EuroCTO CASTLE: Prediction Model for CTO PCI Success Shows Good Discriminating Ability

EuroCTO CASTLE, a new scoring system, may simplify the process of predicting whether PCI for chronic total occlusions (CTOs) will be successful, researchers and outside experts say. The tool performed slightly better than another scoring system used for the same purpose, and it includes more pertinent variables that may minimize subjectivity on the part of operators.

“Although our score is broadly similar to published ones, our study confirms the importance of previously identified predictors of CTO PCI success using contemporary data from a multitude of operators,” write Zsolt Szijgyarto, MSc (London School of Hygiene and Tropical Medicine, United Kingdom), and colleagues in their paper published online January 30, 2019 in JACC: Cardiovascular Interventions. “In order to facilitate usage and uptake, we deliberately simplified the scoring system using a point-based approach. We will also be creating a more precise model using the original coefficients of the model, which will be available in app form.”

The report is particularly timely in light of another study, also published in JACC: Cardiovascular Interventions, showing that about one in seven patients undergoing CTO PCI experience at least one complication that prolongs hospitalization, increases cost, and could lead to cardiac surgery or death.

“We know that hospitals have concerns about the cost of CTO PCI because of the equipment that is necessary, as well as the additional time needed in the cath lab to treat these lesions compared with typical PCI,” the study’s lead author, Adam C. Salisbury, MD, MSc (Saint Luke’s Mid America Heart Institute, Kansas City, MO), told TCTMD. “There has been a gap in the literature with regard to what CTO PCI costs, as well as the impact on overall costs of complications that are known to occur more frequently with CTO PCI.”

EuroCTO CASTLE

The EuroCTO CASTLE score was derived from nearly 15,000 patients in whom CTO PCI was attempted (15.5% failure rate) and who were entered into the EuroCTO registry between 2008 and 2014. Szijgyarto et al used another group of 5,745 patients from the registry (12.2% failure rate) who were treated between 2015-2016 to validate the score and look for improvements in success rates over time.

After 21 variables were put into the model, the researchers chose six that were independently associated with CTO PCI failure as the basis of the CASTLE score: 

• CABG (previous)
• Age (> 70 years)
• Stump anatomy (blunt or no)
• Tortuosity (severe)
• Length of CTO (> 20 mm)
• Extent of calcification (> 50% of the segment)

Each CASTLE parameter was assigned a value of 0 or 1, depending on patient characteristics, and the score was calculated by adding up the total, for a potential risk score of 0 to 6. At a score of 0, the mean predicted risk of failure of CTO PCI was 5.8%. At a maximum score of 6, the predicted risk of failure was 56.5%.

In further analysis, the model was judged to be well calibrated, and the area under the curve (AUC) for patients treated between 2008 and 2014 (derivation cohort) was 0.66. When the score was used in the registry patients treated between 2015 and 2016 (validation cohort), the AUC was 0.68. In a comparison, EuroCTO CASTLE modestly outperformed the J-CTO score, which had AUCs of 0.63 and 0.64 in the same cohorts, respectively.

In an editorial, Stephen G. Ellis, MD (Cleveland Clinic, OH), said the failure to create highly predictive CTO models with C-statistics above 0.80 “likely reflects differences in operator skills, unstudied aspects of the procedure, and the fact that the most skilled operators often take on the most complex cases, thereby mitigating the apparent effect of risk factors for failure when attempting to model outcomes.”

Ajay Kirtane, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, New York, NY), who was not involved in the study, said in an email that the new score is practical and makes good clinical sense. “All of the variables included in the score in my estimation do capture some of the challenges for CTO success, particularly with an antegrade approach. So, in a sense, I do think that the authors accomplished their goal(s),” he observed to TCTMD.

According to Szijgyarto and colleagues, the model may not be applicable to centers where operator experience is lower than that of the operators in the EuroCTO registry.

“I think that it really can help for case selection,” added Kirtane. “For example, for primarily antegrade operators, scores like this can really help to identify which cases ought to be done with assistance (eg, a proctor) or referred to a center with specific expertise that would allow for the ability to overcome some of the failure modes identified with some of the predictors of failure (eg, retrograde, [antegrade dissection and reentry], IVUS guidance).”

Comparisons and Cost Considerations

While other CTO PCI prediction models exists, namely J-CTO and PROGRESS, and use similar variables as EuroCTO CASTLE, Szijgyarto et al note that their score includes two additional and important factors, previous CABG and age, which they say enhances the functional integrity of the score by virtue of being both objective and not open to operator interpretation.

To TCTMD, Kirtane said he views the various scores as complementary rather than competitive.

“There is definitely overlap with other scores and some of the variables that predict success/failure. I am not rigorous about calculating scores per se, but I do integrate all of these factors (and others as well: eg, post-CABG) when presenting to the patient what I think our chances of overall success will be,” he said.

For their study on CTO PCI costs, Salisbury and colleagues examined data from 964 patients treated at 12 centers in the OPEN-CTO trial and found that 14.5% experienced at least one complication, 2.7% experienced two complications, and 1.2% experienced three or more complications. The most common of these were clinically significant perforation (4.8%), access-site hematoma (4.3%), and MI (2.7%).

Average length of procedures was 120 minutes, and the procedural cost averaged $12,280 per patient. Nonprocedural hospital costs added an additional $3,424 and physician fees another $1,344, for a total cost of more than $17,000. While the costs were high, Salisbury and colleagues note that they are comparable to what has been observed in trials of multivessel PCI and treatment of left main disease.

To TCTMD, Salisbury observed that the costs for complications were not “dramatic,” with the incremental cost of a procedure-related complication estimated at $8,600, comparable to the average cost of a complication among unselected Medicare patients undergoing standard PCI.

“An additional takeaway is that because some of these complications are preventable, such as access site hematomas or perforations, and may be addressed by changes in strategy, some of those costs may be preventable as well,” he added.

In an accompanying editorial, Gregory J. Dehmer, MD, and M. Ayoub Mirza, MD (both from Carilion Clinic, Roanoke, VA), observe that even as the success rates for CTO PCI increase, it is important to demonstrate that the procedure provides value beyond economic concerns.

“Proving the value of CTO PCI in our cost-conscious and evidence-based environment will require continued efforts to evaluate the benefits of the procedure while optimizing quality and lowering the costs,” they write.

Photo Credit: Riley RF. Wire crosses proximal cap, but the proximal or distal cap is uncrossable or undilatable: what now? Presented at: TCT 2018. September 24, 2018. San Diego, CA.