“Please feel free to utilise this in your practice,” Michael Dake (University of Arizona Health Sciences, Tucson, USA) urged on the first day of the Cardiovascular and Interventional Radiological Society of Europe (CIRSE) 2020 Symposium (12–15 September, virtual). Speaking during a [email protected] session, he presented a new, interactive, web-based tool built to predict patients’ freedom from target lesion revascularisation (TLR) following treatment with the Zilver PTX (Cook Medical) drug-eluting stent (DES).
Dake and colleagues used patient and lesion factors from five global clinical studies from Cook Medical to develop a prediction model for freedom from target lesion revascularisation (TLR) following use of Zilver PTX. These clinical studies studied both pre-market and post-market outcomes with the Zilver PTX stent in patients with femoropopliteal disease. In total, they collated data came from 2,374 patients, and included 15 risk factors in their creation of the model.
“It is note-worthy that over 50% of the patients included in the model had over five-year follow-up. In fact, this is the first prediction model to estimate the impact of patient and lesion characteristics on freedom from TLR through five years for patients with PAD,” Dake enthused. “Based on unique patient profiles, the model provides expected patient outcomes following treatment with the Zilver PTX DES, and may assist in defining algorithms for patients as the value of population management is increasingly recognised.”
Turning to the freedom from TLR results, Dake informed the CIRSE audience that 94% of the complete dataset (2,227 cases, with a median follow-up time of two years) was used to generate the model. Freedom from TLR was 90.5% at one-year, and 75.2% at five years.
Describing the model more closely, Dake next related the 15 risk factors that were considered when the tool was under construction: sex, age, diabetes, hypertension, hypercholesterolemia, renal disease, smoking status, Rutherford classification, lesion length, dexamethasone, popliteal involvement, total occlusion, calcification, prior interventions, and the number of patient runoff vessels. “Most all of these have been in the past implicated as being high-risk for TLR in studies performed over the last two decades,” Dake disclosed.
He continued: “Risk factors common in PAD patients collectively contributed to overall prognosis. As expected, chronic limb-threatening ischaemia (CLTI), lesion length, and total occlusion have a significant impact on TLR. Other factors, such as diabetes and calcification, do not have a significant impact on TLR.”
He then talked through three example patient profiles to illustrate how listening physicians could use this predictive model in their own practices. The risk factors any given patient may have are fed into the model, which then churns out an estimation of the risk that patient has of maintaining freedom from TLR out to five years, with a standard error given, when treated with the Zilver PTX. These results also translate into a freedom from TLR Kaplan-Meier curve. The physician can then make an informed treatment decision based on individual risk factors.
The baseline data
Teasing apart the larger dataset to examine trends in baseline patient demographics, Dake said: “Looking a bit closer at some of these individual patient demographics, and how they are distributed within the various trials included in the analysis, you can see some differences. Of note, in the Japan post-market study, [there is] a very high frequency of renal disease, and a higher frequency of CLTI. In terms of diabetes, there was a high frequency in all studies, approaching 50%.”
Appraising the baseline lesion characteristics, Dake informed viewers that 42% of patients had total occlusions, and that there was an increased number of prior interventions in both the single-arm study and the Japan post-market study, as both these trials included in-stent restenosis. “Patients with in-stent restenosis were not allowed to enter the other trials,” he explained.
“This is the first step”
Praising Dake for these results, moderator Stefan Müller-Hülsbeck (Diako Hospital, Flensburg, Germany) commented: “I think you have shown once again that paclitaxel is probably safe, and we meet a new level of security when using these kinds of devices, especially when using this kind of predictability model.” He then asked Dake if the model is already in use, and if so, if he was using it for his patients during his daily practice.
“This is actually the first debut of this model and this website,” Dake responded. “The manuscript will be published in Cardiovascular and Interventional Radiology (CVIR).” He encouraged CIRSE attendees to visit the site themselves and “really get a sense of getting comfortable with how individual risk factors can affect TLR going out through five years”.
Referring to his presentation as the launch of this predictive model, Dake closed the discussion by postulating: “Maybe other device manufacturers will look at this as a model and adopt predictive models as a way to counsel patients and their families in the clinic regarding their individual risk factors, modification of those risk factors, how we can really just predict and have them anticipate what results might be with individual different devices. I think this is the first step. I think this is an outcropping of what we have seen as a result of the Katsanos paper [Konstantinos Katsanos (Patras, Greece) et al published a paper in late 2018 that suggested the existence of an increased mortality risk with paclitaxel devices used in the peripheral arteries]. The next level now is [to evaluate the] efficacy in terms of using paclitaxel in individual patients.”