All patients who underwent CTO-PCI at the University of Washington Medical Center between January 2019 and December 2023 were considered for inclusion. As previously described, data were automatically entered into the institutional CathPCI database, which was constructed using customized SQL code to extract information from the hospital’s electronic health record system (Epic, Epic Systems, Verona, WI), the NCDR CathPCI Registry submissions, and other institutional databases. The dataset included demographic and clinical characteristics, laboratory and echocardiographic data, angiographic and procedural details, as well as complications and in-hospital adverse events. After identifying all CTO-PCI cases, an experienced CTO operator independently reviewed each procedure to assess angiographic characteristics and procedural strategies. This study was approved by the University of Washington Institutional Review Board (STUDY00016614). The requirement for informed consent was waived due to the retrospective nature of the study.

The PROGRESS-CTO complication scoring system was developed to estimate the risk of periprocedural adverse events in patients undergoing CTO-PCI. It provides risk prediction for in-hospital MACCE, mortality, pericardiocentesis, and acute MI. Using data from more than 10,000 CTO-PCI procedures in the PROGRESS-CTO registry, the score was built through multivariable logistic regression and internally validated with bootstrapping. The models incorporated clinical, angiographic, and procedural variables including age, sex, left ventricular ejection fraction, prior coronary artery bypass grafting, atrial fibrillation, calcification severity, lesion stump morphology, and crossing strategy (antegrade dissection and re-entry, or retrograde). Weighted points were assigned to each factor according to their odds ratios, allowing simplified bedside calculation. In addition, the PROGRESS-CTO perforation score was specifically designed to estimate the risk of clinical coronary artery perforation during CTO-PCI. Derived from over 9,600 procedures in the PROGRESS-CTO registry, it was developed also using multivariable logistic regression and internally validated with bootstrapping. The score incorporates 5 independent predictors: age ≥ 65 years, moderate/severe calcification, blunt proximal cap, use of antegrade dissection and re-entry or the retrograde approach. Weighted points were assigned according to the magnitude of the odds ratios, yielding a simplified 0–5 risk score.

The BMC2 scoring system leverages machine learning to predict multiple in-hospital adverse events following PCI, including mortality, AKI, new-onset dialysis, stroke, major bleeding, and the need for transfusion. This model utilizes 23 variables and leverages an XGBoost machine learning algorithm, trained on data from the Blue Cross Blue Shield of Michigan Cardiovascular Consortium registry. The BMC2 score was developed using a learning cohort of 107,793 patients undergoing PCI, with external validation performed in 56,583 procedures, and demonstrated strong predictive discrimination.

Although both scores were designed to predict periprocedural complications, they differ substantially in scope, methodology, and outcomes. The BMC2 model was developed in an all-comer PCI population, leveraging machine learning to capture complex nonlinear interactions, whereas the PROGRESS-CTO complication score was specifically created for CTO-PCI using multivariable logistic regression to create an easy-to-calculate integer-based score. Moreover, the only overlapping endpoint predicted by both tools is in-hospital mortality, while the remaining outcomes differ. In this study, our aim was to externally validate both scoring systems in the setting of CTO-PCI and to perform a direct head-to-head comparison of their predictive performance for in-hospital mortality.

CTO were defined as a 100% stenosis with a thrombolysis in MI flow grade of 0 and an estimated occlusion duration of ≥ 3 months BMC2 and PROGRESS-CTO complication scores were calculated for each patient to assess complication risk prediction. The main endpoint of this analysis was in-hospital mortality. All outcomes were defined according to the criteria reported in the original studies, ^, with the exception of MACCE, which was defined as a composite of all-cause mortality, MI, stroke, or pericardiocentesis. Urgent repeat revascularization was not included, in contrast to its incorporation in the PROGRESS-CTO complication score, as it was not captured by our institutional database. Additional outcomes, assessed only for the BMC2 model, included AKI (defined as an absolute increase in post-PCI creatinine of ≥ 0.5 mg/dl), new dialysis requirement (unplanned initiation of dialysis during hospitalization due to post-PCI renal deterioration), stroke (ischemic, haemorrhagic, or unspecified), major bleeding (a drop in haemoglobin >5 g/dl between pre and post-PCI measurements), and transfusion (administration of at least 1 unit of packed red blood cells or whole blood following PCI). MI was defined according to the Society for Cardiovascular Angiography and Interventions criteria. Coronary perforation, as defined in the PROGRESS-CTO perforation score, was considered any perforation requiring treatment, including pericardiocentesis, covered stent implantation, or other therapeutic interventions.

The discriminative performance of each risk model was quantified using the area under the receiver operating characteristic curve (AUC). Comparisons of AUC values between the BMC2 and PROGRESS-CTO scores were conducted with the DeLong test for mortality prediction. For each score additional performance metrics, including accuracy, sensitivity, specificity, and positive and negative predictive values, were calculated. For these secondary metrics optimal cut-off points were derived within the study cohort using the Youden Index, whereas the primary comparison between models was based on discrimination (AUC). Finally, a bootstrap resampling procedure (1,000 iterations) was used to estimate 95% CIs for the AUC of each model and for the paired difference in AUC between models.

Model calibration was evaluated using calibration plots and error indices, while overall goodness of fit was tested with the Hosmer–Lemeshow statistic to assess agreement between predicted and observed event rates. Model accuracy was further quantified by the Brier score, with values approaching zero indicating more accurate probability estimates and values closer to one reflecting poor predictive performance. For variables strictly required for the calculation of either score, cases with missing data were excluded without imputation, as done by the original studies. Conversely, for optional predictors in the BMC2 model (total cholesterol, HDL cholesterol, haemoglobin, and left ventricular ejection fraction), missing values were replaced with the cohort median, in accordance with the methodology applied in the original derivation study.

Statistical analyses were performed using R v 4.1.3 (R Foundation for Statistical Computing) in RStudio environment, version 2022.07.02 (RStudio, PBC).

Between January 2019 and December 2023, 4,679 PCI procedures were performed at our institution. Among them 1,412 (30.2%) were CTO-PCI and were used for the present analysis. Overall, 1,157 of 1,412 (81.9%) procedures had sufficient data for evaluation. All excluded patients were omitted due to missing data necessary for the calculation of both the PROGRESS-CTO and BMC2 scores. Specifically, the variables most frequently missing in the excluded observations were: age = 94; left ventricle ejection fraction = 94; body mass index (height and/or weight) = 99; creatinine level = 3; procedural systolic blood pressure = 49; CSHA scale = 154; patient status during the procedure = 2; diabetes status = 99; NYHA = 142; heart failure type = 205; tobacco use = 124; PCI indication = 100; stress test result = 105. A comparison between included vs excluded patients is provided in the Supplementary Table 1.

Baseline population characteristics are summarized according to MACCE status ( Table 1 ). The median age was 68 years (61-74), and 189 patients (16.3%) were female. The majority of procedures were performed electively (90.7%). Plaque modification devices were used in 17.5% of procedures, most frequently rotational atherectomy (7.7%), whereas intravascular lithotripsy was employed in 3.5%.

The overall in-hospital MACCE rate was 3.0% (35 of 1,157). Among these events, in-hospital death occurred in 11 patients (1.0%), stroke in 8 (0.7%), clinically relevant coronary perforation in 24 (2.1%), pericardiocentesis in 17 (1.5%), and MI in 2 (0.2%) ( Table 2 ). AKI developed in 39 patients (3.4%), with 4 (0.3%) requiring new dialysis. Blood transfusion was required in 22 patients (1.9%), whereas major bleeding was infrequent (0.6%). As anticipated, risk scores were significantly higher in patients with MACCE compared with those without (Supplementary Table 2).

Table 2

Incidence of peri-procedural complications following CTO-PCI

Characteristic	Overall, N = 1,157	No MACCE group, N = 1,122 (97.0%)	MACCE group, N = 35 (3.0%)	P -value
MACCE components	35 (3.0%)	N/A	35 (100%)	<.001
− In-hospital death	11 (1.0%)	N/A	11 (31.4%)	<.001
− In-hospital stroke	8 (0.7%)	N/A	8 (22.9%)	<.001
− In-hospital MI	2 (0.2%)	N/A	2 (5.7%)	<.001
− Pericardiocentesis	17 (1.5%)	N/A	17 (48.6%)	<.001
Coronary perforation	24 (2.1%)	8 (0.7%)	16 (46%)	<.001
AKI	39 (3.4%)	30 (2.7%)	9 (25.7%)	<.001
New requirement for dialysis	4 (0.3%)	2 (0.2%)	2 (5.9%)	.005
Major bleeding	3 (0.6%)	3 (0.6%)	0 (0%)	.912
Transfusion	22 (1.9%)	8 (0.7%)	14 (40.0%)	<.001

Only gold members can continue reading. Log In or Register to continue

Share this:

• X
• Facebook

Related

Related posts:

1. Design and rationale of the clinical trial to obtain the highest efficacy of dual antiplatelet therapy after carotid artery stenting in high bleeding risk patients (CHET): A multicenter, randomized, open-label, superiority trial
2. Remote dietitian counseling with short-term meal delivery improves DASH diet adherence and lowers blood pressure in veterans with hypertension and obesity
3. A comprehensive set of per-protocol analyses of the myocardial ischemia and transfusion (MINT) trial
4. Effect of age on restrictive and liberal transfusion outcomes in patients with anemia and myocardial infarction

Stay updated, free articles. Join our Telegram channel

Join