Risk factors associated with progression of diabetic retinopathy in eyes treated with panretinal photocoagulation

Uncontrolled diabetes has been associated with progression of diabetic retinopathy (DR) in several studies. Therefore, we aimed to investigate systemic and ophthalmic factors related to worsening of DR even after completion of panretinal photocoagulation (PRP). We retrospectively reviewed DR patients who had completed PRP in at least one eye with a 3-year follow-up. A total of 243 eyes of 243 subjects (mean age 52.6 ± 11.6 years) were enrolled. Among them, 52 patients (21.4%) showed progression of DR after PRP (progression group), and the other 191 (78.6%) patients had stable DR (non-progression group). The progression group had higher proportion of proliferative DR (P = 0.019); lower baseline visual acuity (P < 0.001); and higher platelet count (P = 0.048), hemoglobin (P = 0.044), and hematocrit, (P = 0.042) than the non-progression group. In the multivariate logistic regression analysis for progression of DR, baseline visual acuity (HR: 0.053, P < 0.001) and platelet count (HR: 1.215, P = 0.031) were identified as risk factors for progression. Consequently, we propose that patients with low visual acuity or high platelet count are more likely to have progressive DR despite PRP and require careful observation. Also, the evaluation of hemorheological factors including platelet counts before PRP can be considered useful in predicting the prognosis of DR.


Introduction
Diabetic retinopathy (DR) is the leading cause of decreased vision and blindness among working-age adults in most developed countries. [1][2][3] Chronic hyperglycemia in diabetes mellitus (DM) causes impairment of capillaries, resulting in retinal ischemia and increase of vascular permeability. DR progresses from nonproliferative diabetic retinopathy (NPDR) to proliferative diabetic retinopathy (PDR), a stage that can cause blindness due to the formation of retinal neovascularization, resulting in vitreous hemorrhage or tractional retinal detachment. 4,5 Panretinal photocoagulation (PRP) has been the main treatment to prevent severe vision loss in patients with severe NPDR or PDR according to the ndings of studies on DR, and the effectiveness of the treatment was con rmed by the early treatment diabetic retinopathy study (ETDRS). 6 PRP is usually completed in 4 or more sessions. Complete PRP has reduced the ve-year risk of blindness in patients with PDR by over 90%. 7 But, several studies have reported that 45% of the eyes that were treated with PRP needed additional PRP to manage active PDR. 8 Susan et al reported that PDR worsening occurred in 42% of those in the PDR treated with PRP group and in 34% of those in the PDR treated with ranibizumab group. 8 It was revealed that uncontrolled diabetes, diabetic nephropathy, anemia, etc., were associated with progression of PDR in several studies. However, there were a few studies that have focused on systemic factors associated with worsening of PDR after completion of PRP. 8 Therefore, we aimed to explore systemic factors related to worsening of PDR after completion of PRP using the clinical data warehouse (CDW) system. To explore the large database of clinical information of patients with PDR, we used the CDW system of the hospital, which is an electronic data repository of patient and provider information. This is one of the largest studies on the association of systemic Page 3/19 conditions such as comorbid diseases with PDR utilizing clinical data and laboratory test results in Asians. The ndings of this study are expected to provide insight into the pathogenesis of PDR progression.

Results
The data of 1052 subjects with DR were extracted from the CDW system. After excluding patients who did not meet the criteria, a total of 243 eyes of 243 individuals were eligible for this study. Of 243 patients, 52 (21.4%) who showed progression of DR after PRP were de ned as the progression group, and the other 191 (78.6%) patients comprised the non-progression group. Laboratory and ophthalmic ndings Table 1 shows clinical characteristics of all participant and the comparison between the progression and non-progression groups. The mean age of the total study group was 52.6 ± 11.6 years and the mean duration of DM was 15.5 ± 7.85 years, with a mean hemoglobin A1c (HbA1c) level of 8.65 ± 1.80%. Out of 243 patients, there were 50 (20.6%) patients with severe NPDR and 193 (79.4%) patients with PDR. PDR patients were signi cantly higher in the progression group than in the non-progression group (90.4% vs 76.4%, P = 0.019). In particular, the baseline visual acuity was signi cantly lower in the progression group (0.47 ± 0.33) than in the non-progression group (0.69 ± 0.29) (P < 0.001). Platelet count, hemoglobin, and hematocrit were higher in the progression group than in the non-progression group (P = 0.035, 0.014, 0.011, respectively). Phosphorus, uric acid and total cholesterol values were signi cantly different between two groups (P = 0.048, 0.044, 0.042, respectively). Other laboratory ndings did not show any statistical signi cance.

Diagnostic validity of DR progression
To validate the diagnostic performance of DR progression, the area under the receiver operating characteristic curve (AUROC) of visual acuity and platelet count were derived (Fig. 1). The AUROC of the visual acuity was 0.728 (P < 0.001, CI 0.645 -0.811) and platelet count was 0.722 (P < 0.001, CI 0.648 -0.796). Then, the cut-off point was calculated by considering the balance between sensitivity and speci city for the ability to distinguish the progression of DR. The cut-off point was visual acuity, 0.505 (sensitivity: 71.2%; speci city: 71.7%); and platelet counts, 253 (×10 3 /µL) (sensitivity: 75.0%; speci city: 61.2%).

Discussion
This study investigated the predictive markers for DR progression in patients after completion of PRP. A data warehouse was used to analyze a wide range of ophthalmic and laboratory variables in a large patient population. After adjusting for various confounding factors, lower visual acuity at the start of PRP among ophthalmic parameters and higher platelet count among laboratory parameters were found to be predictive of the high probability of DR progression after PRP.
The present study showed that DR signi cantly progressed after PRP when the initial visual acuity was low. While the action mechanism of the PRP is unknown, it is assumed that the PRP reduces metabolic demand and promotes diffusion of oxygen from the choroid to the retina. 19 However, in cases where the macular ischemic condition presenting as low visual acuity is severe, diffusion of oxygen may remain insu cient despite PRP, and DR could actually worsen. In particular, this nding in line with the result of a study that reported that low visual acuity is associated with larger foveal avascular zone size in patients with DR. 20 Poor visual acuity can imply macular ischemia, which is likely to be associated with a progression of DR. In addition, higher platelet count showed signi cant association with worsening of DR in the present study. Altered platelet morphology and function have been observed in diabetes in the form of enhanced platelet activity, which may contribute to this "prothrombotic state". 21 Furthermore, a few studies reported that diabetic patients have higher thrombocyte count and platelet activation can lead to the generation of vascular diseases. 22,23 Thus, more attention needs to be paid to patients with higher platelet count for monitoring DR progression.
DR severity is also well-known as a strong factor associated with DR progression. The ETDRS registered 3711 patients with mild-to-severe nonproliferative or early proliferative DR and randomly assigned each patient to undergo early photocoagulation or delayed treatment until high-risk proliferative retinopathy was detected. At 5 years, rates of severe vision loss were 2.6% with early treatment and 3.7% with deferred treatment. 31 Bressler et al also reported that higher baseline levels of ETDRS retinopathy were associated with higher rates of supplemental PRP in the PRP group. 8 These are consistent with our results in the univariate analysis; however, severity of DR was not a signi cant factor in multivariate analysis. This could be explained by the relatively small proportion of severe NPDRs or the effect of other confounding variables. HbA1c has been known to be a useful marker to determine mean blood glucose levels. 21 However, HbA1c was not predictive of progression of DR after PRP in our study. HbA1c was slightly higher in the progression group than the non-progression group (8.68 ± 1.82 vs 8.55 ± 1.73), but this was not statistically signi cant. This nding is consistent with a report that total glycemic exposure only accounts for approximately 11% of the variation in DR progression, as shown in the Diabetes Control and Complications Trial. 21 We hypothesized that several systemic factors might be associated with progression of DR. Anemia was one of the systemic factors considered by us to be associated with DR progression. Anemia is a decrease in hemoglobin concentration and a sign of reduced oxygen-carrying capacity. In this study, the level of hemoglobin and hematocrit were lower in the progression group than in the non-progression group.
Although not identi ed as a signi cant factor in multivariate regression analyses, the aforementioned results suggest that lower hemoglobin and hematocrit levels support the pathophysiological understanding of the reduction in oxygen-carrying capacity that exacerbates tissue hypoxia. 32 In present study, glaucoma was identi ed in 22 (11.5%) patients in the non-progression group and 6 (11.5%) patients in the progression group. There were 5 neovascular glaucoma (NVG), 6 normal-tension glaucoma (NTG), and 11 primary open-angle glaucoma (POAG) patients in the non-progression group and 4 NVG and 2 NTG patients in the progression group. Glaucoma is known to occur either by ischemia (ischemic theory) as well as through mechanical effect (mechanical theory). We hypothesized that the glaucomatous eye was more susceptible to the ischemic state; thus, the probability of DR progression would increase in glaucoma patients. However, contrary to our assumption, the proportion of NTG and POAG patients in the progression group was not high.
In our study, uric acid and total cholesterol showed signi cant association with DR progression in the univariate analysis but not in the multivariate analysis. Hyperuricemia has been found to independently predict the development of diabetes and mediate insulin resistance in both fructose-dependent and fructose-independent models of metabolic syndrome. 33 Abnormal total cholesterol level is a risk factor for occurrence of sight-threatening DR and diabetic macular edema, and higher serum triglyceride level is a risk factor for PDR in subjects with type 2 diabetes. Dyslipidemia with abnormal levels of total cholesterol, LDL, HDL, and triglyceride is associated with greater risk of incident diabetic macular edema and greater risk for progression to PDR, compared with abnormal levels of individual lipid levels. 34 The exploratory analyses conducted in this report have some limitations. First, PRPs were performed by two different surgeons even though protocols were the same.

Methods
This investigation is based on the Hallym DR Study, an ongoing cohort study conducted at Hallym University Medical Center (HUMC). To explore the large database of hospital-collected clinical information on patients with DR, we used the common integrated CDW system of HUMC, which is an electronic data repository of patients' information. 9 A detailed description of the CDW system was also introduced in previous studies, with some modi cation. 10,11 The common CDW system of HUMC collects and stores extensive electronic medical data including medical records, laboratory results, physical measurements, diagnostic and therapeutic history, and medication history over a period of 10 years. 11 We accessed the CDW system and investigated the medical data of patients who were diagnosed with DR and treated with PRP between January 2009 and December 2015. This study was approved by the institutional review board of HUMC, and all protocols were in accordance with the tenets of the Declaration of Helsinki. The need for informed consent was waived by the Institutional Review Boards of Hallym University Sacred Heart Hospital because of the retrospective nature of the study and the deidenti cation of data by the CDW system before we accessed the database.

Study population
We rst identi ed patients diagnosed with DR, Korean Standard Classi cation of Diseases (KCD) code H34.8, corresponding to the International Classi cation of Diseases, 9 th Revision, Clinical Modi cation (ICD-9-CM) code 362.01 for DR during the study period. Next, to ensure the inclusion of patients who newly underwent PRP during the follow-up period, we veri ed the presence of a previous history of PRP by reviewing the visit data for all eligible patients, beginning from the earliest period for which medical records were provided by the CDW system (January 2009 for HUMC).
DR patients who have completed more than 4 sessions of PRP in at least one eye from 2009 to 2015 and who met these criteria were included: 1) followed up for at least 3 years after completing PRP; and 2) able to con rm the clinical information (underlying systemic comorbidities, physical measurements, and laboratory ndings of blood tests and urine tests). Exclusion criteria were: 1) a history of other retinal disease, neovascular age-related macular degeneration, retinal vein occlusion, posterior uveitis, or ischemic optic neuropathy; 2) a history of intraocular surgery other than uncomplicated cataract surgery; 3) media opacity rendering fundus reading di cult for diagnosis (signi cant cataract, asteroid hyalosis, or vitreous opacity); 4) a history of laser before PRP; and 5) advanced DR with complications requiring immediate surgical treatment, 12 and only those tests whose results were available for more than 80% of the study participants were included in the analyses.
Demographic characteristics included the patients' sex and age at initial PRP. Systemic comorbidities were investigated using the KCD code system. We also investigated the presence of underlying disease that may affect retinal vasculature, including hypertension, ischemic heart disease (IHD), cerebrovascular disease, and chronic kidney disease. Physical measurements included height, weight, systolic blood pressure, diastolic blood pressure, and the body mass index (BMI (Obese). 10,14 systolic blood pressure, diastolic blood pressure were measured in the right arm after a 5minute stabilization period using a standard mercury sphygmomanometer (Baumanometer; Baum, NY, USA). Further, the level of smoking was categorized as ''have never smoked,'' ''previously smoked but no longer smoking,'' or ''currently smoking.'' Outcome measurements, ophthalmic variables All patients underwent comprehensive ocular examinations, including best-corrected visual acuity (BCVA, Snellen chart), intraocular pressure, detailed slit-lamp biomicroscopy and dilated fundus examination after dilatation of the pupils, fundus photography, optical coherence tomography (OCT) imaging, and uorescein angiography. IOP was measured using a non-contact tonometer (CT-80 or CT-1P; Topcon Inc., Tokyo, Japan), and fundus photographs were taken using a 45° digital fundus camera (CR6-45NW; Canon Inc., Utsunomiya, Japan or TRC-NW8, Topcon Inc., Tokyo, Japan). OCT imaging was performed using the swept-source mode of a high-de nition OCT system (DRI OCT Triton, Topcon, Tokyo, Japan). An ultra-wide-eld scanning laser ophthalmoscope (Optos Optomap Panoramic 200MA; Optos PLC, Dunfermline, Scotland) allows wide-angle retinal imaging. During follow-up periods, we checked occurrence of NVG, types and number of intravitreal injection, occurrence of vitreous hemorrhage, tractional retinal detachment, and implementation of pars plana vitrectomy. The presence of any type of glaucoma, POAG, normal-tension glaucoma NTG, NVG, and others was also investigated.

Evaluation and management of diabetic retinopathy
The stage of DR was determined by comparison with standard photographs from the ETDRS. 15 If PDR or progression of severe NPDR was suspected in the fundus photography, uorescein angiography was conducted. Indications of PRP were de ned as PDR, very severe NPDR 15 , or aggravation of severe NPDR.
Intravitreal injections of anti-vascular endothelial growth factor were given in cases of diabetic macular edema with central macular thickness of above 300 µm or vitreous hemorrhage.
According to DRS protocol using a standard argon-type laser PRP, the recommended settings include burns that range approximately 400 μm in size, pulse durations of 100 milliseconds, and 200 mW of power.
Laser burns (1200 to 1600) are evenly beamed or scattered on the retina away from the macula, almost to the equator. Burns were spaced at a one-burn spacing pattern. PRP was performed across 4 treatment sessions, 1 session performed per week. 7 De nition of DR progression In this study, DR worsening was assessed in patients with prior PRP using the previously described composite end point of time to new proliferative event. [16][17][18] This composite end point takes into account clinical outcomes associated with DR worsening as de ned by progression to PDR, any occurrence of newly diagnosed iris or retinal neovascularization, treatment with PRP or vitrectomy for DR-related reasons, or new cases of PDR identi ed by ophthalmoscopy. [16][17][18] The clinical experiences of patients who underwent on-study PRP were assessed by determining the incidence and timing of rst on-study occurrences of vitrectomy, retinal neovascularization, or iris neovascularization. Progressive DR changes were con rmed and agreed on by the same two experienced specialists (S.K, I.W.P), each of whom was masked to the subject's identity and to all other test results.

Progression group and non-progression group
The patients were subdivided into progression group and non-progression group according to progression of DR: the progression group that consisted of eyes exhibiting DR progression (progression to PDR or newly developed NVI or NVE or NVG, or implementation of vitrectomy), and the 'non-progression' group that consisted of eyes exhibiting stationary DR.

Statistical analysis
The baseline demographics and clinical variables were summarized by means and standard deviations or frequencies and percentages, as appropriate. The clinical characteristics of the progression group versus non-progression group were compared using unpaired t-tests or Mann-Whitney U tests for continuous values and the Chi-square test for categorical variables. Univariate and multivariate logistic regression analyses employing a forward conditional method were performed to determine the associations of various factors with progression of DR; hazard ratios (HRs) and 95% con dence intervals (CI) were reported. To avoid multicollinearity, variables correlated signi cantly with each other were not analyzed simultaneously. Instead, the variable with the highest signi cance among correlated variables was chosen. If signi cances were similar between correlated variables, multiple analyses were conducted separately using each variable. Kaplan-Meier survival analysis was used to compare the inter-group cumulative probability of maintenance of the DR without progression, as strati ed by the signi cant variables derived from multivariate logistic regression. All statistical analyses were performed using SPSS version 21.0 (SPSS, Chicago, IL, USA). All P-values were two-sided and considered signi cant when P <0.05.