Skin autofluorescence and cause-specific mortality in a population-based cohort

Abstract

We aimed to assess the association of SAF with cardiovascular mortality in the general population and the possible association between SAF with other disease-specific mortality rates. We evaluated 77,143 participants without known diabetes or cardiovascular disease. The cause of death was ascertained by the municipality database. The associations between SAF and all-cause mortality, cardiovascular mortality and cancer mortality were assessed with Cox proportional hazard analysis.After a median follow-up of 115 months, 1447 participants were deceased (1.9%). SAF and age-adjusted SAF-z score were higher in all mortality groups. Cox regression analysis revealed that the highest quartile of SAF was associated with increased odds of cardiovascular mortality, (HR) 12.6 (7.3–21.7) and after adjusting for age (HR 1.8 (1.0–3.2)). Significance was lost after additional adjustments for sex, smoking status, and BMI (HR 1.4 (0.8–2.5). For cancer-related mortality the highest quartile of SAF was associated with higher probability of mortality in all models (unadjusted HR 8.6 (6.6–11.3), adjusted for age HR 2.1 (1.6–2.8)), adjusted for age, sex, smoking status, and BMI HR 1.7 (1.3–2.4)). SAF is associated with all-cause mortality as well as cardiovascular and cancer-related mortality in the general population.

Introduction

Advanced Glycation End-products (AGEs) are a heterogeneous group of compounds, which are formed through non-enzymatic glycation of proteins, lipids and nucleotides. They accumulate with age, by endogenous accumulation under hyperglycemic conditions, oxidative stress or chronic inflammation¹. Additionally, dietary AGEs and smoking are exogenous sources of AGEs^2,3. AGEs are difficult to be cleared from tissues, and a decrease in renal function contributes to reduced AGE clearance and increased accumulation⁴. AGEs are associated with the development of diabetes-related complications and atherosclerosis^5,6.

A method has been developed to measure fluorescent AGEs noninvasively⁷. The “AGE Reader” can assess AGE accumulation in the skin by measuring skin autofluorescence (SAF) at a specific wavelength after exposure to light. SAF has been shown to be associated with diabetes, and its complications⁸, as well as with the development of cardiovascular disease and many other factors⁹.

A large prospective study in the general population found a significant association between SAF and all-cause mortality¹⁰. Multiple smaller studies targeting a specific study population (i.e., in people with type 1 and type 2 diabetes or end-stage kidney disease) have shown significant associations between SAF and cardiovascular mortality, all-cause mortality, or both^11–13. Recently, a study showed that high SAF can predict new cancers in people with type 2 diabetes¹⁴.

The goal of this study is to investigate the association between SAF and cardiovascular mortality in the general population, as well as the possible associations of SAF with other disease-specific mortality rates.

Methods

Participants

Participants were included from the Lifelines Cohort Study, a large population-based study in the northern region of the Netherlands. The design and selection criteria for the original Lifelines Study have been described elsewhere¹⁵. At baseline, both physical examination and extensive questionnaire data were collected.

Participants were eligible for the present investigation if they were of Western European descent, aged between 18 and 90 years, and underwent baseline investigations with a validated SAF measurements between 2007 and 2013. We did not include participants with extreme values of SAF (< 0.8 n = 5 or > 4.5 n = 27) as these represent measurement errors, those with missing data in the National Office of Statistics Netherlands (CBS) database (n = 32) or those with data mismatch between Lifelines and CBS database (sex mismatch n = 5, mortality mismatch n = 7) and participants with baseline diabetes and / or cardiovascular disease (CVD) (n = 4418). This resulted in 77,143 participants available for analysis (Supplementary Fig. S1).

This study was approved by the medical ethical review committee of the University Medical Center Groningen. All participants provided written informed consent before participating in the study and all methods were carried out in accordance with relevant guidelines and regulations for human subjects.

Clinical examination

Information about health behaviour, and past and current diseases were collected by self-administered questionnaires. Information regarding smoking behaviour (never, former and current smoking) was collected from the questionnaire. Anthropometric measurements and medication use were verified by a certified research assistant. Weight was measured to the nearest 0.1 kg, and height and waist circumference were measured to the nearest 0.5 cm, with participants wearing light clothing and no shoes. BMI was calculated as kg/m². Systolic and diastolic BP and heart rate were measured every minute for 10 min in the supine position using an automated Dinamap monitor (GE Healthcare, Freiburg, Germany); for analysis the average of the last the 3 measurements was used. Medication use was scored according to the Anatomical Therapeutic Chemical Classification System. The vital status of the participants, including cause of death as ICD-10 codes was ascertained by the CBS database. Mortality data were available from 2008 to 2020.

Skin autofluorescence

Skin autofluorescence (SAF) was measured noninvasively using a calibrated AGE reader (Diagnoptics Technologies, Groningen, the Netherlands). The AGE reader contains a light source, emitting light with a wavelength between 300 and 420 nm. While the volar side of the arm rests on the AGE reader, only the emitted and reflected light is measured by a spectrometer in the 300–600 nm range. SAF is calculated as the ratio of emitted light (420–600 nm) and excitation light (300–420 nm), expressed in arbitrary units (AU). In Lifelines, three consecutive SAF measurements were done on the same day within a time frame of approx 2 min, and the average of these three values is used for analyses.

Biochemical measurements

Following an overnight fast, blood samples were taken between 08:00 and 10:00 a.m. Laboratory measurements were performed on the same day. Serum creatinine was measured on a Roche Modular P chemistry analyser (Roche, Basel, Switzerland) and renal function was calculated as estimated (e)GFR with the formula developed by the Chronic Kidney Disease Epidemiology Collaboration¹⁶. Total cholesterol and high-density lipoprotein cholesterol (HDL) were measured using an enzymatic colorimetric method, triacylglycerol (TG) was measured using a colorimetric UV method, and low-density lipoprotein cholesterol (LDL) was measured using an enzymatic method, on a Roche Modular P chemistry analyser (Roche). HbA_1c (EDTA-anticoagulated) was analysed using an NGSP-certified turbidimetric inhibition immunoassay on a Cobas Integra 800 CTS analyser (Roche Diagnostics Nederland, Almere, the Netherlands). Except for vital status all previously mentioned variables, including clinical examination and skin autofluorescence were obtained at baseline.

Definitions and statistical analyses

All analyses were performed in PASW Statistics (Version 25, IBM, Armonk, NY, USA) and Stata Statistical Software (version 16.1; Stata Corp, College Station, TX, USA).

The study population was categorized by SAF quartiles, first quartile (SAF < 1.63), second quartile (SAF 1.63–1.88), third quartile (SAF 1.88–2.18) and fourth quartile (SAF > 2.18). As SAF is strongly affected by age, we also calculated age-adjusted SAF levels (SAF-z score). The associations between SAF and all-cause mortality, cardiovascular mortality and cancer mortality were assessed with Cox proportional hazard analysis. The first quartile of SAF was used as the reference. Kaplan–Meier curves for mortality during follow-up were plotted for SAF quartiles using the Stata STKAP module¹⁷.

Data are presented as means ± SD, or medians and interquartile ranges (IQR) when not normally distributed. Hazard ratios are presented as HR (confidence intervals). Means were compared between groups with ANOVA. When variables were not normally distributed, medians were compared using the nonparametric Mann–Whitney U test. The χ² test was used to analyse categorical variables. Values of p < 0.05 were considered statistically significant. Privacy guidelines of the CBS do not permit publication of data with N < 10 observations.

Results

Baseline characteristics

The study population had a mean (SD) age of 44.2 (11.7) years; 41% were males. The median follow-up duration was 115 months (IQR 105–128 months). On December 31st 2020, 1447 participants were deceased (1.9%). Participants who were deceased were more likely to be male, be older, have a higher BMI and waist circumference, as well as higher systolic and diastolic blood pressure (Table 1). In addition, they had lower eGFR, and higher lipid levels (total cholesterol, LDL and TG, but not HDL) and HbA1c concentrations. Current and former smokers were more common among the deceased, as was usage of blood-pressure lowering drugs and statins (Table 1).

Table 1.

Clinical characteristics of the study population at baseline in relation to mortality.

Characteristic	Alive	Deceased	P value
Sex (n (% deceased); male/female)	30924/44768	732 (2,3%)/719(1,6%)	< 0.001
Age (years)	43.9 ± 11.7	58.6 ± 12.8	< 0.001
BMI (kg/m2)	26.0 ± 4.2	26.8 ± 4.3	< 0.001
Waist circumference (cm)	90 ± 12	94 ± 13	< 0.001
Systolic BP (mmHg)	125 ± 15	134 ± 18	< 0.001
Diastolic BP (mmHg)	74 ± 9	76 ± 10	< 0.001
Heart rate (bpm)	71 ± 11	71 ± 12	0.31
eGFR (ml/min/1.73m2)	97 ± 14	87 ± 16	< 0.001
Total cholesterol (mmol/l)	5.1 ± 1.0	5.4 ± 1.0	< 0.001
HDL-cholesterol (mmol/l)	1.4 (1.2–1.7)	1.4 (1.2–1.7)	0.99
LDL-cholesterol (mmol/l)	3.2 ± 0.9	3.5 ± 0.9	< 0.001
TG (mmol/l)	0.97 (0.71–1.38)	1.11 (0.84–1.59)	< 0.001
HbA1c (mmol/mol)	36 ± 3	38 ± 3	< 0.001
Current smoking (%)	21.2	25.5	< 0.001
Former smoking (%)	30.5	40.0	< 0.001
% using BP-lowering therapy	9.4	25.8	< 0.001
% using statins	3.3	8.8	< 0.001
Skin autofluorescence (AU)	1.91 ± 0.41	2.32 ± 0.50	< 0.001
SAF-z score	0.09 ± 0.69	0.35 ± 0.91	< 0.001

Data are presented as numbers, means ± SD, medians (IQR) or percentages.

BMI body mass index, BP blood pressure, eGFR estimated glomerular filtration rate, HDL high density lipoprotein, LDL low density lipoprotein, TG triacylglycerol, HbA_1c glycated haemoglobin.

Both SAF and SAF-Z were significantly higher in deceased participants compared to those who remained alive. Additionally, mortality was highest in the highest quartile of SAF, accounting for 57% of all mortality. For SAF-z, 32% of all mortality was in the highest quartile (Fig. 1).

Fig.1.

Unadjusted cardiovascular, cancer-related and other causes of death according to SAF (top) and SAF-z (bottom) quartiles. Data are presented as numbers; CVD, cardiovascular disease.

The Kaplan–Meier analysis revealed statistically significant differences in survival probability between the first and other SAF quartiles (P < 0,001, Fig. 2). The lowest survival was in the highest quartile of SAF. After adjustment for age, both the 3rd and 4th quartiles of SAF deviated from the first quartile (Fig. 2).

Fig.2.

Kaplan–Meier survival estimates for the SAF groups. Top, unadjusted; Bottom adjusted for covariates; timescale in months.

These observations were confirmed by Cox proportional hazard analysis (Fig. 3). After additional adjustment for age, BMI, sex and smoking status, survival for participants in only the highest quartile of SAF remained significantly lower compared to the first quartile.

Fig.3.

Cox proportional hazards analysis for all-cause, cardiovascular, and cancer-related mortality according to SAF. Panel 1, adjusted for age. Panel 2, adjusted for age, BMI, sex and smoking status. The 1st SAF quartile was used as the reference.

Association of SAF with cause-specific mortality

Of all individuals who died from cardiovascular causes, 62% were in the highest SAF quartile, for cancer-related mortality, 55% and for other causes of death, 59% (Fig. 1 Top). After adjusting for age, using SAF-z yielded a 40% incidence of cardiovascular mortality in the highest quartile. The cancer-related mortality rate was 38%, and the other causes of death were 32% (Fig. 1 Bottom). SAF and SAF-z were higher in all mortality groups compared to those still alive. (Table 2.).

Table 2.

Skin autofluorescence and mortality due to cardiovascular, cancer-related, other and all-cause mortality.

	N	SAF	SAF-z	P*
CVD	258	2.38 ± 0.51	0.35 ± 0.91	< 0.001
Cancer	855	2.30 ± 0.49	0.35 ± 0.91	< 0.001
Other	338	2.33 ± 0.52	0.34 ± 0.93	< 0.001
Total	1451	2.32 ± 0.50	0.35 ± 0.91	< 0.001
Alive	75,692	1.91 ± 0.44	0.09 ± 0.69

Data are presented as numbers or means ± SD.

CVD cardiovascular disease.

*Compared with those still alive.

Additional causes of death are reported in Supplementary Table S2. Of all mortality due to CVD, SAF was higher in peripheral artery disease (2.41 ± 0.73) and cerebrovascular disease (2.43 ± 0.46) compared to heart disease (2.36 ± 0.51). In addition, SAF was high in mortality due to chronic lower respiratory disease (2.73 ± 0.49) and Alzheimer’s disease (2.39 ± 0.42). SAF was lowest in mortality due to accidents (2.22 ± 0.47) and unknown causes of death (2.18 ± 0.54).

Causes of death by cancer type are reported in Supplementary Table S3. SAF was highest in lung cancer and lowest in breast cancer. Age-adjusted SAF was highest in lung cancer (2.43 ± 0.49) and melanoma (2.34 ± 0.55) and lowest in prostate (2.32 ± 0.39) and pancreas cancer (2.17 ± 0.41) and breast cancer (2.10 ± 0.41).

All cancer groups not significantly associated with SAF-z pooled together were not significantly associated with SAF-z (N = 287 SAF-z = 0.17 ± 0.82 P = 0.08).

Cox regression analysis revealed that the highest quartile of SAF was associated with greater risk of cardiovascular mortality, (unadjusted hazard ratio (HR) 12.6 (7.3–21.7)). After adjusting for age, there was still a higher probability of cardiovascular mortality (HR 1.8 (1.0–3.2)), significance was lost after additional adjustments for sex, smoking status, and BMI (HR 1.4 (0.8–2.5)).

For cancer-related mortality, the highest quartile of SAF was associated with higher odds of mortality in all models, (unadjusted HR 8.6 (6.6–11.3)), after adjusting for age (HR 2.1 (1.6–2.8)) and after additional adjustment for sex, smoking status, and BMI (HR 1.7 (1.3–2.4)).

Discussion

The current study is one of the first to report on the association between SAF and cause-specific mortality in the general population. We observed that SAF was associated with a higher all-cause, cardiovascular and cancer-related mortality. Moreover, age-adjusted SAF values were found to be significantly higher for all cause-specific mortality groups, except for people who died of infectious diseases such as Influenza and pneumonia.

A previous study from our group reported an association between SAF and all-cause mortality in the general population¹⁰. Other studies on mortality have been performed in specific groups of patients, and have reported an association between SAF and cardiovascular mortality in patients with chronic kidney disease¹⁸, chronic hemodialysis^11,19–21, type 1 and type 2 diabetes¹³, and peripheral artery disease¹². The present study extends our earlier observations because we were able to link overall mortality data with specific causes of death, which are registered at the official death certificates. We confirmed the association of SAF and cardiovascular mortality in the general population. This was not surprising, as previous studies have shown a strong association between SAF and CVD, as well as cardiovascular risk factors. In the present study, we found 40% greater odds of cardiovascular mortality; however, after we adjusted for age, sex, smoking status, and BMI, the association was no longer significant, as all these factors by themselves are associated with higher SAF levels.

Recently, a French study showed that higher SAF predicted the new-onset of cancer in individuals with type 2 diabetes¹⁴. Our current study clearly shows an association between increased SAF and cancer mortality, and as such, these findings confirm the association between SAF and cancer. There is evidence that AGEs may play a role in cancer development and progression²². AGEs have been demonstrated in different types of tumors²³. Also, upregulation of the RAGE receptor is associated with tumor size and malignant potential of ovarian and breast cancer^24,25. However, it should be noted that the association between advanced glycation end-products and cancer could be overestimated due to significant shared risk factors between the pathophysiology of AGE accumulation and cancer. An increase in blood-pressure is associated with increased cancer-related mortality²⁶. Obesity is a proven risk factor for CVD, and is estimated to be related with 1 in 5 types of cancer²⁷. There is overwhelming evidence for smoking – a factor well-known to increase SAF- as a risk factor for heart disease and malignancies, and indeed, in our data the highest SAF values were measured in people who subsequently died from lung cancer. In diabetes, an increased incidence of certain types of cancer is observed compared to individuals without diabetes, with increased levels of insulin and insulin-like growth factor as a promotor of cell proliferation as one of the possible explanations²⁸. Conversely, a healthy lifestyle protects against cardiovascular disease as well as incident cancer²⁹. Coincidentally, all these shared risk factors are also associated with higher SAF⁹, while a healthy lifestyle is associated with a lower SAF³⁰. However even after adjusting for confounders SAF remained significantly associated with cancer-related mortality, unlike cardiovascular mortality. This could be partly explained by lower number of cardiovascular deaths.

A striking finding was that SAF was also elevated in non-cardiovascular and cancer-related mortality. The association between SAF and chronic lower respiratory diseases can be explained by the fact that tobacco use is both an important etiologic factor in respiratory disorders, and is also an important source of AGEs. In previous studies, respiratory disorders have been strongly associated with elevated SAF³¹.

For mortality from accidents, clear conclusions cannot be drawn due to the low number of events. A possible confounder is mortality due to falls in frail elderly, as SAF is associated with frailty³². Additionally, mental disorders are a risk factor for accidental death³³ and SAF has been shown to be associated with affective disorders³⁴.

As the association between SAF and different causes of death can be explained by shared risk factors, the role of SAF in the pathophysiology in non-cardiovascular mortality remains unclear and needs further investigation. According to our data, SAF is elevated in every major group of cause-specific mortality except infectious disorders (influenza and pneumonia). This finding supports the possible utility of SAF as a screening tool, not only for cardiovascular disease and diabetes³⁵, but also for mortality by other causes of death. Measuring SAF may help in selecting populations who benefit from cancer screening, as was also suggested by Foussard et al.

Strength and limitations

A strength of the current study is the unparalleled large number of participants. The Lifelines cohort is representative of the populations of the northern provinces of the Netherlands for sociodemographic parameters³⁶. Also, the reliability of cause of death statistics in the Netherlands is high³⁷.

Limitations include the relatively young age of participants in the Lifelines cohort and relatively short follow-up of approximately 10 years, which resulted in a low number of deaths. This resulted in a lack of power to investigate or reject the association between SAF and cause-specific mortality in a fully-adjusted model. The relatively low age of participants in Lifelines could have influenced the results. In the Netherlands, people are more likely to die of cancer than any other cause, while older people die more frequently due to cardiovascular disease³⁸.

Conclusions

SAF is associated with all-cause mortality as well as cardiovascular and cancer-related mortality in the general population.

Author contributions

HEB and BHRW contributed to the study design. HEB and BHRW performed the statistical analyses. All authors contributed to the analyses and interpretation of the data. HEB drafted the initial version of the manuscript. All authors participated in the critical revision of the manuscript and approved the final version. All authors agree to be accountable for all aspects of the work.

Data availability

The manuscript is based on data from the Lifelines Cohort Study. Lifelines adheres to standards for data availability, and allows access for reproducibility of the study results. The data catalog of Lifelines is publicly accessible at www.lifelines.nl. The dataset supporting the conclusions of this article is available through the Lifelines organization (e-mail: research@lifelines.nl). For data access, a fee is required.

Competing interests

AJS is founder and shareholder of Diagnoptics Technologies (Groningen, the Netherlands), the manufacturer of the AGE reader that was used in the present study. The other authors do not have any competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-71037-7.