a b s t r a c t

Objective: The objective of this study us to stratify by gender a new cardiac electrical biomarker (CEB) diagnostic accuracy for detection of acute myocardial Ischemic injury (AMII).

Methods: This is a noninferiority retrospective, case-control, blinded study of 310 archived measured electrocar- diograms (ECGs) acquired from 218 men and 92 women. The CEB is constructed from the derived ECG (dECG) synthesized from 3 leads. Electrocardiograms were included if acquired less than or equal to 1 day from patient presentation. Electrocardiograms were interpreted by 2 Blinded physicians and adjudicated by consensus. Standard ST analyses and computerized ECG interpretation./a>s were active controls. Electrocardiograms were excluded for noise and baseline wander, age younger than 18 years, and ectopic beats in the 10-second ECG acquisition. Diagnostic accuracy measures of sensitivity, specificity, positive and negative predictive values, and likelihood ratios were stratified by gender. Measured vs derived ECG correlations were quantitatively compared using Pearson correlation and qualitatively by percent agreement methodology.

Results: The CEB sensitivities for AMII detection in men and women were 93.9% and 90.5%, respectively, and CEB specificities were 90.7% and 95.2%, respectively, and were superior to active controls. Derived and measured ECGs showed high correlation for both men and women with r = 0.857 and r = 0.893, respectively. Reference standard intra-agreement analysis for measured ECGs and dECGs with AMII was 99.4%.

Conclusions: The CEB demonstrates high diagnostic accuracy for detection of AMII in men and women. The ECG can be derived with accuracy from 3 leads. This technology is an efficient real-time method of identifying patients with AMII who are being monitored in Acute care settings.

(C) 2014

1. Introduction

The concepts of a new and novel cardiac electrical biomarker (CEB) and the derived 15-lead electrocardiogram (ECG) have been recently reported by Schreck and Fishberg [1,2]. Briefly, cardiac electrical activity is reported to be highly dipolar [3,4], and as such, only 3 measured orthogonal leads should be needed to actually derive this composite 15-lead ECG from just 5 body surface electrodes that are connected to a cardiac rhythm monitoring device. This will allow continuous cardiac

? Prior presentations: Concepts in this manuscript have been presented at a prior meet- ing: Schreck DM, Fishberg RD. Detection of acute myocardial ischemic injury by gender using a novel cardiac electrical biomarker. Presented at the American College of Emergen-

cy Physicians Research Forum. Chicago, IL, October 28, 2014.

?? Funding sources/disclosures: This study was unfunded.

? Dr David M. Schreck owns a significant nonmajority interest in VectraCor that consti-

tutes greater than 5% of the entity. VectraCor is the medical device company that manufac- tures the device used in this study.

* Corresponding author. Department of Emergency Medicine, Summit medical group, 1 Diamond Hill Road, Berkeley Heights, NJ 07922. Tel.: +1 908 273 4300.

E-mail address: [email protected] (D.M. Schreck).

rhythm monitoring with the added simultaneous advantage of acquiring the derived 12-lead ECG (dECG) and scalar 3-lead derived vectorcardiogram, a composite 15-lead ECG, instantaneously and in real-time using 1 cardiac rhythm monitoring device. The objective of this study is to identify the CEB diagnostic accuracy, stratified by gender, compared to active controls (ACs).

1. Materials and methods

This is a noninferiority, retrospective, blinded, case-control, paired comparator [5] study of ECGs from 218 men and 92 women. These 310 measured ECGs (mECGs) of various morphologies were obtained from 2 databases including an archived National Institutes of Health-funded Physiobank PTBDB database [6] and a database from Muhlenberg Regional Medical Center (Plainfield, NJ). The Muhlenberg Regional Medical Center database includes consecutive patients who were admitted to the emergency department (ED) with chest pain. Patients included men and women, age 18 years or older. The stan- dard mECGs were acquired using a Marquette MAC-15 machine (GE Healthcare, Waukesha, WI). The study ECGs represent a gender stra- tification subanalysis of a recent prior study by Schreck and Fishberg [1].

http://dx.doi.org/10.1016/j.ajem.2014.12.029

0735-6757/(C) 2014

Fig. 1. ST criteria for ECG interpretation of AMII.

Electrocardiograms were included if acquired less than or equal to 1 day from patient presentation. Electrocardiograms were excluded for exces- sive noise and baseline wander, age younger than 18 years, and paced or ectopic beats in the 10-second ECG period from which a “median beat” is determined.

The dECGs were constructed from 3 measured leads I, II, and V₂, which were converted to a 3-lead orthogonal basis set of {I, aVF, V₂}

[7] using Einthoven triangular geometric relationships. The 15-lead dECGs were synthesized from this orthogonal lead set. The CEB is con- structed from the median beat from each dECG using the VectraplexECG

Assessed for Eligibility: 643

PTBDB: 549 MRMC: 94

Primary Exclusions: 321
Missing data	Wandering baseline	Paced rhythm	Training set	Noise
Missing leads	Duplicate ECGs same day	Lead placement error	PVCs	Age < 18

*Note: Differences between EP , Adjudication and Cardiology sample size is due to diagnosis and associated exclusion criteria

*EP Exclude: ECG > 1 day : 52

TEST SET: 367

MRMC: 46 PTBDB: 321

*Cardiology Exclude: ECG > 1 day: 35

Adjudication: 310

AMI: 67

Non-AMI: 243

*Adjudication Exclude: ECG > 1d: 57

EP < 1 day: 315

AMI: 67

Non-AMI: 248

Cardiology < 1 day: 332

AMI: 57

Non-AMI: 275

ECG = Electrocardiogram

UPTM = Universal patient transformation matrix PVC = Premature ventricular contraction

EP = Emergency physician

AMI = Acute myocardial infarction

Fig. 2. Flow diagram for case enrollment. Abbreviations: UPTM, universal patient transformation matrix; PVC, premature ventricular contraction; AMI, acute myocardial infarction.

Table 1

Characteristics of study population

n 310

Male % 70.3%

Age (all) 54.8 +- 14.7

Median age (male) 54.0 +- 13.5

Median age (female) 58.6 +- 16.2

Non-AMII% 78.4%

Median age non-AMII 52.4 +- 14.4

Male % non-AMII 71.6%

Median age male non-AMII 51.7 +- 13.9

Median age female non-AMII 54.1 +- 15.5

AMII% 21.6%

Median age AMII 63.8 +- 12.0

Male % AMII 65.7%

Median age male AMII 59.7 +- 10.8

Median age female AMII 71.6 +- 10.2

AMII% without Q wave 62.7%

Inferior wall AMII% 49.3%

Lateral wall AMII% 22.4%

Anterior wall AMII% 29.9%

Septal wall AMII% 26.9%

posterior wall AMII% 25.4%

System (VectraCor, Inc, Totowa, NJ) and is a measure of the dipolar energy in the cardiac electrical field.

The CEB was compared to each AC that included the 12-lead ECG computer interpretation and the ST analysis voltage parameters ST0 (J point) and ST area under curve (STSUM). The STSUM points included the lead voltages at the J point and at 20, 60, and 80 milliseconds after the J point (ST0, ST20, ST60, and ST80, respectively).

There were 2 reference standards including 1 board-certified emergency physician and 1 board-certified cardiologist who inde- pendently interpreted the ECGs and were blinded to each other’s inter- pretations. The 2 reference standards also adjudicated the results at study completion. The rationale for this process was to simulate a “real-world” situation where the EP identifies an ECG as consistent with acute myocardial ischemic injury (AMII) and then discusses the ECG findings with the cardiologist for corroboration such that immedi- ate care (ie, cardiac Catheterization laboratory activation or fibrinolytic administration) could be implemented based on facility protocols. The criteria reported by Thygesen et al [8] were used by the reference stan- dards to interpret ECGs for the presence of ST changes consistent with ST-segment elevation myocardial infarction and non-STEMI and are shown in Fig. 1.

Measured vs derived ECGs correlations were quantitatively compared using Pearson correlation coefficient (r) and qualitatively by reference standard percent agreement methodology [9].

The CEB diagnostic accuracy parameters including sensitivity, speci-

ficity, negative and positive predictive values (NPV and PPV), likelihood

ratios (LR+ and LR-), and odd ratios (ORs) were identified and com- pared to the ACs in a noninferiority design [5] and stratified by gender using a 1-sided (? b .025, 1 - ? = 0.90) interval with 95% confidence statistical analysis.

The study was conducted in a community academic teaching hospital. The institutional review board approved this study methodology and exempted the need for informed consent.

1. Results

The case Selection process is shown in Fig. 2 and followed accepted guidelines for acquisition and reporting [10]. Table 1 shows the charac- teristics of the patient population in the study. Table 2 shows the CEB sensitivities, specificities, NPV, PPV, LR+, LR-, and OR for AMII detection in men and women, also stratified by reference standard and adjudication. The detailed CEB sensitivity analyses, also known as the true-positive rate (TPR), and the 1 - specificity analyses, also known at the false-positive rate, are shown in Figs. 3 to 8. These figures show the analyses of the CEB/AC ratios stratified by gender, AC, and ref- erence standard interpretation with adjudication. The CEB is considered a positive test for AMII if greater than 94 and negative for AMII if less than 66. The CEB is considered indeterminate from 66 to 94. As such, the sensitivity is also stratified by “actual” and “worst” case scenarios. The worst-case scenario is such that any CEB in the indeterminate region is considered a false-positive or false-negative result.

Fig. 3 shows the sensitivity (TPR) analyses stratified by gender and AC for the EP reference standard. The CEB was noted to be noninferior by hypothesis testing, but superiority was statistically demonstrated using the actual data compared to the ACs. Fig. 4 shows the false-positive rate (FPR), also known as 1 - specificity. The CEB was noted to be noninferior by hypothesis testing, but superiority was statistically demonstrated using the actual and worst case data com- pared to the ACs.

Figs. 5 to 6 show the same TPR and FPR analyses for the cardiology reference standard. The CEB is shown to be noninferior by hypothesis testing for the TPR analyses. The CEB is shown to superior to ACs in the FPR analyses.

Figs. 7 to 8 show the same TPR and FPR analyses as adjudicated by the reference standards to mimic the real-world situation where the EP and cardiologist both collaborate on the ECG interpretation. The CEB is again shown to be noninferior by hypothesis testing in both the TPR and FPR analyses, but superiority was demonstrated in the actual data TPR analysis. Superiority was also demonstrated for both actual and worst case data FPR analysis.

All CEB diagnostic accuracy measures were significant (P b .025). De- rived vs measured 12-lead ECGs showed high correlation for both men and women with r = 0.857 and r = 0.893, respectively. Because the CEB is constructed from the dECG, it is important to demonstrate the corre- lation between the mECG and dECG to support the diagnostic accuracy

Table 2

Cardiac electrical biomarker diagnostic accuracy parameters

			AMII		CEB		CEB		CEB		CEB		CEB		CEB		CEB		CEB
		n	Prevalence		Sensitivity		Specificity		NPV		PPV		LR(+)		LR(-)		OR		Utility
EP	Men	222	18.5%		92.7%		91.1%		98.1%		71.7%		10.38		0.08		129.2		92.2%
	Women	92	36.6%		95.5%		91.9%		98.3%		80.8%		11.84		0.05		239.4		91.3%
	Total	314	20.1%		93.7%		91.3%		98.1%		74.7%		10.77		0.07		154.9		93.3%
Cardiology	Men	228	15.4%		82.9%		82.8%		96.1%		48.3%		4.83		0.21		23.2		94.3%
	Women	98	32.1%		94.4%		80.0%		98.2%		54.8%		4.72		0.07		68.0		89.8%
	Total	326	16.3%		86.8%		82.0%		96.7%		50.5%		4.82		0.16		29.9		92.9%
Consensus	Men	218	20.9%		93.9%		90.7%		97.5%		79.6%		10.09		0.07		149.2		94.5%
	Women	92	25.0%		90.5%		95.2%		98.3%		76.9%		19.00		0.10		190.0		91.3%
	Total	310	20.6%		92.2%		92.9%		97.7%		78.7%		13.02		0.08		154.9		93.5%

Fig. 3. Emergency physician ECG interpretation: CEB:AC actual and worst case sensitivities for men and women. Abbreviations: TPR, true-positive rate (sensitivity); ?, effect margin; ST0, J- point ST voltage; STSUM, area under curve of ST-segment voltages at J point and 20, 60, and 80 milliseconds after J point.

Fig. 4. Emergency physician ECG interpretation: CEB:AC actual and worst case false positive rates for men and women. Abbreviation: FPR, false-positive rate (1 - specificity).

Fig. 5. Cardiologist ECG interpretation: CEB:AC actual and worst case sensitivities for men and women.

Fig. 6. Cardiologist ECG interpretation: CEB:AC actual and worst case false positive rates for men and women.

Fig. 7. Adjudicated ECG interpretation: CEB:AC actual and worst case sensitivities for men and women.

of the dECG for AMII cases. Table 3 shows the quantitative Pearson r correlations of the mECGs vs the corresponding dECGs by each derived lead.

Table 4 shows reference standard percent inter-agreement and intra-agreement analysis for mECGs and dECGs with AMII.

1. Discussion

The concept of the derived ECG is not new and has been reported by several investigators [11-17]. Now that recent guidelines for STEMI [18] and non-STEMI [19] care have called for frequent serial ECGs to be ac- quired, the utilization of the derived ECG technology may now become more advantageous and clinically useful. The use of the derived ECG may also be cost effective by using less electrodes and minimizing tech- nician and nursing time.

It was very important to note that this study demonstrated high correlations between the mECG and dECG. This is a very important finding because (1) the CEB is constructed from the derived ECG and (2) the dECG must be similar to the mECG to bring clinical value and validation. The high inter-agreement and intra- agreement analysis also lends support for the use of the dECG in the clinical setting.

The concept of the CEB allows a more efficient method for observing patients being evaluated for chest pain equivalents in any acute care setting, particularly the ED. The CEB is obtained continuously and displayed on the cardiac monitor in real time allowing immediate identification of a potential AMII in the proper clin- ical setting.

It was interesting to note that the CEB diagnostic accuracy parame- ters were very favorable in both men and women. Specifically, the

high diagnostic accuracy results found in women were unexpected given the reported difficulties of identifying coronary disease in the fe- male population [20-22].

There are several limitations of this study that should be understood. This is a retrospective study for which Prospective trials are needed and are already underway. In addition, it should be noted that the prevalence of disease in the population studied was approximately 20% and may be considered by some to be higher than expected. However, it was interesting to note that the CEB likelihood ratios and OR, which are not dependent on prevalence [23], were very favorable.

1. Conclusions

The CEB demonstrates high diagnostic accuracy for detection of AMII for men and women when compared to standard ST-segment analysis and ECG computer interpretation ACs. The 12-lead ECG can be derived with accuracy from just 3 leads directly from the cardiac monitor. The measured and derived 12-lead ECGs show high qualitative and quanti- tative correlation. This technology will allow an immediate, cost- effective, and efficient means of identifying patients with AMII who are being monitored in acute care settings.

Acknowledgments

The authors acknowledge Patricia Wright and the Atlantic Health System Overlook Medical Center library staff for their assistance in the extensive literature searching required for the preparation of this manuscript.

Fig. 8. Adjudicated ECG interpretation: CEB:AC actual and worst case false positive rates for men and women.

Table 3

Pearson correlation (r) for each of the measured vs derived leads and the 12-lead ECG

Pearson r

Table 4

Percent agreement results for reference standards

EP dECG vs EP mECG OA% (95% CI) 100 (99.0-99.8)

	Men	Women		PPA% (95% CI) PNA% (95% CI)	100 (94.6-100) 100 (98.5-100)
n	218	92	Cardiology mECG vs cardiology dECG	OA% (95% CI)	99.4 (98.1-99.5)
III	1.00	1.00		PPA% (95% CI)	98.2 (90.4-99.7)
aVR	1.00	1.00		PNA% (95% CI)	99.6 (98.0-99.9)
aVL	1.00	1.00	EP mECG vs cardiology mECG	OA% (95% CI)	92.6 (90.3-94.1)
aVF	1.00	1.00		PPA% (95% CI)	93.1 (85.8-96.8)
V1	0.88	0.90		PNA% (95% CI)	92.5 (88.8,95.0)
V3	0.90	0.89	EP dECG vs cardiology dECG	OA% (95% CI)	92.9 (90.6-94.4)
V4	0.71	0.77		PPA% (95% CI)	95.2 (88.4-98.1)
V5	0.71	0.78		PNA% (95% CI)	92.2 (88.5-94.8)
V6	0.73	0.84	EP dECG vs cardiology mECG	OA% (95% CI)	91.9 (89.2-93.6)
12-lead	0.86	0.89		PPA% (95% CI)	89.8 (79.5-95.3)
				PNA% (95% CI)	92.4 (88.5-95.1)
			EP mECG vs cardiology dECG	OA% (95% CI)	91.3 (88.5-93.0)
				PPA% (95% CI)	72.2 (61.0-81.2)
References				PNA% (95% CI)	97.1 (94.1-98.8)

1. Schreck DM, Fishberg RD. Diagnostic accuracy of a new cardiac electrical biomarker for detection of electrocardiogram changes suggestive of acute myocardial ischemic injury. Ann Noninvasive Electrocardiol 2014;19(2):129-44.
2. Schreck DM, Fishberg RD. Derivation of the 12-lead electrocardiogram and 3-lead vectorcardiogram. Am J Emerg Med 2013:1183-90.
3. Geselowitz DB. Dipole theory in electrocardiography. Am J Cardiol 1964;14:301-6.
4. Gabor D, Nelson CV. Determination of the resultant dipole of the heart from measurements on the body surface. J Appl Phys 1954;25:413-6.
5. Alonzo TA, Pepe MS, Moskowitz CS. Sample size calculations for comparative studies of medical tests for detecting presence of disease. Stat Med 2002;21:835-52.
6. Goldberger AL, Amaral LAN, Glass L, Hausdorff JM, Ivanov PCh, Mark RG, et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 2000;101:e215-20.

Abbreviations: OA, overall agreement; PPA, percent positive agreement; PNA, percent neg- ative agreement; CI, confidence interval.

1. Goldman MJ. Introduction to electrocardiography. In: Goldman MJ, editor. Principles of clinical electrocardiography. 8 ed. Los Altos: Lange Medical Publications; 1973. p. 13-4.
2. Thygesen K, Alpert JS, Jaffe AS, Simoons ML, Chaitman BR, White HD. Third universal definition of myocardial infarction. Circulation 2012;126:2020-35.
3. Clinical and Laboratory Standards Institute document EP12-A2: user protocol for evaluation of qualitative test performance: approved guideline. 2 ed. Pennsylvania: Wayne; 2008.
4. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initia- tive. Ann Intern Med 2003;138:40-4.
5. Dower GE, Yakush A, Nazzal SB, Jutzy RV, Ruiz CE. Deriving the 12-lead electrocar- diogram from four (EASI) electrodes. J Electrocardiol 1988;21:182.
6. Nelwan SP, Kors JA, Meij SH, van Bemmel JH, Simoons ML. Reconstruction of the 12-lead electrocardiogram from reduced lead sets. J Electrocardiol 2004;37:11-8.
7. Drew BJ, Dempsey ED, Joo TH, Sommargren CE, Glancy JP, Benedict K, et al. Pre- hospital synthesized 12-lead ECG ischemia monitoring with trans-telephonic transmission in acute coronary syndromes: pilot study results of the ST SMART trial. J Electrocardiol 2004(37 Suppl.):214-21.
8. Man SC, Maan AC, Kim E, Draisma HHM, Schalij MJ, van der Wall EE, et al. Reconstruc- tion of standard 12-lead electrocardiograms from 12-lead electrocardiograms recorded with the Mason-Likar electrode configuration. J Electrocardiol 2008;41:211-9.
9. Horacek BM, Warren JW, Wang JJ. On designing and testing transformations and derivation of standard 12-lead/18-lead electrocardiograms and vectorcardiograms from reduced sets of predictor leads. J Electrocardiol 2008;41:220-9.
10. Nelwan SP, Carey MG. Reduced and alternative lead sets: clinical and technical challenges. J Electrocardiol 2008;41:451-3.
11. Scherer JA, Jenkins JM, Nicklas JM. Synthesis of the 12-lead electrocardiogram from a 3 lead subset using patient-specific transformation vectors: an algorithmic approach to computerized signal synthesis. J Electrocardiol 1989;22:128.
12. Antman EM, Anbe DT, Armstrong PW, Bates ER, Green LA, Hand M, et al. ACC/AHA Guidelines for the management of patients with ST-elevation myo- cardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines. Circulation 2004;110: 588-636.
13. Amsterdam EA, Wenger NK, Brindis RG, Casey Jr DE, Ganiats TG, Holmes Jr DR, et al. 2014 AHA/ACC Guideline for the management of patients with non-ST-elevation acute coronary syndromes: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines. Circulation 2014;130(25):2354-94.
14. Redberg RF, Shaw LJ. Diagnosis of coronary artery disease in women. Prog Cardiovasc Dis 2003;46(3):239-58.
15. Mosca L, Manson JE, Sutherland SE, Langer RD, Manolio T, Barret-Connor E. Cardio- vascular disease in women. Circulation 1997;96:2468-82.
16. Litwin SE, Priester TC. Diagnosing coronary artery disease in women: an update on cardiac imaging modalities. Obstet Gynecol 2010;115(1):156-9.
17. Jekel JF, Elmore JG, Katz DL. Understanding and reducing errors in clinical medicine. In: Jekel JF, editor. Epidemiology biostatistics and preventative medicine. 1st ed. Philadelphia: W. B. Saunders Company; 1996. p. 86-97.