Abnormal Fractional Amplitude of Low-frequency Fluctuation Changes in Patients With Diabetic Optic Neuropathy: a Functional MRI Study


 The present study was to assess the spontaneous changes in brain activity in patients with DON using functional magnetic resonance imaging (fMRI). 14 DON patients and 14 healthy controls (HCs) were enrolled. All participants underwent resting-state fMRI (rs-fMRI). The fractional amplitude of low-frequency fluctuation (fALFF) method was applied to evaluate neural activity changes. The Hospital Anxiety and Depression Scale (HADS) was used to assess the anxiety and depression status of participants. The independent sample t test and chi-squared test were applied to analyze demographics of DON patients and HCs. Receiver operating characteristic (ROC) curves were applied to analyze the variation in mean fALFF values between DON patients and HCs. The relationships between the fALFF values of brain regions and clinical behaviors in the DON group were analyzed using Pearson’s correlation analysis. In contrast to HCs, the fALFF value of DON patients was significantly higher in the right precentral gyrus (RPCG). However, the fALFF values in right anterior cingulate gyrus (RACG) and left middle cingulate gyrus (LMCG) were markedly decreased in DON patients. The area under the curve (AUC) of ROCs for each brain region showed high accuracy. Pearson's correlation analysis showed that fALFF values of the right anterior cingulate gyrus and left middle cingulate gyrus were negatively correlated with HADS scores, while fALFF values of the left middle cingulate gyrus were negatively correlated with DON disease duration. To sum up, we found abnormal spontaneous brain activities in regions related to cognitive and emotional dysfunction, eye movement disorder, and vision loss in patients with DON. These results may indicate the underlying neuropathological mechanisms of DON, and show that fALFF may be an effective method by which to distinguish patients with DON from healthy individuals.


Introduction
The incidence of diabetes has increased in recent years, making it one of the major diseases affecting public health globally. Diabetes is a metabolic disease with systemic effects. Complications often occur in tissues and organs such as the eyes, kidneys, heart, blood vessels, and nerves 1,2 . Ocular complications include diabetic retinopathy, macular edema, and cataracts. Neuropathy is a complication of diabetes that affects the central and sensory nerve endings, with manifestations such as skin numbness, diabetic nephropathy, and diabetic foot problems. When the optic nerve is involved, it is known as diabetic optic neuropathy (DON). DON is an important clinical blindness factor 3 . DON may be found in the form of optic disc neovascularization, Wolfram syndrome, diabetic papillitis, or anterior ischemic optic neuropathy 4 . Patients with DON have varying degrees of decline in visual function, which will eventually seriously affect their life quality and physical and mental health. Due to anatomical structure, the diagnosis of DON mainly relies on clinical symptoms and routine fundus examination, including visual evoked potential (VEP) and fundus uorescein angiography (FFA) 5,6 . Figure 1 shows the results of retinal fundus photography and FFA of DON patients. However, the diagnosis of DON is di cult due to the lack of speci c clinical symptoms and speci c changes in ophthalmic examination results. Compared with other ocular complications, there are few studies focused on DON, and its pathogenesis remains unclear.
Functional magnetic resonance image (fMRI) is based on traditional magnetic resonance imaging. It can make full use of anatomical, imaging, and functional factors, and provide imaging technical support for clinical magnetic resonance diagnosis from a single morphological study combining morphological and functional research. The technology is non-invasive, non-radioactive, and has the advantages of repeatability and high temporal and spatial resolution 7,8 . In the past, fMRI has been applied to various psychoneurotic diseases, providing new perspectives for further understanding of their neuropathological mechanisms 9,10 . The amplitude of low-frequency uctuation (ALFF) method is a resting state-fMRI (rs-fMRI) approach. Based on blood oxygen level dependent (BOLD) imaging, ALFF indicates the intensity of spontaneous local brain activity in the resting state. Therefore, ALFF may be a useful tool to measure brain activity and to better understand the pathophysiological changes. The fALFF analysis method involves calculation of the ratio of low frequency amplitude to whole brain frequency, and aims to remove the in uence of the cerebrospinal uid noise signal on low frequency amplitude. It is an analysis method for ALFF standardization. In the past, fALFF has been used in post-stroke depression, borderline personality disorder, anorexia nervosa, and Alzheimer's disease 11-14 but has not been applied to DON patients. Therefore, we used the fALFF technique to assess brain activity in DON patients in order to explore the neuropathological mechanism of this disease.

Patients And Methods
Subjects. In total, 14 patients with DON were recruited from the First A liated Hospital of Nanchang University (Nanchang, China). The inclusion criteria for DON were: (1) a history of diabetes; (2) evidence of optic neuropathy with papilledema, ischemic optic neuropathy, optic disc neovascularization, or optic atrophy. DON was diagnosed according to clinical manifestations, visual eld defect, VEP, and FFA; (3) no systemic diseases such as serious heart, liver, or kidney diseases. The exclusion criteria were: (1) current pregnancy; (2) patients with serious neurological and hematopoietic diseases; (3) patients with ocular or orbital diseases, such as corneal ulcers, glaucoma, or orbital tumor; (4) patients with systemic lesions that might cause optic disc edema; (5) history of psychiatric disease; and (6) contraindications for MRI scans, such as pacemakers or other metal implants.
In addition, 14 healthy people of similar age and education level were enrolled. The main inclusion criteria were: (1) no history of eye disease; (2) no liver, kidney, or heart disease; (2) no history of nervous system diseases; (3) no history of mental illness; (4) no contraindications for MRI scans.
All participants were told about the purpose, procedure, and risks of this study and provided signed informed consent. The current study was conducted in accordance with the Declaration of Helsinki and was approved by the Medical Ethics Committee of the First A liated Hospital of Nanchang University.
MRI parameters. A 3-Tesla MRI scanner was used. All participants were asked to breathe smoothly and keep their eyes closed throughout the scan. A 3D metamorphic gradient recalled-echo pulse sequence was used to obtain the functional data and 176 structural images were obtained using the parameters described previously 15 . Functional images (n = 240) were collected using the following scanning parameters: acquisition matrix, 64x64; eld of view, 220x220 mm; thickness, 4.0 mm; gap, 1.2 mm; repetition time, 2,000 msec; ip angle, 90 o ; echo time, 30 msec and 29 axial. rs-fMRI data processing. Statistical parameter maps (Statistical Parametric Mapping, SPM5; http://www. l.ion.ucl.ac. uk/spm) and the REST software (http://www.restfmri.net) were used for fMRI data processing. Initially, the rst 10 time-points were removed to balance the signal, then head motion correction was performed. Data with more than 1.5 mm maximum displacement in the x, y, or z-axes or 1.5 o of angular motion were deleted. Subsequently, functional images were normalized to meet the standards of the Montreal Neurological Institute (MNI) with a voxel size of 3x3x3mm. A linear regression process was applied to remove uncontrollable variables. Finally, the data were ltered in the frequency range of 0.01-0.08 Hz to eliminate the in uence of high-frequency physiological noise. fALFF calculation. For data processing, the REST software was used to calculate the fALFF value of the whole brain, as the ratio of ALFF value in the low-frequency range (0.01-0.08 Hz) to that of the whole frequency range (0-0.25 Hz). This ratio reduced the effect of variance between individuals.
Statistical analysis. SPSS 20.0 software (SPSS, Chicago, IL, USA) was applied for statistical analysis. An independent sample t test and chi-squared test was used to compare data between the two groups. The operating characteristic (ROC) curves were used to assess the accuracy of the fALFF value as a means of distinguishing DON patients from HCs. Pearson's correlation analysis was used to look for any association between regional fALFF values and clinical parameters of DON patients. In all analyses, p value < 0.05 was the threshold for statistical signi cance.

Results
Demographics and visual measurements. No signi cant difference was found in age or gender between DON and HC groups (P>0.05). The mean duration of diabetes was 52.14 ± 27.46 days. However, statistically signi cant differences between the groups were found in best-corrected visual acuity (VA) of each eye, latency (ms), and amplitude (uv) of the monocular VEP from each eye, and the HADS score (P < 0.05). More details are shown in Table 1.  (Fig. 3). All of the AUC values are above 0.9, indicating high diagnostic accuracy.

Discussion
As one method of rs-fMRI, fALFF has been used to investigate several ophthalmic diseases including primary open-angle glaucoma, primary blepharospasm, retinitis pigmentosa, normal-tension glaucoma, and monocular blindness 16-20 ( Table 3). The target of the present research was to assess regional brain function in DON using the fALFF method and to explore the potential pathogenesis of this disease. Our results showed vision de cits in DON patients. In contrast to HCs, we found that the fALFF value of the RPCG in DON patients was increased, while the fALFF values of RACG and LMCG were decreased (Table 4 and Fig. 5).  Composed of many pyramidal neurons, the precentral gyrus (PCG) is the cortical motor center. In the brain of anesthetized monkeys, the PCG was found to be involved in the control of defensive actions, including eye movements 21 . A previous study also showed that cortical electrical stimulation of frontal gyrus and PCG could lead to eye movement 22  indicating that PCG dysfunction may explain eye movement disorder in these patients. Shi et al. 27 found that the lower voxel-mirrored homotopic connectivity (VMHC) value in the precentral gyrus of patients with corneal ulcer than in those without suggests that it may in part explain ocular movement dysfunction in this condition. Using the voxel-based degree centrality (DC) method, Cai et al. 28 studied changes in the brain functional network of primary angle-closure glaucoma patients and found signi cantly higher DC values in PCG after than before glaucoma surgery. Voxel-based DC value in this region is also reportedly decreased in patients with comitant exotropia strabismus, and regional homogeneity (ReHo) value in PCG in patients with strabismus and amblyopia is lower than in controls 29,30 . All of these ndings indicate an association between the PCG and eye movements. In the current study, we found higher fALFF value in the PCG, indicating enhanced brain activity in PCG in patients with DON. On this basis, we speculate that increased fALFF value in PCG may re ect a compensatory mechanism for ocular movement disorder in patients with DON.
The cingulate gyrus consists of four sub regions: anterior cingulate gyrus (ACG), middle cingulate gyrus (MCG), posterior cingulate gyrus, and retrosplenial cingulate gyrus 31 . Having many nerve ber connections with other parts of the brain, the cingulate gyrus participates in the maintenance of arousal, regulating the body's emotional, memory, cognitive, and executive functions 32 . Using independent component analysis of rs-fMRI, researchers have found signi cantly lower activation of cingulate gyrus in early Parkinson's disease patients than in healthy controls 33 . Lower functional connectivity in the ACG and MCG were also observed in patients with schizophrenia 34,35 . In a study conducted by Bürger et al. 36 , whole-brain and a multivariate pattern classi cation analysis of rs-fRMI were used to study brain activity in patients with emotional disorders. They found signi cantly decreased ACG activation in patients with major depressive disorder than in those with bipolar disorder or the healthy controls, which potentially indicates impaired bottom-up emotional processing and abnormal automatic emotion regulation.
Additionally, patients with chronic tinnitus also had abnormal ALFF value in the cingulate gyrus, which was thought to be related to psychological problems including stress, anxiety, inattention, and insomnia 37 . Consistent with the above results, the present study demonstrated signi cantly decreased fALFF values in the RACG and LMCG. Additionally, fALFF values in RACG and LMCG were negatively correlated with HADS scores, and fALFF values in the LMCG were negatively correlated with disease duration in DON, indicating that changed brain activities in the RACG and LMCG are associated with emotional dysfunction in DON patients.
However, activity in the cingulate gyrus may also be related to visual function, since ACG receives afferent neurons from the thalamus. Hence, it can be deduced that cingulate gyrus is related to visual function. A previous study found that the ReHo values of the left marginal lobe/ACG and the right marginal lobe/ACG were signi cantly reduced after eye enucleation 38 , suggesting reduced visual signals after the surgery. Similarly, previous research has also found signi cantly decreased GMV value of RACG in monocular blindness, indicating a disturbance of synchronous nerve activity in these patients 39 . A decrease in fALFF of ACG in monocular blindness further con rmed this conclusion 20 . Zhai et al. 40 analyzed the role of perceptual learning in the treatment of amblyopia using fMRI. They reported that activity of the cingulate gyrus and visual cortex were signi cantly increased in patients after treatment, suggesting de cit of the cingulate cortex as a cause of amblyopia. In the current study, we found lower fALFF value in RACG in DON patients, which suggests that the abnormality of RACG may be part of the neuropathological mechanism of vision loss in DON patients.
ROC curve analysis is often utilized to distinguish diseased from healthy states. In the present analysis, the AUC of fALFF values of the RPCG, RACG, and LMCG were 0.980, 0.990, and 0.939 respectively. All AUC of fALFF values was above 0.9, indicating signi cant differences in fALFF values between DON and control groups and suggesting that the changed fALFF values may be potential diagnostic biomarkers for patients with DON.
The present study has some limitations. First of all, the size of samples included in our study is limited. Secondly, the patient may have made subtle movements during the scanning process, which may have affected the stability of MRI results. Finally, our conclusion needs further veri cation.

Conclusion
In summary, this is the rst study to nd functional brain activity changes in DON patients using fALFF.
The abnormalities in the central anterior gyrus, anterior cingulate gyrus, and middle cingulate gyrus are associated with eye movement, emotional disorder, and vision loss in DON patients. Our ndings may help to explain the pathological mechanisms of DON. Furthermore, the fALFF abnormalities may also serve as diagnostic signs for DON patients, which is of great signi cance.