Intraoperative mortality is now so low that it is hard to quantify.  In contrast, postoperative mortality remains frequent and is arguably the third leading cause of death worldwide.  Bleeding-related and cardiovascular complications are the most common causes of 30-day postoperative mortality. Among cardiovascular complications, myocardial injury after noncardiac surgery is by far the most common. Myocardial injury after noncardiac surgery is defined by troponin elevations thought to be due to cardiac ischemia, with or without signs or symptoms. Myocardial injury after noncardiac surgery usually occurs within 2 days after surgery, and more than 90% is asymptomatic.  In contrast, troponin elevation with ischemic signs or symptoms defines a myocardial infarction. In fact, both result from myocardial necrosis and are usually type 2 (supply–demand) events. Although largely asymptomatic, mortality from myocardial injury after noncardiac surgery is nearly as high as from full postoperative myocardial infarctions. 

Postoperative cardiovascular complications do not occur randomly. Instead, major cardiovascular events nearly always occur in patients with pre-existing cardiovascular disease. A corollary is that baseline characteristics can be used to identify patients at greatest risk for postoperative cardiovascular complications. The concept is hardly new, and predictive systems such as the Revised Cardiac Risk Index have been in clinical use for decades.  The evidence that risk prediction focuses care and improves outcomes remains sparse. But at least in theory, identifying high-risk patients allows clinicians to direct intensive monitoring and care to patients most likely to benefit.

In this issue of Anesthesiology, Zhao et al. present a cardiovascular risk estimation model based on a prospective cohort of 2,820 patients.  Their base model was derived from 25 preoperative patient characteristics including age, type of surgery, and smoking status. The primary outcome was a composite of cardiovascular complications that included myocardial injury, myocardial infarction (a subset of injury), coronary revascularization, congestive heart failure, new atrial fibrillation, cardiac arrest, stroke, pulmonary embolism, and cardiovascular death. The model’s area under the receiver operating characteristics curve was 0.728, indicating moderate discrimination. The model was not internally validated, say with a split population analysis or resampling, much less validated in an external population where it is unlikely to work as well. Other cardiovascular models derived exclusively from baseline characteristics with full external validation report better discrimination. 

By definition, perioperative cardiovascular biomarkers also predict cardiovascular complications—and thus presumably enhance predictions derived exclusively from nonlaboratory patient characteristics. For example, troponin and N-terminal pro-B-type natriuretic peptide (NT-proBNP)  blood concentrations each predict postoperative cardiovascular complications and death.

Zhao et al. extend previous work by asking an important question: To what extent does each biomarker alone and combined enhance predictions? In their analysis, the Revised Cardiac Risk Index alone predicted the authors’ primary composite reasonably well. Their sophisticated analysis, based on continuous biomarker concentrations rather than arbitrary thresholds, showed that each biomarker enhanced prediction accuracy. Furthermore, prediction enhancement was comparable for each, with high-sensitivity troponin T slightly increasing the area under the receiver operating characteristics curve by 0.029 and NT-proBNP slightly increasing the area by 0.028. Unsurprisingly, the combination was yet better, improving the area under the curve by 0.042. None of these area increases is clinically meaningful.

The authors also defined a secondary composite restricted to myocardial infarction, coronary revascularization, cardiac arrest, stroke, and cardiovascular death. Unsurprisingly, the composite of more serious complications was much less common: 2.2% versus 15.4%. For the authors’ secondary composite, the Revised Cardiac Risk Index alone hardly predicted better than chance. In this case, adding biomarkers markedly improved predictions with high-sensitivity troponin T increasing area under the receiver operating characteristics curve by 0.167 and NT-proBNP increasing the area by 0.158. Once again, the combination was yet better, improving area under the curve by 0.196. Each of these area increases is highly meaningful.

For both the primary composite and the secondary composite of more serious complications, the combination of troponin and NT-proBNP improved predictions better than either biomarker alone—but the combinations were only slightly better. From a practical perspective, a single biomarker therefore provides most of the benefit and is therefore all that seems justified. For prediction, either would suffice. But there are compelling reasons to select troponin.

Because myocardial injury after noncardiac surgery is usually asymptomatic, it is rarely detected without troponin screening. Consequently, at least four major organizations including the American Heart Association (Dallas, Texas) recommend troponin screening in moderate- to high-risk surgical patients.  About 20% of surgical inpatients have elevated preoperative high-sensitivity troponin concentrations. Few of these patients are having infarctions, and treatment is rarely needed. Instead, elevated preoperative troponin concentrations identify patients at high cardiovascular risk.

A consequence of frequent high preoperative troponin concentrations is that postoperative troponin concentrations are difficult to interpret without a baseline concentration—and baseline values are therefore recommended. Preoperative troponin concentrations are therefore valuable because they are both necessary to fully interpret postoperative values and because baseline troponin contributes to accurate preoperative risk assessment. However, the analysis of Zhao et al. suggests that NT-proBNP provides little additional predictive value in patients who have preoperative troponin testing. Preoperative troponin alone may thus suffice in most patients and, of the two biomarkers, seems likely to eventually dominate.

Nonetheless, NT-proBNP has some distinct advantages and is likely to remain in common use for some time. For example, NT-proBNP is less expensive (although neither test is especially costly) and available on a point-of-care basis. An additional factor is that progressive cardiovascular risk ranges are well established for NT-proBNP: for example, less than 100 pg/ml (incidence, 0.3%), 100 to less than 200 pg/ml (incidence, 0.7%), 200 to less than 1,500 pg/ml (incidence, 1.4%), and 1,500 pg/ml or more (incidence, 4.0%).  There are comparable thresholds for troponin T, which is a Roche (Switzerland) brand and similar the world over. However, there are also about a dozen brands of troponin I, each of which differs. Cardiovascular risk prediction ranges for most are simply unknown, which means that they cannot easily substitute for NT-proBNP in terms of risk prediction.

Zhao et al. enrolled patients 45 yr and older who were scheduled for major noncardiac surgery. As in many previous analyses, myocardial injury was by far the most common cardiovascular complication and occurred in 12.7% of patients, whereas including all other components of the primary composite only further increased the incidence to 15.4%. The observed incidence of myocardial injury after noncardiac surgery was similar to previous reports in roughly comparable populations. 

The number needed to screen is the reciprocal of the detection fraction. In this case, that corresponds to 1 / 0.127 = 7.9. Thus, only about eight patients need troponin testing to detect myocardial injury after noncardiac surgery in patients older than 45 yr who have elective major surgery. (Myocardial injury after noncardiac surgery is far more common in patients having urgent or emergent surgery, with correspondingly smaller numbers needed to screen.) Just eight patients is a remarkably low number needed to screen, far lower than for nearly all routine laboratory tests.

Laboratory testing is generally considered valuable when results influence care. In the case of myocardial injury after noncardiac surgery, it clearly does, because there is much that can be done for patients who experience myocardial injury. Specifically, patients who have myocardial injury after noncardiac surgery should

  1. Be informed that they had a type of heart attack and that they are at high risk of having another 
  2. Be told that there are things they can do to reduce future risk including smoking cessation, healthful eating, exercise, weight reduction, and diabetes control
  3. Be considered for aspirin, statins, angiotensin-converting enzyme inhibitors, and anticoagulation 
  4. Have inadequately treated hypertension and tachycardia controlled
  5. Be referred to a cardiologists or internists who can provide long-term care

Troponin screening is a high-yield test that meaningfully guides clinical care. Consequently, major societies support troponin screening in moderate- to high-risk surgical patients.  Furthermore, as shown by Zhao et al., baseline troponin concentrations enhance predictions based on nonlaboratory baseline values, especially when predicting serious complications. A twofer! The anesthesia community should be leading the push to make perioperative troponin testing routine—but so far has not. It’s time we do.