Background
Glycation is a process where sugars and proteins are cross-linked because of increased blood sugar concentration. Based on these mechanisms, glycated proteins have been widely used to evaluate changes in blood sugar and the efficiency of blood sugar control. The most representative glycated proteins are glycated haemoglobin (HbA1c) and glycated albumin (GA) [
1]. HbA1c has been used to monitor changes in blood glucose for a longer time than GA. The half-life of HbA1c is 3 to 4 weeks, and HbA1c may reflect changes in blood glucose levels for the previous 3 to 6 months; in contrast, the half-life of GA is as short as 12 to 21 days, and thus, GA may reflect changes in blood glucose for the previous 3 to 4 weeks [
2]. For this reason, the clinical significance of GA in actual clinical practice is gradually increasing.
Meanwhile, once sugars are bound to proteins, conformational changes in proteins may occur, resulting in functional alterations of the proteins in the blood. Functional alteration may result in diseases, and thus, glycated proteins are significant or reasons other than simply observing blood sugar trends [
3]. For example, GA promotes the production of pro-inflammatory cytokines and stimulates protein kinase C, leading to systemic complications and the development of renal insufficiency or nephropathy in diabetes mellitus (DM) patients [
4‐
6]. Thus, GA is known to be a predictor of the systemic complications of DM, in particular diabetic nephropathy [
7].
Contrarily, GA also has clinical implications as a biomarker that reflects the degree of systemic inflammation. It has been reported that GA plays a role in the process of atherosclerosis, which can result in cardiovascular diseases [
8]. GA stimulates the growth of vascular smooth muscle cells and enhances the production of interleukin (IL)-6, a pivotal cytokine associated with atherosclerosis [
9]. For this reason, GA has also been reported to be useful in reflecting the inflammatory burden in non-DM patients with cardiovascular diseases. Moreover, in our previous study, we demonstrated that GA could reflect disease activity in non-DM patients with rheumatoid arthritis. Patients with active rheumatoid arthritis exhibited a significantly higher level of GA than those with inactive rheumatoid arthritis [
10].
It is reported that GA can initiate and accelerate the production of pro-inflammatory cytokines, tumour necrosis factor-alpha, IL-6, and IL-8 [
5,
11]. These pro-inflammatory cytokines are the main molecules involved in the pathogenesis of antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) [
12,
13]. Thus, it can be hypothesised that GA also participates in the pathogenesis of AAV and can reflect the cross-sectional activity of AAV. However, there have been no studies on investigating the clinical significance of GA in patients with AAV. Hence, in this study, we investigated whether GA at diagnosis could reflect the cross-sectional activity of AAV and predict poor outcomes during follow-up in AAV patients without DM. In addition, we compared the clinical significance of GA with that of other glycated proteins such as HbA1c and GA/HbA1c.
Discussion
In this study, GA/HbA1c was selected for two reasons: First, it was previously reported that GA/HbA1c could be used as a factor reflecting long-term glycaemic control along with GA and HbA1c [
19]. Second, although GA/HbA1c did not show a significant difference between DM and non-DM patients, theoretically, elevated GA at diagnosis could reflect the extent of inflammation without confounding effects of hyperglycaemia in AAV patients. GA and GA/HbA1c were selected as biomarkers instead of HbA1c since it showed no significant correlation with any clinical or laboratory variable in the correlation analysis (Table
2). We first demonstrated that GA at diagnosis could reflect BVAS assigned to renal manifestation of AAV, and predict ESRD development during follow-up. In addition, GA/HbA1c also showed a pattern similar to that of GA, but the clinical significance was not as high as GA. It is important to highlight that GA levels may be influenced by decreased liver and renal function. GA levels are usually low in patients with decreased liver function. It seems that liver disease had the least effect on GA/HbA1c compared to GA and HbA1c [
20,
21]. In addition, GA levels are not affected by non-nephrotic range of proteinuria in CKD patients. However, in CKD patients with nephrotic range of proteinuria, GA level is decreased independent of glycaemic status [
22]. Therefore, it is important to interpret GA with caution in AAV patients with abnormal liver function of nephrotic range of proteinuria in CKD patients. In general, GA may be considered as a more useful biomarker that could reflect inflammation that could lead to renal failure in AAV patients.
It is unclear whether GA plays a causal role in the inflammatory process or is a mere consequence of the cascade. Advanced glycation end (AGE) products, including GA, which are mainly formed in a hyperglycaemic state, can be also found under inflammatory conditions [
23,
24]. The receptor for AGEs (RAGE) is a pattern-recognition receptor, and once its ligands, and GA bind to RAGE, various intracellular signalling cascades are initiated [
25,
26]. GA can upregulate the gene expression of monocyte chemoattractant protein-1, IL-6, and IL-8 via nuclear factor-kappa B signalling, and it can also enhance the expression of c-fos and c-jun via extracellular signal-regulated kinases, comparable with tumour necrosis factor-α, IL-1β or lipopolysaccharide [
8,
26]. An increase in the production of GA due to systemic inflammation can further aggravate inflammation, in addition to the deterioration of glucose metabolism, resulting in an amplified vicious cycle. Although we could not definitively determine whether increased GA production might clinically exacerbate AAV activity in the present study, it can be hypothesised upon with reasonable confidence, as the confounding factor of DM was eliminated from this study. Moreover, elevated GA at diagnosis may reflect the extent of inflammation without the confounding effect of hyperglycaemia.
Contrary to the initial assumption, GA was not directly correlated with BVAS in this study. Given that BVAS is composed of nine systemic items, correlations with each systemic item were investigated, and it was confirmed that BVAS assigned to renal manifestation showed a significant correlation. In addition, there was a significant difference in GA between AAV patients with ESRD and those without ESRD, and GA was found to be a possible predictor renal failure leading to ESRD. A previous 5-year prospective population-based study reported that the baseline value of GA independently and significantly predicted renal dysfunction, along with age and uric acid [
27]. Therefore, in the clinical setting, GA, as a biomarker, can reflect the cross-sectional renal manifestation and predict progression to ESRD in AAV rather than directly reflecting AAV activity.
A question arises as to why HbA1c, also a glycated protein, did not show a significant correlation with BVAS assigned to renal manifestation. In a previous study, it was reported that HbA1c may not be a reliable marker for glycaemic state in cases involving renal comorbidities, haemoglobinopathies, and pregnancy, and further, GA could overcome the limitations of HbA1c in these medical conditions [
28]. In this study, AAV activity showed an inverse correlation with haemoglobin. Therefore, the association between the degree of inflammation represented by BVAS and HbA1c may not be significant because the enhanced activity of AAV may exacerbate anaemia due to insufficient production of erythropoietin [
28]. Moreover, renal manifestation itself could reduce the reliability of HbA1c, leading to discordance in the proportionality of the AAV-related inflammatory burden between HbA1c and GA.
In this study, both GA and GA/HbA1c were demonstrated to have potential as biomarkers for assessing the cross-sectional extent of renal involvement in AAV and predicting the progression to ESRD. GA showed significant correlations with renal manifestation, exhibiting a difference between AAV patients with ESRD and those without. Using an optimal cut-off, GA could predict the relative risk of ESRD as well as ESRD occurrence, with the Kaplan–Meier survival analysis. However, GA/HbA1c did not show a significant difference based on the presence or absence of ESRD. Its predictive potential for ESRD occurrence using the Kaplan–Meier survival analysis was not significant. Therefore, we suggest GA rather than GA/HbA1c as a novel biomarker for renal manifestation in AAV patients.
Recently, albumin-adjusted GA (the ratio of GA to serum albumin level) was suggested as a new indicator for glycaemic control [
29]. We know that inflammatory burden initiates and accelerates the production of GA, along with the hyperglycaemic state, but serum albumin falls in an inflammatory state. Therefore, we can expect that albumin-adjusted GA would increase as AAV activity rises, and predict poor outcomes better than GA. First, in terms of the cross-sectional BVAS, we conducted another correlation analysis and found that albumin-adjusted GA showed significant correlation with the cross-sectional BVAS (
r = 0.453,
P < 0.001) and BVAS assigned to renal manifestation (
r = 0.501,
P < 0.001). Therefore, albumin-adjusted GA could be used as a biomarker to directly reflect the cross-sectional both BVAS and BVAS assigned to renal manifestation in AAV patients.
Second, in terms of ESRD occurrence, AAV patients with ESRD exhibited a higher median albumin-adjusted GA than those without ESRD (3.7 vs. 3.4,
P = 0.041). However, in the ROC curve analysis to obtain an optimal cut-off, albumin-adjusted GA (area under the curve 0.752, 95% CI 0.547, 0.957) exhibited a lower area than GA (area under the curve 0.722, 95% CI 0.563, 0.881) (See Supplementary Figure S2, Additional File
3). In addition, the optimal cut-off of albumin-adjusted GA for ESRD occurrence was set at 3.42 with the sensitivity and the specificity of 87.5% and 58.2%, respectively. However, the relative risk of albumin-adjusted GA ≥ 3.42 for ESRD occurrence was lower than that of GA ≥ 14.25% (9.172 vs. 12.040). Although albumin-adjusted GA ≥ 3.42 could significantly predict ESRD occurrence during the follow-up duration based on ESRD, the statistical significance of albumin-adjusted GA did not surpass that of GA (
P = 0.046 vs.
P = 0.020) (See Supplementary Figure S3, Additional File
4). Therefore, GA could predict ESRD occurrence during follow-up better than albumin-adjusted GA. Based on these results, we suggest that GA, rather than albumin-adjusted GA, is more clinically helpful in predicting ESRD occurrence.
There are traditional and conventional risk factors that predicts the progression to ESRD in general population. Serum creatinine and age are well known risk factor for ESRD. We conducted a Cox hazard model analysis to evaluation the predictive ability of several laboratory variables and patient characteristics including age and gender. In univariate analysis, BVAS, white blood cell count, haemoglobin, serum creatinine and GA ≥ 14.25% were statistically significant. In multivariate analysis, only serum creatinine was proven to be a significant predictor for ESRD in AAV patients without DM (HR 1.323, 95% CI 1.019, 1.717,
P = 0.036) (See Supplementary Table S2, Additional File
5). The predictive ability of GA could not surpass that of serum creatinine. However, we believe that GA could give additional clinical information to physicians for predicting ESRD in AAV patients without DM. With a multi-centric and prospective future study with a large number of patients, it will provide more dynamic and clearer information on the clinical usefulness of GA in predicting ESRD in AAV patients and will validate the results of our study further.
This study has several limitations. First, the retrospective study design did not allow for the serial collection of both GA and HbA1c results in non-DM patients with AAV. Second, the number of patients was not large enough to generalise the results of this study for application in all patients with AAV. Third, GA has been reported to be capable of predicting the development of DM in pre-diabetic or euglycemic patients [
30]; however, we could not evaluate it because it was not easy to distinguish the causes of elevated glucose levels (isolated DM versus the drugs for AAV treatment, such as steroids and calcineurin inhibitors that can increase blood sugar. However, for the first time, we demonstrated the predictive capability of GA for the extent of renal involvement in AAV, and thus, our study has clinical significance as a pilot study. A future study with a larger number of patients and with serial results of both GA and HaA1c can validate our study findings and provide more information on the clinical role of GA in AAV.
In conclusion, GA at diagnosis can reflect BVAS assigned to renal manifestation of AAV and predict renal failure to progress to ESRD during follow-up better than HbA1c or GA/HbA1c in non-DM patients with AAV. Therefore, we expect that GA may be used as a biomarker for renal dysfunction and ESRD occurrence during follow-up in AAV patients.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.