UNIVERSITY OF COPENHAGEN FACULTY OF HEALTH AND MEDICAL SCIENCES

PhD Thesis Troels Brynskov, MD

Monitoring of Diabetic Retinopathy and Changes in Related Biomarkers after Bariatric Surgery in Patients with Type 2 Diabetes

FACULTY OF HEALTH AND MEDICAL SCIENCES UNIVERSITY OF COPENHAGEN

Monitoring of Diabetic Retinopathy and Changes in Related Biomarkers after Bariatric Surgery in Patients with Type 2 Diabetes

PhD thesis Troels Brynskov, MD

This thesis has been submitted to the Graduate School at the Faculty of Health and Medical Sciences, University of Copenhagen

1

PhD thesis “Monitoring of Diabetic Retinopathy and Changes in Related Biomarkers after Bariatric Surgery in Patients with Type 2 Diabetes”

Troels Brynskov, MD Department of Ophthalmology, Zealand University Hospital Roskilde, Denmark Faculty of Health and Medical Sciences, University of Copenhagen, Denmark [email protected] +45 50 50 78 89

Submission date: 16 January 2016

Academic advisors Torben Lykke Sørensen, MD, DMSci, Professor Department of Ophthalmology, Zealand University Hospital Roskilde, Denmark Faculty of Health and Medical Sciences, University of Copenhagen, Denmark

Caroline Schmidt Laugesen, MD Department of Ophthalmology, Zealand University Hospital Roskilde, Denmark

Assessment committee Henrik Lund Andersen, MD, DMSci, Professor (Chair) Department of Ophthalmology, Copenhagen University Hospital Glostrup, Denmark Faculty of Health and Medical Sciences, University of Copenhagen, Denmark

Jakob Grauslund MD, PhD, Professor Department of Ophthalmology, Odense University Hospital, Denmark University of Southern Denmark

Stela Vujosevic, MD, PhD, Assistant Professor Department of Ophthalmology, University of Padova, Italy

Cover art: Own montage of a) Iris by Nevermind04/reddit.com and b) fundus image of patient #4.

2 Table of contents

PREFACE 5

ABBREVIATIONS AND ACRONYMS 7

LIST OF PUBLICATIONS 8

INTRODUCTION 9

BACKGROUND 10

TYPE 2 DIABETES 10

BARIATRIC SURGERY 13

DIABETIC RETINOPATHY 14

THE IGF AXIS 20

BARIATRIC SURGERY AND END-ORGAN COMPLICATIONS 20

OBJECTIVES 22

METHODS 23

STUDY DESIGN 23

PATIENTS 23

EXAMINATIONS AND GRADING 24

STATISTICAL ANALYSIS 31

RESULTS 32

RECRUITMENT 32

PREOPERATIVE CHARACTERISTICS 33

CHANGES IN NON-OCULAR OUTCOMES 33

CHANGES IN DIABETIC RETINOPATHY 34

CLINICAL CHANGES IN MACULOPATHY 38

EXPLORATORY ANALYSIS OF SUBCLINICAL CHANGES IN RETINAL THICKNESS 39

CHANGES IN VISUAL ACUITY 42

CHANGES IN VESSEL CALIBERS 42

THE IGF AXIS 43

3 DISCUSSION 48

DIABETIC RETINOPATHY 48

CHANGES IN VISUAL ACUITY 51

SUBCLINICAL OCULAR CHANGES 51

CHANGES IN NON-OCULAR OUTCOMES 53

VALIDITY 54

LIMITATIONS 56

PERSPECTIVES 58

CONCLUSIONS 60

SUMMARY 61

DANSK RESUMÉ (SUMMARY IN DANISH) 63

REFERENCES 65

APPENDICES

PAPER I

PAPER II

PAPER III

4 Preface

I completed the studies within this PhD thesis during my employment at the Department of Ophthalmology, Zealand University Hospital Roskilde 2012–2015. Within this period I have had the fortunate experience of how the field of ophthalmology is undergoing an exciting transformation, as new treatments are emerging for which no or little treatment were previously available. Specifically, the introduction of intravitreal anti- vascular endothelial growth factor for wet age-related macular degeneration, retinal vein occlusion and diabetic retinopathy has revolutionized the prognosis for patients with these diseases. Stem cell therapies and gene therapy are also taking off and multitudes of new imaging modalities are rapidly transforming the field of ophthalmology – it is a very exciting time to be an ophthalmologist!

So many people made this project possible and I am indebted to everyone who was involved from the very beginning to the end. First and foremost, I wish to thank my supervisors Professor Torben Lykke Sørensen and Caroline Schmidt Laugesen from the bottom of my heart! Thank you for trusting me with the project and for welcoming me at the department in Roskilde. Torben, you have been an enormous inspiration, both professionally and personally. Thank you for many great moments shared during these past years and for helping me to see the forest when all I could see was trees and lumber. You have been the greatest possible mentor with an always-open door. Caroline, you are my role model! You have taught me a lot about being an ophthalmologist – every day I strive to become just a little bit more like you.

Also a big thanks to Henrik Kemp who always makes things happen and all staff members at the Department of Ophthalmology in Roskilde, who were always open and supportive. It has been an enormous pleasure to be a part of the growing research unit at the department of ophthalmology in Roskilde. In particular, I would like to thank my fellow PhD-colleagues Amardeep Singh, Mads Falk, Yousif Subhi and Marie Krogh Nielsen, for all the great times and for all the advice, help, moral support and assistance along the way. In particular Yousif who proofread the manuscript for this thesis. Ane Maria Bang Korsholm also deserves praise for laying some of the initial foundation for this project.

Throughout the project, I have had the fortune to be welcomed with a smile and positive attitude, wherever I came. I had the pleasure to collaborate with Andrea Karen Floyd and Annette Lykke Svenningsen from the Department of Bariatric Surgery in Køge. Without their support, there would have been no project at all! The level of recruitment was extraordinary thanks to the efforts of them and everyone else at the Department of Bariatric Surgery in Køge. At the Department of Biochemistry in Roskilde, I could always count on the help and assistance of Marianne Boslev and Birthe Nyberg and their staff. Professor Jan Frystyk provided

5 invaluable expertise with blood samples and provided great feedback for Paper III. His staff from the Medical Research Laboratory at Aarhus Universitetshospital, especially Nils Erik Magnussen and Susanne Sørensen also provided their excellent and attentive assistance with the assays. Malin Lundberg Rasmussen from the Department of Ophthalmology at Odense Universitetshospital was a tremendous help in establishing the image grading procedures. Professor Michael Larsen, Oliver Niels Klefter, Hajer Ahmad Al-Abaiji and Sermed Zyad Khalil Al-Hamdani of the Department of Ophthalmology at Glostrup Hospital generously assisted me with becoming acquainted with the software for the measurement of retinal vessels. The staff at the Department of Biostatistics at Copenhagen University was always very helpful whenever I needed assistance, especially Julie Lyng Forman. Luise Jessen Lundorf provided thoughts and comments for the manuscript for this thesis were sharp as a razor. Of course. Thank you very much.

The generous support of Fight for Sight Denmark (Øjenforeningen), The Bagenkop Nielsen Myopia foundation and the Research Foundation of the Zealand Region made this project economically feasible.

I would also like to thank all the patients who went through all the hassle of scheduling, transportation and tedious examinations in order to make this project happen. This thesis is not mine — it is yours!

Last, but certainly not least: my fantastic backing crew. Above all, my gorgeous wife-to-be Lena and our two energetic dynamos Viggo and Ludvig. And my parents Morten and Tove, and the rest of my family, in-laws and friends. I am so fortunate to have you in my life, I love you very much, and none of this would have been possible, or worthwhile, without you.

Copenhagen, January 2016

Troels Brynskov

6 Abbreviations and acronyms

Abbreviations and acronyms that appear in this thesis except for standard abbreviations in the Diabetologia style guide (http://www.diabetologia-journal.org/webpages/styleguide/abbreviations.html):

ACCORD Action to Control Cardiovascular Risk in Diabetes trial

ACR Albumin Creatinine Ratio

CRAE Central Retinal Artery Equivalent

CRVE Central Retinal Vein Equivalent

DCCT Diabetes Control and Complications Trial

ETDRS Early Treatment Diabetic Retinopathy Study trial

GCL Ganglion Cell Layer

GH Growth Hormone

GLP-1 Glucagon-Like Peptide-1

IGF Insulin-like Growth Factor

IGFBP Insulin-like Growth Factor Binding Protein

INL Inner Nuclear Layer

IPL Inner Plexiform Layer

IRMA Intraretinal Microvascular Abnormality

OCT Optical Coherence Tomography

ONL Outer Nuclear Layer

OS Outer Segment of photoreceptors

RNFL Retinal Nerve Fiber Layer

RPE Retinal Pigment Epithelium

SOS Swedish Obesity Subjects

STAMPEDE Surgical Therapy and Medications Potentially Eradicate Diabetes Efficiently

VEGF Vascular Endothelial Growth Factor

WESDR Wisconsin Epidemiologic Study of Diabetic Retinopathy

UKPDS United Kingdom Prospective Diabetes Study

7 List of publications

Paper I: Brynskov T, Laugesen CS, Svenningsen AL, Floyd AK, Sørensen TL. Monitoring of Diabetic Retinopathy in relation to Bariatric Surgery: a Prospective Observational Study. Obesity Surgery, 2016;26:1279-1286.

Paper II: Brynskov T, Laugesen CS, Svenningsen AL, Floyd AK, Sørensen TL. Thickening of inner retinal layers in the parafovea after bariatric surgery in patients with type 2 diabetes. Acta Ophthalmologica, 2016. Epub ahead of print.

Paper III: Brynskov T, Laugesen CS, Floyd AK, Frystyk J, Sørensen TL. The relationship between diabetic retinopathy and the IGF-axis before and after gastric bypass surgery. Obesity Surgery, 2016. Epub ahead of print.

8 Introduction

Sight is our most precious sense. To prevent sight loss is even more important than to treat sight-threatening diseases when they occur. Diabetic retinopathy is a common disease but predicting progression and vision loss at the individual level is difficult with our present knowledge. The retina is vulnerable during times of metabolic stress. Weight loss surgery – bariatric surgery – is a major metabolic stressor and is often performed in patients with type 2 diabetes due to the association between type 2 diabetes and obesity. Amazingly, most patients with type 2 diabetes achieve normal blood glucose levels immediately after the surgery. If maintained, this is a positive change in the long term on a cohort level, but should we – patients and doctors – worry about the shorter term for the individual patient?

In this PhD-thesis, I will explore this question in more detail by examining the clinical manifestations of diabetic retinopathy as well as the subclinical and systemic biomarkers that may predict patients at risk of diabetic retinopathy in patients with type 2 diabetes undergoing bariatric surgery. I will briefly describe the epidemiology and pathophysiology of diabetes and diabetic retinopathy. I will outline the mechanisms thought to influence worsening of diabetic retinopathy and describe the metabolic effects of two types of bariatric surgery: gastric bypass surgery and vertical sleeve gastrectomy. Thereafter I will summarize the specific objectives that fueled this study, present the results and discuss them. Finally, I will put the findings into perspective. With this approach, I aim to make this thesis a freestanding synthesis of the three papers that appear in the appendix.

The remaining sections of this thesis are in first person plural, as it builds upon the scientific work that I have carried out in collaboration with my co-authors. I will refer to the thesis and the three papers collectively as ‘the study’ and I will refer to each individual paper as ‘Paper’ followed by its roman numeral.

9 Background

The prevalence of type 2 diabetes is increasing worldwide along with the increase in obesity and increased life expectancy. In 2011 an estimated 366 million people had diabetes and that number is expected to exceed 500 million in 2030 [1]. Obesity causes increased insulin resistance and β-cell dysfunction that may lead to type 2 diabetes [2]. It is estimated that 31% of people with morbid obesity (BMI >35 m/kg2) have type 2 diabetes [3]. Adipose tissue releases fatty acids and other pro-inflammatory signals that may be responsible for mediating insulin resistance in susceptible persons [2]. The development of diabetes is also influenced by reduced incretin response from the gut, increased glucose reabsorption in the kidney, increased glucagon secretion by the α-cells of the pancreas as well as dysfunction of central appetite mechanisms that drive obesity which in turn drives diabetes [4].

Type 2 diabetes

Manifest type 2 diabetes was previously diagnosed according to the guidelines of the American Diabetes

Association as HbA1c > 6.5% or fasting plasma glucose > 7.0 mmol/L or plasma glucose > 11.1 mmol/L two hours after an oral glucose tolerance test [5]. From 2012, the Danish National Board of Health adopted the criteria of the World Health Organization to only include an HbA1c above 6.5% [6]. Type 2 diabetes develops as a continuum and the patient has usually had elevated blood glucose levels for many years when the diagnosis of diabetes is made (Figure 1). Prediabetes is defined as HbA1c 5.6–6.4%. In the treatment of diabetes, the goal is usually to keep HbA1c below 7% and some patients with diabetes can maintain this on lifestyle interventions alone.

10

Figure 1. Development of type 2 diabetes. Adapted from [7].

Moderate hyperglycemia does not cause acute damage. However, hyperglycemia causes a slow degrading of many tissues over time (Figure 2). Particularly vulnerable are a) the larger blood vessels causing atherosclerosis and cardiovascular disease, b) the retinal neurovascular unit – that will be discussed at length below – c) the mesangial cells in the renal glomeruli leading to kidney failure and d) the Schwann cells and neurons of the peripheral nerves causing neuropathy. A possibly increased risk of cancer in patients with type 2 diabetes is a matter of ongoing debate [8]. Patients with diabetes have shorter life expectancy due to cardiovascular complications [9,10], but the patients themselves are often more concerned about the risk of losing vision [11].

11

Figure 2. Model of hyperglycemia leading to complications of diabetes [12].

Type 2 diabetes is a chronic systemic disease with organ-specific complications. This has several important implications: a) primary intervention (lifestyle and diet) as well as treatment of hyperglycemia is paramount in the eradication of diabetic complications, b) the treatment of manifest diabetes requires a multidisciplinary approach to reduce diabetes related morbidity, and c) long term reduction of organ related complications and mortality is the ultimate goal in diabetes research. As these complications develop over many years, most intervention trials use a surrogate marker that correlates with end organ damage: HbA1c

[13,14]. Though well-controlled HbA1c is associated with long term reductions in microvascular diabetic complications [14], iatrogenic hypoglycemia is a nontrivial factor to be considered for cardiovascular endpoints. The Action to Control Cardiovascular Risk in Diabetes trial (ACCORD) randomized 10,251 patients with diabetes to receive intensive treatment targeting an HbA1c below 6.0% or conventional treatment targeting an HbA1c 7.0–7.9% and found a higher mortality in the intensively treated arm [15]. Hypoglycemia is therefore a major concern in the current treatment guidelines for diabetes [16]. Also, ‘blood glucose excursions’, as measured using continuous glucose monitoring, could be the trigger for oxidative stress and diabetes related complications, but the evidence is conflicting [17].

12 Bariatric surgery

Unfortunately, long-term weight loss is unattainable in most patients [18]. This is particularly true for the weight loss required to reduce morbid obesity to normal weight or overweight. Strong physiologic counter- mechanisms to weight-loss include hunger and reduced basal metabolic rate [19]. The most viable option to reduce weight permanently is bariatric surgery. The dominant procedures are gastric bypass surgery or vertical sleeve gastrectomy [20] (Figure 3). Anatomically the two procedures are very different, but they incur a similar degree of weight loss and diabetes remission [21]. Interestingly, the surgery appears to change the metabolic signaling and central feedback mechanisms, and weight is not reduced due to the mechanistic restriction in itself [22]. In Denmark, gastric bypass surgery is the most popular procedure [23].

Figure 3. Gastric bypass surgery (A) and vertical sleeve gastrectomy (B) [22].

Bariatric surgery and remission of diabetes

It was a surprising finding when bariatric surgeons discovered that gastric bypass surgery also causes prompt remission of type 2 diabetes [24] independently and ahead of any significant weight loss [25]. The sudden reversal of a prevalent chronic disease that is inherently difficult to treat has naturally prompted much

13 research. The prize catch: to identify the underlying mechanisms and achieve a similar effect without having to undergo a surgical procedure that has numerous side effects and rare but serious adverse effects [26].

Some of the explanation for both weight loss and normalization of HbA1c appears to be an increase of incretin hormones from the distal ileum through a mechanism termed ‘the ileal brake’ where undigested nutrients enter the distal ileum and results in an augmented release of incretin hormones [27]. One such incretin is glucagon-like peptide 1 (GLP-1) that is now marketed as subcutaneous injections for type 2 diabetes treatment [28,29] and shows promise as weight loss treatment in obese patients without diabetes [30].

In the landmark Swedish Obese Subjects (SOS) study the 2-year remission rate of 55 patients who underwent gastric bypass surgery was around 80% but after a median of 17.6 years only 35% remained in remission [31]. Relapse of diabetes was more prevalent in patients with long diabetes duration and the authors therefore recommended that bariatric surgery should be performed early rather than late in the course of type 2 diabetes [31].

Diabetic retinopathy

Diabetic retinopathy continues to be one of the leading causes of vision loss among people in the working age [32]. The prevalence of vision threatening diabetic retinopathy was reported in a study that pooled data from 8 cross sectional studies in the USA to be 8.2% of patients with type 2 diabetes [33] with a higher prevalence in patients with longer duration of diabetes [34]. In the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) study, the 10-year incidence of proliferative diabetic retinopathy in patients with less severe diabetic retinopathy was 17% [35]. Fortunately, vision loss from diabetic retinopathy appears to be declining due to improved diabetes management and improved treatment options for diabetic macular edema [36-39]. However, in absolute terms, the burden of diabetic retinopathy will continue to grow as the number of people with diabetes increases [1].

14

Figure 4. The retina. Adapted from [40].

15 Breakdown of the neurovascular unit

In the retina, the destructive cellular effects of hyperglycemia and possibly the interaction with the supporting Müller cells, neurons and microglia affect an important supporting cell in the retina: the pericytes that surround the endothelium of the inner retinal blood vessels (Figure 4). The loss of pericytes leads to a breakdown of the blood-retina barrier and an influx of macromolecules in the neuronal parenchyma [40]. The interdependence between the neural and vascular tissue is highlighted by the blood flow in the retina being regulated entirely by local metabolite levels, partial pressure of oxygen and Müller cells, known as the retinal autoregulation [41]. Some of the earliest detectable changes in diabetic retinopathy are functional alterations related to the neurosensory retina: impaired dark adaptation [42], reduced electroretinographic responses [43], reduced contrast sensitivity [44] and reduced visual fields [44]. Anatomically, the thinning of the inner retinal layers in early diabetic retinopathy [45] is consistent with loss of retinal ganglion cells and their axons [46]. This has promoted diabetic retinopathy as a disease of the retinal neurovascular unit and not solely as a vascular disease [40]. As a testament to a possible paradigm shift, an ongoing multicenter trial of a new drug for the prevention of diabetic retinopathy was recently approved, with improvements in multifocal electroretinography as the primary outcome, instead of the customary clinical grade of diabetic retinopathy (clinicaltrials.gov NCT01726075).

Clinical diabetic retinopathy

The breakdown of the blood retinal barrier is followed by a long range of events in the retina, that can be appreciated clinically through a dilated pupil using biomicroscopy, fundus photography, fluorescein angiography and, more recently, spectral domain optical coherence tomography (OCT). Diabetic retinopathy can impair vision in two ways: a) non-proliferative or proliferative diabetic retinopathy that can cause intravitreal hemorrhage or scar tissue formation or b) macular edema that can cause swelling of the fovea. In this thesis, the term “diabetic retinopathy” encompasses non-proliferative and proliferative diabetic retinopathy as well as diabetic maculopathy.

Specific retinal signs in patients with diabetes define diabetic retinopathy. The clinically defining feature of diabetic retinopathy is microaneurysms, seen as small red spots in the fundus. Other non-proliferative features follow, such as intraretinal hemorrhages, cotton wool spots, edema with or without hard exudates, venous beading or intraretinal microvascular abnormality (IRMA) vessels. In the proliferative stage, new vessels sprout from the retina, grow into the vitreous cavity, and can give rise to preretinal or intraretinal hemorrhage, scar tissue formation or retinal detachment. The various changes have been grouped

16 empirically forming the gold standard ETDRS scale [47,48]. The ETDRS scale contains 13 steps for one eye, and lesions are graded from 7-field fundus stereo photographs comparing presence of lesions and size/area of lesions according to standard photos.

Screening for diabetic retinopathy

The devastating consequences of blindness make screening for diabetic retinopathy not only worthwhile for the individual but also cost-effective from a public health perspective [49]. Fortunately, vision impairment is a very late event in diabetic retinopathy; it usually takes many years to progress from no diabetic retinopathy to vision threatening disease [50]. Therefore, there is a considerable window of opportunity for timely treatment of disease. In Denmark, the guidelines for screening and treatment are an extension of the American Academy of Ophthalmology guidelines [51,52] which follow this principle: the higher risk of vision threatening disease, the shorter the follow-up. Increasing levels of retinopathy are associated with increasing risk of progression to vision-threatening disease, and these risks determine the appropriate re-screening interval [53]. Table 1 and 2 shows the screening guidelines at the Department of Ophthalmology, Zealand University Hospital Roskilde. The incorporation of more parameters in the screening interval such as duration of diabetes (longer being worse), HbA1c (higher being worse), blood pressure (higher being worse), sex (male being worse), type of diabetes (variable influence) and level of retinopathy at previous screening visit

(fluctuations being worse) are currently being investigated for clinical use [50,54]. Besides HbA1c, no other systemic biomarkers have yet been able to add clinical value in the screening for diabetic retinopathy [55].

17

Table 1. Re-examination intervals of proliferative diabetic retinopathy at Zealand University Hospital Roskilde Disease severity Findings Control interval No apparent retinopathy No abnormalities 12 months Mild non-proliferative Microaneurysms or less than 10 intraretinal dot-blot 12 months diabetic retinopathy hemorrhages Moderate non-proliferative More than mild retinopathy, hard exudates or cotton-wool 12 months diabetic retinopathy spots and less than preproliferative retinopathy Preproliferative diabetic Any of the following: more than 20 intraretinal hemorrhages in 3 months retinopathy each of 4 quadrants; definite venous beading in 2 quadrants; Prominent intraretinal microvascular abnormalities in 1 quadrant and no signs of proliferative retinopathy

Proliferative diabetic One or more of the following: Treat with laser retinopathy neovascularization, vitreous/preretinal hemorrhage, iris neovascularization Previous laser Peripheral scars of panretinal photocoagulation but no 6 months photocoagulation, stable neovascularization Previous laser Peripheral scars of panretinal photocoagulation and active Treat with laser photocoagulation, unstable neovascularization

Table 2. Re-examination schedule of diabetic maculopathy at Zealand University Hospital Roskilde Disease severity Findings Control interval No maculopathy No retinal thickening or hard exudates within the 12 months vessel arcades Mild diabetic Retinal thickening or hard exudates more than 1500 6 months maculopathy µm (one disc diameter) from the center of the fovea Moderate diabetic Retinal thickening or hard exudates 500–1,500 µm 3 months maculopathy from the center of the fovea Clinically significant Retinal thickening or hard exudates <500 µm) from the Treat with anti-VEGF injection. macular edema center of the fovea or an area of thickening more than Foveal sparing: consider central 1,500 µm in diameter laser Previous macular laser Scars from macular photocoagulation, but no clinically 3-6 months photocoagulation, significant macular edema stable Previous macular laser Scars from macular photocoagulation and clinically Treat with anti-VEGF injection. photocoagulation, significant macular edema Foveal sparing: consider central unstable laser

Early Worsening of diabetic retinopathy

Unfortunately, the progression of diabetic retinopathy can be rapid in some patients. Especially under metabolic stressful conditions and in particular when high blood glucose is lowered abruptly. This paradoxical

18 phenomenon is termed early worsening. In type 1 diabetes early worsening was described after intensified glucose treatment goals in the Diabetes Control and Complications Trial (DCCT) [56] and has also been reported during pregnancy [57], after commencing insulin pump [58-60] or after pancreas transplantation [61,62]. In type 2 diabetes worsening has also been reported during pregnancy [63], after commencement of insulin therapy [64-66] or after commencing incretin-based therapy [67,68]. A case report has described worsening in a patient with type 2 diabetes who had severe non-proliferative diabetic retinopathy and clinically significant macular edema who developed proliferative diabetic retinopathy after gastric bypass surgery [69]. Proposed mechanisms for early worsening are:

a) Growth hormone (GH)/insulin-like growth factor-1 (IGF-I) increase. This theory proposes, that a fall in blood glucose causes a reactive surge in systemic IGF-I that acts through the diabetes-degraded blood retina barrier to promote neovascularizations [70-72]. Trials with recombinant IGF-I as an adjunct to insulin reported dose-dependent increases in optic disc swelling and early worsening in both type 1 [73] and type 2 diabetes [74]. Results from trials with GH or IGF-I antagonists for prevention of proliferative diabetic retinopathy have shown promise [75-79], though not unanimously [80,81]. However, all of these studies had less than 25 patients and most included only patients with type 1 diabetes. b) Protracted adaptation to substrate deficiency. The retina relies heavily on glucose and oxygen supply to maintain function. When blood glucose is lowered, the Müller cells of the retina may experience a relative substrate deficiency responding with an increased VEGF release until vascular homeostasis is gradually restored [82]. c) Reduced blood flow. Decreased blood glucose levels and increased insulin levels cause a decrease in blood flow through autoregulation [83] and the decreased blood flow could be responsible for a relative hypoxia, precapillary vasodilation and ultimately breakdown of the blood retina barrier [84]. However, the changes in flow appear both during slow and rapid changes in blood glucose levels [85]. d) Insulin or insulin analogues. Insulin increases intravitreal hypoxia inducible factor-1 transcription that is followed by increased VGEF mRNA and promotes the breakdown of the blood-retinal barrier in rodents [86]. Both insulin and insulin analogues bind with some affinity to the IGF-I receptor [87] and the role of insulin could be an extension of the growth hormone hypothesis above. However, with the affinity of human insulin to the IGF-I receptor indexed to 100%, the affinity of commercially available insulin analogues ranges up to 641% for insulin glargine [88] while human IGF-I has an affinity of up to 75,428% for comparison [87]. However, no difference in worsening of diabetic

19 retinopathy were seen in the registration trials for insulin glargine (Lantus®) compared to human neutral protamine Hagedorn insulin that has lower affinity for the IGF-I receptor [89].

On a cohort level, the long-term effect of improved glucose control appears to outweigh the risk of early worsening of diabetic retinopathy [56], but progression to transient or permanent vision impairment has also been reported [58,90,91].

The IGF axis

IGF-I is involved in the local release of VEGF and as described above, systemic IGF-I and IGF binding proteins (IGFBP) have been associated with diabetic retinopathy with some discrepancy in the results reported in the literature [72]. Vitreous levels of IGF-I are elevated in patients with diabetic retinopathy [92-94]. Total IGF-I can be measured using a commercially available immunoradiometric assay method, but approximately 99% of IGF-I is bound to IGFBP’s, with approximately 75% being bound to IGFBP-3 [95]. The measurement of total IGF-I ignores the binding and modifying effects of IGF-II, the IGFBPs and proteases [96], and this could be the basis of the conflicting results for the association between the IGF-I system and diabetic retinopathy. An alternative approach is to measure bioactive IGF using an IGF-I kinase receptor activation (KIRA) assay that measures the ability of serum to phosphorylate (activate) the IGF-I receptor (IGF-IR) in transfected cell cultures [96] and integrates the whole IGF system into one signal.

Bariatric surgery and end-organ complications

Improvement of hyperglycemia and improvement of dyslipidemia after bariatric surgery are well known, but the important question for the patient is if this translates into a reduction in end-organ disease [97]. Results from non-randomized reports of macrovascular disease [98] and mortality [99] have been positive, but randomized studies with long follow-up are lacking. Few studies have examined microvascular disease. A prospective case-control study of 50 patients undergoing biliopancreatic diversion with 10 years of follow- up reported a positive effect on surgery on microalbuminuria [100] and another prospective case-control study of 70 patients undergoing gastric bypass surgery found improved urine albumin-creatinine ratio in the surgical group [101]. Observational studies have also found improved albumin-creatinine ratio [102,103] and serum creatinine [103]. The only study of diabetic neuropathy was a prospective study reporting no difference between 54 patients undergoing gastric bypass surgery and a matched control group [101]. Recently, a prospective randomized trial found no changes in the development of diabetic retinopathy between the gastric bypass surgery group, the sleeve gastrectomy group and the control group after two

20 years of follow-up [104]. Register-based studies on composite ocular end-points indicate that ocular complications are lowered after bariatric surgery [31,105]. In other studies, the incidence of worsening of diabetic retinopathy in patients with preexisting diabetic retinopathy has ranged from 3% (1 of 32 patients, [101]) to 88% (7 of 8 patients, [106]).

21 Objectives

The objective of this thesis was to examine the perioperative changes in diabetic retinopathy and related biomarkers after bariatric surgery. We approached this from different angles.

Clinical changes in diabetic retinopathy after bariatric surgery (Paper I):

• Assess if a patient with type 2 diabetes who undergo bariatric surgery should be monitored more closely in relation to the surgery than any other patient with diabetes.

Subclinical changes in the diabetic retina after bariatric surgery (Paper II):

• Investigate the background for increased thickness found in Paper I by analyzing layers and regions of the retina in more detail and retinal vessel calibers.

Biomarkers associated with diabetic retinopathy or changes in diabetic retinopathy after bariatric surgery (Paper III):

• Compare baseline levels of biomarkers from the IGF-I system between patients with and without diabetic retinopathy.

• Assess changes in the IGF-I system after gastric bypass surgery and relate these to changes in diabetic retinopathy.

22 Methods

Study design

We recruited patients from the department of bariatric surgery at Zealand University Hospital Køge October 6, 2011–December 3, 2013. We examined the patients with eye examinations, blood pressure measurement and fasting blood samples approximately 14 days prior to their bariatric surgery. The examinations were repeated one, three, six and twelve months after the surgery. Fasting blood samples were taken three and twelve months after surgery. Some patients were enrolled before the blood sampling procedure had been approved and set up and could therefore not be included for Paper III. The time course is presented Figure 5.

2011 2012 2013 2014 2015

Inclusion and follow-up Inclusion and follow-up Follow-up Analysis and including blood samples dissemination

Figure 5. Study recruitment and examinations in calendar time.

Patients

To be considered for bariatric surgery, patients with a history of type 2 diabetes or newly diagnosed diabetes had to have a BMI above 35 kg/m2. Furthermore, they needed to undergo a preoperative weight loss of 8% of their body weight to become eligible for surgery and participation in our study. We excluded patients who failed to attend more than one postoperative examination or withdrew consent. Within the recruitment period, 476 patients underwent bariatric surgery and 96 of those had type 2 diabetes. All medication was adjusted at the discretion of the patient’s general practitioner or the endocrinologists at the bariatric department. For Paper II we excluded patients who had moderate diabetic retinopathy or worse, as these patients are suspected to have different thickness profile of the retina [107]. For Paper III we excluded patients that had vertical sleeve gastrectomy as some studies have found different metabolic changes between the two procedures, such as improved pancreatic β-cell function in gastric bypass surgery [108].

23 Examinations and grading

The patients underwent examinations at each visit as per Table 3.

Table 3. Examination protocol Months from surgery Examination - 0.5 +1 +3 +6 +12 ETDRS visual acuity x x x x x 7-field fundus photos x x x x x OCT x x x x x Fundus biomicroscopy x x x x x Blood pressure x x x x x Weight x x x x

HbA1c, lipids x x x x Fasting biomarkers (Paper III) x x x

Eye-examinations

The patients had thorough eye-examination with visual acuity measured at a distance of 4 m on an ETDRS chart (CC-100, Topcon Corporation, Tokyo, Japan, Figure 6) that is more accurate than the Snellen chart and is superior for statistical analysis [109]. The maximum number of letters being 100 (Snellen 12/6) and 85 letters equaling Snellen 6/6. After dilation of the pupil with topical 1% tropicamide we obtained 7-field digital non-stereo color fundus photographs (Visucam NM, Carl Zeiss Meditec, Jena, Germany [110]).

At each visit, seven 45° fundus images were taken using the ETDRS-grid: field 1 centered on the optic disc, field 2 centered on the macula, field 3 temporal to the macula and field 4-7 tangential to horizontal lines above and below the disc and to a vertical line passing through its center (Figure 7). Furthermore, we took a green-filtered (“red-free”) image centered on the optic disc. This was followed by a 15 x 20° vertical raster fovea centered spectral domain OCT scan (Heidelberg Spectralis, Heidelberg Engineering GmbH, Heidelberg, Germany) with automated eye-tracking (TruTrack™) followed by slit lamp and fundus biomicroscopy examination. Only OCT scans with a quality of 25 dB or more were accepted and scan focus was confirmed to be within 2 diopters of the first scan.

24

Figure 6. ETDRS visual acuity chart. Figure 7. Seven-field ETDRS fundus grid.

Other examinations

We measured blood pressure twice after ten minutes of rest (Omron M5-1, Omron Healthcare Co. Ltd., Kyoto, Japan). Weight was measured with clothes but without shoes (ADE M301020, ADE Gmbh, Hamburg,

Germany). HbA1c was measured at the hospital laboratory using high performance liquid chromatography (Tosoh G8, Tosoh Bioscience, San Francisco, USA).

Grading of fundus images

The fundus images were exported from the proprietary Zeiss Visupac 4.4.1 (Carl Zeiss Meditec, Jena, Germany) database as tiff files to Adobe Photoshop Lightroom 5.7 (Adobe Systems, San Jose, CA, USA) where patient id was masked and each image set was given a unique random identifier generated in Excel 2007 (Microsoft Excel 2007, Microsoft, Redmond, USA) using the RAND function. Each grader graded all image sets from one eye together but was blinded to the sequence and status of the fellow eye. It was allowed to adjust contrast and brightness in each image as deemed appropriate by the grader and it was allowed to manually “sharpen” the image up to 50% [111] and zoom to a maximum of 100% for evaluation of finer details. After completion of the grading, we lifted the masking for further analysis.

25 The fundus images were graded using the ETDRS scale [47,112] with WESDR modifications [113]. Each grader assigned one of 13 levels between no retinopathy (level 10) and vitreous hemorrhage (level 85) for each eye, depending on the presence or absence of features of diabetic retinopathy. Level 60 and above were treated as one collective level (Table 4).

Table 4. Modified Wisconsin Epidemiologic Study of Diabetic Retinopathy grading scale. Level Description 10 No retinopathy. 21 Microaneurysms only, or retinal hemorrhages or soft exudates in the absence of microaneurysms. 31 Microaneurysms and 1 of the following: venous loops 31 µm; questionable soft or hard exudates, intraretinal micro-vascular abnormalities (IRMAs), or venous beading; or retinal hemorrhages. 37 Microaneurysms and one or both of the following: hard or soft exudates. 43 Microaneurysms and ≥1 of the following: hemorrhages/microaneurysms equaling or exceeding those in standard photograph 1 in 4 or 5 fields, hemorrhages/microaneurysms equaling or exceeding those in SP 2A in 1 field, or IRMAs in 1 to 3 fields. 47 Microaneurysms and 1 of the following: both IRMAs and hemorrhage/microaneurysm characteristics from level 43, IRMAs in 4 or 5 fields, hemorrhages/microaneurysms equaling or exceeding those in standard photograph 2A in 2 or 3 fields, or venous beading in 1 field. 53 Microaneurysms and 1 of the following: any 2 or 3 characteristics from level 47, hemorrhages /microaneurysms equaling or exceeding those in standard photograph 2A in 4 or 5 fields, IRMAs equaling or exceeding those in standard photograph 8A, or venous beading in ≥2 fields. 60+ Any of several levels of severity of proliferative retinopathy, including neovascularisations, fibrous proliferations, vitreous and pre-retinal hemorrhage, or obscured field of view due to vitreous hemorrhage.

The eyes were combined for a composite level emphasizing the grade of the worst eye [35]. For instance, a patient without signs of retinopathy in either eye is level 1 (10/10). A patient with microaneurysms in one eye and no signs of diabetic retinopathy in the other eye is level 2 (21/<21). A patient with microaneurysms and cotton wool spots in one eye and no signs of retinopathy in the other eye is level 6 (37/<37) and so forth up to level 15 (60+/60+). Change in diabetic retinopathy was defined as a two-step change from the preoperative level, the smallest change on the scale considered clinically meaningful [113].

Presence of macular edema (Paper I)

We evaluated diabetic macular edema using both fundus images and OCT. On fundus images the presence or absence of clinically significant macular edema were noted - as per the definition of the ETDRS (Figure 8, [114]). OCT replaced stereopsis for determination of thickening without hard exudates. On OCT we used the criteria of the Diabetic Retinopathy Clinical Research Network [115] adjusted 65 µm for Heidelberg

26 equipment [116]. Center-involving macular edema was therefore a center point foveal thickness above 365 µm and subclinical macular edema was a thickness of 290–365 µm or an increase of 50 µm compared to the preoperative level. The center point thickness was measured as the minimal thickness in a fovea centered circle with a diameter of 1 mm. Central subfield thickness was defined as the mean thickness between Bruch’s membrane and the internal limiting membrane of this circle while mean macular thickness was measured in a fovea-centered circle with a diameter of 6 mm. The value was calculated from the volume measurement using cylinder geometry: mean macular thickness = (macular volume mm3 / π x (3 mm)2). We also manually examined each OCT scan for signs of maculopathy.

Figure 8. Definition of clinical significant macular edema from the ETDRS study. Left: Retinal thickening within 500 µm of the centre of the fovea. Middle: Hard exudates within 500 µm of the centre of the fovea if associated with retinal thickening. Right: Retinal thickening measuring more than one disc area within 1,500 µm from the centre of the fovea [117].

Measurement of retinal layers (Paper II).

Eight retinal layers were defined using an automated proprietary software algorithm (Heidelberg Eye Explorer 1.9.10.0, Heidelberg Engineering GmbH, Heidelberg, Germany) with manual verification of each retinal layer (Figure 9). The layers are: the retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer including Henle’s layer (ONL), outer segment of the photoreceptors (OS) and retinal pigment epithelium (RPE). Three concentric regions were defined based on the ETDRS macula grid centered on the fovea [47]: the fovea of 1 mm in diameter, the parafoveal ring 1–3 mm from the umbo and the perifoveal ring 3–6 mm from the umbo (Figure 10). For the calculation of total parafoveal and perifoveal volume, we solved for each volume in the ETDRS grid. For instance, parafovea = total macula – (perifovea + fovea). This revealed that in the ETDRS grid,

27 the fovea is 1/36 of the macula, the parafoveal ring 2/9 of the macula and the perifoveal ring 3/4 of the total ETDRS-defined macula. All automatic OCT segmentations were manually confirmed. Automated segmentation of OCT layers is known to be affected by software inaccuracies in the presence of macular pathology [118] but is less of an issue in the absence of macular pathology [119].

RNFL GCL IPL INL OPL ONL OS RPE

Figure 9. Top: Fovea centered OCT scan. Bottom: Automated delineation of the retinal layers of the same OCT scan. Henle’s fibre layer is not distinguishable, but is located within the outer nuclear layer.

28 A B C

Figure 10. Defintion of fovea (A), parafovea (B) and perifovea (C) in the ETDRS macula grid within Heidelberg Explorer. Mean total thickness for each specified individual layer is calculated automatically in the right panel in black digits (red arrow).

Measurement of vessel calibers (Paper II)

Vessel calibers were traced using the “big six” method, where the six largest arterioles and the six largest venules 0.5–1 disc diameters from the rim of the optic nerve are measured with a semi-automated algorithm (Visiopharm 2.1.3.5, Hørsholm, Denmark). The measures are summarized in a central retinal artery equivalent (CRAE) and a central retinal vein equivalent (CRVE) using an empirically derived formula [120,121] that is a rough estimation of the caliber of the central retinal artery and vein. All vessel caliber measurements were performed in the right eye unless amblyopia or media opacities made the left eye the better choice (n=4).

29

Figure 11. Trace of the six largest arterioles and venules 0.5–1 disc diameter from the rim of the optic disc.

Fasting biomarkers (Paper III)

After an overnight fast, patients had blood samples drawn from an antecubital vein. HbA1c, glucose, LDL, HDL and triglycerides were measured in fresh blood samples at the hospital laboratory. Further samples were centrifuged and transferred to microtubes and stored in a logged and monitored freezer at -80 °C until being analyzed in duplicates en bloc. Bioactive IGF was determined by an in-house kinase receptor activation assay based on human embryonic renal cells (EBNA 293) transfected with the human IGF-I receptor gene [96,122]. Total IGF-I and IGFBP-3 were assayed using chemilumiscence in an IDS-iSYS multi-discipline automated analyzer (Immunodiagnostic Systems Ltd, Tyne & Wear, England). Serum insulin was measured by a time- resolved immunofluorometric assay (AutoDELFIA insulin; PerkinElmer Life Sciences, Waltham, USA). HOMA- IR was calculated using the HOMA2 calculator for Microsoft Excel [123]. The IGF-I/IGFBP3 molar ratio was calculated using the molecular weights of IGF-I and IGFBP3 of 7.65 and 30.5 kDa respectively.

30 Informed consent

Informed oral and written consent was obtained from all participants prior to inclusion in the study. The study adhered to the Declaration of Helsinki and was approved by the local ethical committee #SJ-205 and #SJ-205-1.

Statistical analysis

All patient data were recorded in an Excel spreadsheet and transferred to SAS 9.4 (SAS Institute, Cary, NC, USA) for further analysis. Figures of changes in biomarker were created using Excel 2007 based on the values obtained from SAS. We compared means (proc means) and made histograms (proc univariate) to test for parametric distribution and overall time-effect was visualized per parameter (proc gplot). Values that were skewed towards the lower end were logarithm transformed. In all cases, this resulted in a parametric distribution or a distribution that was close enough to parametric to be considered parametric.

Baseline comparisons were made with unpaired Student’s t-test for continuous variables (proc ttest) or Fisher’s exact test for binary or categorical variables (proc freq) and twelve-month changes for non-ocular outcomes were calculated using paired t-test (proc ttest /paired) and McNemar’s exact test (proc freq /agree) for related samples. To adjust for confounders, baseline analysis of continuous variables was performed using multiple regression analysis with stepwise backwards elimination (proc reg /selection=backwards).

Follow-up measurements were compared to baseline in a mixed model (proc mixed) in Paper I+II. Two mixed models were created: without and with confounders. We also created a model that included both eyes (subject=side(patient)). In Paper III, follow-up measurements were analyzed using paired t-tests with histogram confirmation of parametric distribution and multiple regression to adjust for confounding.

Interobserver agreement was determined using weighted kappa statistics (proc freq /agree) with default Cicchetti-Allison weights.

Power analysis for theoretical sample size was performed using independent proportions (proc power /twosamplefreq) using Pearson Chi approximation.

All reported p-values are two-sided.

31 Results

Recruitment (Paper I+II+III)

The 56 patients who completed Paper I completed 95% (266 of 280) of the protocol-specified follow-up visits. The recruitment and patient flow is presented in Figure 12.

96 eligible patients with type 2 diabetes had achieved 8% preoperative weight loss and a stable HbA1c

6 were not referred for unknown reasons 16 declined to participate 11 were referred too late for participation

63 patients included in the study

7 withdrew consent or were lost to follow-up

56 patients included in Paper I

16 were examined prior to blood sampling 4 had WESDR grade >7 3 had vertical sleeve gastrectomy 1 had missing baseline OCT 1 was suspected of improper fasting

51 patients included in Paper II 36 patients included in Paper III

Figure 12. Inclusion and exclusion of patients in Paper I, II and III.

32 Preoperative characteristics (Paper I+II+III)

We compared data from the 56 participants of Paper I with the referral data of the 40 patients with diabetes who had bariatric surgery but were not included in Paper I (Table 5). The patients who participated were less obese (p<0.0001) and had longer duration of diabetes (p=0.005) and more were taking subcutaneous incretin based therapy (liraglutide, p=0.02) and fewer were controlled with lifestyle intervention (p=0.05).

Table 5. Preoperative characteristics of participants and non-participants, mean±SD or median(IQR) or n(%) Characteristic Participants Non-participants p-value n 56 40 Age 47 (±8) 46 (±7) 0.50 Male sex 23 (41) 14 (36) 0.67 Current smoker 21 (35) 15 (36) 1.00 BMI (kg/m2) 39.6 (±6.1) 47 (±10) <0.0001 Time to achieve preoperative weight loss (months) a 4 (3–5) 4 (2–7) 0.30 Duration of diabetes (years)a 4 (2–10) 2 (0–6) 0.002 Glucose control

HbA1c, (%) 6.4 (±0.9) 6.8 (±1.2) 0.13 s. c. insulin therapy 12 (21) 6 (15) 0.60 s. c. incretin based therapy 21 (38) 9 (23) 0.18 Oral antidiabetic therapy 49 (88) 30 (77) 0.26 Lifestyle intervention only 3 (5) 8 (21) 0.05 Lipid control LDL cholesterol (mmol/l) 2.3 (±0.8) 2.7 (±1.1) 0.05 HDL cholesterol (mml/l) 1.0 (±0.2) 1.0 (±0.2) 0.82 Total cholesterol (mmol/l) 4.1 (±0.9) 4.5 (±1.3) 0.13 Triglycerides (mmol/l) 1.7 (±1.0) 2.0 (±1.2) 0.19 On lipid lowering medication 23 (41) 14 (36) 0.67 Blood pressure control On blood pressure lowering medication 32 (57) 22 (56) 1.00 Student’s unpaired t-test and Fisher’s exact test. a logarithm transformed for analysis.

Changes in non-ocular outcomes (Paper I+II+III)

As expected, bariatric surgery was associated with a decrease in HbA1c, a reduction in medicine use and a reduction in BMI (Table 6). The percentage of patients with hypertension was reduced but failed to reach significance (p=0.17). The diastolic, but not the systolic, blood pressure was lowered with a small but

33 significant amount (p=0.01). The reduction in diastolic blood pressure was most significant in patients who had the highest blood pressure preoperatively (p=0.01, simple linear regression).

Table 6. Pre- and postoperative non-ocular outcomes for all patients (n=56), mean±SEM or n(%) Outcome Preoperatively 12 months postoperatively p HbA1c (%) 6.5±0.1 6.0±0.1 <0.0001 Medication Insulin therapy 13 (23) 4 (7) 0.003 Incretin therapy 22 (39) 0 (0) <0.0001 Oral antidiabetic therapy 51 (91) 22 (39) <0.0001 Only lifestyle intervention for diabetes 3 (5) 33 (59) <0.0001 On blood pressure lowering medication 32 (57) 26 (46) 0.03 On lipid lowering medication 24 (43) 18 (32) 0.01 Systolic blood pressure (mmHg) 126.1±1.2 126.9±1.8 0.58 Diastolic blood pressure (mmHg) 85.5±0.9 83.2±1.1 0.01 Hypertension 38 (68) 33 (59) 0.17 BMI (kg/m2) 39.6±0.8 31.4±0.8 <0.0001 p-values: Student’s paired t-test and McNemar’s exact test. Hypertension was defined as a diastolic blood pressure > 90, a systolic blood pressure > 140, or as the patient receiving antihypertensive medication [124].

Changes in diabetic retinopathy (Paper I)

Of the patients with pre-existing retinopathy (Table 7), there were 5–13% who had worsening at any visit and 9–22% who had improvement at any visit. Both worsening and improvement had a tendency to occur at the visits furthest from the surgery. When all changes were analyzed in a mixed model there was a significant improvement 6 months postoperatively in the patients with preexisting retinopathy (p=0.01 in both univariate and multivariate analysis).

Of the patients without diabetic retinopathy preoperatively (Table 8), only one developed a transient worsening six months postoperatively. The lack of variation from the baseline values was so prevailing, that we could not analyze for changes in our mixed model.

34 Table 7. Evolvement in diabetic retinopathy for the 24 patients with diabetic retinopathy prior to bariatric surgery Preoperatively Postoperatively Patient -2 weeks 1 month 3 months 6 months 12 months 17 2 2 2 2 2 37 2 2 2 2 2 38 2 2 2 2 2 47 2 2 2 2 52 2 1 2 43 2 2 1 1 1 53 2 2 1 1 1 51 2 2 4 (W) 3 3 56 2 2 6 (W) 6 (W) 19 3 2 3 3 2 49 3 3 1 (I) 1 (I) 1 (I) 6 3 3 2 4 6 (W) 7 4 3 3 3 3 3 4 4 2 (I) 1 (I) 30 4 6 (W) 6 (W) 2 (I) 2 (I) 32 5 5 5 5 5 44 6 4 (I) 3 (I) 6 6 50 6 4 (I) 6 1 (I) 2 (I) 61 6 6 6 6 6 5 7 6 8 6 6 26 8 8 8 8 8 25 14 14 14 14 14 15 15 15 15 15 15 (W)b 33 15 15 15 15 (I)a 15 (I)a Unchanged (n, %) 19 (86) 18 (78) 18 (78) 17 (74) Worsening (n, %) 1 (5) 2 (9) 1 (4) 3 (13) Improvement (n, %) 2 (9) 3 (13) 5 (22) 4 (17) Retinopathy compounded for both eyes on a 15-step severity scale with 1 being no retinopathy and 14–15 being proliferative diabetic retinopathy requiring immediate treatment. (W) = worsening compared to the preoperative level. (I) = improvement compared to the preoperative level. Blank indicates a missed visit. aResolution of clinically significant macular edema b Vitreous hemorrhage.

35 Table 8. Evolvement in diabetic retinopathy for the 32 patients without diabetic retinopathy prior to bariatric surgery Preoperatively Postoperatively Patient -2 weeks 1 month 3 months 6 months 12 months 1 1 1 1 1 1 2 1 1 2 3 (W) 2 4 1 1 1 1 2 8 1 1 1 1 1 9 1 1 1 1 1 10 1 1 1 1 1 11 1 1 1 1 1 12 1 1 1 1 1 13 1 1 1 1 14 1 1 1 1 1 18 1 1 1 1 1 20 1 1 1 1 21 1 1 1 1 1 22 1 1 2 2 1 24 1 1 1 1 1 28 1 1 1 1 1 29 1 1 1 1 31 1 1 1 2 1 34 1 1 1 1 35 1 1 1 1 1 39 1 1 1 1 41 1 1 1 45 1 1 1 1 1 54 1 1 1 1 1 55 1 1 1 1 1 57 1 2 1 58 1 1 1 1 1 59 1 1 1 1 1 60 1 1 1 2 1 61 1 1 1 1 2 62 1 1 1 1 1 63 1 1 1 1 Unchanged (n, %) 30 (100) 30 (100) 28 (97) 29 (100) Two-step worsening (n, %) 0 (0) 0 (0) 1 (4) 0 (0) Retinopathy compounded for both eyes on a 15-step severity scale with 1 being no retinopathy and 14–15 being proliferative diabetic retinopathy requiring immediate treatment. (W) = worsening of two steps compared to the preoperative level. Blank indicates a missed visit.

36 Of all patients (Table 9) 11% (n=6) had worsening of diabetic retinopathy at at least one timepoint and for 5% (n=3) the worsened level of retinopathy persisted at one year postoperatively. There were also 11% (n=6) who had improvement at any visit but 7% (n=4) that remained improved at one year.

Table 9. Overall change in retinopathy level compared to baseline 1 month 3 months 6 months 12 months At one visit, n (%) Unchanged 49 (94) 48 (91) 47 (87) 46 (87) Improvement 2 (4) 3 (6) 5 (9) 4 (8) Worsening 1 (2) 2 (4) 2 (4) 3 (6) Cumulative, n (%) a Unchanged 53 (95) 50 (89) 47 (84) 45 (80) Improvement 2 (4) 4 (7) 6 (11) 6 (11) Worsening 1 (2) 2 (4) 4 (7) 6 (11) Cumulative steps b -4 -3 -6 -5 a One patient had both worsening and improvement and is counted as both from visit 6 onwards. b Total change on the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) scale including one-step changes. Negative values indicate clinical improvement. The cases of improvement and worsening that were not reflected in the WESDR scale were not included in cummulative analysis.

The interobserver agreement before resolution of disagreements were 0.78 weighted kappa for the 507 image sets which is substantial agreement [125]. Simple kappa was 0.49, which is moderate agreement [125].

Analysis of individual cases (Paper I)

We analyzed the images of the patients who had progression and worsening and did not find any patients where we needed to change the screening interval [52,126]. The grading scale steps may seem large, but changes often reflected features that were minimally present or absent and most of the changes were clinically mild. One patient both had improvement and worsening, highlighting the unpredictable nature of the disease. One patient had proliferative diabetic retinopathy in both eyes pre-operatively but progressed to vitreous hemorrhage 10 months postoperatively. We counted this patient as a case of worsening despite not changing WESDR level (Figure 13).

37 Pre 1 Pre

3 6

Figure 13. Crop of fundus images (left) and early phase flourescein angiography (right) of patient 15 with proliferative diabetic retinopathy. The patient received pan-retinal laser photocoagulation but had a vitreous haemorrhage 10 months postoperatively. The proliferation was most dilated at the postoperative visits; a sign of active disease. Pre = preoperative visit. 1, 3, and 6 = postoperative months.

Clinical changes in maculopathy (Paper I+II)

No patients met the OCT criteria for center involving macular edema but one patient had clinically significant macular edema pre-operatively that regressed at six months. We counted this patient as improvement six and twelve months postoperatively (Figure 14).

Figure 14. Fundus photo of patient 33 with hard exudates (arrows) preoperatively indicating leakage and edema. There is a gradual remission of the hard exudates postoperatively. Scars from laser photocoagulation 3 years earlier are scattered throughout the retina. Pre = preoperative visit. 1, 3, 6, and 12 = postoperative months.

38 We analyzed the OCT data in a mixed model. There was a retinal thickening that peaked six months postoperatively that was evident in all three parameters measured (Table 10). The fovea also thickened at three and twelve months, albeit less significant.

Table 10. Changes in retinal thickness after bariatric surgery (mean µm) Model 1 month p 3 months p 6 months p 12 months p Centre Uni 0.3±0.6 0.66 1.7±0.6 0.003 3.5±0.6 <0.001 2.3±0.6 <0.001 Multi -0.9±0.8 0.28 1.6±0.8 0.07 3.0±0.9 0.001 2.1±0.9 0.02 Fovea Uni 0.2±0.5 0.65 1.6±0.5 0.001 3.1±0.5 <0.001 1.7±0.5 0.001 Multi -0.7±0.7 0.36 1.7±0.7 0.01 3.0±0.8 <0.001 1.6±0.8 0.04 Macula Uni 0.0±0.4 0.98 0.5±0.4 0.16 1.6±0.4 <0.001 0.1±0.4 0.80 Multi -0.5±0.5 0.34 0.9±0.5 0.09 2.2±0.6 <0.001 0.9±0.6 0.10 Mixed model of the foveal and macular changes after bariatric surgery incorporating both eyes. The final multivariate model adjusted time effect of duration of diabetes, retinopathy level and change in blood pressure, HbA1c change and BMI. Centre: centre point thickness. Fovea: central 1 mm mean thickness. Macula: central 6 mm mean thickness. Uni: Univariate model. Multi: multivariate model.

Exploratory analysis of subclinical changes in retinal thickness (Paper II)

The changes in Table 10 formed the background for the exploratory analysis in Paper II. As the changes for right and left eye were similar, we studied the right eye extensively in the post-hoc analyses. Patients with moderate diabetic retinopathy or worse (WESDR<7) were excluded for this paper. This resulted in 52 patients for further analysis in two mixed models: a univariate and a multivariate model. We made a preliminary multivariate model where we adjusted for the time effect of several possible preoperative confounders: age, sex, duration of diabetes, retinopathy level, change in BMI, HbA1c and blood pressure. After removal of insignificant confounders, the final multivariate model adjusted for the time effect of preoperative retinopathy status and changing HbA1c and blood pressure. The results appear in Table 11 and Table 12.

In the univariate analysis the total macular mean thickness increased 2.7 µm (±0.4, p<0.001) at six months and 1.3 µm (±0.4, p=0.004) at twelve months postoperatively. The RNFL (p=0.001), IPL (p<0.001) and ONL (p<0.001) contributed most to the thickening of the parafoveal region and the ONL of the perifoveal region (p<0.001). The total thickness also increased significantly at three and twelve months in the parafovea, but not in the perifovea.

39

40 40 Table 11. Univariate model of changes in mean thickness of macular regions and layers after bariatric surgery (µm) 1 month post-operatively 3 months post-operatively 6 months post-operatively 12 months post-operatively Estimate SE p Estimate SE p Estimate SE p Estimate SE p Regional thickness Fovea -0.1 0.8 0.94 2.5 0.8 0.002 3.5 0.8 <0.001 2.6 0.8 0.001 Parafovea 0.2 0.4 0.68 1.8 0.4 <0.001 3.3 0.4 <0.001 1.9 0.4 <0.001 Perifovea 0.3 0.4 0.47 0.9 0.4 0.05 2.4 0.4 <0.001 1.0 0.4 0.03 Total macula 0.3 0.4 0.51 1.2 0.4 0.007 2.7 0.4 <0.001 1.3 0.4 0.004 Parafoveal layers RNFL 0.1 0.2 0.54 0.2 0.2 0.36 0.6 0.2 0.001 0.4 0.2 0.03 GCL 0.1 0.4 0.73 -0.4 0.4 0.33 0.5 0.4 0.29 0.4 0.4 0.35 IPL 0.2 0.2 0.22 0.6 0.2 0.001 0.6 0.2 <0.001 0.3 0.2 0.05 INL -0.5 0.2 0.04 -0.4 0.2 0.07 -0.6 0.3 0.02 -0.5 0.3 0.06 OPL 0.2 0.5 0.65 0.8 0.5 0.10 -0.3 0.5 0.56 0.1 0.5 0.77 ONL 0.3 0.5 0.46 0.1 0.5 0.77 1.9 0.5 <0.001 1.0 0.5 0.05 OS -0.2 0.5 0.65 1.0 0.5 0.03 0.6 0.5 0.20 -0.0 0.5 1.00 RPE -0.1 0.1 0.40 0.0 0.1 0.73 0.1 0.1 0.68 0.2 0.1 0.10 Perifoveal layers RNFL 0.0 0.2 0.87 -0.0 0.2 0.91 0.2 0.2 0.39 0.0 0.2 0.88 GCL 0.2 0.1 0.20 0.1 0.1 0.71 0.2 0.1 0.09 0.1 0.1 0.63 IPL -0.0 0.1 0.68 0.1 0.1 0.51 0.1 0.1 0.24 0.1 0.1 0.56 INL -0.2 0.2 0.17 0.2 0.2 0.26 -0.1 0.2 0.48 -0.1 0.2 0.64 OPL -0.1 0.2 0.69 0.1 0.2 0.65 -0.2 0.2 0.24 -0.1 0.2 0.52 ONL 0.7 0.3 0.02 0.5 0.3 0.10 1.7 0.3 <0.001 1.0 0.3 0.001 OS -0.2 0.3 0.46 -0.1 0.3 0.73 0.5 0.3 0.05 -0.1 0.3 0.81 RPE -0.0 0.1 0.74 0.1 0.1 0.47 -0.0 0.1 0.83 0.1 0.1 0.60 Vessel calibers CRAE 1.2 1.1 0.27 -1.0 1.1 0.36 0.6 1.1 0.57 -0.4 1.1 0.71 CRVE -0.9 1.4 0.51 -0.4 1.4 0.80 -1.2 1.4 0.39 -1.9 1.4 0.16 Fovea: circle of 1 mm in diameter centered on the fovea. Parafovea: the ring 1–3 mm from the centre. Perifovea: the ring 3–6 mm from centre. RNFL: Retinal Nerve fibre layer, GCL: Ganglion cell layer, IPL: Inner plexiform layer, INL: inner nuclear layer OPL: Outer plexiform layer, ONL: Outer nuclear layer and Henle’s fibre layer, RPE: Retinal pigment epithelium. P-values calculated in a mixed model on right eye for repeated measures compared to preoperative levels.

Table 12. Multivariate model of changes in mean thickness of macular regions and layers after bariatric surgery (µm) 1 month post-operatively 3 months post-operatively 6 months post-operatively 12 months post-operatively Estimate SE p Estimate SE p Estimate SE p Estimate SE p Regional Thickness Fovea -1.9 1.1 0.09 1.2 1.1 0.28 2.3 1.2 0.06 1.0 1.2 0.44 Parafovea -0.7 0.6 0.28 1.5 0.6 0.02 3.0 0.7 <0.001 1.8 0.7 0.01 Perifovea -0.6 0.6 0.29 0.3 0.6 0.65 1.8 0.7 0.01 1.2 0.7 0.08 Total macula -0.7 0.6 0.24 0.6 0.6 0.30 2.2 0.7 0.001 1.4 0.7 0.04 Parafoveal layers RNFL 0.2 0.3 0.48 0.4 0.2 0.11 0.8 0.3 0.004 0.5 0.3 0.04 GCL 0.0 0.6 0.98 -0.2 0.6 0.80 0.5 0.7 0.44 0.5 0.7 0.48 IPL 0.3 0.2 0.24 0.3 0.2 0.16 0.5 0.3 0.08 0.0 0.3 0.99 INL -0.6 0.4 0.09 -0.2 0.4 0.62 -0.7 0.4 0.10 -0.4 0.4 0.27 OPL 0.1 0.7 0.88 0.2 0.7 0.73 -0.3 0.8 0.65 -0.6 0.7 0.41 ONL 0.1 0.7 0.94 0.2 0.7 0.74 1.6 0.7 0.03 2.0 0.7 0.01 OS -0.5 0.7 0.46 0.7 0.6 0.26 0.4 0.7 0.61 -0.2 0.7 0.79 RPE -0.0 0.2 0.85 0.1 0.2 0.51 0.3 0.2 0.10 0.2 0.2 0.22 Perifoveal layers RNFL -0.3 0.4 0.40 0.3 0.4 0.37 0.2 0.4 0.65 0.4 0.4 0.27 GCL 0.3 0.2 0.20 -0.0 0.2 1.00 0.2 0.2 0.32 0.1 0.2 0.55 IPL -0.1 0.2 0.67 -0.1 0.2 0.60 0.0 0.2 0.98 -0.1 0.2 0.71 INL -0.2 0.3 0.46 0.4 0.3 0.16 0.1 0.3 0.84 0.0 0.3 0.98 OPL -0.2 0.3 0.51 0.1 0.3 0.78 -0.3 0.3 0.34 -0.4 0.3 0.23 ONL 0.2 0.4 0.56 0.6 0.4 0.18 1.1 0.5 0.03 1.1 0.5 0.02 OS -0.4 0.4 0.29 -1.0 0.4 0.01 0.3 0.4 0.51 -0.2 0.4 0.59 RPE 0.0 0.1 0.81 0.1 0.1 0.59 0.3 0.1 0.03 0.2 0.1 0.28 Vessel calibers CRAE 8.6 8.6 0.32 15.6 8.8 0.08 14.0 9.0 0.12 22.1 8.9 0.01 CRVE 14.8 11.3 0.19 -2.3 11.7 0.84 16.7 11.9 0.16 5.2 11.8 0.66 Fovea: circle of 1 mm in diameter centered on the fovea. Parafovea: the ring 1–3 mm from the centre. Perifovea: the ring 3–6 mm from centre. RNFL: Retinal Nerve fibre layer, GCL: Ganglion cell layer, IPL: Inner plexiform layer, INL: inner nuclear layer OPL: Outer plexiform layer, ONL: Outer nuclear layer and Henle’s fibre layer, RPE: Retinal pigment epithelium. P-values calculated in a mixed model on right eye for repeated measures compared to preoperative levels and adjusted for the time effect of preoperative retinopathy status and changing HbA1c and blood pressure. 41 41

Changes in visual acuity (Paper I)

The best corrected visual acuity improved significantly in the worst eye at all postoperative visits in both the univariate and the multivariate analysis (Table 13 and Figure 15). The best eye was unchanged.

Table 13. Changes in visual acuity after bariatric surgery Eye n Model 1 month 3 months 6 months 12 months Best 56 Univariate 0.5±0.5 -0.3±0.5 0.4±0.6 0.0±0.6 Multivariate 0.5±0.5 -0.4±0.5 0.5±0.5 0.3±0.5 Worst 54 Univariate 2.1±0.5 *** 1.6±0.5 ** 2.3±0.5 *** 3.0±0.5 *** Multivariate 2.1±0.5 ** 1.8±0.5 ** 2.3±0.6 *** 3.0±0.6 *** Measured on the ETDRS visual acuity chart. ** p = 0.01-0.001, *** p<0.001

Figure 15. Number of ETDRS letters per visit for each patient for best eye (left) and worst eye (right). Eyes with amblyopia were excluded in this analysis.

Changes in vessel calibers (Paper I)

The vessel calibers remained stable in the univariate model (Table 11) but CRAE widened 22.1 µm (±8.9, p=0.01) in the multivariate analysis 12 months postoperatively in the multivariate model (Table 12). The CRVE showed similar trend as CRAE, but did not reach statistical significance. An increase in blood pressure was associated with a narrowing of CRAE of 1.0 µm (±0.4, p= 0.01) per 10 mmHg and a narrowing of CRVE that

42

failed to reach significance (0.9 µm ±0.5, p=0.07 per 10 mmHg). Other confounders showed either no association or a weak association with the outcomes.

The IGF axis (Paper III)

The baseline differences in patients with and without diabetic retinopathy appear in Table 14. Patients with diabetic retinopathy were leaner and had higher levels of bioactive IGF.

Table 14. Preoperative characteristics stratified by diabetic retinopathy status, mean±SE, median (IQR) or n (%) No diabetic retinopathy Diabetic retinopathy n 18 18 Age (years) 48.0±1.8 47.7±2.1 0.90 Male sex 9 (50) 8 (44) 1.00 BMI (kg/m2) 40.6±1.3 38.0±1.9 0.13 Duration of diabetes (years) 4 (2-9) 7 (2-13) 0.55 Systolic blood pressure (mmHg) 125±3 127±3 0.64 Diastolic blood pressure (mmHg) 85±2 89±2 0.20 Current smoker 8 (47) 5 (29) 0.48 Glucose metabolism HbA1c (%) 6.3±0.2 6.4±0.2 0.82 Glucose (mmol/L) 6.4±0.2 6.8±0.5 0.49 HOMA-IR 1.7 (1.2-2.3) 2.1 (1.4-2.7) 0.26 Lipids HDL (mmol/L) 1.0±0.1 1.0±0.0 0.37 LDL (mmol/L) 2.1±0.2 2.5±0.3 0.23 Triglycerides (mmol/L) 1.6±0.1 1.7±0.3 0.79 IGF system Bioactive IGF (ng/ml) 0.84 (0.68-0.96) 1.27 (0.87-1.44) 0.02 Total IGF-I (ng/ml) 131 (110-140) 137 (114-179) 0.24 Total IGF-II (ng/ml) 531 (434-645) 566 (513-656) 0.87 IGFBP-1 (ng/ml) 8.9 (5.4-17.6) 6.4 (3.4-10.2) 0.17 IGFBP-3 (ng/ml) 3,643 (3,153-4,912) 4,308 (3,189-4,785) 0.86 Total IGF-I/IGFBP-3 molar ratio 0.14 (0.11-0.16) 0.15 (0.11-0.20) 0.30 IGF-I = insulin-like growth factor I, IGFBP = IGF binding protein, HOMA-IR = homeostasis model assessment of insulin resistance. P-values are either unpaired student’s t-test or Fisher’s exact test. For the t-test, parametric distributions for data presented as median (IQR) were achieved using logarithm transformation.

In multiple linear regression analysis, we adjusted the change in each biomarker for age, sex, duration of diabetes and change in HbA1c and diabetic retinopathy status. This analysis revealed that baseline level of bioactive IGF was 61% higher in women than in men (95% CI 43-82%, p<0.001).

43

After adjustment, the association with diabetic retinopathy status was strengthened in the multivariate analysis (p=0.006) compared to the univariate analysis (p=0.03). The changes in each biomarker were not different for patients with and without diabetic retinopathy or patients who had worsening of diabetic retinopathy, but as there were only three patients who had any worsening in Paper III, the analysis was limited in strength. Furthermore, we found no correlation between BMI and bioactive IGF and the postoperative changes for men and women were similar. We therefore analyzed all postoperative changes together.

Post-operative changes in each biomarker appear in Table 15 and Figure 16 A-L. Bioactive IGF increased by a mean of 0.13 µg/l (95% CI 0.00-0.26, p=0.05) at three months but returned to preoperative levels twelve months postoperatively. IGFBP-1 increased by a mean of 8 ng/ml (95% CI 4-11, p<0.001) at three months and 13 ng/ml (95% CI 9-17, p<0.001) at twelve months. IGFBP-3 decreased by a mean of 401 ng/ml (95% CI 209- 595, p<0.001) three months postoperatively and by 613 ng/ml (95% CI 414-811, p<0.001) twelve months postoperatively. There were no significant changes in total IGF-I, but total IGF-II decreased 32 ng/ml (95% CI 14-50, p=0.001) at three months and 49 ng/ml (95% CI 25-72, p<0.001) at twelve months. Glucose and lipid metabolism parameters all improved significantly at either three and/or twelve months. We have plotted each of the three patients that had worsening of diabetic retinopathy for comparison in Figure 16 A-L.

In multivariate analysis of the postoperative changes, bioactive IGF was no longer significantly elevated at three months (p=0.11). Other significant associations remained significant after adjustment in multivariate analysis.

44

Figure 16 A-F. Postoperative changes in systemic biomarkers after gastric bypass surgery. Error bars = 95% confidence limits for the mean change. represent the values for the three patients who experienced worsening of diabetic retinopathy postoperatively.

45

Figure 17 G-L. Postoperative changes in systemic biomarkers after gastric bypass surgery. Error bars = 95% confidence limits for the mean change. represent the values for the three patients who experienced worsening of diabetic retinopathy postoperatively.

46

Table 15. Postoperative Change in Circulating Biomarkers, mean±SE 3 months 12 months postoperatively p postoperatively P Glucose metabolism HbA1c (%) -0.3±0.1 0.01 -0.5±0.1 <0.001 Glucose (mmol/L) -0.2±0.3 0.50 -1.1±0.2 <0.001 HOMA-IR -0.9±0.2 <0.001 -1.1±0.2 <0.001 Lipids HDL (mmol/L) 0.04±0.02 0.12 0.29±0.03 <0.001 LDL (mmol/L) -0.11±0.13 0.40 -0.23±0.10 0.04 Triglycerides (mmol/L) -0.32±0.06 <0.001 -0.49±0.08 <0.001 IGF system Bioactive IGF (ng/ml) 0.15±0.07 0.03 0.06±0.09 0.52 Total IGF-I (ng/ml) -6.0±4.2 0.16 -8.4±5.4 0.13 Total IGF-II (ng/ml) -32±9 0.001 -49±11 <0.001 IGFBP-1 (ng/ml) 7.7±1.6 <0.001 13.2±2.0 <0.001 IGFBP-3 (ng/ml) -401±95 <0.001 -613±98 <0.001 Total IGF-I/IGFBP-3 molar ratio 0.008±0.005 0.10 0.012±0.005 0.01 IGF-I = insulin-like growth factor I, IGFBP = IGF binding protein, HOMA-IR = homeostasis model assessment of insulin resistance. P-values: student’s paired t-test.

47

Discussion

The study provides an extensive examination of the clinical management of patients with type 2 diabetes undergoing bariatric surgery and identifies novel associations for diabetic retinopathy in this group of patients.

Diabetic retinopathy

Preoperative levels of diabetic retinopathy

The proportion of patients without diabetic retinopathy at baseline was similar to other studies of retinopathy in bariatric cohorts except for the recent prospective study by Miras et al. [101] that only had 30% without retinopathy (Table 16). That study did not provide detailed methodology to explain this discrepancy. Overall, our cohort was representative in this aspect. We could not establish cross-sectional differences between the patients with and without diabetic retinopathy. Even though we saw the expected association of longer duration of diabetes in the patients with diabetic retinopathy preoperatively, it did not reach statistical significance (Paper II). This was probably due to study size.

Bioactive IGF was significantly elevated in patients with diabetic retinopathy prior to surgery. This intriguing result was strengthened in the multivariate analysis (p=0.006) and supports the hypothesis that IGF-IR activation in the eye is associated with the pathogenesis of diabetic retinopathy. Previous studies of systemic IGF-I and diabetic retinopathy have been contradicting with studies reporting both elevated [127,128], lowered [129] or similar [130-132] levels of circulating IGF-I in patients with diabetic retinopathy. Some of this discrepancy could be due to the uncertainty involved in the different approaches to measuring IGF-I and the regulatory action of IGFBPs. The KIRA assay employed in Paper III determines the ability of serum to activate the IGF-IR in vitro and thereby integrates these confounders into one signal: IGF-IR activation [96].

Six months - nadir or peak?

As this study was initiated due to a concern that bariatric surgery could have detrimental effect on diabetic retinopathy, it was encouraging to observe a small overall improvement after surgery. This improvement peaked six months postoperatively. In contrast, on OCT there was a maximal thickening six months postoperatively. From our study design, we are unable to determine if those two findings are related or spurious. If they are related, it could point to the subclinical thickening being a positive signal – despite our

48

rationale for the opposite in Paper II – or that we are seeing the result of a biologically intended protective function of the Müller cells releasing VEGF to ensure that the retina remains healthy while there is relative substrate deficiency. If the two are not related, different non-exclusive mechanisms could be at play. For instance, diabetic retinopathy could improve due to fewer hyperglycemic events and the macula increase in thickness due to more hypoglycemic events. Nevertheless, our results point to the retina being more vulnerable six months postoperatively compared to, for instance, the one-month visit where almost no changes were observed.

One year

Our one-year results are compared to other studies of diabetic retinopathy after bariatric surgery in Table 16. Even the prospective study by Miras et al. [101] had the preoperative visit well ahead of the surgery procedure and no other study on the subject have used 7-field fundus photos. The highest prevalence of early worsening was in a Korean study [106] reporting worsening in 88% of the patients with preoperative diabetic retinopathy through up to 3.5 years of follow-up. As that was a retrospective study there was no masking of the graders and the conclusion was based on a sample of 8 patients. However, the patients had a median preoperative HbA1c of approximately 9.4, which raises concern for patients with higher preoperative HbA1c than our cohort had. In contrast, the prospective study of Miras et al. had similar baseline

HbA1c and only reported worsening in 3% of this subgroup. A higher prevalence of diabetic retinopathy in people of Asian ancestry [133,134] has previously been described and the large discrepancy between those two studies at each extreme of the worsening spectrum could be due to a combination of genetic differences and sample size.

49

50

Table 16. Comparison of one-year studies of diabetic retinopathy after bariatric surgery, n (%) No preoperative retinopathy Preoperative retinopathy First Without Exam Images, Un- Wor- Un- Wor- Improve- author year n retinopathy Procedure Design before Exam after grading changed sening changed sening ment Murphy 2015 318 218 (69) RYGB, LSG, Retrospective, ? 353 days (mean) 2-field, 180/218 38/218 115/200 50/200 35/200 [135] DS observational 6-Step (83) (17) (58) (25) (18) Miras 2015 56 17 (30) RYGB Prospective, < 6 12-18 months 2-field, 12/17 5/17 25/32 1/32 (3) 6/32 [101] case-control months 5-step (71) (29) (78) (19) Miras 2012 67 39 (58) RYGB, LSG, Retrospective, ? 12-18 months 2-field, 39/39 0/39 (0) 22/28 1/28 (4) 5/28 [136] LGB observational 15-step (100) (79) (18) Thomas 2014 38 26 (68) RYGB, LSG, Retrospective, ? Approximately 2-field, 22/26 4/26 4/12 (33) 3/12 5/12 [137] LGB, BPD observational 12 months 5-step (85) (15) (25) (42) Varadhan 2012 22 15 (68) RYGB, LSG Retrospective, <12 6-12 months 4-field 13/15 2/15 3/7 (43) 2/7 (29) 2/7 [138] observational months 5-step (87) (13) (29) Kim 2015 20 12 (60) RYGB Retrospective, <3 3,6,9,12 months ?, 10/12 2/12 1/8 (13) 7/8 (88) 0/8 [106] observational months 5-step (83) (17) (0) This study 56 32 (57) RYGB, LGS Prospective, <1 1,3,6,12 months 7-field, 32/32 0/32 17/24 3/24 4/24 observational month 15-step (100) (0) (71) (13) (17) RYGB: Roux-en-Y gastric bypass. LSG: laparoscopic sleeve gastrectomy. LGB: laparoscopic gastric banding. BPD: biliopancreatic diversion. DS: duodenal switch.

Beyond one year

Recent reports from the randomized single center Surgical Therapy and Medications Potentially Eradicate Diabetes Efficiently (STAMPEDE) trial reported that there were no differences between levels of retinopathy between 150 patients who were randomized to either intensive medical care, gastric bypass surgery or gastric sleeve gastrectomy [104]. However, this study did not provide much detail on the ocular examinations and grading. The 15-year results from the SOS study coupled their prospective data with register data on diabetes related ocular complications and found a beneficial effect of bariatric surgery compared to their control group but mostly in patients with short duration of disease [31]. Furthermore, the same study found that 70% had remission of diabetes after 2 years but many patients had relapse and at 15 years 30% were free of diabetes [31]. In addition, the patients most likely to remain free from diabetes during long-term follow-up are those with the shortest duration of diabetes preoperatively. This underscores that an element of ‘metabolic memory’ may also play a role in the long-term results for bariatric surgery: the damage incurred by diabetes is cumulative and irreversible after long exposure to a hyperglycemic state. The phenomenon was described for type 1 diabetes in the DCCT study and for type 2 diabetes in the UKPDS [139,140] and the Danish Steno-2 study [141]. For this reason alone, it would be premature to abandon screening of diabetic retinopathy – or other diabetes related screening – in patients in clinical remission of diabetes until more long-term data emerges for diabetic retinopathy status after bariatric surgery.

Changes in visual acuity

The improvement in visual acuity were unexpected. Being a more subjective measure, the result could be due to a learning effect. Furthermore, the changes were clinically insignificant [109]. The STAMPEDE trial reported no change in visual acuity when comparing to a control group, however this study reported visual acuity in logMar scores and could potentially have missed subtle differences [104]. The lack of increase in the best eye could be due to ceiling effect. No changes in visual acuity were described in the DCCT or UKPDS study [56,140] and if the observed improvement in visual acuity is real, we have no satisfactory explanation for this.

Subclinical ocular changes

Retinal thickness and vessel diameters peaked at six and twelve months respectively. While both signals were clinically modest, we interpreted the direction of the changes as detrimental. Firstly, the retinal thickness is affected by the progressive loss of ganglion cells and nerve fibers in the inner layers with increasing degrees

51

of diabetic retinopathy [142], but beyond mild diabetic retinopathy this thinning is surpassed by vascular leakage that surpasses the thinning of the RNFL, leading to a U-shaped thickness profile for increasing diabetic retinopathy [143]. Secondly, a large prospective study of 1,098 patients described the changes in vessel caliber over four years and then followed the patients for another six years and found that widening of CRVE, but not CRAE, was related to subsequent development of diabetic retinopathy [144]. Those results are in agreement with most large cross-sectional epidemiological studies and with studies investigating the risk of having diabetic retinopathy (Table 17). In studies investigating the risk of developing diabetic retinopathy, the results for both CRVE and CRAE have been mixed, but the overall trend is that widening of either marker is either unrelated or detrimental (Table 17). It is not immediately intuitive why changes in venular caliber appear more important than changes in arterioles. Roy et al. proposed that venular dilation, in contrast to arteriolar dilation, indicates either ischemia, systemic inflammation or endothelial dysfunction [145].

Table 17. Vessel diameter and the risk of having or developing diabetic retinopathy. Association with worse retinopathy levels or worsening of retinopathy First author Year n Diabetes Follow-up (years) CRAE CRVE Presence of DR Klein [146] 2003 996 T1D Narrower Wider Klein [147] 2006 1,370 T2D Not significant Wider Nguyen [148] 2008 892 Mixed Not significant Wider Cheung [149] 2008 645 T1D Wider Wider Jeganathan [150] 2009 428 Mixed Wider Wider Tsai [151] 2011 980 ? Not significant Wider Roy [145] 2011 468 T1D Narrower Wider Cheung [152] 2012 594 Mostly T2D Wider Wider Broe [153] 2014 248 T1D Not significant Wider Development of DR Klein [154] 2004 891 T1D 4,10,14 Wider Wider Alibrahim [155] 2006 166 T1D >2 Wider Not significant Klein [156] 2007 987 T2D 4,10,14 No difference Not significant Cheung [152] 2008 645 T1D 2.5 Wider Not significant Rogers [157] 2008 906 Mostly T2D 5 Wider Not significant Roy [145] 2011 468 T1D 6 No difference Wider Broe [153] 2014 248 T1D 16 Narrower Wider CRAE = Central retinal artery equivalent. CRVE = central retinal vein equivalent. T1D = Type 1 diabetes, T2D: Type 2 diabetes. The comparison excludes patients with impaired glucose tolerance.

52

We would expect the thickening in the inner retinal layers to come from increased vascular leakage, but the two signals did not follow the same time course after surgery, so the association between the two is vague. Nevertheless, the signal for an increase in retinal thickness was statistically convincing. A recent study has found a thickening of the central subfield retinal thickness of 1.5% in 31 patients that commenced continuous subcutaneous insulin infusion pump compared to 20 controls [158] and this is interesting as lowering of blood glucose is the major parallel between these two studies.

We found that the subclinical OCT thickening was most prominent in the inner layers (Paper II). Cross- sectional studies have had similar findings [159]. The same group found that subclinical increases in retinal thickness were associated to the development of microanerysms that in turn was related to increasing levels of diabetic retinopathy [160] and macular edema [161]. Subclinical increases in retinal thickness of a magnitude more than 50 µm have been associated with an increased risk of progression to severe thickening requiring treatment [162].

Changes in non-ocular outcomes

IGF-I related biomarkers (Paper III)

Previous cross-sectional studies of systemic IGF-I in relation to diabetic retinopathy have been contradicting [127-132]. This discrepancy could be related to the methodological differences in the measurement of IGF-I or the regulatory action of IGF-II and the IGFBPs. The kinase receptor activation assay determines the binding capability of the IGF-I receptor and integrates the whole IGF system and its ability to activate the IGF-I receptor [96]. The higher levels of IGF bioactivity in patients with diabetic retinopathy support that systemic IGF-I is involved in the development of diabetic retinopathy.

In Paper III we only had 3 of 36 patients who had worsening of diabetic retinopathy, and the lack of longitudinal association between the systemic markers and worsening of diabetic retinopathy may be due to small sample size. Circulating GH and IGF-I have been implicated to be involved in worsening of diabetic retinopathy in several studies [72]. Intervention trials aiming to inhibit GH secretion, GH receptor activation or IGF-I have shown variable results [80,81,163]. The increase in bioactive IGF three months postoperatively was borderline significant (p=0.05) and did not reach statistical significance in the multivariate analysis (p=0.11). In contrast, IGFBP-1 and IGFBP-3 increased and decreased with high statistical significance at both three and twelve months.

53

Previous studies have found a negative association between IGFBP-1 and impaired glucose tolerance [164,165], diabetes [166] and obesity [167,168] though not unanimously [169]. IGFBP-3 has been found to be positively associated with decreased hepatic insulin sensitivity and development of diabetes [165,169]. IGF-II been shown to decrease during energy restriction-induced weight loss [170,171] and diabetes in a setting of gastric banding [172]. The responses in IGFBP-1, IGFBP-3 and IGF-II after gastric bypass surgery in our study are therefore in line with what would be expected. We are aware of two previous studies that have assessed changes in IGFBP-3 changes after bariatric surgery, and both excluded patients with diabetes [173,174]. These studies found a decrease in IGFBP-3 that was similar to our study. Changes in bioactive IGF, IGF-II and IGFBP-1 after gastric bypass surgery have not previously been reported.

Both IGFBP-1 and IGFBP-3 are known to have several IGF-IR independent actions [175,176]. The dissociation between bioactive IGF and IGFBP-1 and -3 suggests an IGF-IR independent role for these binding proteins. Unfortunately, our study design did not allow us to resolve this further, but we propose that the IGFBP- 1/IGFBP-3 ratio could be a sensitive marker for longitudinal changes in the metabolic profile that should be evaluated in future studies.

Resolution of the metabolic syndrome

Bariatric surgery, and more recently mostly gastric bypass and gastric banding, has proven to be a valuable research tool for the study of obesity itself and for diabetes [22]. Therefore, the available evidence for the effects of surgery on BMI, HbA1c and antidiabetic medication is substantial and our results are similar to what other studies have found [177]. The effect of bariatric surgery on hypertension is nevertheless a matter of ongoing debate [178] and prospective randomized studies are underway to address this specific question [179]. In our cohort, there was no clear signal for resolution of hypertension after twelve months, though we found a trend for a lowering of diastolic blood pressure.

Validity

To assess the generalizability of the results of this study, both internal and external validity are important considerations. Internal validity is the minimization of bias within the study and external validity refers to how the patients were sampled and how the results relate to the clinical spectrum of patients.

54

Internal validity

We have sought to use methods that have been validated and we followed a rigorous protocol. The semi- automated procedures alleviate observer bias and the use of longitudinal measurements ensures that the vessel calibers are measured in the exact same spot as the other images in the series. Grader agreement for diabetic retinopathy was similar to that of the ETDRS [112] and all cases of disagreement were resolved by agreement.

External validity within our own setting

The 56 patients were less obese and had shorter duration of diabetes than the remaining 40 patients in the eligible cohort of 96. This is problematic, as sampling bias could therefore mean that results may not extend to patients that are more obese. The longer duration of diabetes in our sample would probably have led us to overestimate any increased risk of worsening as duration of diabetes is the most important risk factor for development of diabetic retinopathy [180]. We speculate that the 40 patients were more concerned with the weight loss aspect of the surgery and less with the diabetes management aspect and sampling bias may therefore be less of an issue for our setting. The staff who recruited patients for our study share this notion (personal correspondence, Annette Lykke Svenningsen). We consider a drop-out rate of 11% to be low.

External validity for other settings

We compared the baseline characteristics of our cohort to other studies with similar inclusion criteria and found that our cohort was similar in terms of sex distribution, age, duration of diabetes, referral HbA1c and use of antidiabetic medication (Table 18). However, our cohort had lower BMI. We note that there is a trend for lower BMI in more recent studies which could be a result of more focus on treating the complications of obesity rather than weight itself [181]. Most studies used fasting glucose or oral glucose tolerance test or

HbA1c to establish diabetes whereas only one of the referenced studies employed a clinical definition similar to ours [182].

55

Table 18. Comparison of cohorts in prospective studies of bariatric surgery with similar inclusion criteria Duration Men Age HbA1c BMI of diabetes Diet Oral Insulin First author Year n (%) (mean) (%) (kg/m2) (years) (%) (%) (%) Schauer [183] 2003 177 25 48 8.2 50.4 5.3 18 71 29 Torquati [184] 2005 97 32 44.0-46.4 7.5-8.6 48.5-51.7 3.5-4.3 n/a n/a 28 Alexandrides [185] 2007 137 23 41.38 46.1 69 26 5 Hayes [182] 2011 127 35 48.5 7.7 46.8 4.5 11 60 25 Ramos-Levi [186] 2013 141 43 53 7.6 43.7 7.4 n/a n/a 40 This study 2015 56 41 47 7.7 42.8 4 5 74 21 Diet: no antidiabetic medication. Oral: only oral antidiabetic medication. Insulin: on insulin therapy.

Prior to the commencement of this study, the Danish government changed the Danish national guidelines for bariatric surgery. After these changes, government funded surgery could only be provided in patients with a BMI above 50 kg/m2. However, publicly funded surgery remained available to patients with BMI 35– 50 kg/m2 who have complications to obesity such as type 2 diabetes. The implementation of these politically motivated guidelines led to a heated public debate [187,188], and during the course of this study the number of patients undergoing bariatric surgery was steadily declining due to negative press coverage [189]. This has influenced the composition of our cohort and highlights the difficulties involved in generalizing results from a single clinical study to a global spectrum of patients.

Limitations

This study has notable limitations:

1. The lack of comparator group makes it impossible to determine whether the observed subclinical changes on OCT are a diabetes-specific finding. The ideal control group would be that of a randomized trial, which is very difficult in bariatric surgery. Only two randomized trials have been completed to date [21,190] though more studies are underway (clinicaltrials.gov NCT01047735, NCT01041768, NCT01821508 and NCT01974544). Assuming a rate of early worsening comparable to those of the DCCT study [56], with a power of 80 and a two-sided significance level of 0.05, we would have needed 458 patients randomized to bariatric surgery or conventional weight loss to detect a difference. Two groups of 56 in each would have given a power of 0.28, i.e. a 72% probability of rejecting a difference in retinopathy worsening if it actually existed. Furthermore, metabolic memory may interfere with the

56

results of trials of patients with high HbA1c which therefore requires a larger sample of patients to ensure

proper randomization and/or that past HbA1c levels are obtained for comparison [191]. 2. The patients in our cohort were well controlled metabolically after their preoperative weight loss but as we have demonstrated, their level of control is comparable to other reports of patients with type 2 diabetes undergoing bariatric surgery. 3. Our patients were less obese than comparable studies. Though customary at many centers, a preoperative weight loss is not a universally accepted precondition for bariatric surgery. We would not expect BMI in itself to affect diabetic retinopathy, but the difference reduces our ability to extrapolate our findings. 4. We have performed many hypothesis tests in Paper II without adjusting for multiple testing. The overall result was established in a prespecified analysis in Paper I, and the exploratory analysis investigated which components contributed the most to the overall results. An adjustment for multiple testing would raise all the p-values, but the overall trend would remain, i.e. the most significant contributors to the overall change would still be the ones with the lowest p-values.

57

Perspectives

Paper I has established a case for bariatric surgery being safe from an ophthalmologic perspective in patients with type 2 diabetes, especially in well-controlled patients. As there are no indications of a glucose- independent mechanism of early worsening, previous established guidelines for diabetic retinopathy including the guidelines for early worsening cover the bariatric surgery situation as well. Our study provides reassurance for the clinician faced with the perioperative management of patients with diabetes undergoing gastric bypass surgery and is in line with the findings of other prospective studies with a follow-up period exceeding one year postoperatively. There are however limitations, as our sample was not fully generalizable to other clinical settings, and it would be instructive to have these results confirmed in more obese patients with higher HbA1c not undergoing a preoperative weight loss and patients of other ethnicities. One group of patients therefore remains a challenge: patients with high HbA1c and long diabetes duration with the highest a priori risk of worsening of diabetic retinopathy. However, studying such patients would have to be conducted in a multiple center setting to achieve sufficient volume. Furthermore, these particular patients may be more difficult to enroll in clinical studies due to compliance considerations and many research groups tend to be interested in these patients, making recruitment for multiple studies even more challenging.

The OCT findings of Paper II were exploratory and should be confirmed prospectively in a prespecified setup with a control group. If the findings are replicated, they could provide a clue to the link between acute glucose lowering and transient worsening of diabetic retinopathy. Unraveling this enigma could perhaps lead to a better understanding of diabetes in general. Three similar sized groups undergoing bariatric surgery would be ideal: a group without diabetic retinopathy, a group with mild diabetic retinopathy and a group with moderate to severe diabetic retinopathy at three, six and twelve months postoperatively. Alternatively, patients commencing continuous insulin infusion pump therapy could be examined.

The findings of Paper III suggest that bioactive IGF is a biomarker candidate for diabetic retinopathy. This hypothesis could be tested in a larger cohort, ideally with greater emphasis on matching the patients on as many major demographic criteria as possible such as age, sex, duration of diabetes, HbA1c and ethnicity. The volume of patients with diabetes would make such an effort possible in a daily clinical setting and could provide further proof of the hypothesis. The next step would be to obtain vitreous samples of IGF-I, which could be achieved in the setting of vitrectomy for diabetic retinopathy or maybe in the setting of intravitreal anti-VEGF therapy for diabetic macular edema.

58

Unfortunately, our study does not address the long-term ocular consequences of bariatric surgery in patients with type 2 diabetes. To address this important aspect, it would be highly instructive to re-examine our cohort 5 to 10 years after surgery.

59

Conclusions

Bariatric surgery improved all markers of diabetes and the metabolic syndrome and reduced the need for anti-diabetic medication in patients with preoperative type 2 diabetes.

We did not find any clinical indicators of severe worsening of diabetic retinopathy after bariatric surgery and therefore maintain that continued screening for diabetic retinopathy is important after bariatric surgery but can follow standard guidelines of diabetic retinopathy regardless of the surgery. The results extend to patients that are well-controlled metabolically, but closer follow-up may be warranted in patients that are poorly controlled prior to surgery. The most vulnerable period appears to be six months after surgery.

We also identified subclinical indicators of increased retinal thickening in the inner layers of the parafoveal retina that may provide a clue to the early worsening of diabetic retinopathy that has been described in several studies of improved glycemic control. We were unable to associate the changes with changes in the retinal vessel calibers.

Bioactive IGF was higher in patients with preoperative diabetic retinopathy and could be a novel biomarker for diabetic retinopathy. However, we failed to establish a longitudinal relationship between the small number of patients who had worsening of diabetic retinopathy and markers of the IGF-I system.

60

Summary

This thesis explores the perioperative management of diabetic retinopathy after bariatric surgery and examines changes in subclinical retinal and systemic biomarkers associated with the surgery.

Bariatric surgery, in particular gastric bypass surgery and vertical gastric sleeve gastrectomy, rapidly causes remission of type 2 diabetes independent of the ensuing weight loss. Lowering of blood glucose is beneficial for diabetes-related complications on a cohort level, but diabetic retinopathy is a special challenge, as rapid lowering of blood glucose carries a risk of early worsening of diabetic retinopathy in non-bariatric settings. No studies have previously investigated the perioperative retinal status of patients with type 2 diabetes following bariatric surgery.

The aim was to investigate if patients with type 2 diabetes undergoing bariatric surgery should have closer ophthalmologic attention than other patients with diabetes. We also aimed to characterize subclinical postoperative retinal changes and associated changes in the insulin-like growth factor (IGF) system.

We examined 56 patients aged 27-65 years with type 2 diabetes who underwent bariatric surgery. The examinations were performed at five timepoints: prior to surgery and one, three, six and twelve months after surgery. The patients underwent 7-field fundus photographs that were masked and graded using the Wisconsin scale and underwent optical coherence tomography. We also collected fasting blood samples.

We found that diabetic retinopathy was clinically stable after surgery. There was even a trend for improvement six months postoperatively, and no patients needed to have adjustments in their ophthalmologic screening intervals following surgery. However, there was a thickening of the macula that peaked six months postoperatively, indicating vascular leakage. The thickening was most dominant in the inner layers of the parafovea and was clinically negligible. A widening of the retinal arterioles that peaked twelve months postoperatively may also indicate a deleterious effect, but the trend was only borderline significant. Systemic levels of bioactive IGF were higher in patients with diabetic retinopathy preoperatively. Bioactive IGF increased three months postoperatively and its binding protein, IGFBP3, decreased at both three and twelve months postoperatively. However, we could not correlate these changes to changes in diabetic retinopathy. Our patients were similar to other cohorts of patients with type 2 diabetes undergoing bariatric surgery and few had poorly controlled HbA1c preoperatively.

61

In conclusion, diabetic retinopathy was stable after bariatric surgery in a population of patients with a metabolically well regulated type 2 diabetes. We conclude that the metabolically well-regulated patients can follow the standard diabetic retinopathy screening guidelines following bariatric surgery. A clinically negligible, but statistically significant macular thickening in the inner layers of the parafovea six months postoperatively, could be an indicator of vascular leakage. Furthermore, bioactive IGF could be a novel biomarker for diabetic retinopathy.

62

Dansk resumé (Summary in Danish)

Denne afhandling undersøger den perioperative håndtering af patienter med diabetisk retinopati, der gennemgår bariatrisk kirurgi, og de subkliniske nethindeforandringer associeret med operationen.

Bariatrisk kirurgi, særligt gastric bypass og vertical sleeve kirurgi, forårsager en meget hurtig remission af type 2 diabetes uafhængigt af det efterfølgende vægttab. En sænkning af blodsukkeret er godt for diabetes- associerede komplikationer på langt sigt på befolkningsplan, men diabetisk retinopati er en særlig udfordring, da hurtig nedregulering af blodsukkeret medfører en øget risiko for tidlig forværring af diabetisk nethindesygdom i ikke-bariatriske studier. Ingen studier har tidligere beskæftiget sig med den perioperative nethindestatus hos patienter med diabetisk retinopati efter bariatrisk kirurgi.

Formålet var at undersøge om patienter med type 2 diabetes der gennemgår bariatrisk kirurgi bør have foretaget tættere øjenscreening end andre patienter med diabetes. Vi ville også karakterisere subkliniske nethindeforandringer efter operationen og associerede forandringer i insulin-like growth factor (IGF) aksen.

Vi undersøgte 56 patienter i alderen 27-65 år med type 2 diabetes som gennemgik bariatrisk kirurgi. Undersøgelserne blev udført fem gange: før operationen samt en, tre, seks og tolv måneder efter operationen. Patienterne fik taget 7-felts nethindebilleder, som blev maskeret og graderet ved hjælp af Wisconsin-skalaen, og fik foretaget optisk kohærens tomografi scanning. Vi indsamlede også fasteblodprøver ved tre af besøgene.

Efter operationen var diabetisk retinopati stabilt. Der var endda en lille forbedring af retinopati-graden seks måneder postoperativt. Ingen patienter havde behov for ændret screeningsinterval postoperativt. Der var dog en lille fortykkelse af nethinden efter seks måneder, hvilket kunne indikere ødemdannelse. Fortykkelsen var størst i perifoveas indre lag og havde ingen klinisk betydning. Der var desuden en tendens til fortykkelse af de retinale arterioler, hvilket også kunne være et dårligt prognostisk tegn, men tendensen var kun grænsesignifikant. Patienter med diabetisk retinopati før operationen havde et højere niveau af bioaktivt IGF-I. Bioaktivt IGF-I steg seks måneder efter operationen, og IGF-I bindingsprotein 1 og 3 faldt seks og tolv måneder efter operationen. Vi kunne dog ikke korrelere disse ændringer med ændringer i diabetisk retinopati. Vores patienter var sammenlignelige med andre studier vedrørende type 2 diabetes og bariatrisk kirurgi og få havde dårligt reguleret HbA1c præoperativt.

63

Vi konkluderer, at graden af diabetisk retinopati var stabil efter bariatrisk kirurgi i en population af type 2 diabetes patienter, som var metabolisk velregulerede. De metabolisk velregulerede patienter kan således følge normale screeningsretningslinjer efter bariatrisk kirurgi. Desuden fandt vi en klinisk diskret fortykkelse af nethinden seks måneder efter operationen, der var størst i nethindes indre lag der omkranser fovea centralis og kunne indikere øget karlækage. Bioaktivt IGF-I kan være en ny biomarkør for diabetisk retinopati.

64

References

1. Whiting DR, Guariguata L, Weil C, Shaw J. IDF diabetes atlas: global estimates of the prevalence of diabetes for 2011 and 2030. Diabetes Res Clin Pract. 2011;94:311–21. PMID: 22079683

2. Kahn SE, Hull RL, Utzschneider KM. Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature. 2006;444:840–6. PMID: 17167471

3. Hofsø D, Jenssen T, Hager H, Røislien J, Hjelmesæth J. Fasting plasma glucose in the screening for type 2 diabetes in morbidly obese subjects. Obes Surg. 2010;20:302–7. PMID: 19949889

4. Defronzo RA. From the triumvirate to the ominous octet: a new paradigm for the treatment of type 2 diabetes mellitus. 2009;58:773–95. PMID: 19336687

5. American Diabetes Association. Classification and diagnosis of diabetes. Diabetes Care. 2015;38 Suppl:S8–S16. PMID: 25537714

6. Danish National Board of Health. [Regarding the diagnosis of diabetes mellitus using HbA1c]. [in Danish] http://sundhedsstyrelsen.dk/~/media/2A3178A6D31B428FA888E39AA46B0B4E.ashx Accessed October 8th 2015.

7. Ramlo-Halsted BA, Edelman SV. The natural history of type 2 diabetes. Implications for clinical practice. Prim Care. 1999;26:771–89. PMID: 10523459

8. Giovannucci E, Harlan DM, Archer MC, Bergenstal RM, Gapstur SM, Habel LA, et al. Diabetes and cancer: a consensus report. Diabetes Care. 2010;33:1674–85. PMID: 20587728

9. Mulnier HE, Seaman HE, Raleigh VS, Soedamah-Muthu SS, Colhoun HM, Lawrenson RA. Mortality in people with type 2 diabetes in the UK. Diabet Med. 2006;23:516–21. PMID: 16681560

10. Hansen MB, Jensen ML, Carstensen B. Causes of death among diabetic patients in Denmark. Diabetologia. 2012;55:294–302. PMID: 22127411

11. Strain DW. Time to do more for diabetes: clinical inertia and how to beat it. Diabet Voice. 2015;59:36–9.

12. Brownlee M. The pathobiology of diabetic complications: a unifying mechanism. Diabetes. 2005;54:1615–25. PMID: 15919781

13. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993;329:977–86. PMID: 8366922

14. UK Prospective Diabetes Study Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet. 1998;352:837–53. PMID: 9742976

15. Action to Control Cardiovascular Risk in Diabetes Study Group, Gerstein HC, Miller ME, Byington RP, Goff DC, Bigger JT, et al. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med.

65

2008;358:2545–59. PMID: 18539917

16. Snell-Bergeon JK, Wadwa RP. Hypoglycemia, diabetes, and cardiovascular disease. Diabetes Technol Ther. 2012;14 Suppl 1:S51–8. PMID: 22650225

17. Monnier L, Colette C. Glycemic variability: can we bridge the divide between controversies? Diabetes Care. 2011;34:1058–9. PMID: 21447669

18. Dombrowski SU, Knittle K, Avenell A, Araujo-Soares V, Sniehotta FF. Long term maintenance of weight loss with non-surgical interventions in obese adults: systematic review and meta-analyses of randomised controlled trials. BMJ. 2014;348:g2646. PMID: 25134100

19. Maclean PS, Bergouignan A, Cornier M-A, Jackman MR. Biology's response to dieting: the impetus for weight regain. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2011;301:R581–600. PMID: 21677272

20. Angrisani L, Santonicola A, Iovino P, Formisano G, Buchwald H, Scopinaro N. Bariatric Surgery Worldwide 2013. Obes Surg. 2015;26:Epub. PMID: 25835983

21. Schauer PR, Bhatt DL, Kirwan JP, Wolski K, Brethauer SA, Navaneethan SD, et al. Bariatric surgery versus intensive medical therapy for diabetes--3-year outcomes. N Engl J Med. 2014;370:2002–13. PMID: 24679060

22. Miras AD, le Roux CW. Mechanisms underlying weight loss after bariatric surgery. Nat Rev Gastroenterol Hepatol. 2013;10:575–84. PMID: 23835488

23. Borisenko O, Colpan Z, Dillemans B, Funch-Jensen P, Hedenbro J, Ahmed AR. Clinical Indications, Utilization, and Funding of Bariatric Surgery in Europe. Obes Surg. 2015;25:1408–16. PMID: 25528567

24. Pories WJ, Swanson MS, MacDonald KG, Long SB, Morris PG, Brown BM, et al. Who would have thought it? An operation proves to be the most effective therapy for adult-onset diabetes mellitus. Ann Surg. 1995;222:339–50. PMID: 7677463

25. Laferrère B, Teixeira J, McGinty J, Tran H, Egger JR, Colarusso A, et al. Effect of weight loss by gastric bypass surgery versus hypocaloric diet on glucose and incretin levels in patients with type 2 diabetes. J. Clin. Endocrinol. Metab. 2008;93:2479–85. PMID: 18430778

26. Abdeen G, le Roux CW. Mechanism Underlying the Weight Loss and Complications of Roux-en-Y Gastric Bypass. Review. Obes Surg. 2015. PMID: 26530712

27. Dirksen C, Jorgensen NB, Bojsen-Moller KN, Jacobsen SH, Hansen DL, Worm D, et al. Mechanisms of improved glycaemic control after Roux-en-Y gastric bypass. Diabetologia. 2012;55:1890–901. PMID: 22538359

28. Defronzo RA, Ratner RE, Han J, Kim DD, Fineman MS, Baron AD. Effects of exenatide (exendin-4) on glycemic control and weight over 30 weeks in metformin-treated patients with type 2 diabetes. Diabetes Care. 2005;28:1092–100. PMID: 15855572

29. Buse JB, Rosenstock J, Sesti G, Schmidt WE, Montanya E, Brett JH, et al. Liraglutide once a day versus exenatide twice a day for type 2 diabetes: a 26-week randomised, parallel-group, multinational, open-label trial (LEAD-6). Lancet. 2009;374:39–47. PMID: 19515413

66

30. Pi-Sunyer X, Astrup A, Fujioka K, Greenway F, Halpern A, Krempf M, et al. A Randomized, Controlled Trial of 3.0 mg of Liraglutide in Weight Management. N Engl J Med. 2015;373:11–22. PMID: 26132939

31. Sjöström L, Peltonen M, Jacobson P, Ahlin S, Andersson-Assarsson J, Anveden Å, et al. Association of bariatric surgery with long-term remission of type 2 diabetes and with microvascular and macrovascular complications. JAMA. 2014;311:2297–304. PMID: 24915261

32. Congdon NG, Friedman DS, Lietman T. Important causes of visual impairment in the world today. JAMA. 2003;290:2057–60. PMID: 14559961

33. Kempen JH, O'Colmain BJ, Leske MC, Haffner SM, Klein R, Moss SE, et al. The prevalence of diabetic retinopathy among adults in the United States. Arch Ophthalmol. 2004;122:552–63. PMID: 15078674

34. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet. 2010;376:124–36. PMID: 20580421

35. Klein R, Klein BE, Moss SE, Cruickshanks KJ. The Wisconsin Epidemiologic Study of diabetic retinopathy. XIV. Ten-year incidence and progression of diabetic retinopathy. Arch Ophthalmol. 1994;112:1217–28. PMID: 7619101

36. Klein R, Knudtson MD, Lee KE, Gangnon R, Klein BEK. The Wisconsin Epidemiologic Study of Diabetic Retinopathy XXIII: the twenty-five-year incidence of macular edema in persons with type 1 diabetes. Ophthalmology. 2009;116:497–503. PMID: 19167079

37. Liew G, Michaelides M, Bunce C. A comparison of the causes of blindness certifications in England and Wales in working age adults (16-64 years), 1999-2000 with 2009-2010. BMJ Open. 2014;4:e004015–5. PMID: 24525390

38. Klein R, Lee KE, Knudtson MD, Gangnon RE, Klein BEK. Changes in visual impairment prevalence by period of diagnosis of diabetes: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Ophthalmology. 2009;116:1937–42. PMID: 19616855

39. Hovind P, Tarnow L, Rossing K, Rossing P, Eising S, Larsen N, et al. Decreasing incidence of severe diabetic microangiopathy in type 1 diabetes. Diabetes Care. 2003;26:1258–64. PMID: 12663607

40. Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl J Med. 2012;366:1227–39. PMID: 22455417

41. Pournaras CJ, Rungger-Brändle E, Riva CE, Hardarson SH, Stefánsson E. Regulation of retinal blood flow in health and disease. Prog Retin Eye Res. 2008;27:284–330. PMID: 18448380

42. Holfort SK, Nørgaard K, Jackson GR, Hommel E, Madsbad S, Munch IC, et al. Retinal function in relation to improved glycaemic control in type 1 diabetes. Diabetologia. 2011;54:1853–61. PMID: 21516521

43. Bearse MA, Adams AJ, Han Y, Schneck ME, Ng J, Bronson-Castain K, et al. A multifocal electroretinogram model predicting the development of diabetic retinopathy. Prog Retin Eye Res. 2006;25:425–48. PMID: 16949855

44. Jackson GR, Scott IU, Quillen DA, Walter LE, Gardner TW. Inner retinal visual dysfunction is a sensitive marker of non-proliferative diabetic retinopathy. Br J Ophthalmol. 2012;96:699–703. PMID: 22174096

45. van Dijk HW, Kok PHB, Garvin M, Sonka M, DeVries JH, Michels RPJ, et al. Selective loss of inner retinal

67

layer thickness in type 1 diabetic patients with minimal diabetic retinopathy. Invest Ophthalmol Vis Sci. 2009;50:3404–9. PMID: 19151397

46. Kern TS, Barber AJ. Retinal ganglion cells in diabetes. J Physiol. 2008;586:4401–8. PMID: 18565995

47. Early Treatment Diabetic Retinopathy Study Research Group. Grading diabetic retinopathy from stereoscopic color fundus photographs--an extension of the modified Airlie House classification. ETDRS report number 10. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology. 1991;98:786–806. PMID: 2062513

48. Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology. 1991;98:823–33. PMID: 2062515

49. Javitt JC, Aiello LP, Chiang Y, Ferris FL, Canner JK, Greenfield S. Preventive eye care in people with diabetes is cost-saving to the federal government. Implications for health-care reform. Diabetes Care. 1994;17:909–17. PMID: 7956643

50. Leese GP, Stratton IM, Land M, Bachmann MO, Jones C, Scanlon P, et al. Progression of diabetes retinal status within community screening programs and potential implications for screening intervals. Diabetes Care. 2015;38:488–94. PMID: 25524948

51. Wilkinson CP, Ferris FL, Klein RE, Lee PP, Agardh CD, Davis M, et al. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology. 2003;110:1677– 82. PMID: 13129861

52. American Academy of Ophthalmology: Diabetic Retinopathy Preferred practice patterns 2014. www.aao.org/preferred-practice-pattern/diabetic-retinopathy-ppp--2014. Accessed July 7th 2015.

53. Davis MD, Fisher MR, Gangnon RE, Barton F, Aiello LM, Chew EY, et al. Risk factors for high-risk proliferative diabetic retinopathy and severe visual loss: Early Treatment Diabetic Retinopathy Study Report #18. Invest Ophthalmol Vis Sci. 1998;39:233–52. PMID: 9477980

54. Aspelund T, Thornórisdóttir O, Ólafsdottir E, Gudmundsdottir A, Einarsdóttir AB, Mehlsen J, et al. Individual risk assessment and information technology to optimise screening frequency for diabetic retinopathy. Diabetologia. 2011;54:2525–32. PMID: 21792613

55. Zorena K, Raczyńska D, Raczyńska K. Biomarkers in diabetic retinopathy and the therapeutic implications. Mediators Inflamm. 2013;2013:193604. PMID: 24311895

56. The Diabetes Control and Complications Trial Research Group. Early worsening of diabetic retinopathy in the Diabetes Control and Complications Trial. Arch Ophthalmol. 1998;116:874–86. PMID: 9682700

57. Chew EY, Mills JL, Metzger BE, Remaley NA, Jovanovic-Peterson L, Knopp RH, et al. Metabolic control and progression of retinopathy. The Diabetes in Early Pregnancy Study. Diabetes Care. 1995;18:631–7. PMID: 8586000

58. The Kroc Collaborative Study Group. Blood glucose control and the evolution of diabetic retinopathy and albuminuria. A preliminary multicenter trial. N Engl J Med. 1984;311:365–72. PMID: 6377076

59. Lauritzen T, Frost-Larsen K, Larsen HW, Deckert T. Two-year experience with continuous subcutaneous insulin infusion in relation to retinopathy and neuropathy. Diabetes. 1985;34 Suppl 3:74–9. PMID: 4018423

68

60. Brinchmann-Hansen O, Dahl-Jørgensen K, Sandvik L, Hanssen KF. Blood glucose concentrations and progression of diabetic retinopathy: the seven year results of the Oslo study. BMJ. 1992;304:19–22. PMID: 1734985

61. Scheider A, Meyer-Schwickerath E, Nusser J, Land W, Landgraf R. Diabetic retinopathy and pancreas transplantation: a 3-year follow-up. Diabetologia. 1991;34 Suppl 1:S95–9. PMID: 1936707

62. Ramsay RC, Goetz FC, Sutherland DE, Mauer SM, Robison LL, Cantrill HL, et al. Progression of diabetic retinopathy after pancreas transplantation for insulin-dependent diabetes mellitus. N Engl J Med. 1988;318:208–14. PMID: 3275895

63. Rasmussen KL, Laugesen CS, Ringholm L, Vestgaard M, Damm P, Mathiesen ER. Progression of diabetic retinopathy during pregnancy in women with type 2 diabetes. Diabetologia. 2010;53:1076–83. PMID: 20225131

64. Henricsson M, Berntorp K, Berntorp E, Fernlund P, Sundkvist G. Progression of retinopathy after improved metabolic control in type 2 diabetic patients. Relation to IGF-1 and hemostatic variables. Diabetes Care. 1999;22:1944–9. PMID: 10587823

65. Roysarkar TK, Gupta A, Dash RJ, Dogra MR. Effect of insulin therapy on progression of retinopathy in noninsulin-dependent diabetes mellitus. Am J Ophthalmol. 1993;115:569–74. PMID: 8488908

66. Henricsson M, Berntorp K, Fernlund P, Sundkvist G. Progression of retinopathy in insulin-treated type 2 diabetic patients. Diabetes Care. 2002;25:381–5. PMID: 11815514

67. Brooks AMS, Lissett CA. A dramatic deterioration in diabetic retinopathy with improvement in glycated haemoglobin (HbA(1c)) on exenatide treatment. Diabet Med. 2009;26:190–0. PMID: 19236627

68. Varadhan L, Humphreys T, Hariman C, Walker AB, Varughese GI. GLP-1 agonist treatment: implications for diabetic retinopathy screening. Diabetes Res Clin Pract. 2011;94:e68–71. PMID: 21906831

69. Silva RA, Morton JM, Moshfeghi DM. Severe worsening of diabetic retinopathy following bariatric surgery. Ophthalmic Surg Lasers Imaging Retina. 2013;44 Online:E11–4. PMID: 24131131

70. Smith LE, Kopchick JJ, Chen W, Knapp J, Kinose F, Daley D, et al. Essential role of growth hormone in ischemia-induced retinal neovascularization. Science. 1997;276:1706–9. PMID: 9180082

71. Smith LE, Shen W, Perruzzi C, Soker S, Kinose F, Xu X, et al. Regulation of vascular endothelial growth factor-dependent retinal neovascularization by insulin-like growth factor-1 receptor. Nat Med. 1999;5:1390–5. PMID: 10581081

72. Frystyk J. The growth hormone hypothesis - 2005 revision. Horm. Metab. Res. 2005;37 Suppl 1:44–8. PMID: 15918110

73. Quattrin T, Thrailkill K, Baker L, Kuntze J, Compton P, Martha P, et al. Improvement of HbA1c without increased hypoglycemia in adolescents and young adults with type 1 diabetes mellitus treated with recombinant human insulin-like growth factor-I and insulin. rhIGF-I in IDDM Study Group. J Pediatr Endocrinol Metab. 2001;14:267–77. PMID: 11308044

74. Thrailkill KM, Quattrin T, Baker L, Kuntze JE, Compton PG, Martha PM. Cotherapy with recombinant human insulin-like growth factor I and insulin improves glycemic control in type 1 diabetes. RhIGF-I in IDDM

69

Study Group. Diabetes Care. 1999;22:585–92. PMID: 10189536

75. Grant MB, Mames RN, Fitzgerald C, Hazariwala KM, Cooper-DeHoff R, Caballero S, et al. The efficacy of octreotide in the therapy of severe nonproliferative and early proliferative diabetic retinopathy: a randomized controlled study. Diabetes Care. 2000;23:504–9. PMID: 10857943

76. Boehm BO, Lang GK, Jehle PM, Feldman B, Lang GE. Octreotide reduces vitreous hemorrhage and loss of visual acuity risk in patients with high-risk proliferative diabetic retinopathy. Horm. Metab. Res. 2001;33:300–6. PMID: 11440277

77. Mallet B, Vialettes B, Haroche S, Escoffier P, Gastaut P, Taubert JP, et al. Stabilization of severe proliferative diabetic retinopathy by long-term treatment with SMS 201-995. Diabete Metab. 1992;18:438– 44. PMID: 1297600

78. McCombe M, Lightman S, Eckland DJ, Hamilton AM, Lightman SL. Effect of a long-acting somatostatin analogue (BIM23014) on proliferative diabetic retinopathy: a pilot study. Eye. 1991;5 ( Pt 5):569–75. PMID: 1686591

79. Shumak SL, Grossman LD, Chew E, Kozousek V, George SR, Singer W, et al. Growth hormone suppression and nonproliferative diabetic retinopathy: a preliminary feasibility study. Clin Invest Med. 1990;13:287–92. PMID: 2276223

80. Growth Hormone Antagonist for Proliferative Diabetic Retinopathy Study Group. The effect of a growth hormone receptor antagonist drug on proliferative diabetic retinopathy. Ophthalmology. 2001;108:2266– 72. PMID: 11733269

81. Kirkegaard C, Nørgaard K, Snorgaard O, Bek T, Larsen M, Lund-Andersen H. Effect of one year continuous subcutaneous infusion of a somatostatin analogue, octreotide, on early retinopathy, metabolic control and thyroid function in Type I (insulin-dependent) diabetes mellitus. Acta Endocrinol. 1990;122:766–72. PMID: 2197845

82. Kennedy A, Frank RN. The influence of glucose concentration and hypoxia on VEGF secretion by cultured retinal cells. Curr Eye Res. 2011;36:168–77. PMID: 21158590

83. Grunwald JE, Riva CE, Martin DB, Quint AR, Epstein PA. Effect of an insulin-induced decrease in blood glucose on the human diabetic retinal circulation. Ophthalmology. 1987;94:1614–20. PMID: 3323985

84. Lauritzen T, Frost-Larsen K, Larsen HW, Deckert T. Effect of 1 year of near-normal blood glucose levels on retinopathy in insulin-dependent diabetics. Lancet. 1983;1:200–4. PMID: 6130243

85. Sullivan PM, Davies GE, Caldwell G, Morris AC, Kohner EM. Retinal blood flow during hyperglycemia. A laser Doppler velocimetry study. Invest Ophthalmol Vis Sci. 1990;31:2041–5. PMID: 2211001

86. Poulaki V, Qin W, Joussen AM, Hurlbut P, Wiegand SJ, Rudge J, et al. Acute intensive insulin therapy exacerbates diabetic blood-retinal barrier breakdown via hypoxia-inducible factor-1alpha and VEGF. J Clin Invest. 2002;109:805–15. PMID: 11901189

87. Varewijck AJ, Janssen JAMJL. Insulin and its analogues and their affinities for the IGF1 receptor. Endocrine Related Cancer. 2012;19:F63–75. PMID: 22420005

88. Kurtzhals P, Schäffer L, Sørensen A, Kristensen C, Jonassen I, Schmid C, et al. Correlations of receptor

70

binding and metabolic and mitogenic potencies of insulin analogs designed for clinical use. Diabetes. 2000;49:999–1005. PMID: 10866053

89. Rosenstock J, Fonseca V, McGill JB, Riddle M, Hallé JP, Hramiak I, et al. Similar progression of diabetic retinopathy with insulin glargine and neutral protamine Hagedorn (NPH) insulin in patients with type 2 diabetes: a long-term, randomised, open-label study. Diabetologia. 2009;52:1778–88. PMID: 19526210

90. Dandona P, Bolger JP, Boag F, Fonesca V, Abrams JD. Rapid development and progression of proliferative retinopathy after strict diabetic control. Br Med J (Clin Res Ed). 1985;290:895–6. PMID: 3919833

91. Agardh CD, Eckert B, Agardh E. Irreversible progression of severe retinopathy in young type I insulin- dependent diabetes mellitus patients after improved metabolic control. J Diab Comp. 1992;6:96–100. PMID: 1611145

92. Spranger J, Bühnen J, Jansen V, Krieg M, Meyer-Schwickerath R, Blum WF, et al. Systemic levels contribute significantly to increased intraocular IGF-I, IGF-II and IGF-BP3 [correction of IFG-BP3] in proliferative diabetic retinopathy. Horm. Metab. Res. 2000;32:196–200. PMID: 10871161

93. Burgos R, Mateo C, Cantón A, Hernandez C, Mesa J, Simo R. Vitreous levels of IGF-I, IGF binding protein 1, and IGF binding protein 3 in proliferative diabetic retinopathy: a case-control study. Diabetes Care. 2000;23:80–3. PMID: 10857973

94. Grant M, Russell B, Fitzgerald C, Merimee TJ. Insulin-like growth factors in vitreous. Studies in control and diabetic subjects with neovascularization. Diabetes. 1986;35:416–20. PMID: 2420665

95. Clemmons DR. Modifying IGF1 activity: an approach to treat endocrine disorders, atherosclerosis and cancer. Nat Rev Drug Discov. 2007;6:821–33. PMID: 17906644

96. Chen J-W, Ledet T, Orskov H, Jessen N, Lund S, Whittaker J, et al. A highly sensitive and specific assay for determination of IGF-I bioactivity in human serum. Am J Physiol Endocrinol Metab. 2003;284:E1149–55. PMID: 12604504

97. Miras AD, le Roux CW. Metabolic surgery: shifting the focus from glycaemia and weight to end-organ health. Lancet Diabetes Endocrinol. 2014;2:141–51. PMID: 24622718

98. Vest AR, Heneghan HM, Agarwal S, Schauer PR, Young JB. Bariatric surgery and cardiovascular outcomes: a systematic review. Heart. 2012;98:1763–77. PMID: 23077152

99. Sjöström L, Narbro K, Sjöström CD, Karason K, Larsson B, Wedel H, et al. Effects of bariatric surgery on mortality in Swedish obese subjects. N Engl J Med. 2007;357:741–52. PMID: 17715408

100. Iaconelli A, Panunzi S, De Gaetano A, Manco M, Guidone C, Leccesi L, et al. Effects of bilio-pancreatic diversion on diabetic complications: a 10-year follow-up. Diabetes Care. 2011;34:561–7. PMID: 21282343

101. Miras AD, Chuah LL, Khalil N, Nicotra A, Vusirikala A, Baqai N, et al. Type 2 diabetes mellitus and microvascular complications 1 year after Roux-en-Y gastric bypass: a case-control study. Diabetologia. 2015;58:1443–7. PMID: 25893730

102. Amor A, Jiménez A, Moizé V, Ibarzabal A, Flores L, Lacy AM, et al. Weight loss independently predicts urinary albumin excretion normalization in morbidly obese type 2 diabetic patients undergoing bariatric

71

surgery. Surg Endosc. 2013;27:2046–51. PMID: 23292561

103. Fenske WK, Dubb S, Bueter M, Seyfried F, Patel K, Tam FWK, et al. Effect of bariatric surgery-induced weight loss on renal and systemic inflammation and blood pressure: a 12-month prospective study. Surg Obes Relat Dis. 2013;9:559–68. PMID: 22608055

104. Singh RP, Gans R, Kashyap SR, Bedi R, Wolski K, Brethauer SA, et al. Effect of bariatric surgery versus intensive medical management on diabetic ophthalmic outcomes. Diabetes Care. 2015;38:e32–3. PMID: 25715418

105. Johnson BL, Blackhurst DW, Latham BB, Cull DL, Bour ES, Oliver TL, et al. Bariatric surgery is associated with a reduction in major macrovascular and microvascular complications in moderately to severely obese patients with type 2 diabetes mellitus. J Am Coll Surg. 2013;216:545–56. PMID: 23391591

106. Kim YJ, Seo DR, Kim MJ, Lee SJ, Hur KY, Choi KS. Clinical Course of Diabetic Retinopathy in Korean Type 2 Diabetes After Bariatric Surgery: A Pilot Study. Retina. 2015;35:935–43. PMID: 25574784

107. Browning DJ, Fraser CM, Clark S. The relationship of macular thickness to clinically graded diabetic retinopathy severity in eyes without clinically detected diabetic macular edema. Ophthalmology. 2008;115:533–539.e2. PMID: 18067962

108. Kashyap SR, Bhatt DL, Wolski K, Watanabe RM, Abdul-Ghani M, Abood B, et al. Metabolic effects of bariatric surgery in patients with moderate obesity and type 2 diabetes: analysis of a randomized control trial comparing surgery with intensive medical treatment. Diabetes Care. 2013;36:2175–82. PMID: 23439632

109. Kaiser PK. Prospective evaluation of visual acuity assessment: a comparison of snellen versus ETDRS charts in clinical practice (An AOS Thesis). Trans Am Ophthalmol Soc. 2009;107:311–24. PMID: 20126505

110. Neubauer AS, Rothschuh A, Ulbig MW, Blum M. Digital fundus image grading with the non-mydriatic Visucam(PRO NM) versus the FF450(plus) camera in diabetic retinopathy. Acta Ophthalmol. 2008;86:177– 82. PMID: 17944975

111. George LD, Lusty J, Owens DR, Ollerton RL. Effect of software manipulation (Photoshop) of digitised retinal images on the grading of diabetic retinopathy. Br J Ophthalmol. 1999;83:911–3. PMID: 10413691

112. Early Treatment Diabetic Retinopathy Study Research Group. Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Ophthalmology. 1991;98:823–33. PMID: 2062515

113. Klein R, Klein BE, Moss SE. How many steps of progression of diabetic retinopathy are meaningful? The Wisconsin epidemiologic study of diabetic retinopathy. Arch Ophthalmol. 2001;119:547–53. PMID: 11296020

114. Early Treatment Diabetic Retinopathy Study Research Group. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Early Treatment Diabetic Retinopathy Study research group. Arch Ophthalmol. 1985;103:1796–806. PMID: 2866759

115. Bressler NM, Edwards AR, Antoszyk AN, Beck RW, Browning DJ, Ciardella AP, et al. Retinal thickness on Stratus optical coherence tomography in people with diabetes and minimal or no diabetic retinopathy. Am J Ophthalmol. 2008;145:894–901. PMID: 18294608

72

116. Grover S, Murthy RK, Brar VS, Chalam KV. Comparison of retinal thickness in normal eyes using Stratus and Spectralis optical coherence tomography. Invest Ophthalmol Vis Sci. 2010;51:2644–7. PMID: 20007831

117. Kanski JV. Kanski Clinical Ophthalmology. 6 ed. 2007.

118. Hodzic-Hadzibegovic D, Sander BA, Lund-Andersen H. Diabetic macular oedema quantified with spectral-domain optical coherence tomography--evaluation of boundary line artefacts and the effect on retinal thickness. Acta Ophthalmol. 2015;93:74–82. PMID: 25042850

119. Ctori I, Huntjens B. Repeatability of Foveal Measurements Using Spectralis Optical Coherence Tomography Segmentation Software. PLoS ONE. 2015;10:e0129005. PMID: 26076457

120. Taarnhøj NCBB, Larsen M, Sander B, Kyvik KO, Kessel L, Hougaard JL, et al. Heritability of retinal vessel diameters and blood pressure: a twin study. Invest Ophthalmol Vis Sci. 2006;47:3539–44. PMID: 16877426

121. Knudtson MD, Lee KE, Hubbard LD, Wong TY, Klein R, Klein BEK. Revised formulas for summarizing retinal vessel diameters. Curr Eye Res. 2003;27:143–9. PMID: 14562179

122. Reinhard M, Frystyk J, Jespersen B, Bjerre M, Christiansen JS, Flyvbjerg A, et al. Effect of hyperinsulinemia during hemodialysis on the insulin-like growth factor system and inflammatory biomarkers: a randomized open-label crossover study. BMC Nephrol. 2013;14:80. PMID: 23557110

123. Levy JC, Matthews DR, Hermans MP. Correct homeostasis model assessment (HOMA) evaluation uses the computer program. Diabetes Care. 1998;21:2191–2. PMID: 9839117

124. Crim MT, Yoon SSS, Ortiz E, Wall HK, Schober S, Gillespie C, et al. National surveillance definitions for hypertension prevalence and control among adults. Circ Cardiovasc Qual Outcomes. 2012;5:343–51. PMID: 22550130

125. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–74. PMID: 843571

126. Danish Ophthalmological Society: [Clinical guidelines for diabetic retinopathy - screening prevention and treatment] [in Danish] http://www.dansk-oftalmologisk- selskab.dk/Patiensforl%C3%B8b/for%20diabetetisk%20%C3%B8jensygdom.pdf accessed August 3rd 2015.

127. Merimee TJ, Zapf J, Froesch ER. Insulin-like growth factors. Studies in diabetics with and without retinopathy. N Engl J Med. 1983;309:527–30. PMID: 6348545

128. Janssen JA, Jacobs ML, Derkx FH, Weber RF, van der Lely AJ, Lamberts SW. Free and total insulin-like growth factor I (IGF-I), IGF-binding protein-1 (IGFBP-1), and IGFBP-3 and their relationships to the presence of diabetic retinopathy and glomerular hyperfiltration in insulin-dependent diabetes mellitus. J. Clin. Endocrinol. Metab. 1997;82:2809–15. PMID: 9284701

129. Feldmann B, Jehle PM, Mohan S, Lang GE, Lang GK, Brueckel J, et al. Diabetic retinopathy is associated with decreased serum levels of free IGF-I and changes of IGF-binding proteins. Growth Horm. IGF Res. 2000;10:53–9. PMID: 10753593

130. Dills DG, Moss SE, Klein R, Klein BE, Davis M. Is insulinlike growth factor I associated with diabetic retinopathy? Diabetes. 1990;39:191–5. PMID: 2227126

73

131. Hyer SL, Sharp PS, Sleightholm M, Burrin JM, Kohner EM. Progression of diabetic retinopathy and changes in serum insulin-like growth factor I (IGF I) during continuous subcutaneous insulin infusion (CSII). Horm. Metab. Res. 1989;21:18–22. PMID: 2925151

132. Frystyk J, Bek T, Flyvbjerg A, Skjaerbaek C, Ørskov H. The relationship between the circulating IGF system and the presence of retinopathy in Type 1 diabetic patients. Diabet Med. 2003;20:269–76. PMID: 12675639

133. Raymond NT, Varadhan L, Reynold DR, Bush K, Sankaranarayanan S, Bellary S, et al. Higher prevalence of retinopathy in diabetic patients of South Asian ethnicity compared with white Europeans in the community: a cross-sectional study. Diabetes Care. 2009;32:410–5. PMID: 19074992

134. Sivaprasad S, Gupta B, Gulliford MC, Dodhia H, Mann S, Nagi D, et al. Ethnic variation in the prevalence of visual impairment in people attending diabetic retinopathy screening in the United Kingdom (DRIVE UK). PLoS ONE. 2012;7:e39608. PMID: 22761840

135. Murphy R, Jiang Y, Booth M, Babor R, MacCormick A, Hammodat H, et al. Progression of diabetic retinopathy after bariatric surgery. Diabet Med. 2015;32:1212–20. PMID: 25689226

136. Miras AD, Chuah LL, Lascaratos G, Faruq S, Mohite AA, Shah PR, et al. Bariatric surgery does not exacerbate and may be beneficial for the microvascular complications of type 2 diabetes. Diabetes Care. 2012;35:e81. PMID: 23173142

137. Thomas RL, Prior SL, Barry JD, Luzio SD, Eyre N, Caplin S, et al. Does bariatric surgery adversely impact on diabetic retinopathy in persons with morbid obesity and type 2 diabetes? A pilot study. J Diab Comp. 2014;28:191–5. PMID: 24332764

138. Varadhan L, Humphreys T, Walker AB, Cheruvu CVN, Varughese GI. Bariatric surgery and diabetic retinopathy: a pilot analysis. Obes Surg. 2012;22:515–6. PMID: 22246396

139. White NH, Sun W, Cleary PA, Danis RP, Davis MD, Hainsworth DP, et al. Prolonged effect of intensive therapy on the risk of retinopathy complications in patients with type 1 diabetes mellitus: 10 years after the Diabetes Control and Complications Trial. Arch Ophthalmol. 2008;126:1707–15. PMID: 19064853

140. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HAW. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. 2008;359:1577–89. PMID: 18784090

141. Gaede P, Lund-Andersen H, Parving HH, Pedersen O. Effect of a multifactorial intervention on mortality in type 2 diabetes. N Engl J Med. 2008;358:580–91. PMID: 18256393

142. Browning DJ, Fraser CM, Propst BW. The Variation in Optical Coherence Tomography–Measured Macular Thickness in Diabetic Eyes Without Clinical Macular Edema. Am J Ophthalmol. 2008;145:889–93.

143. Biallosterski C, van Velthoven MEJ, Michels RPJ, Schlingemann RO, DeVries JH, Verbraak FD. Decreased optical coherence tomography-measured pericentral retinal thickness in patients with diabetes mellitus type 1 with minimal diabetic retinopathy. Br J Ophthalmol. 2007;91:1135–8. PMID: 17383994

144. Klein R, Myers CE, Lee KE, Gangnon R, Klein BEK. Changes in retinal vessel diameter and incidence and progression of diabetic retinopathy. Arch Ophthalmol. 2012;130:749–55. PMID: 22332203

145. Roy MS, Klein R, Janal MN. Retinal venular diameter as an early indicator of progression to

74

proliferative diabetic retinopathy with and without high-risk characteristics in African Americans with type 1 diabetes mellitus. Arch Ophthalmol. 2011;129:8–15. PMID: 21220623

146. Klein R, Klein BEK, Moss SE, Wong TY, Hubbard L, Cruickshanks KJ, et al. Retinal vascular abnormalities in persons with type 1 diabetes: the Wisconsin Epidemiologic Study of Diabetic Retinopathy: XVIII. Ophthalmology. 2003;110:2118–25. PMID: 14597518

147. Klein R, Klein BEK, Moss SE, Wong TY, Sharrett AR. Retinal vascular caliber in persons with type 2 diabetes: the Wisconsin Epidemiological Study of Diabetic Retinopathy: XX. Ophthalmology. 2006;113:1488–98. PMID: 16828517

148. Nguyen TT, Wang JJ, Sharrett AR, Islam FMA, Klein R, Klein BEK, et al. Relationship of retinal vascular caliber with diabetes and retinopathy: the Multi-Ethnic Study of Atherosclerosis (MESA). Diabetes Care. 2008;31:544–9. PMID: 18070990

149. Cheung N, Rogers SL, Donaghue KC, Jenkins AJ, Tikellis G, Wong TY. Retinal arteriolar dilation predicts retinopathy in adolescents with type 1 diabetes. Diabetes Care. 2008;31:1842–6. PMID: 18523143

150. Jeganathan VSE, Sabanayagam C, Tai ES, Lee J, Lamoureux E, Sun C, et al. Retinal vascular caliber and diabetes in a multiethnic Asian population. Microcirculation. 2009;16:534–43. PMID: 19488921

151. Tsai ASH, Wong TY, Lavanya R, Zhang R, Hamzah H, Tai ES, et al. Differential association of retinal arteriolar and venular caliber with diabetes and retinopathy. Diabetes Res Clin Pract. 2011;94:291–8. PMID: 21864932

152. Cheung N, Wong TY. Predicting risk of diabetic retinopathy from retinal vessel analysis: personalized medicine in transition. Arch Ophthalmol. 2012;130:783–4. PMID: 22801841

153. Broe R, Rasmussen ML, Frydkjaer-Olsen U, Olsen BS, Mortensen HB, Hodgson L, et al. Retinal vessel calibers predict long-term microvascular complications in type 1 diabetes: the Danish Cohort of Pediatric Diabetes 1987 (DCPD1987). Diabetes. 2014;63:3906–14. PMID: 24914239

154. Klein R, Klein BEK, Moss SE, Wong TY, Hubbard L, Cruickshanks KJ, et al. The relation of retinal vessel caliber to the incidence and progression of diabetic retinopathy: XIX: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Arch Ophthalmol. 2004;122:76–83. PMID: 14718299

155. Alibrahim E, Donaghue KC, Rogers S, Hing S, Jenkins AJ, Chan A, et al. Retinal vascular caliber and risk of retinopathy in young patients with type 1 diabetes. Ophthalmology. 2006;113:1499–503. PMID: 16828499

156. Klein R, Klein BEK, Moss SE, Wong TY. Retinal vessel caliber and microvascular and macrovascular disease in type 2 diabetes: XXI: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. Ophthalmology. 2007;114:1884–92. PMID: 17540447

157. Rogers SL, Tikellis G, Cheung N, Tapp R, Shaw J, Zimmet PZ, et al. Retinal arteriolar caliber predicts incident retinopathy: the Australian Diabetes, Obesity and Lifestyle (AusDiab) study. Diabetes Care. 2008;31:761–3. PMID: 18184895

158. Klefter ON. Retinal Physiological Adaptation to Changes in the Glucose Supply in Humans. PhD Thesis. University of Copenhagen. 2015.

75

159. Bandello F, Tejerina AN, Vujosevic S, Varano M, Egan C, Sivaprasad S, et al. Retinal Layer Location of Increased Retinal Thickness in Eyes with Subclinical and Clinical Macular Edema in Diabetes Type 2. Ophthalmic Res. 2015;54:112–7. PMID: 26315448

160. Ribeiro L, Bandello F, Tejerina AN, Vujosevic S, Varano M, Egan C, et al. Characterization of Retinal Disease Progression in a 1-Year Longitudinal Study of Eyes With Mild Nonproliferative Retinopathy in Diabetes Type 2. Invest Ophthalmol Vis Sci. 2015;56:5698–705. PMID: 26322834

161. Vujosevic S, Varano M, Egan C, Sivaprasad S, Menon G, Erginay A, et al. Relevance of Retinal Thickness Changes in the OCT Inner and Outer Rings to Predict Progression to Clinical Macular Edema: An Attempt of Composite Grading of Macular Edema. Ophthalmic Res. 2015;55:19–25. PMID: 26555067

162. Diabetic Retinopathy Clinical Research Network, Bressler NM, Miller KM, Beck RW, Bressler SB, Glassman AR, et al. Observational study of subclinical diabetic macular edema. Eye. 2012;26:833–40. PMID: 22441027

163. Grant MB, Mames RN, Fitzgerald C, Hazariwala KM, Cooper-DeHoff R, Caballero S, et al. The efficacy of octreotide in the therapy of severe nonproliferative and early proliferative diabetic retinopathy: a randomized controlled study. Diabetes Care. 2000;23:504–9. PMID: 10857943

164. Sandhu MS, Heald AH, Gibson JM, Cruickshank JK, Dunger DB, Wareham NJ. Circulating concentrations of insulin-like growth factor-I and development of glucose intolerance: a prospective observational study. Lancet. 2002;359:1740–5. PMID: 12049864

165. Rajpathak SN, He M, Sun Q, Kaplan RC, Muzumdar R, Rohan TE, et al. Insulin-like growth factor axis and risk of type 2 diabetes in women. Diabetes. 2012;61:2248–54. PMID: 22554827

166. Heald AH, Cruickshank JK, Riste LK, Cade JE, Anderson S, Greenhalgh A, et al. Close relation of fasting insulin-like growth factor binding protein-1 (IGFBP-1) with glucose tolerance and cardiovascular risk in two populations. Diabetologia. 2001;44:333–9. PMID: 11317665

167. Frystyk J, Vestbo E, Skjaerbaek C, Mogensen CE, Ørskov H. Free insulin-like growth factors in human obesity. Metab. Clin. Exp. 1995;44:37–44. PMID: 7476310

168. Frystyk J, Brick DJ, Gerweck AV, Utz AL, Miller KK. Bioactive insulin-like growth factor-I in obesity. J. Clin. Endocrinol. Metab. 2009;94:3093–7. PMID: 19470623

169. Frystyk J, Skjaerbaek C, Vestbo E, Fisker S, Ørskov H. Circulating levels of free insulin-like growth factors in obese subjects: the impact of type 2 diabetes. Diabetes Metab. Res. Rev. 1999;15:314–22. PMID: 10585616

170. Espelund U, Bruun JM, Richelsen B, Flyvbjerg A, Frystyk J. Pro- and mature IGF-II during diet-induced weight loss in obese subjects. European Journal of Endocrinology. 2005;153:861–9. PMID: 16322392

171. Belobrajdic DP, Frystyk J, Jeyaratnaganthan N, Espelund U, Flyvbjerg A, Clifton PM, et al. Moderate energy restriction-induced weight loss affects circulating IGF levels independent of dietary composition. Eur J Endocrinol. 2010;162:1075–82. PMID: 20212016

172. Jeyaratnaganthan N, Højlund K, Kroustrup JP, Larsen JF, Bjerre M, Levin K, et al. Circulating levels of insulin-like growth factor-II/mannose-6-phosphate receptor in obesity and type 2 diabetes. Growth Horm. IGF Res. 2010;20:185–91. PMID: 20110184

76

173. Pellitero S, Granada ML, Martínez E, Balibrea JM, Guanyabens E, Serra A, et al. IGF1 modifications after bariatric surgery in morbidly obese patients: potential implications of nutritional status according to specific surgical technique. Eur J Endocrinol. 2013;169:695–703. PMID: 23946276

174. Giusti V, Gasteyger C, Suter M, Heraief E, Gaillard RC, Burckhardt P. Gastric banding induces negative bone remodelling in the absence of secondary hyperparathyroidism: potential role of serum C telopeptides for follow-up. Int J Obes Relat Metab Disord. 2005;29:1429–35. PMID: 16077715

175. Jogie-Brahim S, Feldman D, Oh Y. Unraveling insulin-like growth factor binding protein-3 actions in human disease. Endocrine Reviews. 2009;30:417–37. PMID: 19477944

176. Wheatcroft SB, Kearney MT. IGF-dependent and IGF-independent actions of IGF-binding protein-1 and -2: implications for metabolic homeostasis. Trends Endocrinol. Metab. 2009;20:153–62. PMID: 19349193

177. Buchwald H, Avidor Y, Braunwald E, Jensen MD, Pories W, Fahrbach K, et al. Bariatric surgery: a systematic review and meta-analysis. JAMA. 2004;292:1724–37. PMID: 15479938

178. Wilhelm SM, Young J, Kale-Pradhan PB. Effect of bariatric surgery on hypertension: a meta-analysis. Ann Pharmacother. 2014;48:674–82. PMID: 24662112

179. Schiavon CA, Ikeoka DT, de Sousa MG, Silva CRA, Bersch-Ferreira AC, de Oliveira JD, et al. Effects of gastric bypass surgery in patients with hypertension: rationale and design for a randomised controlled trial (GATEWAY study). BMJ Open. 2014;4:e005702. PMID: 25200559

180. Klein R, Klein BE, Moss SE, Davis MD, DeMets DL. The Wisconsin epidemiologic study of diabetic retinopathy. III. Prevalence and risk of diabetic retinopathy when age at diagnosis is 30 or more years. Arch Ophthalmol. 1984;102:527–32. PMID: 6367725

181. Neff KJ, Chuah LL, Aasheim ET, Jackson S, Dubb SS, Radhakrishnan ST, et al. Beyond weight loss: evaluating the multiple benefits of bariatric surgery after Roux-en-Y gastric bypass and adjustable gastric band. Obes Surg. 2014;24:684–91. PMID: 24362538

182. Hayes MT, Hunt LA, Foo J, Tychinskaya Y, Stubbs RS. A model for predicting the resolution of type 2 diabetes in severely obese subjects following Roux-en Y gastric bypass surgery. Obes Surg. 2011;21:910–6. PMID: 21336560

183. Schauer PR, Burguera B, Ikramuddin S, Cottam D, Gourash W, Hamad G, et al. Effect of laparoscopic Roux-en Y gastric bypass on type 2 diabetes mellitus. Ann Surg. 2003;238:467–85. PMID: 14530719

184. Torquati A, Lutfi R, Abumrad N, Richards WO. Is Roux-en-Y gastric bypass surgery the most effective treatment for type 2 diabetes mellitus in morbidly obese patients? J. Gastrointest. Surg. 2005;9:1112–8. PMID: 16269382

185. Alexandrides TK, Skroubis G, Kalfarentzos F. Resolution of diabetes mellitus and metabolic syndrome following Roux-en-Y gastric bypass and a variant of biliopancreatic diversion in patients with morbid obesity. Obes Surg. 2007;17:176–84. PMID: 17476868

186. Ramos-Levi A, Sanchez-Pernaute A, Matia P, Cabrerizo L, Barabash A, Hernandez C, et al. Diagnosis of diabetes remission after bariatic surgery may be jeopardized by remission criteria and previous hypoglycemic treatment. Obes Surg. 2013;23:1520–6. PMID: 23702908

77

187. Engell C, Ravnbak T.[Background: This is what the Bertel case is about]. [in Danish] dr.dk/nyheder/indland/baggrund-det-handler-bertel-sagen-om Accessed August 1st 2015.

188. Horstmann JR. [Obesity surgeons leave in frustration]. [in Danish] ekstrabladet.dk/kup/sundhed/article4247471.ece Accessed August 1st 2015.

189. Heissel A. [Complications for weight loss surgery are flooding the hospitals]. [in Danish]. dagensmedicin.dk/nyheder/syge-fedmeopererede-valter-ind-pa-hospitalerne/ Accessed August 1st 2015.

190. Mingrone G, Panunzi S, De Gaetano A, Guidone C, Iaconelli A, Leccesi L, et al. Bariatric surgery versus conventional medical therapy for type 2 diabetes. N Engl J Med. 2012;366:1577–85. PMID: 22449317

191. Lind M, Odén A, Fahlén M, Eliasson B. The shape of the metabolic memory of HbA1c: re-analysing the DCCT with respect to time-dependent effects. Diabetologia. 2010;53:1093–8. PMID: 20237754

78