Predictors of no-reflow phenomenon in patients with type 2 diabetes mellitus undergoing elective percutaneous coronary intervention: the emerging role of the triglyceride–glucose index

Introduction

Percutaneous coronary intervention (PCI) has become the cornerstone of coronary revascularisation in patients with obstructive coronary artery disease and is associated with significant improvements in myocardial perfusion, symptom burden and cardiovascular outcomes [1,2]. Despite successful restoration of epicardial coronary artery patency, a subset of patients experience inadequate myocardial tissue perfusion, a condition referred to as the no-reflow phenomenon [3]. This complication is characterised by impaired myocardial reperfusion in the absence of mechanical (epicardial) coronary obstruction and reflects obstruction at the microvascular rather than the epicardial level. It is associated with larger infarct size, impaired ventricular function, heart failure, arrhythmias and increased mortality [4]. The phenomenon has been studied predominantly in the context of primary PCI for acute coronary syndromes, particularly ST-elevation myocardial infarction, where thrombus burden and ischaemia–reperfusion injury are greatest. Its frequency, predictors and clinical relevance in elective PCI for stable coronary artery disease—the setting of the present study—are less well characterised, and the two contexts should not be conflated.

The pathophysiology of no-reflow is multifactorial and involves distal embolization, ischaemia-reperfusion injury, endothelial dysfunction, inflammatory activation, oxidative stress, platelet aggregation and coronary microvascular obstruction [5]. Patients with type 2 diabetes mellitus (T2DM) are particularly vulnerable because of diffuse atherosclerosis, chronic inflammation, endothelial dysfunction, insulin resistance and impaired coronary microvascular integrity [6].

T2DM represents one of the most important global cardiovascular risk factors and is associated with increased morbidity and mortality [7,8,9]. Persistent hyperglycaemia and insulin resistance contribute to vascular inflammation, oxidative stress and endothelial injury, all of which may predispose to impaired myocardial reperfusion following PCI [10].

The triglyceride–glucose (TyG) index, derived from fasting triglyceride and fasting glucose levels, has emerged as a simple and reliable surrogate marker of insulin resistance [11]. Recent studies have demonstrated associations between elevated TyG index and adverse cardiovascular outcomes, coronary artery disease severity and microvascular dysfunction [12]. However, limited data are available regarding its role in predicting no-reflow in diabetic patients undergoing elective PCI [13].

Therefore, the present study aimed to evaluate the prevalence of the no-reflow phenomenon in patients with T2DM undergoing elective PCI and identify clinical, laboratory, angiographic and procedural predictors of no-reflow, with particular emphasis on the predictive value of the TyG index.

Patients and methods

Results

2. Laboratory and echocardiographic findings

Patients with no-reflow exhibited significantly higher random blood glucose, fasting blood glucose, HbA1c, low-density lipoprotein cholesterol, triglycerides and TyG index. In contrast, LVEF was significantly lower in the no-reflow group (Table 2, Figure 1).

Table 2

Haemodynamic parameters, laboratory findings and cardiac function according to reflow status

		Reflow (n = 124)	No-reflow (n = 13)	p-value
HR	Mean ± SD	87.40 ± 11.55	88.38 ± 12.36	0.788
Systolic BP	Mean ± SD	132.83 ± 12.15	134.62 ± 15.87	0.700
Diastolic BP	Mean ± SD	86.65 ± 10.53	88.08 ± 10.91	0.660
SpO2	Mean ± SD	98.63 ± 0.49	98.69 ± 0.48	0.658
RBG (mg/dL)	Mean ± SD	170.23 ± 25.73	198.85 ± 34.39	0.012*
Serum creatinine (mg/dL)	Mean ± SD	0.99 ± 0.32	1.00 ± 0.18	0.849
CrCl (mL/min)	Mean ± SD	129.10 ± 11.98	129.00 ± 15.04	0.982
Hb (g/dL)	Mean ± SD	13.30 ± 0.81	12.56 ± 1.38	0.080
HbA1c (%)	Mean ± SD	7.19 ± 0.71	8.59 ± 1.71	0.012*
Cholesterol (mg/dL)	Mean ± SD	178.48 ± 19.51	181.92 ± 22.51	0.604
LDL (mg/dL)	Mean ± SD	126.00 ± 17.27	151.62 ± 35.91	0.025*
HDL (mg/dL)	Mean ± SD	37.10 ± 4.25	36.83 ± 3.62	0.807
TGs (mg/dL)	Mean ± SD	145.27 ± 20.46	167.31 ± 21.97	0.004*
FBG (mg/dL)	Mean ± SD	96.98 ± 12.35	124.77 ± 34.62	0.014*
TyG index	Mean ± SD	8.84 ± 0.20	9.21 ± 0.26	<0.001*
Ejection fraction (%)	Mean ± SD	50.91 ± 6.11	43.77 ± 4.53	<0.001*

BP, blood pressure; CrCl, creatinine clearance; FBG, fasting blood glucose; Hb, haemoglobin; HbA1c, glycated haemoglobin; HDL, high-density lipoprotein; HR, heart rate; LDL, low-density lipoprotein; RBG, random blood glucose

* Significant

Figure 1

4. Predictive performance of the TyG index

ROC analysis demonstrated that the TyG index had the highest predictive performance for no-reflow among all studied metabolic markers (AUC 0.870, 95% confidence interval [CI] 0.753–0.987, p < 0.001). The optimal cutoff by the Youden index was 9.01, yielding a sensitivity of 76.9% and a specificity of 89.5%. Because the cutoff was derived and evaluated within the same cohort, internal validation was performed using 2000 bootstrap resamples: the bootstrap 95% CI for the AUC was 0.738–0.977, and the optimism-corrected AUC was 0.870, indicating negligible optimism—as expected for a single continuous marker. The bootstrap 95% CI for the optimal cutoff was 8.95–9.20, supporting its stability; nonetheless, the sensitivity and specificity at this cutoff are internal estimates and require external validation (Table 4, Figures 2 and 3).

Table 4

ROC analysis of metabolic laboratory parameters for predicting the no-reflow phenomenon

Parameter	AUC	SE	p-value	95% CI
TyG index	0.870	0.059	<0.001	0.753–0.987
RBG (mg/dL)	0.761	0.067	0.002	0.629–0.893
HbA1c	0.770	0.088	0.001	0.597–0.943
Cholesterol (mg/dL)	0.564	0.093	0.447	0.382–0.746
LDL (mg/dL)	0.755	0.105	0.003	0.550–0.960
HDL (mg/dL)	0.500	0.078	0.997	0.348–0.653
TGs (mg/dL)	0.772	0.078	0.001	0.619–0.925
FBG (mg/dL)	0.808	0.069	<0.001	0.673–0.943

[i] AUC, area under the curve; CI, confidence interval; FBG, fasting blood glucose; HbA1c, glycated haemoglobin; HDL, high-density lipoprotein; LDL, low-density lipoprotein; RBG, random blood glucose; ROC, receiver operating characteristic; SE, standard error.

Figure 2

Figure 3

Discussion

In this cohort of diabetic patients undergoing elective PCI, noreflow occurred in approximately 1 in 10 patients and was associated with poorer glycaemic control, an adverse lipid profile, longer diabetes duration, lower ejection fraction and more severe coronary stenosis. Among all metabolic markers examined, the TyG index showed the strongest discriminatory and independent association with no-reflow.

The mechanisms of no-reflow—distal embolization, ischaemia–reperfusion injury, endothelial and microvascular dysfunction, inflammation and platelet activation—are well described, predominantly in the setting of primary PCI for STEMI [3,4,5]. That context differs materially from elective PCI for stable disease, where thrombus burden and ischaemic time are lower; findings from the STEMI literature are, therefore, cited here only for mechanistic comparison and should not be assumed to transfer directly to the elective setting examined in this study. Elective PCI for stable coronary disease is increasingly performed as a routine, often ambulatory, procedure [16], so anticipating even relatively infrequent complications such as no-reflow remains clinically important.

Because prevalence is our primary outcome, the observed incidence warrants direct comparison. Our rate of 9.5% is higher than that reported in other elective PCI series—2.4% by Jeon et al. [13] in non-acute patients and 5% by Waqar et al. [17] in stable coronary disease—and lower than rates in primary PCI STEMI series (Omar et al. [18] 11.1%; Elrayes et al. [19] 14.6%). The higher rate relative to other elective cohorts is consistent with our exclusively diabetic, higher-risk population, in whom diffuse atherosclerosis and microvascular vulnerability are more pronounced. This intermediate position, between unselected elective and acute STEMI populations, is the expected pattern for an all-diabetic elective cohort.

Metabolic status was strongly linked to no-reflow: affected patients had significantly higher fasting and random glucose, HbA1c, LDL cholesterol, triglycerides and TyG index, and a lower ejection fraction. These associations align with Qu and Guan [20] and with Zhao et al. [6], who likewise reported worse glycaemic control and more severe stenosis in patients developing no-reflow, and with Eldamanhory et al. [5], who linked elevated glucose to impaired systolic function.

Among procedural factors, severe stenosis was significantly more frequent in the no-reflow group, whereas stent length (median 30 mm vs 24 mm, p = 0.292) and stent diameter (p = 0.626) were not associated with no-reflow. This differs from Jeon et al. [13], who reported longer lesion length among no-reflow patients, and indicates that in our cohort, lesion severity, rather than stent geometry, drove the association. All no-reflow events occurred in pre-dilated patients; however, because pre-dilatation is preferentially performed on more complex, severe lesions, it is best interpreted as a marker of lesion complexity rather than an independent contributor, and its complete association with the outcome precluded formal modelling.

The central finding is the performance of the TyG index. It integrates glycaemic and lipid abnormalities into a single fasting-derived surrogate of insulin resistance, a plausible driver of endothelial and microvascular dysfunction. In our data it showed the highest discrimination among metabolic markers (AUC 0.870) and remained independently associated with no-reflow in the parsimonious model (OR 2.05 per 0.1-unit increase, 95% CI 1.26–3.32, p = 0.004), alongside diabetes duration and ejection fraction. This is directionally concordant with prior reports: Yang et al. [21] (STEMI-diabetic) found the TyG index an independent predictor (adjusted OR 2.98), Ma et al. [11] reported an OR of 3.23 in diabetic PCI and Altunova et al. [12] described good discrimination (AUC 0.821). A meta-analysis by Fajar et al. [22] similarly identified diabetes, hyperglycaemia, low ejection fraction and multivessel disease as predictors of no-reflow. The consistency of direction across differing clinical settings supports biological plausibility, while the differences in effect size underline the need for setting-specific validation.

Study limitations

Several limitations should be acknowledged. First, this was a single-centre observational study with a modest sample size, limiting generalisability; moreover, the sample-size estimate was based on a STEMI-derived incidence because elective-PCI diabetic data were unavailable at design, and the lower observed event rate (9.5%) reduced the statistical power for multivariable modelling. Second, only 13 no-reflow events occurred, yielding a low events-per-variable ratio; the multivariable model should, therefore, be regarded as exploratory and hypothesis-generating, and the identified associations require confirmation in larger cohorts. Third, the optimal TyG cutoff was both derived and evaluated within the same cohort; although bootstrap resampling indicated a stable cutoff and negligible optimism in the AUC, the cutoff-specific sensitivity and specificity may be optimistic and require external validation. Fourth, no-reflow was defined using angiographic TIMI flow grade alone; more sensitive indices of microvascular perfusion—myocardial blush grade and the corrected TIMI frame count—were not assessed or were cardiac magnetic resonance or myocardial contrast echocardiography [23], so subclinical microvascular dysfunction may have been underestimated. Fifth, only intracoronary nitroglycerine (Tridil) was administered when impaired flow was encountered; because nitrates act predominantly on epicardial vessels, persistence of impaired flow argues against epicardial spasm but does not exclude a potentially reversible microvascular component that dedicated agents (adenosine, verapamil, sodium nitroprusside) might have addressed. Sixth, stenosis severity and TIMI flow were assessed by visual estimation rather than quantitative coronary angiography or core-laboratory adjudication, introducing potential inter-observer variability. Direct measures of insulin resistance, such as HOMA-IR, were not collected, precluding head-to-head comparison with the TyG index. Finally, long-term clinical outcomes were not evaluated, and residual confounding cannot be excluded; causality cannot be inferred from these observational data.

Future multicentre studies with larger populations and longer follow-up are warranted to validate these findings and determine whether interventions targeting insulin resistance and metabolic optimisation can reduce no-reflow and improve outcomes.

Conclusion

No-reflow occurred in approximately 1 in 10 patients with type 2 diabetes undergoing elective PCI and was associated with longer diabetes duration, poorer glycaemic control, an adverse lipid profile, reduced ejection fraction and severe coronary stenosis. Among all studied biomarkers, the triglyceride–glucose index demonstrated the highest predictive accuracy and remained an independent predictor of no-reflow after multivariable adjustment. The TyG index may, therefore, represent a simple, inexpensive and clinically useful tool for preprocedural risk stratification in diabetic patients undergoing elective PCI. Further large-scale multicentre studies are recommended to validate these findings and determine whether targeted metabolic interventions can improve myocardial reperfusion and long-term cardiovascular outcomes.