Silent Ischemia as a Central Problem: Regional Brain Activation Compared in Silent and Painful Myocardial Ischemia

Methods

Study Sample

No patients in either study group had diabetes or any systemic disease. The baseline characteristics of the patients are shown in Table 1 and Table 5.

Patients with Silent Ischemia

We studied nine right-handed men (mean age ±SD, 62 ± 7 years) with significant coronary artery disease at angiography (at least one stenosis more than 50% of luminal diameter). Five patients had three-vessel disease, two patients had two-vessel disease, and two patients had had coronary bypass grafting but had recently developed occlusion of important grafts. Eight patients had been identified during outpatient investigation of stable exertional breathlessness; breathlessness had been found in two of these eight, who had previously had coronary artery bypass surgery for breathlessness, mild angina, and reduced effort tolerance. One patient had previously had an inferior myocardial infarction. The patients with silent ischemia were enrolled consecutively over a 14-month period. All had painless myocardial ischemia demonstrable by the development of ischemic electrocardiographic changes (> 0.1 mV downsloping or rectilinear ST-segment depression 80 milliseconds after the J point) and new regional wall motion abnormalities during dobutamine stress echocardiography in the complete absence of chest pain. None of the nine patients with silent ischemia showed a “mixed” clinical picture (angina on some occasions and painless ischemia on others). Resting ventricular function was normal in all patients except for the one who had sustained the infarction; inferolateral hypokinesia could be seen in that patient.

Patients with Angina

We also studied nine right-handed patients with stable angina pectoris caused by coronary artery disease (seven men and two women; mean age, 61 ± 7 years) who were recruited from the outpatient clinic. The data on these patients have been previously presented, in part [36]. All nine patients had typical angina pectoris, developed ischemic electrocardiographic changes with exercise stress, and had clinically significant, predominantly single-vessel, coronary artery disease. One patient with angina had had an inferior myocardial infarction with persistent hypokinesis of this wall, but resting ventricular function was normal in the other eight.

Positron Emission Tomography Scanning Protocol

Using a technique described previously [34, 35], we used dynamic positron emission tomography with H₂ ¹⁵O (from inhaled ¹⁵O-labeled carbon dioxide [C¹⁵ O₂]) to make six regional cerebral blood flow measurements for each patient. For each measurement, a 30-second background frame was recorded; scanning was then continued for an additional 2 minutes, during which time C¹⁵O₂ was administered at 500 mL/min and 6 MBq/mL activity. The scanner was an emission computed 931-08/12 positron tomograph (Computer Technology and Imaging-Siemens, Knoxville, Tennessee), whose characteristics have been described previously [38]. Myocardial ischemia was provoked by the increase in cardiac work caused by infusion of the β₁-agonist dobutamine, a drug that does not cross the blood-brain barrier [39] and has a short half-life (2.4 minutes) [40].

Scan Sequence

Before every scan, a 12-lead electrocardiogram and a brief echocardiogram (Challenge 7000, Esaote Biomedica, Florence, Italy) were recorded. Optimal echocardiographic views were chosen, on the basis of the previous dobutamine stress echocardiogram, as those that would best show the development of a new wall motion abnormality. The following sequence of scans was done, with approximately 12 minutes between scans, to allow for the complete decay of ¹⁵O radioactivity. All scans were done while the patient was lying on the scanner couch in a dimly lit room with eyes closed.

1. Baseline scan 1.

2. Placebo scan: done after 6 minutes of a saline infusion. Patients were unaware of the identity of the infused substance and were warned of the possibility of pain or unusual sensations in the chest due to the infusion, which was continued during the scan. The placebo scan was done to control for possible changes in regional cerebral blood flow caused by anticipation.

3. Baseline scan 2.

4. Low-dose dobutamine scan: done during intravenous infusion of low-dose dobutamine. The dose of dobutamine was 5 µg/kg per minute for 3 minutes and then 10 µg/kg per minute for 3 minutes. The latter dose was continued throughout the scan. Patients were blinded to the identity of the infused substance and were warned of the possibility of pain or abnormal sensations in the chest. This infusion was done to control for the effects of dobutamine on regional cerebral blood flow at doses insufficient to cause myocardial ischemia.

5. High-dose dobutamine scan: done during infusion of high-dose dobutamine, from 20 µg/kg per minute to 35 µg/kg per minute, increasing by increments of 5 µg/kg per minute every 3 minutes until myocardial ischemia was produced. From the onset of ischemia (confirmed electrocardiographically and echocardiographically), the infusion was continued at the same concentration (that is, at the ischemia-provoking dose) for the 2.5 minutes of scanning. Immediately after this, the infusion was stopped. In all patients with silent ischemia, chest pain was necessarily absent. However, awareness of an increase in heart rate or the force of contraction or both did not prevent inclusion in the study. In the patients with angina pectoris, the end point was typical angina pectoris associated with ischemic electrocardiographic changes [34]. Patients reported the sensation of pain (or lack of it) immediately before the 2.5 minutes of scanning. This sensation was scored on an arbitrary scale from 0 (no pain) to 10 (unbearable pain).

6. Baseline scan 3: done after ischemia had ceased and the electrocardiogram and echocardiogram had returned to baseline. This scan was done in all patients 13 minutes after the dobutamine infusion was stopped, to allow for complete clearance of the drug.

The echocardiograms were checked by a blinded independent echocardiographer, and the degree of agreement between the echocardiographer and the independent reviewer was noted.

Analysis of Positron Emission Tomographic Images

After realignment, the positron emission tomographic images were transformed into a standard stereotactic space defined by Talairach and Tournoux [41, 42]. The regional cerebral blood flow data were then corrected for global regional cerebral blood flow changes. Comparisons of regional cerebral blood flow were made between the different scans of the study by doing a series of t-tests (more precisely, a block-design analysis of variance) on a voxel-by-voxel basis using statistical parametric mapping (SPM) (Medical Research Council Cyclotron Unit, Hammersmith Hospital, London, United Kingdom) [34, 35]. Statistical parametric maps are spatially extended statistical processes that are used to characterize regionally specific effects in imaging data. The SPM(t) data were transformed to the unit normal distribution (SPM[Z]). The significance of each region was estimated using distributional approximations from the theory of Gaussian fields. This characterization is in terms of the probability that the peak height observed in a particular region could have occurred by chance [P(Zmax more than u)]; that is, it is a corrected P value [35].

The regional cerebral blood flow changes induced by myocardial ischemia were compared with the mean changes seen on baseline scans 1 and 2. In addition, the regional cerebral blood flow changes induced by placebo and low-dose dobutamine were compared with regional cerebral blood flow at baseline. Finally, because myocardial metabolic changes can persist for hours after ischemia [43], we also compared regional cerebral blood flow in the postischemic phase (baseline scan 3) with that in baseline scans 1 and 2 combined. These analyses permitted the construction of statistical parametric maps to describe significant changes in regional cerebral blood flow between the different test stages. Significant changes were identified by applying a statistical threshold of 0.05, corrected for multiple nonindependent comparisons [35]. Quantification of changes in regional cerebral blood flow equivalents (delta rCBF) between myocardial ischemia and baseline was done according to the following formula: Equation 1

After the positron emission tomography data from the two patient groups were separately analyzed, we directly compared the silent ischemia and angina pectoris data. Differences between the two patient groups with respect to the areas of increased regional cerebral blood flow during myocardial ischemia were computed as interactions between the between-group factors (angina compared with silent ischemia) and the within-group conditions (myocardial ischemia compared with baseline scans 1 and 2). Because the interaction effect could be predicted by the main effect (myocardial ischemia − baseline scans 1 and 2) in each group, a less harsh statistical threshold (P < 0.01, without Bonferroni correction) was used for the statistical maps.

Because of the nature of our statistical analysis, our results are considered in the form of pooled data sets for each patient group. Intra- and inter-patient comparisons were not feasible.

This project was approved by the Research Ethics Committee, Hammersmith Hospital, and the U.K. Administration of Radioactive Substances Advisory Committee.

Results

Clinical Observations

Patients with Silent Ischemia

No chest pain or other notable sensations were reported during the baseline scans, the placebo infusion, or the low-dose dobutamine infusion. During the high-dose dobutamine infusion, seven of the nine patients with silent ischemia were aware of a rapid and more forceful heart beat, one patient had slight warmth of the chest wall, and one patient had no unusual sensations. All patients denied that any of their sensations were in any way painful or uncomfortable. In all cases, the occurrence of silent myocardial ischemia was confirmed by electrocardiography and by the development of a new regional wall motion abnormality during real time two-dimensional echocardiography. The final scan (baseline 3) was done in all cases after the complete return of the echocardiogram and electrocardiogram to baseline. The degree of agreement between the echocardiographer and the independent reviewer for assessment of regional wall motion abnormality was 96%.

Patients with Angina Pectoris

Of the nine patients with angina, seven had typical angina pectoris in the form of retrosternal chest pain only; the other two patients also had a slight radiation of the pain to the left arm. The mean pain score (±SD) for the patients with angina was 6.3 ± 1.8. All of the patients with angina reported that the sensation of chest pain, rather than any awareness of a fast or unusually powerful heart beat, occupied their attention during the high-dose dobutamine scan. In all cases, the angina was accompanied by ischemic changes on the electrocardiogram. The final scan (baseline 3) was done in all cases after the complete cessation of pain or other unusual sensations in the chest and after the electrocardiogram had returned to baseline.

Hemodynamic Effects of the High-Dose Dobutamine Infusion

Resting heart rate, systolic blood pressure, diastolic blood pressure, and rate-pressure product (heart rate times systolic pressure) did not differ significantly between the patient groups. Both patient groups received similar intravenous doses of dobutamine to induce myocardial ischemia. Heart rate, systolic blood pressure, and rate-pressure product increased significantly in both patient groups in response to dobutamine, but the changes were similar in both groups. The time to significant ischemic electrocardiographic changes and the extent of these changes were also similar in the two groups. The data are presented in Table 1.

Findings on Positron Emission Tomography

Patients with Silent Ischemia

During myocardial ischemia, regional cerebral blood flow increased bilaterally in the brainstem, thalamus, and left hippocampal gyrus (Brodmann area 30) and, to a lesser extent, in the right and left dorsal and right lateral basal frontal cortical areas (Brodmann areas 6 and 47, respectively). The results are presented in Table 2 and Figure 1. The low-dose dobutamine infusion did not produce echocardiographic or electrocardiographic changes indicative of myocardial ischemia but was associated with a small increase in regional cerebral blood flow bilaterally in the thalami (Z scores, 4.7 [left] and 4.3 [right]) and in the left hippocampal cortex (Brodmann area 19) and right anterior cingulate cortex (Brodmann area 24) (Z scores, 4.7 and 3.9, respectively). The placebo infusion produced a small activation in the right frontal cortex (Brodmann area 46; Z score, 3.8). A comparison of the scan done after ischemia (baseline 3) with the previous baseline scans showed persistent increases in left thalamic and dorsal frontal cortical (Brodmann area 6) regional cerebral blood flow (Z scores, 3.8 and 3.9, respectively) (Figure 2).

Patients with Angina Pectoris

With angina, as previously noted, regional cerebral blood flow increased significantly in the hypothalamus, thalamus, hippocampal gyrus, and right extrastriate cortex and increased bilaterally in the lateral basal frontal (Brodmann area 47), mesial orbitofrontal (Brodmann area 10), and ventral cingulate (Brodmann area 25) cortices. The results of the regional cerebral blood flow changes during angina compared with baseline are presented in Table 3 and Figure 1. Regional cerebral blood flow did not differ between the placebo infusion and the resting state or between the low-dose dobutamine infusion and the resting state. After the cessation of myocardial ischemia (baseline 3), regional cerebral blood flow remained elevated in the thalami and dorsal frontal cortex (Brodmann area 6) (Figure 2).

Differences between Angina Pectoris and Silent Ischemia

An overall comparison of the results for myocardial ischemia and angina with baseline in the two patient groups showed that angina was associated with greater bilateral increases in regional cerebral blood flow in the anterior and ventral cingulate cortex (Brodmann areas 24/32 and 25), mesial orbitofrontal and basal frontal cortex (Brodmann areas 10 and 11), and left temporal pole (Brodmann area 28) (Figure 1, Figure 2, and Table 4).

Discussion

Our study shows that during silent myocardial ischemia, focal cerebral activation can be detected and significant regional cerebral blood flow increases can be seen in subcortical structures and, to a lesser extent, in a small area of the frontal cortex. During overt cardiac pain, however, a much more extensive pattern of activation is seen, especially at the cortical level in the ventral cingulate and basal frontal cortices. In addition, thalamic and dorsal frontal activity persisted through the scan done after ischemia in both groups.

An equivalent stress on the hearts of patients with angina and patients with silent ischemia produced the same degree of thalamic activation, but the patients with silent ischemia had significantly less cortical activation, especially with respect to the anterior and ventral cingulate and basal frontal cortices. These findings show that afferent stimuli from the heart do reach the central nervous system in patients with silent ischemia; therefore, the absence of sensation of chest pain is unlikely to be due to failure of transmission by the peripheral nerves. Our results indicate a difference in the handling by the central nervous system of the afferent signals from the heart in patients with silent ischemia. One hypothesis to account for this would be gating [37] of afferent stimuli at the thalamic level.

The principal differences between the patients with silent ischemia and the patients with angina was the absence of activation of the basal frontal and anterior and ventral cingulate cortices and the left temporal pole. The two patient groups were closely matched for resting ventricular function, dobutamine dose, hemodynamic changes due to dobutamine, time to onset of ischemia, and extent of ischemia. It might therefore be reasonable to suggest that the threshold for myocardial ischemia was similar in both groups. For this reason, we propose that the cerebral areas just described represent specific cortical projections of a pathway mediating the perception of pain from the heart.

The anterior cingulate cortex has been associated with emotional responses to pain [44] and has documented connections to the nucleus of the solitary tract [45], to the dorsal motor nucleus of the vagus, and to the sympathetic thoracic intermediolateral cell column [46]. In addition, studies of the rhesus monkey have shown connections between the frontal cortex and the anterior cingulate cortex [47, 48]. That perception of chest pain involves the cingulate cortex is therefore entirely consistent with current neuroanatomical data. A potential regulatory role for the frontal cortex—modifying the emotional response to angina in light of previous subjective experience of it—is also possible.

Our investigation of both patient groups indicated that the thalamus continues to receive inputs from the heart after myocardial ischemia has ceased; this is consistent with the observation that metabolic disturbances in the heart continue for hours after ischemia has apparently resolved [43]. Presumably, the less intense signal is not transmitted further because it is below the necessary threshold (the Z scores for thalamic activation at this stage were substantially lower than those during myocardial ischemia).

One hypothesis [5, 7, 8, 14, 15] is that silent myocardial ischemia is merely milder than the ischemia that produces angina pectoris. This is not supported by our data. The severity of coronary artery disease in our patients with silent ischemia was similar to if not greater than that in the patients with angina pectoris (more vessels were diseased per patient). The maximum dobutamine dose and the increase in cardiac work that this dose produced were equivalent in the two groups. Finally, in patients with silent ischemia, even low-dose dobutamine infusion evoked a significant bilateral increase in thalamic regional cerebral blood flow. This might suggest that this group had a ready induction of myocardial ischemia compared with the patients with angina. It must be noted, however, that the greater extent of coronary disease in the patients with silent ischemia does not necessarily translate into a greater degree of myocardial ischemia.

Our findings suggest that a stimulus of intensity similar to that of painful or silent ischemia can induce myocardial ischemia that, in certain circumstances (those that make the ischemia silent), may or may not be associated with a painful sensation. We are clearly far from being able to predict the code of afferent stimuli that represents an adequate algogenic signal. This absence of predictability might reflect a complex pattern of afferent impulses related to mechanical or chemical stimulation, or both, rather than the relatively simple involvement of a specific nociceptive pathway [49]. A further influence on the pattern of afferent impulses could be ischemia that is diffuse in origin, such as that in patients who have hypertrophic cardiomyopathy despite a normal coronary arteriogram. It could still be claimed that the afferent signals that reach the thalamus during myocardial ischemia in patients with silent ischemia (which we assume to be gated from further transmission) are not painful signals but are proprioceptive inputs from mechanoreceptors in the myocardium. In addition, as far as is known, no nerve endings in the myocardium have the specific function of mediating pain signals [49]. If proprioceptive inputs to the heart could be shown to be increased during silent myocardial ischemia, this would further support the argument that cardiac nociception depends on the intensity of general afferent signals from the heart rather than on a specific pain pathway (the intensity rather than the specificity theory of cardiac pain mediation [49]). Our study does not address the precise mechanism of gating, which we hypothesize occurs in the thalamus. Further research at the neurotransmitter level might elucidate this.

In an earlier report [36], we compared and contrasted the regional cerebral blood flow changes in angina pectoris with those in somatic pain [50, 51]. The results of our present study are entirely consistent with those in our earlier report. The point raised in that report—that the ventral cingulate and frontal cortices rather than the dorsal cingulate and somatosensory cortices for somatic pain can be considered the central projections for visceral pain—would still hold true.

Study Limitations

Our study had several technical limitations. First, we used only one method to induce myocardial ischemia. Exercise would have been impractical because the patient's position had to be kept constant for optimal scanning. The pharmacologic stressors dipyridamole and adenosine both cause cerebral vasodilatation, which could have confounded the measurements of regional cerebral blood flow. (Whether myocardial adenosine release, caused by myocardial ischemia, influences the pathways identified in this study could be tested by repeating the study with and without an adenosine antagonist, such as aminophylline.) Pacing stress was feasible, but this procedure is excessively invasive.

Second, cerebral regions not involved in pain perception could have been activated, but, averaging across patients, the inclusion of the additional baseline scans and the uniformity of scanning circumstances (each patient had his or her eyes closed in a dimly lit room) should have minimized any effect on the results. The comparison of the two patient groups also reduced the possibility of false-positive areas of activation.

Third, the order in which the scans are done is of methodologic importance. Thus, the placebo and dobutamine infusions should ideally have been administered in a random sequence. We opted for the sequence described above because earlier evidence had shown the long-term persistence of the effects of myocardial ischemia [43]. Again, direct comparison of the results in the two patient groups will have reduced the influence of the scan sequence on the final results.

Fourth, a criticism [52, 53] of our earlier report was that the regional cerebral blood flow changes we observed may merely have been a direct pharmacologic effect of the dobutamine infusion rather than a reflection of a true neurophysiologic response to angina. The present study, in which the study groups had the same maximum ischemia-inducing dose of dobutamine but clear differences in their regional cerebral blood flow responses, refutes this criticism. However, incorporating either normal persons or patients with mild coronary disease (who would not develop myocardial ischemia with a dose of dobutamine similar to that used in our study) would allow more detailed assessment of the significance of the data from our patients with silent ischemia and angina pectoris.

Fifth, the cortical activation observed during angina pectoris may have been an epiphenomenon—a nonspecific, secondary response to the pain—and the essential differences between the patient groups may lie in differences in thalamic activation that cannot be spatially resolved by the methods we used. On purely experimental grounds, it is impossible to determine which activation is a consequence of pain perception and which reflects the perceptual event per se. Such a determination might be possible in a study of patients with angina who perceived pain but felt no affective response toward it (such as patients with angina who had previously had frontal leukotomy). However, it was possible to identify cortical areas that are known, on the basis of invasive experiments in primates, to have direct specific visceral connections (for example, the ventral cingulate [Brodmann area 25]). Thus, we believe that the cortical activations are more than incidental.

Sixth, the extent of coronary artery disease in our patients with silent ischemia was greater than that in our patients with angina (predominantly three-vessel disease compared with predominately single-vessel disease), and this might have affected the afferent signals from the heart. However, because resting ventricular function, hemodynamic changes caused by dobutamine, time to onset of ischemia, and extent of ischemia were similar in the two groups, we suppose that the signals initiated in the hearts of the patients in the two study groups would be equivalent. This, however, needs to be confirmed by other means.

Given the difference in the extent of coronary artery disease in the two study groups, if the pattern of afferent impulses during myocardial ischemia were influenced by whether the ischemia came from a small myocardial region or was more diffuse in origin, there could be implications for the present study.

Seventh, for the purposes of our study, the definition of silent ischemia excluded a “mixed pattern” (myocardial ischemia that was painful on some occasions and painless on others). It is at least theoretically possible that the mechanism of the silence of myocardial ischemia in patients with the mixed pattern is different from that of the patients we studied, who had “pure” silent ischemia. Ideally, our protocol would be repeated in the same patient during both painful and painless ischemia. This could be difficult to achieve, but the use of a moderate stressor, such as mental arithmetic, in severely symptomatic patients with coronary disease might make it feasible. A further point is that our data, obtained from patients with pure silent ischemia, might not be completely applicable to older patients with silent ischemia or to patients with diabetes. In both of these groups, peripheral neuropathy might play a significant role [17, 18].

Eighth, our group of patients with silent ischemia included two patients who had had coronary bypass surgery. It is just conceivable that this surgery could have led to the division of the afferent nerve fibers, which may have affected pain perception. However, omitting these patients made no significant difference to the data.

Ninth, our study would have yielded more information if the same patients had had a somatic painful stimulus and if a direct comparison had been made between the regional cerebral blood flow effects of the visceral and of the somatic stimuli. However, the extra discomfort and exposure to radiation that this would have occasioned were ethically unacceptable, and there was no guarantee that any somatic pain provoked would not in turn induce angina or silent myocardial ischemia. In addition, somatic and myocardial afferent pain signals may have different thresholds [54].

Finally, prospective evaluation of consecutive patients in each study group during the same time period would enhance a study such as this, as would a more formal system of echocardiographic wall motion scoring.

Conclusions

We show that afferent stimuli from the heart do reach the central nervous system during silent myocardial ischemia; this suggests that peripheral neuropathy is an inadequate explanation of the silence of silent ischemia. Gating of the afferent signals at the thalamic level is a possible mechanism. Consistent with the absence of specific myocardial nociceptors, our data support the notion that pain from the heart is felt when the intensity of afferent firing is adequate rather than when a specific group of nerve endings in the myocardium is stimulated. Altered handling by the central nervous system of afferent signals from the heart contributes to the lack of perception of chest pain in patients with silent myocardial ischemia.

This paper was presented in part at the 67th Scientific Sessions of the American Heart Association, Dallas, Texas, November 1994.

Dr. Paulesu: INB-CNR, Istituto Scientifico H San Raffaele, Universita degli Studi, Milano, Italy.

Drs. Nihoyannopoulos and Tousoulis: Clinical Cardiology, Division of Medicine, Hammersmith Hospital, Du Cane Road, London W12 0HS, United Kingdom.

Drs. Frackowiak and Frith: Wellcome Department of Cognitive Neurology, Institute of Neurology, Queen Square, London WC1N 3BG, United Kingdom.