This exploratory pilot study demonstrates the potential for re-stratification of CAD presence and severity using CCTA in a high cardiovascular risk population when compared to the current standard of care, the CCS, alongside its potential impact on clinical management. This is also the first study to measure FAI in a dedicated lipid clinic cohort.

Whilst dyslipidaemia promotes premature atherosclerosis and associated MACE, this risk is heterogeneous, even within the same sub-group of dyslipidaemia. For example, the severity and clinical expression of FH is known to vary even within a family sharing the same pathogenic mutation [37]. Risk factors beyond prolonged exposure to high levels of LDL play a role. Cardiac CT is increasingly recognised as offering a simple quantification of CAD that integrates the interplay of these cumulative risk factors into a single, reproducible and actionable measure of risk [38]. In keeping with this, contemporary AI-enabled quantitative CCTA now allows automated plaque burden and composition measurements that have been linked to outcomes and, importantly, show sex-specific prognostic differences [39].

CCS is currently first-line for asymptomatic patients in guidelines. However, CCS misses non-calcific plaque, lower volume calcific plaque below the threshold for reporting in the Agatston protocol, HRP features and cannot grade stenosis. CCS therefore has no role in the assessment of symptomatic patients, in whom CCTA is first-line [11, 29, 40]. Even for patients with zero CAC, the presence of non-calcified plaque, HRP features and stenosis on CCTA is prognostically superior to CAC testing and has clinical implications for the timing and intensity of preventive therapies [41, 42]. In addition, a recent meta-analysis indicates that CAC density (distinct from score or volume) is inversely associated with cardiovascular risk after adjustment for traditional risk factors and CAC burden, suggesting that denser calcification may reflect more stable plaque biology and should be interpreted alongside (not instead of) total CAC [43].

The SCOT-HEART study tested CCTA as the first-line investigation of stable chest pain versus standard of care [44]. This demonstrated the prognostic benefit of a CCTA strategy, and the largest proportional reduction in non-fatal MI or death from CAD was observed in patients with non-anginal chest pain [45]. For many individuals this may represent incidental identification of CAD that did not account for their non-cardiac symptoms.

In our study of asymptomatic lipid clinic patients, CAC (of any degree) was observed in 64% (29/45) of patients. A subsequent CCTA re-classified CAD severity in a significant proportion of patients (62%), including newly diagnosing CAD in 22% (10/45) of patients with zero CAC. This suggests a potential underestimation of CAD risk when using CCS alone, which may have important clinical implications for risk stratification and management in lipid clinics. Similar findings were observed in the PROMISE and SCOT-HEART studies, where 16% and 17% of patients respectively, had evidence of coronary atherosclerosis on CCTA despite a CCS of zero [41, 42]. The slightly higher proportion in our study (22%) may reflect the higher cardiovascular risk cohort seen in the lipid clinic. Indeed, high Lp(a) was recently shown to accelerate plaque burden and increase both non-calcific plaque and FAI [46]. Moreover, findings from the CONFIRM2 Registry show that quantitative plaque phenotypes carry a stronger relative association with MACE in women than in men, reinforcing the need to look beyond CCS alone when evaluating high-risk female patients [39].

HRP features were identified in 20% of the cohort, present across all lipid diagnoses. HRP was observed in 23% of all patients with a CCS in the zero to mild range (< 100) – patients who would otherwise have been considered lower risk. Identification of subclinical CAD infers support for informed decision-making regarding the use and intensity of preventative medications and therapeutic lifestyle changes by patients. Its impact on outcomes does though require assessment in a prospective, longer-term study. This aligns with recent work showing that premature CAD is characterised by a higher prevalence of non-calcified plaque and multiple HRP features on CCTA compared with incidental plaques in matched controls, highlighting the aggressive, composition-heavy phenotype in younger patients [47].

Though the absolute numbers were low, the presence of a high-intensity statin in all patients with a CCS > 400 where CAD RADS was 1–2 vs. 60% of those where CAD RADS was 3–4 may highlight the need to be cautious in over-interpreting the CCS in patients taking high-intensity statins.

As the first study to quantify the impact of cardiac imaging on the hypothetical management of a real-world lipid clinic cohort, we observed a lowering of the LDL target selected with the incremental data provided by both a CCS and CCTA. International guidelines advise that in asymptomatic individuals with higher CVD risk (including with Agatston score >100 AU), the use of CCTA assessment may reclassify an individual to a higher risk category [7, 11, 48]. We demonstrated a reduction in the hypothetical median LDL target selected with CCTA, even after clinicians had been unblinded to the CCS. Indeed, even in patients with severe CAC (i.e. CCS >400 AU), there was a further lowering of median target LDL with CCTA results.

In the context of the study size and hypothetical nature of the management selected, caution should be applied in over-interpreting the potential clinical impact of CCTA, with study findings serving as hypothesis-generating. Indeed, whilst previous randomised trial data have demonstrated the ability for CCS to significantly re-stratify CVD risk [49], to date this has not been demonstrated with CCTA in asymptomatic patients. In the CONFIRM registry CCTA did not significantly re-stratify risk above CCS [49]. However, the imaging landscape and CCTA techniques continue to evole. The multicentre CONFIRM2 Registry reported significant, sex-differential prognostic associations for quantitative plaque features, suggesting that modern techniques may provide risk information not captured by legacy analyses [39]. This warrants prospective validation in higher-risk, asymptomatic cohorts such as lipid clinic populations.

Increasing focus is placed on recognising patients with proven atherosclerosis as at higher risk of future MACE. European guidelines recommend both the consideration of cardiac imaging to support enhanced risk stratification, and more aggressive targets and treatment options when ASCVD is present. However, these recommendations differ across UK, European and American lipid guidance [50]. In addition, the interpretation of primary versus secondary prevention guidelines in the setting of an asymptomatic patient with proven atherosclerosis but no history of established ischaemic or cerebrovascular disease may vary amongst clinicians. As such, there was significant variation in management selected, demonstrated by the low level of agreement on LDL targets amongst respondents. Given the sex-specific quantitative plaque data and the protective signal of higher CAC density, harmonising guidance on how to integrate plaque composition and calcification density into LDL-target selection may reduce practice variability [43].

Of note, agreement was graded as poor even after the clinical vignette alone. There are a variety of potential factors that may have influenced this finding, including insufficient prior respondent exposure to CTCA results or the potential for ambiguous case vignettes. A key consideration however is whether variability reflects the differing use of guidelines and varying interpretations of cardiovascular risk stratification. If agreement amongst experts in this field is poor, this suggests a need for greater consensus across national and international guidance to improve consistency in patient management. However, the validity of these insights is limited by the hypothetical nature of the survey responses and the small number of respondents, which may not represent broader clinical practice.

The level of clinician agreement also deteriorated with each layer of cardiac imaging. We attempted to improve consistency in the interpretation of CT findings with the provision of educational materials in advance of survey completion (supplementary materials). However, the reduction in agreement may still reflect both variation in clinician’s prior exposure to cardiac imaging and interpretation of results in the context of the differing guidelines available. There is also debate as to what constitutes sufficient plaque within the coronary tree, or indeed across multiple vascular beds, to warrant a lowering of LDL targets and escalation in treatment. Recently published age- and sex-specific nomograms for quantitative plaque volumes provide reference percentiles derived from a large, international cohort and may help standardise interpretation of plaque burden [51].

The variation in baseline LDL target set may impede interpretation of clinician response to subsequent cardiac imaging. Allowing for this and despite the reduction in level of agreement across respondents, there remained a statistically significant decrease in LDL target set with CCTA results even after clinicians had been provided with a CCS. The increasing use of non-invasive cardiac imaging appears to offer an opportunity to differentiate high and very high cardiovascular risk in populations such as those seen in the lipid clinic. Demonstrable evidence of CAD may also improve patient adherence to treatment. These hypotheses require testing in a prospective, randomised study.

As demonstrated in this study, however, there may be a down-stream impact on other services. With increasing burden of CAD detected there were rising cardiology referrals considered. Whilst this may be appropriate for some cases, increased use of cardiac imaging would require a broader discussion around the appropriate investigation and management of asymptomatic CAD, particularly in light of recent evidence [52].

Whilst studies have assessed the role of cardiac imaging and burden of subclinical atherosclerosis in FH, there has been less focus on other dyslipidaemias associated with heightened risk of MACE. Our study observed the potential for cardiac imaging to play a role in the re-stratification of cardiovascular risk across conditions seen in the lipid clinic. Whether this translates into changes in clinical outcomes was beyond the scope of this study.

Recent work has suggested a low short-term MACE risk of heterozygous FH patients with a CCS of 0 [53]. However, this study was limited by a relatively short follow-up period (particularly relevant given the younger age of the population studied) and overall low event rates. Indeed, in the REFERCHOL study, patients with heterozygous FH and no prior history of CVD had a MACE rate of 9.4% at 5 years [53]. This compares with our findings of 9% at just over 4 years, and 75% (3/4) of these had a CCS graded mild or moderate. Further data on longer-term risk for patients assessed with CCTA is required given this patient group’s elevated lifetime risk for MACE. However, as suggested in a clinical practice statement from the American Society for Preventive Cardiology, “as we await studies, a judicious approach to use of CCTA in asymptomatic populations is to target high-risk populations that are currently missed by traditional ASCVD risk factor scoring” [54].

This study demonstrates the potential for further re-stratification of cardiovascular risk in this cohort with layered FAI analysis (available from CCTA). FAI-score is a sensitive biomarker of coronary inflammation known to enhance cardiovascular risk prediction above current standard of care [24], though it should be noted FAI remains an investigational tool in this context requiring prospective validation. In a recent study FAI-score was shown to be an accurate predictor of both cardiac mortality and MACE, independent of other cardiovascular risk factors and the presence or extent of CAD [25]. Indeed, the predicted risk correlated almost perfectly with observed risk across a diverse cohort of patients with varying pretest probabilities, including patients with minimal or non-obstructive CAD (i.e. cases without typical angina from stenotic epicardial disease).

In our study, FAI-score re-stratified the cohort relative to both CCS and CCTA across all lipid diagnoses (Table 2). Interestingly, high FAI-score was observed in 22% of patients with none to mild calcification on CCS (< 100 AU) and 21% of patients with none to mild CAD when defined by stenosis on CCTA (CAD-RADS 0–2). The proportion with a high FAI-score was greater in patients with a CCTA demonstrating no plaque than those with a severe stenosis. It may be that the incremental risk stratification provided by FAI analysis can play a role in the treatment decisions for patients with a lower burden of a disease on baseline imaging. However, the assessment of this is limited by the sample size of the study and requires further work including longer-term outcome data.

FAI is modifiable with treatment. In a sub-analysis of the CRISP-CT study, FAI did not retain its significant association with subsequent MACE in patients who received initial treatment with statins or aspirin after CCTA [24]. These findings were confirmed by other groups [27] who found the perivascular FAI to be a dynamic tool for monitoring response to statin treatment, demonstrating a significant reduction in the pericoronary FAI around non-calcified and mixed plaques but not around calcified plaques. FAI may therefore identify patients with persistent higher risk not responding to treatment, or those with poor adherence.

In our study, high FAI-score was observed in 17% of patients on treatment compared with 33% of patients on no treatment at the time of their imaging. We did not find an association between either any statin or a high-intensity statin and FAI, whilst the low number of patients on PCSK9-inhibitors (2) at the time of CT prevented analysis. The impact of treatment dictated by FAI-score requires further interrogation in a larger cohort tracking MACE.

This exploratory pilot study is limited by its single-centre and retrospective design, limiting the ability to control for confounders. The modest sample size and heterogeneity of the population limit interpretation of findings, particularly sub-analyses, though offer exploratory, hypothesis-generating data. The patient cohort reflects clinician’s discretion as to which patients cardiac imaging were considered in, rather than a random process or systematically applied to all patients, leaving the potential for selection bias. This, combined with the modest sample size and single-centre design may limit the representativeness of the cohort and generalisability of our findings to differing lipid clinic populations.

Allowing for this though, study results demonstrate that the current methods of estimating risk are heterogeneous. The number of raters was modest, and the assessment of impact on management was based on hypothetical decision-making rather than the prospective decisions taken in each case, which may not represent broader clinical practice. However, the use of real-world consecutive clinical cases referred for cardiac CT enhances its application to the lipid clinic and reflects current practice. Further, the involvement of clinicians from a variety of institutions across the UK demonstrates a sampling of clinical practice and adds weight to findings.

The relatively large number of sequential responses required within the survey may have risked a degree of response-fatigue. This was mitigated by allowing respondents to pause, exit and return at the same point, completing the survey over multiple sittings at a time to suit them. A potential lack of clinician routine exposure to CCTA prior to participation results may have influenced responses, though this was partially addressed by the provision of standardised supplemental educational material to all respondents and reflects current practice. Further research will need to consider the health economic and longer-term clinical impact of CCTA +/- FAI versus standard of care in larger, prospective multi-centre cohort studies.