For this EVA, we have identified only one study evaluating CaRi-Heart Risk: Oikonomou EK, Antonopoulos AS, Schottlander D, Marwan M, Mathers C, Tomlins P, et al. Standardized measurement of coronary inflammation using cardiovascular computed tomography: integration in clinical care as a prognostic medical device. Cardiovasc Res 2021;117(13):2677–90.
The focus of the evidence review is, therefore, consideration of the extent to which this study addresses the clinical question defined at scope. As part of this process, we would like to request your input with respect to the ‘appropriateness’ of the variables (additional to FAI score) included in the CaRi-Heart Risk model, and the ‘standard care’ method of risk assessment to which it is compared, specifically:
The above publication describes the CaRi-Heart Risk model as incorporating (in addition to FAI score), atherosclerotic plaque burden (as described by the modified Duke CAD index), diabetes, smoking, hyperlipidaemia and hypertension.
When considering cardiac risk, in patients who are undergoing CTCA for the investigation of suspected CAD:
Are there any additional clinical risk factors (other than diabetes, smoking, hyperlipidaemia and hypertension) that you would routinely consider? Please list any additional clinical risk factors that you consider form part of standard care for risk assessment.
What imaging parameters (available from current standard CTCA) would routinely be reported/considered as part of standard care for risk assessment?
‘Easy answer to the first question. There are a number of other risk factors which are on the QRISK®3 calculation. They include heart attack in first degree relative < 60, chronic kidney disease, BMI, severe mental illness, use of antipsychotic drugs, atrial fibrillation and steroid use. Of these a very strong family history of premature coronary disease is a particularly potent risk factor (genetics)’.
The imaging question you post is relevant. In the UK cardiac imaging is generally performed when patients present with chest pains etc. rather than as a risk assessment tool. I suspect it may have been different in the US cohort in the Cardiovac Research paper (2021). It would be a big leap for general practitioners (GPs) to go from QRISK®3 scoring to sending hundreds of thousands of patients for CT scans that they wouldn’t normally be considered for.
In the Cardiovasc Res (2021) paper, the cardiac death rates were very low at 1.4% over 6 years – this comes to 0.23% per year in the European cohort (i.e. 1 in 450 chance of dying per year). I can’t see any details of the mode of death and in particular whether this was AMI.
For me there are a couple of missing pieces in the jigsaw. (1) Do the people with markers of inflammation in the coronary arteries also have inflammation in their abdominal fat (i.e. is this a systemic effect which is a marker of bad health – potentially related to kidney disease, obesity, mental illness etc. … all of which have mortality implications). (2) They must have looked for people with an inflamed RCA being admitted with a heart attack due to a blocked RCA. I can’t see any data on FAI predicting a heart attack in a specific artery.
Would be interested in others’ views’.
‘The patients who have a CTCA are largely going to be referred for investigation of chest pain. I agree with the comments about Q Risk 3, but these clinical risk models in general overestimate risk. The proposed CaRi-Heart score incorporates both the most important clinical risk factors and CT imaging markers, including information about the atherosclerotic plaque burden, and the fat attenuation index. As was described at our last meeting this is a ‘black box’ and we don’t know the contribution of each of these components. However, this score does outperform the clinical model and the outcome is death. Q-risk is MI or stroke risk rather than death’.
In answer to the specific questions:
In patients undergoing CTCA I suspect that if there is no disease evident and the CTCA is ‘normal’ then treatment would be guided by standard guidelines including Q risk assessment by the GP. If plaque disease is present, then most recommend aspirin and statin, with attention to other cardiac risk factors. I don’t think a risk score would change this recommendation in the presence of anatomical plaque. One key question is whether the CaRi-Heart score can improve this stratification perhaps most importantly in that large cohort with ‘normal’ CTCA because those with plaque are going to be treated anyway. That population with ‘normal’ CTCA may benefit from refined risk assessment.
Atherosclerotic plaque burden is reported – usually in a subjective way. Although there is probably variation in practice, it would be interesting to hear views on how this is used by colleagues. It doesn’t directly alter my approach to recommending treatment in the presence of anatomical disease, recognising that those with more plaque are likely to do worse, and if we could better stratify that risk it might be helpful – although I don’t know what we would do differently at this stage other than aspirin, high-dose statins and addressing the other modifiable risk factors.
‘I am less concerned than Gerald about the vessel specific prediction of event by FAI, and I think that one of the key drivers for the improved risk assessment maybe of the CaRi-Heart score is that the anatomical extent of disease is incorporated into the black box model, and we may not be able to separate those components’.
‘Replying from the perspective of a radiologist reporting CTCA:
(1) What imaging parameters (available from current standard CTCA), would routinely be reported/considered as part of standard care for risk assessment?
In addition to severity of stenosis and length of stenosis, I report the position of the lesion (e.g. left main stem and proximal LAD are particularly important) and whether it is fully calcified/mixed density/soft tissue density (i.e. can be induced to calcify with statins). The overall plaque burden is subjectively reported (i.e. overall amount, what proportion of vessels, age is taken into account in the emphasis – the younger with disease being more worrisome). Whether the lesion has features of vulnerability (aka napkin ring sign) or an obvious dissection is apparent, or whether a lesion appears more as vessel wall irregularity or thickening with perivascular fat stranding – i.e. implying there is localised vasculitis. Whether there is an anatomical variation putting the patient at more severe risk from a particular plaque is commented up if present.
We use HeartFlow for all lesions subjectively assessed as moderate or greater in severity of stenosis. Utilising computational fluid dynamics modelling to compute an estimate of lesions fractional flow reserve (CT-FFR), this in effect assesses whether the stenoses are tight and long enough to cause ‘flow limiting disease’ that is likely to be symptomatic. We also see a falloff in CT-FFR in distal vessels which is currently thought of equivocal significance but may possibly represent otherwise unmeasurable microscopic diffuse CAD. (One thought is that since HeartFlow also estimates myocardial volume for each vascular territory, what it might be reflecting is whether a vessel is large enough to supply that amount of myocardium.)
Note that the widely (but not universally) used CADRADs system of reporting CTCA plaque severity now includes in it’s recent ‘2.0’ update scoring of overall plaque burden.
https://pubs.rsna.org/doi/10.1148/ryct.220183
(2) Are there any additional clinical risk factors (other than diabetes, smoking, hyperlipidaemia and hypertension) that you would routinely consider? Please list any additional clinical risk factors that you consider form part of standard care for risk assessment.
Gerald’s list of risk factors as implied by the QRISK®3 calculator is quite reasonable and I see these listed in CTCA referrals to me. I would comment anecdotally that we have large numbers of patients with high BMI on our lists, who turn out to have no obvious CAD, whereas all too many thin patients who have lots of disease. I suspect BMI as a risk factor is more about downstream demand upon the heart rather than not of CAD itself. Furthermore, genetics is not the only reason for the familial risk factor – lifestyle habits and exposure (both food and pathogens) frequently have commonality in families.
I find the comments about RCA
FAI and inflammation elsewhere (e.g. abdominal vasculature) intriguing. One increasingly important set of risk factors perhaps not yet fully recognised to be put on the CTCA request history by the referrers (i.e. cardiologists at my hospital) are the systemic inflammatory diseases. I see plenty of referrals with ‘COVID’, but are yet to see any with psoriasis, rheumatoid arthritis and gout which are growing in interest as risk factors for CAD due to their possible vascular aetiologies. (Many systemic diseases like these may however be reflected in the ‘use of steroids’ risk factor.)
www.ncbi.nlm.nih.gov/pmc/articles/PMC7462628/
There is a chicken and egg question here. Do the perivascular fat changes represent a response to CAD, or does it represent a measurable feature that is a precursor to the actual development of plaque? Perhaps both. If the latter is true – CaRi-Heart therefore could potentially be useful in identifying patients where a treatment could be aimed at settling the inflammation before it develops into (initially soft tissue) coronary arterial plaques. Statins are thought to convert soft-tissue plaques more quickly into calcified plaques, which is not a cure but a mitigation hoping to reduce the risk of future plaque rupture. For those of the panel who were at the BSCI conference in Bath: – anti-inflammatory medications such as colchicine (used to treat gout flare-ups) were mentioned as potential anti-inflammatory treatment.
Lastly CaRi-Heart does not have to be just applied to CTCA scans. Having worked with a GE revolution scanner which was being used principally for A&E traumas, the newer generation of scanners is producing remarkably good images of the heart (although not intentionally) when visualising the thorax for other reasons. Just as the BSCI has previously advised a review of the degree of calcification of coronary arteries when the thorax is imaged (under the premise that young with calcified disease are at increased risk of future coronary events, whereas the old with none are conversely at much lower risk), CaRi-Heart may potentially be a useful way of screening patients for risk of developing plaques in future from pre-existing scans’.
www.birpublications.org/doi/10.1259/bjr.20200894
‘I’m answering from a radiologist’s perspective so will try not to stray too far into territory outside my expertise! And note that some good points have been made already.
When considering cardiac risk, in patients who are undergoing CTCA for the investigation of suspected CAD:
Are there any additional clinical risk factors (other than diabetes, smoking, hyperlipidaemia and hypertension) that you would routinely consider? Please list any additional clinical risk factors that you consider form part of standard care for risk assessment.
As per comments below there are various additional risk factors and scores which are relevant although still relatively crude. Going forward, I think more accurate risk prediction is going to play a greater role in clinical medicine. Moving towards personalised medicine in an increasingly multimorbid population with ever more treatments and interventions available could provide benefits to individuals and society but requires such prediction models to be better tailored to the individual. An interesting paradigm in medicine is the need to have better diagnostic tools to assess the impact of novel treatments and interventions; therefore you often can’t have one without the other and these develop in parallel. Having said that, not all diagnostic investigations will find a role beyond a research setting if they don’t have a use on a patient-by-patient basis.
What imaging parameters (available from current standard CTCA) would routinely be reported/considered as part of standard care for risk assessment?
As Rob said, practice is heterogeneous and will vary by centre and individual expertise. Coronary calcium score has been best validated as an additional risk factor in an asymptomatic population and not routinely performed in patients referred for chest pain assessment. CADRAD-2 is probably the best ‘template’ for a comprehensive CT coronary angiogram report but recently described and therefore use won’t be widespread. Even where it is in use, much of the risk is subjective and we know that interobserver variability for many imaging findings is generally poor and interpretation subject to bias. Standardisation is therefore limited in this context. HeartFlow is useful for predicting whether anatomical stenoses are causing symptoms and can help determine which patients may benefit from revascularisation but doesn’t predict the overall vascular risk. I suspect in some centres the CT report will focus mainly on functionally significant stenosis and role for interventions rather than a more holistic view of overall burden of atheroma. Again, not standard of care and I’m not sure what is commercially available, but software to assess high-risk plaques may also have a role for risk prediction.
Finally, as we discussed at the meeting, there may be scope for the use of CaRi-Heart outside the population and parameters studied, for example asymptomatic patients with some risk factors, acute chest pain presentations, non-cardiac gated studies, but obviously evidence is not currently available in these patients/settings’.
When considering cardiac risk, in patients who are undergoing CTCA for the investigation of suspected CAD:
Are there any additional clinical risk factors (other than diabetes, smoking, hyperlipidaemia and hypertension) that you would routinely consider? Please list any additional clinical risk factors that you consider form part of standard care for risk assessment.
The advantage of Qrisk over Euroscore (predicts mortality) etc. is the inclusion of post code which includes IMD status. In Scotland they use ASSIGN.
I believe the initial CRISP study publication compared to standard risk model and found a small incremental benefit in AUC for using the FAI model.
(Oikonomou EK, Marwan M, Desai MY, Mancio J, Alashi A, Hutt Centeno E, et al. Non-invasive detection of coronary inflammation using computed tomography and prediction of residual cardiovascular risk (the CRISP-CT study): a post-hoc analysis of prospective outcome data. Lancet 2018;
392
:929–39.)
However, this improved AUC was lost if you gave the patient a statin.
So the real Q is how Cari-Heart helps in the 30% that you are not going to advise a statin (normal coronaries). The metanalysis Mani sent around touches on this’.
What imaging parameters (available from current standard CTCA) would routinely be reported/considered as part of standard care for risk assessment?
‘The risk parameters include:
This is ‘best standard of care’ – and should be reported on every CCTA scan report. However, I suspect the majority of report in the UK don’t include this level of detail or nuance. Nor do the people receiving the report understand the nuances. Finally, there is no final % risk given in the report. We can’t do this at the moment. You can say the relative risk is 32× if you have 3 HRP features – but what does that mean?!
I am aware that Caristo have ORFAN running and a NHSE award in 4 trusts so that they may be able to answer many of these Q in the future. They also now are incorporating plaque quantification. However, none of these data are available’.
