Myocardial Infarction after Noncardiac Surgery 

After we received institutional review board approval and written informed patient consent, we enrolled in this prospective, observational study 323 patients aged 50 yr or older who had ischemic heart disease (IHD) and were admitted to the hospital for elective noncardiac surgery. Entry criteria included the presence of any of the following: typical angina history diagnosed by an admitting or a consulting physician, unequivocal history of previous myocardial infarction (MI) documented by an admitting or a consulting physician, ECG evidence of previous MI consisting of Q waves lasting >or= to 0.04 s and 1-mm deep in at least two contiguous leads, or angiographic or scintigraphic evidence of IHD documented in the medical chart. Exclusion criteria included the presence of unstable angina, MI < 6 months, anticipated need for treatment in the postoperative intensive care unit, skeletal muscle disease, or hemolytic anemia.

During surgery, patients received routine pre-, intra-, and postoperative care at the discretion of their attending physicians and surgeons without the knowledge of study enzyme and ECG results. After surgery, for the purposes of the study, patients had daily clinical assessments (at 10:00 A.M.), and daily 12-lead ECGs (at 10:00 A.M.). Isoenzyme measurements included creatine kinase (CK; activity and mass), CK-2 activities, CK-2 mass, and Troponin-T, all obtained on the operative night (at 8:00 P.M.), twice daily on PODs 1–4 (at 8:00 A.M. and 8:00 P.M.), and then daily on PODs 5–7 (11:00 A.M.). Troponin-T became available for use starting with patient 93. The study nurse (T.V.N.) did the daily clinical assessment; by direct questioning, she sought evidence of the presence of chest pain, dyspnea, palpitations, light-headedness, or diaphoresis occurring at any time since the previous assessment. The study nurse also reviewed the chart for information on medication changes and tabulated 24-h opioid use in morphine equivalents. She also determined the amount of surgical pain the patients were experiencing using a visual analog scale (0 = no pain, 10 cm = worst pain possible) and assessments of heart rate and blood pressure. The hospital laboratory measured total CK and CK-2 activity spectrophotometrically, whereas total CK, CK-2 mass, and Tropronin-T were measured using monoclonal antibody immunoassay. A technetium-99m pyrophosphate scan would be ordered in the presence of equivocal enzymatic and ECG findings before Troponin-T measurement became available. The pyrophosphate scans were performed 2–3 days after the suspected event and were interpreted by a nuclear medicine physician who was unaware of the patient's clinical course.

The diagnosis of PMI was defined prospectively to require the presence of total CK > 174 L/l and two of the following:(1) CK-2/CK (mass or activity) > 5%, (2) presence of new Q waves lasting >or= to 0.04 s and 1-mm deep in at least two contiguous leads, (3) Troponin-T > 0.2 micro gram/l (upper limit normal at our institution), or (4) a positive pyrophosphate scan. A diagnosis of PMI was made by a cardiologist (J.E.B.) unaware of the patient's clinical condition. Timing of the PMI was based on enzyme and ECG changes interpreted by the cardiologist. Patients were also followed at 1 yr with a telephone interview done by the study nurse to determine any changes in their cardiac status.

After the study was complete, a post hoc analysis and comparison of the different diagnostic criteria was also done. Statistical analysis consisted of comparisons with unpaired t tests for parametric data and chi-square testing for nonparametric data. A P value < 0.05 was considered significant.

Eighteen of the 323 patients (5.6%) had a PMI, and three (17%) of them were fatal. Table 1lists the clinical characteristics of all the patients having a PMI, including the symptoms present, the method of diagnosis (excluding mandatory elevated total CK), ECG findings (Q wave vs. non-Q wave), and cardiac outcome at 1 yr. Chest pain developed in eight patients (44%) during the postoperative period, but only 3 of 18 (17%) had chest pain at the time of the MI. Ten (56%) of the patients developed significant clinical findings including atrial fibrillation (n = 2), hypotension (n = 4), or pulmonary edema (n = 4). Seven patients (39%) exhibited no clinical evidence of infarction at the time of their MIs. Figure 1shows the timing of the PMIs. Most PMIs occurred on the operative night and first POD, which is a non-normal distribution (P < 0.01). The ECG classification of the PMI was determined to be Q wave in 6, non-Q wave in 10, and indeterminate (left bundle branch block, inadequate recording) in 2. Of the three fatal PMIs, two were non-Q-wave infarctions and one was a Q-wave infarction.

(Table 2) shows a comparison between the demographic variables of patients in whom PMI developed and those in whom it did not. The PMI patients were significantly older, shorter, weighed less, and used more nitrates than did the patients who did not have a PMI. There were, however, no differences in the presence of IHD risk factors, method of IHD diagnosis, gender or procedure distribution, type of anesthetic received, intraoperative or postanesthesia care unit hypotension, or in the use of beta blockers or calcium channel blockers between the two groups of patients. Table 3lists the postoperative data, including opioid use, pain scores, and heart rate. Although there was no difference in pain scores between the two groups, there was a strong trend in which PMI patients required less opioid. The PMI patients also had significantly higher heart rates and no difference in blood pressure (data not shown).

At the 1-yr follow-up evaluation, 2 of the 15 (13%) surviving PMI patients had had a change in their cardiac status. One of seven asymptomatic survivors had suffered a fatal MI, whereas one of the clinically evident PMIs had developed unstable angina. The other 13 patients had no change in their cardiovascular status at the time of their follow-up telephone interview. Twenty-two of 280 (7.9%) non-PMI patients interviewed at 1 yr had sustained a cardiac event (fatal MI, n = 4; nonfatal MI, n = 8; and unstable angina, n = 10), whereas 25 were lost to follow-up. This event rate was similar to that in the PMI patients (P = 0.44).

Because the diagnosis of PMI has varied among different investigators, we did a post hoc analysis using different MI diagnostic criteria;Table 4shows the results. If the criteria for diagnosis of PMI required the presence of Q waves, or CK-2 elevation > 5% with ECG changes, or an autopsy diagnosis (DX-2), three additional PMIs would have been diagnosed, but patients 4, 12, 69, and 285 would have been removed because of indeterminate or no ECG changes yielding one less PMI on POD 2. At 1 yr, one of these patients had suffered a fatal MI, one a nonfatal MI, and one had developed unstable angina. The use of a PMI diagnosis similar to DX-2 but one that used Troponin-T > 0.2 micro gram/l with ECG changes (DX-3) instead of CK-2 > 5% would have led us to diagnose 26 of 231 (11.2%) patients as PMIs, with a peak incidence on POD 1. The 1-yr follow-up evaluation of these patients indicated that one had suffered a fatal MI, one a nonfatal MI, and one had developed unstable angina. If the ECG requirement with the elevated enzymes was removed from DX-2 and DX-3 (DX-4), 67 of 323 (20.7%) patients would have been diagnosed as having a PMI, also with a peak incidence on POD 1. Of these 67 patients, two had suffered fatal MIs, two had sustained nonfatal MIs, and two had developed unstable angina, whereas five were lost to follow-up at 1 yr. All four diagnostic criteria would have diagnosed the three fatal PMIs. The difference in PMI diagnosis rate between DX-1, DX-2, DX-3, and DX-4 is significant (P < 0.001), whereas the difference in event rates at 1 yr was not significant (P = 0.64), nor was the rate of fatal PMIs (P = 0.21).

We also compared the diagnostic value of CK-2 and Troponin-T measurements in the 231 patients in whom both were assayed. In patients who had postoperative ECG changes with one or both of these enzyme increases, 11 patients had elevations of CK-2 and Troponin-T, whereas an additional 2 patients only had CK-2 elevations and 10 patients had only Troponin-T elevations. Finally, four patients had pyrophosphate scans of which three were positive (two PMIs in patients 5 and 186), and one was negative (PMI patient 216). The patient with a positive result of a pyrophosphate scan but who was not diagnosed as PMI had an elevated total CK and a borderline elevation of CK-2 mass.

In this study of patients with known IHD, we observed a 5.6% incidence for PMI after noncardiac surgery. This event rate is consistent with earlier studies. [1]This incidence of PMI is consistent with other studies of patients with documented IHD. [10,11]This is in contrast to the lower event rates of 2–4% noted when study patients only have risk factors for IHD [9,12,13]and approaches zero when patients do not have known IHD. [14]*

A major finding of our study was that the peak incidence of PMI occurred during the first postoperative night, with a decreasing incidence thereafter. This is different from earlier studies that noted a peak incidence at about the second POD. These studies by von Knorring [11]and Mangano et al., [9]however, only obtained isoenzyme measurements when clinical indicators were present. The observation that PMI may be an earlier event has, however, been noted by other researchers. [8,12,15]However, the study by Raby et al. [8]only involved the first 72 h, whereas Landesberg et al. [15]only measured isoenzymes regularly for the first 24 h and then only when clinically indicated. Because our study comprehensively measured isoenzymes for the first 7 POD, we documented more conclusively that PMI is an early postoperative event in patients with known IHD.

Another major difference in our study was the criteria for PMI diagnosis. There is no gold standard for the diagnosis of PMI. Our criteria were similar to another recent publication [7]and were designed to require the presence of an indicator of high sensitivity (elevated total CK) and at least two indicators of high specificity (elevated CK-2, elevated Troponin-T, Q waves, or a positive result of a pyrophosphate scan). We chose to exclude ST-T wave ECG changes because of their potential to be obscured by common postoperative factors such as fluid and electrolyte imbalances, surgical dressings, and abdominal or thoracic complications (e.g., paralytic ileus, pneumothorax). In fact, the addition of ECG ST-T changes as described in DX-2 and used by many authors [5,6,8–14,16–21]resulted in three additional PMIs but four exclusions, for a net loss of one PMI and no change in the peak POD incidence. The use of the more sensitive Troponin-T assay, [7,22]instead of CK-2 with ECG ST-T changes as in DX-3, would have nearly doubled the incidence of PMI, whereas the peak POD occurrence of PMI would have become POD 1. This is still earlier than noted by other investigators. However, Troponin-T assays were not available when the study began, and their increased sensitivity has only recently been documented with echocardiographic controls. [22]The use of enzyme elevations independently, as in DX-4, was recently adopted by some authors. [3]Interestingly, this is a change from these authors' earlier studies [5,9,13,18,21]that was made without any explanation. The use of elevated CK-2 alone as in DX-4 was also recently shown to offer no clinically important information. [23]This opinion would be substantiated by the lack of a difference in the rates of fatal PMI and 1-yr cardiovascular events when used with our database.

Other findings of note from our study include the fact that only 17% of patients presented with chest pain at the time of their MI. This may have been a result of their concomitant use of opioids for their surgical pain. However, these patients received fewer narcotics but reported similar pain scores. This apparent increased pain tolerance has been noted in nonsurgical patients with silent ischemia. [24,25]With the exception of the study by von Knorring, [11]which was done before isoenzymes could be measured, our low rate of chest pain is consistent with studies documenting the presenting features of PMI patients. Most PMI patients did manifest other signs and symptoms of cardiovascular origin, as other investigators have noted. [9,13,15,16,26] 

Our study found a preponderance of non-Q-wave infarctions. This has also been noted by others in recent studies. [12,15,17,26]Again, the only study without this finding was that by von Knorring, [11]who probably missed many PMIs based on his inability to use isoenzyme assays, as noted before. This preponderance of non-Q-wave infarctions is different from nonsurgical patients presenting to emergency rooms, where the predominant type of MI is a Q-wave infarction. [27]Mechanistically, this may indicate that PMIs are more often the result of prolonged ischemia rather than thrombotic occlusion of a coronary artery, [17]similar to the presumed pathophysiology of silent ischemia. [28]The adage of avoiding hypotension, hypertension, and tachycardia, all of which alter myocardial oxygen supply and demand, therefore should continue to be applied to the postoperative period. Our finding of elevated heart rates in the PMI group is compatible with this hypothesis, although our method could not determine if the elevated heart rate was a cause or an effect.

Finally, we tried to compare the long-term outcome of silent and symptomatic PMI patients. At our 1-yr follow-up evaluation, there was no difference in cardiac outcome between silent and symptomatic patients. Yeager et al. [26]made a similar observation, although they also reviewed patients at 4 yr and noted a higher cardiac complication rate in the symptomatic patients at this time. The implication is that the symptomatic PMIs have worse long-term prognoses (> 1 yr) but similar short-term ones, for reasons that are unclear at this time. We also noted no difference in cardiac event rates at 1 yr between PMI and non-PMI patients, which differs from previous findings by Mangano et al. [18]This difference may be due to the fact that our study included only patients with known IHD and therefore at a higher long-term risk in comparison with Mangano et al.'s [18]inclusion of patients at risk for IHD with a subsequently lower long-term risk. Conversely, Mangano et al. [18]monitored patients with visits by a study physician, in contrast to our use of a telephone interview, with its likelihood of missing significant information. This approach has, however, been used before for similar cardiac long-term outcome studies. [6,19] 

There was a lack of uniformity in isoenzyme measurement for all patients in our study. We added Troponin-T measurements in an attempt to use this more specific marker of myocardial injury [22]when it became clinically available. Conversely, a strength of our study is that the only previous studies to enroll more than 300 patients at one institution used either no isoenzyme measurement [11]or CK-MB alone. [10,16,26] 

In conclusion, in this study, which intensely followed patients with known IHD who were having noncardiac surgery, we found that PMI was an early event only occasionally associated with chest pain and usually non-Q wave in nature. One-year follow-up evaluation suggested that patients with silent or symptomatic PMIs have similar short-term outcomes.

This study was initiated by the late Dr. Richard Knill. The authors thank Catherine Hawke for her dedicated secretarial assistance.