The influence of age, anthropometric and metabolic variables on LDIFLARE and corneal confocal microscopy in healthy individuals

Introduction The laser Doppler imaging (LDI) FLARE and corneal confocal microscopy (CCM) are reliable markers of small fibre function (SFF) and structure (SFS), respectively, but the impact of potential confounding variables needs to be defined. The objective of this study was to determine the influence of age, anthropometric and biochemical variables on LDI and CCM. Methods 80 healthy volunteers (43 males) (age: 39.7±15.2 yrs.) underwent LDIFLARE and CCM assessment and the effect of age, anthropometric and biochemical variables was determined using multivariate analysis. Results There was an age-related decline in LDIFLARE (0.07cm2/yr; R2 = 0.669; p = <0.0001) and CCM parameters (CNFD: 0.05 fibres/mm2 /yr; R2 = 0.590; p = <0.0001, CNBD: 0.06 branches/mm2/yr; R2 = 0.549; p = 0.001and CNFL 0.07 mm/mm2/yr; R2 = 0.369; p = 0.009). BMI did not influence SFF (p = 0.08) but had a significant independent association with CNFD (p = 0.01). Fasting triglycerides (TG) independently influenced the LDIFLARE (βc:-0.204; p = 0.008) and all CCM indices (βc:-0.191 to -0.243; p = <0.05). HbA1c was significantly associated with CNFD only (p = 0.001) but not with LDIFLARE, CNBD or CNFL (p = ≥0.05). Blood pressure and total cholesterol were not associated with LDIFLARE or any CCM parameters. There was a significant correlation between LDIFLARE and all CCM parameters (p = ≤0.01). Conclusions This study shows that in healthy controls, both SFF measured by LDIFLARE and SFS assessed by CCM showed a significant inverse correlation with age and triglycerides, perhaps suggesting the use of age-specific normative values when interpreting these outcomes. Furthermore, this study shows that in healthy controls, despite measuring different neural parameters, both methods correlated significantly with each other.


Conclusions
This study shows that in healthy controls, both SFF measured by LDI FLARE and SFS assessed by CCM showed a significant inverse correlation with age and triglycerides, perhaps suggesting the use of age-specific normative values when interpreting these a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
Diabetic polyneuropathy (DPN) is the commonest chronic complication of diabetes mellitus with a prevalence of up to 50% [1]. Prospective studies in both type-1 (TIDM) and type-2 (T2DM) diabetes have shown that the prevalence of DPN increases with the duration of diabetes [2][3][4] and is associated with significant morbidity and mortality and health economic cost [5].
There is an increasing drive for the earlier detection of DPN by quantifying large myelinated Aα and Aβ and in particular Aδ and C fibres [6,7]. Traditionally, neuropathic symptoms and signs along with nerve conduction studies were used to diagnose DPN and assess therapeutic outcome [8]. However, there is now increasing evidence to suggest that small fibre dysfunction and damage may precede large fibre neuropathy in subjects with impaired glucose tolerance [9,10] and diabetes [11,12] and predicts future-incident DPN [13]. This has led to the development of newer non-invasive techniques to assess small nerve fibre integrity.
Small nerve fibre integrity can be evaluated by quantifying small fibre structure (SFS) and function (SFF). SFS can be evaluated via skin biopsy by measuring IENFD and in the cornea using corneal confocal microscopy (CCM). SFF can be evaluated using the laser Doppler imager flare technique (LDI FLARE ), quantitative sensory testing (QST) of warm and cold thermal thresholds and the quantitative sudomotor axon reflex test (QSART) [14]. However, IEFND is invasive and requires considerable expertise for accurate quantification and QST is relatively time consuming, subjective and have a high coefficient of variation [15]. QSART remains limited to research centers due to the technical complexity, and again is a relatively time consuming technique [16].
CCM is a rapid, non-invasive ophthalmic technique that can accurately quantify corneal nerve morphology in diabetic patients with minimal and more advanced neuropathy [17], correlates with IENFD [18] and can be quantified objectively and reproducibly using automated algorithms [19,20]. The LDI FLARE method has also been shown to be a sensitive method for the detection of early small fibre dysfunction and also correlates with dermal nerve fibre density [21]. Furthermore, we have shown that the flare size is influenced by long term glycaemia in T1DM whilst features of the metabolic syndrome appear more important determinants in IGT and T2DM [22,23].
Given the potential utility of CCM and LDI FLARE in clinical screening and early identification of DPN, it is important to identify variables which may influence these measurements in a healthy population. Whilst corneal sensitivity may decline with age [24,25], corneal nerve fibre density has been related in some [26][27][28] but not all [29-31] studies. There are a limited number of studies showing age-dependent deterioration of both the LDI FLARE and thermal threshold testing [32,33]. This is the first study to evaluate the effect of age and other clinical variables on both small fibre function and structure in a cohort of healthy controls.

Ethical approval
This study was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of the NRES Committee East of England-Norfolk, UK. (REC reference: 13/ EE/0162). All subjects provided written informed consent.

Study design and selection of participants
Eighty (80) healthy subjects, aged 18 to 80 years, were recruited by invitation. This study is part of a larger longitudinal neuropathy study currently in progress in our centre involving both healthy subjects alongwith Type-1 and Type-2 diabetes patients. The current data presented was obtained as part of this larger study between 2015-2017. Participants were unselected from the local population including hospital staff where the ethnicity is 97% Caucasian.. Subjects with a history of impaired fasting glycaemia, pre-diabetes or diabetes (type-1, type-2 or other specific types) as per diagnostic criteria of the American Diabetes Association, 2017, were excluded [34] based on a fasting blood glucose (FBG) and glycosylated haemoglobinA 1c (HbA 1c ). Subjects were also excluded if there was any history of hypertriglyceridaemia, alcohol abuse, vitamin B12 and folate deficiency, thyroid or connective tissue disorders, renal or hepatic failure, malignancy, inherited neuropathy or exposure to any toxins including chemotherapy.

Clinical and biochemical assessment
All participants underwent testing for fasting glucose HbA 1c , lipids, thyroid, renal and hepatic profiles, vitamin B12 and folate. Overall, five patients-three due to pre-diabetes and two due to subclinical hypothyroidism were excluded. All screened patients underwent detailed neurological examination with the modified Neuropathy Disability Score (NDS) [35]. Vibration perception threshold was measured on the toes using the Neurothesiometer (Horwell Scientific, UK) and expressed as millivolts (mV). Sural nerve conduction velocity (SNCV) and amplitude (SNAP) were assessed in one leg (NC-stat|DPNCheck system (Neurometrix, Waltham, MA).

Corneal confocal microscopy (Fig 1)
CCM was performed using the laser scanning Heidelberg Retina Tomograph III confocal microscope with Rostock Corneal Module (Heidelberg Engineering GmbH, Dossenheim, Germany). Prior to scanning, topical anaesthesia (Minims 1 : Oxybuprocaine hydrochloride 0.4%; Bausch & Lomb, Kingston upon Thames, Surrey, UK) was applied to the cornea of both eyes followed by topical application of a viscous gel (Viscotears 1 : Carbomer 980 polyacrylic acid 0.2%; Alcon Eye Care, Frimley, Surrey, UK) to form a aqueous medium between the applanated corneal surface and disposable TomoCap covering the objective lens. Images were obtained from both eyes using our established methodology [36, 37] with a total duration for examination of both eyes between five to ten minutes. Six best images from both eyes were selected manually and automated image analysis was performed using purpose-written, proprietary software (CCM Image Analysis tool v1.1: Imaging Science and Biomedical Engineering, University of Manchester, Manchester, UK) [19]. The specific parameters measured per frame were: CNFD (number of fibres/mm 2 ), CNBD (number of fibres / mm 2 ), and CNFL (length in mm/ mm 2 ) [36]. LDI FLARE (Fig 2) LDI FLARE was undertaken using our established technique [38]. The procedure is carried out in a temperature-controlled room (25 ± 1˚C) after acclimatization [38]. A 1 cm 2 heating probe was applied to the dorsum of the foot, 2-3cm proximal to the first inter-phalangeal space and the skin was sequentially heated for 6 minutes in a step-wise algorithm: 44˚C for 2 minutes, 46˚C for 1minute, and 47˚C for 3 minutes and the reference area was scanned using a laser Doppler imager (Moor Instruments, Axminster, UK). On the flux image, the area of the LDI-FLARE was outlined using the free drawing tool and the results were expressed in cm 2 and measured using the Moor V 5.3 software.

Calculation of rate of decline of small fibre structure and function with age
The age range in our cohort of HC was between 19 and 74 years and was normally distributed. Using the age-related reference intervals suggested by Wright et al [39], we segregated our HC into four age groups (<30 years, 30-44, 45-59 and >60 years). The percentage difference between the 50th centile of the youngest and oldest group was divided by a factor of 6, covering the six decades, to establish the percentage rate of decline per decade using the formula: ½1 À fðMedianf lare areainOldestgroupÞ=ðMedianf lare areainyoungestgroupÞgx100 Numberof decadesinyears

Statistical analysis
Statistical analysis was performed using SPSS (version 20 for Windows) and StatsDirect version 3. Clinical characteristics as quantitative variables are expressed as mean ± standard deviation. Normal distribution of the data was determined with the Kolmogorov-Smirnov test.
To examine independent effect of age, BMI, HbA 1c and lipid indices on both LDI FLARE and CCM parameters, multivariate linear regression was used. Pearson's bivariate correlation was used, to assess the association of small fibre function and structure large fibre methods VPT, SNAP, SNCV) and to examine the relationship between LDI FLARE and CCM outcomes. The co-efficient of variation for CCM and LDI flare is 7.8% and 8.7%, respectively. and CNFL (p = 0.009; βc:-0.508). There was a weak but significant association between BMI and CNFD. Gender and both systolic and diastolic blood pressures did not correlate with any of the small fibre assessments (p = !0.05).

Rate of decline in small fibre function and structure
The distribution centiles for the four age categories are shown in Table 2. The median (50 th ) centile of the youngest and oldest group was used to derive the rate of decline.

Discussion
The techniques of CCM and LDI FLARE allow rapid, non-invasive detection of early small fibre deficits in DPN [11]. This study demonstrates that age has a highly significant independent  [31] found no significant association with age, whilst Grupcheva et al found an independent age-related reduction in corneal nerve density [26]. Neiderer et al found an age-related decrease of sub-basal corneal nerve density of 0.9% per year [27]. In a 3-year prospective study of 49 healthy individuals Dehgani et al reported a significant linear decrease of 0.05 mm/mm 2 in CNFL per one year increase in age [44], which is comparable to the decrease observed in our study. In a recent multicentre collaborative study there was a significant linear age-dependent decrease in CNFD (-0.164/mm 2 per year for men, and -0.161/ mm 2 per year for women; p = <0.0001)) and CNFL (R 2 = 0.026; p = 0.003),but no change in CNBD; height, weight, and BMI had no influence [28]. Lin et al [33] have commented that age is the most significant factor in determining sensory thresholds compared with the other factors of gender and anthropometric parameters and similarly, Hafner et al [45]have shown that thermal threshold detection increases with age. We have previously demonstrated a significant age-dependent decline in LDI FLARE size (r = -0.42; p = <0.0001) of 0.056 cm 2 per year, giving a percentage loss of 5.5% per decade.
The present study found no influence of gender, BMI or blood pressure on both small fibre function and structure in keeping with other studies of CCM and the LDI FLARE in HC. [28,46,47]. The effects of glycaemic measures on CCM and the LDI FLARE were inconsistent; HbA 1c showed a significant inverse correlation with CNFD but not with the LDI FLARE . In contrast, FBG correlated significantly with LDI FLARE and CNBD but not with other CCM parameters. Other studies of CCM in healthy controls also show conflicting results. A recent multinational data set report by Tavakoli et al [28] did not find any association between CCM and HbA 1c although FBG was not done; however in another study by Wu et al [40], HbA 1c was show to be the only independent clinical factor to account for variations in corneal nerve fibre length, independent of age. We did not expect an association between FBG and the LDI FLARE ; our previous study with a different cohort of HC did not demonstrate any association of HBA 1c or FBG with the LDI FLARE but CCM was not performed in this study. Further studies including longitudinal studies will be necessary to determine whether the association between fasting glucose and the LDI FLARE is real or a chance statistical finding.
Both the LDI FLARE and CCM correlated significantly with fasting triglycerides, but not total cholesterol. The association of TG with diabetic neuropathy has been previously highlighted in both observational [48][49][50] and interventional [51][52][53] studies. Indeed we have previously reported an inverse correlation between LDI FLARE and TG in normoglycaemic individuals [46]. To our knowledge, the present study is the first to find a relationship between TG and both SFF and SFS in healthy controls.
An intriguing finding of this study is that all three CCM indices (CNFD, CNBD and CNFL) correlated significantly with the LDI FLARE despite examining anatomically very different areas and tissues. This suggests that these measures are influenced by common determinants of neural health and could be used interchangeably in healthy controls. However, whether these are affected in the same manner in people with diabetes needs to be determined before suggesting that they can be used interchangeably to estimate early small fibre diabetic neuropathy; additionally, the effect of variables such as glycaemic control, TG, hypertension, other metabolic and inflammatory markers will need to be better defined. This will require high quality longitudinal studies. Additional measures of small fibre structure (IENFD) and function (QST) may provide a more robust and comprehensive analysis.
Another limitation of this study is that the participants were healthy volunteers; therefore, this study cannot be considered to be a true population-based study. Hence, it is not powered to study the effect of other demographic factors like geographical location including altitude, lifestyle choices (e.g. athletes' vs sedentary living), dietary patterns and smoking history. Whilst these are important factors which might further define the utility of these methods in the early detection of DPN, it will require much larger multicentred studies in diverse populations to understand their possible roles.
To conclude, this large cross-sectional study confirms the importance of age when interpreting LDI FLARE and CCM to diagnose small fibre neuropathy. Furthermore, it identifies the influence of glycaemia and in particular triglycerides on both small fibre structure and function, even in healthy individuals. Finally, the significant correlation between and CCM in healthy subjects validates their use in studies evaluating small fibre neuropathy.