Flow measurement at the aortic root - impact of location of through-plane phase contrast velocity mapping

Cardiovascular magnetic resonance (CMR) is considered the gold standard for cardiac volumetric measurements [1]. The most widespread method for measuring flow with CMR is free-breathing through-plane phase-contrast velocity mapping. In general, aortic flow is often measured at the sinotubular (ST) junction [2, 3] even though placing the slice at the tip of the valve cusps in systole may be more precise [4, 5]. An exception to this is in case of aortic stenosis where it is recommended to do phase velocity encoding at valve level to avoid the disturbed flow further downstream as the jet breaks down [2, 6]. The work by Chatzimavroudis et al. argue that regurgitant flow is more precise when measured at valve level [5] but we have been unable to detect any research illuminating which level is more precise with regards to forward flow. Flow measured at different levels in the aorta is used equally in the clinic and the literature provides examples of papers where both levels of measurement are used [7, 8], but in our experience, they are by no means equal when compared (see Fig. 1 for example). This difference could to some extend be explained by the fact that the coronary arteries originate between the two imaging planes. The coronary blood flow is normally estimated to represent 5 % of the cardiac output [9] and would hence explain such a difference.

The difference between the two slices can be appreciated in the images in Fig. 2. The lumen at the ST-junction is round which is easily measured using automatic software. The lumen at valve level is shaped like a three leaf clover making delineation more difficult and this most often requires manual tracing, which is much more time consuming than automatic analyses.

We wished firstly to quantify how much the flow measurements vary at different levels in the aortic root and secondly to investigate which level corresponds best with the gold standard, left ventricle volumetry. We chose a group of patients with multiple morbidities (see Methods) to make results applicable to a broad population of patients.

Initially, 152 patients were included in the study, 22 were excluded because of arrhythmias (multiple ventricular extra systoles, strong respiratory sinus arrhythmia) and 9 had AS. Accordingly, data from 121 patients were analyzed, and 63 showed signs of mitral regurgitations and were excluded from the second analysis (see Table 1 for demographics). The patients also had en echocardiogram performed, and these were reviewed for mitral regurgitation. The group with MR identified on echo did not match the CMR group with MR, but defining MR from echo did not change the results (data not shown).

In the present study, we measured flow with free-breathing through-plane phase-contrast velocity mapping at two different levels in aorta, the ST-junction and at valve level. We compared the two flow measurements to each other and in turn with the gold standard, left ventricle volumetric measurements. We found that flow measured in aorta is highly dependent on the location of the slice in the aortic root. Our results indicate that measuring flow at the ST-junction underestimates flow compared both to volumetric measurements and flow at valve level. If flow measured at the ST-junction is subsequently used to estimate for example mitral regurgitations this could result in a systematic overestimation. Our results indicate a 13–16 % underestimation of aortic flow when measured at the ST-junction compared to valve level. Even though the coronary ostia are situated between the two planes, the coronary blood flow only comprises approximately 5 % of the total cardiac output [9], e.g 4.3 mL as in the case of a stroke volume of 85.4 mL.

Even though volumetry is considered gold standard, it is subject to assumptions and simplifications resulting in possible errors. Two important sources of error are the amount of trabeculation included in the blood pool, especially difficulties in segmenting trabeculations at end-systole and delineation of the basal areas of the ventricle, which comprise a large portion of the total volume.

Measuring aortic flow at valve level corresponded best with the volumetric measurements in the present study. Still, SV measured volumetrically was on average approximately 6 mL or 8 % (CI 6–11 %) higher than when measured with flow at valve level. When comparing results obtained with different methods, as we have done in this study, it is important to keep in mind under which conditions these were acquired. During flow measurements the patients are instructed to breathe normally, however for SSFP cine sequences the patients are instructed to hold their breath during end-expiration of a normal breathing pattern. It is essential that breath is held at a normal breathing pattern, since a large lung volume will result in decreased SV due decreased venous return to the heart [13]. Van den Hout et al. have found that SV decreased during inspiration and increased during expiration in healthy subjects [14]. This was in the level of approximately 7 mL and may provide an explanation for the difference we found between volumetric measurements and flow measurements. Since breath-hold data acquisition has been demonstrated to be accurate and reliable [15] this difference is likely due to patients not complying completely with instructions of breath-hold at normal end-expiration breathing pattern, hence performing a slight Valsalva manoeuvre and increasing their cardiac output slightly in the first 10 to 15 s. An alternative explanation could be that small mitral or aortic regurgitations are not visible on SSFP images. A jet of some magnitude is required to be visible on SSFP images [16]. Hence, since mitral regurgitations were determined visually, the difference between SV_ref and SV_V could be due to small missed mitral regurgitations. Through-plane velocity mapping relies upon the imaging plane being perpendicular to the flow being measured. Throughout the cardiac cycle the movements of the heart causes the aorta to move, which may cause slight angulation to the imaging plane. This angle could be different for the ST-junction and valve level, which may cause measurements to differ.

There is evidence to suggest that the myocardial blood flow is increased in the elderly compared to a younger population [17]. This increase can account for some of the here described difference but in our opinion, it is unlikely that the coronary blood flow in the elderly accounts for 15 % of the total cardiac output.

As an interesting observation, we saw that in the relatively small number of patients with AS excluded from the main analyses, flow measurements at valve level tended to overestimate SV compared to the volumetric measurements contrasting the underestimation seen in the rest of the population (see Fig. 3). Furthermore we found that flow measured at the ST-junction seemed to correspond better to measurements in patients without AS. These observations have to be studied with the reservation that the patients on average had increased SVs presumably due to hypertrophy of the left ventricle caused by their AS. Since only a small number of patients with AS were studied, these results should be investigated further and confirmed in a larger population.

Another interesting finding was that the mean differences between SV_ref–SV_V and SV_ref-SV_ST showed similar variances and hence equal accuracy. So even though measuring flow at valve level may be more time consuming, we found no evidence that this method is less accurate than measuring flow at the ST-junction.

When using CMR one has to be aware that results are highly dependent on several factors, including placement of the slice. Especially in patients with numerous extra systoles flow measurements are unreliable. We conclude that: