Det medisinske fakultet

Diastolic functional imaging


Asbjørn Støylen, Professor, Dr. med.

Department of Circulation and Medical Imaging,
Faculty of Medicine,
NTNU Norwegian University of Science and Technology

The page is part of the website on Strain rate imaging
Contact address:

Website updated:      This section updated:     

This section:

  References for all sections

Diastolic function

Mitral flow

Traditionally evaluation of left ventricular diastolic function has been by mitral flow (82) as shown below. The late filling (A) is more about left ventricular compliance (126) in the passive phase.

Diastolic function diagram shown by pressure and flow traces.

A: Relation between mitral flow indices and pressure in the normal situation. Mitral flow (red curve) is dependent on the pressure gradients between the left ventricle and the atrium, which is created by left ventricular relaxation. The decline in pressure gradient during IVRT ( = relaxation constant tau) after AVC determines the length of the isovolumic relaxation time. The decline in the pressure difference between atrium and ventricle as the ventricular pressure increases, determines the deceleration time.  This again is dependent on the relaxation rate, as the active relaxation, creating a quick pressure drop in the ventricle, a high gradient and a short deceleration time.  Atrial pressure increases during atrial systole, forcing blood to flow again in the A wave.
B: Slower relaxation leads to a decrease in the tau, and thus a longer IVRT before the mitral valve opens; Increased IVRT. In addition, the slower relaxation leads to a less profound but longer drop in LV pressure, leading to a reduced E amplitude and a prolonged dec-t.  The lower filling volume leads to a higher atrial volume at the start of atrial contraction, and thus a higher atrial stroke volume (perhaps by the Frank-Starling mechanism), and a higher A- wave. The E/A ratio is reversed. (Light gray flow curve is from A, for comparison).
C: Decreased LV compliance due to fibrosis or dilation, leads to a higher increase in LV pressure from the injected volume from the atrium. This leads to an earlier equilibration of LV and LA pressure, and an abbreviated A-wave, which can be seen by comparing with the duration of the reverse A wave in the pulmonary veins. Decreased LV compliance shows up first in end diastole, as this is the phase where the ventricle is  at the highest volume.  (Light gray flow curve is from B, for comparison). D: Increased filling pressure (Left atrial pressure) due to filling problems, will decrease IVRT as shown here, as the pressure gradient between LA and LV is less. In addition, the gradient is higher in early filling, due to the higher LA pressure, with a subsequent higher E-wave. But then LV pressure increases faster in response to the filling from the LA, due to both the increased filling rate, slower relaxation and finally less compliant ventricle already during diastasis. The filling time and dec-t is shortened. Finally the A wave is blunted, due to the higher LV pressure at the start of LA systole, and the E/A ratio reverses back.  When the mitral flow looks normal due to delayed relaxation compensated by higher pressure it is called pseudonormalisation, when the E/A, ratio is higher than normal, and the IVRT and Dec-t is shorter than normal, it is called restrictive filling. Restrictive filling is usually a sign of  reduced compliance already in early diastole; i.e. severely reduced compliance leading to early pressure increase. (Light gray flow curve is from CB, for comparison).

This means that the rapidity of relaxation is a measure of pressure decline, and is reflected in the peak velocity during early filling. However, in order to normalise for total filling (stroke volume), the conventional measure has been the ratio of early versus late filling E/A. As early filling declines due to decreased relaxation, the A wave increases as contributor to the total filling. As shown above, the IVRT and Dec-t are other measures.

Normal values for diastolic mitral flow indices from the HUNT study (conventional diastolic values)

For completeness I have added the normal values for the mitral flow indices from the HUNT study. The values were published in (165).

Mitral E
Mitral A
Feasibility N (%) 657 (99%) 657 (99%) 657 (99%) 657 (99%) 653 (98%)
<40 years, N=208, mean (SD) 80 (16) 48 (15) 1.85 (0.76) 212 (55) 85 (16)
40-60 years, N=336, mean (SD) 74 (15) 59 (15) 1.32 (0.40) 220 (66) 95 (20)
>60 years, N=119, mean (SD) 69 (16) 75 (18) 0.96 (0.32) 244 (79) 105 (23)
All, N=663, mean (SD) 75 (16) 58 (18) 1.42 (0.62) 218 (66) 93 (21)
Feasibility N (%)
599 (99%)
599 (99%)
599 (99%)
599 (99%)
597 (99%)
<40 years, N=126, mean (SD)
75 (15)
44 (14)
1.86 (0.64)
217 (65)
91 (17)
40-60 years, N=327, mean (SD)
64 (15)
52 (14)
1.30 (0.42)
232 (81)
100 (21)
>60 years, N=150, mean (SD)
61 (14)
65 (18)
0.99 (0.34)
269 (97)
118 (29)
All, N=603, mean (SD)
66 (15)
54 (17)
1.34 (0.54)
238 (85)
103 (24)

As is evident, the early diastolic indices decline with age as does the systolic, A increases.

Diastolic function by tissue Doppler

Thus, it is evident that mitral flow gives information about LV relaxation, but the secondary changes in pressure tends to complicate the picture. With a low E/A ratio and long dec-t, it is obvious that the filling pressure is NOT increased, and no further information is necessary. Pseudonormalisation will camouflage delayed relaxation, and the restrictive pattern can be seen also in the young, due to a very quick relaxation (although seen in the old, it should be seen as pathological).

Diastolic function seen by tissue Doppler and M-mode of the mitral ring. To the left a normal subject showing normal e' velocity and normal e'/a' ratio, to the right a patient with hypertension, showing reduced e' velocity as well as e'/a'. The delayed relaxation is evident also in the M-modes, but may be more difficult to measure, if the deflection between the diastasis and the late diastolic displacement is less sharp.

Fundamentally, the physiology discussed her will not be representative for situations where there may be marked asynchrony of the e' waves, such as bundle branch block or pacing.

A: Patient < 30 with normal diastolic function. E/A > 1, Short Dec.T and IVRT, high e'. In this patient it is normal for age, but might have been severe heart failure with restrictive filling, even given the patient's age.  Compare with patient F. In this case the tissue Doppler helps to discern.
B: Patient about 50 years with near normal diastolic function. for age. E/A = 1, somewhat longer Dec-T and IVRT.
C: Slightly impaired relaxation. Patient at about 70, with slightly delayed relaxation due to a history of hypertension. Prolonged IVRT, dec-T, reduced E and E/A ratio < 1.  Also reduced e'. D: Severely impaired relaxation. Patient with heart failure (and normal EF and LV EDV), but with normal filling pressure due to diuretic and ACE inhibitor treatment. Severely reuced relaxation with prolonged IVRT, dec-t, decreased E and E/A ratio <<1. Very low e'.
E: Pseudonormalisation. Patient age 69 with history of hypertension. Mitral flow (top left) shows normal values for E, A and Dec. time, and the IVRT (top right) is also normal. Tissue Doppler (bottom) shows impaired relaxation, (E/e' about 15), indicating that the atrial pressure is elevated. This is demasked by doing a mitral flow acquisition during Valsalva manouver (decreasing venous return and hence, atrial pressure) below: F: Patient with restrictive pattern (actually same patient as in C, but before treatment, and then with increased LVEDV  and low EF), due to high filling pressure. Short IVRT, dec-t, high E and E/A ratio. e' still low showing that there is delayed relaxation, despite the high E and E/A. Compare with A, little difference, but taking the patient's age into consideration, it is actually evident that this is restrictive filling, even without tissue Doppler showing a low e'.

Valsalva manouver in patient E above, demasking pseudonormalisation and showing the typical pattern of impaired relaxation.

As shown above, the global diastolic function is more robustly assessed by tissue Doppler of the mitral ring, being the resultant of the local relaxation events, than of regional diastolic strain rate (except possibly for diastolic strain rate propagation shown below). The early diastolic annular velocity (e') has been shown to be related to tau, and to be less preload dependent than mitral flow (69, 70, 71). This means that the tissue velocity can be used to separate impaired relaxation with increased filling pressure from normal situations, in impaired relaxation the e' remains low despite increased atrial pressure. Thus diastolic function of the LV (relaxation) can be more directly measured by the e' than E, and the closest correlate to relaxation rate.

Normal values for spectral tissue Doppler annular left and right ventricular diastolic velocities by age and gender from the HUNT study. From (165).

Left ventricle, mean of 4 walls Right ventricle (free wall)

e' (pwTDI)
a' (pwTDI)
a' (pwTDI)

< 40 years
14.6 ( 2.3)
8.8 (1.9)
14.7 (2.9)
12.4 (3.5)
40 - 60 years
11.3 (2.4)
10.0 (1.9)
13.1 (2.9)
15.0 (3.5)
> 60 years
8.2 (3.2)
10.6 (1.9)
11.0 (2.3) 16.1 (3.1)
11.8 (3.2)
9.7 (2.0)
13.3 (3.0)
14.4 (3.7)

< 40 years
14.1 (2.7)
9.1 (1.7)
14.5 (2.9)
12.3 (3.5)
40 - 60 years
10.7 (2.3)
10.4 (1.6)
12.5 (3.2)
14.3 (3.7)
> 60 years
8.2 (1.9)
11.1 (1.6)
11.0 (3.0)
15.8 (4.2)
10.8 (3.0)
10.3 (1.7)
12.5 (3.3)
14.2 (3.9)
Annular velocities by sex and age. Values are mean (SD).  pwTDI: Pulsed Tissue Doppler recorded at the top of the spectrum with minimum gain.  Values of e' decline with age, a' increase. Normal range is customary defined as mean ± 2 SD.

The study is based on 1266 healthy individuals from the HUNT study by Dalen et al (165). The age dependency of values is evident. Colour tissue Doppler gives mean values, which are consistently lower than pulsed wave values, as discussed here.

E is pressure driven, but e' is not to the same degree, the relaxation actually being the cause of the pressure drop. However, this load independency is not absolute, as discussed below.

This means that the ratio of E/e' can be used for assessment of left atrial pressure (71, 72, 177). If both E and e' increases, the ratio remains unchanged, and the increase in flow is due to higher relaxation rate. (For instance in exercise in normals (29).)  However, if E increases without e' increasing simultaneously , the increase in flow  must be driven by increased filling pressure  instead of by relaxation (as for instance in exercise in patients with impaired relaxation reserve (160), and the increase in the E/e' ratio is related to the increase in filling (atrial) pressure. However, if E remains unchanged and e' decreases, it is not physiologically meaningful to take the increased ratio as a measure of increased filling pressure. In fact, in transition from supine to sitting the E/e' increases while filling pressure decreases (160). Thus the E/e' relates only to filling pressure when E increases.

An E/e' < 8 is considered normal, while E/e' > 15 is considered a sign of elevated LA pressure.

Normal values for left ventricular E/e' from the HUNT study (From 165).

The E/e' ratio was age dependent, as has been shown previously in a smaller study (166), and confirmed in the larger HUNT study:

< 40
40 - 60
> 60
5,6 (1,3)
6,5 (1,7)
8,2 (2,6)
6,6 (2,1)
Values are mean (SD), and is the average of four walls.  The E/e' can be seen to increase with age.

The age dependency of E/e' is evident. This is actually one of the reasons for the "grey zone" between 8 and 15, as normal range for the age group < 40 is below 8.2 (mean + 2SD), between 40 and 60 it would be < 9.9 and above 60 years it would be below 13.4.

In a smaller sample of 100, cTDI was compared to pwTDI, and the values by cTDI were 2.2 lower than pwTDI. The ratio in the septum was 2.5 lower than in the lateral wall, but very similar to the mean of four walls. It is evident that by a normal range of mean ± 2SD, the normal range in the youngest group is 3.0 - 8.2 and in the oldest group 3.0  - 13.4. This explains previous findings of the ambiguity of the interval from 8 - 15 concerning relation to filling pressure. It should be age adjusted.

Where should measurements of e' be done?
The differences between walls are even greater in early diastolic than in systolic velocities. Thus, the e' and hence, the E/e' is highly site dependent. This has been shown several times, in the largest normal material being the HUNT study (165). In the HUNT study the e' did show the following normal values for pulsed Doppler per wall

05.March 2012:Thanks to observant reading by Charlotte Bjørk Ingul it was discovered that the values given in the table blow were S' values, not e'. Now the correct values (which are in accordance with the normal values given in the table above) are given below:

e' (SD) cm/s
11.3 (3.2)
9.9 (2.9)
11.6 (3.7)
12.5 (3.5)
11.2 (3.5)
The general principles of the site specific variability, however, still applies, and the E/e' values below were correct from the start.

This means, of course, that the E/e' also varies with the site of e' measurement:
E/e' from pwTDI according to age and site of e' measurement

Mean of four points
Mean septum-lateral
Septal Anterior Lateral Inferior
6,6 (2,1)
6,6 (2,1)
7,5 (2,4)
6,6 (2,4)
6,1 (2,2) 6,8 (2,3)
<40 years 5,6 (1,3)
5,6 (1,3)
6,5 (1,7)
5,4 (1,6)
5,1 (1,3)
5,7 (1,6)
40-59 years 6,5 (1,7)
6,5 (1,8)
7,4 (2,0)
6,5 (2,0)
6,0 (1,8)
6,7 (2,0)
>60 years 8,2 (2,6)
8,2 (2,7)
9,0 (3,1)
8,5 (3,0)
7,6 (3,0)
8,4 (2,9)

 It is evident that there is considerable site dependency of the E/e' as well. It is also evident that there is little difference between mean of four points versus two points, when only mean and SD are considered. However, the Standard deviation is a large population study reflects biological rather that measurement variability. The study of Thorstensen et al (154)did show an improvement in reproducibility of about 15% of e' measurement using the mean of septal and lateral, compared to either of them alone, and a further 30% (p<0.001) using four-point compared to two point averages.

Also, all systolic measurements of MAE and systolic peak velocity have been established from the start as being the mean of four points, although two points seem to work equally well in terms of mean, if not in terms of reproducibility. Thus, in the interest of robustness and to harmonise systolic and diastolic measures, the logical thing would be to chose four point average for e' as well. But logic has not got anything to do with it.

Rodriguez (69) in one of the first observational studies used the lateral point. In the early invasive validation studies; Nagueh (71, 196, 200) and Sundereswaran (197)  used the lateral wall alone, Sohn the septal point (70, 198, 199)   while Ommen (177) studied both the septum and the lateral point, as well as the mean. He found the best correlation between E/e' and filling pressure using the septum alone.  Present recommendations, however, favors mean of septal and lateral (195). It is argued by some that the invasive validation work has been done with one-site measurements, but at least, the HUNT provides normal data for all sites.

Load dependency of diastolic tissue velocity (e')

However, the e' is not entirely load independent. As normal subjects are sat upright on a bicycle (not using their leg muscles, and thus reducing venous return), the filling pressure drops, and so does e' (29, 160). The drop in filling pressure is evident by decreased mitral flow E, decreased LVEDV and increased HR (160). This has also been shown as e' changes after dialysis (178) and in applying lower body negative pressure (179).

AS the relaxation rate declines, and atrial pressure increases, however, there will surely be a cross over point where the load takes over as the main lengthening mechanism. In this case the diagram above will no longer be valid, and the e' relates to filling pressure more than relaxation and elasticity. In this case the e' relates to lenghthening load.

Splashing humpback whale in Wilhelmina Bay, Antarctica.

Load dependency of E/e'

As the e' is load dependent, even the E/e' may be be. At low pressures, the e' actually changes more than the E, thus increasing as LA pressure decreases, as has been shown consistently by the supine to sitting transition (29, 160). This may be due to the mechanism of load dependency being different, and e' may be more load dependent at low loads, which could be explained by the diagram above.

Ischemic post systolic shortening in diastole

As seen above, ischemic post systolic shortening will also affect diastolic filling:

As seen here, the post systolic shortening in the apex occurs simultaneously with the elongation of the normal base, where filling in principle should occur.
Looking at the colour M-mode, we can e an intraventricular flow of short duration, at end ejection. But as seen here, in this case there is some early onset of diastolic filling, which is cut short by the intraventricular flow, thus the intraventricular filling is delayed by the PSS.

Mitral ring and flow velocities from the case just aboveabove. In this case, timing of the early filling was fairly similar, despite PSS in the seoptum: Time to onset of e' wave was 480ms in the septum, 477 in the lateral wall and 489 to mitral flow E. Tme to peak e values were fairly similar too: Septum 535, Lateral 541 and flow 563 ms. Thus in this case, the inflow to LV is less affected, mitral flow is slightly later than tissue velocities, which is as expected. However intraventricular flow can be seen to be hindered.
This is the values from the case shown previously. In this case there is delayed onset and peak of e' wave in the septum where there is PSS, compared with lateral wall, and even earlier onset of E wave in mitral flow, possibly indicating elevated atrial pressure. In this case the mitral E was 65 cm/s, septal e' was 6 cm/s (E/e' 10.8) , lateral e' 9 cm/s (E/e' 7.2) (giving a mean e' of 7.5, mean E/e' of 6.7).

Comparison of pulmonary venous flow and mitral flow A waves, indicates increased end diastolic pressure.

In the case of ischemic PSS, however, there may be imitations to the use of E/e' ratio in estimating filling pressures, as they may not be simultaneous, neither in onset nor peak.

It is evident that Ischemic PSS may interfere with filling pressure. However, some authors have suggested that PSS is actually the cause of diastolic dysfunction. This is really putting the cart before the horse, to say it mildly. From the discussion above, it is really the other way around, delayed relaxation in ischemia that is the cause of ischemic PSS.

E and A fusion

In hemodynamic thinking, it is customary to start the heart cycle with ejection, and the to proceed to diastolic filling, hence S - E - A. This is the way tissue Doppler is presented as well. However, each hart cycle start with a sinus node activation, followed by an atrial activation and atrial systole, and this is the customary way of describing the ECG, hence P - QRS - T. But this corresponds to the sequence of A - S - E, which may be a help in describing the relation of E and A in relation to heart rate as illustrated below.

Four heart cycles illustration the relation between E and A with heart rate and PQ time.  Cycle I to II show normal PQ time and RR-interval from I to II, i.e. normal heart rate, resulting in a normal diastasis period between E of I and and A of II.  Cycle II to III shows higher shorter RR-interval, i.e. higher heart rate. As heart rate is increased,  it means that  P and hence A of cycle III , comes earlier after cycle II. Thus, this explains why it is the diastasis that is shortened with increasing heart rate. Cycle III to IV, shows the same RR-interval as I to II, i.e. same heart rate, but with longer PQ time.  Heart rate regulation modulates the RR (or, actually PP) interval, but in this case the PQ interval is prolonged in relation to the RR interval. This has the same effect as reduced RR interval; the PQ as fraction of RR decreases, and the diastasis is abolished.

Recordings from a patient with 1st degree AV block, PQ time of 272 ms at a HR of 73.  There is partial EA fusion at rest, showing up in mitral flow (left), pw and colour tissue Doppler (middle and it also changes the annulus motion pattern, as the diastole moves more or less continuously.

Thus, as heart rate increases, it is the diastasis that is shortened first, however, after the diastasis interval is zero, the next step is fusion of the E and A waves as shown below:

E/A fusion with increasing heart rate (or PQ time). (NB: the numbers on this image are not related to the numbers on the image above.) 1: No fusion and discernible diastasis. 2: Shortening of RR-interval first abolishes the diastasis. 3: Further shortening of the RR-interval leads to partial fusion of the E and A wave and e' and a', respectively. the peak E and e' is still separate, but the A and a' are atrial velocities added to the remaining velocities of the early phase, as shown by the arrows in the upper diagram. In the velocity curves, this is seen as the A/a' wave "climbs" up (down) the descending limb of the E wave. 4: At higher heart rate, the E and A are completely fused, and the separate effect of ventricular relaxation ant atrial contraction can no longer be discerned.

Patient with Wenckebach block, showing progressive E and A fusion as the P-wave comes closer to previous T-wave,  until one beat is dropped. Prolongation of PQ-interval reduces the Q-P interval. Thus, the diastasis varies inversely with PQ-time.

Although there is adaptive shortening of both QT interval and PR interval, this does not compensate for the shortened RR interval, with fusion the diastolic period shortens more with decreasing RR interval (29) and below.

Left ventricular diastolic filling period (DFP) and ejection period (LVET) in relation to heart rate during exercise.  Below HR 110, the RR interval and DFP shortens in parallel, showing the the diastasis is shortened first, while ejection time shortens much less. Above 110, there is parallel shortening of LVET and DFP, both contributing to the shortening of RR interval. The study also showed that the LVET and RR interval was only linear below HR 100 (29).
E and A with increasing heart rate during an exercise test in one patient. At HR 65,there is separate E, a and diastasis, both in mitral flow and in tissue Doppler as evident by the fact that tissue velocity is 0 between e' and a'. At HR 88 there is partial fusion, neither E nor e' reaches 0 before the start of A and a', respectively, and the A and a' are higher in absolute values due to this.  At HR 94 there is more fusion, but the peak of the E and e' are still discernible, and can be measured, as a measure of ventricular diastolic function. The E/A and e'/a' ratios, however, are useless, as the A and a' are summation velocities. The A and a' are increased further. At HR 121, the E/A and e'/a', respectively, are nearly completely fused. The peak E and e' can no longer be discerned. The peak diastolic velocity is far higher (in absolute values) that the E or e' and cannot be compared.

With partial fusion, peak E and e'
, and thus the effect of ventricular relaxation can still be seen and measured. However, peak A and a' are now a sum of the velocities due to atrial systole and the remaining velocities due to ventricular relaxation. This means:
  1.  The A and a' are higher than when separate, and no longer a measure of atrial function, and thus:
  2. The E/A ratio is no longer a measure of the relative contribution of the two mechanisms, and is nearly useless.
With total fusion, the E and e' velocities, and thus ventricular diastolic function cannot be measured separately. The diastolic function measured by the fused wave, is the sum of ventricular relaxation and atrial contraction velocities, and can still be taken as a measure of atrioventricular  diastolic function. But this means:
  1. Ventricular diastolic function cannot be measured separately.
  2. In an exercise or inotropic test , when  heart rate becomes high enough, the fused wave cannot be compared with the E wave at lower heart rates.
This  point is important in three situations:

  1. Children have higher heart rates, and partial fusion is common at rest, and even total fusion in neonates.
  2. First degree AV-block may give fusion at normal resting heart rates.
  3. In exercise testing, the increasing heart  rate leads to total fusion, usually at HR around 100. This means that for diastolic dysfunction, exercise tests are not as useful at HR > 100. In dysfunction due to ischemia at higher heart rates, however, ischemic stunning may persist for some time, while heart rate falls, and may still be useful.

Some invasive studies, however, seem to indicate that using the ratio between the fused EA wave and the fused e'a' wave, the filling pressure can still be estimated (198), as well as in atrial fibrillation where the a wave is absent, and the E and e' waves are higher (199).

Left bundle branch block in diastole:

Looking at the diastolic measures, there is some evidence that the asynchrony in left bundle branch block affects diastole as well: The use of the E/e' ratio, relates tissue velocity to flow velocity, assuming that the discrepancy should be an expression of filling pressure. But that also presupposes that the peaks are simultaneous, which they are very often not, when there is LBBB. And in fact, if the peaks are not simultaneous, the ratio has no physical existence, and is not meaningful at all.

Comparison of septal mitral velocity and mitral flow. As there is septal early flash, late systolic stretch and post systolic recoil as explained in another section, the e' is delayed and the E/e' is meaningless as a physical entity.

However, this may vary:

Looking at the spectral Doppler traces aligned by ECG (first vertical line)  there is post systolic motion (PSS) in the septal trace, which corresponds to a delayed e' wave. Laterally, there is no PSS, but there is corresponding delay in the peak e', the e' wave is seen slowly sloping towards the peak. The mitral flow is slightly affected as well, the peak E is similarly delayed. This leads to a partial fusion of E and A, as in 1st degree AV-block, as the next P and Awave comes early in relation to the delayed E wave. Thus, there may be potential for developing partial AV-block, despite normal PQ-time. In this case, however, there is similar dely of the septal e' wave, but the lateral peak e' is seen slightly earlier, abd the E wave corresponding to the lateral e'. It seems evident that in this case, the septal E/e' ratio is not strongly associated with filling pressure, as they are not simultaneous. In this case, the septal PSS is delaying the septal e' even more, but again with no delay in the lateral e'. And the mitral flow peak E is earlier than both. It seeems evident that in this case, meither E/e' ratio is not as strongly associated with filling pressure, as they are not simultaneous. And the fact that the flow E is earlier than the tissue e', might indicate that filling pressure actually is elevated.
Another case, but with lower S'in the septum, indicating a poorer septal function, again the e' in the septum can be sen to be delayed, and later that the mitral flow E, which in this case is more synchronous with the lateral e'. It is difficult to see that the E/e' ratio, at least the septal one expresses filling pressure, as the peaks are not simultaneous. In this case, the E/e' was 17.5 for the septal e', 7.7 for the lateral e' and 10.7 for the mean e' (however, the validity of taking the mean of non simultaneous peaks is dubious). As the lateral e' is simultneous with peak E, this should be a better measure of filling pressure????

However, in the last case, the Valsalva indicates an elevated atrial pressure:

During Valsalva, the E/A ratio drops from 0.78 to 0.33 and the flow pattern becomes that of typical delayed relaxation.

Limitations of the E/e' ratio

As the investigantions have continued, there has been shown severe limitations of the E/e' in estimating atrial pressure(332).

E/A fusion:
The most evident is explained above, where there is total E/A (and e'/a') fusion, the combined e'/a' wave do no longer measure the relaxation alone, but the combination of ventricular relaxation and atrial systole. Also the E/A wave do relate to atrial pressure, but no longer the mean  pressure but the peak pressure during atrial systole, which may differ.

Position dependency:
AS we have shown earlier, the E/e' increases in the sitting position, while filling pressure drops. Thus the studies are mainly valid for supine acquisitions.

High filling pressures:
If the filling pressures are high, the filling pressure (lengthening load) may take over as the main mechanism for the early motion of the mitral ring, as discussed above. This is in accordance with
Mullens (272), who found no clear relation between filling pressures and E/e' in a heart failure population (except that most had high E/e', of course).

In constrictive pericarditis, the filling pressures are generally elevated, but longitudinal diastolic function (e') intact (273), giving a lower E/e' for any given filling pressure.

Left bundle branch block:
And of course, left bundle branch block, as discussed above.

Diastolic strain rate

Looking at the velocity and displacement traces, even with the addition of the protodiastolic motion event, the diastole looks fairly straightforward, after AVC, the three fundamental phases known from Doppler flow can be seen: Early filling phase (E), seen as the first negative phase (e') after AVC, diastasis with little or no motion, and the atrial systole (A) seen as the second negative velocity spike (a'). The atrial displacement of the ring may be described as the atrium pulling the ring away from the apex, and in addition the added volume pushed into the ventricle by atrial (esp. auricular) contraction pushing the atrioventricular plane. The relative contribution of the two mechanisms is uncertain.

Taken from the mitral ring, diastolic ventricular displacement and velocity show the left ventricular diastolic global function.

Velocity and displacement in the base of the septum, showing systolic motion toward the apex, protodiastolic motion away, and the the two basis diastolic phases, early (e') and late (a') motion way from the apex, separated by diastasis.
In strain and strain rate, the pattern can be seen to be much more complex in these tracings from the base alone. There are at least four positive spikes (elongation) during diastole, this is reflected by much more "steps" towards zero in the strain curves. As strain rate is fairly susceptible to noise, this might have been interpreted as noise (as is the small negative spikes between) , but integrating to strain eliminates the random noise, and shows what is real. 

However, using strain and strain rate, the diastole can be seen to be far more complex, showing a sequence of events that are different, and with different timing in the different segments. Thus  is seen due to the better spatial resolution, as deformation imaging eliminates the effects of the tethering of the base to the more apical parts. In addition, these events interact, to result in the simpler pattern seen in motion traces, and the main finding is that there are more than one peak in each of the two phases of E and A, and also, the peaks are not simultaneous in all parts of the ventricle.

An average measure of diastolic strain rate, however, can be obtained with using a maximal ROI length, in addition to a maximal strain length, which then will cover at least 2/3rds of the ventricle

The finding of a complex pattern in diastole, shows that no single strain rate measurement parameter can be used as a criterion for diastolic function. Regional early strain rate might be taken as an indication of regional diastolic function, but only if care is taken to identify the elongation spike, and avoid the return wave. And as the traces above show, there are differences in both the amplitude and timing of early diastolic strain rate, the implication being that there is no meaningful way of averaging the values into a more global function measure. The e', being the resultant velocity of the mitral plane, however, is a truly global measure, being the summation of all local measurements and taking the time differences into account, as well as being less pressure dependent, is a more robust measure of diastolic function as discussed below.

For global diastolic function, diastolic tissue velocity is still the most important measure, as this is the resultant global peak measure. This is not the average, but the resultant of all the local (non-simultaneous) diastolic strains AND the propagation along the wall.


Strain and strain rate in the atria

As the outer contour of the heart is relatively constant, the apex is stationary, and the atria is attached to the large veins, the atrioventricular plane has to be the piston of a reciprocating pump as discussed above), expanding the atria while the ventricle shortens and shortening the atria while the ventricle expands. This is energetically useful, as the work used to decrease the volume, in additon to ejection, also moves the blood from the veins into the atria. If the heart had worked by squeezing changing outer contour to a high degree, the work would have been used to shift the rest of the thoracic contents especially lungs inwards in each systole, work that would have been waisted. Thus, most of the filling volume to the ventricles, is a function of the AV-plane pumping.

Basically, the deformation of both chambers reflects the motion of the atrioventricular plane. In systole, the ventricle shortens while the atria expands. This is a function of ventricular contraction. In early diastole there is elongation of the ventricles and shortening of the atria, the active component of this is the ventricular relaxation. In late diastole, there is further elongation of the ventricles and shortening of the atria, but in this phase the active component is the atrial contraction. However, deformation of both chambers are reciprocating, both reflecting the atrioventricular function, and for  the elongation of the atria during ventricular systole is not an independent parameter, and is mainly due to the systolic function (shortening) of the left ventricle.

Deformation (strain rate) in the atria is reciprocally related to the deformation of the ventricles. as both  apex and the atrial roof are relatively immobile, the ventricle shortens  (yellow) while atria elongates (cyan) during systole, ventricle elongates (cyan) and atria shortens (yellow) during the diastolic phases. In diastasis, there is no deformation, both are green.

Strain rate curved M-mode going through both ventricular and atrial septum shows the reciprocal colours of the atria and ventricles. Strain and strain rate in atrium (yellow) and ventricle (cyan) are seen as almost mirror images of each other. However, the absolute values in the atria are higher, as the atrioventricular plane motion is a greater percentage of the smaller atria as illustrated below. Strain rate curves are also basically mirror images of each other. As deformation is active in one chamber and passively transmitted to the other, the peak values may be higher in the active chamber, and there will be a time delay of events as waves propagate as shown in the ventricle during diastole.

Thus, deformation during ventricular systole, reflects ventricular contraction, early diastolic deformation reflects ventricular deformation and A wave reflects atrial contraction, irrespectively of which chamber the measurements are done, although values may vary. as shown below.

In this subject there is a ventricular systolic strain of 15%, while atrial strain during ventricular systole is 38%. However, taking the different lengths of the atrium and the ventricle, and calculating the absolute change in length, it can be seen to be the same within the limit of accuracy.  This is simply the MAE, reflecting both shortening of the ventricle and (longitudinal) expansion of the atrium.

Thus: The AV-plane motion is determined by the systolic function of the ventricles. It has been proposed that atrial strain is a measure of atrial "reservoir function". This rubbish means that the AV plane motion does mean different things when looked at from opposite sides, which is obviously rubbish. The atrial wall being thin, there is little resistance to AV-plane motion, and does not express distensibility of the atria. If one choses to measure strain in the atria instead of the ventricles, the ventricular strain being MAE/LV length, the atrial strain will be MAE/ atrial length. Thus the atrial strain will reflect the relative change in LA length, but this is due to ventriculoar contraction and says nothing of the atrium itself.

Atrial strain during ventricular systole

Reduced "reservoir function" (atrial strain during ventricular systole) in the atria in atrial dilation, is simply a function of the increased atrial length, given normal ventricular systolic function. And this will be relative, the absolute increase is the same, given the same MAE. In fact: the same MAE will result in the same absolute longitudinal expansion of the atria, and thus, much the same volume increase irrespectively whether the atria is long or short.

The relative expansion,  however has different contents.
  • Atrial strain is MAE divided by the atrial length. Thus, in reduced systolic function, the MAE is reduced, and so is atrial strain.
  • In atrial dilation, the atrial strain during ventricular systole will be reduced even with normal MAE, as atrial length is increased.
Thus atrial strain during ventricular systole is a composite of longitudinal ventricular function and atrial size. Left ventricular filling pressure (LA pressure), is elevated in reduced LV function, especially in heart failure. Thus, studies showing reduced left atrial strain with increased filling pressure may thus suffer from confounding with LV function.

Also, the left atrial size per se, is a sensitive index of chronic atrial pressure over time (195), and thus, even in normal LV function, the LA strain may correlate with LA pressure (and indeed may be a function of LA pressure), but the parameters are not independent. Thus, atrial strain during ventricular systole does not add new information, as recently confirmed clinically(245)

Being a composite parameter, it may be more sentitive than sigle parameters, but not independent. Long axis shortening is the most sensitive parameter of both function and prognosis (36, 190, 191, 192, 193), far more than EF.

  • The longitudinal expansion of the atria corresponds to a greater volume dependent on the width (cross sectional area/diameter).  The atrial expansion times the cross sectional area, would thus reflect a part of the reservoir volume in the atria. However, the auricles would not be included, and the volume function would not be complete.
The wall length should probably be related to a curved strain length (as in 2D strain). However, because of its thin wall, strain rate imaging in the atria from the apical position is extra prone to artifacts due to low lateral resolution at the large depth, which may affect transverse tracking in the atria, and result in under estimation of the wall shortening anyway.

Thanks to Mads Ersbøll of Copenhagen, who pointed out the connection to LA pressures in studies, leading to the discussion in this paragraph being extended, incorporating the relation to atrial size.

His group has recently confirmed in a large prognostic study confirming that atrial strain during ventricular systole does not add independent prognostic information (245).

Atrial strain during early diastole
Early filling phase is likewise related to ventricular diastolic function, the mechanisms being elastic recoil modulated by the rate of calcium removal from the cytoplasm as discussed below. Thus, the amount of the systolic atrial strain being reversed in early diastole is also a property of the ventricle, divided by the atrial length.

Atrial strain during late diastole
Finally, the atrial contraction is a property of the atria.

It has been proposed that the main function of the atrial systole is to pull the mitral ring back to the end diastolic point, thus pulling the mitral ring over a volume of blood contributing to the end diastolic volume(13). However, this model disregards the finding by Doppler flow that there is an active flow component as well. And, as the pressure in the ventricles increase during the atrial systole, there is evidence for the vis a tergo mechanism being an important component also of the motion of the mitral ring, i.e. pressure being the driving force. Thus, the peak A is the volume flow due to the contraction of the atrium, but modified by the rate of pressure increase in the LV, being a function of the LV compliance. However, the AV-plane motion and  a', would be measures of atrial function. And, as this is pumping volume driven the total atrial action (including the pumping of the auricles) would be the driving force.

As e'/a' ratio decreases, the a increases, so this function is not independent on LV function, but comes closer than other measures. And is of little use where there is partial fusion of e and a. as the velocities then are combined partially of ventricular relaxation.

Atrial strain and strain rate are simply displacement and velocity normalised for atrial length as shown below.

The true atrial contractile function is the length change during atrial systole. Changing hte start of tracking to the start of the a-phase, shows atrial strain to be 13%. Peak strain rate can be measured independent of the starting point for tracking.

Back to section index
Back to website index.


Editor: Asbjørn Støylen Contact address: