Validation of randomized controlled trial-derived models for the prediction of postintervention outcomes in chronic limb-threatening ischemia

Artificial Intelligence to Predict Quality-of-Life Outcomes for Vascular Intervention of the Leg

BACKGROUND

Artificial intelligence (AI) tools created to enhance decision-making may have a significant impact on treatment algorithms for peripheral arterial disease (PAD). A Markov-based AI model was developed to predict optimal therapy based on maximization of calculated quality of life (cQoL), a patient-centered system of assessment designed to report outcomes directly linked to health-related quality of life.

STUDY DESIGN

The AI model was prospectively interrogated immediately after individual interventions for PAD over a 12-year period to test predictive performance. Patient cQoL was determined at each patient follow-up visit.

RESULTS

A total of 1,143 consecutive patients were evaluated, with a median follow-up of 18 months. Observed mean annualized cQoL was higher than predicted by the model (0.85 ± 0.38 vs 0.79 ± 0.18, p < 0.0001). Of 5 potential clinical outcomes, the AI model correctly predicted final status in 71.3% of patients, with insignificant model performance deterioration over time (-0.15% per month, r = -0.49, p = 0.063). The chance of having the condition predicted by the model was 0.57 ± 0.32, compared with a theoretical maximum of 0.70 ± 0.19 (p < 0.0001, mean ratio 0.79). The AI model performed better in patients with claudication than limb-threatening ischemia (75.5% vs 63.6%, p = 0.014) but equally well for open or endovascular intervention (69.8% vs 70.5%, p = 0.70). Graft or artery patency and amputation-free survival were better for patients with claudication and those treated with endovascular techniques.

CONCLUSIONS

AI can successfully predict treatment for PAD that maximizes patient quality of life in most cases. Future application of AI incorporating better estimates of patient anatomic and physiological risk factors and refinement of model structure should further enhance performance.

Validation of JCLIMB, SPINACH, and VQI Calculators for Prediction of Two Year Survival in Patients With Chronic Limb Threatening Ischaemia After Infra-Inguinal Surgical or Endovascular Revascularisation

OBJECTIVE

This study aimed to evaluate three survival prediction models: the JAPAN Critical Limb Ischaemia Database (JCLIMB), Surgical Reconstruction Versus Peripheral Intervention in Patients With Critical Limb Ischaemia (SPINACH), and Vascular Quality Initiative (VQI) calculators.

METHODS

Multicentre data of patients who underwent infrainguinal revascularisation for chronic limb threatening ischaemia between 2018 and 2021 were analysed retrospectively. The prediction models were validated using a calibration plot analysis with the intercept and slope. The discrimination was evaluated using area under the curve (AUC) analysis. The observed two year overall survival (OS) was evaluated by the Kaplan - Meier method. The two year OS predicted by each model at < 50%, 50 - 70%, and > 70% was defined as high, medium, and low risk, respectively.

RESULTS

A total of 491 patients who underwent infra-inguinal revascularisation were analysed. The rates of surgical revascularisation, endovascular therapy, and hybrid therapy were 26.5%, 70.1%, and 5.5%, respectively. The average age was 75.6 years, and the percentages of patients with diabetes mellitus and dialysis dependent end stage renal disease were 66.6% and 44.6%, respectively. The tissue loss rate was 85.7%. The intercept and slope were -0.13 and 1.18 for the JCLIMB, 0.11 and 0.82 for the SPINACH, and -0.15 and 1.10 for the VQI. The AUC for the two year OS of JCLIMB, SPINACH, and VQI were 0.758, 0.756, and 0.740, respectively. The observed two year OS rates of low, medium, and high risk using the JCLIMB calculator were 80.1%, 61.1%, and 28.5%, respectively (p < .001), using the SPINACH calculator were 81.0%, 57.0%, and 38.1%, respectively (p < .001), and using the VQI calculator were 77.8%, 45.8%, and 49.6%, respectively (p < .001).

CONCLUSION

The JCLIMB, SPINACH, and VQI survival calculation models were useful, although the OS predicted by the VQI model appeared to be lower than the observed OS.

Artificial intelligence-based predictive models in vascular diseases

Cardiovascular disease represents a source of major health problems worldwide, and although medical and technical advances have been achieved, they are still associated with high morbidity and mortality rates. Personalized medicine would benefit from novel tools to better predict individual prognosis and outcomes after intervention. Artificial intelligence (AI) has brought new insights to cardiovascular medicine, especially with the use of machine learning techniques that allow the identification of hidden patterns and complex associations in health data without any a priori assumptions. This review provides an overview on the use of artificial intelligence-based prediction models in vascular diseases, specifically focusing on aortic aneurysm, lower extremity arterial disease, and carotid stenosis. Potential benefits include the development of precision medicine in patients with vascular diseases. In addition, the main challenges that remain to be overcome to integrate artificial intelligence-based predictive models in clinical practice are discussed.

A systematic review of patient-reported outcome measures patients with chronic limb-threatening ischemia

Chronic limb-threatening ischemia (CLTI) causes significant morbidity with profound negative effects on health-related quality of life. As the prevalence of peripheral artery disease and diabetes continue to rise in our aging population, the public health impact of CLTI has escalated. Patient-reported outcome measures (PROMs) have become common and important measures for clinical evaluation in both clinical care and research. PROMs are important for the measurement of clinical effectiveness and cost effectiveness and for shared decision-making on treatment options. However, the PROMs used to describe the experience of patients with CLTI are heterogeneous, incomplete, and lack specific applicability to the underlying disease processes and diverse populations. For example, certain PROMs exist for patients with extremity wounds, and other PROMs exist for patients with pain, and still others exist for patients with vascular disease. Despite this multiplicity of tools, no single PROM encompasses all of the components necessary to describe the experiences of patients with CLTI. This significant unmet need is evident from both published reports and contemporary large-scale clinical trials in the field. In this systematic review, we review the current use of PROMs for patients with CLTI in clinical practice and in research trials and highlight the gaps that need to be addressed to develop a unifying PROM instrument for CLTI.

Applicability of the Vascular Quality Initiative (VQI) mortality prediction model for infrainguinal revascularization in a tertiary limb preservation center population

OBJECTIVE

Patients undergoing revascularization for chronic limb-threatening ischemia (CLTI) are at elevated risk for both mortality and limb loss. To facilitate therapeutic decision-making, a mortality prediction model derived from the Vascular Quality Initiative (VQI) database has stratified patients into low, medium, and high risk, defined by 30-day mortality estimated of ≤3%, 3-5%, or >5% and 2-year mortality estimates of ≤30%, 30-50%, or ≥50%, respectively. The purpose of this study was to compare expected mortality risk derived from this model with observed outcomes in a tertiary center.

METHODS

Consecutive patients treated at a single center between 2016 and 2019 were analyzed. Baseline demographics, approach, and mortality events were reviewed. Observed mortality was obtained using life-table methods and compared using a log-rank test with the expected mortality risk which was calculated using the VQI model.

RESULTS

This study cohort consisted of 195 revascularization procedures in 169 unique patients stratified into 128 (66%) low, 50 (26%) medium, and 17 (8%) high-risk cases based on the VQI model. 90% of revascularizations were performed for tissue loss. Compared with the VQI population, comorbidities were prevalent and included unstable angina or myocardial infarction within 6 months (6% vs. 2.4% in VQI; p<0.001), congestive heart failure (30% vs. 23%; p<0.001), and dialysis dependence (14% vs. 0.9%; p<0.001). Patients were also older (31% vs. 21% ≥80 years old; p<0.001) and more likely to be frail (45% vs. 64% independent; p<0.001). High-risk patients were more prevalent in the endovascular group (11% of 132 endovascular interventions vs. 3% of 63 bypasses; p=0.056). 30-day observed mortality exceeded expected VQI prediction model mortality in all groups, although was not statistically significant. The VQI model adequately stratified the studied population into risk groups (p<0.001). Low risk CLTI patients (65% of the overall cohort) experienced 2- year mortality of 18.9%. However, observed mortality for medium and high-risk VQI strata were similar. After a median follow-up of 28 months, medium-risk patients incurred a significantly higher mortality than predicted (53.5%±2.1% vs. 36.8%±1.1%; p=0.016).

CONCLUSIONS

The VQI mortality prediction model discriminates mortality risk after limb revascularization in CLTI, accurately identifying a majority subgroup of patients who are suitable for either open or endovascular intervention. However, it may underestimate mortality in a tertiary referral population with high comorbidity burden and was not well calibrated for the medium-risk group. It may be more appropriate to dichotomize CLTI patients who are candidates for limb salvage into an average risk and high-risk group.

External validation of the Vascular Quality Initiative prediction model for survival in no-option chronic limb-threatening ischemia patients

OBJECTIVE

Chronic limb-threatening ischemia (CLTI) is associated with high morbidity and mortality rates. More than 50% of all CLTI patients die within 5 years after presentation. Patient-specific survival prediction is critical for informing treatment strategies, even for those without a clear option for revascularization. We validated a survival prediction model, developed in a revascularized Vascular Quality Initiative (VQI) cohort, in a Western European no-option CLTI cohort.

METHODS

The VQI survival prediction model was applied to the validation cohort (N = 150) to compare estimated mortality and observed mortality at 2 years after baseline. Performance of the VQI model was tested by evaluating discrimination using the receiver operating characteristic area under the curve and calibration using the Hosmer-Lemeshow goodness-of-fit test.

RESULTS

The 2-year survival rate was 79% in the validation cohort compared with 83% in the VQI cohort. Baseline characteristics were significantly different for 13 of 17 variables. The C statistic was 0.86 (95% confidence interval, 0.78-0.95), which indicates good discrimination. The Hosmer-Lemeshow goodness-of-fit test had a P value of .30, which indicates good fit.

CONCLUSIONS

This is the first external validation of the VQI survival prediction model. The good model performance suggests that this model can be used in different CLTI populations, including no-option CLTI, and underlines its contributory role in this challenging population.