Skin Intrinsic Fluorescence and Selected Measures of Visual Function and Aging in Older Adults

Abstract

Purpose:

Functional, structural and metabolic decline in many systems and in combination contribute to biologic aging and may be manifest as increased risk of morbid events such as neuropathy, albuminuria, and coronary artery disease or mortality. A biologic marker of aging may be a useful tool in identifying persons at increased risk of morbidity or mortality. We have measured skin intrinsic fluorescence (SIF) in a group of older adults to determine whether this easily determined measure could serve as such a biomarker.

Methods:

Survivors of a population based study of older adults in a moderate sized Midwestern town. Of the 1181 persons participating, 939 had measures of skin intrinsic fluorescence (SIF) and at least one functional or diagnostic characteristic at the most recent examination. Characteristics such as blood pressure, forced expiratory volume, vision, time to walk a standard course and medical history and their associations with SIF measures were examined. Mortality after the last examination with respect to SIF was also investigated. There were 118 deaths among those who participated in this phase of the study. All analyses pertinent to these findings were adjusted for age.

Results:

SIF measures were significantly associated with low contrast sensitivity, more errors on frequency doubling technology testing (loss of peripheral vision), self-reported poor vision, slow gait, poor forced expiratory volume, and self-reported poor health. SIF was also associated with increased risk of death. All of these analyses were adjusted for age.

Conclusions:

Skin intrinsic fluorescence provides easily obtained markers of age-related functional outcomes, suggesting SIF measurements may be useful to identify persons who may benefit from more frequent medical scrutiny to decrease morbidity and mortality.

Introduction

Vision along with other neurologic, metabolic, and functional characteristics tends to diminish with increasing age in adults.^1–4 Further, most older adults tend to manifest signs of functional decline in more than one system. Persons with dysfunctional deficiencies have been found to be more likely to have medical problems such as cardiovascular disease, renal disease, diabetes, arthritis. A biologic marker of a state of functional declines would be useful in identifying persons for whom closer observation or preventive measures might be operationalized to decrease overt medical conditions.

Skin intrinsic fluorescence (SIF) is a characteristic of many substances including animal and vegetable tissues. The fluorescence may be elicited by exposing the tissue to a given wave length of light which excites electrons or several molecules to fluores. Its measurement can be accomplished without causing tissue damage.⁵ To further explore the possibility of the associations of SIF with a number of functional measures of aging, we have measured skin intrinsic fluorescence in survivors of a population based study in Beaver Dam, Wisconsin.

Materials and Methods

A private census of the population of Beaver Dam, Wisconsin, was performed from 1987 to 1988, to identify all persons aged 43 to 84 years.⁶ The nearly 6,000 people identified were 99% white. Examinations have occurred roughly every 5 years starting in 1988. Tenets of the Declaration of Helsinki were followed, institutional review board approval was granted, and informed consent was obtained from each subject. During each study visit, standard measurements were made, and a questionnaire was administered with occasional additions dictated by specific aims of an exam phase. The most recent examination phase, June 2014–September 2016, included measures of skin intrinsic fluorescence. Of the 1181 persons participating, there were 1036 with SIF measurements. Reasons for not obtaining SIF measurements were remote examinations (N=61) where we did not have the SCOUT device or the participant could not spend the time needed to complete the study examination including the SIF measurement (N=84). Data are missing for at least one critical covariate in an additional 97 participants. There were 939 participants whose data are included in at least one analysis (Table 1). While there were statistically significant differences in several characteristics between those whose data were included and those whose data were not, the mean values of those included differed little from the entire participant group. There were no significant differences between the SIF measures for those who were and those who were not included. Because of previous associations of some SIF measures with characteristics of diabetes (e.g., complications), diabetes was included in multivariable models.

Table 1:

Participant Characteristics

	All Study Participants		Included in Analyses		p-value*
Characteristic	N	Mean/% (SD)	N	Mean/% (SD)	(age-adj)
Age (years)	1181	78.6 (6.8)	939	78.1 (6.4)	<.001
Sex (%M)	1181	40.0%	939	43.2%	<.001
BMI (kg/m2)	1154	30.7 (6.1)	929	31.0 (5.9)	0.01
Systolic BP (mmHg)	1108	134.3 (19.6)	932	134.7 (19.7)	0.11
Diastolic BP (mmHg)	1108	71.6 (9.9)	932	71.9 (10.0)	0.06
Hypertension	1156	80.7%	936	80.3%	0.81
Ever Smoke	1181	48.0%	939	48.1%	0.73
Cardiovascular Disease	1178	17.4%	939	16.9%	0.51
Diabetes	1004	25.3%	939	24.2%	0.003
HbA1c	989	5.93 (0.80)	924	5.92 (0.80)	0.65
Visual Acuity, #letters (better eye)	1115	51.7 (12.5)	938	52.5 (11.4)	0.001
Contrast Sens, Triplet (better eye)	994	1.64 (0.16)	874	1.65 (0.16)	0.03
PSD from FDT**	735	3.52 (1.25)	684	3.54 (1.27)	0.36
Gait Time, sec	989	3.23 (1.25)	880	3.18 (1.08)	<.001
FEV1***	1093	350.2 (135.4)	932	360.7 (135.3)	<.001
Fallen (%yes)	1180	12.3%	939	11.1%	0.12
SkinTone (MRef)	1036	237.5 (36.5)	939	236.9 (36.5)	0.04
SIF03 (log)	1036	2.68 (0.21)	939	2.68 (0.20)	0.23
SIF15 (log)	1036	0.34 (0.24)	939	0.33 (0.23)	0.45

^*p-value is age adjusted comparison of those included to those not included in analyses

^**pattern standard deviation from frequency doubling technology

^***forced expiratory volume

Functional measures

Overall self-reported health was obtained using the rating scale of excellent, good, fair and poor. Analyses were done combining excellent with good and fair with poor. Distance visual acuity was measured according to a modification of the Early Treatment Diabetic Retinopathy Study (ETDRS) protocol⁷ for each eye and was denoted as best-corrected visual acuity. Contrast sensitivity was measured using the Pelli-Robson charts and protocol.⁸ Overall self-reported vision was obtained using the rating scale of excellent, very good, good, fair and poor. Analyses were done combining excellent and very good with good and fair with poor.

Frequency doubling technology assesses visual field loss using a method that relies on detecting the presence of a flickering low spatial frequency and high temporal frequency grating at selected locations in the visual field.⁹ For analyses reported in this investigation the second generation Humphrey Matrix FDT perimeter F24–2 threshold strategy was used.

The forced expiratory volume (FEV1) was measured using the Jaeger AM1 device. This was performed three times, with the participant standing, taking as deep a breath as possible and blowing as hard and fast as s/he can. The best value (greatest flow rate) was used in the analysis.

The time to walk a measured course (gait-time) was obtained, as was a self-reported history of falls in the past year (excluding falls while doing sports).¹⁰

Disease related measures

Creatinine was measured from serum using the Roche COBAS 6000 chemistry analyzer (Roche Diagnostics, Indianapolis, IN) using an IDMS-traceable calibration. Estimated glomerular filtration rate (eGFR) was calculated using the CKD-EPI equation.¹¹ Glycosylated hemoglobin A1c was measured from EDTA packed cells using non-porous Ion Exchange High Performance Liquid Chromatography (HPLC) using the Tosoh Automated HPLC Analyzer HLC-723G8 which is certified by the National Glycohemoglobin Standardization Program (NGSP).

Biomarker measurement

Skin intrinsic fluorescence (SIF) was measured from the underside of the left forearm near the elbow using the SCOUT device.¹² Skin intrinsic fluorescence was excited with light emitting diodes (LEDs) centered at 375, 405, 416, 435 and 456 nm with the resulting emission measured over the 435 to 655 nm spectral range. A white LED was used to measure skin reflectance over the 435 to 655 nm spectral range to both correct for the optical distortion of skin and provide an objective measure of skin tone. The SIF utilizes the measured skin reflectance at both the excitation wavelengths and the emission wavelengths with two correction factors (Kx and Km) to mathematically adjust the measured fluorescence for optical distortion due to absorption of light from melanin and hemoglobin as well as light scattering. A variety of correction factors can be applied. SIF measures where the Km correction factor is larger than the Kx correction factors seems to control for intra-individual variations and be less affected by variations in melanin. For this paper, we focus on the measure from LED1 (centered at 375 nm) with Km=0.7 and Kx=0.4 (SIF3) and from LED5 (centered at 456 nm) with Km=0.9 and Kx=0.5 (SIF15). The highest and lowest centered values were selected because they represent excitation from different skin fluorphores. The low wavelength (from LED1) excites fluorescent advanced glycation endproducts (AGEs), nicotinamide adenine dinucleotide hydride (HADH) and flavin adenine dinucleotide (FAD). The highest wavelength (LED5) excites fluorescence from collagen cross-links and FAD.¹³

Covariate/confounder measures

Age was defined as the participant’s age at the time of the current examination. Diabetes was defined by glycosylated hemoglobin A1c >= 6.5% or self-reported diabetes. Height and weight were measured with a Health-o-Meter Scale. Blood pressures were the means of two readings taken by protocol. Individulas were asked if they had smoked more than 100 cigarettes in their lifetimes.

Definitions

Snellen equivalent of 20/40 or worse in the better seeing eye was considered visual impairment. Low contrast sensitivity was defined as triplet log score < 1.55 in the better seeing eye. Slow gait was considered present for persons requiring an aid for walking (walker, cane, arm, wheelchair) or if it took longer than 3.35 seconds for women or 3.22 seconds for men to walk the course. Poor lung function was considered present when FEV1 was < 250 for women or < 375 for men. The cutpoints for these analyses were based on the quartiles in all examined participants. Visual field deficiency was defined by a significant p-value for the pattern standard deviation (PSD) measure (as displayed on the FDT output) which is predominantly influenced by localized visual field loss. Hypertension was defined as 140/90 or current use of blood pressure lowering medication. A current smoker was defined as persons who reported having smoked more than 100 cigarettes in their lifetimes and reports still smoking.

Mortality

The population is followed regularly with annual phone contact, with the last round completed June 1, 2017. In addition, regular review of obituaries identified persons who had died as well as those who were listed as survivors. This population has strong ties to the community; obituaries are commonly printed in the local paper even for those who have moved away. In the 30 years of following the population, we have confirmed these methods successfully capture survival. Assessment of survival time used date of death (DOD) or last date not known to be dead (NKD). The information included in these analyses includes DOD and NKD dates through June 1, 2018. There were 125 deaths among the 939 participants with SIF measures.

Statistical analyses

Logarithmic transformations of SIF measures was performed. Binary versions of outcomes were chosen for all functional measures and standard logistic regression was performed. Minimally adjusted models were adjusted for age, gender, presence of diabetes, and skin tone as measured by the SCOUT device. Cox proportional hazards models were performed to assess mortality (time to death or censoring). SAS version 9 was used for all analyses.

Results

While adjusting for age, gender, diabetes status, and skin tone, higher SIF 3 and SIF 15 were significantly associated with self-reported poor vision, low contrast sensitivity, more errors on frequency doubling technology (FDT) visual field, slow gait, poor forced expiratory volume in 1 second, and poor self-reported health (Table 2). Further adjustment for lens status (presence or absence of intraocular lens), presence of self-reported cataract, presence of late age-related macular degeneration, and adjustment for other specific ocular diseases, estimated glomerular filtration rate, or smoking status did not materially affect the associations for either SIF measure with any of the functional outcomes (data not depicted).

Table 2:

Association of SIF with Each Functional Outcome (adjust for age, sex, diabetes, Skin Tone)

			SIF03†			SIF15††
Outcome	N	Event %	OR	(95% CI)	P-val	OR	(95% CI)	P-val
Visual Impairment (20/40+)	938	7.0	1.20	(0.90,1.61)	0.22	1.53	(1.13,2.06)	0.006
Low Contrast Sensitivity	874	17.0	1.54	(1.23,1.94)	<.001	1.59	(1.26,2.00)	<.001
FDT,* PSD** significant Loss	684	39.2	1.28	(1.05,1.56)	0.01	1.27	(1.05,1.55)	0.02
Self Report Poor Vision	938	14.2	1.36	(1.09,1.69)	0.006	1.64	(1.31,2.04)	<.001
Slow Gait	903	34.1	1.43	(1.19,1.73)	<.001	1.80	(1.48,2.20)	<.001
Poor FEV1	932	29.4	1.49	(1.24,1.78)	<.001	1.80	(1.50,2.16)	<.001
Fall 2+ times (past year)	939	11.1	0.98	(0.77,1.23)	0.84	1.19	(0.94,1.51)	0.16
Self Report Poor Health	933	16.3	1.27	(1.03,1.56)	0.02	1.60	(1.30,1.97)	<.001

OR is for the association of the SIF measure, per 1 standard deviation on the log scale, with the outcome

^*frequency doubling technology

^**pattern standard deviation

^†SIF3 is from LED1 (centered at 375 nm) with Km=0.7 and Kx=0.4

^††SIF15 is from LED5 (sentered at 456 nm) with Km=0.9 and Kx=0.5

There was a systematic increase in mortality with increasing SIF measurement. The maximum follow-up of observation was about 3.75 years. The HR was 1.41; 1.17,1.70 for SIF 3 and HR 1.50; 1.30,1.74 for SIF 15 adjusting in each model for age, sex, diabetes, skin tone, systolic blood pressure, diastolic blood pressure, forced expiratory volume, and smoking.

Discussion

In these analyses there are significant associations of measures of SIF with abnormalities of function in older adults. In addition, both SIF measures were significantly and substantially associated with mortality over a short follow-up (on average 3.75 years in this analysis) even after adjusting for specific markers of aging and age itself. There were 118 deaths among the 925 subjects who participated at the start of this follow-up. SIF is a measure of cross-linked proteins that fluoresce when exposed to light. These cross-linked proteins may result in various changes to cell function. In other studies, photo-aging has been related to sunlight exposure presumably due to oxidation.¹⁴ Oxidative stress has been thought the mechanism responsible for the development of increased intrinsic fluorescence associated with aging¹⁵ in other tissues such as the lens. Lerman attributed this to prolonged exposure to UV light above 295 nm resulting in the generation of a specific fluorogens tightly bound to a peptide within the lens protein.^{16, 17}

While it may seem intuitive that oxidation in skin exposed to sunlight would be associated with SIF, the association of SIF with dysfunction in other tissues and organs is not intuitive. However, there are suggestive data from animal models that caloric restriction may limit exposure to reactive oxygen species, a potential cause of tissue oxidation.¹⁸

SIF is associated with neuropathy, albuminuria, coronary artery disease, and retinopathy in persons with type 1 diabetes.^{16, 19–21} This may be partially explained by the additional glycemic exposures present with diabetes, although formation of advanced glycation endproducts occurs in persons who do not have diabetes. Whether there is an underlying association of intrinsic biologic aging manifesting as higher SIF levels or whether higher SIF levels are causally related to negative environmental exposures cannot be determined from our analyses. However it is likely that further research approaches, especially designs that have longitudinal measures with concomitant exposure measures, may provide insights into this question and might suggest alterations that individuals might make to retard progression of biologic aging and its associated functional abnormalities.

The association of SIF with mortality was very strong and appears to be linear, such that even moderate levels of SIF appears to be associated with increased mortality. While the length of follow-up is short, the age of the population allows for enough deaths to feel confident in these associations. The significance/strength of the SIF association with mortality was very similar to the age association. An association of skin autofluorescence has been reported to be associated with all cause mortality in a study by de Vos.²²

This study has inherent limitations. Participants were, on average, 78 years of age (range from 68 to 102 years) at the time of the exam. Because we did not have the capability of measuring SIF at earlier examinations we are unable to extrapolate to the relationship of SIF to functional characteristics in younger people or to determine if higher SIF preceeds functional changes. Participants in this exam phase are survivors within this study population, but are representative of people who survive to this age in general. In addition, virtually all participants were of European extraction limiting the generalizability of our findings to a more racially diverse US population. We cannot assess the influence of this characteristic on our findings but at this time there is no specific extant data indicating that this would alter the association of SIF measures with the outcomes we examined. We look forward to research in other ethnic groups to examine this possibility.

In summary, the findings of these analyses suggest that measures of SIF provide an easily obtained marker of age-related frailty. By inference this suggests that measurement of SIF in older persons may in the future be a useful tool to indicate persons who may benefit from more frequent preventive medical care because frailer individuals are at relatively increased risk of morbidity and mortality.^23–25

Submission Statement: This submission has not been published anywhere previously and it is not simultaneously being considered for any other publication. An abstract of this study was presented during the 2016 ARVO Annual Meeting, May 1–5, 2016, in Seattle, Washington, USA.