Blood Glucose Levels Moderate the Associations Between IGF-1 Levels and Choroidal Metrics in Patients With Diabetes With Acromegaly Without Diabetic Retinopathy

Abstract

Purpose: To examine the effects of serum growth hormone (GH) and insulin-like growth factor-1 (IGF-1) on choroidal structures with different blood glucose levels in patients with diabetes mellitus (DM) with acromegaly without diabetic retinopathy.

Methods: Eighty-eight eyes of 44 patients with acromegaly were divided into a nondiabetic group (23 patients, 46 eyes) and a diabetic group (21 patients, 42 eyes). Forty-four age- and sex-matched healthy controls and 21 patients with type 2 DM without diabetic retinopathy were also included. Linear regression models with a simple slope analysis were used to identify the correlation and interaction between endocrine parameters and choroidal thickness (ChT), total choroidal area (TCA), luminal area (LA), stromal area (SA), and choroidal vascular index (CVI).

Results: Our study revealed significant increases in the ChT, LA, SA, and TCA in patients with acromegaly compared with healthy controls, with no difference in the CVI. Comparatively, patients with DM with acromegaly had greater ChT than matched patients with type 2 DM, with no significant differences in other choroidal parameters. The enhancement of SA, LA and TCA caused by an acromegalic status disappeared in patients with diabetic status, whereas ChT and CVI were not affected by the interaction. In the diabetic acromegaly, higher IGF-1 (P = 0.006) and GH levels (P = 0.049), longer DM duration (P = 0.007), lower blood glucose (P = 0.001), and the interaction between GH and blood glucose were associated independently with thicker ChT. Higher GH levels (P = 0.016, 0.004 and 0.007), longer DM duration (P = 0.022, 0.013 and 0.013), lower blood glucose (P = 0.034, 0.011 and 0.01), and the interaction of IGF-1 and blood glucose were associated independently with larger SA, LA, and TCA. As blood glucose levels increased, the positive correlation between serum GH level and ChT diminished, and became insignificant when blood glucose was more than 7.35 mM/L. The associations between serum IGF-1 levels and LA, SA, and TCA became increasingly negative, with LA, becoming significantly and negatively associated to the GH levels only when blood glucose levels were more than 8.59 mM/L.

Conclusions: Acromegaly-related choroidal enhancements diminish in the presence of DM. In diabetic acromegaly, blood glucose levels are linked negatively with changes in choroidal metrics and their association with GH and IGF-1.

Translational Relevance: We revealed the potential beneficial impacts of IGF-1 and GH on structural measures of the choroid in patients with DM at relatively well-controlled blood glucose level, which could provide a potential treatment target for diabetic retinopathy.

Introduction

Acromegaly, a chronic disease primarily caused by pituitary adenoma, is characterized by increased serum levels of growth hormone (GH) and insulin-like growth factor-1 (IGF-1),¹ which leads to numerous systemic complications. As the most common metabolic comorbidity of acromegaly, diabetes mellitus (DM) has been found in 15.8% to 37.6% of patients with acromegaly.^2–5 Acromegaly-related DM is distinct from typical type 2 DM (T2DM), in which GH levels are actually decreased owing to increased somatostatin secretion.⁶ In acromegaly, GH impacts both insulin secretion and action. Although it stimulates insulin secretion, the predominant effect of inducing insulin resistance elevates glucose levels and subsequently increases the prevalence of prediabetes and DM.^7,8 

Previous clinical observations have revealed the potential role of both circulating and retinal system-based IGF-1 in the pathogenesis of diabetic retinopathy (DR). For circulating IGF-1 specifically, a transient increase in total serum IGF-1 in patients with DM in the initial stages of retinal neovascularization has been recorded in early studies.^9,10 A pregnancy-induced increase in serum IGF-1 was associated with the progression of DR in women with type 1 DM.¹¹ DR was decreased significantly in patients after hypophysectomy.¹² As a natural disease model affected by excessive circulating GH and IGF-1, acromegaly serves as an ideal opportunity to explore the impact of those hormones on retinal and choroidal, structures as well as their interactions with blood glucose. 

Choroidal impairment has long been determined to be an essential factor in the development and progression of DR.^13,14 Changes in the microvascular structures can be detected as early as before the emergence of any clinically visible retinal signs, such as microaneurysms.^15,16 Choroidal thickness (ChT) can be either increased or decreased in patients with DM without retinopathy, although there is a general tendency to decrease.¹⁷ 

Additionally, since they are active metabolically and vascular rich, choroidal structures are more susceptible to circulating macromolecules than the retina because they are not protected by the blood-retinal barrier. Our team first confirmed that ChT can be increased in treatment-naïve patients without DM with acromegaly. Curiously, the degree of thickening is positively related to the serum IGF-1 level, disease duration, and IGF-1 burden,¹⁸ which was then confirmed by later studies.^19,20 None of these studies, however, analyzed glucose metabolism abnormalities as a confounding factor. Therefore, we designed this cross-sectional clinical observational study to explore the structural parameters of the choroid in patients with acromegaly with and without DM compared with healthy controls and patients with T2DM. Our hypothesis is that IGF-1 exhibits varying associations with choroidal structures depending on the presence of DM and may demonstrate distinct correlations with these structures across different glucose levels. Moreover, unlike acromegaly, the impact of DM on choroidal parameters, such as ChT, the choroidal vascular index (CVI), and DR progression have been shown to be more related to disease status and duration than to serum biomarker levels at a single time point.^21,22 We, therefore, evaluated the interaction and moderating effects of DM and acromegaly as categorical variables as well as serum biomarkers as continuous variables in different statistical models. 

Methods

Subjects

This cross-sectional, case-control study included 44 patients with acromegaly (88 eyes) initially diagnosed and treated at Peking Union Medical College Hospital from August 2018 to May 2023 and 44 sex- and age-matched normal controls (88 eyes). The healthy controls were recruited from the hospital staff who underwent annual physical examinations and had normal GH and IGF-1 levels. Patients with acromegaly were divided according to the status of glucose metabolism into the following subgroups: (1) 23 patients (46 eyes; nondiabetic acromegaly [NDA] group) with no DM and (2) 21 patients (42 eyes; diabetic acromegaly [DA] group) with DM and no DR. In addition, 21 patients with T2DM without DR (42 eyes; T2DM group) who were sex and age matched with the DA group were also recruited. 

The inclusion criteria for patients with acromegaly were the endocrine diagnostic criteria for the disease,²³ namely, a fasting GH level of 2.5 ng/mL or greater, a lack of suppression of GH to less than 1 ng/mL after oral administration of 75 g glucose, and a high level of serum IGF-1 controlled for age and sex; (2) enhanced magnetic resonance imaging showing pituitary adenoma; (3) typical acromegaly clinical symptoms; and (4) pituitary GH adenoma confirmed by postoperative pathology. For patients with acromegaly and controls, DM, impaired fasting glycaemia and impaired glucose tolerance were diagnosed according to the American Diabetes Association criteria²⁴ by endocrinologists. The exclusion criteria included (1) acromegaly treated with surgery, radiotherapy, or somatostatin treatment before the first consultation; (2) prior treatment other than exercise, diet, and oral medication before ophthalmologic evaluation; (3) first record of elevated serum glucose before the onset of acromegaly symptoms; (4) fundus photography (see the Methods section for details) showing any signs that may be diagnosed as DM retinopathy according to the International Clinical Diabetic Retinopathy Disease Severity Scale²⁵; (5) severe lens opacities (best-corrected visual acuity [BCVA] of <0.5 logarithm of the minimum angle of resolution); other diseases known to cause choroidal structure changes, such as hypertension, eye trauma, glaucoma, nanophthalmos, choroidal, and retinal diseases, and optic nerve diseases; or those who underwent ophthalmologic surgeries; (6) refraction error (spherical equivalent refraction) exceeding ±3.0 diopters; (7) visual field test showing a visual field defect caused by chiasmal compression; (8) exogenous glucocorticoid treatment in the previous year; and (9) imaging examination of the posterior segment that did not meet the analysis requirements after evaluation by two researchers. 

Noticeably, DM disease duration was acquired from medical records. Accordingly, only patients with a DM duration shorter than the acromegaly disease duration were considered to have secondary DM and were thus enrolled in this study. Patients with impaired fasting glycaemia and impaired glucose tolerance were included in the NDA group, with details described in the Results. Patients with acromegaly and patients with T2DM were recruited prospectively from the neurosurgery department and endocrine department, respectively. For patients with DM with or without acromegaly, only patients with diet or oral medications were included. Data on oral medications were reviewed and extracted from the patients’ past medical records. All procedures performed in this study involving human participants were conducted in accordance with the ethical standards of Peking Union Medical College Hospital Ethical Committee and with the tents of the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Ethical approval was obtained from Peking Union Medical College Hospital Ethical Committee with reference number S-K673. Informed consent was obtained from all individual participants included in the study. 

Baseline Examination

All enrolled patients underwent a detailed ophthalmologic examination within 7 days before surgery. BCVA, refractive error and anterior and fundus examinations were performed. Intraocular pressure was measured between 3:00 PM and 4:00 PM using a noncontact tonometer (CT-800 computerized tonometer, Topcon Corporation, Tokyo, Japan) and reported as the average of three measurements. All visual field tests were conducted with the Octopus 101 G2 program (Interzeag, Schlieren, Switzerland) after correction of refractive errors for at least 2 reliable results. Axial length was measured by an IOL Master 700 (Carl Zeiss Meditec AG, Jena, Germany). Fundus photographs were obtained by a skilled technician (WS) 30 minutes after pupil dilation (tropicamide 0.5%). DR stage was determined by double reading according to the Diabetic Retinopathy Disease Severity Scale²⁵ on confocal color fundus photographs covering at least the seven Early Treatment Diabetic Retinopathy Study fields (Visucam, Carl Zeiss Meditec AG). In cases of disagreement between the two readers (JY and XZ), the photographs were evaluated by a third retinal specialist (XYZ). 

Endocrine Parameters

The term disease duration of acromegaly was defined as the time from the first emergence of clinical symptoms (according to the medical history acquired from the patients) to surgical or medical treatment. DM disease duration was acquired from the medical records. For patients with acromegaly, serum GH, serum IGF-1, fasting blood glucose, and hemoglobin A1c (HbA1c) were measured routinely inpatient before surgery at 7:00 AM on at least two consecutive days after fasting. The result obtained at the measurement time closest to the ophthalmology test was used in the correlation analysis. 

For patients with T2DM, fasting blood glucose and HbA1c were tested within one week at 7:00 AM after fasting in the endocrinology clinic at Peking Union Medical College Hospital before ophthalmology examination and were used for the correlation analysis. 

All patients with acromegaly underwent 7:00 AM fasting anterior pituitary hormone examinations, including a 75-g oral glucose tolerance test and fasting blood glucose test, 2-hour postprandial blood glucose test, and serum free cortisol, adrenocorticotropic hormone, thyroid-stimulating hormone, T3, T4, free T3, free T4, prolactin, follicle-stimulating hormone, luteinizing hormone, testosterone, and estradiol tests. The lowest GH level based on the pretreatment 75-g oral glucose tolerance test was defined as the nadir GH level. 

Enhanced Depth Imaging Optical Coherence Tomography (EDI-OCT)

All subjects underwent OCT imaging (Spectralis, Heidelberg Engineering GmbH, Heidelberg, Germany) scanning from 2:00 PM to 4:00 PM to avoid the possible impact of diurnal fluctuation on the results within 7 days before surgery. All OCT scans were conducted in EDI mode: 30° lens, high-resolution scan mode (1536 × 496 A-scans), scan line length 9.5 mm, and single-line scanning in the horizontal direction through the fovea. Each single-line scan ensures that the image ART reaches 100; thus, the final image is composed of 100 two-dimensional images. All scans were completed by the same experienced specialist (XZ). According to the OSCAR-IB quality control standard,²⁶ OCT images with signal strength of 15 or greater obtained with a correctly centered scan can be used for analysis; otherwise, the scanning is repeated until a satisfactory scanning image is obtained. 

ChT was measured under the fovea manually by two specialists (XZ and HW) who were masked to the clinical patient data and each other's observations. Total ChT was defined as the vertical perpendicular distance from the hyper-reflective line of the choroid–scleral interface (Fig. 1A). Each specialist measured ChT three times, and the average was used for statistical analysis. 

Figure 1.

 

The ChT, SA, LA, TCA, and CVI were measured. (A) EDI-OCT scans showing the ChT measurement in a horizontal scan. Total ChT was defined as the vertical perpendicular distance from the hyperreflective line of the choroid-scleral interface. (B–D) Choroidal image binarization. (B) Original OCT B-scan in the horizontal meridian with LabelMe (http://github.com/wkentaro/labelme) used to select an area of 1500 µm on the nasal side and 1500 µm on the temporal side with the fovea as the midpoint. The distance of the high reflective band outside the retinal pigment epithelium layer was marked as the upper boundary, and the choroid-sclera junction was marked as the lower boundary. The selected area was added to the region of interest (ROI) manager. (C) Binarized image with outline of choroidal area using Niblack's autolocal threshold algorithm. (D) The subfoveal ROI choroidal area was selected. (E) Overlay of the binarized choroidal area on the original image and all computational processes implemented with MATLAB (R2018a).

Image Binarization and Calculation of the Choroid Index

Referring to the measurement method of Agrawal et al.,²⁷ first, the polygon selection tool LabelMe (http://github.com/wkentaro/labelme) was used to select an area of 1500 µm on the nasal side and 1500 µm on the temporal side with the fovea as the midpoint. The distance of the high reflective band outside the retinal pigment epithelium layer was marked as the upper boundary, and the choroid-sclera junction was marked as the lower boundary. The selected area was added to the region of interest manager, and the total choroidal area (TCA) under the macular fovea was calculated (Fig. 1B). Then, the image was converted to 8 bits and binarized using Niblack's auto local threshold algorithm.²⁸ In the binarized image, the bright pixel area represented the stromal area (SA), and the dark pixel area represented the vascular luminal area (LA) (Fig. 1C). The image conversion tool was used to convert the image to RGB format, the color threshold tool was used to set the upper limit value of brightness to 0, the lower limit value was adjusted to less than 255, and the threshold value was set to black to select dark pixels and add them to the region of interest manager. In the region of interest manager, the AND intersection of the two added areas was used to calculate the LA (Fig. 1D). The SA was calculated by subtracting the LA from the TCA, and the CVI was calculated by dividing the choroidal vascular LA by the TCA. 

The image processing and calculation were independently performed by two doctors (XZ and HW) under the supervision of a software engineer (KZ). Both doctors were masked to the patient information. All computational processes were implemented with MATLAB (R2018a). 

To determine the reliability of the choroidal measurements, the first reader repeated this procedure for all eyes once in a masked fashion at a 1-month interval. The result from the first reading was used for the correlation analysis. The interobserver and intraobserver reliabilities of the data were calculated separately (Supplementary Table S1). 

Statistics

Data that conformed to a normal distribution are described as the mean ± standard deviation, while data that did not conform to a normal distribution are described as the median (Q1, Q3). For comparison across cohorts, the Student t tests and χ² tests were used for the baseline demographic characteristics, whereas the Student t test (for normally distributed variables) and the independent sample Mann-Whitney U test (for nonnormally distributed variables) were applied for choroidal parameters. 

As described in the Introduction, the impact of DM on choroidal parameters has been shown to be more related to disease status than to serum biomarkers at a single time point. First, the overall interaction of serum IGF-1/GH and blood glucose/HbA1c was evaluated as the acromegalic status and diabetic status, respectively, in all individuals in this study by two-way analysis of variance. Then, a subgroup linear regression was conducted with diabetic status as a categorical variable in patients with acromegaly. 

Both eyes were included in the analysis to comprehensively assess the systemic impact of acromegaly. We performed a generalized estimating equation (GEE) analysis accounting for intereye correlation, with ChT, CVI, LA, SA, and TCA as dependent variables and GH levels, IGF-1 levels, blood glucose, HbA1c, and their interactions as predictors, adjusting for age, sex, disease duration, DM disease duration, treatment, and axial length. The serum IGF-1, serum GH, fasting blood glucose, and HbA1c levels were centered around the mean. Simple slope analysis was used to further interpret significant interaction terms. Then, the Johnson–Neyman technique was used to identify the interval of the moderator where the slope of the predictor changed from nonsignificant to significant with R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria). All other statistical analyses were performed using SPSS software version 26.0 (IBM, Chicago, IL). 

For ChT, SA, LA, CVI and TCA, interobserver reliability and intraobserver reliability are expressed as intragroup correlation coefficients. A P value of less than 0.05 (bilateral) was considered statistically significant. 

Results

Demographic Characteristics

There was no significant difference in age, refractive error, BCVA, intraocular pressure, or axial length between the acromegaly group (44 individuals, 88 eyes) and the normal control group (44 individuals, 88 eyes) or between the NDA (23 individuals, 46 eyes) and DA (21 individuals, 42 eyes) subgroups. The demographic and clinical characteristics are shown in Table 1. All the pituitary thyroid, pituitary–adrenal, and pituitary–gonadal axes of the enrolled patients were within the normal range. 

Table 1.

 

Demographic Characteristics

 

Differences Between Patients With Acromegaly and Healthy Controls

As shown in Table 1, the ChT, LA, SA, and TCA (all P < 0.001) increased significantly in patients with acromegaly compared with healthy controls. However, the CVI did not differ between the two groups (P = 0.338). For patients with DM with acromegaly (DA group), only ChT showed a significant increase compared with that of healthy controls (P < 0.001). 

Differences Between Patients Without DM With Acromegaly and Patients With DM With Acromegaly

No significant differences were detected in acromegaly disease duration, serum GH levels or serum IGF-1 levels (all P > 0.05) (Table 1) between the DA and NDA groups. The average fasting serum glucose and HbA1c levels were significantly higher in the DA group than in the NDA group (P < 0.001) (Table 1). No significant differences were detected in the ChT (P = 0.846) or CVI (P = 0.216) between the NDA and DA groups. LA, SA, and TCA (all P < 0.001) (Table 1) were significantly lower in the DA group than in the NDA group. 

Differences Between Patients With DM With Acromegaly and Patients With T2DM

Twenty-one sex- and age-matched patients with T2DM (42 eyes) were enrolled additionally for comparison with 21 patients with DM with acromegaly (DA group; 42 eyes). Twelve patients in the DA group and 17 patients in the T2DM group were on oral medications (specific types are shown in Table 2), whereas no patient was under insulin management in either group. The ChT was significantly greater in the DA group than in the T2DM group (P = 0.036) (Table 2). No significant differences were detected in LA, SA, CVI, or TCA between the DA group and the T2DM group (all P > 0.05) (Table 2). 

Table 2.

 

Clinical Characteristics and Choroidal Structure of Patients With DM With Acromegaly and Patients With T2DM

 

Interaction Effect of Diabetic Status and Acromegalic Status on Choroidal Parameters

Two-way analysis of variance was conducted for the four groups (HC, NDA, DA, and T2DM). The NDA and DA groups were labelled acromegalic/nondiabetic and acromegalic/diabetic, respectively, whereas the HC and T2DM groups were labelled nonacromegalic/nondiabetic and nonacromegalic/diabetic, respectively. As shown in Figure 2, LA (P < 0.001; η² = 0.069), SA (P = 0.001; η² = 0.055), and TCA (P < 0.001; η² = 0.064) were affected significantly by the interaction of acromegalic status and diabetic status, whereas ChT (P = 0.779; η² = 0.000) and CVI (P = 0.869; η² = 0.004) were not affected significantly. The simple effect analysis is shown in the Supplementary Results. In other words, the enhancement of SA, LA and TCA by acromegalic status disappeared in the presence of diabetic status. 

Figure 2.

 

Interaction between acromegalic status and diabetic status. The NDA and DA groups were labelled acromegalic/nondiabetic and acromegalic/diabetic, whereas healthy controls and patients with T2DM were labelled nonacromegalic/nondiabetic and nonacromegalic/diabetic. (A and D) CT and CVI were not significantly affected by the interaction of acromegalic and diabetic labels. (B, C, and E) The interaction was significant in LA, SA, and TCA with similar patterns. The enhancement of SA, LA and CVI owing to acromegalic status decreased significantly when diabetic status was also present.

Correlations Between Serum IGF-1, GH, Blood Glucose, and HbA1c Levels and Choroidal Parameters in Patients With DM With Acromegaly

In the DA group, regression models with GEE analysis revealed that higher IGF-1 levels (P = 0.049) and GH levels (P = 0.006), longer DM disease duration (P = 0.007), and lower blood glucose (P = 0.001) (Supplementary Table S2) were independently associated with greater ChT. Higher GH levels (P = 0.016, 0.004, and 0.007, respectively) and longer DM disease durations (P = 0.022, 0.013, and 0.013, respectively) were associated independently with greater SA, LA, and TCA. Blood glucose was negatively associated with SA, LA, and TCA (P = 0.034, 0.011, and 0.019, respectively). In contrast, only a longer duration of acromegaly (P = 0.034) (Supplementary Table S2) was associated independently with a lower CVI. 

The interaction of serum GH level and blood glucose level was associated significantly with ChT (P < 0.001), whereas the interaction of IGF-1 level and blood glucose level was associated significantly with SA (P = 0.009), LA (P = 0.002), and TCA (P = 0.005) (Supplementary Table S2). 

Interaction and Moderating Effects of Serum GH/IGF-1 and Blood Glucose on Choroidal Parameters in Patients With DM With Acromegaly

To interpret further the effect of the interaction between GH and blood glucose in patients with DM with acromegaly, we conducted a simple slope analysis stratified by blood glucose within 0.5 standard deviations on ChT. The associations between serum GH levels and ChT became less positive as the blood glucose level increased (Fig. 3A). We used the Johnson–Neyman technique to identify the interval of significant associations. Only the data from right eyes were included because GEE cannot be applied to the Johnson–Neyman technique. As shown in Figure 4A, GH was positively associated with ChT in patients with DM with acromegaly only when serum blood glucose levels were less than 7.35 ng/mL. When the serum blood glucose level increased, the positive response of ChT to the serum GH level decreased. When the blood glucose level was greater than 7.35 mM/L, the correlation was no longer significant. We further conducted the same analysis of the left eye data and found a similar result (significant interval, 0–6.83 mM/L) (Supplementary Fig. S1A). 

Figure 3.

 

Simple slope study analyses demonstrating the effect of the interaction between blood glucose and serum GH or IGF-1 levels on choroidal parameters in patients with diabetes and acromegaly. (A) Adjusted prediction of ChT against GH. (B) Adjusted prediction of LA against IGF-1. (C) Adjusted prediction of SA against IGF-1. (D) Adjusted prediction of TCA against IGF-1; scatterplots were divided by blood glucose of more than 0.5 standard deviations (SD) (green), blood glucose from −0.5 SD to +0.5 SD (blue), and blood glucose of less than −0.5 SD (orange).

Figure 4.

 

The Johnson–Neyman N) technique demonstrating the interval of blood glucose where significant associations between serum GH/IGF-1 levels and choroidal parameters were detected. (A) Predicted association between the slope of GH on ChT and blood glucose. When the serum blood glucose levels were less than 7.35 ng/mL, GH was significantly positively associated with ChT in patients with DM with acromegaly. When the serum blood glucose level increased, the positive response of the ChT to the serum GH level decreased. (B) Predicted association between the slope of IGF-1 on LA and blood glucose. When the serum blood glucose levels were higher than 8.59 mM/L, serum IGF-1 levels were associated significantly and negatively with LA in patients with DM with acromegaly. When the blood glucose level was elevated further, the negative response of the LA to serum IGF-1 levels also increased. The green area is the interval of blood glucose where the correlation between IGF-1/GH and choroid parameters was significant (P < 0.05). The red area is the interval of blood glucose where the correlation between IGF-1/GH and choroid parameters was not significant (P ≥ 0.05). The bold line is the range of blood glucose in patients with acromegaly in this study. GLU, glucose; n.s., not significant.

We also explored the interaction between IGF-1 and blood glucose in patients with DM with acromegaly and conducted simple slope analysis stratified by blood glucose within 0.5 standard deviations on LA, SA, and TCA. The findings suggested that the associations of IGF-1 with SA, LA, and TCA became increasingly negative as the blood glucose level increased (Figs. 3B–D). With only right eye data included in the Johnson–Neyman technique, IGF-1 did not show any associations with LA at lower blood glucose levels. The serum IGF-1 level was associated significantly and negatively with LA in patients with DM with acromegaly only when serum blood glucose levels were higher than 8.59 mM/L (Fig. 4B). When the blood glucose level was increased further, the negative response of the LA to serum IGF-1 level also increased. We also conducted the same analysis on the left eye data but failed to identify the interval, and the result is shown in Supplementary Figure S1B. No significant intervals could be identified for SA or TCA. 

Correlations of Serum GH, IGF-1, Blood Glucose, and HbA1c Levels With Choroidal Parameters in the NDA Group

In the NDA group, GEE analysis with linear regression models showed that higher serum IGF-1 levels and lower HbA1c were associated independently with thicker ChT. Higher serum GH levels were independently associated with larger SA, LA, and TCA. CVI, in contrast, did not show any association with GH, IGF-1, blood glucose, or HbA1c. Details are shown in Supplementary Table S2. No interaction of GH/IGF-1 and glucose/HbA1c was detected for any choroidal parameters (all P > 0.05) (Supplementary Table S2). 

In the cohort comprising all patients with acromegaly, linear regression analysis was performed. The outcomes indicated a positive correlation between GH and IGF-1 levels with ChT. Conversely, the duration of acromegaly exhibited a negative association with CVI. These findings are detailed in Supplementary Table S3. 

Discussion

In a previous study, we demonstrated that choroid thickness was significantly thicker in patients without DM with acromegaly and was associated significantly and positively with serum IGF-1 levels.¹⁸ In this study, we further validated that LA, SA, and TCA were also enhanced in both patients with DM with acromegaly and patients without DM with acromegaly. We also demonstrated that serum IGF-1 levels were positively associated with ChT and that serum GH levels were positively associated with LA, SA, and TCA in patients with DM with acromegaly. These results suggest a potential role for GH and IGF-1 in choroidal remodeling. 

Blood glucose, in contrast, was suggested to be associated negatively with ChT, LA, SA, and TCA in patients with DM with acromegaly in our study. The ChT of diabetic eyes without retinopathy can vary notably; a few studies reported a reduced central ChT,²⁹ whereas others found the decrease insubstantial.^14,30 Furthermore, LA was detected to be decreased significantly in patients with DM but without DR,²¹ suggesting that choroidal hypoperfusion precedes retinopathy. Our results suggested that an increase in blood glucose did have a negative effect despite the high GH and IGF-1 levels on both luminal parameters and stromal parameters of the choroid in patients with DM with acromegaly. 

However, our study showed that the difference in CVI remained nonsignificant between the acromegaly groups and healthy controls and did not show any correlation with serum GH or IGF-1 levels, blood glucose, or HbA1c. CVI, the ratio of vascular area to TCA, reflects the inner structural modifications of the choroid beyond the volume and has been determined to be a sensitive biomarker in detecting DR and monitoring DR severity.³⁰ Mixed results have been reported regarding the correlation of blood glucose and HbA1c with the CVI in the eyes of patients with DM.^31,32 The results of our study suggest that, in patients with diabetes and patients without DM with acromegaly, although the lumina area increased, the proportion of vascular structure to stroma was not affected significantly by serum IGF-1, GH, blood glucose, or HbA1c. 

Our study further explored the impact of the interaction of serum IGF-1 level, serum GH level, blood glucose, and HbA1c on choroidal parameters of patients with acromegaly. Among all individuals, the significant enhancement of acromegaly status on the LA, SA, and TCA ceased when DM status was also present. Moreover, in patients with DM with acromegaly, the associations of IGF-1 and SA, LA, and TCA became increasingly negative as the blood glucose level increased. The association of GH and ChT became less positive as blood glucose increased. To our knowledge, this study is the first to attempt to interpret the interaction between IGF-1, GH, and blood glucose in patients with DM with acromegaly. 

IGF-1, GH, and blood glucose share a complicated and fascinating relationship in numerous disease models. In disease models of DR, the roles of IGF-1 and GH have always been unclear. Some studies suggest that they promote the progression of DR, whereas others suggest that they have a protective effect. IGF-1 was found to increase in the vitreous of patients with proliferative DR.^33–35 Octreotide, a somatostatin analogue, has been shown to inhibit the progression of DR,³⁵ but failed to be confirmed in a randomized, double-blind, placebo-controlled trial with patients with moderate-to-severe nonproliferative DR to low-risk proliferative DR.^36,37 The potential protective effect of circulating IGF-1 has also been described. The severity of DR was related inversely to serum IGF-1 levels in patients with type 1 DM.³⁸ A positive correlation was detected between the IGF-1 standard deviation score and retinal vessel density in children with type 1 DM.³⁹ 

Our results provide a possible explanation for these conflicting results. Although excessive IGF-1 and GH were associated positively with the structural components of the choroid, including LAs and SAs, in patients without DM with acromegaly and patients with DM with acromegaly, the impact was moderated negatively by blood glucose. In our study, this positive association was completely overwhelmed with the presence of diabetic status. For IGF-1 specifically, with blood glucose above the normal range (our interval, 0–8.59 mM/L), the correlation between serum IGF-1 level and LA, SA, and TCA changed from insignificant to significantly negative, suggesting that with high blood glucose, IGF-1 may act as a risk factor for choroid hypoperfusion. 

These results can in turn explain the DR rate in patients with DM with acromegaly. A high incidence of proliferative retinopathy was found in acromegaly cohorts, but the incidence of nonproliferative retinopathy was the same as that in patients with T2DM or patients with impaired fasting glycaemia.^40–42 The possible higher blood glucose levels in the advanced stage of DR may lead to a more significant decrease in choroid lumina and SAs when interacting with high serum IGF-1 levels and result in a higher incidence of proliferative retinopathy. Additionally, there might be a possible influence of inner blood-retinal barrier breakdown, which possibly leads to alterations in the permeability to small peptides, such as GH and IGF-1, that would contribute to DR progression. 

The choroid is a rich vascular system and houses abundant different cell types, including fibroblasts, melanocytes, contractile pericytes/smooth muscle cells, and infiltrating immune cells. High expression of IGF-2 was detected in cell clusters with fibroblasts, and high expression of insulin-like growth factor binding protein 4 (IGFBP4) was detected in choroidal endothelial cells by single-cell transcriptomics.⁴³ IGF-I and IGF-IR messenger RNA expression was found in choriocapillary endothelial cells in both normal individuals and patients with diabetes.⁴⁴ Excessive serum IGF-1 may modulate the function and circulation level of vascular endothelial growth factor to affect choroidal perfusion, since elevated levels of IGF-1 may contribute to elevated vascular endothelial growth factor levels⁴⁵ or enhanced local vascular endothelial growth factor expression.^46,47 Second, the proliferation of fibroblasts in the choroid might be enhanced by IGF-1, thus increasing the thickness of the stroma.⁴⁸ The impact of blood glucose on the choroid was more complicated and unclear. Increased choriocapillaris permeability,⁴⁹ the overexpression of cytokines activated by inflammation and oxidative stress,⁵⁰ choriocapillaris dysfunction,⁵¹ increased vessel resistance, and unstable choroidal perfusion⁵² may all contribute to choroid remodeling. 

The expression of the IGF system in the choroid with and without DM might be different, and messenger RNA for IGFBP6 was observed at a much lower abundance in mice with diabetes than in mice without diabetes.⁵³ Additionally, in T2DM, proinflammatory cytokines could be responsible for the disruption in the insulin–IGF-1 signaling pathways in peripheral tissues and the pancreas.⁵⁴ Both retinal hypoxia and relative hypoxia have been reported to alter the gene expression of the IGF system by decreasing IGF-I but increasing IGF-IR, IGFBP2, and IGFBP3 messenger RNA.⁵⁵ The mechanism underlying the negative moderating effect of blood glucose on the effects of IGF-1 and GH on choroidal parameters should be investigated further. 

Our study has some limitations. First, the sample size is small, especially regarding patients with acromegaly with DM, which may lead to bias in the conclusions. We limited the strict inclusion criteria concerning DM and acromegaly disease duration, treatment for DM and acromegaly, and the status of DR to decrease bias, which added difficulty in cohort enrolment. We adopted GEEs to decrease bias and provide cohesive other eye analyses in the Supplementary data. However, studies with larger sample sizes should be conducted in the future. Second, this was a cross-sectional study. Longitudinal studies should be conducted to avoid the effects of individual variation and bias. Third, the OCT device we used was an EDI-OCT machine, and thus, we were unable to perform angiographic analysis of the choroidal vessels. Last, but not least, two patients from the T2DM patient group had a history of using a sodium glucose cotransporter-2 inhibitor (dapagliflozin), which may influence the SA, LA, TCA, and CVI results.⁵⁶ However, after conducting sensitivity analysis by excluding those two patients, the conclusion of our study did not change significantly (Supplementary Table S4). 

In conclusion, this study is the first to our knowledge to evaluate choroidal structures in patients with acromegaly with or without DM. The results emphasized that the increases in ChT and the luminal and SAs owing to acromegalic status was reduced significantly when DM was also present. In patients with DM with acromegaly, blood glucose not only was associated negatively with ChT, SA, LA, and TCA, but also negatively moderated the response of ChT, SA, LA, and TCA to serum GH and IGF-1. 

The management of DR remains a critical concern in ophthalmology. Our findings indicate that the association between IGF-1 and GH with choroidal structures varies with blood glucose levels in patients with DM, suggesting a potential heterogeneous response. Further research is needed to explore the mechanisms underlying these associations. Such insights could provide valuable insights for developing new strategies for the management of DR. 

Acknowledgments

The authors thank Peter H Scanlon, Nuffield Department of Clinical Neuroscience, University of Oxford, for his valuable advice and guidance in the preparation of this article. The authors also thank Rui Gao from the University of Oxford for her valuable advice regarding the endocrinology part of this article. We express our gratitude to all the patients and participants who generously gave their time and contributed to this study. 

Disclosure: X. Zhang, None; H. Wang, None; K. Zhang, None; J. Ma, None; H. He, None; S. Song, None; E. Shao, None; B. Chen, None; J. Yang, None; X. Zhao, None; W. Sui, None; M. Wang, None; S. Liu, None; X. Guo, None; H. Zhu, None; Y. Yao, None; Y. Zhong, None; B. Xing, None