Feasibility of cardiovascular population-based CT screening
Vonder, Marleen
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Vonder, M. (2018). Feasibility of cardiovascular population-based CT screening. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
9
General discussion
The results of the research described in this thesis show that cardiovascular population-based CT screening is feasible, but it requires standardization and validation of the dose-reduced CT protocols for accurate screening. We provide a validated coronary artery calcium (CAC) acquisition protocol for low-dose CT screening for cardiovascular disease (CVD) and a workflow for acquiring the CAC imaging biomarker from preparation of the participant up to data analysis of the biomarker. Integrated screening of imaging biomarkers of the Big-3 requires the combination of CT protocols to reduce radiation dose. However, there is subtle balance in maintaining imaging biomarker validity of the Big-3 protocol and reducing the total radiation dose of such a combined acquisition. Radiation dose can also be reduced by optimizing the acquisition and reconstruction parameter settings. We showed that radiation dose can be reduced with 48% in CAC scanning by using high-pitch spiral mode compared to sequential mode in participants with a regular high heart rate. In general, with radiation dose reduction techniques in CAC scanning a dose reduction of 50% or more can be reached, but results in reclassification rates of 3% to 21%. In phantom studies, we could validate that the radiation dose of CAC protocols could be reduced with 60.6% and 73.4% for low tube voltage combined with iterative reconstruction and for spectral shaping with tin-filter respectively. However, risk stratification cannot be validated in phantom studies. In repetitive scanning for CVD screening, we found that progression of CAC can only be concluded if Agatston score increases >12.7% for different generations of dual-cource CT (DSCT) systems. However, in third generation DSCT, progression can already be concluded from >3.4% increase.
Although CAC has been used as an imaging biomarker over more than two decades, there is a wide variety in imaging protocols [Chapter 3]. For electron beam CT (EBCT) studies, acquisition and reconstruction settings were highly standardized, whereas for multi-detector CT (MDCT) or DSCT studies, information about parameter settings was often poorly documented or the protocols revealed differences in parameter settings [Chapter 3]. Besides, the risk stratification categories applied on MDCT and DSCT acquisitions today, were originally used and extensively validated based on EBCT acquisitions [1–3][Chapter 2, 3].
In this thesis we describe the development of a standardized CAC imaging protocol that is used in the ROBINSCA trial on second generation DSCT at various sites. Furthermore, we optimized the CAC imaging protocols by reducing the radiation dose by using a different scan mode. Moreover, we evaluated the potential of other dose reduction techniques to further reduce the radiation dose of CAC scanning. However,
175
GENERAL DISCUSSION
9
wide-spread implementation of CAC as an imaging biomarker for screening would most certainly require site and CT device specific adaption of the imaging protocol as provided in chapter 3. Nevertheless, no clear guidelines exist that can be followed to guarantee the validity of CAC as an imaging biomarker across the different sites and CT devices.
Besides site and CT specific adaptation of the CT protocol, it has become more and more important to reduce the radiation dose to as low as reasonably achievable (ALARA principle). To achieve this, it requires adaptation of the protocol [Chapter 5-7]. Therefore, we performed several validation studies in which we sought to show non-inferiority of dose-reduced protocols compared to conventional full dose protocols. However, many of these studies could only be performed on phantoms, and it is unclear how the results of these studies would be translated into population screening. Consequently, large patient studies to achieve power threshold are needed to show true non-inferiority of such dose reduced protocols as we performed and is described in chapter 4. Nevertheless, these studies with statistically significant results are very time consuming and labor intensive, and therefore in vivo validation of adapted and dose reduced protocols is seldom performed.
Recently, the Quantitative Imaging Biomarker Alliance (QIBA) of the Radiological society of North America (RSNA) has published profiles about Nodule volume assessment and monitoring in CT screening (profile stage 3), see Figure 1. In these profiles the clinical context, profile claims, groundwork and profile details are addressed to the specific imaging biomarker [4]. The next step towards the following profile stage entails development of protocols in conjunction with these profiles. These protocols will clearly describe the standardized imaging procedures to achieve reproducible endpoints on tests performed using systems that meet specific performance claims outlined in these profiles. Unfortunately, no such profiles and protocols exist yet for CAC in CT screening [5]. Nevertheless, for each new protocol the impact on the CAC quantification should be determined. In this thesis we show that the impact on the CAC quantification can be determined on several levels and with various methods. In the next paragraphs, we propose a number of issues that should be evaluated for each adapted or dose-reduced CAC protocol to warrant imaging biomarker validity in CVD screening.
First of all, the impact of an adapted protocol on the number of participants with an Agatston score of zero (impact on zero score) should be determined. Budoff et al. showed in the MESA study that individuals with a zero score have a very low
Imaging biomarker profiling process
Figure 1 – Imaging biomarker profiling process of the Quantitative Imaging Biomarker Alliance (QIBA)
[11]. The final profiles are intended to increase acceptance of quantitative imaging biomarkers by the ima-ging community, clinical trial industry, and regulatory agencies as proof of biological, proof of changes in pathophysiology, and surrogate end-points for changes in the health status of patients. For the imaging bio-marker of lung cancer: Lung nodule volumetry, the profile process is at stage 3, and technical confirmation of results and resolutions are currently ongoing. In this thesis we provide and propose a number of issues that could be incorporated in the future imaging biomarker profile of CAC.
CVD risk. The CVD risk increased 3-fold for individuals with CAC score of 1 to 10, relatively to individuals with a zero score [6]. Besides, in a study by Blaha et al., the authors showed that a zero score resulted in a much larger downward shift in estimated CVD risk than would have been predicted based on ACC/AHA2013 Pooled Cohort Equations [7]. Moreover, even in patients with stable cardiac symptoms, but with a zero score, obstructive CAD is rare and prognosis over the long-term is excellent [8].
Pr oces s s tadi um b y Q IB A: L un g n od ule v ol um et ry Thi s Th esi s: C oro na ry a rter y c alci um
177
GENERAL DISCUSSION
9
Therefore any change in the percentage of a population with a zero score resulting from adaptation of the CT acquisition protocol should be avoided or reported. Hence, in chapter 4 of this thesis we analyzed the number of participants in which the CAC score shifted from a zero to positive score and vice versa for adapted protocols compared to the conventional protocol. Moreover, in phantom studies in which this was not possible, we analyzed the change in number of small calcifications that could be detected as a surrogate of zero score change. Repetitive scanning of phantoms showed some variability in the number of detected calcifications, even for the conventional protocol, with a trend towards increased variability for the adapted protocols [Chapter 6, 7]. Also, the number of detected calcifications decreased for increasing patient size [Chapter 8]. Unfortunately, very few studies report whether an adapted protocol impacts the number of individuals with a zero score [Chapter 5].
Secondly, the impact of an adapted protocol on the Agatston score should be determined. A systematic bias of the Agatston score for an adapted protocol or scanner can lead to systematic over- or underestimation. For instance, in many studies the mean Agatston score of a population scanned with a low dose protocol and with a full dose protocol was compared, and in some other studies the median Agatston score was compared [Chapter 5]. The use of median would be most appropriate, since the Agatston score is not normally distributed in any population. A paired-wise median difference and median absolute difference between two protocols can be calculated to show systematic differences between protocols [Chapter 4].
Thirdly, the impact of an adapted protocol on risk stratification should be determined. A systematic difference in Agatston score could also lead to systematic differences in risk stratification. Even for a population with similar mean or median Agatston score of two protocols, a difference in risk stratification could be present on an individual level. In the ROBINSCA trial we used absolute calcium score cut-offs for risk stratification, rather than percentile age-sex-race/ethnicity percentiles [Chapter 3]. Although risk prediction based on percentile ranking is provided in many CAC software packages, and is still in use in clinical practice by cardiologists, studies by Budoff et al. and Akram et al. showed that absolute Agatston score values predict CHD events better than percentile ranking [9, 10][Chapter 2]. The future results of the ROBINSCA trial will show whether and how effective this risk stratification based on Agatston score is. Moreover, based on those results the impact of reclassification should be determined, and ideally should provide some boundaries for maximum allowed reclassification. Reclassification can be defined as the percentage of individuals that shifts towards a risk category higher or lower based on the adapted protocol, compared to the
reference protocol [Chapter 4]. Unfortunately, many studies examining the impact of an adapted CAC protocol do not report on risk stratification [Chapter 5]. Those studies in which the impact on risk stratification though was analyzed, only reported the agreement in risk stratification between two different protocols. However, even if there is a high agreement (κ≥0.8) present for the total population, a considerable number of individuals might have been reclassified to higher or lower risk categories. Therefore, besides agreement analysis also reclassification rates should be determined. Since the clinical impact of a reclassification rate is still unknown, we propose to keep the difference between the reclassification rate of any new protocol with the reference conventional protocol as low as possible.
Finally, the impact of an adapted protocol on the reproducibility of the Agatston score should be determined. The higher the Agatston score variability of an adapted protocol or scanner, the higher the threshold of calcium increase needs to be before CAC
progression can be concluded. Although CAC progression is associated with coronary
and cardiovascular event rates [11], Radford et al. and Lehmann et al. recently showed that it adds only weakly to risk prediction [12], and the most recent absolute CAC score should be used for risk prediction [13]. Nevertheless, in terms of prognosis, no consensus is yet reached about the warranty period [Chapter 2] with reported repeat scan intervals between 5 to 15 years [14–17]. The impact on the reproducibility of the Agatston score of an adapted protocol and/or the use of a different scanner is also of importance for CVD risk stratification. A systematic bias of a new protocol could lead to either over- or underestimation of CVD risk, but a low reproducibility could also lead to over- or underestimation by random upward or downward shifts of individuals with a Agatston scores near the boundary of a risk category.
Clinical implications & future perspectives
The final outcome of the ROBINSCA trial will show whether population-based screening of high risk CVD by CAC scoring followed by risk reducing treatment can reduce coronary artery disease related morbidity and mortality. Recently all ~13,000 participants have undergone CAC CT scan and study inclusion has been finalized. After the follow-up period of five-years, large-scale implementation might be proven effective if this period provides enough evidence for the net-effectiveness of population-based CT screening for CVD in an asymptomatic population. In that case, clear guidelines and criteria for CAC screening are urgently needed. Besides the need for a CAC imaging biomarker profile, also guidelines regarding follow-up (warranty period) and treatment are then needed and have to be developed.
179
GENERAL DISCUSSION
9
Currently, 12,000 randomly assigned individuals from the Dutch LifeLines cohort are enrolled in a large population-based imaging study: ImaLife. All participants will undergo ultra-low-dose chest CT to evaluate the presence of CAC, lung nodules, bronchial wall thickness and/or emphysema. Contrary to the ROBINSCA and NELSON screening trial in which only participants at elevated risk were included, the ImaLife study aims to assess reference values and prevalence of early imaging biomarkers of the Big-3 diseases in a general population. Because all ImaLife participants are part of the large LifeLines cohort, numerous other clinical and laboratory parameter values of these participants are accessible. Potentially, from the results of the ImaLife study, individuals can be pooled based on the imaging biomarker value, and among the numerous other parameters, new risk factors might be revealed and extracted. These new risk factors can then be used for even more effective selection of people at elevated risk, and thereby improve the eligibility criteria for screening. In this way, population-based screening could become more and more effective.
In conjunction with the ImaLife study, the Netherlands-China Big-3 study (NELCIN-B3) is currently enrolling thousands of participants in China. Of these Chinese participants, similar clinical and laboratory parameters and imaging biomarkers are gathered as of the Dutch participants in the ImaLife study. Thus, big-data of the Western and Chinese population can be used to develop or optimize (imaging) biomarkers to integrate these into personalized health strategies (such as screening) in the general population of the Chinese urban and Western population. To conclude, cardiovascular population-based CT screening is feasible with a standardized and validated imaging biomarker protocol. Combining screening protocols, optimizing scan parameters and using latest generation of DSCT can significantly reduce radiation dose. Before cardiovascular population-based CT screening can be implemented, an imaging biomarker profile for CAC is needed to ensure proper use of dose-reduced protocols.
References
[1] R. Vliegenthart, M. Oudkerk, A. Hofman, et al., Coronary Calcification Improves Cardiovascular Risk Prediction in the Elderly, Circulation. 112 (2005) 572–577. doi:10.1161/ CIRCULATIONAHA.104.488916.
[2] J.J. Carr, J.C. Nelson, N.D. Wong, et al., Calcified Coronary Artery Plaque Measurement with Cardiac CT in Population-based Studies: Standardized Protocol of Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in Young Adults (CARDIA) Study, Radiology. 234 (2005) 35–43. doi:10.1148/radiol.2341040439.
[3] A. Schmermund, N. Lehmann, L.F. Bielak, et al., Comparison of subclinical coronary atherosclerosis and risk factors in unselected populations in Germany and US-America, Atherosclerosis. 195 (2007) e207–e216. doi:10.1016/j.atherosclerosis.2007.04.009.
[4] Quantitative Imaging Biomarker Alliance;, RSNA;, QIBA Profiles and Protocols, (2018). https://www. rsna.org/QIBA-Profiles-and-Protocols/ (accessed April 25, 2018).
[5] Quantitative Imaging Biomarker Alliance;, RSNA, QIBA Committee Organization Chart February 2018, (2018). http://qibawiki.rsna.org/index.php/Committees (accessed April 25, 2018).
[6] M.J. Budoff, R.L. McClelland, K. Nasir, et al., Cardiovascular events with absent or minimal coronary calcification: The Multi-Ethnic Study of Atherosclerosis (MESA), Am. Heart J. 158 (2009) 554–561. doi:10.1016/j.ahj.2009.08.007.
[7] M.J. Blaha, M. Cainzos-Achirica, P. Greenland, et al., Role of Coronary Artery Calcium Score of Zero and Other Negative Risk Markers for Cardiovascular Disease, Circulation. 133 (2016) 849–858. doi:10.1161/CIRCULATIONAHA.115.018524.
[8] T.K. Mittal, A. Pottle, E. Nicol, et al., Prevalence of obstructive coronary artery disease and prognosis in patients with stable symptoms and a zero-coronary calcium score, Eur. Hear. J. - Cardiovasc. Imaging. 44 (2017) 1–8. doi:10.1093/ehjci/jex037.
[9] M.J. Budoff, K. Nasir, R.L. McClelland, et al., Coronary Calcium Predicts Events Better With Absolute Calcium Scores Than Age-Sex-Race/Ethnicity Percentiles. MESA (Multi-Ethnic Study of Atherosclerosis), J. Am. Coll. Cardiol. 53 (2009) 345–352. doi:10.1016/j.jacc.2008.07.072.
[10] K. Akram, S. Voros, Absolute coronary artery calcium scores are superior to MESA percentile rank in predicting obstructive coronary artery disease, Int. J. Cardiovasc. Imaging. 24 (2008) 743–749. doi:10.1007/s10554-008-9305-5.
[11] Quantitative Imaging Biomarker Alliance;, RSNA;, Profiling Process, (2018). http://qibawiki.rsna.org/ index.php/Process (accessed April 25, 2018).
[12] M.J. Budoff, R. Young, V.A. Lopez, et al., Progression of Coronary Calcium and Incident Coronary Heart Disease Events, J. Am. Coll. Cardiol. 61 (2013) 1231–1239. doi:10.1016/j.jacc.2012.12.035. [13] N. Lehmann, R. Erbel, A.A. Mahabadi, et al., Value of Progression of Coronary Artery Calcification for
Risk Prediction of Coronary and Cardiovascular Events, Circulation. 137 (2018) 665–679. doi:10.1161/ CIRCULATIONAHA.116.027034.
181
GENERAL DISCUSSION
9
[14] N.B. Radford, L.F. DeFina, C.E. Barlow, et al., Progression of CAC Score and Risk of Incident CVD, JACC Cardiovasc. Imaging. (2016). doi:10.1016/j.jcmg.2016.03.010.
[15] J.H. Lee, D. Han, B. ó Hartaigh, et al., Warranty Period of Zero Coronary Artery Calcium Score for Predicting All-Cause Mortality According to Cardiac Risk Burden in Asymptomatic Korean Adults, Circ. J. 80 (2016) 2356–2361. doi:10.1253/circj.CJ-16-0731.
[16] V. Valenti, B. ó Hartaigh, R. Heo, et al., A 15-Year Warranty Period for Asymptomatic Individuals Without Coronary Artery Calcium, JACC Cardiovasc. Imaging. 8 (2015) 900–909. doi:10.1016/j. jcmg.2015.01.025.
[17] A.J. Taylor, M. Cerqueira, J.M. Hodgson, et al., ACCF/SCCT/ACR/AHA/ASE/ASNC/NASCI/SCAI/ SCMR 2010 Appropriate Use Criteria for Cardiac Computed Tomography, J. Cardiovasc. Comput. Tomogr. 4 (2010) 407.e1-407.e33. doi:10.1016/j.jcct.2010.11.001.
