PROGRESS IN CARDIOLOGY

Determination of left ventricular ejection fraction using ultrafast computed tomography

Stuart Rich, M.D., Eva V. Chomka, M.D., Richard Stag& M.D., Jeffrey G. Shanes, M.D., George T. Kondos, M.D., and Bruce H. Brundage, M.D. Chicago, 111.

Measurement of left ventricular ejection fraction has been a widely utilized parameter of left ventric- ular function that reflects the hemodynamics and prognosis of a variety of cardiovascular diseases. For this reason there has been much interest in the development of noninvasive modalities to determine left ventricular ejection fraction accurately and pro- vide the clinician with this valuable information.

One noninvasive imaging modality that theoreti- cally should be able to provide this information is x-ray transmission computed tomography (CT). However, until recently the prolonged scan acquisi- tion times required for imaging have rendered CT scanning of limited application to cardiac disease. Methodologic strategies such as ECG gating have enhanced the application of CT to cardiac work,’ but it is the recent development of a scanner able to image at ultrafast scanning speeds (50 msec/slice) that has now made CT a viable noninvasive modali- ty by which to image cardiac structures.* This study reports an initial evaluation of ultrafast CT as a modality to image the left ventricle and, utilizing simple methodology, to determine left ventricular ejection fraction in a group of patients with cardio- vascular disease.

METHODOLOGY

Study group. Patients enrolled in the study con- sisted of 16 individuals (seven men and nine women, ages 45 -t 14 years) with congenital or acquired heart disease who underwent cardiac catheterization at the University of Illinois Hospital (see Table I). Within 48 hours prior or post catheterization, the

From the Section of Cardiology, Department of Medicine, University of Illinois College of Medicine at Chicago.

Rereived for publication Dec. 3, 198.5; accepted Feb. I, 1986.

Reprint requests: Stuart Rich, M.D, Section of Cardiology. University uf Illirroi~. P.O. Rex 6998, Chicago. IL 60680.

patients underwent a CT evaluation of their cardiac function. All patients gave written informed consent for participation in the study.

Ultrafast tine CT imaging system. An Imatron 100 tine CT scanner was used in this study. This device consists of a scanning electron beam x-ray tube, a stationary detector array, and a computer system for data acquisition, reconstruction, and display. In this system an electron beam is accelerated along the axis of the x-ray tube where coils magnetically focus and steer the beam through a 210-degree arc along a curved target. Four target rings are available and are typically scanned serially to obtain a multiple slice examination. Source collimators restrict the x-ray beam to a wide fan, 2 cm thick. An approximately 30-degree fan-shaped sector of the x-ray beam is attenuated by the scanned object and the resulting intensity distribution is measured by a curved detec- tor array. The digitized detector output is stored in a bulk memory and is used for computerized recon- struction. Spatial resolution of the acquired images is less than 2 mm, comparable to conventional CT scanners.

Since the detector array consists of two contigu- ous rings of 216 degrees, a single scan will produce a pair of side-by-side tomographic slices at 50 msec per scan with an 8 msec interscan delay. For each traverse of the electron beam on any one of the four x-ray targets, a pair of 8 mm thick slices are imaged. The collection of eight slices covers a region approx- imately 8 cm thick, large enough to encompass all of the left ventricle of most patients. The radiation delivered is approximately 0.35 rad/scan, locaiized to the scanning regions with little scatter to adjacent areas.

The patient table is designed to provide optimal positioning of the patient in the scan field, and can be moved forward or backward, up or down and vertically to provide axial tilt up to 20 degrees and lateral slew up to as much as 10 degrees.

392

Volume 112

Number 2

Scanning protocol. Patients undergoing a CT study have a peripheral intravenous contrast line started in either the antecubital or external jugular vein, and are positioned within the scanner using the left ventricular impulse as a reference point and are positioned with an axial tilt of 15 degrees and a lateral slew of 10 degrees to better approximate the long axis of the left ventricle. If an initial “scout” image of the patient showed the LV axis to be markedly skewed, the patient was repositioned accordingly. Circulation time from injection site to the left ventricle is determined prior to the study by using either a test dose (10 cc) of RenograGn-76 or indocyanine green dye (1 mg), which is detected with an ear-oximeter. Once the circulation time is determined, the left ventricle is imaged by injecting a 25 ml bolus of Renografln-76 via the peripheral intravenous site with the scanning sequence initi- ated at the time of the estimated peak enhancement of the left ventricle. Scans are repeated every 58 msec for one entire cardiac cycle, obtained over four levels. Depending on the heart size of the patient, this scanning sequence may be repeated one or two more times so that the entire left ventricle is evalu- ated.

Post processing and computation. An off-line con- sole provides reconstruction, display, and archival functions independently from the operator’s con- sole. This console consists of a cathode ray tube (CRT) screen, alpha-numeric and dedicated func- tion keys, and a track ball for image display and analysis. The sequential images for this study can be displayed individually in a frame-by-frame matrix (see Fig. 1) or continuously in a movie mode.

The CT scans obtained were analyzed by two reviewers without knowledge of the left ventriculo- graphic findings. The entire scanning sequence of the left ventricle was reviewed and a single slice was selected that best represented the mid-left ventricu- lar level. This was chosen as a slice at the papillary muscle level, just caudad to the mitral valve. Endo- cardial borders were outlined, using the track ball in diastole (the slice with the largest area) and systole (smallest area). Multiple scans were traced at the same level to assure that the largest and smallest areas were measured. Endocardial edges of the scans were determined visually based on the different CT numbers between the left ventricular cavity and myocardium that could be adjusted for each study (see Fig. 1). Ejection fraction was then determined by calculating the percent difference between the systolic and diastolic areas of the left ventricle (EF = diastolic area - systole area/diastolic area).

Ventriculographic analysis. The ejection fraction

Table I. Patient characteristics

1 32F VSD

6~ .---.-l,i--- --

2 53F CAD J 69 (3 i

3 69F CAD 66 Ii:! 4 46%’ LV thrombus :; I ii:,

5 23F ASD t-d5 :; h 6 24F Tetrology of Fallot 6; ,-:;

7 53M CAD :,;j .ilT

8 27M Variant angina 6:: I;.:

9 46M CAD : I ,.I :,z

70 33M Constrictive * ,i, _, I

pericarditis 11 34M CAD 3; 4:: 12 53F CAD 82 . . -, ! 13 49F CABG 4t :\!I 14 6OM CAD .\ .A “8 15 66M VSD ,. I:: Ii? 16 48F CAD G&l ‘2

MD = atria1 septal defect; CAHC = cor~n;tr\i .rrrery h) pa5s croft: CAD = coronary artery disease: CT = tompr~ted tww~raph\;: \‘s;I) = ww tricular septal defect.

was derived from left ventriculography by means of 45 ml of Renografin-76 injected with a power injec- tor into the left ventricular cavity at the time of cardiac catheterization. Biplane images were ob- tained by positioning the x-ray tube at approximate- ly 30 degrees for the right anterior oblique view and at 60 degrees for the left anterior oblique view, with an axial tilt of 20 degrees. Ejection fraction was computed using the Dodge area-length formula from the ventriculographic images displayed on a 35 mm tine projector (Vanguard) by tracing the systolic and diastolic outlines with a digitized light pen system. None of the patients analyzed had discrete ventricular aneurysms.

Statistical analysis. The mean and standard devia- tions for the ejection fractions obtained by both methodologies were determined, and the Pearson product moment correlation coefficient determined between the two.

OBSERVATIONS

The mean left ventricular ejection fraction for the patients determined by ultrafast CT was 58.1 +- 15% and ranged from 24% to 84%. The mean ejection fraction determined by left ventriculogra- phy on the same patients was 59 +- 8.3%) and ranged between 28% and 77 % (see Table I). There was an excellent correlation between left ventricular ejection fraction by CT and by ventriculography (see Fig. 2), with a correlation coefficient of r = 0.91.

394 Rich et at. August, 1966

American Heart Journal

Fig. 1. One entire cardiac cycle is displayed as sequential images at the mid-left ventricular level. Contrast enhancement of the left ventricular (LV) cavity allows clear distinction of the endocardial border. Using a track ball, the endocardial outlines of the largest (diastolic) and smallest (systolic) areas were manually traced with the areas and their differences were computed by the off-line console.

s 80 li ’ 60

” ‘“I 20 40 60 80 100

L.V. ANGIO EJECTION FRACTION (46)

Fig. 2. The correlation between LV ejection fraction determined by computed tomography (CT) and ventricu- lography are plotted. A strong correlation exists through both extremes of ventricular function.

and a slope of y = 1.1x - 8.5 (p < 0.001). The corre- lation between CT and left ventriculography ap- peared to be strong, both at the high and low ends of the ventricular function range.

COMMENTS

There have been many methodologies developed to evaluate ejection fraction in patients with cardiac disease. Left ventriculography has served as the “gold standard” for comparison,3 but the method is invasive and carries with it the morbidities assoc- iated with such techniques. Echocardiography is a noninvasive and safe technique that has been prov- en to correlate well with ventriculography, but it has problems in edge definition.4 Additionally, the chest characteristics in many patients render the tech- nique non-quantifiable. Multiple gated acquisition studies with radionuclides is another popular meth- od to determine left ventricular ejection fraction, but suffers from poor spatial resolution.5

Ultrafast CT is another modality by which left ventricular ejection fraction can be reliably mea- sured.6 Ultrafast CT offers advantages in superb spatial resolution and three-dimensional image capability that currently exceed radionuclide imag- ing and left ventriculography. In addition, as a tomographic method, localized wall motion abnor- malities and changes in wall thickness can be readily evaluated. We chose a methodology that was simple

volume 112 Number 2

and fast, utilizing a single slice through the left ventricular cavity to demonstrate that in the major- ity of adult patients undergoing cardiac evaluation this simple methodology will provide a reasonably accurate estimate of left ventricular function. Actual left ventricular volumes can be determined by sum- mating the areas of each individual slice obtained through the left ventricle, but this method is more time-consuming.6 Our method required less than 10 minutes to obtain an ejection fraction for each patient utilizing the computerized off-line console that accompanies the ultrafast CT scanner.

The excellent correlation between global ejection fraction by ventriculography and our single-slice method may seem surprising. However, if one postu- lates the left ventricle to be shaped like a bullet, as has been applied by echocardiography, then a short-axis area ejection fraction should correlate well with the global ejection fraction.8 Recently it has been shown that a single-slice, mid-left ventric- ular area ejection fraction correlates well with global ejection fraction by CT in normal adults, which also supports this premise.g The attractiveness of this simple method is that it will provide the clinician both a quick and reliable indication of left ventricu- lar performance that can be applied towards clinical decision making. Should regional wall motion analy- sis be necessary, this can then be determined level- by-level from base to apex.

There are some limitations in the CT technique. It requires an intravenous line and contrast injection, which can be associated with adverse reactions. In addition, although radiation delivered to the patient is relatively small (0.35 rad/slice), it must be consid- ered potentially harmful.

The simple methodology chosen for this study has limitations as well. Marked wall motion abnormali- ties or geometric distortions in the left ventricular cavity may be left out of the analysis when a single slice is selected for processing that would affect the assessment of left ventricular function. This could be overcome, however, by evaluating and summating the area ejection fraction of each tomographic slice.6 Patients with rapid resting heart rates (e.g., greater than 120 per minute) would allow less images to be attained per beat, which could also lead to inaccura- cies.

We conclude that ultrafast CT is a new modality that can provide an accurate correlate of left ventric- ular ejection fraction utilizing a simple methodolo- gy. Although one would not likely utilize CT solely to evaluate a single level area ejection fraction, given the other proven and potential applications of CT for cardiovascular diseases,7 this measurement

[,$’ ejection fractioo 9)’ ltl:riil(lSt (‘7’ 395

should add additional useful information to the complete evaluation of patients having a variety of cardiovascular disorders.

SUMMARY

We evaluated ultrafast CT as a method to mea- sure left ventricular ejection fraction in 16 adults with congenital or acquired heart disease who underwent cardiac catheterization. CT scanning of the left ventricle was performed at 4 to 12 adjacent 1 cm levels (depending on heart size) at 50 msec/scan for one cardiac cycle, with the table positioned with an axial tilt of 10 to 20 degrees and a lateral slew of 5 to 10 degrees to best approximate the long axis of the left ventricle. Image enhancement was achieved by an injection of 25 ml of Renografin-76 via a peripheral vein, with scanning timed to coincide with maximal enhancement of the left ventricular cavity. Ejection fraction was computed by measur- ing the percent change in area of the left ventricle from diastole (largest area) to systole (smallest area) in a single slice at the mid-left ventricular level. Mean ejection fraction for the group was 58.1 & 15.1% (range 24% to 84% ). The ejection fraction from left ventriculography, computed from biplane images using the Dodge (area-length) formula, was 59.6 + 12.3% (range 28% to 77%). There was an excellent correlation between left ventricular ejec- tion fraction by CT and ventriculography (r = 0.91, y = 1.1x - 8.5, p < 0.001). This study demonstrates that ultrafast CT can provide an accurate measure of left ventricular ejection fraction by simple meth- odology.

REFERENCES

1

2.

3.

.i.

3.

6.

7.

8.

<Johnson GA, Godwin JD, Fram EK. (;ated multiplanar cardiac computed tomography. Radiology i&:195, 1982. Boyd D, Gould R, Quinn J, Sparks Ii. Stanley R, Hermanns- feldt W: A proposed dynamic cardiac :{-I) densitometer for easy detection and evaluation of heart disease. IEEE Tram Nucl Sci NS 26:2724, 1979. Kennedy JW, Baxley WA, Figley MM, Dodge HT. Blackman JR: Quantitative angiocardiography. I. The normal left ven- tricle in man. Circulation 34:272, 1966. Schnittger I, Fitzgerald PJ, Daughters !;T. Ingels NB. Kan- trowitz NE, Schwarzkopf A, Mead CW. Popp RL: Limita- tions of comparing left ventricular volumes by two dimen- sional echocardiography, myocardial markers and cineangi- ography. Am J Cardiol 50:512, 1982. Slutsky R, Karliner J, Ricci D, Kaiser K, I’tisterer M. Gordon D, Peterson K, Ashburn W: Left ventricidar volumes by gated equilibrium radionuclide angiography: :\ now met.hod. Circu- lation 60:556, 1979. Carlson E, Lipton MJ, Brundage B. Doherty P, Berninger WH, Redington RW: The diaznostic uotential of cardiac computed tomography. Appl Ridiol 7: Iof;. 19%. Guere TP, Wyatt HL, Meerbaum S, (‘orday F,: A practical two-dimensional echocardiographic model to assess volume in the ischemic left ventricle. Am .I Cardiol 45:471, 1980. Rich S. Sheikh A. Gallastegui .I, Kontfc’> c;T. Mason T, 1,am

Rich et al. August, 1966 American Heart Journal

W: Determination of left ventricular ejection fraction by visual estimation during real-time two-dimensional echocar- diography. AM HEART J 104603, 1982.

9. Rumberger JA, Feiring AJ, Skorton DJ, Collins SM, Rees MR, Noel MP, Eli SR, Marcus ML: Patterns of segmental

ejection fraction in normal adults within ventricular levels. Circulation 72:111-181, 1985.

10. Brundage BH, Rich S, Spigos D: Computed tomography of the heart and great vessels: Present and future. Ann Intern Med 101:801, 1984.

-

The pharmacodynamic and pharmacokinetic differences of the D- and L-isomers of verapamil: Implications in the treatment of paroxysmal supraventricular tachycardia

Timothy J. Hoon, Pharm.D.41 Jerry L. Bauman, Pharm.D.,*§ Keith A. Rodvold, Pharm.D.,*$ Jose Gallestegui, M.D.&i and Robert J. Hariman, M.D.3 Memphis, Term., and Chicago, Ill.

Many of the drugs prescribed today are produced by organic synthesis. This often results in the produc- tion of not one compound, but two compounds that are structurally mirror images, known as stereoisom- ers. Frequently, drugs are administered to interact with biochemically produced enzymes, transport systems, and receptor sites within the body that tend to be stereoselective. Advanced work in the area of structure-activity relationships has shown that slight differences in spatial arrangement may drastically affect receptor binding aflinity and phar- macologic action. Similarly, differences in transport system, plasma protein, or metabolic enzyme inter- action may result in differing pharmacokinetic pro- files. Recent interest in this area has prompted research to look for these isomeric differences in many drugs.l

The molecular structure of verapamil contains a carbon atom with four distinctly different side chains (Fig. 1). This asymmetric carbon allows the

From the Department of Pharmacy Practice,* the Clinical Pharmacokinet- its Laboratory,$ and the Department of Medicine, Section of Cardiology& University of Illinois at Chicago; and the Department of Clinical Pharma- cy,lJ University of Tennessee. Received for publication Jan. 17, 1986; accepted Feb. 20, 1986. Reprint requests: Jerry L. Bauman, Pharm.D., Dept. of Pharmacy Prac- tice. University of Illinois College of Pharmacy, 833 S. Wood St., Chicago, IL 60612.

396

production of two nonsuperimposable mirror images with the same molecular structure (Fig. 2). These stereoisomers demonstrate unique physical proper- ties. In solution, they differ in their ability to rotate polarized light. The dextro (D-) isomer rotates polarized light to the right as it passes through solution, while the levo (L-) isomer rotates polarized light to the left. The differences between these isomers surpass these physical characteristics alone.

As verapamil is marketed as a racemic mixture of the D- and L-isomers, stereospecific characteristics may be of clinical significance in select therapeutic situations. One of the indications for employing verapamil is in the management of paroxysmal supraventricular tachycardia (PSVT). Intravenous dosing in the acute management of PSVT is very effective, while chronic oral dosing has demon- strated variable success in preventing episodes of PSVT. This discrepancy in therapeutic response may be partially explained by the stereospecific differences between the D- and L-isomers. This review of the isomeric studies of verapamil will discuss the reported stereospecific pharmacodynam- ic and pharmacokinetic characteristics of verapamil. In addition, the suspected clinical implications of verapamil stereospecificity in the treatment of PSVT will be discussed.