Utility of pre-operative cardiac biomarkers to predict myocardial infarction and injury after non-cardiac surgery

Although non-cardiac surgery has its own benefits, it initiates inflammatory, hypercoagulable, stress, catabolic, and bleeding states that can precipitate perioperative cardiovascular complications.¹ Moreover, a large proportion of adults undergoing non-cardiac surgery have the substrate (e.g. coronary artery disease) that puts them at risk of cardiovascular complications. A study of 10 million adults, aged ≥45 years having major non-cardiac surgery in the USA, demonstrated that 24% had atherosclerotic disease and 45% had ≥2 cardiovascular risk factors.²

Given the high proportion of patients at risk of perioperative cardiovascular complications, accurate pre-operative risk predication is important. This facilitates informed decision-making about the appropriateness of surgery and can inform the surgical approach (e.g. open, laparoscopic), type of anaesthesia (e.g. general, regional), the optimal post-operative monitoring setting (e.g. surgical floor, step-down ward), and the need for post-operative monitoring (e.g. cardiac troponin measurements).

Research has informed the value of the following four approaches to pre-operative cardiovascular risk prediction: clinical risk indices, functional capacity, non-invasive cardiac testing, and cardiac biomarkers. In this edition of the journal, Meister and colleagues³ make an important contribution to this patient-important issue. Several clinical risk indices exist, but research demonstrates they have limited discrimination capabilities and commonly underestimate risk.⁴ This finding is further demonstrated in Meister’s paper.

Several studies have assessed patient self-reported, physician estimation of, or index scores of functional capacity.^5,6 These studies have demonstrated mixed results, and the most favourable results suggest, at best, modest predictive capability. The only large study in this area demonstrated that self-reported functional capacity was independently associated with major adverse cardiovascular complications but did not improve discrimination over a clinical risk index.⁷

Because of the inconvenience and cost associated with pre-operative non-invasive cardiovascular testing, patients should only have such testing if it will provide independent prognostic information beyond a simpler method of risk prediction (e.g. clinical risk index, biomarker) and improve overall risk reclassification. Studies demonstrate that pre-operative echocardiographic parameters do not improve pre-operative risk prediction beyond a clinical index or pre-operative N-terminal pro-B-type natriuretic peptide (NT-proBNP) measurement.⁴

An international, prospective, blinded, cohort study of 955 patients who had pre-operative coronary computed tomography angiography (CCTA) demonstrated that pre-operative CCTA improved perioperative risk prediction beyond a clinical risk index in patients who had the primary outcome (i.e. cardiovascular death or myocardial infarction); however, CCTA findings were >5 times as likely to inappropriately overestimate risk among patients who did not suffer the primary outcome.⁸ The net absolute effect is that pre-operative CCTA will inappropriately estimate risk in 8% of patients compared with using only a clinical risk index.

Studies evaluating the pre-operative prognostic capabilities of pharmacological stress echocardiography, pharmacological radionuclide imaging, and cardiopulmonary exercise stress testing have typically been small with methodological limitations (e.g. lack of blinding).⁴ The best data do not support that these non-invasive cardiovascular stress tests enhance perioperative cardiovascular risk reclassification beyond pre-operative clinical risk factors or biomarkers.⁴

Given this research demonstrating the uncertain, suboptimal, and net inappropriate perioperative cardiovascular risk reclassification with pre-operative clinical risk indices, functional capacity, and non-invasive cardiac testing, and the strong prognostic capabilities of cardiac biomarkers in the non-surgical setting, researchers have evaluated pre-operative cardiac biomarkers. To date, three cardiac biomarkers have been evaluated in large high-quality studies.

Before discussing these three biomarkers, there is a need to review relevant nomenclature. Myocardial infarction after non-cardiac surgery was the first reported major perioperative cardiovascular complications, with studies published in the 1920s and 1930s.^9,10 One of these studies reported that most patients who had a perioperative myocardial infarction did not experience ischaemic chest pain and speculated that narcotics had masked the ischaemic symptoms.¹⁰

Given that most patients having a perioperative myocardial infarction do not experience ischaemic symptoms and ischaemic electrocardiography (ECG) changes may be missed due to the lack of symptoms to trigger obtaining an ECG at the time of ischaemia,¹¹ in 2014, to avoid missing prognostically important ischaemic myocardial injuries, the VISION investigators established the diagnosis of myocardial injury after non-cardiac surgery (MINS).¹² Their analysis demonstrated that myocardial injury adjudicated as resulting from ischaemia (i.e. secondary to supply–demand mismatch or thrombosis) that did not fulfil the universal definition of myocardial infarction was independently associated with 30-day mortality, hazard ratio (HR), 3.30; 95% confidence interval (CI), 2.26–4.81.¹² Based on this, the authors’ definition of MINS included myocardial infarction and ischaemic myocardial injuries that did not fulfil the diagnostic criteria of myocardial infarction, occurring during or within 30 days after non-cardiac surgery. Myocardial injury after non-cardiac surgery did not include an elevated perioperative cardiac troponin measurement that was due to a non-ischaemic aetiology (e.g. sepsis, rapid atrial fibrillation, pulmonary embolism, chronic elevation).

In 2018, the BASEL-PMI investigators, led by Dr Mueller, introduced the diagnosis of perioperative myocardial infarction/injury (PMI).¹³ This diagnosis included all prognostically important acute myocardial injuries after non-cardiac surgery, regardless of whether they were caused by an ischaemic or non-ischaemic aetiology. These authors subsequently demonstrated that various subtypes (e.g. ischaemic, non-ischaemic) of PMI were associated with an increased risk of mortality and major adverse cardiac events at 1-year follow-up.¹⁴

The diagnostic criteria are more inclusive moving from perioperative myocardial infarction, to MINS, to PMI. All three of these perioperative cardiovascular complications are prognostically important, and researchers and clinicians may preferentially choose to focus on one of these outcomes depending on their goals. For example, if the goal is to identify all prognostically relevant perioperative myocardial injuries regardless of the underlying mechanism, then PMI is the preferable outcome. If the goal is to capture all prognostically important perioperative myocardial injuries with the common shared pathophysiology of ischaemia, to ultimately have a greater potential to facilitate the identification of effective therapies, then MINS is the preferable outcome.

N-terminal pro-B-type natriuretic peptide was the first biomarker with substantial data to inform its perioperative prognostic capabilities. The VISION NT-proBNP study measured pre-operative NT-proBNP in 10 402 patients at 16 hospitals in 9 countries.¹⁵ Patients, healthcare providers, and study personnel were blinded to the pre-operative NT-proBNP measurements. The authors did not simply evaluate an arbitrary threshold for NT-proBNP [e.g. 99th percentile of the upper reference limit (URL)]. Rather, they undertook Cox proportional hazard models in which the dependent variable was vascular mortality or MINS, and the independent variables were the Revised Cardiac Risk Index score and iterative thresholds of pre-operative NT-proBNP values to objectively identify optimal NT-proBNP thresholds.

Multi-variable analyses demonstrated that compared with pre-operative NT-proBNP <100 ng/L (the reference group), values of ≥100 to <200, ≥ 200 to <1500, and ≥1500 were associated with adjusted HR of 2.27 (95% CI 1.90–2.70), 3.63 (95% CI 3.13–4.21), and 5.82 (95% CI 4.81–7.05), and corresponding incidences of the primary outcome in 12.3, 20.8, and 37.5%, respectively. Pre-operative NT-proBNP improved risk prediction among patients who did and did not suffer the primary outcome. Moreover, the pre-operative NT-proBNP thresholds substantially differentiated the risk of perioperative myocardial infarction, cardiovascular death, and total mortality.

The VISION growth differentiation factor-15 (GDF-15) study evaluated the prognostic capabilities of pre-operative GDF-15 in 5238 patients at 9 centres in 4 countries.¹⁶ This study used the same methodology and statistical approach as was used in the VISION NT-proBNP study. Multi-variable analyses demonstrated that compared with pre-operative GDF-15 < 1000 ng/L (the reference group), values of ≥1000 to <1500, ≥ 1500 to <3000, and ≥3000 were associated with adjusted HR of 1.93 (95% CI 1.50–2.48), 3.04 (95% CI 2.41–3.84), and 4.8 (95% CI 3.76–6.14) for the primary outcome (i.e. vascular death or MINS), and corresponding incidences of the primary outcome in 12.1, 20.5, and 34.1%, respectively. GDF-15 improved perioperative risk prediction among patients who did and did not have a major perioperative cardiovascular complication, and GDF-15 substantially differentiated the risk of perioperative myocardial infarction and total mortality.

Meister and colleagues report the first large prospective cohort study to evaluate the prognostic capabilities of pre-operative high-sensitivity cardiac troponin T (hs-cTnT). In their derivation cohort of 6944 patients, they demonstrated that pre-operative hs-cTnT improved risk prediction discrimination for PMI and MINS beyond a clinical risk index. Their validation cohort of 722 patients demonstrated a lower area under the receiver operating characteristic curve (AUC) for the PMI model with hs-cTnT (0.64) compared with the AUC in the derivation cohort (0.79). This may be explained by the small validation cohort and the exclusive enrolment of patients undergoing elective major abdominal surgery. In the cohort of 966 patients, they also demonstrated that pre-operative hs-cTnI improved risk prediction discrimination for PMI beyond a clinical risk index.

Considering the four methods of pre-operative cardiovascular risk predication, cardiac biomarkers are the most predictive, convenient, and least expensive (Figure 1). Among the three cardiac biomarkers, pre-operative cardiac troponin costs less and facilitates immediate interpretation of whether an elevated post-operative cardiac troponin value is an acute event. The encouraging data from Meister and colleagues study suggest the possibility that pre-operative cardiac troponin measurement will become the pre-operative biomarker of choice. What limits this position at present is the limited knowledge regarding prognostication across the spectrum of pre-operative cardiac troponin values, and data from one troponin assay cannot be assumed to inform the thresholds of another troponin assay. Until further pre-operative cardiac troponin research occurs, NT-proBNP is more available than GDF-15, it provides robust risk estimates across several established thresholds that substantially differentiate risk, and it can be measured with point-of-care devices in pre-operative clinics.

Figure 1

Preoperative risk prediction for noncardiac surgery.

Open in new tabDownload slide

References

Author notes