Correlation
Between Myocardial Ischemia and Changes in Arterial Resistance During Coronary Artery Bypass Surgery
Rephael Mohr, MD, lnge Dinbar, MD, Yaron Bar-El, MD, Uri Goldbourt, PhD, Martin Abel, MD, and Daniel A. Goor. MD The arterial resistometer provides continuous on-line monitoring of changes in arterial resistance. Resistance index (Ri), which bears a direct relationship to systemic vascular resistance (SVR), is defined by the equation Ri = P’/(dP’/dt), where dP’/ dt is the peak dP/dt of the arterial waveform, and P’ is the pressure at dP’/dt. In 42 patients with unstable angina, changes in Ri were studied at six periods during aortocoronary bypass surgery before tracheal intubation, during tracheal intubation, leg elevation, presternotomy, sternotomy, and dissection of the internal mammary artery. Thirty-four episodes of ischemia (0.1 mV ST segment changes) were observed in 26 patients. All ischemic episodes were associated with increased Ri (mean increase, 102 2 52%). Elevation of the pulmonary capillary wedge pressure corre-
T
HE MOST widely used monitor of myocardial ischemia during anesthesia is the electrocardiogram (ECG). The diagnosis of ischemia is based on the development of ST segment elevation or depression of 1 mm or more, most often seen in leads V, and II.‘,’ Changes in arterial pressure, heart rate: pulmonary capillary wedge pressure (PCWP), rate pressure product,3 and triple product’ have been used as adjuvants in the detection of ischemia. Although the detection of new regional wall motion abnormalities by transesophageal echocardiography has been shown to be a more sensitive and earlier marker of myocardial ischemia than ECG changes,6 this technique is not feasible during the induction of anesthesia. The authors have shown recently that a new device, the arterial resistometer,’ which monitors the behavior of the peripheral resistance (referred to in this article as resistance index is capable of detecting myocardial ischemia in [Ril), patients during coronary balloon angioplasty’. In the present study of patients undergoing coronary artery bypass surgery, a significant increase in Ri was found during periods of myocardial ischemia detected by ECG changes. This finding suggests that an alternative method of detecting ischemia intraoperatively may now be available.
MATERIALS
AND METHODS
Forty-two consecutive patients with unstable angina were studied during surgery for aortocoronary bypass grafting. There were 34 men and 8 women, ranging from 34 to 74 years of age. Eighteen of the patients required an intravenous infusion of nitroglycerin to control rest angina. Ten of 42 patients underwent emergency surgery. One patient entered the operating room with angina pectoris and ischemic changes on the ECG, which were reversed before induction of anesthesia with intravenous nitroglycerin and sublingual nifedipine. Two patients who entered the operating room with ischemia that could not be controlled before induction of anesthesia were excluded from the study because a baseline nonischemic period could not be obtained. There were 8 patients with left main coronary artery stenosis, 28 patients had a history of previous myocardial infarction, 7 patients had left ventricular (LV) dysfunction (ejection fraction < 40%), and 9 patients were hypertensive.
Journalof Cardiothoracic and VascularAnesthesia,
lated with ischemia during the preintubation, intubation, and sternotomy periods, but not in the remaining periods. Changes in arterial pressure and heart rate were not good predictors of ischemia. The prevalence of ST segment changes increased markedly during all periods of anesthesia with increase in Ri (P < 0.05). Ninety-one percent of ST segment changes were associated with a 25% increase from the baseline Ri. Raising the cutoff point to a 2 75% increase in Ri improved the specificity of Ri in ischemia detection from 61% to 92%. An increase of 2 75% in Ri occurred in only 6% of cases without ST segment changes. It was found that an increase in Ri as depicted by the arterial resistometer was the best hemodynamic correlate of myocardial ischemia. Copyright o 1992 by W. B. Saunders Company Preanesthetic medication included diazepam, 10 mg, morphine, 10 mg, or meperidine, 75 mg, plus atropine, 0.5 mg, or scopolamine, 0.4 mg. All patients receiving beta-adrenergic blocking drugs preoperatively received the last dose on the morning of surgery. The anesthetic management was at the discretion of the anesthesiologist, and consisted of fentanyl, 10 to 40 Kg/kg, diazepam, 2.5 to 25 mg, flunitrazepam, 0.6 to 2 mg, and halothane, nitrous oxide, and oxygen. Muscle relaxation for tracheal intubation was achieved with succinylcholine, 1 mgikg, or pancuronium bromide, 0.1 mg/kg, and the latter agent was used for additional relaxation during surgery. The anesthetic plan was designed to produce a decrease in systolic blood pressure of less than 20 mm Hg compared with preinduction values, and to maintain the heart rate close to normal preoperative values.“.” Nitroglycerin (1 to 3 mg/h) was given intravenously to all patients except those with a left main coronary lesion. In addition, nitroglycerin boluses of 0.1 mg were given to control episodes of ischemia. The time between patient entry into the operating room and the beginning of extracorporeal circulation was divided into seven periods: (1) preinduction (control), (2) pretracheal intubation (the period from start of induction until the administration of succinylcholine), (3) tracheal intubation, (4) leg lift (while the legs were raised for application of povidine), (5) presternotomy (the time when the saphenous vein was removed), (6) sternotoniy, and (7) mammary (dissection of the internal mammary artery). In 10 patients the internal mammary artery was not dissected; therefore, data for this last stage could not be obtained. Intraarterial pressures were monitored via a 17-gauge, 20-cm long catheter (Bard-I-Cath, CR Bard International Ltd., Sunderland, England), introduced percutaneously into the femoral artery and connected by a 4-foot rigid pressure monitoring line (Cobe Laboratories Inc., Lakewood, CO) to a physiological pressure transducer (Mennen Medical 922-122-010, Rehovot, Israel). The
From the Departments of Cardiac Surgery, Anesthesiology, and Epidemiology and Preventive Medicine, The Chaim Sheba Medical Center, Tel Hashomer, Israel, and the Department of Anesthesiology, Mayo Clinic, Rochester, MN. Address reprint requests to Daniel A. Goor, MD, Professor of Surgery, Director, Department of Thoracic and Cardiovascular Surgery, The Chaim Sheba Medical Center, Tel Hashomer, Israel 52621. Copyright 0 1992 by W B. Saunders Company 1053-0770/92/0601-0009$03.OOiO
Vol6, No 1 (February), 1992: pp 33-41
33
34
MOHR ET AL
natural frequency of the monitoring system was 20 to 30 Hz and the damping coefficient ranged from 0.12 to 0.2, as determined by the Intraflo (Burron Medical Inc., Bethlehem, PA) fast flush method. ECG and pressure readings were obtained using an ECG and pressure monitoring unit (Mennen Medical 741, Rehovot, Israel). Standard leads II and V, were monitored visually on the ECG. Both leads were routinely compared with the preoperative ECG. To increase the sensitivity of ST segment change detection, the operating room monitors were calibrated so that deflection of 2 cm was equivalent to 1 mV. Monitors were protected against interference from electrosurgical devices, and equipped with noise reduction units that work via an averaging process. ECG changes were diagnosed directly from the monitor and were not recorded. ST segment depression or elevation were considered significant when they were at least 0.1 mV (horizontal or downsloping) or at least 0.15 mV (upsloping) 80 milliseconds after the J point.“.” Using the resistometer,* the first derivative of the peak arterial pressure pulse (dP’/dt) was obtained by electronic differentiation that provided a linear frequency response to 100 Hz. No correction was made for the frequency response of the catheter because only relative changes of dP’/dt were sought. Ri was monitored continuously with the same unit that detected the dP’/dt. The equation used by the arterial resistometer79 for calculation of the Ri is: P’ Ri = dP’/dt where P’ is the pressure at the time of peak dP/dt, dP’/dt is the peak dP/dt of the central (femoral) arterial waveform, and Ri is resistance index. AI1 hemodynamic and ECG parameters were measured and recorded every minute in the first 5 minutes, and every 2 minutes thereafter. For statistical analysis, one set of hemodynamic measurements consisting of heart rate, arterial pressure, peak dP’/dt, PCWP, and maximal Ri value was taken in each period from all patients. The preinduction period served as a baseline and ischemic changes and hemodynamic parameters in the remaining six periods were compared with the corresponding variables in this baseline, nonischemic period. After dividing the patients into those with ECG evidence of ischemia at any stage (group A) and those remaining ischemia-free (group B), a repeated measure analysis of variance (ANOVA) was performed. The difference between ischemic and nonischemic patients was examined using the seven consecutive determinations of the previously mentioned hemodynamic parameters. The repeated measure ANOVA examined the relationship between the presence of ischemia and the mean values of the various hemodynamic parameters. It was also used to evaluate the relationship between the stage of anesthesia on hemodynamic findings and time-ischemia interactions.14 Subsequently, for purposes of demonstration, a Student’s t test was used post-hoc to test the significance of mean differences in the various hemodynamic parameters between patients with and without ischemic changes. To calculate the specificity and sensitivity of Ri in the diagnosis of ischemia, the prevalence of ST-T wave changes was also examined in quartiles of Ri and dP’/dt. A x2 test was used to evaluate the nominal type I error (probability for independence between ST-T wave changes on the ECG and Ri or dP’/dt despite the evidence in the study sample to the contrary). Given the multiple comparison setting, these were viewed in light of the repeated measures ANOVA findings for between-patient and within-patient effects.
RESULTS
Twenty-six of 42 patients (62%) developed ST segment changes during the study, and comprised group A (isch-
emit). The remaining 16 patients (38%) never developed ST segment changes, and comprised group B (nonischemit). There were 34 ischemic episodes; 13 occurred before tracheal intubation, 7 new episodes started during sternotomy, 5 during leg lifting, 4 during mammary dissection, 3 during intubation, and 2 before sternotomy (Fig 1, Table 1). Data were obtained in 32 patients for all seven periods. Repeated measures ANOVA (Table 2) shows that for Ri there was a highly significant relationship with ischemia (between-subjects effect, P = 0.0002), but no dependency on either the stage at which Ri was determined, or stage-ischemia interactions (P = 0.85). In 10 patients, internal mammary artery dissection was not performed; therefore, only six periods were available for analysis. If the final period (mammary) is excluded from the analysis, results in 42 patients, for the first six stages, confirm a substantial between-subjects effect (P < O.OOOl), and no independent stage role or stage-ischemia interaction Given these findings, a post-hoc observation of the changes during each of the stages of anesthesia reveals that, during all stages of anesthesia, mean Ri levels as well as relative and absolute changes in Ri were significantly greater in patients with ECG ischemic patterns (Fig 2). All 34 ischemic episodes diagnosed by ST-T wave changes were associated with an increased Ri (Fig 1A). Most ischemic episodes were preceded by a moderate increase of Ri compared with the preinduction period (20%, P < 0.001) (Table 1). During ischemia the mean Ri increase was 102 + 52% (P < 0.001) and after reversal of ECG changes Ri decreased to 24 k 32%. The ANOVA also shows that absolute values of dP’/dt were significantly lower in patients in the ischemia group. However, changes from baseline values were not always significantly different from those found in patients without ischemia (Fig 3, Table 2). Whereas the ANOVA showed a significant effect of stage and stage-ischemia interaction on PCWP changes, elevations in PCWP demonstrated borderline statistical significance between groups (P = 0.08, Table 2). These changes were significant for ischemia during preintubation. intubation, and sternotomy, but failed to discriminate ischemic from nonischemic cases during the other three periods (Fig 4). Some of the ischemic episodes were associated with decreased PCWP, and mean changes in this parameter before and during ischemic episodes were not significant predictors of ischemia (Fig 1B). Changes in blood pressure were not a good ischemia predictor (Fig 1C). They were significant only during the presternotomy period. Figure 4 shows significantly lower blood pressure in patients with ischemia during this period. No significant difference in blood pressure was found between ischemic and nonischemic patients in the remaining periods. Heart rate also was not a useful indicator of ischemia in any period (Figs ID and 4). Maximum values of blood pressure and heart rate achieved during all periods of anesthesia were not different between ischemic patients (group A) and patients who did not develop ischemia (group B); mean * SEM, 155 2 5 versus 149 f 7 mm Hg, and 94 2 6 versus 97 _t 5 beatsimin. Patients in group A had significantly higher maximum levels maximal of PCWP (17 k 2 KS 13 f 1 mm Hg, P < O.OOS),
ISCHEMIA DETECTION WITH ARTERIAL RESISTOMETER
35
(m 19)
B
lal-
lM1%
nillO-
‘\,
‘s,
on m 9
I
:I
33
BEFORE INDUCTION
10 0 40
1
I
im
m-
BEF6RE ISCHEMIC EPlwDE
D
m-
L&?&BEFORE INWCllON
ISCHEW EPISODE
ISCHEYIC EPISODE
lsz% EPISODE
6E;ORE lNMlcnON
Fig 1. Hemodynamic changes before, during, and after ischemic episodes (ST segment changes). (A) Changes in Ri; (6) PCWP; (C) mean blood pressure; and (D) heart rate. (A) Preintubation; (*) intubation; (:) leg-lift; (0) presternotomy; (m) sternotomy; (0) mammary.
Ri (125 ? 6 vs 79 2 4 milliseconds, P < O.OOl), and significantly lower minimum achieved peak dP’/dt levels (1040 f 90 vs 1370 f 100 mm Hg/s, P < 0.05). Comparison between maximal Ri values achieved during anesthesia
induction and Ri observed before induction shows that patients in group A had both a greater absolute and relative percentage increase in normal values of Ri than patients in group B (65 f 6 1~s26 2 4 milliseconds, and 118 f 13% vs
36
MOHR
ETAL
Ri During Inductionof Anesthesia
Table 1. Time-courseof Patient NO.
Premduction
1
47
Preintubation
lntubation
65
72
63
70
78
Leg Lifting
Presternotomy
68
85*
88*
94*
88*
80"
79*
sternotomy
MSllm.Sy
92*
94*
90*
2
54
59
60
58
74
76
70
61
58
56
72
74
70
96*
lOO*
102*
105*
ill*
108*
3
59
63
69
71
78
82
74
73
75
78
78
80
79
105*
109*
104
82
84
80
4
68
65
71
73
50
48
44
47
46
44
46
47
45
63
75
54
48
45
46
5
42
78
79
76
68
61
63
80
84*
85*
88*
94*
92"
75
68
64
48
42
38
6
78
102*
107*
105*
93
9J*
lOO*
lOl*
104*
108*
lOO*
104*
99*
76
64
68
7
71
140*
169*
138*
llO*
108+
102*
119*
118*
125*
107*
100
98
184*
192*
174'
160*
133*
128*
8
56
71
64
63
80
81
77
85
86
89
80
77
78
82
88
77
182*
117*
168*
164*
160*
126*
122*
120*
132*
138*
134*
116'
113*
102"
84
80
77
104
107
130*
134'
129"
104
93
90
91
100
104*
103*
88
95
9
88
157"
10
78
96
11
49
69
74*
J4*
47
56
54
66
68
59
54
48
51
69*
68*
65
60
61
53
12
54
46
44
39
65
70*
77*
82*
88*
90*
SIX
92*
88
91*
92*
89*
91*
93*
90*
13
29
43
45
44
37
40
39
39
38
36
56
48
49
70
76
78
74
82*
84*
14
29
58
76
64
66
65
66
44
39
37
62
60
59
65
79
77
69
72
71
15
84
102*
105*
105*
158"
162*
138*
99
96
91
93
95
87
96
91
88
16
40
52*
53*
51*
58*
62*
63*
59*
56'
49*
39*
36
35
21
20
20
42*
46*
45*
17
56
29
41
40
33
32
31
44
48
50
41
35
37
49
50
51
50
43
41
18
50
59
60
55
50
53
49
65
69
70
80
82
62
44
43
44
35
34
29
19
60
66
70*
72*
76*
73*
74*
78*
83*
91*
94*
97*
63
38
36
34
93*
98*
lOl*
20
47
79
88
91*
lOO*
lOl*
95*
88
95
113
121*
124*
119*
116*
96
81
79
70
63
21
31
39
39
38
40
43
43
36
39
39
51
52
52
51
50
50
48
57
59
22
67
78
79
80
102*
161*
147*
98
52
55
88
93
85
56
54
48
23
45
60
62
63
68
70
69
85
91*
98*
79
76
71
102*
129*
103*
88
44
47
24
49
67
70
75
70
70
65
48
54
49
50
48
46
40
42
43
42
41
42
25
51
JO
75
78
91
92
86
76
74
71
70
66
63
55
54
50
63
79
91*
26
67
66
65
60
68
61
60
61
62
63
68
70
67
63
63
61
47
45
43
27
65
67
63
60
99*
61*
86
83
80
78
74
73
45
48
63
28
41
51
52
52
76
73
71
50
49
48
49
45
47
57
66
63
46
44
45
29
68
8
86
84
66
65
64
60
61
58
89
92
91
90
88
85
30
68
103*
124*
120*
117,
115*
97*
94
97
96
91
90
85
69
63
64
73
81
85
31
48
46
48
45
39
38
38
46
50
51
51
50
47
68
84
87
32
50
173*
177*
180*
169*
174*
180*
159*
149*
147*
164*
171*
184*
179*
163*
151*
33
77
113
127*
130*
87
80
77
118*
123*
127*
74
68
59
55
54
53
34
49
37
36
35
36
35
36
48
50
52
32
34
32
40
42
40
41
46
48
36
73
99
lOO*
120'
122*
120*
121*
135*
146*
142*
135*
127*
138*
143*
140*
137*
130'
129*
37
60
69
74
70
84
70
76
74
73
70
73
77
82
80
75
84
38
44
39
40
39
39
40
39
40
57*
74*
47
44
43
51
53
48
39
68
178
197
156
151
148
137
80
87
82
80
78
71
201*
265*
195*
40
40
53
95
116
80
74
89
51
48
44
47
38
42
65
115
116
72
63
51
41
57
60
68
66
75
79
77
89
96
99
69
58
39
42
40
41
51
72
77
42
37
65
79
98
90
82
88
94
98
99
98
96
88
99
104
92
74
65
63
43
37
43
52*
51*
50
48
46
44
53
57
35
33
29
60*
99*
100*
58*
64* 118*
71* 101
70* 94
50
*ECGischemicchanges
54 ? 9%, P < 0.001) (Fig 5). However, changes in dP’/dt levels compared with the preinduction period were not significantly different between the two groups (Fig 5). The probability of ischemia was examined in quartiles of Ri during the various periods of anesthesia. The prevalence of ST segment changes increased markedly in all stages with the increase in Ri (P < 0.05) (Fig 6). Table 3 relates the number of patients with ST-T wave changes to quartiles of maximum Ri values and minimum dP’/dt values measured for each patient. This analysis was performed to evaluate the relationship between relative changes in Ri and dP’/dt and the prevalence of ST segment changes. The probability for independence between ST segment changes and Ri was 0 (x’ analysis), and for dP’/dt, p < 0.001. One patient among those with maximum Ri above the median (102
milliseconds) did not show an ischcmic pattern on the ECG, whereas only one patient with maximum Ri at the lower quartile ( < 80 milliseconds) had ischemia, suggesting a good specificity and sensitivity for the index. The correlation between percentage increase in Ri and ECG evidence of ischemia was evaluated in Table 4. Using a cutoff point of 25% increase of Ri, a sensitivity of 85% to 100% (mean 91%) and a specificity of 55% to 69% (mean 61%) was obtained. A Ri increase of more than 25% was observed in 91% of ischemic episodes. Better specificity (82% to 95%, mean 92%) was obtained when the cutoff point was set at a Ri increase of 75% (Table 5). The sensitivity with this cutoff point decreased to 38% to 69% (mean 51%). The probability of nonischemic patients developing a Ri increase of more than 75% is only 8%.
ISCHEMIA DETECTION WITH ARTERIAL RESISTOMETER
37
Table 2. The Effect of lschemia on Hemodynamic Parameters.
2
Analysis Performed Using a Repeated Measures ANOVA
mmHg/sec
Effect of Effect of lschemia (between
Ri absolute values
subjects effect)
F value
(excluding mammary
P
18.7 0.0002
Anesthesia
stage-
stage
lschemia
(time effect)
Interaction
0.2
0.31
0.93
0.85
stage) ‘Ri absolute values
F value
(including mammary) Ri % Change (excluding mammary) ?? Ri
9.1
0.61
0.28
0.0045
0.65
0.87
0.24
0.48
0.94
0.78
PCWP
Abbreviations:
10.57
P
0.0024
F value
2.36
7.44
0.8
2240
8
3
11
1880-2240
5
5
10
1260-1880
1
10
11
< 1260
1
9
10
26
42
Total P < 0.001
-20
-40
-I
Fig 5. The relative changes in resistance index (IX) and the first derivative of the peak arterial pressure (dP’/dt). Comparison between ischemic (I@ and nonischemic (0) patients.
peak dP/dt, and dP’/dt is the peak dP/dt.7~Y The relationship P’/(dP/dt) has been termed the “Ri” (see Appendix). When “E” and “a” are not determined by the thermodilution technique, fluctuations in SVR, but not the absolute values, can still be followed by monitoring Ri.” A previous study’ has suggested that to obtain more reliable measurements of Ri, the arterial catheter from which the waveform is derived should be as central as possible. The femoral artery is probably superior to the radial artery or other No.
of
Patients
PREINTUBATION P-0.2
-0.1-18.4
P5.4
!I,
k’)
Abbreviations: Max Ri, maximal Ri; Min dP’/dt, lowest dP’/dt values.
PCO.
-60
16
41.2
P 0.1 mV ST segment deviation) also showed a significant increase in Ri. To determine the
ISCHEMIA DETECTION WITH ARTERIAL RESISTOMETER
39
Table 4. Diagnostic Accuracy of an Increased Ri* in Predicting lschemia (ST-T wave changes) Using a Cutoff of a 25% Increase in Ri Increase in Ri (%)
A Preintubation < 25% > 25% Total
lschemia
lschemia TOtal
(+)
1-l
> 25% Total
> 25% Total
Total
6. lntubation 2
18
16
1
17
12
12
24
13
12
25
28
14
42
29
13
42
Sensitivity = 92%; Specificity = 55% D. Leg Lift
16
2
18
20
0
13
11
24
9
13
22
29
13
42
29
13
42
Sensitivity = 85%; Specificity = 55% E. Stemotomy 75% Total
Total
24
8
32
5
5
10
29
13
42
Sensitivity = 38%; Specificity = 82% D. Leg Lift 27
29
5
34
2
6
8
2
9
11
31
13
42
29
13
42
4
31
Sensitivity = 60%; Specificity = 93%
Sensitivity = 46%; Specificity = 94% E. Stemotomy
Between Myocardial Ischemia and Changes in Arterial Resistance During Coronary Artery Bypass Surgery
Rephael Mohr, MD, lnge Dinbar, MD, Yaron Bar-El, MD, Uri Goldbourt, PhD, Martin Abel, MD, and Daniel A. Goor. MD The arterial resistometer provides continuous on-line monitoring of changes in arterial resistance. Resistance index (Ri), which bears a direct relationship to systemic vascular resistance (SVR), is defined by the equation Ri = P’/(dP’/dt), where dP’/ dt is the peak dP/dt of the arterial waveform, and P’ is the pressure at dP’/dt. In 42 patients with unstable angina, changes in Ri were studied at six periods during aortocoronary bypass surgery before tracheal intubation, during tracheal intubation, leg elevation, presternotomy, sternotomy, and dissection of the internal mammary artery. Thirty-four episodes of ischemia (0.1 mV ST segment changes) were observed in 26 patients. All ischemic episodes were associated with increased Ri (mean increase, 102 2 52%). Elevation of the pulmonary capillary wedge pressure corre-
T
HE MOST widely used monitor of myocardial ischemia during anesthesia is the electrocardiogram (ECG). The diagnosis of ischemia is based on the development of ST segment elevation or depression of 1 mm or more, most often seen in leads V, and II.‘,’ Changes in arterial pressure, heart rate: pulmonary capillary wedge pressure (PCWP), rate pressure product,3 and triple product’ have been used as adjuvants in the detection of ischemia. Although the detection of new regional wall motion abnormalities by transesophageal echocardiography has been shown to be a more sensitive and earlier marker of myocardial ischemia than ECG changes,6 this technique is not feasible during the induction of anesthesia. The authors have shown recently that a new device, the arterial resistometer,’ which monitors the behavior of the peripheral resistance (referred to in this article as resistance index is capable of detecting myocardial ischemia in [Ril), patients during coronary balloon angioplasty’. In the present study of patients undergoing coronary artery bypass surgery, a significant increase in Ri was found during periods of myocardial ischemia detected by ECG changes. This finding suggests that an alternative method of detecting ischemia intraoperatively may now be available.
MATERIALS
AND METHODS
Forty-two consecutive patients with unstable angina were studied during surgery for aortocoronary bypass grafting. There were 34 men and 8 women, ranging from 34 to 74 years of age. Eighteen of the patients required an intravenous infusion of nitroglycerin to control rest angina. Ten of 42 patients underwent emergency surgery. One patient entered the operating room with angina pectoris and ischemic changes on the ECG, which were reversed before induction of anesthesia with intravenous nitroglycerin and sublingual nifedipine. Two patients who entered the operating room with ischemia that could not be controlled before induction of anesthesia were excluded from the study because a baseline nonischemic period could not be obtained. There were 8 patients with left main coronary artery stenosis, 28 patients had a history of previous myocardial infarction, 7 patients had left ventricular (LV) dysfunction (ejection fraction < 40%), and 9 patients were hypertensive.
Journalof Cardiothoracic and VascularAnesthesia,
lated with ischemia during the preintubation, intubation, and sternotomy periods, but not in the remaining periods. Changes in arterial pressure and heart rate were not good predictors of ischemia. The prevalence of ST segment changes increased markedly during all periods of anesthesia with increase in Ri (P < 0.05). Ninety-one percent of ST segment changes were associated with a 25% increase from the baseline Ri. Raising the cutoff point to a 2 75% increase in Ri improved the specificity of Ri in ischemia detection from 61% to 92%. An increase of 2 75% in Ri occurred in only 6% of cases without ST segment changes. It was found that an increase in Ri as depicted by the arterial resistometer was the best hemodynamic correlate of myocardial ischemia. Copyright o 1992 by W. B. Saunders Company Preanesthetic medication included diazepam, 10 mg, morphine, 10 mg, or meperidine, 75 mg, plus atropine, 0.5 mg, or scopolamine, 0.4 mg. All patients receiving beta-adrenergic blocking drugs preoperatively received the last dose on the morning of surgery. The anesthetic management was at the discretion of the anesthesiologist, and consisted of fentanyl, 10 to 40 Kg/kg, diazepam, 2.5 to 25 mg, flunitrazepam, 0.6 to 2 mg, and halothane, nitrous oxide, and oxygen. Muscle relaxation for tracheal intubation was achieved with succinylcholine, 1 mgikg, or pancuronium bromide, 0.1 mg/kg, and the latter agent was used for additional relaxation during surgery. The anesthetic plan was designed to produce a decrease in systolic blood pressure of less than 20 mm Hg compared with preinduction values, and to maintain the heart rate close to normal preoperative values.“.” Nitroglycerin (1 to 3 mg/h) was given intravenously to all patients except those with a left main coronary lesion. In addition, nitroglycerin boluses of 0.1 mg were given to control episodes of ischemia. The time between patient entry into the operating room and the beginning of extracorporeal circulation was divided into seven periods: (1) preinduction (control), (2) pretracheal intubation (the period from start of induction until the administration of succinylcholine), (3) tracheal intubation, (4) leg lift (while the legs were raised for application of povidine), (5) presternotomy (the time when the saphenous vein was removed), (6) sternotoniy, and (7) mammary (dissection of the internal mammary artery). In 10 patients the internal mammary artery was not dissected; therefore, data for this last stage could not be obtained. Intraarterial pressures were monitored via a 17-gauge, 20-cm long catheter (Bard-I-Cath, CR Bard International Ltd., Sunderland, England), introduced percutaneously into the femoral artery and connected by a 4-foot rigid pressure monitoring line (Cobe Laboratories Inc., Lakewood, CO) to a physiological pressure transducer (Mennen Medical 922-122-010, Rehovot, Israel). The
From the Departments of Cardiac Surgery, Anesthesiology, and Epidemiology and Preventive Medicine, The Chaim Sheba Medical Center, Tel Hashomer, Israel, and the Department of Anesthesiology, Mayo Clinic, Rochester, MN. Address reprint requests to Daniel A. Goor, MD, Professor of Surgery, Director, Department of Thoracic and Cardiovascular Surgery, The Chaim Sheba Medical Center, Tel Hashomer, Israel 52621. Copyright 0 1992 by W B. Saunders Company 1053-0770/92/0601-0009$03.OOiO
Vol6, No 1 (February), 1992: pp 33-41
33
34
MOHR ET AL
natural frequency of the monitoring system was 20 to 30 Hz and the damping coefficient ranged from 0.12 to 0.2, as determined by the Intraflo (Burron Medical Inc., Bethlehem, PA) fast flush method. ECG and pressure readings were obtained using an ECG and pressure monitoring unit (Mennen Medical 741, Rehovot, Israel). Standard leads II and V, were monitored visually on the ECG. Both leads were routinely compared with the preoperative ECG. To increase the sensitivity of ST segment change detection, the operating room monitors were calibrated so that deflection of 2 cm was equivalent to 1 mV. Monitors were protected against interference from electrosurgical devices, and equipped with noise reduction units that work via an averaging process. ECG changes were diagnosed directly from the monitor and were not recorded. ST segment depression or elevation were considered significant when they were at least 0.1 mV (horizontal or downsloping) or at least 0.15 mV (upsloping) 80 milliseconds after the J point.“.” Using the resistometer,* the first derivative of the peak arterial pressure pulse (dP’/dt) was obtained by electronic differentiation that provided a linear frequency response to 100 Hz. No correction was made for the frequency response of the catheter because only relative changes of dP’/dt were sought. Ri was monitored continuously with the same unit that detected the dP’/dt. The equation used by the arterial resistometer79 for calculation of the Ri is: P’ Ri = dP’/dt where P’ is the pressure at the time of peak dP/dt, dP’/dt is the peak dP/dt of the central (femoral) arterial waveform, and Ri is resistance index. AI1 hemodynamic and ECG parameters were measured and recorded every minute in the first 5 minutes, and every 2 minutes thereafter. For statistical analysis, one set of hemodynamic measurements consisting of heart rate, arterial pressure, peak dP’/dt, PCWP, and maximal Ri value was taken in each period from all patients. The preinduction period served as a baseline and ischemic changes and hemodynamic parameters in the remaining six periods were compared with the corresponding variables in this baseline, nonischemic period. After dividing the patients into those with ECG evidence of ischemia at any stage (group A) and those remaining ischemia-free (group B), a repeated measure analysis of variance (ANOVA) was performed. The difference between ischemic and nonischemic patients was examined using the seven consecutive determinations of the previously mentioned hemodynamic parameters. The repeated measure ANOVA examined the relationship between the presence of ischemia and the mean values of the various hemodynamic parameters. It was also used to evaluate the relationship between the stage of anesthesia on hemodynamic findings and time-ischemia interactions.14 Subsequently, for purposes of demonstration, a Student’s t test was used post-hoc to test the significance of mean differences in the various hemodynamic parameters between patients with and without ischemic changes. To calculate the specificity and sensitivity of Ri in the diagnosis of ischemia, the prevalence of ST-T wave changes was also examined in quartiles of Ri and dP’/dt. A x2 test was used to evaluate the nominal type I error (probability for independence between ST-T wave changes on the ECG and Ri or dP’/dt despite the evidence in the study sample to the contrary). Given the multiple comparison setting, these were viewed in light of the repeated measures ANOVA findings for between-patient and within-patient effects.
RESULTS
Twenty-six of 42 patients (62%) developed ST segment changes during the study, and comprised group A (isch-
emit). The remaining 16 patients (38%) never developed ST segment changes, and comprised group B (nonischemit). There were 34 ischemic episodes; 13 occurred before tracheal intubation, 7 new episodes started during sternotomy, 5 during leg lifting, 4 during mammary dissection, 3 during intubation, and 2 before sternotomy (Fig 1, Table 1). Data were obtained in 32 patients for all seven periods. Repeated measures ANOVA (Table 2) shows that for Ri there was a highly significant relationship with ischemia (between-subjects effect, P = 0.0002), but no dependency on either the stage at which Ri was determined, or stage-ischemia interactions (P = 0.85). In 10 patients, internal mammary artery dissection was not performed; therefore, only six periods were available for analysis. If the final period (mammary) is excluded from the analysis, results in 42 patients, for the first six stages, confirm a substantial between-subjects effect (P < O.OOOl), and no independent stage role or stage-ischemia interaction Given these findings, a post-hoc observation of the changes during each of the stages of anesthesia reveals that, during all stages of anesthesia, mean Ri levels as well as relative and absolute changes in Ri were significantly greater in patients with ECG ischemic patterns (Fig 2). All 34 ischemic episodes diagnosed by ST-T wave changes were associated with an increased Ri (Fig 1A). Most ischemic episodes were preceded by a moderate increase of Ri compared with the preinduction period (20%, P < 0.001) (Table 1). During ischemia the mean Ri increase was 102 + 52% (P < 0.001) and after reversal of ECG changes Ri decreased to 24 k 32%. The ANOVA also shows that absolute values of dP’/dt were significantly lower in patients in the ischemia group. However, changes from baseline values were not always significantly different from those found in patients without ischemia (Fig 3, Table 2). Whereas the ANOVA showed a significant effect of stage and stage-ischemia interaction on PCWP changes, elevations in PCWP demonstrated borderline statistical significance between groups (P = 0.08, Table 2). These changes were significant for ischemia during preintubation. intubation, and sternotomy, but failed to discriminate ischemic from nonischemic cases during the other three periods (Fig 4). Some of the ischemic episodes were associated with decreased PCWP, and mean changes in this parameter before and during ischemic episodes were not significant predictors of ischemia (Fig 1B). Changes in blood pressure were not a good ischemia predictor (Fig 1C). They were significant only during the presternotomy period. Figure 4 shows significantly lower blood pressure in patients with ischemia during this period. No significant difference in blood pressure was found between ischemic and nonischemic patients in the remaining periods. Heart rate also was not a useful indicator of ischemia in any period (Figs ID and 4). Maximum values of blood pressure and heart rate achieved during all periods of anesthesia were not different between ischemic patients (group A) and patients who did not develop ischemia (group B); mean * SEM, 155 2 5 versus 149 f 7 mm Hg, and 94 2 6 versus 97 _t 5 beatsimin. Patients in group A had significantly higher maximum levels maximal of PCWP (17 k 2 KS 13 f 1 mm Hg, P < O.OOS),
ISCHEMIA DETECTION WITH ARTERIAL RESISTOMETER
35
(m 19)
B
lal-
lM1%
nillO-
‘\,
‘s,
on m 9
I
:I
33
BEFORE INDUCTION
10 0 40
1
I
im
m-
BEF6RE ISCHEMIC EPlwDE
D
m-
L&?&BEFORE INWCllON
ISCHEW EPISODE
ISCHEYIC EPISODE
lsz% EPISODE
6E;ORE lNMlcnON
Fig 1. Hemodynamic changes before, during, and after ischemic episodes (ST segment changes). (A) Changes in Ri; (6) PCWP; (C) mean blood pressure; and (D) heart rate. (A) Preintubation; (*) intubation; (:) leg-lift; (0) presternotomy; (m) sternotomy; (0) mammary.
Ri (125 ? 6 vs 79 2 4 milliseconds, P < O.OOl), and significantly lower minimum achieved peak dP’/dt levels (1040 f 90 vs 1370 f 100 mm Hg/s, P < 0.05). Comparison between maximal Ri values achieved during anesthesia
induction and Ri observed before induction shows that patients in group A had both a greater absolute and relative percentage increase in normal values of Ri than patients in group B (65 f 6 1~s26 2 4 milliseconds, and 118 f 13% vs
36
MOHR
ETAL
Ri During Inductionof Anesthesia
Table 1. Time-courseof Patient NO.
Premduction
1
47
Preintubation
lntubation
65
72
63
70
78
Leg Lifting
Presternotomy
68
85*
88*
94*
88*
80"
79*
sternotomy
MSllm.Sy
92*
94*
90*
2
54
59
60
58
74
76
70
61
58
56
72
74
70
96*
lOO*
102*
105*
ill*
108*
3
59
63
69
71
78
82
74
73
75
78
78
80
79
105*
109*
104
82
84
80
4
68
65
71
73
50
48
44
47
46
44
46
47
45
63
75
54
48
45
46
5
42
78
79
76
68
61
63
80
84*
85*
88*
94*
92"
75
68
64
48
42
38
6
78
102*
107*
105*
93
9J*
lOO*
lOl*
104*
108*
lOO*
104*
99*
76
64
68
7
71
140*
169*
138*
llO*
108+
102*
119*
118*
125*
107*
100
98
184*
192*
174'
160*
133*
128*
8
56
71
64
63
80
81
77
85
86
89
80
77
78
82
88
77
182*
117*
168*
164*
160*
126*
122*
120*
132*
138*
134*
116'
113*
102"
84
80
77
104
107
130*
134'
129"
104
93
90
91
100
104*
103*
88
95
9
88
157"
10
78
96
11
49
69
74*
J4*
47
56
54
66
68
59
54
48
51
69*
68*
65
60
61
53
12
54
46
44
39
65
70*
77*
82*
88*
90*
SIX
92*
88
91*
92*
89*
91*
93*
90*
13
29
43
45
44
37
40
39
39
38
36
56
48
49
70
76
78
74
82*
84*
14
29
58
76
64
66
65
66
44
39
37
62
60
59
65
79
77
69
72
71
15
84
102*
105*
105*
158"
162*
138*
99
96
91
93
95
87
96
91
88
16
40
52*
53*
51*
58*
62*
63*
59*
56'
49*
39*
36
35
21
20
20
42*
46*
45*
17
56
29
41
40
33
32
31
44
48
50
41
35
37
49
50
51
50
43
41
18
50
59
60
55
50
53
49
65
69
70
80
82
62
44
43
44
35
34
29
19
60
66
70*
72*
76*
73*
74*
78*
83*
91*
94*
97*
63
38
36
34
93*
98*
lOl*
20
47
79
88
91*
lOO*
lOl*
95*
88
95
113
121*
124*
119*
116*
96
81
79
70
63
21
31
39
39
38
40
43
43
36
39
39
51
52
52
51
50
50
48
57
59
22
67
78
79
80
102*
161*
147*
98
52
55
88
93
85
56
54
48
23
45
60
62
63
68
70
69
85
91*
98*
79
76
71
102*
129*
103*
88
44
47
24
49
67
70
75
70
70
65
48
54
49
50
48
46
40
42
43
42
41
42
25
51
JO
75
78
91
92
86
76
74
71
70
66
63
55
54
50
63
79
91*
26
67
66
65
60
68
61
60
61
62
63
68
70
67
63
63
61
47
45
43
27
65
67
63
60
99*
61*
86
83
80
78
74
73
45
48
63
28
41
51
52
52
76
73
71
50
49
48
49
45
47
57
66
63
46
44
45
29
68
8
86
84
66
65
64
60
61
58
89
92
91
90
88
85
30
68
103*
124*
120*
117,
115*
97*
94
97
96
91
90
85
69
63
64
73
81
85
31
48
46
48
45
39
38
38
46
50
51
51
50
47
68
84
87
32
50
173*
177*
180*
169*
174*
180*
159*
149*
147*
164*
171*
184*
179*
163*
151*
33
77
113
127*
130*
87
80
77
118*
123*
127*
74
68
59
55
54
53
34
49
37
36
35
36
35
36
48
50
52
32
34
32
40
42
40
41
46
48
36
73
99
lOO*
120'
122*
120*
121*
135*
146*
142*
135*
127*
138*
143*
140*
137*
130'
129*
37
60
69
74
70
84
70
76
74
73
70
73
77
82
80
75
84
38
44
39
40
39
39
40
39
40
57*
74*
47
44
43
51
53
48
39
68
178
197
156
151
148
137
80
87
82
80
78
71
201*
265*
195*
40
40
53
95
116
80
74
89
51
48
44
47
38
42
65
115
116
72
63
51
41
57
60
68
66
75
79
77
89
96
99
69
58
39
42
40
41
51
72
77
42
37
65
79
98
90
82
88
94
98
99
98
96
88
99
104
92
74
65
63
43
37
43
52*
51*
50
48
46
44
53
57
35
33
29
60*
99*
100*
58*
64* 118*
71* 101
70* 94
50
*ECGischemicchanges
54 ? 9%, P < 0.001) (Fig 5). However, changes in dP’/dt levels compared with the preinduction period were not significantly different between the two groups (Fig 5). The probability of ischemia was examined in quartiles of Ri during the various periods of anesthesia. The prevalence of ST segment changes increased markedly in all stages with the increase in Ri (P < 0.05) (Fig 6). Table 3 relates the number of patients with ST-T wave changes to quartiles of maximum Ri values and minimum dP’/dt values measured for each patient. This analysis was performed to evaluate the relationship between relative changes in Ri and dP’/dt and the prevalence of ST segment changes. The probability for independence between ST segment changes and Ri was 0 (x’ analysis), and for dP’/dt, p < 0.001. One patient among those with maximum Ri above the median (102
milliseconds) did not show an ischcmic pattern on the ECG, whereas only one patient with maximum Ri at the lower quartile ( < 80 milliseconds) had ischemia, suggesting a good specificity and sensitivity for the index. The correlation between percentage increase in Ri and ECG evidence of ischemia was evaluated in Table 4. Using a cutoff point of 25% increase of Ri, a sensitivity of 85% to 100% (mean 91%) and a specificity of 55% to 69% (mean 61%) was obtained. A Ri increase of more than 25% was observed in 91% of ischemic episodes. Better specificity (82% to 95%, mean 92%) was obtained when the cutoff point was set at a Ri increase of 75% (Table 5). The sensitivity with this cutoff point decreased to 38% to 69% (mean 51%). The probability of nonischemic patients developing a Ri increase of more than 75% is only 8%.
ISCHEMIA DETECTION WITH ARTERIAL RESISTOMETER
37
Table 2. The Effect of lschemia on Hemodynamic Parameters.
2
Analysis Performed Using a Repeated Measures ANOVA
mmHg/sec
Effect of Effect of lschemia (between
Ri absolute values
subjects effect)
F value
(excluding mammary
P
18.7 0.0002
Anesthesia
stage-
stage
lschemia
(time effect)
Interaction
0.2
0.31
0.93
0.85
stage) ‘Ri absolute values
F value
(including mammary) Ri % Change (excluding mammary) ?? Ri
9.1
0.61
0.28
0.0045
0.65
0.87
0.24
0.48
0.94
0.78
PCWP
Abbreviations:
10.57
P
0.0024
F value
2.36
7.44
0.8
2240
8
3
11
1880-2240
5
5
10
1260-1880
1
10
11
< 1260
1
9
10
26
42
Total P < 0.001
-20
-40
-I
Fig 5. The relative changes in resistance index (IX) and the first derivative of the peak arterial pressure (dP’/dt). Comparison between ischemic (I@ and nonischemic (0) patients.
peak dP/dt, and dP’/dt is the peak dP/dt.7~Y The relationship P’/(dP/dt) has been termed the “Ri” (see Appendix). When “E” and “a” are not determined by the thermodilution technique, fluctuations in SVR, but not the absolute values, can still be followed by monitoring Ri.” A previous study’ has suggested that to obtain more reliable measurements of Ri, the arterial catheter from which the waveform is derived should be as central as possible. The femoral artery is probably superior to the radial artery or other No.
of
Patients
PREINTUBATION P-0.2
-0.1-18.4
P5.4
!I,
k’)
Abbreviations: Max Ri, maximal Ri; Min dP’/dt, lowest dP’/dt values.
PCO.
-60
16
41.2
P 0.1 mV ST segment deviation) also showed a significant increase in Ri. To determine the
ISCHEMIA DETECTION WITH ARTERIAL RESISTOMETER
39
Table 4. Diagnostic Accuracy of an Increased Ri* in Predicting lschemia (ST-T wave changes) Using a Cutoff of a 25% Increase in Ri Increase in Ri (%)
A Preintubation < 25% > 25% Total
lschemia
lschemia TOtal
(+)
1-l
> 25% Total
> 25% Total
Total
6. lntubation 2
18
16
1
17
12
12
24
13
12
25
28
14
42
29
13
42
Sensitivity = 92%; Specificity = 55% D. Leg Lift
16
2
18
20
0
13
11
24
9
13
22
29
13
42
29
13
42
Sensitivity = 85%; Specificity = 55% E. Stemotomy 75% Total
Total
24
8
32
5
5
10
29
13
42
Sensitivity = 38%; Specificity = 82% D. Leg Lift 27
29
5
34
2
6
8
2
9
11
31
13
42
29
13
42
4
31
Sensitivity = 60%; Specificity = 93%
Sensitivity = 46%; Specificity = 94% E. Stemotomy
