Prognostic irrelevance of plaque vulnerability following plaque sealing in high-risk patients with type 2 diabetes: an optical coherence tomography study

We enrolled 81 patients with known T2DM undergoing coronary angiography with OCT-guided PCI at the Department of Internal Medicine I, University Hospital of the RWTH Aachen, Germany between May 2011 and January 2014. Indication for coronary angiography and PCI was ACS in 27 patients and stable angina in 54 patients. Stable coronary artery disease (CAD) was defined as no progression of symptoms in the last 6 weeks; in this case, identification of the target lesion relied on stress imaging and/or FFR. Previous studies partially relied on these data [3, 4, 9, 12, 17, 18]. Further inclusion criteria were age > 30 years and written informed consent to the study protocol. Exclusion criteria included left main coronary artery stenosis, graft stenosis, hemodynamic or rhythmic instability, acute or chronic renal insufficiency (serum creatinine level > 1.5 mg/dl), systemic acute or chronic infections, pregnancy, anti-inflammatory medications such as steroids and chronic total occluded, severely tortuous or calcified vessels, which impeded proper advancement of the OCT catheter. A further exclusion criterion was incomplete follow-up data; these were acquired telephonically using a standardized questionnaire recording, among other, death of the patient and time of death. Furthermore, cardiovascular death was recorded; this was defined as death resulting from myocardial infarction, sudden cardiac death, death due to heart failure, death due to stroke, death due to cardiovascular procedures, death due to cardiovascular hemorrhage, and death due to other cardiovascular causes, according to previous definitions [26].

The study was approved by the local Ethics Committee and conforms with the Declaration of Helsinki on ethical principles for medical research involving human subjects.

OCT image acquisition of the target vessel was performed by using the Frequency Domain-OCT C7XR system and the DragonFly catheter (St. Jude Medical, LightLab Imaging Inc, Westford, Massachusetts, USA) as previously described [27]. Target/culprit lesion was defined as previously described [4]. OCT-image analysis was executed throughout the entire lesion frame by frame in a 0.2 mm interval by two independent observers applying the proprietary software provided by LightLab Imaging. Determination of intraluminal and plaque morphologic parameters by OCT occurred as previously described [4].

Statistical analysis was performed using SPSS Statistics (IBM Corp., Armonk, NY, USA). Categorical variables are reported as n and percentage, continuous variables as mean ± standard deviation. A comparison of continuous variables was performed with t-test, while a comparison of nominal variables took place using Pearson’s χ-squared test.

Follow-up duration was determined by reverse Kaplan–Meier analysis. Uni- and multivariable Cox regression analyses were performed to calculate hazard ratios (HR) for clinical and OCT-derived parameters to predict mortality. For multivariable analysis, significant predictors in the univariable analysis were included in the model and then eliminated stepwise through backward-selection. In order to avoid collinearity of two or more parameters, a selection according to statistical and biological significance was performed before inclusion in the multivariable model. Specifically, we selected calcium volume index among all parameters assessing calcium quantity in the target/culprit lesion which were significant in the univariable analysis (presence of a calcific plaque, presence of macrocalcifications, number of macrocalcifications, mean calcium arc, calcium length). In summary, we included several clinical (age, BMI, HDL-c, GFR, diabetic polyneuropathy, insulin therapy, metformin therapy) and plaque-morphological (lesion length, stenosis eccentricity, calcium volume index) parameters in the multivariable Cox regression model. These parameters were then selected stepwise backwards if presenting p > 0.10; this resulted in a multivariable model including 5 variables (age, diabetic polyneuropathy, insulin therapy, lesion length, calcium volume index). The final model was validated, accounting for the variable selection process, by the bootstrap as proposed by Harrell [28]. Beforehand, missing values were multiply imputed (5 sets). Optimism in the c-statistic (area under ROC curve) was about 6%, e.g. from 0.86 (original) to 0.81 (bias-corrected) with a slope of 0.61. Thus, model coefficients should be shrinked by 39% (multiply by 0.61) to get better predictions. For plaque morphological features relevantly associated to excess mortality, we performed receiver operating curve (ROC) analysis and determined optimal cut-off values for mortality after 5 years. Optimal cut-off was defined as the value with the highest Youden index; diagnostic efficiency was classified as previously described [29]. Kaplan–Meier analyses were performed using these cut-off values. Statistical significance was awarded by a p-value < 0.05.

A total of 81 patients with T2DM underwent coronary angiography due to stable/unstable CAD. Following OCT-guided PCI, patients were followed for a median follow-up of 66.0 (IQR = 8.0) months. In this period of time, 24 (29.6%) patients died. The baseline patient and lesion characteristics are shown in Tables 1 and 2, including a comparison of non-survivors with survivors. The most relevant data are also graphically shown in Fig. 1. 

In order to find predictors of mortality following PCI, univariable Cox regression analyses were performed. We found several clinical (Table 3) and morphological plaque characteristics (Table 4) to be predictors of mortality following PCI. However, classical features of plaque vulnerability such as thickness of the fibrous cap overlying the necrotic lipid core, the extent of the necrotic lipid core, the presence of micro- or spotty calcifications and the presence of macrophages had no significant predictive value for mortality (all p = ns), neither in patients with ACS nor in patients with stable CAD (p = ns).

To evaluate which of the above-mentioned parameters independently predict mortality following PCI, multivariable analysis was performed. In this analysis the clinical parameters age (HR 1.16 per year, 95% CI 1.07–1.26, p < 0.001), diabetic polyneuropathy (HR 3.58, 95% CI 1.44–8.93, p = 0.006) and insulin therapy (HR 3.25, 95% CI 1.44–8.70, p = 0.019) were significant predictors of mortality. Regarding plaque characteristics, only calcium volume index (HR 1.71 per 1000°*mm, 95% CI 1.21–2.41, p = 0.002) and lesion length (HR 1.93 per 10 mm, 95% CI 1.02–3.67, p = 0.044) were independent significant morphological predictors of mortality following PCI.

To further analyze the role of these parameters in predicting cardiovascular events, we focused on the prediction of cardiovascular mortality. Here, only age (HR 1.14 per year, 95% CI 1.02–1.28, p = 0.018) and calcium volume index (HR 2.04 per 1000°*mm, 95% CI 1.29–3.23, p = 0.002) could predict cardiovascular mortality in univariable Cox regression analysis. Performing multivariable Cox regression analysis, both age (HR 1.16 per year, 95% CI 1.02–1.32, p = 0.025) and calcium volume index (HR 1.94 per 1000°*mm, 95% CI 1.24–3.02, p = 0.003) still remained significant independent predictors of cardiovascular mortality.

In order to determine the diagnostic value of morphologic predictors of mortality after 5 years, ROC analyses were performed. In these analyses, calcium volume index predicted mortality following PCI with good diagnostic efficiency (AUC 0.769, 95% CI 0.525–0.845, p = 0.001; sensitivity: 64.7%, specificity: 80.4% at the optimal cut-off = 1399.8°*mm). Furthermore, lesion length predicted mortality following PCI with sufficient efficiency (AUC 0.685, 95% CI 0.641–0.897, p = 0.023; sensitivity: 56.3%, specificity: 83.9% at the optimal cut-off = 20.65 mm). The ROC-curves are represented in Fig. 2.

To analyze the diagnostic value of these cut-offs to distinguish between high- and low-risk patients, Kaplan–Meier analyses were performed. In these analyses, calcium volume index > 1399.8°*mm and lesion length > 20.65 mm were associated with a hazard ratio of 4.30 (95% CI 1.51–12.21, p = 0.006) and 3.22 (95% CI 1.12–9.24, p = 0.030), respectively. Graphical representation is shown in Fig. 3a, b. In order to assess the diagnostic value of minimal FCT, the most widely accepted parameter of plaque vulnerability, to distinguish between high- and low-risk patients following PCI, we performed Kaplan–Meier analyses using the accepted thresholds of 65 and 80 µm [22]. This showed no significant association of this feature of plaque vulnerability with mortality following PCI (HR 1.09, 95% CI 0.32–3.77, p = 0.885 for FCT < 65 µm; HR 1.20, 95% CI 0.32–4.56, p = 0.787 for FCT < 80 µm, see Fig. 3c).

In the present study, significant independent clinical predictors of mortality were age, insulin therapy and diabetic polyneuropathy. Whereas age appears quite self-explanatory, the other two factors do require a brief discussion. Our results show that an insulin-based therapy is associated with an increased mortality risk in our patient population. This association is in line with the findings of previous studies in CAD patients [34, 35] and may reflect advanced T2DM requiring insulin therapy. This hypothesis is further supported by the significantly higher mortality risk of patients with diabetic polyneuropathy, as a typical late-onset complication of advanced T2DM. These results are in line with findings of previous studies describing an association of cardiovascular autonomic neuropathy and an excess mortality in patients with T2DM [36‐38].

Among all plaque-morphological parameters investigated, only parameters assessing extent of calcification (e.g. calcium volume index) and lesion length as markers of advanced atherosclerosis predicted mortality independently in T2DM patients following PCI.

The extent of coronary calcification has been related to adverse prognosis both in patients with [39] and without known CAD [40]; these studies relied on computed tomography to evaluate the calcium burden in the coronary arteries [39, 40]. However, it is unknown if coronary plaque calcification represents a vulnerable plaque feature in itself or merely reflects more advanced CAD. In addition, semi-quantitative assessment of coronary calcification performed with coronary angiography has been associated with increased mortality following PCI [41]. Our findings are in line with this body of evidence, confirming the negative prognostic role of extensive coronary calcification as determined by OCT. This confirms the hypothesis that a more extensive calcification, which is typical for an advanced stage of atherosclerosis, characterizes a high-risk population.

The length of the atherosclerotic lesion was a further parameter independently associated with mortality in our population. Previous data on this feature are sparce, but in the light of the global findings of our study it is tempting to speculate that lesion length may represent yet another marker of a more diffuse and advanced atherosclerosis as it has been described previously in patients with T2DM [42]. These patients with an advanced atherosclerosis are therefore more frequently prone to further cardiovascular events.

Interestingly, in our study, “classical” features of plaque vulnerability (as FCT, extent of the necrotic lipid core, presence of TCFA, macrophage infiltration or microcalcifications) are not associated with a worse prognosis following PCI of the target/culprit lesion. Previously, Niccoli et al. described in patients with ACS an adverse effect of ruptured plaques compared to plaque erosions on prognosis following primary PCI, which is mainly driven by unstable angina and target vessel failure [43]. As the authors of this study discuss, the different outcomes of patients presenting with plaque rupture and plaque erosion as the morphologic correlate of ACS may be due to a multiplicity of factors, including residual thrombus and/or plaque burden causing suboptimal PCI results through stent malapposition or residual disease at stents’ edge. In the present study, we performed routine PCI optimization, which excluded relevant malapposition or residual plaque burden following intervention. Furthermore, the prognostic relevance of plaque rupture and plaque erosion could not be conclusively determined in our study possibly due to low case number in our mixed cohort of patients with ACS and stable angina. However, in the study by Niccoli et al. no other plaque morphological features were associated with adverse outcome in univariate analysis following PCI, which is in line with the findings of the present study.

Furthermore, the prognostic irrelevance of features of plaque vulnerability following PCI, which we could demonstrate in the present study, may suggest that sealing vulnerable plaques neutralizes its effects on prognosis. The principle of plaque sealing is based on treating plaques by PCI to avoid total vessel occlusion and prevent future events due to plaque rupture or thrombosis [44, 45]. So far, prospective interventional data assessing plaque sealing as a potential strategy to prevent major cardiovascular events in patients with chronic coronary syndrome showed no benefit of such an approach [44]. Remarkably, the lesions treated in this previous study were selected without the use of intravascular imaging and based only on angiographic evidence of intermediate lumen stenosis [44]. This evidence, however, needs to be re-evaluated in the light of the current data. In fact, the CLIMA and LPS studies, in analogy to the data from the PROSPECT study, showed a higher incidence of major cardiovascular events in vulnerable plaques [13, 14, 20]. In contrast, the present study suggests that PCI yields a neutralizing effect on the negative prognostic impact of plaque vulnerability in patients with T2DM. In the light of these novel and potentially clinically relevant findings, the present data highlight the need for prospective, randomized studies evaluating the prognostic effect of vulnerable plaque sealing using OCT-guided PCI. This may be of particular interest in populations with an increased cardiovascular risk, such as patients with T2DM.