The Relationship Between Retinal and Choroidal Thickness and Adiponectin Concentrations in Patients With Type 2 Diabetes Mellitus

Abstract

Purpose: To evaluate the association between retinal and choroidal thickness and serum and aqueous humor (AH) adiponectin concentrations in patients with diabetic retinopathy (DR).

Methods: This prospective study enrolled diabetic patients without DR (group 1, n = 46) and with DR (n = 130). Central foveal thickness (CFT), subfoveal choroidal thickness (SCT), and adiponectin in serum and AH concentrations were compared. For subgroup analysis, the DR group was divided into four subgroups: mild (group 2), moderate (group 3), severe nonproliferative DR (group 4), and panretinal photocoagulation (group 5).

Results: The log-transformed serum and AH adiponectin concentrations in patients with DR (groups 2–5) were higher than in patients without DR (all Ps < 0.001). In addition, serum and AH adiponectin concentrations showed a positive linear correlation with DR severity (P < 0.001 and P = 0.001, respectively). In univariate analysis between serum or AH adiponectin concentrations and CFT or SCT, AH adiponectin significantly correlated with CFT and SCT (all Ps < 0.001). However, serum adiponectin concentration significantly correlated with SCT (P = 0.041) but not with CFT (P = 0.337). In multivariate analysis, AH adiponectin concentration significantly correlated with CFT, but serum adiponectin concentration did not (P = 0.002 and 0.309, respectively). In contrast, serum and AH adiponectin concentrations significantly correlated with SCT (P = 0.048 and 0.041, respectively).

Conclusions: Serum and AH adiponectin concentrations are positively associated with DR development and progression. Additionally, SCT looks related to the serum and AH adiponectin concentrations, whereas CFT looks related to AH adiponectin concentrations.

Diabetes mellitus (DM) is a systemic inflammatory condition that can cause various vascular complications.¹ Diabetic retinopathy (DR), one of the microvascular complications of DM, is related to multiple inflammatory cytokines such as vascular endothelial growth factor (VEGF), platelet derivative growth factor, matrix metalloproteinases, IL-6, pentraxin-3, adiponectin (APN), and adhesive molecules including intercellular adhesion molecules and vascular cell adhesion molecules (VCAM).^2–5 VEGF, one of the major target cytokines in treating DR complications, affects DR development and progression through various mechanisms, including loss of vascular wall integrity, vascular leakage, and fluid accumulations.^6–8 VEGF produced by the retinal pigment epithelium can affect the adjacent tissue, choroid, and retina by inducing principal structural changes in the retina and choroid.⁹ In fact, several optical coherence tomography (OCT) studies have demonstrated that increases in retinal and choroidal thickness depends on DR progression.^10–14 Chen et al.¹⁵ also report that macular choroidal thickness is significantly correlated with aqueous VEGF but not with serum VEGF in myopic eyes. 

Obesity can cause primary microvascular dysfunction in many types of tissues, including the eyes.¹⁶ Adipocytes can produce pro-angiogenic molecules, including TNF-α, IL-6, VEGF, and anti-angiogenic factors including serpins.¹⁷ Thus adipocytes may be implicated in the pathogenesis of vasculopathy in type 2 DM. APN is one of the most abundant cytokines secreted from adipocytes and is a relatively small molecule compared with VEGF.^18–20 Previously, we reported the positive relationship between APN in serum and aqueous humor (AH) and DR development and progression.⁵ Omae et al.²⁰ also reported that serum APN levels showed increased retinal blood flow in early DM patients. Therefore, although systemic and local APN may play important roles in retinal and choroidal structural changes, related studies using OCT have not yet been reported. In this study, we have investigated the relationship between APN in serum and AH and central foveal thickness (CFT) or subfoveal choroidal thickness (SCT) using OCT in patients with DR. 

Methods

Study Design

Patients (≥40 years old) scheduled for cataract surgery were consecutively recruited from the outpatient clinic of the Division of Ophthalmology at the Ophthalmology Department of Veterans Health Service Medical Center, South Korea, between March 2020 and February 2022. The ethics committee of the Veterans Health Service Medical Center (approval numbers: 2021-05-012-001) approved this study, and informed consent was obtained from all participants. The research protocol complied with the tenets of the Declaration of Helsinki. All patients had a basic physical examination, blood sampling, and a complete eye examination, which included best-corrected visual acuity, tonometry, bio-microscopy, and ophthalmoscopy using a 90-diopter lens or indirect ophthalmoscopy after pupil dilation to check for DR grade and retinal abnormality before cataract surgery. As described in a previous study, 0.1 mL AH was collected during cataract surgery from all patients.⁵ Patients with serious medical problems such as uncontrolled hyperlipidemia, hypertension, cardiovascular diseases, hepatobiliary diseases, immunological disorders, history of cancer, severe infectious conditions, renal failure, or severe ocular diseases were excluded in this study. We excluded patients with total cholesterol ≥240 mg/dL, LDL cholesterol ≥160 mg/dL, triglycerides ≥200 mg/dL, and HDL cholesterol <40 mg/dL for men and <50 mg/dL for women who receive proper medications including statins for uncontrolled dyslipidemia. And for hypertension, we excluded patients with systolic blood pressure greater than 140 mm Hg or diastolic blood pressure greater than 90 mm Hg, even if on antihypertensive therapy. We also excluded patients with active liver disease or pancreatitis, a history of autoimmune diseases or immunosuppressive treatments, a history of cancer or currently receiving treatment for cancer, active infections or a recent history of severe infections, creatinine levels above the normal range, or on dialysis. However, the described criteria were modified and finally determined by an internal medicine doctor in cases of patients with comorbidities such as cardiovascular disease and stroke. Furthermore, we excluded patients with severe retinopathy including proliferative diabetic retinopathy (PDR), severe diabetic macular edema (CFT >350 µm), retinal detachment, wet age-related macular degeneration, or patients undergoing eye surgery for conditions such as glaucoma. 

This study enrolled 130 DM patients with DR and 46 age- and body mass index (BMI)–matched DM patients without DR (group 1) and underwent all examinations. For subgroup analysis, DR patients were divided into five subgroups: mild non-proliferative diabetic retinopathy (NPDR, group 2), moderate NPDR (group 3), severe NPDR (group 4), proliferative DR (PDR, excluded by not being eligible for cataract surgery), and patients who underwent panretinal photocoagulation (PRP, group 5), based on a fundus examination during screening or immediately after the surgery. DM without DR was graded in the absence of overt retinopathy, and mild, moderate, and severe NPDRs were graded based on the early treatment diabetic retinopathy study. The patients who underwent PRP at least 6 months before the surgery were enrolled in group 5. Two trained ophthalmologists (H.Y.H. and Y.J.C.) blinded to the study design performed all fundus and OCT examinations and graded patients based on their consensus. One day before cataract surgery, spectral-domain OCT (Spectralis; Heidelberg Engineering, Heidelberg, Germany) examinations were conducted for CFT and SCT during 16:00–18:00. CFT and SCT measurements were conducted based on our previous studies.^21,22 In addition, only one eye from each patient was selected using digitalized randomization. Laboratory measurements were also conducted following our previous study.⁵ 

Statistical Analysis

Statistical analysis was performed using SPSS Statistics version 17 (SPSS Inc. Chicago, IL, USA). An independent t-test was used to compare serum and AH APN concentrations in patients with and without DR. Levene's test was performed for homogeneity of variance of the test parameter, and the homogeneity of variance was assumed when P > 0.05. One-way analysis of variance was used for subgroup comparisons. Post hoc analysis was performed to compare baseline characteristics among the five groups using Tukey's test and Bonferroni correction. In nonparametric analysis, categorized parameters were analyzed using the χ² independent test, and continuous parameters were analyzed using the Kruskal-Wallis test. The Pearson correlation coefficient was used for trend analysis, according to the characteristics of the data. Variables that do not show a normal distribution were converted to a normal distribution by substituting a logarithm. Multivariate analysis was used to remove the effect of each variable on other variables. Moreover, the variable's value was maximized while neutralizing the effect of multicollinearity by calculating the serum APN and the AH APN variables separately. A test for linear trend was calculated based on the average levels of APN in serum and AH within each group and subgroup. Statistical significance was set at P < 0.05. Intra-assay precision was determined for identical samples by evaluating each sample at least six times. Interassay precision was determined for assaying aliquots of each sample in four to six independent analytical runs, as described in the previous study.⁵ 

Results

Baseline Characteristics

Table 1 summarizes the clinical characteristics of the study population, including laboratory data and OCT parameters. Among all participants, 135 (76.7 %) were male and 41 (23.3%) were female, with a mean age of 74.5 ± 4.9 years. The sex, age, BMI, statin use, and lipid profiles were not significantly different between the five groups (P = 0.333, P = 0.254, P = 0.966, P = 0.158, and all P ≥ 0.100, respectively). However, the duration of DM, HbA1c, log serum APN, log AH VEGF, and log AH APN values were significantly different among the five groups (all Ps < 0.001). Of the total of 176 participants in the present study, the serum APN concentrations (log APN in serum) for group 1 (n = 46), group 2 (n = 41), group 3 (n = 42), group 4 (n = 21), and group 5 (n = 26) were 5.51 ± 3.57 µg/mL (6.67 ± 0.26), 8.38 ± 6.75 µg/mL (6.83 ± 0.25), 7.41 ± 3.03 µg/mL (6.83 ± 0.20), 8.59 ± 4.10 µg/mL (6.88 ± 0.23), and 8.99 ± 7.32 µg/mL (6.85 ± 0.30), respectively (P < 0.001). The AH APN concentrations (log APN in AH) for groups 1, 2, 3, 4, and 5 were 8.00 ± 9.00 ng/mL (3.71 ± 0.39), 13.25 ± 11.00 ng/mL (3.98 ± 0.37), 17.80 ± 14.14 ng/mL (4.11 ± 0.36), 46.90 ± 25.83 ng/mL (4.60 ± 0.26), and 32.05 ± 22.2 ng/mL (4.38 ± 0.37), respectively (P < 0.001). The log-transformed serum and AH APN and AH VEGF in patients with DR showed higher concentrations than in patients without DR (Fig. 1, all Ps < 0.001). However, log serum VEGF did not differ between non-DR and DR patients (P = 0.214). There is a significant correlation between the duration of diabetes and the levels of both serum APN and AH APN for groups 1–4 (R = 0.161 and 0.220; P = 0.049 and 0.007; respectively). 

Table 1.

 

Demographic Data and Values of Various Parameters (Mean ± Standard Deviation) Depending on the DR Grades

 

Figure 1.

 

Comparison of log-transformed APN (A, B) or VEGF (C, D) concentrations in serum and AH of patients with or without DR. The P values were estimated using an independent t-test. The middle lines of the boxes represent the median of each group.

Correlation Between Two OCT Parameters (CFT or SCT) and Two Cytokines (APN or VEGF) in Serum and AH

CFT (Table 2, groups 1–5) showed significant positive correlations with APN and VEGF concentrations in AH (R = 0.367 and 0.348; P < 0.001 and P < 0.001; respectively). However, CFT did not significantly correlate with the two cytokines, VEGF and APN, in serum (R = 0.073 and 0.110; P = 0.337 and 0.145, respectively). SCT (Table 2, groups 1–5) showed significant positive correlations with APN and VEGF concentrations in AH (R = 0.253 and 0.236; P = 0.001 and 0.002, respectively). SCT also showed a weak correlation with serum APN concentration (R = 0.154; P = 0.041) but did not significantly correlate with serum VEGF concentration (R = 0.006; P = 0.938). Figure 2 shows the correlation between APN and VEGF in serum and AH and retinal and choroidal thickness. 

Table 2.

 

Pearson Correlation Between Clinical Characteristics Including Various Serum and AH Cytokines and Optical Coherence Tomographic Findings, Such as CFT and SCT in DR, Including Eyes Undertaken Pan-Retinal Photocoagulation (n = 176 Eyes)

 

Figure 2.

 

Comparison of log-transformed APN (A, B) or VEGF (C, D) concentrations in serum and AH with CFT (blue color) and SCT (orange color) in patients with type 2 DM. The P values were estimated using Pearson correlation coefficient. *P < 0.05.

Tables 3 and 4 show the multivariate analysis between CFT and SCT and clinical parameters. AH APN and VEGF concentrations showed a significant correlation with CFT, but those in serum did not correlate (Table 3; P = 0.002, 0.001, 0.309, and 0.358, respectively). For SCT, however, serum and AH APN concentrations showed a significant correlation with SCT, but serum and AH VEGF concentrations did not correlate (Table 3; P = 0.048, 0.041, 0.801, and 0.076, respectively). 

Table 3.

 

Multivariate Linear Regression Analysis with Backward Elimination to Determine Clinical Factors, Including Cytokines in Serum and AH, For Relating Central Foveal Thickness (CFT)

 

Table 4.

 

Multivariate Linear Regression Analysis With Backward Elimination To Determine Clinical Factors, Including Cytokines in Serum and AH, for Relating the SCT in DM Patients With or Without DR

 

Correlation Between CFT or SCT and DR Grades

Table 2 shows the correlation between clinical data. CFT was significantly correlated with all DR grades. In all groups (group 1–5), CFT was significantly correlated with DR grade (R = 0.275; P < 0.001), and SCT was not correlated with DR grade (R = 0.167; P = 0.027). In addition, in the non-PRP group (Table 5), CFT and SCT showed a better significant correlation with DR grade (R = 0.465 and 0.271; P < 0.001 and 0.001, respectively). 

Table 5.

 

Pearson Correlation Between Clinical Characteristics Including Various Serum and AH Cytokines and Optical Coherence Tomographic Findings, Such as CFT and SCT in Non-Treatment DR Eyes (n = 150 eyes)

 

Correlation Between DR Grades and APN or VEGF in Serum and AH

Figures 3A and 3B show the relationship in all groups (groups 1–5), and Figures 3C and 3D show the relationship in non-PRP groups (groups 1–4). In the group analysis, APN and VEGF concentrations in AH showed a positive correlation with DR grades (all Ps < 0.001; Figs. 3B, 3D). In addition, serum APN concentration showed a significant positive correlation with DR grades (all Ps ≤ 0.001; Figs. 3A, 3C), but serum VEGF concentration failed to show a correlation with DR grades (all Ps ≥ 0.052). For all comparisons between DR grades and APN or VEGF in serum and AH, non-PRP groups (groups 1–4) correlated more than all groups (Fig. 3). 

Figure 3.

 

Correlation using Pearson correlation coefficient between DR grade (A, B: group 1–5 with PRP; C, D: group 1–4 without PRP) and log-transformed APN (orange color) and VEGF (blue color) in serum (A, C) and AH (B, D). *P < 0.05.

Figure 4 depicts the receiver operating characteristic curves for DR progression using log APN and log VEGF concentrations in AH and serum. For AH and serum APN, the area under the receiver operating characteristic curves were 0.805 (P < 0.001) and 0.694 (P < 0.001), respectively (Fig. 4A). With a cutoff value of 4437.8 pg/mL for AH APN and 4645462.9 pg/mL for serum APN, the diagnostic sensitivity and specificity for DR progression were 60.9% and 90.0% for AH APN and 54.3% and 77.7% for serum APN, respectively. For AH and serum VEGF, the area under the receiver operating characteristic curves were 0.717 (P < 0.001) and 0.548 (P = 0.338), respectively (Fig. 4B). With a cutoff value of 64.2 pg/mL for AH VEGF concentration and 32.2 pg/mL for serum VEGF concentration, the diagnostic sensitivity and specificity for DR progression were 54.3% and 80.0% for AH VEGF and 34.8% and 86.2% for serum VEGF, respectively. 

Figure 4.

 

Receiver operating characteristics (ROC) curves for log APN (A) and VEGF (B) in serum and AH for predicting the presence of diabetic retinopathy in patients with type 2 DM.

Intra- and Inter-Assay Precision

The intra-assays of three analytes for APN, VCAM-1, and VEGF in AH were 4.7%, 3.1%, and 2.6%, respectively. Inter-assays of three analytes for APN, VCAM-1, and VEGF in AH were 3.3%, 4.3%, and 3.1%, respectively. The intra-assays of three analytes for APN, VCAM-1, and VEGF in serum were 4.1%, 3.2%, and 4.0%, respectively. Inter-assays of three analytes APN, VCAM-1, and VEGF in serum were 3.6%, 4.1%, and 3.5%, respectively. All the intra-assay and interassay precision looked acceptable for duplication analysis using a Luminex 200 (Luminex corp., Austin, TX, USA). 

Discussion

In this study, we found the following results: (1) AH APN concentrations were significantly correlated with CFT and SCT, (2) serum APN was weakly correlated with SCT and CFT, (3) in serum and AH samples, APN and VEGF showed a similar pattern of relationship to CFT and SCT, but APN showed more correlation with CFT and/or SCT than VEGF, (4) DR grade was significantly correlated with serum and AH APN concentrations, and (5) CFT and SCT were positively correlated with DR grade, but this tendency was slightly reduced after PRP. To the best of our knowledge, this is the first study to assess the relationship between serum or AH APN concentrations and CFT and SCT in DR patients. 

Because DR is a multifactorial disease, it is challenging to identify a single molecular target in this chronic and multifactorial disease. Currently, VEGF is the most popular biomarker for DR progression and complications. In addition, anti-VEGF is an essential treatment option for edema in DR and DR progression. However, the anti-VEGF response is limited in some cases, so various molecular targets including platelet derivative growth factor, pentraxin-3, C-reactive protein, IL families, and matrix metalloproteinases have been proposed.^2–5 We also summarized the role of APN in the inflammation cascade and suggested its potential role in DR development and progression.^5,23,24 

APN has been shown to act as a proinflammatory growth factor with anti-inflammatory and anti-atherosclerotic effects in injured major vessels.^23,25,26 However, as previously stated, the results on microvascular complications in the eye were inconsistent.⁵ In addition, several studies have demonstrated a positive correlation between APN and DR severities, which is consistent with our findings.^5,20,26–34 In an experimental study, APN levels were found to be higher in animal models with retinal vein occlusion and oxygen-induced retinopathy, and anti-APN antibody injections reduced retinal edema and ischemia.³⁵ Furthermore, elevated serum APN, as a marker of microvascular complications in patients with DM, indicates poor glycemic control and the development of complications, including retinopathy, neuropathy, and nephropathy.³⁶ Thus there is a conflicting issue about the relationship between APN and DR. It is exceedingly difficult to answer the question, “Why is APN increased in patients with microvascular complications, especially retinopathy?” It is possible that the increase in APN levels observed in diabetic retinopathy may be due to a number of factors, including reactive compensation after the progression of the disease, an increase in circulatory APN as a result of impaired renal function, or proinflammatory functions in the eyes.^31,32 However, previous research has shown that blood vessels in normal human tissue do not express APN and that inhibitors of APN can slow retinal neovascularization in animal models.^37,38 This suggests that an increase in APN levels may be associated with the development of neovascularization in DR. Further research is needed to fully understand the mechanisms behind this association. 

Rapid advances in OCT technology may allow researchers to investigate correlations between retinal subcellular pathophysiology in vivo and clinical and laboratory findings, thereby addressing unanswered pathophysiological questions in the field. Among many OCT parameters, retinal and choroidal thicknesses have been extensively studied for the diagnosis, treatment, and prognosis of various intraocular diseases. Thus finding the correlation between APN and these two OCT parameters would effectively explain APN's role in retinopathy. Gungel et al.³² showed a positive relationship between serum APN and central macular thickness using OCT. In their study, the mean concentration of serum APN was higher in the subgroup (choroidal thickness ≥ 300 µm) than in the subgroup (choroidal thickness < 300 µm). Omae et al. also reported that high serum APN concentration was associated with increased retinal blood flow in patients with early-stage DR.³⁹ A recent study also reported a decrease in serum APN and a relatively thin choroid in obese patients.⁴⁰ Another study revealed that choroidal thickness was positively correlated with a reduction in BMI after bariatric surgery.⁴¹ In addition, Oner et al. discovered that obese patients had reduced choroidal thickness and choroidal perfusion.⁴² Hence, the relationship between serum APN and choroidal thickness can be indirectly inferred. To support this, in this study, CFT and SCT were simultaneously measured and compared with AH APN concentrations in univariate (P < 0.001 and P = 0.001, respectively) and multivariate analyses (P = 0.002 and P = 0.041, respectively). CFT and SCT were found to have a positive correlation with AH APN. However, serum APN was correlated with SCT (P = 0.041 in univariate analysis and P = 0.048 in multivariate analysis) but not with CFT (P = 0.145 in univariate analysis and P = 0.309 in multivariate analysis). In univariate analysis, AH VEGF concentration showed a significant positive correlation with CFT and SCT (P < 0.001 and P = 0.002, respectively). However, in multivariate analysis, AH VEGF correlated only with the subfoveal retinal thickness and not with SCT (P = 0.001 and P = 0.076). In univariate and multivariate analyses, serum VEGF concentration was not correlated with the two OCT parameters (all Ps ≥ 0.337). Therefore serum and AH APN correlated more with the two OCT parameters than serum and AH VEGF. 

DR severity has been positively related to retinal and choroidal thickness.^10,11,43 VEGF is a growth factor responsible for this relationship.^10,44 In addition, anti-VEGF treatment reduced choroidal thickness.^44–46 In our previous study, we discovered a link between DR severity and AH VEGF, AH APN, and serum APN.⁵ This study also showed a similar relationship between DR severity and OCT parameters and between DR severity and AH VEGF and/or AH APN. Interestingly, the relationship between DR severity, CFT, and SCT was stronger in groups 1 to 4 without PRP (R = 0.465; P < 0.001 and R = 0.271; P < 0.027) than in groups 1–5 with PRP (R = 0.275; P < 0.001 and R = 0.167; P = 0.027). Therefore, CFT and SCT showed an increasing tendency before PRP and a decreasing tendency after PRP (Figs. 3A, 3B vs. Figs. 3C, 3D). In addition, DR severity showed a significant correlation with serum APN but not with serum VEGF in both groups with and without PRP (Tables 2 and 3, Fig. 3). The relationship between DR severity and AH APN or serum APN was stronger in groups 1–4 without PRP (Table 5, R = 0.598, P < 0.001; and R = 0.293, P < 0.001) than in groups 1–5 with PRP (Table 2; R = 0.581, P < 0.001; and R = 0.247, P = 0.001). Serum APN had a weak relationship with SCT (Table 2; R = 0.154; P = 0.041) and CFT (Table 5; R = 0.191; P = 0.019). However, Figure 3 shows that serum VEGF did not correlate with any OCT parameters or DR severity following previous studies (all Ps ≥ 0.337). Figure 4 also shows a better function of AH and serum APN concentrations than the diagnostic power of corresponding VEGF in AH and serum as a biomarker of DR development. It is unknown why systemic APN levels correlate with DR activity or the two OCT parameters but not with systemic VEGF levels. This may be due to the differences in functional molecule size, charges when passing through the blood-retinal barrier, and concentrations of molecules in the bloodstream. In fact, VEGF-A, the most abundant and biologically active form, is 36 to 46 kDa, and the major form of APN is 28 kDa.^47–49 However, further studies are needed that provide more clues about this phenomenon. 

Despite the prospective nature of this study based on a relatively large patient sample, there are limitations worth noting. First, this study is limited by the fact that it only included patients with diabetes and scheduled for cataract surgery, and did not include patients with more severe forms of diabetic retinopathy (PDR) or diabetic macular edema. This may limit the ability to apply the findings to a wider population and may have affected the variability of the results comparing with the previous articles. Second, the APN subtype was not considered in evaluating the results; hence, this might have affected the heterogeneity of our results. This might have affected the heterogeneity of our results. However, the exact role of APN subtypes has rarely been evaluated; therefore there is no gold standard for adequate measurement of APN isoforms.⁵⁰ Follow-up studies should be conducted with precisely controlled disease models for the grade of DR severity and delicate analytical tools for evaluating the function of various APN isotypes. 

In conclusion, our study suggests that increasing concentration of AH APN results in the thickening of CFT and SCT. Furthermore, APN concentrations in AH and serum significantly correlates with DR progression. In addition, serum APN concentration shows a weak correlation with CFT in the non-PRP group and SCT in all groups. These findings suggest that APN concentration may be related to the structural changes and disease progression in DR. Future studies should validate the role of AH and serum APN concentrations and their potential therapeutic strategy for DR progression by controlling their concentrations. 

Acknowledgments

Supported by a VHS Medical Center Research Grant, Republic of Korea (grant VHSMC20029). The sponsor or funding organization had no role in the design or conduct of this research. 

Disclosure: H.S. Yang, None; Y.J. Choi, None; H.Y. Han, None; H.S. Kim, None; S.H. Park, None; J. Kim, None; S. Choi, None