Acta Ophthalmologica 2015
ACTA OPHTHALMOLOGICA THESIS http://www.actaophthalmologica.com
Thesis Early risk stratification in pediatric type 1 diabetes
Rebecca Broe Department of Ophthalmology Odense University Hospital Sdr. Boulevard 29, DK-5000 Odense C
aos_93_A102_Acknowlegements_Layout 1 03-02-2015 07:34 Page 1
ACTA OPHTHALMOLOGICA 2015
Acknowledgements The work presented in this PhD thesis was carried out at the Department of Ophthalmology, Odense University Hospital, Denmark, in 2010-2014. Fortunately, I was not alone on this trip but accompanied by an extended team of colleagues and friends, always willing to help and motivate me. For this I am very grateful. My greatest source of inspiration both in research and how to be a clinician came from the most amazing doctor I have ever met – Anne Katrin SjŒlie. She always had time for me and just a few moments with her could turn the most unmanageable problems into minor bumps on the road. I am deeply grateful for the time we worked together and for all the laughs we shared. You are truly missed. I was fortunate to have Jakob Grauslund take over as main supervisor after Anne Katrin. I am amazed at how much work you do in a day and also convinced you have found the way to stop time. Thank you for fitting me in your busy schedule, for all your great ideas and your continuous support and help with my research. I hope we will keep working together for many years more. I was, furthermore, blessed with a co-supervisor who has a gift when it comes to keeping PhD-students on the right track, Tunde Peto. At several meetings I had ideas and plans all over the place and no overview of my time schedule - something I do not thrive with at all. An hour later I could leave the meeting feeling calm and on top of things – thatÊs just priceless. Also thanks for letting me visit Moorfields. I also had the pleasure of collaborating with two incredibly friendly and helpful pediatricians from Herlev, Birthe Susanne Olsen and Henrik BindesbŒl Mortensen. Without your splendid work in the 1980s and 90s this study would never have been possible. I am very grateful for this opportunity to follow up on these patients and for your enthusiasm, helpfulness and positive spirit throughout my time as a researcher. I had the privilege of visiting the Centre for Eye Research Australia in Melbourne, Australia, where Lauren Hodgson and Ryo Kawasaki trained me in retinal vessel analyzes. I am thankful for their great kindness and helpfulness. A special thanks to Tien Wong for allowing this visit and for great inputs on one of the papers. I wish to thank Kelvin Kamp Mortensen for allowing this study to be conducted at the Department of Ophthalmology at Odense University Hospital. I hope to continue working here for many years on. I would also like to thank Toke Bek and Henrik Lund-Andersen for allowing me to examine some of the participants at the Departments of Ophthalmology in Aarhus and Glostrup. A special thanks goes out to my dear friends and colleagues in the research unit: Anders HŒjslet Vestergaard, who continuously tried to develop type 2 diabetes by consuming massive amounts of chocolate; Kristian Lundberg for your forever positive mind and great sense of humor, even after corresponding with the PhD school on salary issues; Ulrik FrydkjÕr-Olsen, who with great enthusiasm took upon himself to become my personal slave in his spare time and manually digitalized around 1500 slides; Malin Lundberg Rasmussen, who truly beats me when it comes to power shopping; Christina DŒfler Poulsen, who will always lend you a screw driver from her office drawer when needed; and Rasmus SŒgaard Hansen, who takes any task upon him with great enthusiasm and a positive approach. Keep the smiles up, lovely people. Furthermore, I would like to thank Sasikala Thineshkumar for being a lovely friend during these years. It has been a pleasure to talk things over with you and to be your roommate at FAYO meetings. Also thanks to Hanne Jeppesen for her great patience and kindness when it comes to young researchers who doesnÊt know their way around FPAS and Cosmic. Charlotte Olsen and Lone Hansen – thank you for the times you helped me draw blood from patients with veins that clearly had a life of their own. Also thanks for letting me use your lab in the evenings. Thank you, Kirsten Ulla Bahrt and Anette Tyrsted Mikkelsen for your massive help with storing and analyzing the collected blood and urine from the patients, and for all the funny remarks and smiles. Great thanks to my family and friends - for your endless support and believe in me. A special thanks goes out to my beautiful daughter; Isabella, who keeps me smiling even at rough times. You mean the world to me. Thanks to the Danish Cohort of Pediatric Diabetes 1987 for the time you spent on this project. Hopefully it will be of help to many future patients with type 1 diabetes mellitus. This work was only made possible by generous donations from: Fight for Sight, Denmark; Synoptic Foundation; Medivit Aps; Gangsted Foundation; The Legacy of Karen Svankjaer Yde; Lykfeldts Grant; The Danish Society of Ophthalmology; The A.P. Moeller Foundation for the Advancement of Medical Science; Alcon and Novartis travel grants; The Region of Southern Denmark; and The University of Southern Denmark. Thank you. And finally thanks to you, dear reader. You have so far read two pages of my dissertation. I hope you will enjoy the rest. Rebecca Broe Odense, May 2014
aos_93_A102_contents_Layout 1 09-02-2015 17:15 Page 1
Contents
ACKNOWLEDGEMENTS
1
LIST OF PUBLICATIONS/PAPERS
3
ABBREVIATIONS
4
INTRODUCTION Diabetes mellitus Epidemiology of microvascular complications in type 1 diabetes mellitus New approaches for prevention and treatment of microvasculopathy
5 5
6
OBJECTIVES
7
PATIENTS AND METHODS The Danish Cohort of Pediatric Diabetes 1987 Study population Inclusion- and exclusion criteria Drop-out analysis Ethics Data acquisition Study design Baseline examination in 1995 Interview General examination Eye examination Follow-up examination in 2011 General examination Eye examination The Danish Patient Registry Data analysis Diabetic peripheral neuropathy - calculations and definitions Diabetic nephropathy – calculations and definitions Diabetic retinopathy – grading and definitions Retinal vessel calibers Retinal vascular fractal dimension Statistical methods Common for all papers Paper I Paper II+III
8 8 9 9 9 10 11 11 11 11 11 11 12 12 12 12 12
RESULTS Clinical characteristics of the cohort
5
12 12 13 15 17 19 19 19 19 21 21
The 16-year incidence, progression and regression in diabetic retinopathy (Paper I) The incidence of diabetic retinopathy Regression in diabetic retinopathy Progression in diabetic retinopathy The incidence of proliferative diabetic retinopathy Retinal vessel calibers and microvascular complications (Paper II) CRAE and CRVE as predictors of microvascular complications Diabetic peripheral neuropathy Diabetic nephropathy Proliferative diabetic retinopathy ROC analyzes for estimation of cutoffs Retinal vascular fractal dimension and microvascular complications (Paper III) The retinal vascular fractal dimension as predictor of microvascular complications Diabetic peripheral neuropathy Diabetic nephropathy Proliferative diabetic retinopathy ROC analyzes for estimation of cutoffs
22 22 22 23 23 25 27 27 27 27 29 31 33 33 33 33 33
DISCUSSION The burden of diabetic retinopathy in type 1 diabetes mellitus (Paper I) Risk factors for diabetic retinopathy and treatment modalities (Paper I) Early detection of high-risk patients (Paper II+III) Strengths and limitations Paper I Paper II+III
36 37 39 40 40
OVERALL CONCLUSIONS
41
PERSPECTIVES
42
SUMMARY IN ENGLISH
43
SUMMARY IN DANISH (DANSK RESUMÉ)
45
REFERENCES
47
PAPER I PAPER II PAPER III
36 36
aos_93_A102_List of Paper_Layout 1 03-02-2015 07:39 Page 1
ACTA OPHTHALMOLOGICA 2015
List of publications/papers This thesis is based on the following papers: Paper I Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Peto T, Grauslund J. The 16-year incidence, progression and regression of diabetic retinopathy in a young population-based Danish cohort with type 1 diabetes mellitus: The Danish cohort of pediatric diabetes 1987 (DCPD1987). Acta Diabetologica 2013, doi: 10.1007/s00592-013-0527-1 Paper II Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Hodgson L, Wong TY, Peto T, Grauslund J. Retinal vessel calibers predict long-term microvascular complications in type 1 diabetes mellitus: The Danish cohort of pediatric diabetes 1987 (DCPD1987). Diabetes 2014 (Epub ahead of print) Paper III Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Peto T, Grauslund J. Retinal vascular fractals predict long-term microvascular complications in type 1 diabetes mellitus: The Danish cohort of pediatric diabetes 1987 (DCPD1987). Diabetologia 2014 (Epub ahead of print)
aos_93_A101_List of abbreviations_Layout 1 30-01-2015 16:38 Page 1
ACTA OPHTHALMOLOGICA 2015
List of abbreviations ACR
Albumin/creatinine ratio
AER
Albumin excretion ratio
AUC
Area under the curve
BMI
Body mass index
BP
Blood pressure
CI
Confidence interval
CRAE
Central Retinal Arteriolar Equivalent
CRVE
Central Retinal Venular Equivalent
DCCT
The Diabetes Control and Complications Trial
DCPD1987
The Danish Cohort of Pediatric Diabetes 1987
DN
Diabetic nephropathy
DPN
Diabetic peripheral neuropathy
DR
Diabetic retinopathy
ETDRS
Early Treatment Diabetic Retinopathy Study
FD
The retinal vascular fractal dimension
HbA1c
Hemoglobin A1c
NPDR
Non-proliferative diabetic retinopathy
OR
Odds ratio
PDR
Proliferative diabetic retinopathy
ROC
Receiving operator characteristics
SD
Standard deviation
VPT
Vibration perception threshold
WESDR
The Wisconsin Epidemiologic Study of Diabetic Retinopathy
Acta Ophthalmologica 2015
Thesis Early risk stratification in paediatric type 1 diabetes Rebecca Broe Department of Ophthalmology, Odense University Hospital, Odense C, Denmark
ABSTRACT. In the late 1980s all Danish children with type 1 diabetes were invited for a nationwide evaluation of glycemic control. Approximately 75% (n = 720) participated and have later been referred to as The Danish Cohort of Pediatric Diabetes 1987 (DCPD1987). The results were surprisingly poor glycemic control among these young patients which lead to a great emphasis on glycemic control in the Danish Pediatric Departments. In 1995 the participants were invited for yet another evaluation but this time with main focus on early signs of microvascular complications – 339 participated. The mean HbA1c had remained at high levels (9.6%) and 60% of the participants had some level of Diabetic Retinopathy (DR). However, as the patients with DR mostly had the very milder forms it was believed that stricter glycemic control would reverse or at least stop progression of the disease in accordance with results from the large intervention study DCCT. This was investigated further at follow-up in 2011. The first study in the present thesis aimed to describe the 16-year incidence, progression and regression of DR in 185 participants from the DCPD1987 cohort. The 16-year incidence of proliferative retinopathy (PDR), 2-step progression and regression of DR was 31.0, 64.4, and 0.0%, respectively. As expected, the participants with PDR at follow-up had significantly higher HbA1c-values at both baseline and follow-up than those without PDR. However; a significantly larger decrease in HbA1c was also observed in the group with PDR over the study period, which in accordance with DCCT should have prevented the development of PDR to some extent. A surprisingly high incidence of proliferative retinopathy amongst young patients with type 1 diabetes in Denmark was found despite improvements in HbA1c over time. The improvement in HbA1c was either too small or happened too late. This study highlights that sight-threatening diabetic retinopathy remain a major concern in type 1 diabetes and the importance of early glycemic control. Identifying high-risk patients at a very early stage is not only desired for prevention of diabetic retinopathy – neuropathy and nephropathy similarly remain frequent in type 1 diabetes. Early risk stratification will allow for timely implementation of effective interventions and for individualized screening and diabetes care. The second and third studies of this thesis provide the longest prospective studies to date on both retinal vessel calibers and retinal fractal dimensions and their predictive value on diabetic microvascular complications. Semi-automated computer software has been developed to measure smaller changes in the retinal vessels on retinal photographs. Two of the first parameters to be reliably estimated by these programs were retinal vessel calibers and retinal vascular fractal dimensions (a quantitative measure on vascular complexity). There is very limited knowledge on their predictive value on diabetic complications thus
1
Acta Ophthalmologica 2015
far. In the second and third study, a consistent relation between narrower retinal arteriolar calibers, wider retinal venular calibers, lower fractal dimensions and the 16-year incidences of diabetic neuropathy, nephropathy and proliferative retinopathy was found. This has never been shown before. The results on vessel analyzes provides indications of a common pathogenic pathway for diabetic microvascular complications and therefore a possibility of universal risk estimation for development of neuropathy, nephropathy and retinopathy in type 1 diabetes. Acta Ophthalmol. 2015: 93, thesis 2: 1–19 ª 2015 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
doi: 10.1111/aos.12702
Introduction Diabetes mellitus
Diabetes mellitus is a chronic disease characterized by elevated glucose levels in blood and urine. This is either caused by lack of insulin production (type 1) subsequent to an autoimmune destruction of pancreatic beta cells or by insulin resistance in the peripheral cells (type 2) induced by an unhealthy lifestyle. Most children diagnosed with diabetes have type 1, while type 2 most often has its onset during adulthood. It was estimated that in 2011, there were around 366 million people with diabetes worldwide – a number predicted to increase to an alarming 552 million by 2030 (Whiting et al. 2011). Worldwide, the number of children with type 1 diabetes aged 0–14 years was estimated to be 497 000 in 2013 with more than a quarter coming from Europe (the prevalence was around 129 000 children – 1500 of these were Danish) (Patterson et al. 2014). Along with the other Scandinavian countries, in 2013, Denmark was placed in the top 10 of countries with the highest incidence rates for type 1 diabetes in children under 15 years of age with an annual rate of 25 per 100 000 children (Patterson et al. 2014). Epidemiology of microvascular complications in type 1 diabetes mellitus
As the global prevalence of diabetes increases so will the number of patients with diabetes-related microvascular complications, which can affect the nerves (neuropathy), kidneys (nephropathy) and the eyes (retinopathy). Estimates on the burden and risk of diabetic neuropathy and nephropathy
2
can be difficult to compare between studies due to great variations in assessment methods, and thus, it is unclear exactly how many patients suffer from these complications. For instance, a large American study on patients with type 1 diabetes has reported a prevalence of 32.0% after 25 years of diabetes duration for nephropathy (Pambianco et al. 2006), while a Danish study found a prevalence of 10.6% after 44 years of duration (Grauslund et al. 2009a,b). For neuropathy, the prevalence was 52.7% after 44 years in the Danish study (Grauslund et al. 2009a,b) and 35–50% depending on year of diagnosis in the American study after 25 years (Pambianco et al. 2006). For diabetic proliferative retinopathy (PDR), the methodology is more consistent. Retinal images are gold standard. A Danish study on type 1 diabetes reported a 25-year cumulative incidence of PDR of 42.9% (Grauslund et al. 2009a,b), much in line with the 42.0% figure from the Wisconsin epidemiologic study of diabetic retinopathy (WESDR) (Klein et al. 2008). In the latter, a decrease in incidence rates with increasing year of diabetes diagnosis was seen, pointing towards overall improvements, but this was not the case in another American study (Pambianco et al. 2006) which additionally reported a 25-year incidence of 53.0%. New approaches for prevention treatment of microvasculopathy
and
Intervention studies have revealed beneficial effect of improvement in glycaemic control (Anonymous 1993; Reichard et al. 1993; Wang et al. 1993) and antihypertensive treatment
(Anonymous 1998; Chaturvedi et al. 1998) on development and progression in diabetic retinopathy (DR) and nephropathy. Nonetheless, diabetic microvascular complications still occur, and it is therefore needed to search for other means of intervention. Furthermore, it is important to identify the patients at high risk earlier to allow for timelier implementation of current treatment modalities. Identifying new biomarkers might be helpful in this respect and would also support development of new treatments as a greater understanding of the underlying pathogenesis could be achieved. Retinal vessel analyses could potentially deliver such biomarkers. The retinal vasculature provides a unique opportunity to non-invasively observe the circulatory system in the human body and has for many years been a key element in observing progression in diabetic vascular damage. In the later years, measurements of retinal vessel calibres have been linked to diabetic complications (Wong et al. 2004a,b; Klein et al. 2006a,b; Grauslund et al. 2009a,b; Cheung et al. 2012a,b; Ding et al. 2012a,b), but as most are cross-sectional designs or short-term follow-up studies, there is very limited information on the predictive value of these measurements. To truly define whether there is a link between retinal vessel calibres and the development of microvascular complications, prospective studies with long follow-up periods are needed. In addition to calibre measurements, a way of assessing a more global parameter of the retinal vasculature by semiautomatic computer software became available recently, that is assessment of the retinal vascular fractal dimension
Acta Ophthalmologica 2015
(FD). This is a quantitative measurement of the complexity and density of branching structures that are characterized by self-similarity and retain a similar level of complexity despite changes in magnification. These structures are frequent in nature (snowflakes, tree branches, lightning), but it is also a recognized pattern in the eye – the retinal vascular tree has fractal properties (Masters 2004). The FD increases with increasing structural complexity and space-filling properties of the retinal circulation. Possible links between vascular complexity and microvascular diabetic complications are sparsely investigated by some cross-sectional reports (Daxer 1993; Cheung et al. 2009; Grauslund et al. 2010; Lim et al. 2013) and one shortterm prospective study (Lim et al. 2009). The predictive value of FD is therefore rather undocumented, leaving a need for further studies.
Objectives The overall aims of this thesis were to examine a population-based paediatric cohort of Danish patients with type 1 diabetes mellitus to
•
• •
describe the 16-year cumulative incidence of DR, regression and progression in DR, as well as the incidence of PDR in relation to various baseline risk factors (Paper I). investigate the predictive value of retinal vascular calibres on the 16year development of diabetic microvascular complications (Paper II). examine whether retinal vascular fractal analysis could be a potential tool for long-term prediction of diabetic microvasculopathy (Paper III).
Patients and methods The Danish Cohort of Pediatric Diabetes 1987
In the 1980s, inadequate metabolic control was frequently seen in children with type 1 diabetes, especially during adolescence. Studies had shown how metabolic control deteriorates in puberty (Daneman et al. 1981; Bloch et al. 1987), and this was most often attributed to poor compliance and dietary neglect. However, decreased sensitivity to insulin was shown in both children with and without diabetes (Amiel et al.
1986; Bloch et al. 1987), and thus, the combined effect of puberty and diabetes on insulin resistance was speculated to be responsible for the impaired blood glucose control seen in teenagers with type 1 diabetes. To obtain more knowledge on this topic, a Danish research group of paediatricians initiated a nationwide cross-sectional study in 1986–1987 to evaluate haemoglobin A1c (HbA1c) levels in relation to age, sex, body mass index (BMI), duration of diabetes and insulin dose in Danish children with type 1 diabetes. In the 21 (of 22) paediatric departments that took part, all children and adolescents with type 1 diabetes, aged 19 years or below, were asked to participate and 884 were included (Mortensen et al. 1988). It was discovered that the current treatment of childhood diabetes nationwide was rather unsatisfactory (mean HbA1c = 9.7% in adolescents (>12 years of age) and 8.9% in preadolescents). In the early 1980s, it was furthermore shown that microalbuminuria is an early predictor of nephropathy (Viberti et al. 1982). Concomitantly, it was shown that arterial blood pressure (BP) and HbA1c values were significantly increased in young people with type 1 diabetes and microalbuminuria (versus normoalbuminuria) (Wiseman et al. 1984). The Danish study group therefore initiated another cross-sectional study in 1989 to determine the prevalence of microalbuminuria in Danish children with type 1 diabetes in relation to age, duration of diabetes, BP, BMI and HbA1c. This time all 22 paediatric departments treating children and adolescents with type 1 diabetes took part and 1020 patients were included (Mortensen et al. 1990). Mean HbA1c for all participants was 9.6%, and microalbuminuria was seen in 4.3% of these patients. In the early 1990s, striking results from especially the Diabetes Control and Complications Trial (DCCT) (Anonymous 1993) revealed the importance of glycaemic control in the pathogenesis to the microvascular late complications seen in diabetes. With the knowledge derived from the previous two studies and from the DCCT, the paediatric study group anticipated that the Danish children with type 1 diabetes were at high risk of developing microvasculopathy. They aimed to investigate whether there were early signs of neu-
ropathy, nephropathy and retinopathy at a follow-up study in 1995. A total of 720 patients had participated at the examinations in both 1987 and 1989 and were thus decided eligible for follow-up in 1995 – these patients constitute the Danish Cohort of Pediatric Diabetes 1987 (DCPD1987). Study population
Of the 720 patients from DCPD1987, 339 participated at follow-up in 1995 (Olsen et al. 1999). Despite great emphasis on glycaemic control since the previous studies, the mean HbA1c had not changed and 60% of the patients showed early signs of DR (Olsen et al. 2000). However, as the patients with DR mostly had the very mild forms, it was believed that stricter glycaemic control would reverse or at least stop progression of the disease in accordance with the DCCT results (Anonymous 1993). This was something we aimed to investigate 16 years later when we decided to re-examine the cohort with respect to DR and other late complications. Inclusion- and exclusion criteria
With main focus being DR, we decided to base the studies in the present thesis on the participants from 1995 (n = 339) where retinal photography was included in the examinations. The only inclusion criterion used was therefore participation at baseline in 1995, and the only exclusion criterion was missing baseline retinal photographs. Dropout analysis
The DCPD1987 initially comprised approximately 75% of all Danish children aged 19 or below with type 1 diabetes (n = 720). At the study in 1995, 339 (47.1%) participated and 381 did not take part for various reasons (Fig. 1). A dropout analysis showed that those who did not participate in 1995 were significantly older and had poorer glycaemic control in the studies in 1987–1989 compared to the participants. Of the 339 patients, 15 had no retinal photographs available from 1995 and were thus excluded from the 2011 follow-up. A total of 76 of the remaining 324 patients were not reachable at follow-up due to emigration (n = 6), research protection (n = 57) or because they had died (n = 13). Of the
3
Acta Ophthalmologica 2015
Cohort (1987-1989), n = 720
Deceased (n = 2) Emigrated (n = 11) Not traceable (n = 2)
Potential for follow-up (1995), n = 705
Declined (n = 108) Did not respond (n = 152) Failed to show (n = 106)
Participated at baseline (1995), n = 339
No retinal photos (n = 15) Deceased (n = 13) Emigrated (n = 6) Researcher protected (n = 57)
Potential for follow-up (2011), n = 248
Declined (n = 22) Did not respond (n = 38) Failed to show (n = 3)
Participated at follow-up (2011), n = 185 Fig. 1. The Danish Cohort of Pediatric Diabetes 1987 over time.
248 patients eligible for participation in 2011, 185 (74.6%) took part (Fig. 1). Comparison of data from 1995 for the participating (n = 185) and non-participating (n = 139) patients in 2011 showed that the non-participants were significantly younger (mean age 20.2 3.2 versus 21.0 3.3 years), had higher levels of HbA1c (mean 10.1 1.6 versus 9.5 1.6%) and lower vibration perception threshold (VPT; mean 7.4 3.0 versus 8.1 3.0 volt) when compared to the participants. No difference was seen with respect to sex, duration of diabetes, BP, BMI, presence of albuminuria, use of antihypertensive treatment or level of DR.
4
Ethics
The study was performed in accordance with the criteria of the Helsinki II Declaration and was approved by the local scientific ethics committee. All patients gave a written informed consent at both baseline and follow-up examinations.
Data acquisition Study design
The overall study design is a population-based prospective cohort study of type 1 diabetes mellitus in childhood and adolescence in Denmark. Moreover, it is a study of risk factors for
early development of microvascular complications. It was designed as a 16-year follow-up investigation based on clinical examinations in 1995 and 2011. Baseline examination in 1995 Interview. All participants underwent
an interview, where information on sex, age, diabetes duration and use of antihypertensive medicine was recorded. General examination. Height and weight were measured and BMI calculated. With a digital BP meter (UA-751; Takeda Medical, Tokyo, Japan), the arterial BP was measured with the patient in sitting and resting position. VPT was measured on the apex of the first right toe in triplicate by a hand-held biothesiometer (Bio-Medical Instrument, Newbury, OH, USA). Blood samples were collected, and HbA1c was analysed centrally by automated highpressure liquid chromatography (HPLC; HI-AUTO A1c, normal ranges 4.3–5.8, mean 5.3%). Participants handed in two consecutive overnight timed urine samples for determination of mean albumin excretion ratio (AER). Albumin concentration was analysed by an immunoturbidimetric method with a detection limit of 1 mg/l and an interassay CV of 7% (Ames, Milan, Italy). Eye examination. Colour retinal photographs were taken in mydriasis by trained operators with 40–60° retinal cameras in accordance with the EURODIAB recommendations (Aldington et al. 1995); thus, two fields were taken of each eye, a macular-temporal field and a disc-nasal field. The film slides were later digitalized with DigitDia 5000 Filmscanner (Reflecta, Rottenburg, Germany) to a resolution of 4.926 9 3.090 pixels. Follow-up examination in 2011 General examination. Participants
underwent a clinical examination where VPT was measured in triplicate on the apex of the first right toe with a hand-held biothesiometer (Bio-Medical Instrument). Spot urine samples were collected at the end of the examinations unless the patient was menstruating or had ongoing infections – those patients were asked to hand in samples at a later time. The urine samples were analysed to determine the albumin/ creatinine ratio (ACR). Measurements of albumin and creatinine were per-
Acta Ophthalmologica 2015
formed on an Abbott Architect analyzer (Abbott, Deerfield, IL, USA) by immunoturbidimetric and enzymatic assays, respectively. Blood samples were collected, and HbA1c was analysed on an automated HPLC analyzer (Tosoh HLC-723 G8; Alere, Oslo, Norway). Eye examination. Mydriatic non-stereo digital colour retinal photographs were taken by the same trained operator using 3D OCT-2000 Spectral Domain OCT (Topcon, Tokyo, Japan) at a resolution of 4288 9 2848 pixels. Seven 45° fields were taken of each eye in accordance with the Early Treatment Diabetic Retinopathy Study (ETDRS) standards (Anonymous 1985). The Danish Patient Registry. For comparisons between participants and nonparticipants, The Danish Patient Registry provided data on pan-retinal laser photocoagulation (procedure code KCKC15), renal transplant (procedure codes KKAS00, KKAS10 and KKAS20), and dialysis (procedure codes BJDF00 and BJDF20) in every patient in DCPD1987. Data analysis Diabetic peripheral neuropathy – calculations and definitions
At both baseline and follow-up, the mean of the last two VPT measurements was calculated and used as an estimate of neuropathy. Abnormal VPT was defined as mean VPT >25 volt, and participants with abnormal VPT were classified as having developed diabetic peripheral neuropathy (DPN) – both at baseline and at follow-up. Diabetic nephropathy – calculations and definitions
Baseline micro- and macroalbuminuria was defined as mean AER 20–200 lg/ min and >200 lg/min, respectively, in at least two timed overnight urine collections. At follow-up, participants with ACR ≥300 mg/g (macroalbuminuria), history of kidney transplant or who had received dialysis were all classified as having diabetic nephropathy (DN).
(Aldington et al. 1995). Retinopathy was overall classified into five levels: 0 – none. 1 – minimal non-proliferative diabetic retinopathy (NPDR; microaneurysms and/or minor haemorrhages and/or hard exudates). 2 – moderate NPDR (moderate to severe haemorrhages/microaneurysms and mild cotton-wool spots and/or intraretinal microvascular abnormalities and/or venous beading). 3 – severe NPDR (severe haemorrhages/microaneurysms and severe cotton-wool spots and/or intraretinal microvascular abnormalities and/or venous beading). 4 – PDR (scars of photocoagulation, new vessels, fibrous proliferations, vitreous- or preretinal haemorrhages). When comparing participants to non-participants in this study, this grading was used; however, we combined level 1–3 into ‘NPDR’. On the 185 participants from 2011, all baseline retinal photos were regraded and follow-up photos were graded for DR by a single grader (ML Rasmussen, trained at Moorfields Eye Hospital Reading Centre, London, UK) with a modified ETDRS protocol, allowing for non-standard photography at baseline (Anonymous 1991; Broe et al. 2013). The ETDRS levels used are shown in Table 1. Patients with ETDRS level 61 or above were classified as having PDR in all analyses. The development in DR (Paper I) over the study period was assessed by
Retinal vessel calibres
Retinal vascular calibres were graded by the same certified grader (R Broe, trained and certified at Centre for Eye Research Australia, Melbourne, Australia) with semi-automated computer software (IVAN, Department of Ophthalmology Visual Science, University of Wisconsin, Madison, WI, USA) following a previously validated protocol (Hubbard et al. 1999; Wong et al. 2004a,b). The program automatically places a grid on top of the optic disc with two zones marked. All vessels coursing completely through the zone 0.5–1.0 disc diameter (zone B) from the disc margin are traced and marked either blue (venules) or red (arterioles) by the program. The grader provides necessary adjustments with respect to grid placement, inaccurate tracing, missing tracings and identification of vessel types (Fig. 2).
Table 1. The modified Early Treatment Diabetic Retinopathy Study (ETDRS) levels used for grading retinopathy in the DCPD1987 study Level Level Level Level
10 20 35 43
Level 47 Level 53
Level 61 Level 65 Level 71
DR – grading and definitions
In 1995, all retinal photographs were graded for DR by the same trained grader (AK Sjølie) with the EURODIAB Hammersmith grading system
comparison between follow-up and baseline photos on the worst eye at follow-up. For progression and regression in DR, a two-step minimum was chosen – that is, participants had to move at least two levels up or down on the ETDRS scale in the study period to be classified as having progressed/ regressed. Participants were classified as having PDR in the studies on vessel diameters and retinal fractal dimensions (papers II and III) if they at follow-up had ETDRS level 61 or above on the eye included in the vessel analyses.
Level 75 Level 85
No retinopathy Microaneurysms only Hard exudates, cotton-wool spots and/or mild retinal haemorrhages Moderate haemorrhages in four fields/severe in one field or mild intraretinal microvascular abnormalities (IRMA) in one to three fields Both characteristics from level 43 and/or IRMA in four fields, severe haemorrhages in two to three fields and venous beading in one field Two or more characteristics from level 47 and/or severe retinal haemorrhages in four fields, moderate to severe IRMA in one field and venous beading in at least two fields New vessels elsewhere (NVE)
ACTA OPHTHALMOLOGICA THESIS http://www.actaophthalmologica.com
Thesis Early risk stratification in pediatric type 1 diabetes
Rebecca Broe Department of Ophthalmology Odense University Hospital Sdr. Boulevard 29, DK-5000 Odense C
aos_93_A102_Acknowlegements_Layout 1 03-02-2015 07:34 Page 1
ACTA OPHTHALMOLOGICA 2015
Acknowledgements The work presented in this PhD thesis was carried out at the Department of Ophthalmology, Odense University Hospital, Denmark, in 2010-2014. Fortunately, I was not alone on this trip but accompanied by an extended team of colleagues and friends, always willing to help and motivate me. For this I am very grateful. My greatest source of inspiration both in research and how to be a clinician came from the most amazing doctor I have ever met – Anne Katrin SjŒlie. She always had time for me and just a few moments with her could turn the most unmanageable problems into minor bumps on the road. I am deeply grateful for the time we worked together and for all the laughs we shared. You are truly missed. I was fortunate to have Jakob Grauslund take over as main supervisor after Anne Katrin. I am amazed at how much work you do in a day and also convinced you have found the way to stop time. Thank you for fitting me in your busy schedule, for all your great ideas and your continuous support and help with my research. I hope we will keep working together for many years more. I was, furthermore, blessed with a co-supervisor who has a gift when it comes to keeping PhD-students on the right track, Tunde Peto. At several meetings I had ideas and plans all over the place and no overview of my time schedule - something I do not thrive with at all. An hour later I could leave the meeting feeling calm and on top of things – thatÊs just priceless. Also thanks for letting me visit Moorfields. I also had the pleasure of collaborating with two incredibly friendly and helpful pediatricians from Herlev, Birthe Susanne Olsen and Henrik BindesbŒl Mortensen. Without your splendid work in the 1980s and 90s this study would never have been possible. I am very grateful for this opportunity to follow up on these patients and for your enthusiasm, helpfulness and positive spirit throughout my time as a researcher. I had the privilege of visiting the Centre for Eye Research Australia in Melbourne, Australia, where Lauren Hodgson and Ryo Kawasaki trained me in retinal vessel analyzes. I am thankful for their great kindness and helpfulness. A special thanks to Tien Wong for allowing this visit and for great inputs on one of the papers. I wish to thank Kelvin Kamp Mortensen for allowing this study to be conducted at the Department of Ophthalmology at Odense University Hospital. I hope to continue working here for many years on. I would also like to thank Toke Bek and Henrik Lund-Andersen for allowing me to examine some of the participants at the Departments of Ophthalmology in Aarhus and Glostrup. A special thanks goes out to my dear friends and colleagues in the research unit: Anders HŒjslet Vestergaard, who continuously tried to develop type 2 diabetes by consuming massive amounts of chocolate; Kristian Lundberg for your forever positive mind and great sense of humor, even after corresponding with the PhD school on salary issues; Ulrik FrydkjÕr-Olsen, who with great enthusiasm took upon himself to become my personal slave in his spare time and manually digitalized around 1500 slides; Malin Lundberg Rasmussen, who truly beats me when it comes to power shopping; Christina DŒfler Poulsen, who will always lend you a screw driver from her office drawer when needed; and Rasmus SŒgaard Hansen, who takes any task upon him with great enthusiasm and a positive approach. Keep the smiles up, lovely people. Furthermore, I would like to thank Sasikala Thineshkumar for being a lovely friend during these years. It has been a pleasure to talk things over with you and to be your roommate at FAYO meetings. Also thanks to Hanne Jeppesen for her great patience and kindness when it comes to young researchers who doesnÊt know their way around FPAS and Cosmic. Charlotte Olsen and Lone Hansen – thank you for the times you helped me draw blood from patients with veins that clearly had a life of their own. Also thanks for letting me use your lab in the evenings. Thank you, Kirsten Ulla Bahrt and Anette Tyrsted Mikkelsen for your massive help with storing and analyzing the collected blood and urine from the patients, and for all the funny remarks and smiles. Great thanks to my family and friends - for your endless support and believe in me. A special thanks goes out to my beautiful daughter; Isabella, who keeps me smiling even at rough times. You mean the world to me. Thanks to the Danish Cohort of Pediatric Diabetes 1987 for the time you spent on this project. Hopefully it will be of help to many future patients with type 1 diabetes mellitus. This work was only made possible by generous donations from: Fight for Sight, Denmark; Synoptic Foundation; Medivit Aps; Gangsted Foundation; The Legacy of Karen Svankjaer Yde; Lykfeldts Grant; The Danish Society of Ophthalmology; The A.P. Moeller Foundation for the Advancement of Medical Science; Alcon and Novartis travel grants; The Region of Southern Denmark; and The University of Southern Denmark. Thank you. And finally thanks to you, dear reader. You have so far read two pages of my dissertation. I hope you will enjoy the rest. Rebecca Broe Odense, May 2014
aos_93_A102_contents_Layout 1 09-02-2015 17:15 Page 1
Contents
ACKNOWLEDGEMENTS
1
LIST OF PUBLICATIONS/PAPERS
3
ABBREVIATIONS
4
INTRODUCTION Diabetes mellitus Epidemiology of microvascular complications in type 1 diabetes mellitus New approaches for prevention and treatment of microvasculopathy
5 5
6
OBJECTIVES
7
PATIENTS AND METHODS The Danish Cohort of Pediatric Diabetes 1987 Study population Inclusion- and exclusion criteria Drop-out analysis Ethics Data acquisition Study design Baseline examination in 1995 Interview General examination Eye examination Follow-up examination in 2011 General examination Eye examination The Danish Patient Registry Data analysis Diabetic peripheral neuropathy - calculations and definitions Diabetic nephropathy – calculations and definitions Diabetic retinopathy – grading and definitions Retinal vessel calibers Retinal vascular fractal dimension Statistical methods Common for all papers Paper I Paper II+III
8 8 9 9 9 10 11 11 11 11 11 11 12 12 12 12 12
RESULTS Clinical characteristics of the cohort
5
12 12 13 15 17 19 19 19 19 21 21
The 16-year incidence, progression and regression in diabetic retinopathy (Paper I) The incidence of diabetic retinopathy Regression in diabetic retinopathy Progression in diabetic retinopathy The incidence of proliferative diabetic retinopathy Retinal vessel calibers and microvascular complications (Paper II) CRAE and CRVE as predictors of microvascular complications Diabetic peripheral neuropathy Diabetic nephropathy Proliferative diabetic retinopathy ROC analyzes for estimation of cutoffs Retinal vascular fractal dimension and microvascular complications (Paper III) The retinal vascular fractal dimension as predictor of microvascular complications Diabetic peripheral neuropathy Diabetic nephropathy Proliferative diabetic retinopathy ROC analyzes for estimation of cutoffs
22 22 22 23 23 25 27 27 27 27 29 31 33 33 33 33 33
DISCUSSION The burden of diabetic retinopathy in type 1 diabetes mellitus (Paper I) Risk factors for diabetic retinopathy and treatment modalities (Paper I) Early detection of high-risk patients (Paper II+III) Strengths and limitations Paper I Paper II+III
36 37 39 40 40
OVERALL CONCLUSIONS
41
PERSPECTIVES
42
SUMMARY IN ENGLISH
43
SUMMARY IN DANISH (DANSK RESUMÉ)
45
REFERENCES
47
PAPER I PAPER II PAPER III
36 36
aos_93_A102_List of Paper_Layout 1 03-02-2015 07:39 Page 1
ACTA OPHTHALMOLOGICA 2015
List of publications/papers This thesis is based on the following papers: Paper I Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Peto T, Grauslund J. The 16-year incidence, progression and regression of diabetic retinopathy in a young population-based Danish cohort with type 1 diabetes mellitus: The Danish cohort of pediatric diabetes 1987 (DCPD1987). Acta Diabetologica 2013, doi: 10.1007/s00592-013-0527-1 Paper II Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Hodgson L, Wong TY, Peto T, Grauslund J. Retinal vessel calibers predict long-term microvascular complications in type 1 diabetes mellitus: The Danish cohort of pediatric diabetes 1987 (DCPD1987). Diabetes 2014 (Epub ahead of print) Paper III Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Peto T, Grauslund J. Retinal vascular fractals predict long-term microvascular complications in type 1 diabetes mellitus: The Danish cohort of pediatric diabetes 1987 (DCPD1987). Diabetologia 2014 (Epub ahead of print)
aos_93_A101_List of abbreviations_Layout 1 30-01-2015 16:38 Page 1
ACTA OPHTHALMOLOGICA 2015
List of abbreviations ACR
Albumin/creatinine ratio
AER
Albumin excretion ratio
AUC
Area under the curve
BMI
Body mass index
BP
Blood pressure
CI
Confidence interval
CRAE
Central Retinal Arteriolar Equivalent
CRVE
Central Retinal Venular Equivalent
DCCT
The Diabetes Control and Complications Trial
DCPD1987
The Danish Cohort of Pediatric Diabetes 1987
DN
Diabetic nephropathy
DPN
Diabetic peripheral neuropathy
DR
Diabetic retinopathy
ETDRS
Early Treatment Diabetic Retinopathy Study
FD
The retinal vascular fractal dimension
HbA1c
Hemoglobin A1c
NPDR
Non-proliferative diabetic retinopathy
OR
Odds ratio
PDR
Proliferative diabetic retinopathy
ROC
Receiving operator characteristics
SD
Standard deviation
VPT
Vibration perception threshold
WESDR
The Wisconsin Epidemiologic Study of Diabetic Retinopathy
Acta Ophthalmologica 2015
Thesis Early risk stratification in paediatric type 1 diabetes Rebecca Broe Department of Ophthalmology, Odense University Hospital, Odense C, Denmark
ABSTRACT. In the late 1980s all Danish children with type 1 diabetes were invited for a nationwide evaluation of glycemic control. Approximately 75% (n = 720) participated and have later been referred to as The Danish Cohort of Pediatric Diabetes 1987 (DCPD1987). The results were surprisingly poor glycemic control among these young patients which lead to a great emphasis on glycemic control in the Danish Pediatric Departments. In 1995 the participants were invited for yet another evaluation but this time with main focus on early signs of microvascular complications – 339 participated. The mean HbA1c had remained at high levels (9.6%) and 60% of the participants had some level of Diabetic Retinopathy (DR). However, as the patients with DR mostly had the very milder forms it was believed that stricter glycemic control would reverse or at least stop progression of the disease in accordance with results from the large intervention study DCCT. This was investigated further at follow-up in 2011. The first study in the present thesis aimed to describe the 16-year incidence, progression and regression of DR in 185 participants from the DCPD1987 cohort. The 16-year incidence of proliferative retinopathy (PDR), 2-step progression and regression of DR was 31.0, 64.4, and 0.0%, respectively. As expected, the participants with PDR at follow-up had significantly higher HbA1c-values at both baseline and follow-up than those without PDR. However; a significantly larger decrease in HbA1c was also observed in the group with PDR over the study period, which in accordance with DCCT should have prevented the development of PDR to some extent. A surprisingly high incidence of proliferative retinopathy amongst young patients with type 1 diabetes in Denmark was found despite improvements in HbA1c over time. The improvement in HbA1c was either too small or happened too late. This study highlights that sight-threatening diabetic retinopathy remain a major concern in type 1 diabetes and the importance of early glycemic control. Identifying high-risk patients at a very early stage is not only desired for prevention of diabetic retinopathy – neuropathy and nephropathy similarly remain frequent in type 1 diabetes. Early risk stratification will allow for timely implementation of effective interventions and for individualized screening and diabetes care. The second and third studies of this thesis provide the longest prospective studies to date on both retinal vessel calibers and retinal fractal dimensions and their predictive value on diabetic microvascular complications. Semi-automated computer software has been developed to measure smaller changes in the retinal vessels on retinal photographs. Two of the first parameters to be reliably estimated by these programs were retinal vessel calibers and retinal vascular fractal dimensions (a quantitative measure on vascular complexity). There is very limited knowledge on their predictive value on diabetic complications thus
1
Acta Ophthalmologica 2015
far. In the second and third study, a consistent relation between narrower retinal arteriolar calibers, wider retinal venular calibers, lower fractal dimensions and the 16-year incidences of diabetic neuropathy, nephropathy and proliferative retinopathy was found. This has never been shown before. The results on vessel analyzes provides indications of a common pathogenic pathway for diabetic microvascular complications and therefore a possibility of universal risk estimation for development of neuropathy, nephropathy and retinopathy in type 1 diabetes. Acta Ophthalmol. 2015: 93, thesis 2: 1–19 ª 2015 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd
doi: 10.1111/aos.12702
Introduction Diabetes mellitus
Diabetes mellitus is a chronic disease characterized by elevated glucose levels in blood and urine. This is either caused by lack of insulin production (type 1) subsequent to an autoimmune destruction of pancreatic beta cells or by insulin resistance in the peripheral cells (type 2) induced by an unhealthy lifestyle. Most children diagnosed with diabetes have type 1, while type 2 most often has its onset during adulthood. It was estimated that in 2011, there were around 366 million people with diabetes worldwide – a number predicted to increase to an alarming 552 million by 2030 (Whiting et al. 2011). Worldwide, the number of children with type 1 diabetes aged 0–14 years was estimated to be 497 000 in 2013 with more than a quarter coming from Europe (the prevalence was around 129 000 children – 1500 of these were Danish) (Patterson et al. 2014). Along with the other Scandinavian countries, in 2013, Denmark was placed in the top 10 of countries with the highest incidence rates for type 1 diabetes in children under 15 years of age with an annual rate of 25 per 100 000 children (Patterson et al. 2014). Epidemiology of microvascular complications in type 1 diabetes mellitus
As the global prevalence of diabetes increases so will the number of patients with diabetes-related microvascular complications, which can affect the nerves (neuropathy), kidneys (nephropathy) and the eyes (retinopathy). Estimates on the burden and risk of diabetic neuropathy and nephropathy
2
can be difficult to compare between studies due to great variations in assessment methods, and thus, it is unclear exactly how many patients suffer from these complications. For instance, a large American study on patients with type 1 diabetes has reported a prevalence of 32.0% after 25 years of diabetes duration for nephropathy (Pambianco et al. 2006), while a Danish study found a prevalence of 10.6% after 44 years of duration (Grauslund et al. 2009a,b). For neuropathy, the prevalence was 52.7% after 44 years in the Danish study (Grauslund et al. 2009a,b) and 35–50% depending on year of diagnosis in the American study after 25 years (Pambianco et al. 2006). For diabetic proliferative retinopathy (PDR), the methodology is more consistent. Retinal images are gold standard. A Danish study on type 1 diabetes reported a 25-year cumulative incidence of PDR of 42.9% (Grauslund et al. 2009a,b), much in line with the 42.0% figure from the Wisconsin epidemiologic study of diabetic retinopathy (WESDR) (Klein et al. 2008). In the latter, a decrease in incidence rates with increasing year of diabetes diagnosis was seen, pointing towards overall improvements, but this was not the case in another American study (Pambianco et al. 2006) which additionally reported a 25-year incidence of 53.0%. New approaches for prevention treatment of microvasculopathy
and
Intervention studies have revealed beneficial effect of improvement in glycaemic control (Anonymous 1993; Reichard et al. 1993; Wang et al. 1993) and antihypertensive treatment
(Anonymous 1998; Chaturvedi et al. 1998) on development and progression in diabetic retinopathy (DR) and nephropathy. Nonetheless, diabetic microvascular complications still occur, and it is therefore needed to search for other means of intervention. Furthermore, it is important to identify the patients at high risk earlier to allow for timelier implementation of current treatment modalities. Identifying new biomarkers might be helpful in this respect and would also support development of new treatments as a greater understanding of the underlying pathogenesis could be achieved. Retinal vessel analyses could potentially deliver such biomarkers. The retinal vasculature provides a unique opportunity to non-invasively observe the circulatory system in the human body and has for many years been a key element in observing progression in diabetic vascular damage. In the later years, measurements of retinal vessel calibres have been linked to diabetic complications (Wong et al. 2004a,b; Klein et al. 2006a,b; Grauslund et al. 2009a,b; Cheung et al. 2012a,b; Ding et al. 2012a,b), but as most are cross-sectional designs or short-term follow-up studies, there is very limited information on the predictive value of these measurements. To truly define whether there is a link between retinal vessel calibres and the development of microvascular complications, prospective studies with long follow-up periods are needed. In addition to calibre measurements, a way of assessing a more global parameter of the retinal vasculature by semiautomatic computer software became available recently, that is assessment of the retinal vascular fractal dimension
Acta Ophthalmologica 2015
(FD). This is a quantitative measurement of the complexity and density of branching structures that are characterized by self-similarity and retain a similar level of complexity despite changes in magnification. These structures are frequent in nature (snowflakes, tree branches, lightning), but it is also a recognized pattern in the eye – the retinal vascular tree has fractal properties (Masters 2004). The FD increases with increasing structural complexity and space-filling properties of the retinal circulation. Possible links between vascular complexity and microvascular diabetic complications are sparsely investigated by some cross-sectional reports (Daxer 1993; Cheung et al. 2009; Grauslund et al. 2010; Lim et al. 2013) and one shortterm prospective study (Lim et al. 2009). The predictive value of FD is therefore rather undocumented, leaving a need for further studies.
Objectives The overall aims of this thesis were to examine a population-based paediatric cohort of Danish patients with type 1 diabetes mellitus to
•
• •
describe the 16-year cumulative incidence of DR, regression and progression in DR, as well as the incidence of PDR in relation to various baseline risk factors (Paper I). investigate the predictive value of retinal vascular calibres on the 16year development of diabetic microvascular complications (Paper II). examine whether retinal vascular fractal analysis could be a potential tool for long-term prediction of diabetic microvasculopathy (Paper III).
Patients and methods The Danish Cohort of Pediatric Diabetes 1987
In the 1980s, inadequate metabolic control was frequently seen in children with type 1 diabetes, especially during adolescence. Studies had shown how metabolic control deteriorates in puberty (Daneman et al. 1981; Bloch et al. 1987), and this was most often attributed to poor compliance and dietary neglect. However, decreased sensitivity to insulin was shown in both children with and without diabetes (Amiel et al.
1986; Bloch et al. 1987), and thus, the combined effect of puberty and diabetes on insulin resistance was speculated to be responsible for the impaired blood glucose control seen in teenagers with type 1 diabetes. To obtain more knowledge on this topic, a Danish research group of paediatricians initiated a nationwide cross-sectional study in 1986–1987 to evaluate haemoglobin A1c (HbA1c) levels in relation to age, sex, body mass index (BMI), duration of diabetes and insulin dose in Danish children with type 1 diabetes. In the 21 (of 22) paediatric departments that took part, all children and adolescents with type 1 diabetes, aged 19 years or below, were asked to participate and 884 were included (Mortensen et al. 1988). It was discovered that the current treatment of childhood diabetes nationwide was rather unsatisfactory (mean HbA1c = 9.7% in adolescents (>12 years of age) and 8.9% in preadolescents). In the early 1980s, it was furthermore shown that microalbuminuria is an early predictor of nephropathy (Viberti et al. 1982). Concomitantly, it was shown that arterial blood pressure (BP) and HbA1c values were significantly increased in young people with type 1 diabetes and microalbuminuria (versus normoalbuminuria) (Wiseman et al. 1984). The Danish study group therefore initiated another cross-sectional study in 1989 to determine the prevalence of microalbuminuria in Danish children with type 1 diabetes in relation to age, duration of diabetes, BP, BMI and HbA1c. This time all 22 paediatric departments treating children and adolescents with type 1 diabetes took part and 1020 patients were included (Mortensen et al. 1990). Mean HbA1c for all participants was 9.6%, and microalbuminuria was seen in 4.3% of these patients. In the early 1990s, striking results from especially the Diabetes Control and Complications Trial (DCCT) (Anonymous 1993) revealed the importance of glycaemic control in the pathogenesis to the microvascular late complications seen in diabetes. With the knowledge derived from the previous two studies and from the DCCT, the paediatric study group anticipated that the Danish children with type 1 diabetes were at high risk of developing microvasculopathy. They aimed to investigate whether there were early signs of neu-
ropathy, nephropathy and retinopathy at a follow-up study in 1995. A total of 720 patients had participated at the examinations in both 1987 and 1989 and were thus decided eligible for follow-up in 1995 – these patients constitute the Danish Cohort of Pediatric Diabetes 1987 (DCPD1987). Study population
Of the 720 patients from DCPD1987, 339 participated at follow-up in 1995 (Olsen et al. 1999). Despite great emphasis on glycaemic control since the previous studies, the mean HbA1c had not changed and 60% of the patients showed early signs of DR (Olsen et al. 2000). However, as the patients with DR mostly had the very mild forms, it was believed that stricter glycaemic control would reverse or at least stop progression of the disease in accordance with the DCCT results (Anonymous 1993). This was something we aimed to investigate 16 years later when we decided to re-examine the cohort with respect to DR and other late complications. Inclusion- and exclusion criteria
With main focus being DR, we decided to base the studies in the present thesis on the participants from 1995 (n = 339) where retinal photography was included in the examinations. The only inclusion criterion used was therefore participation at baseline in 1995, and the only exclusion criterion was missing baseline retinal photographs. Dropout analysis
The DCPD1987 initially comprised approximately 75% of all Danish children aged 19 or below with type 1 diabetes (n = 720). At the study in 1995, 339 (47.1%) participated and 381 did not take part for various reasons (Fig. 1). A dropout analysis showed that those who did not participate in 1995 were significantly older and had poorer glycaemic control in the studies in 1987–1989 compared to the participants. Of the 339 patients, 15 had no retinal photographs available from 1995 and were thus excluded from the 2011 follow-up. A total of 76 of the remaining 324 patients were not reachable at follow-up due to emigration (n = 6), research protection (n = 57) or because they had died (n = 13). Of the
3
Acta Ophthalmologica 2015
Cohort (1987-1989), n = 720
Deceased (n = 2) Emigrated (n = 11) Not traceable (n = 2)
Potential for follow-up (1995), n = 705
Declined (n = 108) Did not respond (n = 152) Failed to show (n = 106)
Participated at baseline (1995), n = 339
No retinal photos (n = 15) Deceased (n = 13) Emigrated (n = 6) Researcher protected (n = 57)
Potential for follow-up (2011), n = 248
Declined (n = 22) Did not respond (n = 38) Failed to show (n = 3)
Participated at follow-up (2011), n = 185 Fig. 1. The Danish Cohort of Pediatric Diabetes 1987 over time.
248 patients eligible for participation in 2011, 185 (74.6%) took part (Fig. 1). Comparison of data from 1995 for the participating (n = 185) and non-participating (n = 139) patients in 2011 showed that the non-participants were significantly younger (mean age 20.2 3.2 versus 21.0 3.3 years), had higher levels of HbA1c (mean 10.1 1.6 versus 9.5 1.6%) and lower vibration perception threshold (VPT; mean 7.4 3.0 versus 8.1 3.0 volt) when compared to the participants. No difference was seen with respect to sex, duration of diabetes, BP, BMI, presence of albuminuria, use of antihypertensive treatment or level of DR.
4
Ethics
The study was performed in accordance with the criteria of the Helsinki II Declaration and was approved by the local scientific ethics committee. All patients gave a written informed consent at both baseline and follow-up examinations.
Data acquisition Study design
The overall study design is a population-based prospective cohort study of type 1 diabetes mellitus in childhood and adolescence in Denmark. Moreover, it is a study of risk factors for
early development of microvascular complications. It was designed as a 16-year follow-up investigation based on clinical examinations in 1995 and 2011. Baseline examination in 1995 Interview. All participants underwent
an interview, where information on sex, age, diabetes duration and use of antihypertensive medicine was recorded. General examination. Height and weight were measured and BMI calculated. With a digital BP meter (UA-751; Takeda Medical, Tokyo, Japan), the arterial BP was measured with the patient in sitting and resting position. VPT was measured on the apex of the first right toe in triplicate by a hand-held biothesiometer (Bio-Medical Instrument, Newbury, OH, USA). Blood samples were collected, and HbA1c was analysed centrally by automated highpressure liquid chromatography (HPLC; HI-AUTO A1c, normal ranges 4.3–5.8, mean 5.3%). Participants handed in two consecutive overnight timed urine samples for determination of mean albumin excretion ratio (AER). Albumin concentration was analysed by an immunoturbidimetric method with a detection limit of 1 mg/l and an interassay CV of 7% (Ames, Milan, Italy). Eye examination. Colour retinal photographs were taken in mydriasis by trained operators with 40–60° retinal cameras in accordance with the EURODIAB recommendations (Aldington et al. 1995); thus, two fields were taken of each eye, a macular-temporal field and a disc-nasal field. The film slides were later digitalized with DigitDia 5000 Filmscanner (Reflecta, Rottenburg, Germany) to a resolution of 4.926 9 3.090 pixels. Follow-up examination in 2011 General examination. Participants
underwent a clinical examination where VPT was measured in triplicate on the apex of the first right toe with a hand-held biothesiometer (Bio-Medical Instrument). Spot urine samples were collected at the end of the examinations unless the patient was menstruating or had ongoing infections – those patients were asked to hand in samples at a later time. The urine samples were analysed to determine the albumin/ creatinine ratio (ACR). Measurements of albumin and creatinine were per-
Acta Ophthalmologica 2015
formed on an Abbott Architect analyzer (Abbott, Deerfield, IL, USA) by immunoturbidimetric and enzymatic assays, respectively. Blood samples were collected, and HbA1c was analysed on an automated HPLC analyzer (Tosoh HLC-723 G8; Alere, Oslo, Norway). Eye examination. Mydriatic non-stereo digital colour retinal photographs were taken by the same trained operator using 3D OCT-2000 Spectral Domain OCT (Topcon, Tokyo, Japan) at a resolution of 4288 9 2848 pixels. Seven 45° fields were taken of each eye in accordance with the Early Treatment Diabetic Retinopathy Study (ETDRS) standards (Anonymous 1985). The Danish Patient Registry. For comparisons between participants and nonparticipants, The Danish Patient Registry provided data on pan-retinal laser photocoagulation (procedure code KCKC15), renal transplant (procedure codes KKAS00, KKAS10 and KKAS20), and dialysis (procedure codes BJDF00 and BJDF20) in every patient in DCPD1987. Data analysis Diabetic peripheral neuropathy – calculations and definitions
At both baseline and follow-up, the mean of the last two VPT measurements was calculated and used as an estimate of neuropathy. Abnormal VPT was defined as mean VPT >25 volt, and participants with abnormal VPT were classified as having developed diabetic peripheral neuropathy (DPN) – both at baseline and at follow-up. Diabetic nephropathy – calculations and definitions
Baseline micro- and macroalbuminuria was defined as mean AER 20–200 lg/ min and >200 lg/min, respectively, in at least two timed overnight urine collections. At follow-up, participants with ACR ≥300 mg/g (macroalbuminuria), history of kidney transplant or who had received dialysis were all classified as having diabetic nephropathy (DN).
(Aldington et al. 1995). Retinopathy was overall classified into five levels: 0 – none. 1 – minimal non-proliferative diabetic retinopathy (NPDR; microaneurysms and/or minor haemorrhages and/or hard exudates). 2 – moderate NPDR (moderate to severe haemorrhages/microaneurysms and mild cotton-wool spots and/or intraretinal microvascular abnormalities and/or venous beading). 3 – severe NPDR (severe haemorrhages/microaneurysms and severe cotton-wool spots and/or intraretinal microvascular abnormalities and/or venous beading). 4 – PDR (scars of photocoagulation, new vessels, fibrous proliferations, vitreous- or preretinal haemorrhages). When comparing participants to non-participants in this study, this grading was used; however, we combined level 1–3 into ‘NPDR’. On the 185 participants from 2011, all baseline retinal photos were regraded and follow-up photos were graded for DR by a single grader (ML Rasmussen, trained at Moorfields Eye Hospital Reading Centre, London, UK) with a modified ETDRS protocol, allowing for non-standard photography at baseline (Anonymous 1991; Broe et al. 2013). The ETDRS levels used are shown in Table 1. Patients with ETDRS level 61 or above were classified as having PDR in all analyses. The development in DR (Paper I) over the study period was assessed by
Retinal vessel calibres
Retinal vascular calibres were graded by the same certified grader (R Broe, trained and certified at Centre for Eye Research Australia, Melbourne, Australia) with semi-automated computer software (IVAN, Department of Ophthalmology Visual Science, University of Wisconsin, Madison, WI, USA) following a previously validated protocol (Hubbard et al. 1999; Wong et al. 2004a,b). The program automatically places a grid on top of the optic disc with two zones marked. All vessels coursing completely through the zone 0.5–1.0 disc diameter (zone B) from the disc margin are traced and marked either blue (venules) or red (arterioles) by the program. The grader provides necessary adjustments with respect to grid placement, inaccurate tracing, missing tracings and identification of vessel types (Fig. 2).
Table 1. The modified Early Treatment Diabetic Retinopathy Study (ETDRS) levels used for grading retinopathy in the DCPD1987 study Level Level Level Level
10 20 35 43
Level 47 Level 53
Level 61 Level 65 Level 71
DR – grading and definitions
In 1995, all retinal photographs were graded for DR by the same trained grader (AK Sjølie) with the EURODIAB Hammersmith grading system
comparison between follow-up and baseline photos on the worst eye at follow-up. For progression and regression in DR, a two-step minimum was chosen – that is, participants had to move at least two levels up or down on the ETDRS scale in the study period to be classified as having progressed/ regressed. Participants were classified as having PDR in the studies on vessel diameters and retinal fractal dimensions (papers II and III) if they at follow-up had ETDRS level 61 or above on the eye included in the vessel analyses.
Level 75 Level 85
No retinopathy Microaneurysms only Hard exudates, cotton-wool spots and/or mild retinal haemorrhages Moderate haemorrhages in four fields/severe in one field or mild intraretinal microvascular abnormalities (IRMA) in one to three fields Both characteristics from level 43 and/or IRMA in four fields, severe haemorrhages in two to three fields and venous beading in one field Two or more characteristics from level 47 and/or severe retinal haemorrhages in four fields, moderate to severe IRMA in one field and venous beading in at least two fields New vessels elsewhere (NVE)
