Catheterization and Cardiovascular Diagnosis 11 :473-481 (1985)

Arteriovenous Shunt Measured by Bolus Dye Dilution: Reproducibility and Comparison Between Two Injection Sites

Paulo Rocha, MD, Jean-Claude Kahn, MD, Gerard Dongradi, MD, Bernard Baron, MD, and Jean-Pierre Fendler, MD

Twenty-eight brachial arteriovenous fistulae (AVF) flows were assessed by the Stew- art and Hamilton method by bolus dye injection. These measurements were divided in two groups: a first group with dye injection into the AVF artery and a second group with dye injection into the efferent vessel of the AVF in close proximity. The increase and the decrease of dye concentration were regular and the circulation occurred very late in both groups. Reproducibility was assessed by the usual index: the mean of the differences between two successive measurements of each series related to the first of these two and expressed as a percentage, m(Q, - Q, - l)/Q,O/o. In the two groups, the reproducibility index was at 10.1%, similar to the index applied to Grim- by’s results, measuring successive cardiac output by dye bolus injection at 9.8%. Theoretical criteria of validity of the Stewart and Hamilton method were checked for all measurements. Even when the duration of the measurement was very short, arteriovenous flow fulfilled the criteria of validity in the same way as cardiac output. Two AVF flows were measured successively at both injection sites with no difference between the obtained values. The same reproducibility of the efferent vessel injec- tion site group suggests that an arterial puncture is not necessary for a correct assessment of AFV flow.

Key words: dye dilution, regional blood flow, arteriovenous fistula, hemodialysis

INTRODUCTION

Arteriovenous fistulae (AVF) in chronically hemodialyzed patients are an obvious cause of increase in cardiac output and appear to be capable of producing cardiac failure [l]. The flow of the AVF has often been measured by methods of poor precision or high complexity, ie, by variation of the cardiac output after AVF occlusion [2-41, by electromagnetic flowmetry [5,6], by plethysmography [7], or by Doppler effect [8].

From the Poissy Hospital, Poissy, France.

Address reprint requests to Dr. J.C. Kahn, Cardiology Department, HBpital de Poissy, 78303 Poissy, France.

Received March 11, 1985; revision accepted July 11, 1985.

0 1985 Alan R. Liss, Inc.

474 Rocha et al

Electromagnetic flow measurement is probably a very accurate method of calculat- ing AVF blood flows, but it requires surgical exposure of the vessels. Furthermore, repeated measurements have shown that flows increase quickly later and that a first assessment during surgery cannot predict the output of the mature fistula [5,6]. Variation of cardiac output after AVF occlusion is indeed an easy technique, but results seem much lower than the real value of AVF flow [9].

In the last few years, some investigators have employed tracer dilution, assuming that the thrill felt over the AVF supposes a mixer site at this level, with injection of tracer by bolus [9,10] or by continuous infusion [ I 1-13]. However, for many reasons, the formulas of Stewart and Hamilton [14] with tracer injected by bolus cannot be employed in this kind of flow without reservations, particularly in short curves. We review here the criteria of validity for this method in AVF flow and compare the reproducibility of measurements with two injection tracer sites, in the artery feeding the fistula and in the efferent vessel of the fistula, this latter avoiding an arterial puncture.

PATIENTS AND METHOD

The study involved 28 chronically hemodialyzed patients, 16 men and 12 women. Written consent for investigations was obtained after detailed description of the procedure. Ages ranged from 20 to 64 years (mean = 40), duration of dialysis ranged from 1 to 63 months (18 k 17 months; mean +_ standard deviation), body areas ranged from 1.35 to 1.88 m2 (1.63 k 0.14 m2), and hemoglobinemias ranged from 4.2 to 10.2 g/100 nil (7.0 +_ 1.7 g/100 ml). Twenty-one patients had a side to end AVF, four had bovine carotid grafts, two had saphenous vein grafts, and one had a polytetrafluorethylene graft. AVF was always in the arm or forearm. Two AVF were studied twice.

We assumed that there was a mixing chamber in the AVF, and we were able to measure the AVF flow by the Stewart and Hamilton method. A rapid injection of a bolus of 0.50-1.25 mg of cardiogreen into 0.2-0.5 ml of isotonic glucose solution was done through a plastic catheter of 10 cm length and 1 mm diameter. Blood sampling was done downstream by another catheter of the same dimensions at a constant rate of 23 ml/min (Watson Mariow Pump 200) through a Water dichromatic cuvette X02 connected to a Water Instruments densitometer MD-41 and recorded over an X-Y table. After the first passage through the cuvette, blood was reinfused in another vein. The calibration of the cuvette’s responsiveness was determined by passing three samples of each patient’s blood with different concentrations of dye (three times 5 ml of blood). The flow was assessed by the formula

Q=mS;C.dt[l4], where Q = flow, m = amount of dye injected, and C.dt =

mathematical integration of the concentration of the tracer in the blood during its first passage. The mean transit time (MTT) was calculated by DOW’S method [15]; t was the duration of the pulsation. Each measurement was repeated three times. The direction of the flow in all radial arteries was studied by Doppler effect upstream and downstream to the AVF.

The 30 measurements were divided in two groups. In the first group (n = 14), the tracer was injected into the afferent artery of the AVF, 10 cm upstream, and the concen-

s,

Arteriovenous Shunt Measured by Bolus Dye Dilution 475

tration of the tracer in the blood was studied in the efferent vein, 10 cm downstream to the AVF (Fig. 1). In the second group (n = 16), the tracer was injected into the efferent vein of the AVF, or in the upper segment of the graft linking artery and vein, with the needle going against the direction of flow. The sampling site was about 10 cm lower down in the same vessel (Fig. 2) . Homogenous mixing of dye in the blood is followed by an exponential decay of the concentration. For all 30 measurements, dye concentra- tions were aligned on logarithmic schedule paper, and exponential decay was confirmed by an exponential correlation program (H.P. 41).

Reproducibility was assessed by one index employed by Olsson et a1 [16]. This index was the mean of the differences between two successive measurements related to the first of these two and expressed as a percentage: m (Q, - Q,- ,)/Q,%. AVF reproducibility index was compared to the same index applied to the results of Grimby et a1 1171 measuring serial cardiac output by the dilution bolus in steady state.

MTT/t was calculated for each measurement. Reproducibility indexes of the pa- tients at MTT/t < 4 (short curves) were compared to the reproducibility indexes of all measurements. All data are expressed in the text as mean _+ SEM and the tables as each measurement value with mean k SEM. Statistical comparisons were carried out using Student’s t test for paired numerical data.

RESULTS

All the results are shown in Tables I and 11. The general shape of the dilution curves was the same whether the tracer was injected into the afferent artery or into

Fig. 1. bolus injection into the AVF artery.

Arteriovenous fistula (AVF) flow measured by Stewart and Hamilton method by dye

476 Rocha et al

Fig. 2. Arteriovenous fistula (AVF) flow measured by Stewart and Hamilton method by dye bolus injection into the efferent vessel.

the efferent vessel, and the increase in the dye concentration was regular with the recirculation occurring very late (Fig. 3) . In each case, except No. 6, the downward part of the curve, before the dye recirculation, was an exponential curve.

The mean transit time of the curves was on average 4.9 sec when the tracer was injected in the afferent artery and 4.2 sec when it was injected in the efferent vessel. In six cases (Nos. 1, 8, 14, 16, 24, and 26) the mean transit time (MTT) was short compared to the period of the pulsatile flow (t) and MTT/t was between 3 and 4.

The reproducibility index was less than 20% for each measurement, except Nos. 6 and 28. This mean was 10.1 % in both groups. For the two AVF studied twice (with both ways of injection), there was no difference between the means of the measure- ments (13 compared to 15 and 14 compared to 16).

THEORETICAL REVIEW Theoretical Conditions of Validity of the Stewart and Hamilton Method in Stationary Flow [15,18,19]

the concentration.

first passage of the tracer.

1. Mixing of the tracer in the blood must be homogenous with exponential decay of

2 . The recirculation of the tracer has to come late to permit the extrapolation of the

3. All the tracer has to cross the mixing site.

Arteriovenous Shunt Measured by Bolus Dye Dilution 477

TABLE 1. Measurements of Arteriovenous Fistula Flows (&VF) by Dye Bolus Injection Into the Afferent Arterv of AVF'

Difference between two Index o f

measurements (%) duration of the measurements

Flow measurements for each AVF (literirnin) 6 3 - 6 2 r M T T Q2 - 61 ~.

No. Type QI Q2 Q3 .. Mean SD 6, Q2 (set) t

I AVF 3.70 4.18 3.66 3.85 0.29 13 12 2.6 3.6 2 AVF 1.78 2.05 2.37 2.07 0.29 15 16 7.7 12.4 3 AVF 1.66 1.95 1.65 1.75 0.17 17 15 4.0 6.1 4 AVF 1.42 1.13 1.02 1.19 0.20 20 17 4.1 6.1 5 AVF 2.30 2.11 2.14 2.18 0.10 8 I 4.7 7.2 6 AVF 0.59 0.72 0.46 0.59 0.13 22 36 5.4 6.7 I AVF 0.62 0.67 0.72 0.67 0.05 8 7 9.2 12.3 8 AVF 1.61 1.68 1.57 1.62 0.06 4 7 3.0 3.6 9 AVF 0.79 0.76 0.71 0.75 0.04 4 7 5.5 8.2 10 AVF 0.97 0.95 0.96 0.96 0.01 2 I 5.0 5.3 I1 AVF 1.45 1.20 1 . 1 1 1.25 0.18 17 8 4.3 4.8 12 AVF 2.38 2.46 2.34 2.39 0.06 5 5 4.7 7.2 13 AVF 0.99 1.06 1.06 1.04 0.04 7 0 4.7 6.0 14 AVF 0.41 0.38 0.30 0.39 0.02 7 3 3.3 3.1 Mean 1.48 1.52 1.44 1.48 10.12 10.6 9.6 4.9 6.6 SD 10.89 k0.99 kO.92 k0.93 kO.10 k6.6 f9 .4 k1.8 k2.9

*Abbreviations: MTT, mean transit time; Q,, 6, Q.3, first, second, and third tlow fistulae measurements: t , duration of pulsatile flow (cardiac cycle duration). For this group, the ycan of the differences between two successive measurements expressed in percentage, m(Q, ~- Q,, . l)/Qr,, is 10. I%.

4. The flow of the mixer site has to be constant. 5 . The flow of the withdrawing vessel has to be constant as well. 6. The conditions of the measurements must not change the measured flow. When these conditions are satisfied, flows can be measured by the Stewart and

Hamilton formula.

Theoretical Considerations in the Measurements of a Nonstationary Flow

In the circulation, the pulsatile nature of the t!ow exposes the application of the formulas of Stewart and Hamilton to two basic errors 120,221. I ) Wrong mewz error: The instantaneous bolus of tracer can be mixed in the tlow of one systole or one diastole and does not represent the real mean flow. 2) Disfunce ifisfortion error: This depends on the distance between the mixer site and the sampling site. The systolic flow of the mixer site can pass in front of the sampling site at the speed of the diastolic tlow or vice versa.

These two possible errors are negligible if the duration of the measurement is higher than four times the duration of the pulsation [20,22,23].

DISCUSSION

An analysis of the dilution curves of 30 fistula tlow measurements has been done according to the theoretic criteria of validity of the Stewart and Hamilton method.

478 Rocha et al

TABLE II. Measurements of Arteriovenous Fistula Flows (hVF) by Dye Bolus Injection Into the Efferent Vessel of AVF’

Difference between two Index of

measurements duration ofthe ______ measurements

Flow measurements for each AVF (liter/min) Q2 - Ql Q3 - Qz MTT MTT No. Type Q, & 6, Mean SD Q, 4 2 (set) t

15 AVF 1.20 1.10 0.82 1.04 0.20 8 25 5.4 7.0 16 AVF 0.41 0.38 0.39 0.39 0.02 7 3 3.3 3.3 17 AVF 1.34 1.40 1.53 1.42 0.10 4 9 4.0 5.3 18 AVF 2.00 1.95 2.03 1.99 0.04 3 4 5.1 6.9 19 AVF 1.36 1.33 1.35 1.35 0.01 2 2 4.4 5.8 20 AVF 0.82 0.76 0.85 0.81 0.05 7 12 4.6 7.5 21 AVF 0.90 0.90 0.95 0.92 0.03 0 6 5.1 7.0 22 AVF 2.74 2.37 2.70 2.60 0.20 14 14 4.5 7.5 23 AVF 1.33 1.33 1.09 1.25 0.13 0 18 3.9 5.3 24 CAR 1.07 0.91 1.04 1.01 0.09 15 14 3.2 3.2 25 CAR 2.21 2.48 2.88 2.52 0.34 12 16 4.2 8.4 26 CAR 0.48 0.42 0.43 0.44 0.03 13 2 2.7 3.8 27 CAR 1.68 1.74 1.76 1.73 0.04 4 1 5.2 6.9 28 SAP 2.51 1.99 3.34 2.61 0.68 21 68 3.2 4.7 29 SAP 1.01 0.95 0.93 0.96 0.04 6 2 3.8 5.6 30 PTFE 0.49 0.53 0.52 0.51 0.02 8 2 4.0 5.4 Mean 1.35 1.28 1.41 1.35 0.13 7.8 12.4 4.2 5.9 SD k0.71 10.66 k0.90 k0.75 k0.17 k5.9 k16.5 k0.8 +1.6

*Abbr+atjons; CAR, bovine carotid fistula; MTT, mean transit time; PTFE, polytetrafluorethylene graft; Q,, Q, Q3, first, second, and third flow fistulae measurements; SAP, saphenous vein graft; t, duration of cardiac cycle. For this group, the mean of the differences between two measurements related to the first is: m ( Q - Q - ,)la = 10. I %.

Conditions of Validity in Stationary Flow [15,18,19]

1. Even if there was a thrill over the efferent vessel of the AVF, bolus dye injection has always been done against the AVF direction to try to create by itself a rapid and homogenous mixture of the dye in the blood. An exponential decay of the concentra- tion in 29 of 30 series of measurements suggests this homogenous mixing at the withdrawing site. This was confirmed by a similar reproducibility of the results.

2. Recirculation of the tracer was later than in cardiac output curves and extrapo- lation of its first passage was easily done in 29 of 30 measurements.

3. Some measurements of AVF flows of the first group, with arterial injection tracer site, did not fulfill the third criterion: all the tracer has to cross the mixing site. In eight of 14 of these fistulae, the Doppler effect showed that the radial artery, while feeding the fistula, went downstream to feed the hand too. Thus one part of the tracer, going to the hand, did not pass through the fistula in its first circulation. However, these eight fistulae (Nos. 2, 6, 8, 10, 11, 12, 13, and 14) had a reproducibility index of 9.7 % , similar to the full group index. This suggests a turbulent flow in the arterial segment of the fistula; otherwise, the fraction of dye going to the hand would not be constant. If, in fact, there is one mixing chamber in the artery feeding the fistula, the possible loss of tracer is necessarily proportional to the flow to the hand, which is usually less than 20 ml/min [7] and negligible compared to fistula flow. Two of these fistulae had been measured twice, by dye injection in both sites: Nos. 13 and 14

Arteriovenous Shunt Measured by Bolus Dye Dilution 479

i 1 Fig. 3. Curves obtained in the same AVF with injection into afferent artery (left three) and efferent vessel (right three).

measured by arterial bolus injection compared to Nos. 15 and 16, measurements of the same fistulae by dye injection in the efferent fistula vessels. Identical values were obtained in the two injection sites. For these two fistulae, the loss of dye was in fact negligible, and it could not be appreciated by this method. For the six other measure- ments in the first group, there was no loss of tracer; all fistulae were fed by both arterial segments upstream and downstream to the anastomosis. There was no possi- bility of other entering flows between the injection and the sampling sites. The collaterals of the vein are sutured by the surgeon who made the AVF to improve the development of the vein and the prostheses are not permeable. Even if there was one collateral in the efferent vessel, the pressure in the AVF would be higher than in the other veins, similar to an arterial pressure, and would avoid all entering flow.

4 and 5 . These conditions require a stationary flow, and these criteria are never satisfied in the circulation. However, except for the very short curves, the oscillatory linked errors are as disturbing for fistulae as for cardiac output measurements.

6. The tracer volume injected by bolus was between 0.2 and 0.5 ml, a very small amount compared to the flow itself, which was about 25 ml/sec with a mean time measurement of 4.5 sec and thus with a dilution volume of more than 100 ml.

The sampling pump withdrew blood at 0.4 ml/sec, a small flow compared to the mean fistula flow. However, this pump flow could disturb the smaller fistula flows.

480 Rocha et al

Nevertheless, pumping action has never decreased pressure in the efferent vessel, which could provoke an increase in local flow.

Conditions of Validity in Nonstationary Flow [20-231

The two possible errors induced by a nonstationary flow are negligible if the transit time between the injection site and the sampling site (here the mean transit time, MTT), is four times longer than the duration of one pulsation (here the duration of one cardiac cycle, t) [21-231. The relation MTT/t was at 6.6 for the first group and at 5.9 for the second group. Nevertheless, in six cases (Nos. 1, 8, 14, 16, 24, and 26) this relation was less than 4. However, their reproducibility index was at 9 .9%, not worse than for the other fistulae.

The reproducibility index was similar in the shorter curves and exactly the same in both tracer injection sites. This index, for all the 30 fistulae at 10.1 %, was similar to the index of Grimby’s traditional measurements of five successive cardiac outputs by dye bolus injection at 9.8%. However, even if our measurements satisfy the underly- ing criteria for validity of dilution technics, they are not exactly validated. Validation would require an independent and accurate technic easily applicable to this kind of flow.

CONCLUSIONS

Theoretical criteria of validity of the Stewart and Hamilton method are fulfilled for arteriovenous brachial fistulae as well as for cardiac output measurement with the same reproducibility. Injection of tracer in the efferent vessel of the fistulae avoids an arterial puncture and is a guarantee of a passage for all the tracer in the fistula mixing chamber.

REFERENCES

I. Anderson CB, Codd JE, Graff RJ, Groce MA, Harter HR, Newton WT: Cardiac failure in upper extremity arteriovenous dialysis fistulae. Arch Intern Med 136:292. 1976.

2. Nenno AD, Zizzi J, Hodson J: An evaluation of the radial arteriovenous fistula as a substitute for the quinton shunt in chronic hemodialysis. Transplant [Am Soc Artif Intern Organs] 13:62, 1967.

3. Johnson G Jr, Blythe WB: Hemodynamic effects of arteriovenous shunts used for hemodialysis. Ann Surg 171:715, 1970.

4. Thompson BW, Babour G, Bisset J: Internal arteriovenous fistula for hemodialysis. Am J Surg 124:785, 1972.

5. Ehrenfeld WK, Grausz H, Wylie EG: Subcutaneous arteriovenous fistulae for hemodialysis. Am J Surg 124:200, 1972.

6. Anderson CB, Etheredge EE, Harter HR, Codd JE, Graff RJ, Newton WT: Blood flow measure- ments in arteriovenous dialysis fistulas. Surgery 81 :459-46 I, 1977.

7. Hurwich BJ: Plethysmographic forearm blood flow studies in maintenance hemodialysis patients with radial arteriovenous fistulae. Nephron 6:673, 1969.

8. Ahearn DJ, Maher JF: Heart failure as a complication of hernodialysis arteriovenous fistula. Ann Intern Med 77:201, 1972.

9. Dongrddi G, Rocha P, Kahn JC, Baron B, Marichez P, Fendler JP: Arteriovenous fistula blood flow in chronic hemodialysis patients. Comparison of results obtained by the dye-dilution method and by variation in cardiac output during temporary occlusion of the AVF. Proc EDTA 16:690, 1979.

10. Kult von J, Scheitza E, Crosswendt J, Klustsch K: Das reale shuntvolurnen subkultaner arteriove- nous fisteln bei chronisch hemodialysierten patienten. Z Kardiol 62: 1%- 164, 1973.

Arteriovenous Shunt Measured by Bolus Dye Dilution 481

1 I . Gottlieb S, Garcia E, Cold B, Vanderwerf B: Radiotracer method for nonsurgical measurement of blood flow in bovine graft arteriovenous fistulae. Proc Clin Dial Transplant Forum 6: 107-109, 1976.

12. Kaye M, Lemaitre P, O’Regan S: A new technique for measuring blood flow in polytetrafluorethy- lene grafts for hemodialysis. Clin Nephrol 8:533-534, 1977.

13. Revillon L, O’Regan S, Robitaille P, Ducharme G, Davignon A: The effects of brescia-cimino fistulae and cardiac function in transplanted pediatric patients. Clin Nephrol l2:26-3 1, 1979.

14. Stewart GN: The output of the heart in dogs. Am J Physiol 57:27, 1921. 15. Dow P: Estimations of cardiac output and central blood volume by dye-dilution. Physiol Rev 36:77-

102, 1956. 16. Olsson B, Pool J, Vandemoten P, Varnauskas E, Wassen R: Validity and reproducibility of

determination of cardiac output by thermodilution in man. Cardiology 55: 136-148, 1970. 17. Grimby G, Nilsson NJ, Sanne H: Serial determinations of cardiac output at rest. Br Heart J 28: 118-

121, 1966. 18. Zierler KL: Theoretical basis of indicator dilution methods for measuring flow and volume. Circ

Res 10:393-407, 1962. 19. Guyton AC, Jones CE, Coleman TG: Indicator dilution methods for determining cardiac output. In

Guyton AC (ed): “Circulatory Physiology: Cardiac Output and Its Regulation. ” Philadelphia: W.B. Saunders Company, 1973, p 40.

20. Cropp GJA, Burton AC: Theoretical considerations and model experiments on the validity of indicator dilution methods for measurements of variable flow. Circ Res 18:26-48, 1966.

21. Lowe RD: Use of local indicator dilution technique for the measurement of oscillatory flow. Circ Res 22:49-56, 1968.

22. Bassingthwaighte JB, Knopp TJ, Anderson DU: Flow estimation by indicator dilution (bolus injection). Circ Res 27:277-290, 1970.

23. Pavek E, Pavek K, Boska D: Mixing and observation errors in indicator-dilution studies. J Appl Physiol 28:733-740, 1970.