) 11.0 ± 7.0; Control + OADs OADs treatment group: 4.9%
group: 11.0 ± 6.0 versus 6.4%; not sta-
tistically analyzed
Pfeffer et al. (2015) 6068 Treatment group: Treatment group: Lixisenatide + SOC Placebo + SOC 2.1 (median) Serious eye AEs
(3034; 3034) 9.2 ± 8.2; Control 10.5%; Control trended lower in
group: 9.4 ± 8.3 group: 10.9% treatment group: 0.3%
versus 0.4%; not sta-
tistically analyzed
Seino et al. (2016) 257 14.51 ± 8.73 N/A Liraglutide + insulin Placebo + insulin 0.7 ↓ Rate of DR AEs in
(127, 130) treatment group: 9
(7.1%) versus 13
(10.0%); not statisti-
cally analyzed
Husain et al. (2019) 3183 14.9 ± 8.5 N/A Semaglutide + SOC Placebo + SOC 1.3 (median) ↑ Rate of DR AEs in
(1591; 1592) treatment group: 113
(7.1%) versus 101
(6.3%)
Post-hoc analyses
Vilsbøll et al. 8105; from 8 studies N/A SUSTAIN 6: 29.4%; Semaglutide Placebo +/− OADs N/A SUSTAIN 6: ↑ DR
(2018) of the SUSTAIN remaining studies: complication risk
clinical trial pro- 3.7−14.5% with semaglutide in
gram patients with baseline
DR; associated with
the rapid ↓ in HbA1c
within 16 weeks of
treatment onset; even
↑ risk with insulin
treatment
Studies other than SUS-
TAIN 6: no differ-
ences in DR AE rates
Table 3 (continued)
Bethel et al. (2020) 14,752 (7356; 7396) Ν/Α N/A Exenatide + SOC Placebo + SOC 3.2 (median) No difference in the
rate of DR events
(HR, 0.89 [95% CI,
0.74–1.07], p = 0.22);
regardless of DR his-
tory or HbA1c reduc-
tion from baseline
Varadhan et al. 165 12.0 58.8% Exenatide N/A; liraglutide and 0.6 ↑ Rate of DR progres-
(2011) semaglutide were sion versus improve-
excluded ment with treatment:
49 (29.7%) versus
32 (19.4%), p <
0.005; ↑ rate of DR
worsening versus
improvement with
treatment-induced
reduction in HbA1c:
47/133 (35.3%) ver-
sus 23/133 (17.3%),
p < 0.001; ↑ rate
of DR progression
with HbA1c reduc-
tion versus increase:
35.3% versus 6.2%, p
< 0.001
Varadhan et al. 49; subgroup of Var- 12.5 100.0% (DR and ↓ Exenatide N/A; liraglutide and 1.2 (mean) New-onset DR upon
(2014) adhan et al. 2011 HbA1c while on semaglutide were follow-up: 10/14
study exenatide) excluded (71.0%) improved;
3/14 (21.0%)
remained stable
Pre-existing DR: 15/25
(60.0%) that had
progressed during the
2011 study improved
on follow up and 5/25
(20.0%) remained
stable
13
Table 3 (continued)
13
Douros et al. 77.115 1.0 ± 2.4 0.0% GLP-1RAs exclud- ≥ 2 OADs/insulin 2.8 (median) No difference in DR
(2018) ing semaglutide ± incidence with
OADs GLP-1RAs versus ≥
2 OADs (aHR, 1.00
[95% CI, 0.85–1.17]);
↑ risk of DR inci-
dence between 6.1
and 12.0 months
(aHR, 1.44 [95%
CI, 1.06–1.95]); risk
not affected by T2D
duration or HbA1c
levels; ↓ risk of
DR incidence with
GLP-1RAs compared
to insulin (HR, 0.67
[95% CI, 0.51–0.90]);
requiring at least 12
months of use
Fadini et al. (2018) 114,814 FAERS N/A N/A GLP-1RAs excluding Any non-GLP-1RA 13.0 (FAERS reports) ↓ DED-related AEs
reports semaglutide antidiabetic with GLP-1RAs:
2.53/1000 versus
6.62/1000 (PRR, 0.38
[95% CI, 0.34–0.43],
p < 0.0001)
Wang et al. (2018) 130,110; (19,916; N/A Treatment group: New prescription of New prescription of 0.59–0.67 (vs. insu- No difference in preva-
110,194) 10.7–13.3% Control any GLP-1RA other insulin (82,849), lin); 0.58–0.78 (vs. lence of advanced DR
group: 13.0–14.2% than semaglutide TZDs (27,345) TZDs) requiring treatment
with GLP-1RAs ver-
sus TZDs: HR, 0.75
[95% CI, 0.53–1.06]
↓ Risk with GLP-
1RAs versus insulin:
HR, 0.50 [95% CI,
0.39–0.65])
Table 3 (continued)
Ueda et al. (2019) 18,280 (6650; 11,630) N/A 100.0% GLP-1RAs except DPP-4Is 1.8 (mean) No difference in risk
semaglutide of DR complications
(onset of retinal pho-
tocoagulation/intravit-
real agents, VH, RD,
RB, or vitrectomy):
HR, 1.07 [95% CI,
0.95–1.20])
Wang et al. (2019) 389 FAERS reports N/A N/A GLP-1RAs except One or more OADs 12.4 (FAERS reports) No difference in DR
semaglutide except GLP-1RAs risk with GLP-1RA
use
Gaborit et al. 3.154 12.0 Severe DR: 10.0%; GLP-1RAs OADs other than 2.0 No significant correla-
(2020) Νon-severe or no GLP-1RAs ± tion between GLP-
DR: 90.0% insulin 1RA exposure and
severe DR status, on
multivariate analysis
(OR, 1.139 [95% CI,
0.8–1.622], p = 0.47);
no association with
duration of GLP-1RA
exposure
Meta-analyses
Dicembrini et al. 39,078 (21,782; N/A N/A GLP-1RAs Anti-diabetics other 0.8 (mean) No difference in DR
(2017) 17,296); from than GLP-1RAs, incidence (onset
37/113 RCTs placebo of retinal photoco-
agulation/intravitreal
agents, DME, VH,
RD, DED-related
blindness);↓ DR risk
with GLP-1RAs
compared to SUs
(MH-OR, 0.58 [95%
CI, 0.34–0.99], p =
0.045); ↑ DR risk
with semaglutide
(MH-OR, 1.75 [95%
CI,1.10–2.78], p =
0.018); ↓ DR risk
with liraglutide
(MH-OR, 0.70 [95%
CI, 0.49–0.99], p =
0.043)
13
Table 3 (continued)
13
Andreadis et al. 9501 N/A N/A Semaglutide (SQ) Anti-diabetics other N/A; > 0.5 mean, DR risk trended higher
(2018) than semaglutide or checked manually with semaglutide
placebo (OR, 1.32 [95% CI,
0.98–1.77])
Avgerinos et al. 60,077; from 60 RCTs N/A N/A GLP-1RAs Anti-diabetics other N/A No difference in DR
(2018) than GLP-1RAs, incidence with
placebo GLP-1RAs treatment
versus any control
intervention; the same
applied to the inci-
dences of DME, RD,
retinal hemorrhage;
↑ VH incidence with
GLP-1RAs treat-
ment versus placebo
(OR, 1.93 [95% CI,
1.09–3.42])
Tang et al. (2018) 100,928; from 37 8.7 (mean) N/A GLP-1RAs Placebo 1.5 (median) DR risk trended higher
RCTs with GLP-1RAs in
pairwise meta-anal-
ysis; no difference to
controls in network
meta-analysis
Zhang et al. (2018) 33,457; 9.3–13.9 (mean) N/A GLP-1RAs ± other Placebo ± other N/A; range = 0.2–1.4 No difference in DR
(16,706; 16,751) OADs OADs risk versus placebo
(OR, 1.15 [95%
CI, 0.83–1.60], p =
0.394), with signifi-
cant heterogeneity
across trials (I2 =
73.5%, p = 0.023)
Avgerinos et al. 9890 3.4–16.2 (range) N/A Semaglutide (oral) ± Placebo ± other N/A; > 0.5 mean, No difference in DR
(2020) other OADs OADs checked manually incidence, both versus
placebo (OR, 1.22
[95% CI, 0.92–1.60])
and active compara-
tors (OR, 1.08 [95%
CI, 0.74–1.57])
Table 3 (continued)
Tsapas et al. (2020) 95,664; from 12/453 6.9 (median) N/A GLP-1RAs Placebo 0.5 (median) No difference in risk
RCTs of a DR-related event
versus placebo; ↑ risk
only with SQ sema-
glutide (OR, 1.75
[95% CI, 1.10–2.78])
in patients at ↑ CVD
risk, on metformin-
based background
therapy
Bethel et al. (2021) 49,936; (24,943; 10.0–15.0 (range) N/A; not in all trials; GLP-1RAs Placebo 3.4 (median) No difference in DR
24,993) 9.0–28.2% range, outcomes on random-
when recorded effects meta-analysis
(OR, 1.10 [95% CI,
0.93–1.30]), with
high heterogeneity
among studies (I2 =
52.2%, p = 0.063); ↑
risk of DR outcomes
for every every 1%
reduction in HbA1c
on meta-regression
analysis, during the
entire follow-up
period (LnOR slope,
0.77 [95% CI, 0.21–
1.34], p = 0.007)
Alexander et al. 71,517 N/A N/A GLP-1RAs Placebo or other 1.7 (mean) No difference in risk
(2022) OADs of DR-related events,
versus both placebo
(RR, 1.10 [95% CI,
0.93–1.29], p = 0.08)
and OADs (RR, 0.72
[95% CI, 0.52–1.00],
p = 0.6)
13
Table 3 (continued)
13
Wang et al. (2022) 22,096 8.9 (median) N/A Semaglutide Placebo or other N/A; > 0.5 mean, No difference in DR
OADs checked manually risk, when all trials
are analyzed together:
RR, 1.14 [95% CI,
0.98–1.33]; subgroup
analysis found ↑ DR
risk when semaglu-
tide was compared
to placebo (RR, 1.24
[95% CI, 1.03–
1.50]),and ↑ DR risk
with semaglutide
with age ≥ 60 years
(RR, 1.27 [95% CI,
1.02–1.59]) and with
T2D duration ≥ 10
years (RR, 1.28 [95%
CI, 1.04–1.58])
Wei et al. (2022) 60,081 N/A N/A GLP-1RAs Placebo N/A; > 0.5, checked No difference in risk of
manually DR/retinal arteriopa-
thy (RR, 1.54 [95%
CI, 0.74–3.23], p =
0.25)
Yoshida et al. 2519 (mean); 4 RCTs 10.5 N/A; 9.0–28.2% GLP-1RAs; liraglu- Placebo or other 1.67 ↑ risk of rapid DR
(2022) range, when tide, dulaglutide, OADs worsening, in RCTs
recorded semaglutide with CVD benefits
(OR, 1.23 [95% CI,
1.05–1.44]); subgroup
analysis showed ↑
DR risk in RCTs
with follow-up >
52 weeks (OR, 1.2
[95% CI, 1.0–1.43]),
and versus placebo
(OR, 1.22 [95% CI,
1.05–1.42]); DR
risk trended higher
with T2D duration
≥ 10 years, and with
multi-nationality of
participants
Table 3 (continued)
Albert et al. (2023) 49,936 from 6 RCTs 12.0 ± 0.8 (6 RCTs); N/A; 9.0–31.0% GLP-1RAs (6 RCTs Placebo or other 1.3 ± 0.5 (6 RCTs); 6 RCTs with CVD
with CVD benefits); 10.6 ± 4.0 (11 range, when with CVD benefits); OADs 3.1 ± 1.3 (11 RCTs) benefits including
11,894 from 11 RCTs) recorded semaglutide (11 all GLP-1RAs: ↑
RCTs on semaglu- RCTs) DR risk only with
tide SQ semaglutide
(RR, 1.73 [95% CI,
1.10–2.71], p = 0.02);
subgroup analysis
showed ↑ DR risk
with decreases in
HbA1c (logRR,
−0.67 [ΔHbA1c], p =
0.076)
11 RCTs on semaglu-
tide: ↑ DR risk with

SQ semaglutide
when given for > 1
year (RR, 1.559 [95%
CI, 1.068–2.276], p
= 0.022), and with
↓ in HbA1c > 1.0%
(RR, 1.59 [95% CI,
1.092–2.316], p =
0.016)
Abbreviations: GLP-1RAs, glucagon-like peptide-1 receptor agonists; T2D, type 2 diabetes; DR, diabetic retinopathy; DED, diabetic eye disease; RCT, randomized controlled trial; SOC, stand-
ard of care; HR, hazard ratio; CI, confidence intervals; VH, vitreous hemorrhage; N/A, not applicable/available; DN, diabetic nephropathy; AEs, adverse events; OADs, oral anti-diabetic drugs;
HbA1c, hemoglobin A1c; aHR, adjusted hazard ratio; FAERS, FDA Adverse Event Reporting System; PRR, proportional reporting ratio; DPP-4Is, dipeptidyl peptidase 4 inhibitors; RD, retinal
detachment; RB, retinal bleeding; OR, odds ratio; GLDs, glucose-lowering drugs; TZDs, thiazolidinediones; SGLT-2Is, sodium glucose co-transporter-2 inhibitors; ROR, reporting odds ratio;
DME, diabetic macular edema; SUs, sulfonylureas; MH-OR, Mantel–Haenszel odds ratio; SQ, subcutaneous; CVD, cardiovascular disease
13
retrospective studies available, as they were published prior In 2019, the double-blind multi-center CARMELINA
to or shortly after its FDA approval [20, 73–78] (Table 3). trial included a composite microvascular endpoint with
A review of 114,814 records from the FDA Adverse Event several ocular components, such as the time until first laser
Reporting System, spanning a continuous 13-year period, photocoagulation, intra-vitreal injection of anti-vascular
demonstrated decreased DED-related AEs [74]. Another endothelial growth factor for DR, vitreous hemorrhage,
investigation on the same FDA database, performed by a and diabetes-related blindness[95] (Table 4). The study was
different group over the same period, found no association designed to evaluate major cardiovascular outcomes of lina-
between GLP-1RAs and DR [78]. Meanwhile, the preva- gliptin versus placebo over a mean follow-up of 2.2 years,
lence of advanced DR requiring treatment and DR incidence, in 6979 T2D participants with advanced cardiovascular and
both identified via procedural and diagnostic codes in a renal risk. Ocular complications trended lower in the lina-
cohort of 213,652 US Medicare beneficiaries, revealed no gliptin cohort, although not significantly (HR 0.73 [95% CI
differences when GLP-1RAs were compared to either TZDs 0.47–1.12]). Meanwhile, the composite microvascular end-
or insulin; however, the median duration of each medication point co-evaluating renal with retinal complications was
usage was rather short (0.58–0.78 years for GLP-1RAs ver- found significantly reduced (HR, 0.86 [95% CI 0.78–0.95])
sus TZDs and 0.59–0.67 years for GLP-1RAs versus insulin) [95]. On the other hand, the TECOS trial showed a slightly
[20]. Analogous findings were published from a European higher DR AE rate with sitagliptin versus placebo (2.8%
cohort with baseline DR, suggesting no association between vs. 2.2%) over 3 years of follow-up, without further statisti-
GLP-1RA use and code-retrieved DR complications, with cal analysis [93]. DN (9.2% vs. 9.4%) and diabetes-related
DPP-4Is as the active comparator [75]. Fundoscopic evalu- blindness (0.3%) were similar between the groups, yet
ations in a French cohort revealed a trend towards higher patients with severe hypoglycemia or renal insufficiency
GLP-1RA use among patients with severe DR, and interest- had been excluded [93].
ingly, metformin demonstrated a protective effect against The available retrospective data are not sufficient to reveal
severe DR [79]. In a different study of 77,115 patients, GLP- a clear association between DPP-4Is and the retina, either
1RA use was not associated with changes in overall DR (Table 4). In a Taiwanese cohort of 40,888 T2D patients
rates, with the only exception of a significant increase within with pre-existing DR, the risk of both vitreous hemorrhage
the first 6–12 months of treatment [73]. Two smaller-scale and DME increased significantly over 2.5 years of DPP-4I
studies of 165 patients also revealed significantly higher DR use, indicating accelerated DR progression [96]. A com-
incidence and progression over the first 10 months of exena- posite DR outcome that included intravitreal injections, vit-
tide treatment; although at 15 months, 10 out of the 14 new rectomy, and photocoagulation was also found significantly
DR cases had improved [76, 77]. elevated in the DPP-4I group; however, patients with any
In terms of the meta-analyses studying GLP-1RAs, only history of retinal disease or intervention had been excluded
a minority of their pooled data represent original studies [96]. Two studies over the same Korean database yielded
with an endpoint or AE dedicated to DR (Table 2). The inconsistent results in terms of DR; the majority of which
majority of those meta-analyses found no significant differ- did not reach statistical significance [97, 98]. DPP-4Is were
ences in DR risk [18, 72, 80–87]. A small number of stud- associated with a higher trend towards a composite DR
ies suggested a significantly higher DR risk versus placebo, outcome, which only became significant within the first 12
associated either specifically with semaglutide [19, 80, 85], months of treatment [97]. In the other study though, after
or with the entire GLP1-RA class [88]. On the other hand, adjustment for confounders, addition of DPP-4Is to met-
Dicembrini and colleagues showed a lower DR risk with formin trended towards lower DR progression, compared
GLP-1RAs, when compared to sulfonylureas (SUs) [84]. to the combination of metformin with SUs [98]. A smaller
study of 6030 T2D patients with no baseline DR suggested
DPP‑4 inhibitors (gliptins) a significantly more favorable profile of DR incidence with
DPP-4Is again versus SUs [99]. Yet, all the aforementioned
Normally, GLP-1 is rapidly metabolized by DPP-4, another studies had several limitations, such as the absence of data
membrane-bound enzyme of the intestinal brush border and on glycemic control, Diabetic Retinopathy Severity Scale
liver [89]. Sitagliptin was the first FDA-approved DPP-4 (DRSS) assessment, and fundus photographs [97, 98], as
inhibitor (DPP-4I), in 2006; since then, four additional DPP- well as the exclusion of patients with any history of micro-
4Is have become commercially available in the USA. Per vascular complication [99].
os administration once daily, and the relatively low rates of Meanwhile, four retrospective studies from different groups
hypoglycemic episodes have increased DPP-4Is use for T2D found no significant differences in DR incidence and/or progres-
patients with unstable glucose levels [10, 90]. RCT data on sion with DPP-4Is use, compared to other oral anti-diabetics [20,
the risk of macro- and microvascular outcomes with DPP- 100–102]. In one of the aforementioned cohorts (n = 138), Sim-
4Is demonstrate a slightly protective effect [91–95]. mons and colleagues evaluated the effects of oral anti-diabetic
13
Table 4 Key methodological parameters and retinal outcomes in association with DPP-4Is use
First author, year Sample size (n) T2D duration (years) Baseline DR/DED Anti-diabetic(s) tested Control Follow-up/duration DR/DED outcome rates
intervention(s) (years) (treatment group versus
control group)

Rosenstock et al. 6979; treatment Treatment group: 15.0 N/A Linagliptin Placebo 2.2 (median) ↓ Composite micro-
(2019) group: 3494; control ± 9.6 vascular endpoint
group: 3485 Control group: 14.5 (ocular and renal
± 9.3 function) in treatment
group: HR, 0.86 [95%
CI, 0.78–0.95], p =
0.003
Composite ocular end-
point (time till first
intervention for DR,
vitreous hemorrhage,
diabetes-related
blindness) trending ↓
in treatment group
RCTs reporting retinal/eye AEs
Green et al. (2015) 14,671; treatment 11.6 ± 8.1 N/A Sitagliptin Placebo 3.0 (median) ↑ total DED rates in the
group: 7332; control treatment group: 3.1%
group: 7339 versus 2.5%; ↑ DR in
the treatment group:
2.8% versus 2.2%;
same diabetes-related
blindness between
the groups (0.3%);
no statistical analysis
performed
Kolaczynski et al. 6030; treatment Treatment group: 0.0% (exclusion Vildagliptin SUs Treatment group: ↓ DR incidence in the
(2016) group: 3015; control 3.1± 3.4 criterion) 2.2 ± 1.1 Control treatment group: HR,
group: 3015 Control group: 3.2 group: 2.9 ± 1.7 0.55 [95% CI, 0.39–
± 3.4 0.77], p = 0.0004;
↓ neuropathy inci-
dence in treatment
group: HR, 0.71 [95%
CI, 0.60–0.85], p =
0.0001; ↓ nephropa-
thy and DFU risk
↓ Incidence of com-
posite microvascular
outcome in treatment
group: OR, 0.70 [95%
CI, 0.61–0.82], p <
13
0.0001
Table 4 (continued)
13
control group)
Kim et al. (2018) 29,044; treatment Treatment group: 4.6 Treatment group: Use of DPP-4Is at Any anti-diabetic 2.3 (median) ↑ Risk of a composite
group: 14,522; con- ± 3.3 13.7% Control least once other than DPP-4Is DR outcome (vitreous
trol group: 14,522 Control group: 4.4 group: 13.7% hemorrhage, vitrec-
± 3.4 tomy/laser, injec-
tion, blindness) in
the treatment group:
HR, 1.08 [95% CI,
0.93–1.26]; risk ↑ in
patients using DPP-
4Is for < 12 months,
especially in the sup-
population without
baseline DR
Risk of each compo-
nent of DR outcome
unaffected by DPP-4I
use; composite and
individual DR out-
comes unaffected by
baseline DR
Wang et al. (2018) 200,006; treatment N/A Treatment group: New prescription of New prescription of 0.75–0.87 (DPP-4Is No difference in preva-
group: 90,702; con- 11.6% Control any DPP-4I SUs (87,973), TZDs vs. SU); 0.74–0.80 lence of advanced DR
trol group: 109,304 group: 9.4–12.2% (22,231) (DPP-4Is vs. TZDs) requiring treatment
No difference in DR
incidence
Chung et al. (2019) 11,200; treatment N/A N/A DPP-4Is + metformin SUs + metformin 2.0 ↓ Risk of DR progres-
group: 6136; control (19.7 months, mean) (20.3 months, sion in the treatment
group: 5064 mean); TZDs + group, after adjust-
metformin (18.6 ment for age and
months, mean) sex: HR, 0.74 [95%
CI, 0.62–0.89], p =
0.001; ↓ risk of DR
progression after
secondary adjustment
for comorbidities,
duration of metformin
therapy, intravitreal
injections, and calen-
dar index year
Table 4 (continued)
control group)
Kang et al. (2021) 40,888; treatment Treatment group: 11.2 100.0% DPP-4Is ≥ 80% for Any anti-diabetic 2.5 (mean) ↑ Risk of compos-
group: 20,444; con- ± 2.8 9.0% with baseline the first 6 months of other than TZDs ite DR outcome
trol group: 20,444 Control group: 11.2 PDR treatment (vitreous hemor-
± 2.9 rhage, tractional RD,
DME, interventional
retinal therapies)
in treatment group:
SHR, 1.23, [95%
CI, 1.06–1.44], p =
0.008;↑ risk of VH
(SHR, 1.24, [95%
CI, 1.05–1.48], p =
0.013) and DME
(SHR, 1.48, [95%
CI, 1.03–2.13], p =
0.035) in treatment
group
↑ Risk of composite
retinal interven-
tion outcome (laser,
vitrectomy, intravit-
real injection), as well
as for each retinal
intervention sepa-
rately in treatment
group (p < 0.001)
Ueki et al. (2021) 5150; treatment Treatment group: 8.33 N/A Alogliptin Other oral anti- 3.0 No differences in DR
group: 3395; control Control group: 7.70 diabetics incidence and/or
group: 1755 progression
Deng et al. (2022) 8252; treatment N/A Treatment group: Sitagliptin (added Glimepiride (4318), 2.0 No differences in DR
group: 2993; control 5.1% to > 8 weeks met- liraglutide (690), risk (HR)
groups: 5259 Control groups: formin) insulin (251)
4.2–8.8%
Simmons et al. 138; stratified per N/A 78% (baseline CME) DPP4-Is (started at SUs, metformin, 2.1 (median) ↓ CME prevalence
(2022) treatment (7% on a CRVO onset) insulin, TZDs when sitagliptin was
DPP4-I at starting used at CRVO onset,
point) p < 0.01; no benefit
in visual outcomes
or treatment burden
for T2D patients with
CRVO
13
Table 4 (continued)
13
control group)
Meta-analyses
Tang et al. (2018) 100,928; 8/37 RCTs 8.7 (mean) N/A Any DPP-4I Placebo 1.5 (median) Risk of DR events (DR,
on DPP-4Is DME, VH, onset
of diabetes-related
blindness, intravit-
real injection/laser)
trended ↑ in treatment
group versus placebo
on network meta-
analysis; ↑ risk of DR
events in treatment
group versus placebo
on pairwise meta-
analysis: OR, 1.27,
[95% CI, 1.05–1.53]
Abbreviations: DPP-4Is, dipeptidyl peptidase 4 inhibitors; DR, diabetic retinopathy; DED, diabetic eye disease; RCT, randomized controlled trial; N/A, not applicable/available; HR, hazard
ratio; CI, confidence intervals; SUs, sulfonylureas; DFU, diabetic foot ulcer; TZDs, thiazolidinediones; PDR, proliferative diabetic retinopathy; RD, retinal detachment; DME, diabetic macular
edema; SHR, sub-distribution hazard ratio; VH, vitreous hemorrhage; OR, odds ratio; T2D, type 2 diabetes; CME, cystoid macular edema; CRVO, central retinal vein occlusion
administration starting at the time of central retinal vein occlu- [116–118] (Table 5). A Korean cohort of 50,756 T2D
sion (RVO) diagnosis [101]. Although CME was not studied patients, receiving either SGLT-2Is or DPP-4Is, was stud-
as a dedicated endpoint corresponding to DR status, DPP4-Is ied in terms of DR incidence and progression over a median
were associated with lower CME prevalence at final follow-up. period of less than 1 year [116]. DR incidence was found
However, none of the anatomic benefits achieved with oral anti- significantly lower with SGLT-2Is in the patient sub-group
diabetics could be translated into significantly improved visual with no baseline DR; meanwhile, there was no significant
outcomes or treatment burden [101]. Finally, network meta- difference in DR progression in the subgroup of patients
analysis of 8 trials did not reveal a significantly higher risk of with pre-existing disease. Other than its non-diverse popula-
DR events with DPP-4Is; compared to placebo though, this risk tion sample, this study had several limitations including the
was found significantly increased in pairwise meta-analysis [37] lack of glycemic control data, DR reporting that was mainly
(Table 4). based on ICD-10 codes, and incomplete history of intravit-
real injections [116]. In a different cohort of 11,053 T2D
SGLT‑2 inhibitors (gliflozins) patients from Taiwan, newly prescribed SGLT-2Is (empa-
gliflozin, dapagliflozin, canagliflozin) were associated with
SGLT-2 is a transmembrane protein located at the early proxi- lower DME risk, compared to GLP-1RAs (liraglutide, dula-
mal tubule, responsible for reabsorption of approximately 90% glutide) [118]. Using the “Archimedes” simulation model for
of glucose filtered by healthy kidneys [103]. Phlorizin, a com- human diseases, Dziuba et al. projected the outcomes of a
pound isolated from apple trees, was the first SGLT-2 inhibitor 20-year trial comparing dapagliflozin with standard of care
(SGLT-2I) discovered in 1835, yet, it was not until 1973 that its in T2D patients. The model estimated a significant decrease
glycosuric effect and contribution to glucose homeostasis were (9.8%) in DR incidence in the dapagliflozin group, based
fully understood [104, 105]. The first SGLT-2I, canagliflozin, on 29,878 patient records from the US National Health and
was approved by the FDA in 2013, followed by more substances Nutrition Examination Survey database [117].
in the same class [103, 105]. Due to their beneficial effect on The majority of data from five meta-analyses revealed no
cardiovascular morbidity, mortality, and DN [63, 106–114], significant differences among SGLT-2Is and various compara-
SGLT-2Is are nowadays promoted as first-line treatment for tors [18, 19, 22, 119, 120] (Table 5). A non-significantly lower
T2D patients with chronic kidney disease or cardiovascular DR risk was demonstrated with SGLT-2Is upon both network
comorbidities [10]. and pairwise meta-analyses of 5 original studies [18]. The same
To this moment, no RCT has incorporated a dedicated meta-analysis suggested a significantly higher DR risk with SUs,
endpoint investigating DED with SGLT-2Is use (Table 5). in comparison to either placebo or SGLT-2Is on network meta-
A composite microvascular outcome consisting of both reti- analysis [18]. The meta-analysis by Li and colleagues showed
nal and renal events was analyzed during the EMPA-REG similar results, with no significant differences in total eye AEs or
OUTCOME study, a multi-center double-blind RCT, which DR risk with SGLT-2Is [119]. However, not every study in this
mainly evaluated the effects of empagliflozin on DN [113]. meta-analysis included DR data at both baseline and throughout
Among the 6968 participants with advanced T2D and cardi- the study period [119]. Finally, the meta-analysis by Zhou and
ovascular history followed over 3.1 years, the microvascular associates indicated a significantly protective effect of empa-
outcome was found significantly lower in the empagliflozin gliflozin in terms of DR risk versus SUs/placebo, yet, it also
group (14.0% vs. 20.5%, p < 0.001). Yet, this was mainly reported a significantly higher risk of vitreous disease with cana-
driven by the difference in the incidence or worsening of gliflozin [22] (Table 5).
diabetic nephropathy (12.7% vs. 18.8%, p < 0.001); instead, Meanwhile, SGLT-2Is were recently associated with
the rates of retinal photocoagulation (0.9% vs. 1.2%, p = higher risk of RVO. A cohort of 94,738 T2D Korean
0.134), vitreous hemorrhage (0.6% vs. 0.7%, p = 0.815), and patients, with ETDRS-graded DR at baseline, was retro-
DED-related blindness (0.1%) were not different between spectively followed over a mean period of 2.6 years [121].
the study groups [113]. These results were consistent with a RVO risk was found to be significantly more pronounced
post hoc analysis of the EMPA-REG OUTCOME trial, this among new SGLT-2I users, compared to other anti-diabetics
time focusing on the first occurrence of a composite DR (HR 1.264 [95% CI 1.056–1.513]), even though baseline
outcome entailing retinal photocoagulation, vitreous hemor- DR rates were significantly lower in the SGLT-2I group.
rhage, diabetes-related blindness, and intravitreal injection Additional analysis identified older age (> 60 years), and
[115] (Table 5). The risk of the aforementioned outcome advanced chronic kidney disease as the main drivers of the
trended lower in the empagliflozin cohort, yet, the majority observed difference in risk. Despite limitations such as the
of its constituents were plainly reported as AEs, without unknown compliance to prescribed anti-diabetics, RVO
systematic DR grading or routine retinal imaging. diagnosis based on modified ICD-10 codes, and the lack of
Although scarce, the retrospective data on SGLT-2Is routine fundoscopy, such a complication definitely requires
and DR also suggests a potentially protective correlation further investigation [121].
13
Table 5 Key methodological parameters and retinal outcomes in association with SGLT-2Is use
13
control group)

Wanner et al. (2016) 7018; Treatment group: N/A Empagliflozin + SOC Placebo + SOC 3.1 (median) ↓ Rate of compos-
Treatment group: > 10.0 (2286); ite microvascular
4685; Control group: Control group: > outcome (onset of
2333 10.0 (1156) retinal photocoagula-
tion, VH, DED-related
blindness, plus new/
worsening DN):
14.0% versus 20.5%,
HR, 0.62 [95% CI,
0.54–0.70], p < 0.001
Secondary outcomes
related to DR/DED
(onset of retinal
photocoagulation, VH,
DED-related blind-
ness), trended lower in
treatment group
Post-hoc analyses
Inzucchi et al. 7020 > 10.0 (57.0%) 22.0% Empagliflozin + SOC Placebo + SOC 3.1 (median) First occurrence of
(2019) a composite DR
outcome (retinal
photocoagulation,
VH, DED-related
blindness, intravitreal
injection) trended
lower in treatment
group: HR, 0.78 [95%
CI, 0.54–1.12], p =
0.1732
Dziuba et al. (2014) 29,878; (NHANES 7.0 (mean) N/A Dapagliflozin + SOC SOC (SUs, TZDs, 20.0 (simulation ↓ DR incidence pro-
database) DPP-4Is, GLP- model) jected in treatment
1RAs, insulin) group and subgroups:
9.8%
Table 5 (continued)
control group)
Chung et al. (2019) 50,756; Cohort 1: N/A Cohort 1: 0.0% SGLT-2Is DPP-4Is < 1 (median) Cohort 1: ↓ DR inci-
41,430 (20,175 per Cohort 2: 100.0% dence in treatment
treatment group); group; HR, 0.89
Cohort 2: 9326 [95% CI, 0.83–0.97],
(4663 per treatment p = 0.006; DR risk
group) remained mostly ↓
after stratification for
confounders such as
CVD history, serum
glucose < 126 mg/dL,
SBP < 40 mmHg
Cohort 2: risk of DR
progression (need for
PRP/vitrectomy, VH,
RD, NV glaucoma) no
different between the
groups
Lee et al. (2021) 94,738; Treatment Treatment group: Treatment group: SGLT-2Is; dapagli- Oral anti-diabetics 2.6 (mean) ↑ RVO incidence (ICD-
group: 47,369; Con- 6.12 ± 3.84; Con- 13.62% Control flozin; ipragliflozin; other than SGLT-2Is 10) in treatment group
trol group: 47,369 trol group: 6.07 group: 14.48% empagliflozin (log-rank p = 0.0104)
± 3.6 ↑ RVO risk in new
SGLT-2Is users (HR,
1.264 [95% CI, 1.056–
1.513]), with age >
60 (HR, 1.523 [95%
CI, 1.198–1.936]),
and with eGFR < 60
(HR, 3.134 [95% CI,
1.554–6.318])
Su et al. (2021) 11,053; Treatment N/A Treatment group: SGLT-2Is: empagli- GLP-1RAs: liraglutide 3.0 ↓ DME incidence
group: 9986; Control 6.9% Control flozindapagliflozin dulaglutide in treatment group
group: 1067 group: 7.7% canagliflozin starting SGLT-2Is
(HR, 0.75 [95% CI,
0.64–0.88])
13
Table 5 (continued)
13
control group)
Meta-analyses
Tang et al. (2018) 100,928; 5/37 RCTs 8.7 N/A SGLT-2Is Placebo/SUs 1.5 (mean) DR risk trended lower
(10,957 patients) on with SGLT-2Is in both
SGLT-2Is pairwise and network
meta-analysis
↑ DR risk with SUs
versus SGLT-2Is in
network meta-analy-
sis: OR, 2.11 [95% CI,
1.07–4.17]
Tsapas et al. (2020) 95,664 (assessed for 6.9 (median) N/A SGLT-2Is Placebo 0.5 (median) No difference in risk
DR); from 12/453 of a DR-related event
RCTs compared to placebo
Li et al. (2021) 39,982; 9 RCTs on N/A N/A (only 4/9RCTs) SGLT-2Is Placebo 2.8 (mean) No difference in risk
SGLT-2Is of total ocular events
(RR, 0.97 [95% CI,
0.85–1.11]) or DR
(RR, 0.98 [95% CI,
0.84–1.16]) with
SGLT-2Is; no differ-
ences in correlation
with T2D history or
relevant comorbidities
Ma et al. (2022) 82,684; Treatment N/A N/A SGLT-2Is Placebo N/A; > 0.5 after No difference in DR
group: 47,070; Con- manual check of incidence
trol group: 35,614 trials ↓ DR incidence in
treatment group, when
T2D duration < 10
years: OR, 0.32 [95%
CI, 0.13–0.76], p =
0.01
Table 5 (continued)
control group)
Zhou et al. (2022) 58,236; Treatment N/A N/A SGLT-2Is Placebo and oral anti- 0.3–8.0 (range)* ↓ DR risk with
group: 33,106; Con- diabetics other than empagliflozin, versus
trol group: 25,130 SGLT-2Is controls (4 RCTs):
25/41 RCTs on RR, 0.44 [95% CI,
retinal diseases 0.2–0.99], p = 0.05
↓ Risk of retinal dis-
ease with ertugliflo-
zin, versus controls:
RR, 0.47 [95% CI,
0.26–0.86], p = 0.01
↑ DR risk of vitreous
disease with canagli-
flozin, versus placebo
RR, 4.5 [95% CI,
1.14–17.7], p = 0.03
Abbreviations: SGLT-2Is, sodium glucose co-transporter-2 inhibitors; DR, diabetic retinopathy; DED, diabetic eye disease; RCT, randomized controlled trial; N/A, not applicable/available; SOC,
standard of care; VH, vitreous hemorrhage; DN, diabetic nephropathy; HR, hazard ratio; CI, confidence intervals; NHANES, National Health and Nutrition Examination Survey; SUs, sulfony-
lureas; TZDs, thiazolidinediones; DPP-4Is, dipeptidyl peptidase 4 inhibitors; GLP-1RAs, glucagon-like peptide-1 receptor agonists; CVD, cardiovascular disease; SBP, systolic blood pressure;
PRP, pan-retinal photocoagulation; RD, retinal detachment; NV, neovascular; eGFR, estimated glomerular filtration rate; DME, diabetic macular edema; OR, odds ratio; RR, relative risk; T2D,
type 2 diabetes. *Follow-up duration was shown to have no effect on sub-group analysis in the study by Zhou et al. [22]
13
Fig. 2 Graphical abstract summarizing the main retinal outcomes anti-diabetic class. Abbreviations: DR, diabetic retinopathy; RCT,
from all clinical studies we reviewed, in a qualitative manner. Seven randomized controlled trial; T2D, type 2 diabetes; PPAR-γ, peroxi-
(7) anti-diabetic classes that received FDA approval in or after 1995 some proliferator-activated receptor gamma; DME, diabetic macular
were examined, in chronological order of their introduction to the edema; DED, diabetic eye disease; GLP-1RAs, glucagon-like pep-
market; the only exception was the class of twincretins, due to its tide-1 receptor agonists; AEs, adverse events; DPP-4Is, dipeptidyl
pharmacological similarity with GLP-1RAs. Green, gray, and red peptidase 4 inhibitors; SGLT-2Is, sodium glucose co-transporter-2
overlays on the retina animations stand for protective, inconclu- inhibitors
sive, and harmful association respectively, with the corresponding
Ongoing trials Discussion
The effects of newer anti-diabetics on the retina are being Despite being a fundamental component of T2D guidelines
assessed by ongoing studies in the field. Based on the [10], newer-generation anti-diabetics have not received
indications of DR worsening with semaglutide [19, 60, enough granular investigation in terms of their effects on DR,
61, 80, 85], the FOCUS trial (NCT03811561) has been DED, and the retina in general. The randomized prospective
designed to specifically address semaglutide’s effects data studying this relationship are scarce for all the anti-dia-
on the retina. In this phase 3, placebo-controlled RCT, betic classes we reviewed, while the majority of such evidence
1500 T2D participants will be followed for up to 5 years, was collected via AE reporting (Table 1). For instance, no
ophthalmologic monitoring will include DR grading, VA large RCTs have systematically evaluated DR endpoints with
measurements, DME, and the need for ophthalmologic the use of alpha-glucosidase inhibitors, TZDs, or amylin ana-
intervention. Two smaller-scale observational cohorts are logs to date (Table 1). From these anti-diabetic categories, the
currently investigating retinal outcomes following GLP- retrospective data correlating TZDs with higher DME inci-
1RA use: the SEVERAL study (NCT05136287) with dence [39, 40], although not replicated in all studies, should
360 participants and expected due-date in 2022 and the probably be taken under consideration in clinical practice,
“Observational Study of ASCVD Risks of Type 2 Diabe- particularly in patients with DME history and in those who
tes in East China” (NCT04866667) with 800 participants, receive TZDs in combination with insulin [40].
concluding in 2024. An ongoing study in the SGLT-2I In the GLP-1RA and dual GLP-1/GIP receptor agonist
field, the SMARTEST trial (NCT03982381), includes a (tirzepatide) categories, the risk of retinal complications
dedicated outcome for DR occurrence or progression; remains inconclusive, with the only exception of semaglu-
the latter will be assessed among 4300 T2D participants tide. The latter was indeed associated with a significantly
receiving either dapagliflozin or metformin for approxi- higher risk of DR-related outcomes in two RCTs [60, 61],
mately 36 months. one of them with a dedicated DR endpoint [61], and in
five subsequent meta-analyses [19, 80, 84, 85, 88]. The
13
DR-worsening effect of semaglutide might be attributed to The association between SGLT-2Is and DR also remains
the rapid decline it causes in blood sugar levels. Neverthe- to be further elucidated. In the EMPA-REG OUTCOME
less, the signal remains strong, and thus, additional caution study, the only relevant RCT to date, the rates of DR-related
is recommended as it regards to DR monitoring in the early endpoints were almost identical between the empagliflo-
stages of semaglutide administration; especially within the zin and the placebo cohorts [113]. Of note, the majority
first 18 months of treatment, in which the risk of DR wors- of evidence available from post-hoc analyses [115], retro-
ening would be higher, based on the landmark findings of spective studies [116–118], and meta-analyses [18, 22, 119]
the DCCT study [6]. Interestingly, semaglutide had been points towards a protective effect of SGLT-2Is (Table 5);
excluded from all the RCTs that showed lower DR risk with with the only exception of the significantly higher risk of
GLP-1RA use, based on data collected from AE reports vitreous disease with canagliflozin, as reported by Zhou
[69–71]. and colleagues [22]. On the other hand, the retrospective
Meanwhile, a closer examination of the RCT data on other association of SGLT-2Is with RVO in elderly and patients
members of this class could raise suspicion of retina-specific with advanced kidney disease should be further investigated
toxicity for the entire family of incretin-mimicking anti-dia- via a prospective randomized design [121]. T2D patients
betics. For instance, the GLP-1RA liraglutide had opposing in the aforementioned groups receiving SGLT-2Is should
effects on the two most prevalent microvascular complica- be educated to recognize RVO symptoms and immediately
tions of T2D, DR, and diabetic nephropathy, even though the seek ophthalmologic care for any change in their vision,
underlying pathophysiology is considered to be similar in both especially if additional factors predisposing to RVO, such
conditions. Liraglutide was associated with non-significantly as hypertension, smoking, elevated blood viscosity, and
higher DR rates versus a significant benefit over diabetic vasospasm, are present.
nephropathy [62]. A similar pattern was observed with dula- In most RCTs we reviewed, follow-up duration was
glutide in the REWIND trial, when looking at the incidence shorter than three years (Tables 2–5), and data on pre-exist-
of DR versus renal complications [57]. The unpublished ing DR as well as DR severity were not consistently col-
randomized data on tirzepatide also indicate increased DR lected. This could over-estimate the early worsening effect
AEs, especially at doses higher than 10 mg (NCT03861039, of glucose lowering on DR [6], potentially masking the true
NCT03861052 ). Thus, future trials in the field should aim at outcomes of newer anti-diabetics on the retina. Unfortu-
further elucidating the potential association between incretin- nately, the majority of DR-related events in the available
receptor agonists and DR complications. RCTs originate from AE reporting and not from direct imag-
The existing evidence on the relationship between DR ing of the retinal vasculature. In our review, there were two
and DPP4-Is entails much higher variability (Table 4). For anti-diabetic classes (GLP1-RAs and DPP-4Is) where the
instance, the systematically evaluated retinal endpoint in findings from RCTs with at least one dedicated DR end-
the CARMELINA trial trended lower for linagliptin [94], point were almost contradictory to the data from AE reports,
whereas sitagliptin has been associated with significantly retrospective studies, and meta-analyses (Tables 3 and 4).
increased DR AEs in the TECOS trial [93]. The retrospec- Also, there is a long-lasting limitation in the design of RCTs
tive data on this topic are also quite variable [20, 96–102], in the anti-diabetics field, which only recently began to be
even in the case of studies referring to the same population discussed. Historically, those RCTs have aimed to primar-
[97, 98]. More specifically, the effects of DPP4-Is on DR ily assess glucose-lowering and cardiovascular outcomes,
ranged anywhere from significantly protective [99, 101], to and for decades, glycemic control has been the key outcome
non-significantly protective [20, 98], non-significantly harm- measure for clinical efficacy. In fact, it was not until the
ful [97], and significantly harmful [96]. Such inconsistent Guidance for Industry recommendations of 2008 that the
findings may reflect fundamental differences in the retinal FDA requested the incidence of major cardiovascular events
safety profiles of individual drugs in the DPP4-I category, to be considered in the prospective evaluation of every new
which remain unknown. anti-diabetic [111, 112, 122]. The updated FDA’s Guid-
A subset of the aforementioned retrospective studies sug- ance for Industry document is currently available online
gests a more favorable DR profile for DPP-4Is, when com- for review [123, 124]. However, the inconsistent evaluation
pared to SUs in particular [20, 96–99]. These findings can of dedicated DR outcomes in trials investigating new anti-
be correlated with similar data on GLP-1RAs and SGLT-2Is; diabetics has not been addressed by this update, despite the
both of which have also been associated with lower DR com- fact that microvascular outcomes are implicated in both the
plication rates versus SUs, in three meta-analyses [18, 22, safety and the effectiveness of those drugs. At this point,
84]. Given the rapidly expanding industry that targets T2D, retina specialists could come forward to advocate the sys-
a longitudinal study co-evaluating the retinal outcomes of tematic assessment of DR status in all such trials, especially
older versus new anti-diabetics would be strongly recom- with the non-invasive imaging options and telemedicine
mended at this point.
13
platforms available nowadays, which could make DR screen- References

ing and monitoring more convenient than ever.
Furthermore, it is important to highlight two limitations 1. International Diabetes Federation (2021) IDF Diabetes Atlas,
which were not restricted to the RCTs discussed in our review. 10th edn. Brussels, Belgium
2. GBD (2019) Blindness and Vision Impairment Collaborators
In certain studies, it was unclear whether DME was evaluated (2021) Causes of blindness and vision impairment in 2020 and
as an integral part of the DR continuum or as a distinct clinical trends over 30 years, and prevalence of avoidable blindness in
entity. Alarmingly, the latter may represent a frequent and per- relation to VISION 2020: the Right to Sight: an analysis for the
sistent misconception in the DR literature. Nevertheless, DME Global Burden of Disease Study. Lancet Glob Health 9:144–160
3. UK Prospective Diabetes Study (UKPDS) Group (1998) Inten-
should be consistently addressed as a DR manifestation and sive blood-glucose control with sulphonylureas or insulin com-
preferably assessed alongside T2D duration, DR history, and pared with conventional treatment and risk of complications in
ophthalmologic signs recorded during comprehensive retinal patients with type 2 diabetes (UKPDS 33). Lancet Lond Engl
exams. A similar approach should be applied to any alterations 352:837–853
4. Stratton IM, Adler AI, Neil HA et al (2000) Association of gly-
in retinal morphology ensuing secondarily to T2D. caemia with macrovascular and microvascular complications of
Meanwhile, distinguishing the retina-specific effects that type 2 diabetes (UKPDS 35): prospective observational study.
can be attributed to each drug’s pharmacology and biochem- BMJ 321:405–412
istry, from secondary retinal outcomes associated with the 5. ACCORD Study Group, ACCORD Eye Study Group, Chew EY
et al (2010) Effects of medical therapies on retinopathy progres-
systemic glucose-lowering dynamics, presents an additional sion in type 2 diabetes. N Engl J Med 363:233–244
challenge. Differences in the magnitude, duration, and consist- 6. The Diabetes Control and Complications Trial Research Group
ency of glycemic control, which may vary among anti-diabetic (1998) Early Worsening of Diabetic Retinopathy in the Dia-
medications, add to the complexity of this problem. Unless betes Control and Complications Trial. Arch Ophthalmol
116(7):874–886
appropriately addressed by the study design, these parameters 7. Aiello LP, DCCT/EDIC Research Group (2014) Diabetic retin-
may introduce a considerable degree of bias. Notably, numer- opathy and other ocular findings in the diabetes control and
ous articles in our review did not control for plasma glucose complications trial/epidemiology of diabetes interventions and
levels across treatment and control groups. In this setting, and complications study. Diabetes Care 37:17–23
8. Hainsworth DP, Bebu I, Aiello LP et al (2019) Risk factors for
to enhance the clinical relevance of their results, future studies retinopathy in type 1 diabetes: the DCCT/EDIC study. Diabetes
should contemplate strategies to control for variability in the Care 42:875–882
glycemic effects among different anti-diabetic classes or even 9. Diabetes Control and Complications Trial (DCCT)/Epidemiology
among medications within the same category. of Diabetes Interventions and Complications (EDIC) Research
Group, Lachin JM, White NH et al (2015) Effect of intensive
In conclusion, despite their documented benefits in glycemic diabetes therapy on the progression of diabetic retinopathy in
control and cardiovascular risk, the effects of newer-generation patients with type 1 diabetes: 18 years of follow-up in the DCCT/
anti-diabetics on DR remain to be fully elucidated. In fact, they EDIC. Diabetes 64:631–642
may not always be beneficial, as indicated by the observed DR 10. American Diabetes Association Professional Practice Commit-
tee (2022) 9. Pharmacologic approaches to glycemic treatment:
worsening within the first year of semaglutide use, and the posi- standards of medical care in diabetes—2022. Diabetes Care
tive correlation between TZDs and DME. Randomized pro- 45:S125–S143
spective data are currently scant on dedicated DR endpoints. 11. Saw M, Wong VW, Liew G (2019) New anti-hyperglycaemic
Future RCTs would require systematic evaluation of at least agents for type 2 diabetes and their effects on diabetic retinopa-
thy. Eye 33:1842–1851
one endpoint specific to DR and longer follow-up duration to 12. El Mouhayyar C, Riachy R, Khalil AB et al (2020) SGLT2
better address this topic. Patients started on semaglutide may inhibitors, GLP-1 agonists, and DPP-4 inhibitors in diabetes
require closer monitoring especially within the first 12 to 18 and microvascular complications: a review. Int J Endocrinol
months of use. Elderly and patients with advanced kidney dis- 2020:1–11
13. Moher D, Liberati A, Tetzlaff J et al (2009) Preferred reporting
ease placed on SGLT-2Is should be aware that there may be a items for systematic reviews and meta-analyses: the PRISMA
slightly increased risk for retinal vein occlusion. statement. BMJ 339:b2535–b2535
14. Petticrew M, Roberts H (2006) Systematic reviews in the social
sciences. Blackwell Publishing Ltd, Oxford, UK
Funding This study was funded by the John S. Latsis Public Benefit 15. Jarrett RJ (1986) Duration of non-insulin-dependent diabetes
Foundation, the Bodossaki Foundation, and the American Macular and development of retinopathy: analysis of possible risk fac-
Degeneration Foundation (AMDF). tors. Diabet Med 3:261–263
16. Harris MI, Klein R, Welborn TA et al (1992) Onset of NIDDM
Declarations occurs at Least 4–7 yr Before Clinical Diagnosis. Diabetes Care
15:815–819
Ethical approval This article does not contain any studies with human 17. Porta M, Curletto G, Cipullo D et al (2014) Estimating the delay
participants or animals performed by any of the authors. between onset and diagnosis of type 2 diabetes from the time
course of retinopathy prevalence. Diabetes Care 37:1668–1674
18. Tang H, Li G, Zhao Y et al (2018) Comparisons of diabetic retin-
Conflict of interest The authors declare no competing interests. opathy events associated with glucose-lowering drugs in patients
13
with type 2 diabetes mellitus: a network meta-analysis. Diabetes 37. Tang H, Shi W, Fu S et al (2018) Pioglitazone and bladder can-
Obes Metab 20:1262–1279 cer risk: a systematic review and meta-analysis. Cancer Med
19. Tsapas A, Avgerinos I, Karagiannis T et al (2020) Comparative 7:1070–1080
effectiveness of glucose-lowering drugs for type 2 diabetes. Ann 38. Erdmann E, Charbonnel B, Wilcox RG et al (2007) Pioglitazone
Intern Med 173:278–286 use and heart failure in patients with type 2 diabetes and preexist-
20. Wang T, Hong J-L, Gower EW et al (2018) Incretin-based thera- ing cardiovascular disease. Diabetes Care 30:2773–2778
pies and diabetic retinopathy: real-world evidence in older U.S. 39. Fong DS, Contreras R (2009) Glitazone use associated with dia-
adults. Diabetes Care 41:1998–2009 betic macular edema. Am J Ophthalmol 147:583–586
21. OCEBM Levels of Evidence Working Group (2011) The Oxford 40. Idris I, Warren G, Donnelly R (2012) Association between
levels of evidence 2. Oxford Centre for Evidence-Based Medi- thiazolidinedione treatment and risk of macular edema among
cine. https://www.cebm.ox.ac.uk/resources/levels-of-evidence/ patients with type 2 diabetes. Arch Intern Med 172:1005–1012
ocebm-levels-of-evidence. Accessed 13 Jun 2023 41. Motola D, Piccinni C, Biagi C et al (2012) Cardiovascular, ocular
22. Zhou B, Shi Y, Fu R et al (2022) Relationship between SGLT- and bone adverse reactions associated with thiazolidinediones: a
2i and ocular diseases in patients with type 2 diabetes mel- disproportionality analysis of the US FDA adverse event report-
litus: a meta-analysis of randomized controlled trials. Front ing system database. Drug Saf 35:315–323
Endocrinol 13:907340. https://d oi.o rg/1 0.3 389/fendo.2 022. 42. Shen LQ, Child A, Weber GM et al (2008) Rosiglitazone and
907340 delayed onset of proliferative diabetic retinopathy. Arch Oph-
23. U.S. Food and Drug Administration (2022) FDA approves thalmol 126:793–799
novel, dual-targeted treatment for type 2 diabetes. https:// 43. Vallabhajosyula S, Vallabhajosyula S, Vallabhajosyula S et al
www.f da.g ov/n ews-events/p ress-a nnoun cemen ts/f da-a ppro (2016) Development of vascular complications and bladder car-
ves-novel-dual-t argeted-treatment-type-2-diabetes. Accessed cinoma in diabetics using pioglitazone: a five-year Indian review.
10 May 2023 Med J Armed Forces India 72:253–257
24. Dirir AM, Daou M, Yousef AF et al (2021) A review of alpha-glucosi- 44. Shani M, Feldman L, Kaiserman I et al (2011) Is Rosiglitazone
dase inhibitors from plants as potential candidates for the treatment use associated with an increase in intensive eye treatment in
of type-2 diabetes. Springer Netherlands 21(4):1049–1079 diabetic patients? A community based study. Eur J Gen Pract
25. Quianzon CCL, Cheikh IE (2012) History of current non-insulin 17:205–209
medications for diabetes mellitus. J Community Hosp Intern Med 45. Ambrosius WT, Danis RP, Goff DC et al (2010) Lack of asso-
Perspect 2(3). https://doi.org/10.3402/jchimp.v2i3.19081 ciation between thiazolidinediones and macular edema in type 2
26. Holman RR, Coleman RL, Chan JCN et al (2017) Effects of acar- diabetes. Arch Ophthalmol 128:312
bose on cardiovascular and diabetes outcomes in patients with 46. Gower EW, Lovato JF, Ambrosius WT et al (2017) Lack of longi-
coronary heart disease and impaired glucose tolerance (ACE): a tudinal association between thiazolidinediones and incidence and
randomised, double-blind, placebo-controlled trial. Lancet Dia- progression of diabetic eye disease: the ACCORD EYE study.
betes Endocrinol 5:877–886 Am J Ophthalmol 187:138–147
27. American Diabetes Association Professional Practice Commit- 47. Chew EY, Ambrosius WT, Howard LT et al (2007) Rationale,
tee (2022) 8. Obesity and weight management for the prevention design, and methods of the Action to Control Cardiovascular
and treatment of type 2 diabetes: standards of medical care in Risk in Diabetes Eye Study (ACCORD-EYE). Am J Cardiol
diabetes—2022. Diabetes Care 45:S113–S124 99:103–111
28. Chiasson J-L, Josse RG, Gomis R et al (2002) Acarbose for 48. Clark A, Charge SBP, Badman MK et al (1996) Islet amyloid
prevention of type 2 diabetes mellitus: the STOP-NIDDM ran- polypeptide: actions and role in the pathogenesis of diabetes.
domised trial. Lancet Lond Engl 359:2072–2077 Biochem Soc Trans 24:594–599
29. Chiasson J-L, Josse RG, Gomis R et al (2003) Acarbose treat- 49. Edelman S, Maier H, Wilhelm K (2008) Pramlintide in the
ment and the risk of cardiovascular disease and hypertension treatment of diabetes mellitus. BioDrugs 22:375–386
in patients with impaired glucose tolerance: the STOP-NIDDM 50. Westermark P, Andersson A, Westermark GT (2011) Islet amy-
trial. JAMA 290:486–494 loid polypeptide, islet amyloid, and diabetes mellitus. Physiol
30. Segal P, Eliahou HE, Petzinna D et al (2005) Long-term efficacy Rev 91:795–826
and tolerability of acarbose treatment in patients with type 2 51. Ratner RE, Want LL, Fineman MS et al (2002) Adjunctive
diabetes mellitus. Clin Drug Investig 25:589–595 therapy with the amylin analogue pramlintide leads to a com-
31. Stumvoll M, Häring H-U (2002) Glitazones: clinical effects and bined improvement in glycemic and weight control in insulin-
molecular mechanisms. Ann Med 34:217–224 treated subjects with type 2 diabetes. Diabetes Technol Ther
32. Lalloyer F, Staels B (2010) Fibrates, glitazones, and peroxisome 4:51–61
proliferator-activated receptors. Arterioscler Thromb Vasc Biol 52. Spreckley E, Murphy KG (2015) The L-cell in nutritional sensing
30:894–899 and the regulation of appetite. Front Nutr 2:1–17
33. Gale EAM (2001) Lessons from the glitazones: a story of drug 53. Nauck MA, Meier JJ, Bochum R (2016) The incretin effect in
development. Lancet 357:1870–1875 healthy individuals and those with type 2 diabetes : physiology,
34. Nanjan MJ, Mohammed M, Prashantha Kumar BR et al (2018) pathophysiology , and response to therapeutic interventions.
Thiazolidinediones as antidiabetic agents: a critical review. LANCET Diabetes Endocrinol 4:525–536
Bioorganic Chem 77:548–567 54. Neumiller JJ (2009) Differential chemistry (structure), mecha-
35. Home PD, Pocock SJ, Beck-Nielsen H et al (2009) Rosiglitazone nism of action, and pharmacology of GLP-1 receptor agonists
evaluated for cardiovascular outcomes in oral agent combina- and DPP-4 inhibitors. J Am Pharm Assoc 49:S16–S29
tion therapy for type 2 diabetes (RECORD): a multicentre, ran- 55. Min T, Bain SC (2021) The role of tirzepatide, dual GIP and
domised, open-label trial. The Lancet 373:2125–2135 GLP-1 receptor agonist, in the management of type 2 diabetes:
36. Mahaffey KW, Hafley G, Dickerson S et al (2013) Results of a the SURPASS clinical trials. Diabetes Ther 12:143–157
reevaluation of cardiovascular outcomes in the RECORD trial. 56. Del Prato S, Kahn SE, Pavo I et al (2021) Tirzepatide versus
Am Heart J 166:240–249 insulin glargine in type 2 diabetes and increased cardiovascular
13
risk (SURPASS-4): a randomised, open-label, parallel-group, 73. Douros A, Filion KB, Yin H et al (2018) Glucagon-like peptide
multicentre, phase 3 trial. Lancet Lond Engl 398:1811–1824 1 receptor agonists and the risk of incident diabetic retinopathy.
57. Gerstein HC, Colhoun HM, Dagenais GR et al (2019) Dulaglu- Diabetes Care 41:2330–2338
tide and cardiovascular outcomes in type 2 diabetes (REWIND): 74. Fadini GP, Sarangdhar M, Avogaro A (2018) Glucagon-like pep-
a double-blind, randomised placebo-controlled trial. The Lancet tide-1 receptor agonists are not associated with retinal adverse
394:121–130 events in the FDA adverse event reporting system. BMJ Open
58. Hernandez AF, Green JB, Janmohamed S et al (2018) Albiglutide Diabetes Res Care 6:1–8
and cardiovascular outcomes in patients with type 2 diabetes 75. Ueda P, Pasternak B, Eliasson B et al (2019) Glucagon-like pep-
and cardiovascular disease (Harmony Outcomes): a double-blind, tide 1 receptor agonists and risk of diabetic retinopathy complica-
randomised placebo-controlled trial. The Lancet 392:1519–1529 tions: cohort study in nationwide registers from two countries.
59. Holman RR, Bethel MA, Mentz RJ et al (2017) Effects of once- Diabetes Care 42:92–94
weekly exenatide on cardiovascular outcomes in type 2 diabetes. 76. Varadhan L, Humphreys T, Hariman C et al (2011) GLP-1 ago-
N Engl J Med 377:1228–1239 nist treatment: implications for diabetic retinopathy screening.
60. Husain M, Birkenfeld AL, Donsmark M et al (2019) Oral sema- Diabetes Res Clin Pract 94:68–71
glutide and cardiovascular outcomes in patients with type 2 dia- 77. Varadhan L, Humphreys T, Walker AB et al (2014) The impact of
betes. N Engl J Med 381:841–851 improved glycaemic control with GLP-1 receptor agonist therapy
61. Marso SP, Bain SC, Consoli A et al (2016) Semaglutide and on diabetic retinopathy. Diabetes Res Clin Pract 103:37–39
cardiovascular outcomes in patients with type 2 diabetes. N Engl 78. Wang T, Lu W, Tang H et al (2019) Assessing the association
J Med 375:1834–1844 between GLP-1 receptor agonist use and diabetic retinopathy
62. Marso SP, Daniels GH, Brown-Frandsen K et al (2016) Liraglu- through the FDA adverse event reporting system. Diabetes Care
tide and cardiovascular outcomes in type 2 diabetes. N Engl J 42:21–23
Med 375:311–322 79. Gaborit B, Julla J-B, Besbes S et al (2020) Glucagon-like pep-
63. Packer M, Anker SD, Butler J et al (2020) Cardiovascular and tide 1 receptor agonists, diabetic retinopathy and angiogenesis:
renal outcomes with empagliflozin in heart failure. N Engl J Med the angiosafe type 2 diabetes study. J Clin Endocrinol Metab
383:1413–1424 105:1–23
64. Pfeffer MA, Claggett B, Diaz R et al (2015) Lixisenatide in 80. Albert SG, Wood EM, Ahir V (2023) Glucagon-like peptide
patients with type 2 diabetes and acute coronary syndrome. N 1-receptor agonists and A1c: good for the heart but less so for
Engl J Med 373:2247–2257 the eyes? Diabetes Metab Syndr 17:102696. https://doi.org/10.
65. Sattar N, McGuire DK, Pavo I et al (2022) Tirzepatide cardiovas- 1016/j.dsx.2022.102696
cular event risk assessment: a pre-specified meta-analysis. Nat 81. Alexander JT, Staab EM, Wan W et al (2022) The longer-term
Med 28:591–598 benefits and harms of glucagon-like peptide-1 receptor ago-
66. U.S. Food and Drug Administration (2021) FDA approves new nists: a systematic review and meta-analysis. J Gen Intern Med
drug treatment for chronic weight management, First Since 37:415–438
2014. https://www.fda.gov/news-events/press-announcements/ 82. Andreadis P, Karagiannis T, Malandris K et al (2018) Semaglu-
fda-approves-new-drug-treatment-chronic-weight-management- tide for type 2 diabetes mellitus: a systematic review and meta-
first-2014. Accessed 13 Jun 2023 analysis. Diabetes Obes Metab 20:2255–2263
67. U.S. Food and Drug Administration (2021) FDA approves weight 83. Avgerinos I, Karagiannis T, Malandris K et al (2019) Glucagon-
management drug for patients aged 12 and older. FDA, In https:// like peptide-1 receptor agonists and microvascular outcomes in
www.fda.gov/drugs/news-events-human-drugs/fda-approves- type 2 diabetes: a systematic review and meta-analysis. Diabetes
weight-management-drug-patients-aged-12-and-older. Accessed Obes Metab 21:188–193
13 Jun 2023 84. Dicembrini I, Nreu B, Scatena A et al (2017) Microvascular
68. Vilsbøll T, Bain SC, Leiter LA et al (2018) Semaglutide, reduc- effects of glucagon-like peptide-1 receptor agonists in type 2
tion in glycated haemoglobin and the risk of diabetic retinopathy. diabetes: a meta-analysis of randomized controlled trials. Acta
Diabetes Obes Metab 20:889–897 Diabetol 54:933–941
69. Kaku K, Rasmussen MF, Nishida T et al (2011) Fifty-two-week, 85. Wang F, Mao Y, Wang H et al (2022) Semaglutide and diabetic
randomized, multicenter trial to compare the safety and efficacy retinopathy risk in patients with type 2 diabetes mellitus: a meta-
of the novel glucagon-like peptide-1 analog liraglutide vs glib- analysis of randomized controlled trials. Clin Drug Investig
enclamide in patients with type2 diabetes. J Diabetes Investig 42:17–28
2:441–447 86. Wei J, Yang B, Wang R et al (2022) Risk of stroke and retinopa-
70. Rosenstock J, Hanefeld M, Shamanna P et al (2014) Beneficial thy during GLP-1 receptor agonist cardiovascular outcome tri-
effects of once-daily lixisenatide on overall and postprandial als: an eight RCTs meta-analysis. Front Endocrinol 13:1007980.
glycemic levels without significant excess of hypoglycemia in https://doi.org/10.3389/fendo.2022.1007980
type 2 diabetes inadequately controlled on a sulfonylurea with 87. Zhang X, Shao F, Zhu L et al (2018) Cardiovascular and micro-
or without metformin (GetGoal-S). J Diabetes Complications vascular outcomes of glucagon-like peptide-1 receptor agonists
28:386–392 in type 2 diabetes: a meta-analysis of randomized controlled
71. Seino Y, Kaneko S, Fukuda S et al (2016) Combination therapy cardiovascular outcome trials with trial sequential analy-
with liraglutide and insulin in Japanese patients with type 2 dia- sis. BMC Pharmacol Toxicol 19:58. https://doi.org/10.1186/
betes: a 36-week, randomized, double-blind, parallel-group trial. s40360-018-0246-x
J Diabetes Investig 7:565–573 88. Yoshida Y, Joshi P, Barri S et al (2022) Progression of retin-
72. Bethel MA, Mentz RJ, Merrill P et al (2020) Microvascular and opathy with glucagon-like peptide-1 receptor agonists with car-
cardiovascular outcomes according to renal function in patients diovascular benefits in type 2 diabetes-a systematic review and
treated with once-weekly exenatide: insights from the EXSCEL meta-analysis. J Diabetes Complications 36:108255. https://doi.
trial. Diabetes Care 43:446–452 org/10.1016/j.jdiacomp.2022.108255
13
89. Deacon CF (2019) Physiology and pharmacology of DPP-4 in 106. Cannon CP, Pratley R, Dagogo-Jack S et al (2020) Cardiovascu-
glucose homeostasis and the treatment of type 2 diabetes. Front lar outcomes with ertugliflozin in type 2 diabetes. N Engl J Med
Endocrinol 10:80. https://doi.org/10.3389/fendo.2019.00080 383:1425–1435
90. Gallwitz B (2019) Clinical use of DPP-4 inhibitors. Front Endo- 107. Heerspink HJL, Stefánsson BV, Correa-Rotter R et al (2020)
crinol 10:1–10 Dapagliflozin in patients with chronic kidney disease. N Engl J
91. Scirica BM, Bhatt DL, Braunwald E et al (2013) Saxagliptin Med 383:1436–1446
and cardiovascular outcomes in patients with type 2 diabetes 108. McMurray JJV, Solomon SD, Inzucchi SE et al (2019) Dapagli-
mellitus. N Engl J Med 369:1317–1326 flozin in patients with heart failure and reduced ejection fraction.
92. White WB, Cannon CP, Heller SR et al (2013) Alogliptin after N Engl J Med 381:1995–2008
acute coronary syndrome in patients with type 2 diabetes. N Engl 109. Neal B, Perkovic V, Mahaffey KW et al (2017) Canagliflozin and
J Med 369:1327–1335 cardiovascular and renal events in type 2 diabetes. N Engl J Med
93. Green JB, Bethel MA, Armstrong PW et al (2015) Effect of sit- 377:644–657
agliptin on cardiovascular outcomes in type 2 diabetes. N Engl J 110. Perkovic V, Jardine MJ, Neal B et al (2019) Canagliflozin and
Med 373:232–242 renal outcomes in type 2 diabetes and nephropathy. N Engl J Med
94. Rosenstock J, Kahn SE, Johansen OE et al (2019) Effect of lina- 380:2295–2306
gliptin vs glimepiride on major adverse cardiovascular outcomes 111. Sharma A, Pagidipati NJ, Califf RM et al (2020) Impact of
in patients with type 2 diabetes: the CAROLINA randomized regulatory guidance on evaluating cardiovascular risk of new
clinical trial. JAMA - J Am Med Assoc 322:1155–1166 glucose-lowering therapies to treat type 2 diabetes mellitus: les-
95. Rosenstock J, Perkovic V, Johansen OE et al (2019) Effect of sons learned and future directions. Circulation 141(10):843–862
linagliptin vs placebo on major cardiovascular events in adults 112. Turner JR (2021) FDA’s New guidance for industry addressing evalu-
with type 2 diabetes and high cardiovascular and renal risk: the ation of the safety of new drugs for improving glycemic control: a
CARMELINA randomized clinical trial. JAMA - J Am Med case study in regulatory science. Ther Innov Regul Sci 55:1–5
Assoc 321:69–79 113. Wanner C, Inzucchi SE, Lachin JM et al (2016) Empagliflozin
96. Kang EYC, Kang C, Wu WC et al (2021) Association between and progression of kidney disease in type 2 diabetes. N Engl J
add-on dipeptidyl peptidase-4 inhibitor therapy and diabetic Med 375:323–334
retinopathy progression. J Clin Med 10(13):2871. https://doi. 114. Wiviott SD, Raz I, Bonaca MP et al (2019) Dapagliflozin and
org/10.3390/jcm10132871 cardiovascular outcomes in type 2 diabetes. N Engl J Med
97. Kim NH, Choi J, Choi KM et al (2018) Dipeptidyl peptidase-4 380:347–357
inhibitor use and risk of diabetic retinopathy: a population-based 115. Inzucchi SE, Wanner C, Hehnke U et al (2019) Retinopathy
study. Diabetes Metab 44:361–367 outcomes with empagliflozin versus placebo in the EMPA-REG
98. Chung Y, Ha KH, Kim HC et al (2019) Dipeptidyl peptidase-4 OUTCOME trial. Diabetes Care 42:53–55
inhibitors versus other antidiabetic drugs added to metformin 116. Chung YR, Ha KH, Lee K et al (2019) Effects of sodium-glucose
monotherapy in diabetic retinopathy progression: a real world- cotransporter-2 inhibitors and dipeptidyl peptidase-4 inhibitors
based cohort study. Diabetes Metab J 43:640. https://doi.org/10. on diabetic retinopathy and its progression: a real-world Korean
4093/dmj.2018.0137 study. PLoS ONE 14:1–15
99. Kolaczynski WM, Hankins M, Ong SH et al (2016) Microvas- 117. Dziuba J, Alperin P, Racketa J et al (2014) Modeling effects of
cular outcomes in patients with type 2 diabetes treated with SGLT-2 inhibitor dapagliflozin treatment versus standard dia-
vildagliptin vs. sulfonylurea: a retrospective study using German betes therapy on cardiovascular and microvascular outcomes.
electronic medical records. Diabetes Ther 7:483–496 Diabetes Obes Metab 16:628–635
100. Deng Y, Polley EC, Wallach JD et al (2022) Emulating the 118. Su Y-C, Shao S-C, Lai EC-C et al (2021) Risk of diabetic macu-
GRADE trial using real world data: retrospective comparative lar oedema with sodium-glucose cotransporter-2 inhibitors in
effectiveness study. BMJ 379:e070717. https://doi.org/10.1136/ type 2 diabetes patients: a multi-institutional cohort study in
bmj-2022-070717 Taiwan. Diabetes Obes Metab 23:2067–2076
101. Simmons K, Singh P, Borkar DS et al (2022) The effect 119. Li C, Zhou Z, Neuen BL et al (2021) Sodium-glucose co-trans-
of select systemic medications on outcomes in diabetics porter-2 inhibition and ocular outcomes in patients with type 2
with central retinal vein occlusion. Ther Adv Ophthalmol diabetes: a systematic review and meta-analysis. Diabetes Obes
14:25158414211063076. https://doi.org/10.1177/2515841421 Metab 23:252–257
1063076 120. Ma Y, Lin C, Cai X et al (2022) The association between the
102. Ueki K, Tanizawa Y, Nakamura J et al (2021) Long-term safety use of sodium glucose cotransporter 2 inhibitor and the risk of
and efficacy of alogliptin, a DPP-4 inhibitor, in patients with type diabetic retinopathy and other eye disorders: a systematic review
2 diabetes: a 3-year prospective, controlled, observational study and meta-analysis. Expert Rev Clin Pharmacol 15:877–886
(J-BRAND Registry). BMJ Open Diabetes Res Care 9:e001787. 121. Lee M-K, Kim B, Han K et al (2021) Sodium–glucose cotrans-
https://doi.org/10.1136/bmjdrc-2020-001787 porter 2 inhibitors and risk of retinal vein occlusion among
103. Choi C (2016) Sodium-Glucose Cotransporter 2 (SGLT2) Inhibi- patients with type 2 diabetes: a propensity score–matched cohort
tors from natural products: discovery of next-generation antihy- study. Diabetes Care 44:2419–2426
perglycemic agents. Molecules 2:1136. https://doi.org/10.3390/ 122. U.S. Food and Drug Administration (2008) Guidance for industry
molecules21091136 on diabetes mellitus-evaluating cardiovascular risk in new anti-
104. Vick H, Diedrich DF, Baumann K (1973) Reevaluation inhibi- diabetic therapies to treat type 2 diabetes. Fed. Regist, In https://
tion of renal by phlorizin tubular analogs glucose transport. Am www.f edera lregi ster.g ov/d ocume nts/2 008/1 2/1 9/E 8-3 0086/g uida
J Physiol 224:552–557 nce-f or-i ndust ry-o n-d iabet es-m
ellit us-e valua ting-c ardio vascu lar-
105. Saisho Y (2020) SGLT2 inhibitors: the star in the treatment of risk-in-new-antidiabetic. Accessed 13 Jun 2023
type 2 diabetes? Diseases 8:14. https://doi.org/10.3390/disea 123. U.S. Food and Drug Administration (2020) Type 2 diabetes
ses8020014 mellitus: evaluating the safety of new drugs for improving
13
glycemic control guidance for industry. US Food Drug Adm, Publisher’s note Springer Nature remains neutral with regard to
In https://www.fda.gov/regulatory-information/search-fda-guida jurisdictional claims in published maps and institutional affiliations.
nce-documents/type-2-diabetes-mellitus-evaluating-safety-new-
drugs-improving-glycemic-control-guidance-industry. Accessed Springer Nature or its licensor (e.g. a society or other partner) holds
13 Jun 2023 exclusive rights to this article under a publishing agreement with the
124. U.S. Food and Drug Administration (2020) Type 2 diabetes mel- author(s) or other rightsholder(s); author self-archiving of the accepted
litus: evaluating the safety of new drugs for improving glycemic manuscript version of this article is solely governed by the terms of
control; Draft Guidance for Industry; Availability; Establishment such publishing agreement and applicable law.
of Public Docket; Request for Comments. https://www.regul
ations.gov/docket/FDA-2019-N-6084/comments. Accessed 13
Jun 2023
13

Review Anti Diabetic Drugs

Uploaded by

Copyright:

Available Formats

Review Anti Diabetic Drugs

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Review Anti Diabetic Drugs

Uploaded by

Copyright:

Available Formats

Graefe's Archive for Clinical and Experimental Ophthalmology

Effects of newer‑generation anti‑diabetics on diabetic retinopathy:

Received: 20 June 2023 / Revised: 30 August 2023 / Accepted: 7 September 2023

Dimitrios P. Ntentakis and Victor San Martin Carvalho Correa

Introduction demonstrated by the ACCORD study, in 2856 T2D partici-

RCTs with systematic N/A N/A N/A 3 1 1

5385 Control group: 3.2 ence) and insulin agulation/vitrectomy):

RCTs systematically assessing at least one retinal endpoint

vitreal therapy and

tide: ↑ DR risk with

RCTs systematically assessing at least one retinal endpoint

RCTs systematically assessing at least one retinal endpoint

Ongoing trials Discussion

platforms available nowadays, which could make DR screen- References

You might also like