One hundred and four subjects (: 63.9 ± 11 years, 54% male) underwent clinically indicated TSE, with Doppler studies of mitral inflow performed at rest and after exercise.
The subjects were categorized into two groups depending on the appearance of ischemia, which was defined as an increase in the WMSI from pre to post-exercise.
Table 1 presents the baseline summary statistics of DMIP variables, heart rate, age and sex. Ischemia developed in 33 (32%) subjects. The ischemic subjects were older [t(96) = –2.37, p = 0.020] and predominantly males [F(1,102) = 5.03, p = 0.027]. None of the baseline DMIP values or HR (denoted by an index of 1) differed significantly between the two groups: E1, t(102) = –0.12, p = 0.903; DR1, t(102) = –0.02, p = 0.981; A1, t(102) = –1.57, p = 0.119; HR1, t(46) = –0.13, p = 0.900.
Table 2 presents the summary statistics of the exercise variables and changes (postexercise–baseline) in E, DR, and A. The duration of exercise [t(102) = 0.44, p = 0.664; peak HR: t(102) = 1.72, p = 0.088] and TIME [t(102) = –0.57, p = 0.572] were not statistically different between the two groups. ΔHR was significantly higher in patients without ischemia [t(102) = 2.85, p = 0.005].
The ΔE [t(102) = –2.29, p = 0.024] and ΔDR [t(102) = –2.64, p = 0.010] were significantly higher in the ischemic subjects than in the nonischemic subjects. The changes in A were not statistically different between the ischemic and nonischemic subjects [t(47) = 1.78, p = 0.082].
Table 1
Baseline Summary Statistics of Doppler and Clinical Variables
|
Group 1
nonischemics
|
Group 2
ischemics
|
p-value
|
N
|
71 (68%)
|
33 (32%)
|
-
|
Age, years
|
62.5 ± 12.0
|
67.0 ± 7.2
|
0.020
|
Males
|
33 (47%)
|
23 (70%)
|
0.027
|
E1, cm/s
|
72.3 ± 21.4
|
72.8 ± 21.4
|
0.903
|
A1, cm/s
|
68.2 ± 17.5
|
74.4 ± 21.1
|
0.119
|
DR1, cm/s2
|
388.6 ± 177.2
|
389.5 ± 194.8
|
0.981
|
HR1, beat/min
|
61.3 ± 8.5
|
61.6 ± 12.7
|
0.900
|
Values are mean ± SD or N (%) |
The p-value indicates the significance of the differences in the mean values between the two groups. N = number of subjects; E = E-wave peak velocity; A = A-wave peak velocity; DR = E-wave deceleration rate; HR = heart rate. Index 1 denotes baseline values.
Table 2
Summary Statistics of Exercise Variables and Changes (Postexercise – Baseline) in Doppler Variables
|
Group 1
nonischemics
|
Group 2
ischemics
|
p-value
|
ΔE (E2 – E1), cm/s
|
24.2 ± 21.6
|
34.2 ± 19.4
|
0.024
|
ΔA (A2 – A1), cm/s
|
15.0 ± 14.8
|
7.9 ± 20.9
|
0.082
|
ΔDR (DR2 – DR1), cm/s2
|
225.7 ± 234.5
|
348.0 ± 183.9
|
0.010
|
Duration of exercise, min
|
7.9 ± 2.7
|
7.7 ± 2.5
|
0.664
|
Peak HR, beats/min
|
141.1 ± 21.7
|
133.7 ± 17.0
|
0.088
|
ΔHR (HR2 – HR1), beats/min
|
26.3±10.7
|
19.5±12.9
|
0.005
|
TIME, min
|
2.9 ± 1.1
|
3.1 ± 1.1
|
0.572
|
Values are the mean ± SD |
p-values for the differences in the means between the two groups |
Δ = difference between postexercise and baseline values; Index 2 denotes postexercise, and index1 baseline values; E = E-wave peak velocity; A = A-wave peak velocity; DR = E-wave deceleration rate; HR = heart rate; Peak HR = HR at cessation of exercise; TIME = time of obtaining Doppler variables measured from the cessation of exercise. |
Multiple linear regression was calculated to predict ΔE and ΔDR from age, sex, peak HR, ΔHR, exercise duration, IR (IR = 0 for nonischemics, and IR = 1 for ischemics), TIME, and the TIME*IR interaction. Table 3 presents a summary of this regression analysis. Forward selection criteria were applied with IR, TIME, and TIME*IR as an initial set of predictors. Except for age, none of the other independent variables introduced one by one and in various combinations contributed significantly on their own.
Table 3
Summary of Regression Analysis for Variables Predicting Changes in E and DR
|
ΔE
|
ΔDR
|
Age
|
NS⇑ (p = 0.065)
|
⇑
|
IR
|
NS⇓ (p = 0.242)
|
NS⇓ (p = 0.206)
|
TIME
|
⇓⇓⇓
|
⇓⇓
|
TIME*IR
|
⇑
|
⇑
|
Adjusted R2
|
0.162
|
0.175
|
Directions of arrows indicate direction of association (upward means positive) |
⇑: p < 0.05; ⇑⇑: p < 0.01; ⇑⇑⇑: p < 0.001
|
NS = nonsignificant (p > 0.05)
|
IR = ischemic response: for nonischemics IR = 0, and for ischemics IR = 1; TIME = time of obtaining Doppler variables measured from the cessation of exercise; R2 = coefficient of determination; Δ = difference between postexercise and baseline values; E = E-wave peak velocity; and DR = E-wave deceleration rate.
|
This model showed a marginally significant influence of age on ΔE [F(4,99) = 5.972, sig. F < 0.001, coeff. 0.336, p = 0.065, 95% CI: –0.021–0.694]. Age was a significant predictor of ΔDR [F(4,99) = 6.479, sig. F < 0.001, coeff. 3.989, p = 0.038, 95% CI: 0.232–7.745]. The main effect of ischemia on ΔE and ΔDR was not significant. The main effects of TIME on ΔDR [F(4,99) = 6.479, sig. F < 0.001, coeff. –80.445, p = 0.001, 95% CI: –125.319 to –35.571] and on ΔE [F(4,99) = 5.972, sig. F < 0.001, coeff. –7.988, p < 0.001, 95% CI: –12.262 to –3.714] were statistically significant. The values of ΔE and ΔDR obtained later were lower than those obtained earlier.
There was no correlation between IR and TIME (r = 0.06, p = 0.57), and the frequency distribution of ischemics was not statistically different between and within groups of TIME (1–6 min) [F(5,98) = 1.49, p = 0.20)]. There was no bias in collecting data regarding TIME between ischemics and nonischemics.
The TIME*IR interaction was statistically significant for ΔDR [F(4,99) = 6.479, sig. F < 0.001, coeff. 90.473, p = 0.025, 95% CI: 11.832–169.113] and for ΔE [F(4.99) = 5.972; sig. F < 0.001; coeff. 7.799; p = 0.041, 95% CI: 0.309–15.289]. The effect of IR on the values of ΔE and ΔDR relied on the time after cessation of exercise at which data were collected.
To determine the time course of ΔE = E2 – E1 and ΔDR = DR2 – DR1 after the exercise, multiple linear regressions were performed to predict ΔE and ΔDR from the IR, TIME, and TIME*IR interaction as predictors. The general form of the regression function was as follows:
\({\text{Y=}}{{\text{C}}_{\text{0}}}{\text{+}}{{\text{C}}_{\text{1}}} \cdot \left( {{\text{TIME--}}{{\text{T}}_{{\text{shift}}}}} \right){\text{+}}{{\text{C}}_{\text{2}}} \cdot {\text{IR + }}{{\text{C}}_{\text{3}}} \cdot {\text{ IR}} \cdot \left( {{\text{TIME--}}{{\text{T}}_{{\text{shift}}}}} \right)\)
where Y represented either ΔE or ΔDR, Tshift is the time shift (Tshift = 0 denotes linear regression with respect to the end of the exercise, Tshift = 1 min denotes linear regression with respect to the time coordinate having zero in the first minute, continuing similarly for later times). The outcome of the multiple regression model for nonischemics (IR = 0) was calculated according to
\({{\text{Y}}_{{\text{nonischemic}}}}{\text{= }}\left[ {{{\text{C}}_{\text{0}}}{\text{-- }}{{\text{C}}_{\text{1}}} \cdot {{\text{T}}_{{\text{shift}}}}} \right]{\text{ + }}{{\text{C}}_{\text{1}}} \cdot {\text{TIME}}\)
where, coefficient C1 denotes the slope.
For ischemics (IR = 1), the outcome is
\({{\text{Y}}_{{\text{ischemics}}}}{\text{=}}\left[ {{{\text{C}}_{\text{0}}}{\text{--}}{{\text{C}}_{\text{1}}} \cdot {{\text{T}}_{{\text{shift}}}}{\text{+}}{{\text{C}}_{\text{2}}}{\text{--}}{{\text{C}}_{\text{3}}} \cdot {{\text{T}}_{{\text{shift}}}}} \right]{\text{ +}}\left[ {{{\text{C}}_{\text{1}}}{\text{+ }}{{\text{C}}_{\text{3}}}} \right] \cdot {\text{TIME}}\)
where the slope is C1 + C3.
Note that the intercept is a function of Tshift and IR: it changes from \(\left[ {{{\text{C}}_0}--{{\text{C}}_{\text{1}}} \cdot {{\text{T}}_{{\text{shift}}}}} \right]\) for nonischemics to \(\left[ {{{\text{C}}_0}--{{\text{C}}_{\text{1}}} \cdot {{\text{T}}_{{\text{shift}}}}+{{\text{C}}_{\text{2}}}--{{\text{C}}_{\text{3}}} \cdot {{\text{T}}_{{\text{shift}}}}} \right]\)for ischemics.
Figures 2 and 3 show slopes for the group of ischemics (coefficients C1 + C3) and nonischemics (coefficient C1).
The slopes of ΔE or ΔDR in ischemic subjects were not significantly different from zero. Contrarily, slopes of ΔE (p < 0.001) and ΔDR (p = 0.001) in the nonischemic subjects were significantly different from zero and negative. Our model showed that there was a significant IR*TIME interactive effect on ΔE (adj. R2 = 0.14, F(3,100) = 6.64, sig., F < 0.001, coeff. 8.1, p = 0.037, 95% CI: 0.51–15.66) and ΔDR [adj. R2 = 0.15, F(3,100) = 6.92, sig. F < 0.001, coeff. 93.8, p = 0.022, 95% CI: 13.9–173.7]. Consequently, regression slopes were significantly different between the ischemic and nonischemic subjects (coefficient C3 for the IR*TIME interactive effect). Figure 2 shows the measured ΔE as a function of TIME and IR and the regression lines for the ischemic and nonischemic subjects, and Figure 3 shows the same for ΔDR. At Tshift = 0, the intercepts of the regression lines of ΔE and ΔDR were not significantly different between the two groups. As shown by the regression lines for the ischemic group, the values of ΔE and ΔDR remain constant within the first 6 min, whereas in the nonischemic group, E and DR tend to return to the baseline values. Consequently, the differences in ΔE and ΔDR between the two groups increased in TIME; hence, the question was at what time did these differences become significant, i.e., the time at which the simple main effect of IR became relevant.
Table 4 presents the p-values for the simple main effect of IR on ΔE and ΔDR, calculated at different Tshift values. The results indicated that until the third minute, the effect of ischemia was not significant. At ≥3 min, positive associations of ischemia with ΔDR and ΔE were highly significant, which means that the values of these variables were significantly higher in the ischemic subjects.
Table 4
P-values for the Simple Main Effect of Ischemia on ΔE and ΔDR as a Function of Tshift
Tshift, min
|
0
|
1
|
2
|
3
|
4
|
5
|
6
|
ΔE, cm/s
|
0.271
|
0.532
|
0.651
|
0.012
|
0.001
|
0.002
|
0.005
|
ΔDR, cm/s2
|
0.236
|
0.516
|
0.578
|
0.005
|
0.000
|
0.001
|
0.002
|
Tshift = time shift; Δ = difference between postexercise and baseline values; E = E-wave peak velocity; and DR = E-wave deceleration rate.