Predicting the risk of abrupt vessel closure after angioplasty in an individual patient.

OBJECTIVES

We proposed to examine the relation between angiographic morphologic characteristics and abrupt closure after coronary angioplasty and to develop an empirically based risk stratification system.

BACKGROUND

Certain lesion morphologic characteristics are associated with higher rates of abrupt closure after coronary angioplasty. Previous approaches have been limited by relatively small sample sizes and an inability to combine multiple characteristics to predict risk in an individual patient.

METHODS

Lesion morphology was determined for 779 lesions in 658 patients undergoing an elective first angioplasty. Abrupt closure occurred in 63 lesions (8.1%). Variables associated with abrupt closure were identified by univariate and stepwise multiple logistic regression analysis, and internal validity was assessed by use of bootstrapping. An empirically based scoring system was developed by assigning different weights to each predictive characteristic and was then validated.

RESULTS

Almost all lesion characteristics previously labeled "adverse" were associated with an increased risk of abrupt closure, but only total occlusion, location at a branch point, increasing lesion length, evidence for thrombus and right coronary artery location were statistically significant independent predictors. Despite the large sample size, the study was underpowered to detect even a 50% increase in risk with many characteristics. Using a scoring system, we assigned each lesion a specific risk of abrupt closure. The distribution of risk was broad, with 20% of patients having < or = 2.5% risk and 25% having > 10% risk. Internal validation techniques revealed that when 10% of patients were randomly eliminated from the sample in multiple iterations, the risk estimates varied, again pointing to the need for a larger sample.

CONCLUSIONS

Empirically based weighting of lesion characteristics could quantify the risk of abrupt closure for individual patients, but a very large sample will be required to understand the interplay of complex lesion characteristics in altering expected outcomes.