Long-term impact of diabetes on mortality in patients undergoing unprotected left main PCI: a propensity score-matched analysis from the BIA-LM registry

Abstract

Background

There is modest data on long-term impact of diabetes on left main coronary artery (LMCA) percutaneous coronary intervention (PCI). This observational study, based on the largest single-center registry of LMCA PCI in Poland, evaluated the impact of diabetes on long-term survival following PCI in a real-world setting.

Methods

We retrospectively analyzed 998 patients who underwent LMCA PCI between December 27, 2007, and February 21, 2022. Diabetes and insulin dependence were defined based on medical history, prior records, and prescribed treatment. The endpoint was all-cause mortality at the longest available follow-up (mean 4.2 years). Survival analysis was conducted in the overall cohort and a one-to-one propensity score-matched (PSM) population. Moderation effects and differences between subgroups were analysed in predefined groups of PSM cohort.

Results

The median age was 71 (63–79) years; 212 (28.2%) were women, and 250 (33.2%) had diabetes. In the overall cohort, diabetes was associated with a worse prognosis (HR 1.35, 95% CI 1.03–1.76, P = 0.03). PSM resulted in 214 well-balanced pairs (median age 73 years (66–79)), with no significant difference in all-cause mortality between groups (adjusted HR 1.27, 95% CI 0.91–1.77, P = 0.16). After PSM subgroup analysis showed worse outcomes for patients with diabetes undergoing two-stent angioplasty (HR 3.70, 95% CI 1.64–8.34, P = 0.002) and elective PCI (HR 2.07, 95% CI 1.29–3.31, P = 0.003). Conversely, among patients presenting with myocardial infarction (MI), people with diabetes had better survival than the control group (HR 0.56, 95% CI 0.35–0.90, P = 0.02). No significant differences in outcomes were observed in patients with heart failure (HR 1.29, 95% CI 0.88–1.89, P = 0.19), chronic kidney disease (HR 1.08, 95% CI 0.69–1.71, P = 0.19), intravascular imaging use (HR 1.38, 95% CI 0.70–2.71, P = 0.35), or concomitant multivessel disease (HR 1.14, 95% CI 0.75–1.73, P = 0.53).

Conclusions

No significant association was observed between diabetes and overall mortality following LMCA PCI. Sensitivity analyses showed worse survival outcomes in diabetic patients treated with two-stent techniques and undergoing non-emergency PCI. These findings suggest the overall safety of PCI for LMCA in people with diabetes and highlight the need for randomized trials, especially investigating indicated high-risk subgroups.

Graphic abstract

Background

Ischaemic heart disease (IHD) accounts for 13% of all global deaths, making it the leading cause of mortality worldwide, according to the latest WHO data [1]. Notably, patients with diabetes constitute a particularly vulnerable subgroup, as they are at an increased risk of both developing and experiencing worse outcomes from cardiovascular disease [2]. Given the global annual increase in diabetes prevalence of 1.56% with further rises anticipated [3], it is imperative to evaluate this population and establish standardised treatment guidelines rigorously.

One of the most dangerous occlusion sites is the left main coronary artery (LMCA), as its obstruction poses an exceptionally high risk due to its critical role in myocardial perfusion, supplying between 75% and 100% of the left ventricular myocardium, depending on the coronary artery dominance. Significant stenosis or occlusion of the LMCA can precipitate extensive myocardial ischaemia and acute haemodynamic compromise. The prognostic implications of left main coronary artery disease (LMCAD) are reflected in historical studies demonstrating a five-year mortality rate exceeding 50% in patients managed with medical therapy alone, underscoring the necessity of timely revascularisation [4].

Moreover, the anatomical distribution of coronary artery disease (CAD) in diabetic patients differs significantly from that of non-diabetic individuals. This distinction may influence both the selection of an optimal revascularisation strategy and long-term prognosis [5]. Given these unique considerations, determining the most effective approach for managing LMCAD in diabetic patients remains a clinical priority.

The optimal revascularisation strategy for LMCAD has been the subject of longstanding debate. Coronary artery bypass grafting (CABG) has historically been considered the standard of care due to its established safety profile and durable long-term outcomes. However, advances in percutaneous coronary intervention (PCI) have increasingly positioned it as a viable alternative in selected patients. Notably, patients with diabetes constitute a particularly complex subgroup, representing up to 35% of those undergoing elective PCI [6]. If long-term PCI outcomes prove equivalent to those of CABG, PCI may emerge as the preferred strategy, particularly for patients requiring expedited functional recovery [7].

Despite extensive evidence comparing PCI and CABG, the impact of diabetes on long-term mortality in patients undergoing PCI for LMCAD remains unclear. To address this gap, our study aimed to analyse the largest single-center registry of LMCA percutaneous revascularisations in Poland, providing a comprehensive assessment of the relationship between diabetes and long-term survival outcomes following PCI in a real-world setting.

Methods

Study population

The design and study population of the BIA-LM Registry was described before [6]. For the needs of the current analysis, out of the 998 patients undergoing LMCA PCI from December 27, 2007 to February 21, 2022, primarily 815 patients with unprotected LMCA were included. After the exclusion of patients with no information on concomitant DM, final cohort of 752 patients was analysed. A detailed study flowchart is presented in Fig. 1.

Fig. 1

Flowchart of the study. CABG coronary artery bypass grafting; DM diabetes mellitus; LMCA left main coronary artery; PCI percutaneous coronary intervention; PS propensity score

The study was approved by the Bioethics Committee of the Medical University of Białystok, Poland (approval no. APK.002.78.2022 obtained on 10.02.2022), adheres to Helsinki Declaration as revised in 2013 and the STROBE guidelines (Supplemental Table 1).

Definitions and outcome

The diagnosis of co-morbidities was based on the anamnesis, previous reports, and ECGs performed during the hospitalization, as well as on on concomitant treatment and laboratory tests for seleceted diseases. Diagnosis of diabetes and insulin-dependent diabetes was based on the anamnesis, previous reports and prescribed treatment: oral hypoglycemic medications or insulin.

Endpoint was survival at the longest available follow-up, which was obtained for all of the patients from Centre for Information Technology, Minister of Digital Affairs, Poland and is valid as for 13.06.2022.

Statistical analyses

Baseline characteristics were presented as counts (N) and percentages (%) for categorical variables. For continuous variables with a normal distribution, the mean and standard deviation (SD) were reported, while those with a non-normal distribution were summarized using the median and interquartile range (IQR). Categorical variables were analysed using the chi-square test, or Fisher’s exact test when the expected frequency was less than 5. Continuous variables were compared using either Student’s t-test or the Wilcoxon rank-sum test, depending on the data distribution.

Mortality analysis was performed in the overall and propensity score (PS) matched population. To balance for confounders, firstly LASSO cross validation was performed to determine mortality predictors [8]. Then, all significant variables identified in multivariable Cox proportional hazard regressions were included in a one-to-one nearest neighbor matching with replacement (caliper 0.2) [8].

Differences in survival were assessed using Cox proportional hazard regression models in the unmatched and matched cohorts. Further moderation effects and differences between subgroups were analysed in predefined groups of PS-matched cohort. The proportionality assumption of the Cox model was calculated and assessed visually based on Schoenfeld residuals [8]. Results are presented as hazard ratio (HR) with a 95% confidence interval (CI).

For all analyses, the level of statistical significance was set at P < 0.05. Missing values for baseline variables were not imputed using any methods.

All statistical analysis was performed using StataNow/SE version 18.5 for Mac (StataCorp. 2023. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC.).

Results

Overall population

The final cohort consisted of 752 patients, aged 71 years (63–79) of which 212 (28.2%) were women. Chronic coronary syndrome was the most prevalent indication for PCI (68.7%). There was a total of 250 (33.2%) patients diagnosed with diabetes, including 59 (7.9%) treated with insulin. During a mean follow-up of 4.2, years there were 402 deaths (40.8%). Results of LASSO cross-validation and Cox multivariable analysis are presented in Supplementary Fig. 1 and Table 2.

Patients with DM were significantly older [73 (66–79) vs. 70.5 (61–80), P = 0.02], had higher BMI, more prevalent CKD, history of previous PCI or stroke and were considered higher risk assessed by EuroSCORE II. People with diabetes had more commonly multivessel disease and required modification of calcified plaque by rotational atherectomy. Patients without diabetes were more likely to receive a larger diameter of stent implanted to LM. There were no significant differences regarding peri-procedural complications or post-PCI prescription of potent P2Y₁₂ inhibitors. Detailed baseline characteristics are presented in Tables 1 and 2.

At 5 years, diabetes was associated with 35% higher risk of death (HR 1.35, 95% Cl 1.03–1.76, P = 0.03).

Propensity score matching

One-to-one propensity score (PS) matching resulted in 214 pairs of patients with and without concomitant diabetes aged 73 (66–80). There were no significant differences in baseline characteristics between the two groups, except for body mass index. Procedural data analysis showed a reduction in most imbalances, with significant disparities regarding higher mean LMCA stent diameter and complication rates in non-diabetics. Detailed information on quality of matching and baseline characteristics are presented in Tables 1 and 2, and Supplementary Figs. 2 and 3.

At 5 years, there were no significant differences in mortality rates between the two groups (adjusted HR 1.27, 95% Cl 0.91–1.77, P = 0.16). Cox proportional hazard assumption was not violated (P = 0.19). The results of mortality analysis are shown in Fig. 2 and graphical representation of Schoenfeld residuals and log-log plot of survival are presented in Supplementary Figs. 4 and 5.

Fig. 2

Kaplan–Meier curves showing five-year survival before (A) and after propensity score matching (B). CI confidence interval; HR hazard ratio

Subgroup analysis showed mostly consistent directional trend toward worse prognosis amongst patients with diabetes. Patients with diabetes had worse outcomes than patients without diabetes in two-stent angioplasty techniques (P_difference=0.003, P_interaction=0.004) and elective PCI (both P_difference and P_interaction <0.001). There were no significant differences in long-term prognosis, subgroup analyses, or interaction analyses between patients with insulin-dependent and non-insulin-dependent diabetes. Forest plots for outcomes in selected subgroups are presented in Figs. 3 and 4. Additional subgroup analyses based on insulin-dependence status are presented in Supplementary Figs. 6, 7.

Fig. 3

Forrest plot showing selected subgroup analysis after propensity score matching. CI confidence interval; CKD chronic kidney disease; HR hazard ratio; ULMCA unprotected left main coronary artery

Fig. 4

Forest plot showing subgroup analysis of mortality outcomes in patients with insulin-dependent and non-insulin-dependent diabetes. CI confidence interval; CKD chronic kidney disease; HR hazard ratio; IDDM insulin-dependent diabetes mellitus; NIDDM non-insulin-dependent diabetes mellitus

Discussion

Our analysis encompassed the largest single-center registry of percutaneous revascularisation procedures for the LMCA in Poland. It evaluated survival outcomes of patients with diabetes following PCI compared to those without diabetes in this population. The main findings of this real-world clinical registry analysis include: (1) A 35% worse survival rate was observed in the diabetes group before the application of propensity score matching; after matching, no significant differences in survival were found between the groups. (2) When a two-stent technique was employed, better survival was noted in the non-diabetic group. (3) A significant disparity based on sex assigned at birth was evident in the frequency of PCI procedures, with men undergoing PCI more often. (4) In patients with MI, people with diabetes have been observed to have better survival, whereas in patients undergoing elective revascularisation, diabetes was noted to be associated with an increased risk of mortality. (5) Heart failure, chronic kidney disease, intravascular imaging, and concomitant multivessel disease appear to have no significant impact on outcomes between patients with and without diabetes.

The number of large observational studies evaluating PCI outcomes for LMCAD remains limited, with few providing long-term follow-up data on real-world revascularisation effects ​ [9, 10]​. The European Society of Cardiology currently recommends CABG as the preferred treatment for patients with chronic coronary syndrome and significant multivessel disease with diabetes (Class I, Level A). However, for patients at very high surgical risk, irrespective of anatomical complexity, PCI is recommended as an alternative to conservative management (Class IIa, Level B), representing an upgrade from the previous Class III recommendation. This shift reflects the growing recognition of PCI as a valuable therapeutic option when CABG is deemed excessively high-risk [11]. Importantly, the withdrawal of the SYNTAX score criterion [11] has further expanded access to PCI for patients with LMCAD, broadening its applicability in clinical practice. Our study is among the first to focus on this specific patient group.

Comparison of mortality in patients with diabetes

In our analysis, before propensity score matching five-year survival was lower in patients with diabetes than in those without. However, after matching, this difference was no longer significant, suggesting that excess mortality in the diabetes cohort could be attributed to comorbid conditions rather than diabetes itself. These findings align, yet differ in nuance, with a pooled analysis of four randomised trials (NOBLE, EXCEL, SYNTAX, and PRECOMBAT), which reported higher all-cause mortality in diabetic patients undergoing revascularisation for LMCAD. Notably, cardiovascular mortality was higher in patients with diabetes and high SYNTAX scores undergoing PCI, while no significant difference was observed in those without diabetes [12].

Furthermore, when stratified by insulin dependence, we observed no significant differences in long-term prognosis, nor in subgroup or interaction analyses between insulin-dependent and non-insulin-dependent patients. These findings are consistent with two recent meta-analyses of randomised trials, which demonstrated that insulin therapy in patients with type 2 diabetes does not significantly increase the risk of all-cause or cardiovascular mortality and major adverse cardiovascular events [13, 14].

Interpreting these results requires consideration of trial design. The pooled analysis included patients deemed equally suitable for PCI or CABG, limiting generalisability to routine practice, where treatment preferences often exist. Adding further complexity to the interpretation of trial outcomes, PCI and CABG techniques have evolved since these randomised trials were published. The advent of next-generation drug-eluting stents (DES), alongside refinements in procedural techniques and adjunctive pharmacotherapy, has significantly enhanced PCI outcomes [7, 15]. For instance, the emergence of newer-generation DES has been linked to a possible reduction in thrombotic and restenosis complications [15], potentially influencing treatment outcomes and limiting the comparability between patients treated with earlier and newer stent models. On the other hand, randomized studies reported intravascular imaging guidance in up to 76% of cases, while contemporary registry reports show significantly lower rates, ranging from 38 to 59%, which aligns with our data [6]. Furthermore, PCI offers advantages such as reduced invasiveness, shorter hospitalisation and faster recovery, potentially improving patients’ quality of life [7]. Concurrently, advancements in surgical techniques and perioperative care have likely contributed to enhanced long-term outcomes in CABG. Demographic differences further complicate extrapolation, as trial populations had a median age of 66 years.

In contrast, real-world cohorts, including ours (median age 70.5–73 years), tend to be older with greater comorbidity burden [6, 10]. Finally, the detrimental effect of diabetes after PCI is especially associated with an increased risk of thrombosis and restenosis in small-caliber arteries. Thus, this impact may not be apparent in LMCA, which is usually an artery with the largest area.

These findings reinforce the critical role of multidisciplinary teams and individualised treatment strategies, particularly in evolving revascularisation technologies and techniques.

Impact of the two-stent technique on outcomes

In this study, a higher survival rate was observed in non-diabetic patients compared to those with diabetes who underwent treatment with the two-stent technique. While the impact of diabetes on PCI outcomes in bifurcation lesions remains insufficiently studied, existing evidence suggests a worse prognosis in diabetic patients. Strepkos et al. [2] reported a significantly higher incidence of major adverse cardiac events in diabetic patients, although their analysis did not specifically assess outcomes associated with the two-stent technique. To address this gap, Song et al. [16] investigated this revascularisation strategy and identified diabetes and LMCA bifurcation treatment as independent predictors of target vessel failure. Our findings provide additional evidence suggesting a negative impact of diabetes on PCI outcomes in LMCAD bifurcations treated with the two-stent technique, highlighting the need for dedicated analyses, given the heterogeneity of bifurcation populations in previous studies.

Sex assigned at birth disparity

A significant disparity based on sex assigned at birth was observed in our cohort, with women comprising only 28.2% of patients undergoing PCI for LMCA stenosis. This proportion aligns with population-level trends, as LMCAD is more prevalent in men [12]. While prior studies have reported higher rates of major adverse cardiac events and MI in women, no significant sex-based differences in mortality following PCI for unprotected LMCA have been identified [17]. In our study, which focused solely on mortality, survival outcomes did not differ between sexes, suggesting that the observed disparity does not influence overall results.

Myocardial infarctions

In our analysis, diabetic patients with MI were observed to have better survival than controls, whereas in patients undergoing elective revascularisation, diabetes was found to be associated with an increased risk of mortality. These findings are important and may be prone for misinterpretation, thus requiring careful discussion.

Disparities in mortality may in part be attributable to observed differences in age at MI onset between patients with different diabetes status. In our cohort, the median age at the time of MI was 74 years among people with diabetes and 76 years among those without diabetes. This difference was especially significant in patients hospitalized after 2015, as people with diabetes were 74 (66–80) years old while median age of patients without diabetes was 80 (67–84). Consequently, observed interaction may be related with excess risk of death among patients without diabetes due to the older age during MI occurrence. It is plausible that younger patients with fewer comorbidities, greater physiological reserve, and more preserved cardiac function, could exhibit increased resilience following an acute coronary event. Moreover, another possible explanation for this finding is the demonstrated higher prevalence and greater extent of collateral circulation in individuals with microvascular complications in diabetes [18], which has been linked to improved survival following MI [19].

On the other hand, diabetic patients undergoing elective PCI may face significantly higher mortality rates due to the complex and diffuse nature of atherosclerosis in this population [20]. Silent ischemia is more common in diabetes and could further delay CAD detection, often resulting in a more advanced disease state at the time of intervention [21].

Coexistence of heart failure, chronic kidney disease, and additional lesions

In the present study, no significant difference in survival was observed between diabetic and non-diabetic patients with two or more coexisting coronary artery lesions. However, prior research indicates that diabetes is associated with more extensive coronary involvement, affecting both proximal and distal segments [5]. A study by Oqab et al. [22] focusing on the population with ST-elevation myocardial infarction (STEMI) evaluated the impact of complete revascularisation, finding that adverse events were more frequent in diabetic patients. Notably, this cohort exhibited a greater burden of comorbidities compared with non-diabetic patients. In our analysis, propensity score matching was employed to minimise baseline differences between diabetic and non-diabetic groups, with BMI being the only significant variation between them. This development is likely to reduce the influence of confounding variables on survival outcomes.

Our analysis did not demonstrate significant differences in survival between patients with and without diabetes or heart failure (HF). For this study, we did not differentiate between patients with preserved or reduced ejection fraction in the context of HF. There is a notable paucity of evidence from randomised clinical trials comparing outcomes in patients with HF undergoing PCI for LMCAD to those without HF. This is particularly relevant in the subset of patients with coexisting diabetes, where HF prevalence can reach up to 20% [23]. It has been shown that patients with HF and advanced coronary CAD involving the LMCA have lower survival probabilities compared to those with HF and either single-vessel disease or normal/non-obstructive CAD [24].

Chronic kidney disease (CKD) is frequently considered a risk factor for repeat revascularisation, although the underlying mechanisms remain unclear. CKD is hypothesised to exacerbate chronic inflammation, while diffuse and complex vascular changes may result in incomplete revascularisation. These effects are likely compounded by comorbid hypertension and diabetes, increasing the possibility of repeat revascularisation after PCI or CABG ​ [25]. CKD is associated with major adverse cardiac and cerebrovascular events following PCI in patients undergoing revascularisation for LMCAD [26]. Although data from other studies suggest that CKD elevates the risk of recurrent adverse events, our findings did not demonstrate an independent effect of CKD on overall mortality following PCI for LMCAD.

The absence of significant differences in survival across all analysed subgroups may partly be attributed to the limited sample size in our analysis. The high procedural volume at our facility enables the refinement of techniques and strategies, potentially contributing to improved outcomes. However, these findings are hypothesis-generating and suggest subgroups for further investigation. Given the aging population, shifts in the profile of comorbidities, and advancements in pharmacotherapy, there remains a critical need for large-scale, randomised clinical trials with extended follow-up to understand the long-term effects of PCI better. Patients with diabetes represent a particularly high-risk cohort due to their growing prevalence and elevated cardiovascular risk. Targeted research is necessary to define the optimal revascularisation strategy for these patients.

Strengths and limitations

Apart from limitations related to the registry which were listed earlier, some specific to the current study which need to be discussed [6]. Firstly, given the observational and non-randomised nature of analysis, there were some inherent limitations and bias, which were tried to be minimized by propensity score matching. Although this method reduces the risk of confounding, it is not exhaustive and unmeasured factors may still influence results. This single-centre registry was conducted in a high-volume tertiary referral hospital serving a population of over two million and covered almost 15 years of observation. This setting ensures robust data quality but requires cautious interpretation when generalizing findings to larger populations or directly comparing with other registries. The analysed endpoint included all-cause mortality, which on one hand is the most reliable and clinically meaningful outcome, eliminating the risk of misclassification that may occur with more complex measures, but on the other may make direct comparisons with other studies investigating composite endpoint difficult. To ensure consistency and clarity, we only included patients with clearly defined diabetes status. We included the percentage of patients treated with insulin, which improves the characterisation of the diabetic cohort and may facilitate comparison with other study populations. However, as the study covered almost fifteen years of observation, some clinical data, i.e., glycated haemoglobin levels or prescribed oral hypoglycaemic agents, were unavailable for analysis.

Finally, we used angiographic classification based on the number of vessels involved rather than the SYNTAX score, what may hinder comparisons with other studies. However, it is notable that the most recent ESC guidelines have de-emphasized using the SYNTAX score to guide PCI decisions in patients with LMCAD, reflecting a shift away from reliance on the subjective score to guide clinical management in these populations.

Conclusions

The present study has revealed indications that diabetes may not have an impact on overall mortality in patients undergoing unprotected LMCA PCI. Nevertheless, it is important to interpret this finding with caution due to the limited number of patients and the retrospective nature of the study. Conversely, patients with diabetes have been observed to experience worse survival outcomes when a two-stent technique is employed. Notably, factors such as heart failure, chronic kidney disease, the utilisation of intravascular imaging, and the presence of two or more atherosclerotic lesions do not appear to exert a significantly different impact on outcomes between diabetic and non-diabetic patients. While people with diabetes were observed to have better survival in myocardial infarction, they experienced higher mortality in elective revascularisation. Validating these findings requires randomised trials with large cohorts and extended follow-up, focusing on high-risk subgroups such as those with diabetes, heart failure, and chronic kidney disease, while assessing the long-term effects of PCI to identify the most effective treatment regimens for these populations.

Table 1 Procedural details before and after propensity score matching

Table 2 Procedural details before and after propensity score matching