Subclinical Atherosclerosis Is Associated With Incident Atrial Fibrillation

A Systematic Review and Meta-analysis

Kit Engedal Kristensen; Cille Cederholm Knage; Liv Havgaard Nyhegn; Bart A. Mulder; Michiel Rienstra; Isabelle C. Van Gelder; Axel Brandes


Europace. 2020;22(7):991-1000. 

In This Article


The findings in this systematic review and aggregate level meta-analysis provide evidence that subclinical atherosclerosis identified through measuring cIMT by carotid ultrasound or CACS by non-contrast computed tomography is significantly associated with the incidence of AF.

To our knowledge, this is the first systematic review on this topic. At present, there are only few studies available. In particular, there are very few studies describing the association between CACS and incident AF. The findings from this study are consistent with findings from previous cross-sectional studies and a previous meta-analysis, which all measured cIMT at baseline.[12,18,19] The meta-analysis by Chen et al. processed data from the ARIC, MESA, and Rotterdam studies and found an overall HR of 1.37 (95% CI 1.17–1.61), which is comparable to the overall HR in the present study.[12]

In contrast to our analysis, the population-based cohort Cardiovascular Health Study (CHS), found no significant association between cIMT and incident AF.[20] This study was not included in this systematic review and meta-analysis, because no data on cIMT measurements were reported. The differences between the findings from CHS compared to the other population-based cohort studies included in this systematic review might be explained by the fact that the population in the CHS cohort was older and that self-reported cases of AF were included.

There are other parameters for identification of subclinical atherosclerosis beyond cIMT and CACS, e.g., the ankle-brachial index and maximum carotid plaque height. A low ankle-brachial index (<1.0) was an independent risk factor for AF.[21,22] Several cohort studies also found a significant association between maximum carotid plaque height and incident AF.[12] These findings also confirm the association of subclinical atherosclerosis with the development of incident AF.

The present analysis included two studies measuring CACS. The study by Vinter et al.[16] showed a weaker association between increased CACS and incident AF than the study by O'Neal et al.[13] The variation of these results might be explained by the differences between the two studies. Vinter et al. investigated a larger and younger population with a higher proportion of women, while O'Neal et al. had a longer follow-up period. In addition, the population in the study by Vinter et al. consisted of individuals, which were suspected of CAD and, thus, do not represent the general population. Although this is conflicting with our aim to investigate subclinical atherosclerosis, this study was nonetheless included, because all individuals were proven free of significant CAD. The above-mentioned differences between the two studies could possibly result in an unequal confounder distribution.

The overall HR for cIMT was based on the results from five studies.[11,12,14,15] Results from the MESA cohort study showed the strongest association between increased cIMT and incident AF, and the results from the ARIC cohort study showed the weakest association.[12] In both studies, the HR was calculated comparing the 1st and the 5th quintile, but there were some differences. The population of the ARIC study was younger, twice as large, and the follow-up period twice as long compared to the MESA study. This could possibly explain the different results.

The lowest incidence rate was found in the population investigated by Adamsson Eryd et al.[14] However, this was expected, since they only looked at the first hospitalization for AF as a measure of incident AF.

The population investigated by Losi et al. consisted of hypertensive patients.[15] Hypertension is a well-known risk factor for AF and atherosclerosis, why one would expect to find higher baseline measurements of cIMT and a higher incidence rate of AF in a hypertensive population.[19] The baseline measurements of cIMT were notably higher in the hypertensive population, compared to the other studies, but this could also be due to different measurement methods. The incidence rate in the study by Losi et al. was surprisingly not different from incidence rates in the ARIC study and the MESA study.[12,15] This might be explained by the fact that the hypertensive population was younger than the other populations. Moreover, the calculation of the HR was different between the studies. While Losi et al. calculated the HR comparing two groups categorized by cIMT values higher or lower than 1.5 mm, the other studies compared the highest with the lowest quartile or quintile of cIMT values, which expectedly would lead to a higher HR.[11,12,14,15] However, their HR was comparable to the results of the other studies, which might be explained by the above-mentioned differences in measurement methods, population characteristics, and study design.

Strengths and Limitations

The principal strength of this study is that the systematic literature search was conducted by three independent reviewers, who systematically searched three comprehensive medical literature databases. Other strengths are the consistency of the results across independent cohort studies and across the two measurement methods cIMT and CACS. Furthermore, all included studies have a relatively high quality assessed by the NOS.

However, this study also has some limitations. First, only few studies have been published on the topic, which is why the meta-analyses consist of results from only five studies on cIMT and two studies on CACS. The small number of studies resulted in two studies being weight more than 85% of the overall HR for cIMT. Second, a notable limitation is that all the included studies possibly miss cases of paroxysmal AF, because of the time-dependent nature of this condition. Third, there were some differences between the studies regarding ascertainment of baseline and outcome measurements, and the calculation of HRs, which resulted in decreased comparability. Fourth, the five studies investigating the association between cIMT and incident AF used different measurement methods and definitions of the presence of a carotid plaque, which may have had impact on their results. Fifth, as this is a systematic review and meta-analysis based on cohort studies bias may be present. Because of the low number of included studies, it is difficult to estimate the risk of publication bias, which could be relatively high, because the contradictory result found in the CHS study, was not included, due to lack of accessible data.


Current guidelines on AF management do not recommend specific treatment of subclinical atherosclerosis detected by cIMT or CACS measurement to reduce the risk of AF,[1] because this has yet to be demonstrated. Nevertheless, our findings could put additional focus on risk factor management in general. Further research is specifically needed to test the measurement of cIMT as screening tool in cardiovascular risk assessment. There is also an urgent need for standardization regarding the definition and measurement of cIMT as well as high quality data on its reliability, before it might be used as routine screening tool. More research is also needed to investigate the benefit of screening for AF in patients with subclinical atherosclerosis and vice versa.