European Journal of

Nuclear Medicine

Original article

Accuracy and repeatability of left ventricular systolic and diastolic function measurements using an ambulatory radionuclide monitor Leonardo Pace 1, Alberto Cuocolo 1, Antonio Nappi 1, Emanuele Nicolai 1, Bruno Trimarco 2, and Marco Salvatore 1'3 1 Cattedra di Medicina Nucleare, Istituto di Scienze Radiologiche, 2 Istituto di Clinica Medica, II Facolt& di Medicina e Chirurgia, Universit& degli Studi "FEDERICO I1", and 3 Istituto Nazionale Tumori "Fond. Sen. G. Pascale", Naples, Italy Received 28 N o v e m b e r 1991 and in revised form 22 February 1992

Abstract. The accuracy and repeatability of a new ambulatory radionuclide detector (VEST) for left ventricular systolic (ejection fraction) and diastolic (peak filling rate) measurements were assessed. Seventeen patients underwent equilibrium radionuclide angiography immediately before and immediately after a VEST study. The accuracy was evaluated at the beginning and at the end of the VEST studies. Limits of agreement for the ejection fraction were - 1 % : 2% at the beginning of the VEST study and - 4 % : 4% at the end. Limits of agreement for the peak filling rate were - 0 . 6 : 0 . 6 at the beginning of the VEST study and -0.7:0.5 at the end. For both measurements the limits of agreement were well within the clinical range. Repeatability was evaluated in a second group of 11 patients who underwent VEST studies in 2 separate days. The coefficient of repeatability (twice the standard deviation of the differences between the 2 studies) was 13 for the ejection fraction and 0.4 for the peak filling rate. Thus, the VEST is an accurate and repeatable method to measure both the ejection fraction and peak filling rate.

1988; de Yang et al. 1991). In all these studies the VEST was used to monitor LV systolic function, namely the ejection fraction (EF). It is well-known that abnormalities of LV diastolic function are present in coronary artery disease (CAD; Bonow etal. 1981a; Pace etal. 1989) and in other cardiac diseases (Bonow et al. 1981 b; Cuocolo et al. 1990; Soufer et al. 1985). The VEST could be used to monitor LV diastolic function by measuring the peak filling rate (PFR). While the accuracy and repeatability of LV EF measurements by the VEST have been studied (Wilson et al. 1983; de Yang et al. 1991), to our knowledge no study has been performed to assess the accuracy and repeatability of PRF measurement by VEST. The aim of this study was, therefore, to determine the accuracy and repeatability of the VEST in evaluating PFR. Moreover, the accuracy of VEST measurements of LV systolic and diastolic function was also evaluated at the end of VEST acquisition.

Key words: VEST - Left ventricular systolic function -

Methods

Left ventricular diastolic function

Eur J Nucl Med (1992) 19:800-806

Introduction An ambulatory device (VEST, Capintec, Inc., USA) to monitor changes in left ventricular (LV) function has recently been described (Wilson et al. 1983). In the past few years the VEST has been used for a variety of clinical applications (Bairey et al. 1990; Breisblatt et al. 1988; Kayden et al. 1990; Kiess et al. 1988; Tamaki et al. 1987, Offprint requests to: L. Pace, Sanseverino 5/A, 1-80128 Napoli, Italy

Patient population. Two groups of patients were studied. Group 1 was used to evaluate the accuracy of VEST measurements in comparison with equilibrium radionuclide angiography (ERNA). This group consisted of 17 subjects (all men, mean age 53 i-11 years). All patients were affected by coronary artery disease (8 patients with a previous myocardial infarction). Group 2 was used to assess VEST repeatability. Eleven patients (8 men and 3 women, mean age 54_+10 years) with coronary artery disease (7 patients with a previous myocardial infarction) composed this group. Study protocol. The accuracy of VEST measurements was evaluated in patients of group I at the beginning and at the end of the VEST acquisition. E R N A was performed immediately before and after the VEST study. In order to assess VEST repeatability, the patients of group 2 underwent VEST studies on 2 separate days (mean interval 7 410 days, range 3-11 days).

© Springer-Verlag 1992

801 adequate. In patients of group 1, the first and the last 10 min of the data were then considered for the analysis in the accuracy study (only the first 10 rain for the repeatability study in patients of group 2). The radionuclide and ECG data were summed for 30-s intervals, and the heart rate (HR) was displayed graphically to individuate a period of stable heart rate comparable to that recorded during the corresponding ERNA. EF and P F R were then computed only in this limited part of the VEST study in the first and in the last 10 rain in the accuracy study (only in the first i0 min in the repeatability study). EF was computed as the stroke counts divided by the background corrected end-diastolic counts. Background was determined by matching the initial resting VEST EF value to that obtained by the gamma-camera. P F R was obtained from the Fourier curve and computed as the inflection point after end-systole where the second derivative changes sign from positive to negative.

Data acquisition Radionuclide angiography. In vivo labelling of red blood cells was performed with 555 MBq of technetium-99m (15 mCi). E R N A was performed in the 45 ° left anterior projection with a 15 ° craniocaudal tilt with the patient in the supine position. A small field-of-view gamma-camera (Starcam 300A/M, General Electric, USA) equipped with a low-energy, all-purpose collimator was used. Data were recorded at a frame rate of 30 frames/cardiac cycle on a dedicated computer system (General Electric, USA). At least 200000 counts/frame were acquired. E R N A was performed under control conditions immediately before (ERNA-1) and immediately after (ERNA-2) the VEST study.

Vest. The VEST consists of two radionuclide detectors : one (sodium iodide crystal and parallel-hole collimator) was used to monitor the left ventricle, and the other (cadmium telluride and a fiat-field collimator) was used to monitor activity in the lung. Other components of the VEST inlcude an electrocardiographic (ECG) recorder (2 leads), a gating device, a cassette recorder and a microcomputer. A vest like garment was used to hold in place the 2 detectors. The optimal placement of the VEST was determined by using the gamma-camera, as previously described (Kayden et al. 1990). The patients wore the VEST for at least 3 h. During this time the subjects were allowed to move freely in the department, except for the first (VEST-l) and the last (VEST-2) 10 min when they lay under control conditions in the supine position for the accuracy group (group 1). Patients in the repeatability group (group 2) were studied under control conditions in the supine position during the first 10 rain of study on day 1 (study-I) and on day 2 (study-2).

Statistical analysis. Values are expressed as mean_+ 1 standard deviation. The paired t-test was used to assess differences in the mean values o f H R , EF and P F R measured by E R N A and VEST studies. The mean difference between the measurements obtained with the 2 procedures was computed (EF by E R N A - I and VEST-l, EF by ERNA-2 and VEST-2, P F R by ERNA-1 and VEST-1 and PFR by ERNA-2 and VEST-2). Correlation analysis was used as a first approach to test the accuracy of VEST in comparison with E R N A at the beginning and at the end of the VEST studies. However, Bland and Altman (1986) stressed that correlation analysis (1) measures the strength of a relation between 2 variables, not the agreement between them; (2) is not affected by a change in the scale of measurements, which in turn certainly affects agreement; (3) could give quite a high correlation coefficient for data which are in poor agreement. They proposed a different approach based upon the analysis of a plot of the difference between the methods against their mean. This plot also allows us to investigate any possible relationship between the measurement error and the "true value" (in this case the mean of the two methods is the best estimate of the true value). The differences would follow a normal distribution, and thus 95% of them will lie between_+ 1.96 standard deviations. If differences within-+ 1.96 standard deviations are not clinically relevant, the 2 methods could be used interchangeably, and it is possible to refer to these as the "limits of agreement". Since the limits of agreement are only estimates of the values which apply to the whole population, confidence intervals should be then computed to see how precise the estimates of the limits of agreement are. Therefore, our data were analyzed using the method proposed by Bland and Altman (1986). Repeatability was examined by plotting the differences in parameter values between the two VEST analyses against their mean. A repeatability coefficient (British Standard Institution 1979) was then computed as twice the standard deviation of the differences.

Data AnaIys& Radionuclide angiography. E R N A studies were analyzed using a standard commercial software (General Electric, USA). Briefly, LV regions of interest were automatically drawn for each frame. A background region of interest was also computer-delineated on the end-systolic frame. After background correction, a LV timeactivity curve was generated. EF was computed on the raw timeactivity curve, while P F R was calculated after a Fourier expansion with 4 harmonics. P F R was computed as the maximum positive peak after end-systole on the first derivative of the LV time-activity curve and normalized by the end-diastolic counts.

VEST. At the end of the VEST study, data were reviewed for technical adequacy. Briefly, the average count rate (decay corrected) of the entire study was displayed: if this curve had < 10% deviation from a straight line, the VEST study was considered

Table 1. Accuracy of VEST measurements (VEST-I versus E R N A - I )

Variable

VEST-1

ERNA-1

P

d

LA

95% CI LLA

95% CI U L A

HR EF PFR

79 +20 43 -+15 1.8_+ 0.6

79 +20 43 -+14 1.8_+ 0.7

NS NS NS

0.7-+5 0 -+l 0 _+0.3

--9:11 --1:2 -0.6:0.6

--14:--5 --2:0 -0.9:-0.3

7:16 1:3 0.3:0.9

HR, Heart rate (bpm); EF, ejection fraction (%); PFR, peak filling rate (end-diastolic counts/s); d, mean difference; LA, limits of agreement; 95% CI LLA, 95% confidence interval of the lower limit of agreement; 95% CI ULA, 95% confidence interval of the upper limit of agreement

802 120

MEAN

.....

+/-

2 SD

20 1O0, P'4 . . . . . . . . . . . .

10

"~

"O

. . . . . . . . . . . . .

80'

• /

60'

•

/ 0~ •

•

p < 0.05

40

u

40

•

0

y = 4,2 + 0,96x r= 0.97 n= 17

60

I

u

80

100

-10 120

HR-ERNA 1 -20

I

40

80

I

60 AVERAGE

I

80 HR by VEST

100 and BP

120

60' 6 40

4

20

2 /

0

0

p < 0.05

m 40 EF-ERNA 1

20

6u0

80

ol

-2 -4 -6

I

3

I

20

10

I

30

AVERAGE

40 EF by VEST

I

I

50

60

and

70

BP

2

11,

i

L :E

•

•o . . . . .

n= 17 p < 0.05 0

I

1

I

2 PFR-ERNA 1

I

3

a

•

Fig. 1. Comparison of equilibrium radionuclide angiography (ERNA-1) and VEST (VEST-l) measurements of heart rate (top panel), ejection fraction (middlepanel) and peak filling rate (bottom

•

go

panel)

Results Accuracy VEST-1 versus ERNA-1. Table 1 shows the results obtained with VEST and E R N A in patients of group 1. H R , EF and P F R were not statistically different between the 2 procedures. The limits of agreement were well within the clinical range. The 95% confidence intervals o f b o t h lower and upper limits of agreement are narrow, showing an acceptable degree of agreement between VEST-1 and E R N A - I for the 3 parameters considered in this study. Figure 1 shows the plots of H R , EF and P F R with the 2 methods. To gain m o r e information on the agreement, we plotted the differences in the parameter measurements against their m e a n (Fig. 2). No

0

I

|

I

1

2

3

AVERAGE

PFR

by VEST

and

4 BP

Fig. 2. Plot of the differences between the two methods (ERNA-1 and VEST-t) against their mean. Results obtained for heart rate (top panel), ejection fraction (middle panel) and peak filling rate (bottom panel) are shown. There are no relations between the differences and the mean for the 3 parameters

relationship between the difference and the mean was found, suggesting the lack of any relationship between the measurement error and the estimate of the true value.

VEST-2 versus ERNA-2. Table 2 shows the results obtained with VEST and E R N A in patients of group 1. H R , EF and P F R were not statistically different between the 2 procedures. The limits of agreement were well with-

803 Table 2. Accuracy of VEST measurements (VEST-2 versus ERNA-2) Variable

VEST-1

ERNA-1

P

d

LA

95% CI LLA

95% CI U L A

HR EF PFR

76 +_17 44 +_17 1.7-t- 0.6

78 +_18 44 +16 1.8_+ 0.7

NS NS NS

-2_+5 0_+2 -0.1+_0.3

-12:8 -4:4 -0.7:0.5

-17:-8 -6:-2 -1:0.4

4:13 2:6 0.2:0.8

120

.....

MEAN

+/- 2 SD

15 1 O0

~o

=, 80

10 m

. . . . . . .

i

....

. . - . m

. . . . . .

5 y = 9,95

60

+

0,85x

0

r= 9.97 n= 17 p < 0.05

4O

i

i

•

-5

i

40

60

8o HR-ERNA 2

100

120 ¢1

-10 -15

80

40

50 '

70 '

6'0

8 '0

9 '0

1

I O0 '

0

AVERAGE H R b y VEST a n d BP 60 @I

6

40

4 20

f |

•

20

2

n= 17 p/0.05

| 40

•

60

•0 80

EF-ERNN 2

@

•

0

i

a

-2 @

4

-4 -6

I

0

~ ~

•

I

50

I

60

70

y = 0,3 + 0,79x

•

II=17

r

0

|

40

"

I

0 i

!

30

AVERAGE E F b y VEST a n d BP

21 4

20

p < 0.05

.

, I

.

, 2

.

, 3

. 4

PFR-ERNA 2

@@

Fig. 3. Comparison of equilibrium radionuclide angiography (ERNA-2) and VEST (VEST-2) measurements of heart rate (top panel), ejection fraction (middlepanel) and peak filling rate (bottom

g

•

•

@ @

panel)

in the clinical range. The 95% confidence intervals of both lower and upper limits of agreement are narrow, showing an acceptable degree of agreement between VEST-2 and ERNA-2 for the 3 parameters considered in this study. Figure 3 shows the plots of HR, EF and PFR with the 2 methods. To gain more information on the agreement, we plotted the differences in the pa-

-I 0

i

!

1

2

3

AVERAGE P F R b y VEST a n d BP Fig. 4. Plot of the differences between the two methods (ERNA-2 and VEST-2) against their mean. Results obtained for heart rate (top panel), ejection fraction (middle panel) and peak filling rate (bottom panel) are shown. There are no relations between the differences and the mean for the 3 parameters

804 Table 3. Repeatability of VEST measurements (VEST study-1 versus VEST study-2) Variable

Study-1

Study-2

P

d

MEAN

20 15

CR

......

.....

.._..@

+ / - 2 SD

...............

10

HR EF PFR

68 + 1 2 50 + 1 3 1.7± 0.6

67 ± 1 0 51 ___13 L 7 ± 0.7

NS NS NS

-0.8 ±7 1.5 ± 3 -0.02±0.2

13 6.5 0.42

i

ol

5 0 -5

CR, Coefficient of repeatability

-10 -15 -20

100

I

40

i

I

50

I

60

a

70

80

90

AVERAGE H R b y VEST STUDY 1 a n d STUDY 2

80 10

60

5

Y=

19,6 + 0,Tx r = 0.85 n= 11 p < 0.05

9

40 6

40

80

9

f

0 1 O0

HR VEST STUDY 1 -5

m

m

m

m

W .

.

.

.

.

.

.

.

.

.

.

m

.

.

.

.

.

.

.

~

~

.

.

.

.

.

80 -10 0

60

$

I

i

20

30

i

I

40

I

50

~,VERAGE E F b y VEST STUDY 1

60

70

STUDY 2

and

40

0,6 20

~

4~

r = 0.97

.....................

n= 11 p < 0.05

,

41---

0,3

i

2'0

4o

6'0

6o

EF VEST STUDY I

0,0

~

-0,3

3

-0,6 0

AVERAGE

J

r= 0.~ n=ll

J

p < 0.05 0

0

i

!

i

1

2

3

PFR

I

i

1

2

by

VEST STUDY

3 1 and

STUDY

2

Fig. 6. Plot of the differences between the two VEST studies against their mean. Results obtained for heart rate (top panel), ejection fraction (middle panel) and peak filling rate (bottom panel) are shown. There are not relations between the differences and the mean for the 3 parameters

PFR VEST STUDY 1

Fig. 5. Comparison of VEST study-1 and VEST study-2 measurements of heart rate (top panel), ejection fraction (middle panel) and peak filling rate

(bottom panel)

rameter measurements against their mean (Fig. 4). No relationship between the difference and the m e a n was found, suggesting the lack of any relationship between the measurement error and the estimate o f the true value.

Repeatability Table 3 shows the results obtained with VEST study-I and VEST study-2 in patients o f group 2. H R , EF and P F R were not statistically different between the 2 studies. The coefficient of repeatability was well within the clinical range. Figure 5 shows the plots of H R , EF and P F R with the 2 VEST studies. No relationship was found between the differences in p a r a m e t e r measure-

805 merits by the 2 VEST studies and their mean, again suggesting no relationship between the m e a s u r e m e n t error and the estimate of the true value (Fig. 6).

and diastolic LV function. In addition, our analysis demonstrated that VEST is a repeatable method to calculate EF and PFR.

Discussion

References

The design of this study allowed us to assess the accuracy of VEST m e a s u r e m e n t of LV function under 2 different conditions : at the beginning and at the end of the VEST acquisition. The rationale to analyse VEST accuracy in the last period of data acquisition is based u p o n the consideration that the patient's movements could affect the stability o f the data recorder. Our findings demonstrate that the patient's m o v e m e n t s do not affect the accuracy of VEST measurements of LV systolic and diastolic funciton. Since it is well-known that the most relevant advantage of the VEST is its ability to m o n i t o r the EF continuously over time, the majority of authors have addressed the accuracy and repeatability of EF measurement by VEST (Wilson et al. 1983; de Yang et al. 1991). H o w ever, LV diastolic abnormalities (i.e. P F R ) are often present in patients without evident systolic disfunction (Bonow et al. 1981a; Cuocolo et al. 1990; Pace et al. 1989). In fairness, others (Miller et al. 1987) have not found a correlation between diastolic parameters and C A D , after corrections for a g e + h e a r t rate. Moreover, P F R has been used to m o n i t o r the effects of p h a r m a c o logical intervention on LV function (Betocchi et al. 1987; Inouye et al. 1984). Thus, it appears clinically relevant to m o n i t o r P F R continously. The VEST offers this unique possibility. To our knowledge, however, the accuracy and repeatability of P F R measurement by VEST have not been assessed so far. The present study showed that the VEST is an accurate and repeatable method to evaluate LV diastolic function. Prior studies have used linear regression and correlation analysis to examine the reliability of the VEST (Wilson et al. 1983; Tamaki et al. 1987). Other authors have outlined the drawback of these methods (Altman and Bland 1983; Bland and Altman 1986). They stressed that the correlation analysis measures the strength of a relation between 2 variables and not the agreement between 2 methods. The relevant question which arises when a new method is introduced is how m u c h it differs f r o m the old. In particular, these authors proposed to measure the "limits of a g r e e m e n t " and to use them as an estimate of a m e t h o d ' s accuracy. Moreover, linear regression and correlation analyses are not well suited to assess repeatability, which is better examined through the determination o f a "coefficient of repeatability" (Altman and Bland 1983; Bland and Altman 1986; British Standard Institution 1979). In the present study we found that the "limits of a g r e e m e n t " of EF and P F R measurements by VEST are well within the clinical range, thus suggesting that VEST is an accurate method to evaluate b o t h systolic

Altman DG, Bland JM (1983) Measurement in medicine: the analysis of method comparison studies. Statistician 32:307-317 Bairey CN, Yang L de, Berman DS, Rozanski A (1990) Comparison of physiologic ejection fraction responses to activities of daily living: implications for clinical testing. J Am Colt Cardiol 16: 847-854 Betocchi S, Cuocolo A, Pace L, et al. (1987) Effects of intravenous verapamil administration on left ventricular diastolic function in systemic hypertension. Am J Cardiol 59 : 624-629 Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurements. Lancet I: 307-310 Bonow RO, Bacharach SL, Green MV, et al. (1981a) Impaired left ventricular diastolic filling in patients with coronary artery disease : assessment with radionuclide angiography. Circulation 64:315-323 Bonow RO, Rosing DR, Bacharach SL, et al. (1981 b) Effects of verapamil on left ventricular systolic function and diastolic filling in patients with hypertrophic cardiomyopathy. Circulation 64:787-796 Breisblatt WM, Weiland FL, McLain JR, Tomlinson GC, Burns MJ, Spaccavento LJ (1988) Usefulness of ambulatory radionuclide monitoring of left ventricular function early after acute myocardial infarction for predicting residual myocardial ischemia. Am J Cardiol 62:1005-1010 British Standard Institution (1979) Precision of tests methods I: guide for the determination and reproducibility for a standard test method (BS 5497). BSI, London Cuocolo A, Sax FL, Brush JE, Maron BJ, Bacharach SI, Bonow RO (1990) Left ventricular hypertrophy and impaired diastolic filling in essential hypertension. Diastolic mechanisms for systolic dysfunction during exercise. Circulation 81:978-986 Kayden DS, Wackers FJT, Zaret BL (1990) Silent left ventricular dysfunction during routine activity after thrombolytic therapy for acute myocardial infarction. J Am Coll Cardiol 15:15001507 Kiess MC, Dimsdale JE, Moore RH, Liu P, Newell J, Barlai-Kovach M, Boucher CA (1988) The effects of stress on left ventricular ejection fraction. Eur J Nucl Med 14:12-16 Inouye IK, Massie BM, Loge D, Simpson P, Tubau JF (1984) Failure of antihypertensive therapy with diuretics, beta-blocking and calcium channel-blocking drugs to consistently reverse left ventricular diastolic filling abnormalities. Am J Cardiol 53:1583-1587 Miller TR, Fountos A, Biello DL, Ludbrook PA (1987) Detection of coronary artery disease by analysis of ventricular filling. J Nucl Med 28:832843 Pace L, Bacharach SI, Bonow RO, Cannon RO III, Green MV, Larson SM (1989) Diagnosis of coronary artery disease by radionuclide angiography: effect of combining indices of left ventricular function. J Nucl Med 30:1966-1971 Soufer R, Wohlgelernter D, De Vita NA, et al. (1985) Intact systolic left ventricular function in clinical congestive heart failure. Am J Cardiol 55:1032-1037 Tamaki N, Gill JB, Moore RH, Yasuda T, Boucher CA, Strauss HW (1981) Cardiac responses to daily activities and exercise

806 in normal subjects assessed by an ambulatory ventricular function monitoring. Am J Cardiol 59 : 1164-1169 Tamaki N, Yasuda T, Morris RH, Gill JB, Boucher CA, Hutter AM, Gold HK, Strauss HW (1988) Continuous monitoring of left ventricular function by an ambulatory radionuclide detector in patients with coronary artery disease. J Am Coll Cardiol 12: 669-679 Wilson RA, Sullivan PJ, Moore RH, Zielonka JS, Alpert NM,

Boucher CA, McKusick KA, Strauss HW (1983) An ambulatory ventricular function monitor: validation and preliminary results. Am J Cardiol 52:601-606 Yang L de, Bairey CN, Berman DS, Nichols KJ, Odom-Mayron T, Rozanski A (1991) Accuracy and reproducibility of left ventricular ejection fraction measurements using an ambulatory radionuclide left ventricular function monitoring. J Nucl Med 32: 796-802