Prognostic impact of prediabetic glycated hemoglobin levels in nondiabetic patients undergoing percutaneous coronary intervention: A systematic review and meta-analysis

Abstract

BACKGROUND

Glycated hemoglobin (HbA1c) is a well-established biomarker for diagnosing and managing diabetes. However, its prognostic significance in patients without diagnosed diabetes undergoing percutaneous coronary intervention (PCI) remains uncertain. This systematic review and meta-analysis evaluates the association between elevated HbA1c levels in the prediabetic range (≥ 5.7%) and adverse cardiovascular outcomes in this population.

AIM

To investigate the association between elevated HbA1c levels in the prediabetic range and adverse outcomes in patients without diagnosed diabetes undergoing PCI.

METHODS

We systematically searched PubMed, EMBASE, and Cochrane Central through April 2025 for studies comparing clinical outcomes in coronary artery disease (CAD) patients without a prior diabetes diagnosis, stratified by HbA1c levels (≥ 5.7% vs < 5.7%). Risk ratios (RR) with 95% confidence intervals (CI) were pooled using a random-effects model. Statistical analysis was performed using R software (version 4.3.2). Primary outcomes were long-term all-cause mortality and major adverse cardiovascular events (MACE); secondary outcomes included short-term mortality and cardiac death.

RESULTS

Ten studies involving 32403 patients (mean age: 60 years; 29% female) were included. Elevated HbA1c levels in patients without diagnosed diabetes were significantly associated with increased risk of long-term all-cause mortality (RR: 1.30; 95%CI: 1.10-1.54; P < 0.01; I² = 41%) and MACEs (RR: 1.31; 95%CI: 1.01-1.69; P = 0.04; I² = 61%). Although the risks of short-term all-cause mortality (RR: 1.16; 95%CI: 0.88-1.53; P = 0.29; I² = 1%) and cardiac mortality (RR: 1.76; 95%CI: 0.85-3.67; P = 0.13; I² = 94%) were elevated, they did not reach statistical significance. Sensitivity analyses confirmed the robustness of the findings despite moderate to high heterogeneity in some outcomes.

CONCLUSION

Among CAD patients without diagnosed diabetes, elevated HbA1c levels in the prediabetic range (≥ 5.7%) are independently associated with worse long-term outcomes following PCI. HbA1c may serve as a valuable biomarker for post-PCI risk stratification in this metabolically at-risk group.

Key Words: Glycated hemoglobin; Percutaneous coronary intervention; Coronary artery disease; Non-diabetic patients; Major adverse cardiovascular events

Core Tip: Patients without a formal diabetes diagnosis may still exhibit elevated glycated hemoglobin (HbA1c) levels in the prediabetic range (5.7%-6.4%), which may contribute to worse cardiovascular outcomes. This systematic review and meta-analysis examined the prognostic impact of elevated HbA1c in non-diabetic patients undergoing percutaneous coronary intervention (PCI). We found that HbA1c ≥ 5.7% is independently associated with increased long-term all-cause mortality and major adverse cardiovascular events. These findings highlight the potential value of HbA1c as a simple, cost-effective marker for post-PCI risk stratification and emphasize the need for early intervention in this high-risk group.

INTRODUCTION

Coronary artery disease (CAD) remains a leading cause of global morbidity and mortality, despite substantial advancements in both pharmacological and interventional strategies[1,2]. Since its inception in 1977, percutaneous coronary intervention (PCI) has become a cornerstone of coronary revascularization, offering improved survival and reduced procedural complications[3]. Glycated hemoglobin (HbA1c) is a widely used biomarker for the diagnosis and management of diabetes mellitus[4]. Among diabetic patients, elevated HbA1c levels have consistently been associated with adverse cardiovascular outcomes following PCI[5]. However, the prognostic value of HbA1c in individuals without diagnosed diabetes undergoing PCI remains unclear, with prior studies reporting inconsistent results[6,7].

Importantly, individuals without a formal diagnosis of diabetes may still present with elevated HbA1c levels within the range defined as prediabetes by international guidelines (5.7%-6.4%)[4]. This subclinical dysglycemia is linked to insulin resistance, endothelial dysfunction, and chronic inflammation-all of which may contribute to adverse cardiovascular events. Therefore, referring to these individuals solely as “non-diabetic” may obscure their underlying metabolic risk. A clearer distinction is needed to better understand outcomes in this metabolically vulnerable subgroup.

Previous meta-analyses, such as the one by Li et al[8], attempted to address this issue but were limited by smaller sample sizes, a narrower study population [e.g., acute myocardial infarction (AMI) patients], and lack of stratification by follow-up duration. Given these limitations and the growing body of new evidence, we conducted an updated and comprehensive meta-analysis to evaluate whether elevated HbA1c levels-specifically in the prediabetic range-are independently associated with adverse clinical outcomes in patients without diagnosed diabetes undergoing PCI. This study aimed to incorporate a broader, contemporary dataset, distinguish between short- and long-term outcomes, and provide more definitive insights into this clinically relevant question.

MATERIALS AND METHODS

This meta-analysis adhered to the PRISMA[9] and was prospectively registered on the international prospective register of systematic reviews (PROSPERO) (CRD420251041323).

Data sources and search strategy

A systematic search of studies listed in the PubMed, EMBASE and Cochrane Central databases from inception to April 2025 was conducted using the following search terms: ("Percutaneous Coronary Intervention" OR "PCI" OR "Coronary Angioplasty" OR "Coronary Revascularization" OR "Coronary Stenting" OR "Stent Placement" OR "Balloon Angioplasty" OR "Angioplasty, Balloon, Coronary") AND ("Hemoglobin A1c" OR "HbA1c" OR "Glycated Hemoglobin" OR "Glycosylated Hemoglobin" OR "Glycohemoglobin" OR "Hemoglobin A1*" OR "HbA1*" OR "A1c"). All references in the retrieved articles were also scanned to identify other potentially available studies. The detailed search strategy is provided in Supplementary Table 1.

Eligibility criteria and data extraction

Studies were considered eligible to be included in our systematic review and meta-analysis if they met the following inclusion criteria: (1) Enrolled patients without a prior diagnosis of diabetes with CAD undergoing PCI; (2) Evaluated elevated HbA1c levels (defined as > 5.7%) as the intervention group; (3) Included a control group with normal HbA1c levels (< 5.7%); (4) Reported at least one of the outcomes of interest; and (5) Provided raw data to allow estimation of risk ratios (RR) with 95% confidence intervals (CI).

We defined elevated HbA1c as ≥ 5.7%, corresponding to the internationally recognized threshold for prediabetes, in accordance with American Diabetes Association guidelines. This allowed us to investigate outcomes in patients within the prediabetic spectrum, despite the absence of a formal diabetes diagnosis. Patients without diagnosed diabetes were defined as individuals with no prior diagnosis of diabetes at the time of admission and no history of receiving anti-diabetic therapy. Exclusion criteria were: (1) Experimental studies; (2) Conference articles, case reports, systematic reviews and meta-analyses; (3) Insufficient outcome data provided; (4) Duplicate reports of literature research; and (5) Animal studies.

The following data were reviewed and extracted from each study: (1) Study details such as the first author's last name, year of publication, country, and sample size; (2) Baseline characteristics of the patient population, including age, gender, body mass index, smoking habits, presence of hypertension, dyslipidemia, and family history of CAD; (3) The timing of HbA1c measurement; (4) Clearly stated inclusion and exclusion criteria; (5) Potential for selection bias; (6) Follow-up completeness; (7) Reported adverse clinical outcomes; and (8) Whether confounding variables were adjusted for in multivariate analyses.

Definition of outcomes

The primary adverse clinical outcomes evaluated in our study were long-term all-cause mortality and major adverse cardiovascular events (MACEs), while the secondary outcomes were short-term all-cause and cardiac mortality. Long-term outcomes were defined as those with a follow-up period of at least one year, whereas short-term outcomes included events occurring during hospitalization or within 30 days post-procedure. MACEs encompassed all-cause mortality, non-fatal myocardial infarction, target lesion revascularization, target vessel revascularization, recurrent AMI, hospitalization for heart failure, and stent thrombosis. Cardiac mortality was defined as any death due to any cardiac cause.

Quality assessment

Two independent reviewers (Shahid MH and Raja HAA) evaluated the risk of bias using the Newcastle-Ottawa Scale (NOS) for observational studies. Observational studies were assessed using the NOS, which evaluates methodological quality based on three broad perspectives: (1) Selection of study groups; (2) Comparability of groups; and (3) Ascertainment of either the exposure or outcome of interest. Studies were rated According to the number of stars awarded in each domain, with higher scores indicating better methodological quality. Any disagreements were resolved through discussion with a third reviewer (Shahid S). Publication bias was evaluated through visual inspection of funnel plots and quantified using the Luis Furuya-Kanamori (LFK) asymmetry index.

Statistical analysis

We used R software (version 4.3.2) to perform data synthesis for all reported outcomes. A P value of < 0.05 was considered statistically significant. To evaluate heterogeneity among the included studies, we applied the Cochran’s Q test and the I² statistic[10]. The random-effects model (Der Simonian and Laird method) for calculating the pooled RR and 95%CI was used. Significant heterogeneity was defined as a P value < 0.05 for Cochran’s Q or an I² value greater than 50%. Sensitivity analyses were carried out by sequentially excluding individual studies to evaluate the robustness of the findings.

RESULTS

Study and patient characteristics

The preliminary search yielded 2324 articles. After removing duplicates, 1944 articles remained for title and abstract screening. Of these, 1798 were excluded based on irrelevance or failure to meet the inclusion criteria. The remaining 140 articles were subjected to full-text review to assess eligibility. Ultimately, ten studies comprising a total of 32403 patients met the inclusion criteria and were included in the final analysis[11-20]. The process of identifying and selecting relevant studies is illustrated in the flow chart (Figure 1).

Figure 1 PRISMA flow chart of the screening process.

Eight of the included ten studies reported long-term all-cause mortality[11-16,19,20], five studies had long-term MACE outcomes at follow-up[14-16,18,19], three of the ten studies investigated short-term all-cause mortality[11,15,16] and two studies assessed cardiac mortality[19,20]. Among the ten studies, five were from Asia (three from China, one from Japan, and one from South Korea), two from the Netherlands, one from Germany, and one from the United States. Baseline characteristics of the included studies are shown in Table 1.

Table 1 Baseline characteristics of included studies.

Ref.	Country	Sample size	Mean age in years (MD)	Males %	Normal HbA1c	Abnormal HbA1c	Follow up	Cardiac diagnosis	Study outcomes
Aggarwal et al[13]	United States	1163 (652 vs 511)	NR	NR	< 5.7	5.7 < HbA1c < 6.5	3 years	STEMI	All-cause deaths
Chen et al[15]	China	267 (64 vs 263)	NR	83.5	< 6	6.0 < HbA1c < 6.5	178 days	STEMI/NSTEMI	All-cause deaths, repeated PCI, recurrent AMI, HF requiring hospitalization
Dykun et al[17]	Germany	1692 (1188 vs 510)	66.1 (11.4)	77.1	6.3 (1.2)	≥ 7.8%	2.8 years	STEMI/NSTEMI	association of HbA1c with ACM in unadjusted and multivariable adjusted modeling
Kok et al[16]	Netherland	2103 (234 vs 1869)	NR	NR	< 6	6.0 < HbA1c < 6.5	1 year	Obstructive coronary disease	All-cause deaths, MI, revascularization, and stent thrombosis)
Naito et al[12]	Japan	452 (231 vs 221)	NR	NR	< 5.7	5.7 < HbA1c < 6.5	4.7 years	ACS	Primary endpoint was a composite of all-cause death and non-fatal MI
Park et al[18]	South Korea	1318 (120 vs 1198)	64.8 (8.9) vs 60.6 (11.3)	48.02	< 5.7	6.5 ≤ HbA1c < 7.0	1.83 years	CAD	Survival free from MACE
Shin et al[14]	South Korea	2470 (1475 vs 995)	NR	NR	< 5.7	5.7 < HbA1c < 6.5	1 year	STEMI	All-cause deaths, MACEs (all-cause mortality, non-fatal MI, TLR, TVR)
Timmer et al[11]	Netherland	4176 (1024 vs 3152)	62 (13) vs 63 (13)	74	< 5.8	> 5.8	3.3 ± 1.5 years	STEMI	30-day mortality, 1-year mortality, and infarct size
Wen et al[19]	China	7033 (3530 vs 3503)	NR	74.3	< -0.506	≥ 0.179	2 years	ACS + SCAD	Mortality, ACM, CM, MACEs, MACCEs
Zhou et al[20]	China	11729 (4976 vs 6753)	51.78 (6.97) vs 49.07 (7.21)	71.9	< 5	> 6	4.62 years	Premature CAD	Long-term all-cause mortality, cardiac mortality

PCI: Percutaneous coronary intervention; ACM: All-cause mortality; CM: Cardiac mortality; MACE: Major adverse cardiac event; MI: Myocardial infarction; MACCE: Major adverse cardiac and cerebrovascular events; NR: Not reported; TLR: Target lesion revascularization; NSTEMI: Non-ST segment elevation myocardial infarction; STEMI: ST-segment elevation myocardial infarction; ACS: Acute coronary syndrome; CAD: Coronary artery disease; SCAD: Stable coronary artery disease; HbA1c: Glycated hemoglobin.

Quality appraisal and publication bias assessment

Risk of bias was generally low across studies based on the NOS, with most scoring 8 out of 9 stars. Two studies[12,19] scored 9, reflecting high methodological quality. One study[20] scored 7, indicating a slightly higher risk of bias in at least one domain (Supplementary Table 2). Publication bias was assessed using funnel plots and quantified via the LFK asymmetry index. For short-term all-cause mortality (LFK = -0.35) and MACE (LFK = 0.62), funnel plots showed minimal asymmetry, indicating low risk of publication bias. However, the analysis for long-term all-cause mortality (LFK = 3.62) revealed significant asymmetry, suggesting potential publication bias. This may reflect under-representation of small neutral/negative studies in the literature (Supplementary Figures 1-4).

Outcomes of interest

Long-term all-cause mortality: The outcome of long-term all-cause mortality was reported by 8 out of 10 selected studies with a total of 20503 patients[11-16,19,20]. The pooled analysis demonstrated that abnormal HbA1c was associated with increased risk of long-term all-cause deaths compared to those with normal HbA1c levels (RR: 1.30; 95%CI: 1.10-1.54; P value < 0.01; I² = 41%) (Figure 2A). There was a moderate statistical heterogeneity among the studies. We performed a sensitivity analysis by excluding each study in turn, with no significant impact on heterogeneity (Supplementary Figure 5).

Figure 2 Forest plots comparing clinical outcomes between patients with abnormal and normal glycated hemoglobin levels. A: Long-term all-cause mortality with subgroup analysis; B: Major adverse cardiovascular events with subgroup analysis; C: Short-term all-cause mortality; D: Cardiac mortality.

MACE: The outcome of MACE was reported by 5 out of 10 selected studies with a total of 13281 patients[14-16,18,19]. The pooled analysis demonstrated that abnormal HbA1c was associated with an increased risk of MACEs compared to those with normal HbA1c levels (RR: 1.31; 95%CI: 1.01-1.69; P value = 0.04; I² = 61%) (Figure 2B). There was a marked heterogeneity in the data (I² = 61%). We conducted a sensitivity analysis by excluding each study in turn for heterogeneity. We found that the marked heterogeneity originated from Kok et al's study[16], likely due to the HbA1c cut-off levels of 6.0% to 6.5% used in their analysis (Supplementary Figure 6).

Short-term all-cause mortality: The outcome of short-term all-cause mortality was reported by 3 out of 10 selected studies with a total of 7809 patients[11,15,16]. The pooled analysis demonstrated that abnormal HbA1c were associated with increased risk of short-term all-cause deaths compared to those with normal HbA1c levels. However, this association was not statistically significant (RR: 1.16; 95%CI: 0.88-1.53; P value = 0.29; I² = 1%) (Figure 2C). There was no significant statistical heterogeneity observed among the included studies. Sensitivity analysis confirmed the robustness of the findings, as exclusion of individual studies did not substantially alter the overall effect estimates or heterogeneity levels (Supplementary Figure 7).

Cardiac mortality: Cardiac mortality was reported by 2 out of the 10 selected studies[19,20]. The pooled analysis showed that abnormal HbA1c was associated with a higher risk of cardiac mortality compared to normal HbA1c levels; however, this association did not reach statistical significance (RR: 1.76; 95%CI: 0.85-3.67; P value = 0.13; I² = 94%) (Figure 2D). Considerable heterogeneity was observed among the included studies. Sensitivity analysis demonstrated that excluding individual studies did not significantly change the overall results or reduce heterogeneity (Supplementary Figure 8).

Subgroup-analysis

Subgroup analysis was performed based on country of origin. It demonstrated that, elevated admission HbA1c levels were significantly associated with an increased risk of long-term all-cause mortality (RR 1.77, 95%CI: 1.32-2.30, P < 0.01) and MACE (RR 1.94, 95%CI: 1.36-2.76, P < 0.01) in patients from the Netherlands. In contrast, no statistically significant associations were observed in other countries, including China, South Korea, Germany, the United States, and Japan, suggesting possible geographic variation in the prognostic impact of abnormal HbA1c levels (Figure 2A and B). Subgroup analysis was not performed for the outcomes of short term all-cause and cardiac mortality due to an insufficient number of studies, with fewer than two studies available per subgroup.

DISCUSSION

This systematic review and meta-analysis indicate that elevated HbA1c levels (≥ 5.7%) are substantially correlated with an increased risk of long-term all-cause mortality and MACE in non-diabetic patients receiving PCI. Short-term all-cause and cardiac mortality risks were raised, although not statistically significant. Small sample sizes (three studies for short-term mortality and two for cardiac death), short follow-up periods, and few events may explain this. These limitations diminish statistical power and may obscure a true association, necessitating validation in larger prospective cohorts. The findings suggest that even modest increases in HbA1c below the diabetic threshold may reflect underlying metabolic or vascular dysfunction that predisposes patients to adverse long-term cardiovascular outcomes.

Our results validate and expand upon the findings of Li et al[8], who first identified the prognostic relevance of increased HbA1c levels in non-diabetic individuals after PCI. Nonetheless, their analysis was limited by a reduced sample size (n = 8385), a lower number of included studies (n = 5), and a primary emphasis on AMI patients, without differentiating outcomes based on follow-up time. Conversely, our research included three supplementary trials and included 12118 more patients, culminating in a combined cohort of 20503 people for long-term mortality assessment and 13281 for MACE. We investigated outcomes across both short- and long-term durations and incorporated cardiac death as an additional endpoint, which was not addressed in the previous meta-analysis. Our dataset encompasses a diverse group of CAD individuals with differing manifestations, hence enhancing the generalizability of our findings.

Furthermore, the results align with individual studies conducted by Aggarwal et al[13] and Geng et al[21], both of which recognized HbA1c as an independent predictor of worse cardiovascular outcomes in non-diabetic patients following PCI. Xu et al[5] also documented an increase in in-hospital and long-term mortality linked to higher HbA1c levels in non-diabetic patients having PCI for ST-segment elevation myocardial infarction (STEMI). In contrast, Cicek et al[22] identified no independent correlation between HbA1c and significant poor outcomes; however, their results may be constrained by a smaller sample size, a predominance of stable CAD patients, a shorter follow-up period, and variations in the scheduling of HbA1c evaluation.

A significant aspect of our study is the inclusion of patients from various geographic locations, especially from China, which was lacking in the previous meta-analysis by Li et al[8]. Subgroup analysis indicated a statistically significant correlation between elevated HbA1c levels and adverse outcomes in Dutch patients-specifically regarding long-term mortality (RR: 1.77, 95%CI: 1.32-2.30, P < 0.01) and MACE (RR: 1.94, 95%CI: 1.36-2.76, P < 0.01)-but not in other nations such as China, South Korea, the United States, Germany, or Japan, implying possible regional disparities in the prognostic significance of HbA1c. Future subgroup analyses stratified by clinical presentation (e.g., STEMI vs non-ST segment elevation myocardial infarction vs stable CAD), baseline cardiovascular risk profiles, or stent type (DES vs BMS) could further refine these geographic differences and uncover patient-specific modifiers of risk.

Sensitivity analyses corroborated the robustness of our results. Sequential elimination of individual studies did not significantly modify impact estimates for both long-term mortality and MACE. The heterogeneity in the MACE analysis (I² = 61%) was mostly affected by the study done by Kok et al[16], which utilised a broader HbA1c range (6.0%-6.5%) in contrast to the standardised criteria of ≥ 5.7% applied in other segments of our analysis. The elimination of this study led to a decrease in heterogeneity, hence confirming its designation as an outlier. Conversely, minimal heterogeneity was noted for short-term mortality (I² = 1%) and moderate heterogeneity for long-term mortality (I² = 41%), highlighting the uniformity of results across the majority of included studies.

Nonetheless, additional sources of heterogeneity should be considered. Variability in HbA1c cutoff values across studies may reflect differing definitions of prediabetes, which could influence the pooled effect sizes. Similarly, inconsistency in follow-up durations might affect the detection and timing of outcomes, thereby contributing to between-study variability. While we explored heterogeneity through sensitivity analyses, further subgroup analyses stratified by HbA1c thresholds or follow-up duration were limited by the availability and consistency of data across included studies. Future meta-analyses with access to individual participant data may help better delineate these sources and strengthen the validity of findings.

Several molecular pathways may link high HbA1c to unfavourable cardiovascular events in non-diabetics. Endothelial dysfunction due to reduced nitric oxide bioavailability, low-grade chronic inflammation, oxidative stress, and enhanced platelet reactivity are likely contributors. These mechanisms promote atherogenesis, thrombogenicity, and vascular injury, even in the absence of overt diabetes. Periodic HbA1c monitoring in non-diabetic PCI patients may help stratify risk. Integrating HbA1c into existing risk scoring systems, such as the SYNTAX score, could improve their predictive performance. A ≥ 5.7% threshold might serve as a target for early intervention in this population.

Several limitations of this meta-analysis must be acknowledged. First, the majority of included studies were observational in nature, which inherently increases the risk of residual confounding despite multivariable adjustments. Importantly, key confounders such as markers of inflammation (e.g., C-reactive protein, interleukin-6) and insulin resistance indices (e.g., homeostatic model assessment for insulin resistance) were not consistently adjusted for, potentially affecting the interpretation of HbA1c as an independent predictor. Procedural aspects such as PCI techniques, stent technology, and institutional protocols varied across studies and timeframes, potentially affecting comparability. In addition, the use and reporting of adjunct pharmacotherapies-particularly antiplatelet and anticoagulant regimens-were inconsistent and may have influenced clinical outcomes. Furthermore, discrepancies in HbA1c thresholds, timing of measurements, duration of follow-up, and outcome definitions likely contributed to inter-study variability. The limited number of studies reporting short-term mortality (n = 3) and cardiac mortality (n = 2) may have reduced statistical power and the reliability of subgroup analyses for these endpoints. Lastly, evidence of publication bias was noted in the analysis of long-term all-cause mortality, as indicated by funnel plot asymmetry and a LFK index of 3.62, suggesting possible underreporting of smaller studies with null or negative findings. In contrast, analyses of MACE (LFK = 0.62) and short-term mortality (LFK = -0.35) demonstrated minor asymmetry, indicating a lower likelihood of publication bias for these outcomes.

This meta-analysis highlights that HbA1c levels within the non-diabetic range (≥ 5.7%) may serve as a valuable biomarker for cardiovascular risk stratification following PCI. Given its affordability, widespread availability, and prognostic value, incorporating HbA1c into routine follow-up protocols could help identify high-risk individuals who may benefit from targeted interventions. These could include structured lifestyle programs, closer monitoring, and potentially early pharmacological therapy to improve metabolic health. The frequency of HbA1c monitoring and the optimal threshold for triggering interventions should be evaluated in future research.

Future research should also explore several directions: (1) Prospective studies measuring serial HbA1c changes post-PCI; (2) Randomized controlled trials targeting pre-diabetic HbA1c levels to assess benefit from early intervention; and (3) Mechanistic studies investigating HbA1c's role in endothelial dysfunction, plaque instability, and thrombosis risk.

CONCLUSION

In conclusion, increased HbA1c levels, defined as ≥ 5.7%, correlate with worse long-term results in non-diabetic CAD patients after PCI. Regular evaluation of HbA1c may provide a straightforward and efficient method for risk stratification, even in non-diabetic patients. Future prospective trials are warranted to ascertain if targeted therapies in individuals with increased HbA1c might enhance cardiovascular outcomes.