Screening for diabetic retinopathy with fluorescein angiography in patients with type 1 diabetes from adolescence to adult life. A retrospective study of the past 30 years of clinical practice in a tertiary Belgian centre

Abstract Background The aim of the present study was to describe the prevalence and progression of DR diagnosed by fluorescein angiography (FA) in patients with type 1 diabetes (T1D) during a 30‐year follow‐up, and the relationship with glycated haemoglobin (HbA1c). Materials and methods We included 4325 FA reports representing 851 patients with T1D with a mean age at diagnosis of 10.4 years (range: 0.0–49.9) and followed between 1986 and 2015. Clinical characteristics of the population were collected from patients’ files. The HbA1c level was measured within a maximum period of ±1 year from the date of FA. Descriptive statistics were realized to study prevalence and progression of DR. Results At diagnosis of incipient abnormalities, mean age was 22.8 years (range 13.7–46.9) and mean diabetes duration was 13 years (range: 4.3–29.6). Lesions requiring treatment were observed in 5.9% of the patients at a mean age of 32.4 years (range: 30.4–34.3) and a mean diabetes duration of 23.8 years (range: 19.4–28.1). On average, it took 12.9 years (range: 12.2–13.5) to progress from an incipient abnormality to a lesion requiring treatment. Mean HbA1c ± SD was 7.8 ± 1.5% over a period of 30 years. Conclusions While it could have been expected to observe a higher prevalence of DR, our study described by far the lowest results of prevalence comparing to similar studies, probably due to a good average HbA1c over 30 years.

A lot of epidemiological studies have described DR and its relationships to various risk factors. [3][4][5][6][7][8] Only a few have been realized to study the prevalence of DR by fluorescein angiography (FA). 4 In 2018, the International Society for Pediatric and Adolescent Diabetes (ISPAD) guidelines recommended that screening for retinopathy should start from age 11 years and after 2-5 years of diabetes duration. Minimal assessment for retinopathy should be performed by ophthalmoscopy through dilated pupils by an experienced observer. The frequency of retinopathy screening should occur annually or more frequently in case of high-risk features for visual loss. 9 The retinal FA allows detection of early abnormalities that are undetectable by regular ophthalmoscopy. It enables the study of the vascular walls and the perturbations of the dynamic circulation. In our experience, compared with regular ophthalmoscopy, FA doubles the diagnosis of incipient retinopathy. 10 FA allows the detection of retinal changes approximately 4 years earlier than ophthalmoscopy. 11 A recent comparison of digital colour fundus imaging and fluorescein angiographic findings for the early detection of DR in young patients with T1D confirmed the superiority of FA. 12 The aim of this retrospective 30-year study was to describe the prevalence and progression of DR detected by FA in patients with T1D followed at the Diabetology Clinic of the University Children's Hospital Queen Fabiola and the Department of Ophthalmology of the Brugmann University Hospital in Brussels.

| Reports
The database is a gathering of DR screening reports stored at the were excluded because of a lack of FA in duration groups of more than 30 years. In total, 4325 FA reports were included ( Figure 1).
Through three decades, insulin treatments (but not the paediatric diabetologist, head of the team, HD), angiographic material and retina specialists protocoling FA reports, have changed.
Currently, a mix of one third phenylephrine and two thirds tropicamide is used to dilate pupils.
Detailed methodology for FA has been previously describe. 4,8,13 Over time, there has been a technical evolution in angiography, from cameras using silver print (Zeiss, Canon, Topcon) to digital print acquisition as by Scanning Laser Ophthalmoscopy (SLO Heidelberg).
These differences were not reported in FA protocols.
Macular and peripheral areas were described separately for each eye. While constituting the database, it has been noticed that different ophthalmologists wrote reports following different classifications which have evolved over time. Therefore, the decision was made to transfer observed lesions into three categories approved by all ophthalmologists.  Table S1, we reported a condensed version of the scale. 14 If more than one FA was realized in the same year, only the one including the higher number of lesions was recorded. The 4422 FA reports were classified into groups of years of DM duration.

| Subjects
In total, 4325 FA reports (a mean of 5.1 reports per patient) representing 851 patients with T1D (371 females, 480 males), followed over the past 30 years, compose the database.

| Clinical and biological assessments
A retrospective review of these 851 patients' files was performed.

| Prevalence of diabetic retinopathy
The first aim of this study was to describe the prevalence of abnormalities (categories: 0, 1 and 2, graded according to the DRSS Classification), in macular and peripheral areas, throughout years of DM duration. The different required ophthalmic treatments were described, as well as side effects due to fluorescein injection. In order to complete this first aim, data of the 851 patients were used.

| Progression of diabetic retinopathy
The second aim of this work was to describe the progression of initial abnormalities (categories 1 and 2) and time of progression.
To study this, three new inclusion criteria were added: having at least one normal FA at the beginning of the screening; having at least 10 years of DM duration; having at most a 2-year period between the first abnormal FA and the last one known as normal.

| Data analysis
The statistical analysis was performed using IBM SPSS® Statistics 23. Descriptive statistics were calculated for baseline characteristics of the subjects (mean, SD, median, minimum, and maximum), prevalence of retinopathy for each year of DM duration group and progression of diabetic lesions.

| RE SULTS
A total of 851 patients with T1D underwent 4325 DR screening examinations by FA between 1986 and 2015. Baseline characteristics of these subjects appear in Table 1. Age at DM diagnosis was 10.4 ± 6.6 years (range: 0-49.9). DM duration at the end of the F I G U R E 1 Construction of the database which is a gathering of DR screening reports from 1986 to 2015. Patient data were used differently to reach the first and the second aim of this study. 1 These 198 FA reports were excluded because they constituted years of DM duration groups containing less than 20 FA reports by group. 2 As the aim is to specify the moment of appearance of lesions, it was decided to specially add this third criterion 4523 FA reports were classified into groups of years of DM duration (from 1 to 34 years).
198 reports were excluded to preserve statistical validity 1 . Finally, 4325 FA reports representing 851 type 1 diabetic patients were available in order to reach 2 different aims: Description of the prevalence of diabetic retinopathy in the different groups of years of DM duration.  In total, 274 type 1 diabetic patients matched these criteria and were recruited to complete this second aim. 1 = These 198 FA reports were excluded because they constituted years of DM duration groups containing less than 20 FA reports by group. 2 = As the aim is to specify the moment of appearance of lesions, it was decided to specially add this third criterion.  and 9 had mild proliferative DR (grade 61). 15 The treatment mainly consisted in photocoagulation therapy, sometimes associated with non-steroidal anti-inflammatory eye drops, acetazolamide tablets or anti-vascular endothelial growth factor intravitreal injection. Two patients required vitrectomy.

| Prevalence of diabetic retinopathy
The mean HbA1c, as well as the number of FA reports according to DM duration can be found in Table S2.
Considering side effects, 17 patients among 4424 (0.4%) had nauseas. In addition, one of them had reactive urticaria and another one had Quincke's oedema. No death occurred.

| Progression of diabetic retinopathy
Data from 274 patients with T1D were used to achieve the second aim.
Considering a maximum period of 1.5 years between the first abnormal FA and the last one known as normal, age at diagnosis of category On average, it took 12.9 ± 1 years (range 12.2-13.5) to move from an incipient retinal lesion to a lesion requiring treatment (Table 3).

| DISCUSS ION
This study provides a unique overview of long-term clinical practice.
The main feature of this work is the description of prevalence and progression of DR diagnosed by a sensitive method, FA, 8 in a large young population, during a very long follow-up with available data.
Our aim was to use the massive available database to publish a report looking back over the last 30 years, taking stock of what has been achieved in terms of screening.
However, this work has limitations due to a lack of data on risk factors such as blood pressure, body mass index, lipids levels, smoking, puberty and maternity. Also, DR is the only studied complication. Based on these results obtained with a less sensitive method, they suggested that screening for DR should start at 15 or 5 years after DM onset but, thanks to FA, we described earlier lesions starting at 13 years of age and after 4 years of DM.

| Prevalence of diabetic retinopathy
The relatively low prevalence of DR in our study could be due to the degree of glycaemic control (HbA1c 7.8 ± 1.5% (range: 4-18.7)).
Objectively, our study shows that moderate and severe side effects of FA are very rare in youth.

| Progression of diabetic retinopathy
This study described the apparition of the first signs of DR at 13 years of age and after 4 years of DM, approximately at the same period 2018 ISPAD guidelines recommended to start screening. 9 Our results were consistent with a study we published more than 25 years ago. 17 From our point of view, the practical interest of an early diagnosis of DR is to motivate both the patient and the multidisciplinary paediatric diabetology team to obtain the best possible metabolic control to avoid serious and irreversible complications. The HbA1c targets, expectations and goals that diabetes care professionals have for their patients is determinant in metabolic outcomes. 18 Considering the ophthalmological point of view, the optimal period for therapy is often missed when visual impairment occurs. Therefore, early recognition, prevention and treatment of DR should be of major consideration. 19

| Recent breakthroughs in screening
New imaging systems could be useful in screening, in grading, in the treatment and the follow-up of DR. While enhancement of smartphone-based retinal imaging holds great promise to increase the accessibility of retinal screening, 20 it seems to be more useful in DR lesions follow-up then in early diagnosis in children. It is based on wide field mydriatic retinal photographs which has been demonstrated less sensitive than FA in revealing information about retinal vascular pathology after a short disease time. 13 Also, this article showed that identifying minor DR signs at a stage with unaffected visual acuity constitutes an alert for a stricter glycaemic control and more frequent fundus examination.
TA B L E 2 Description of the mean prevalence of diabetic retinopathy found in macular and peripheral areas after 5, 10, 15, 20,

ACK N OWLED G M ENTS
The authors would like to thank Mr William Stanbury for supervising the English language as well as ophthalmology and diabetology teams for technical assistance.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

E TH I C S A PPROVA L S TATE M E NT
This is a retrospective study without new intervention-whether physical, psychological, or chemical-concerning the routine followup of young patients with diabetes.

PE R M I SS I O N TO R E PRO D U CE M ATE R I A L FRO M OTH E R S O U RCE S
None to declare. We don't reproduce any material from other sources.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are personal and come from the patients' medical reports.