The Development of a Control Method for a TotalArtificial Heart Using Mixed Venous Oxygen Saturation

*Makoto Nakamura, *Toru Masuzawa, *Eisuke Tatsumi, *Yoshiyuki Taenaka,*Tomomichi Nakamura, *Bin Zhang, *Takeshi Nakatani, *Hisateru Takano, and

†Takashi Ohno

*Department of Artificial Organs, National Cardiovascular Center Research Institute, Osaka; and †Tokyo DenkiUniversity, Tokyo, Japan

Abstract: For physiological control of a total artificialheart (TAH), applying mixed venous oxygen saturation(SvO2) as a parameter for TAH control is a promisingapproach regarding sensitivity to the recipient’s oxygendemand and the practical possibility of continuous moni-toring using near infrared rays through transparent bloodpump housings. To develop a control method for the TAHusing SvO2, the relationship between SvO2 and cardiac out-put (CO) was investigated in a normal calf, and a controlalgorithm was developed based on this correlation. Thenthe feasibility of this method (SvO2 mode) was evaluated in

a calf implanted with a pneumatic TAH and comparedwith the fixed drive control mode (fixed mode) in whichthe drive parameters were unchanged. The calf performeda graded exercise test in both modes. The CO was effec-tively increased from 7.3 to 13.0 L/min in the SvO2 mode,and the capacity for exercise was augmented compared tothe fixed mode. We conclude that this SvO2 mode is fea-sible and may be effectively applied in TAH control. KeyWords: Total artificial heart—Mixed venous oxygen satu-ration—Cardiac output—Control—Exercise.

A physiological control method is an importantissue in the development of a total artificial heart(TAH) system. A TAH system completely replacesthe patient’s native heart, the function of which isadequately controlled according to the patient’s flowdemands by the natural circulatory feedback systemwith its neural or humoral regulation. However, thecontrol of the TAH operation is disconnected fromthis loop. Several physiological problems have beenreported in the recipients or experimental animalsimplanted with TAHs, and it has been pointed outthat one of the important factors responsible forsuch disorders is the mismatched flow supply of theTAH for the recipients’ flow demands (1,2). There-fore, to prevent abnormal reactions, the TAH output

should meet the recipients’ flow demands and meta-bolic needs.

To achieve such physiological TAH output con-trol, it is necessary to monitor the parameters thatreflect the recipients’ flow demands sensitively.However, these parameters must be measured con-tinuously, for long periods, inside the body, and withdurable and reliable sensors, to determine their prac-tical use.

We have proposed the possibility of applyingmixed venous oxygen saturation (SvO2) as a suitableparameter for TAH control (3) because the SvO2

reflects the recipient’s oxygen metabolic conditionclosely and sensitively. Moreover, the measurementof SvO2 has already been established in the oxymet-ric catheters and in the pacemakers using this pa-rameter (4–7). Because SvO2 can also be measuredthrough the transparent blood pump housing withnear infrared rays, continuous monitoring may bepossible for long periods without contact with theblood.

The purpose of this study was to develop a physi-ological control method for the TAH using SvO2 and

Received September 1998.Presented in part at the 26th Congress of the Japanese Society

for Artificial Hearts and Assisted Circulation, held March 6 and 7,1998 in Yamagata, Japan.

Address correspondence and reprint requests to Dr. MakotoNakamura, Department of Artificial Organs, National Cardiovas-cular Center Research Institute, 5-7-1, Fujishiro-dai, Suita, Osaka565-8565, Japan.

Artificial Organs23(3):235–241, Blackwell Science, Inc.© 1999 International Society for Artificial Organs

235

to evaluate its feasibility in a series of animal experi-ments.

MATERIALS AND METHODS

Experiment 1: Investigation of the relationshipbetween SvO2 and cardiac output in a normal calfto determine the control algorithm

Preparing the animalIn a normal calf weighing 48 kg at the time of

operation, pressure monitoring lines were placed inthe intermammalian vein and artery and in the rightand left atria through thoracotomy. An oxymetriccatheter (Swan-Ganz oxymetry catheter, BaxterHealthcare Co., Irvine, CA, U.S.A.) was also in-serted into the main pulmonary artery (mPA) tomonitor the pulmonary arterial pressure (PAP) andSvO2. An electromagnetic flowmeter probe wasplaced on the ascending aorta to measure the cardiacoutput (CO). The SvO2 was measured with the SAT2 (Baxter Healthcare Co. and the CO with an elec-tromagnetic flowmeter (MFV2100, Nihon Kohden,Tokyo, Japan).

Treadmill exercise testTwo weeks later, graded treadmill exercise tests

were performed. The treadmill exercise protocol in-cluded 6 consecutive stages in which the speed of thetreadmill was increased from 1 to 6 km/h by 1 km/hat every stage. Each stage lasted for 3 min. The SvO2

and the CO were continuously measured and re-corded from approximately 1 min before the start ofexercise to 3 min after the end of exercise at a sam-pling interval of 5 ms. The data for every 5 s wereaveraged and used for the analysis.

Analysis of the dataThe absolute values of SvO2 and CO and their

differences from the values at the beginning of theexercise were compared and evaluated. The relation-ship between SvO2 and CO was also assessed to ob-tain basic data to establish a TAH control algorithm,that is, the SvO2 based driving mode (SvO2 mode).

Experiment 2: Evaluation of the SvO2 baseddriving mode (SvO2 mode)

The control algorithm, which was based on theresults of Experiment 1, was installed in the control-ler of the pneumatic TAH as the SvO2 mode. Thismode was evaluated in a series of animal experi-ments by treadmill exercise tests.

Preparing the animalA calf implanted with a pneumatic TAH was pre-

pared for this study. A polyurethane made dia-phragm type pneumatic TAH was implanted into a

calf weighing 73 kg. The effective stroke volumeswere approximately 100 ml for both right and leftpumps. The system was actuated with an externaldriver (AISIN-CA03-K, AISIN Human Co., Ltd.,Kariya, Japan). Under general anesthesia and car-diopulmonary bypass, the native ventricles werecompletely removed and the TAH was implanted.The pressure monitoring lines were fixed in the sameway as they were for the normal calf in Experiment1. An oxymetric catheter (Opticath, Abbot Labora-tory, Chicago, IL, U.S.A.) was inserted into themPA, and the probe of the ultrasonic flowmeter(Transonic T108, Transonic Systems, Inc., IthacaNY, U.S.A.) was placed on the mPA to measure theCO. After the operation, the TAH was driven undera fixed drive mode with partial filling and full eject-ing status.

Treadmill exercise testTwo weeks later, a series of treadmill exercise

tests were performed. The pneumatic TAH was ac-tuated in 2 different driving modes. One was thefixed driving mode (fixed mode), in which the driveparameters including the pump rate, positive/negative pressures, and systolic/diastolic ratio werefixed at the preexercise settings during exercise.Both pumps were actuated at partial filling and fullejecting conditions, in which the CO was passivelyaffected by the preload change via the change in thestroke volume. The second mode was the SvO2 baseddriving mode (SvO2 mode), which was derived fromthe results of Experiment 1. The treadmill exerciseprotocol was almost the same as that in Experiment1, but exercise was continued until the calf was ex-hausted and collapsed. The following were evalu-ated: exercise capacity represented by the endurancetime, the maximal stage, and the peak oxygen con-sumption; CO and SvO2; right arterial pressure(RAP), PAP, left arterial pressure (LAP), aorticpressure (AoP), and the systemic vascular resistance(SVR); and the blood lactate level measured every 3min. Finally, the SvO2 mode was compared with thefixed mode.

RESULTS

Experiment 1Exercise tests were performed 5 times on postop-

erative Days 18, 20, 23, 27, and 31 when the weight ofthe calf was from 54 to 64 kg.

The changes of CO and SvO2 during exercise inthe 5 treadmill tests are shown in Fig. 1A and B,respectively. The CO increased according to thegrade of exercise, and the SvO2 decreased inverselywith the intensity of exercise. The changes of the

M. NAKAMURA ET AL.236

Artif Organs, Vol. 23, No. 3, 1999

difference in the CO (d-CO) and those in the SvO2

(d-SvO2) compared to the preexercise values duringthe treadmill test are shown in Fig. 1C and D, re-spectively. The variation among the 5 tests wasmarkedly reduced for both CO and SvO2.

The relationships between SvO2 and CO and be-

tween d-SvO2 and d-CO are shown in Fig. 2A and B,respectively. Although some variation was foundamong these 5 tests in the relationship between SvO2

and CO in absolute values, the correlation was betterin the relationship between d-SvO2 and d-CO in rela-tive values. The coefficient of correlation between

FIG. 1. Shown are the changes in cardiac output (CO) (A), mixed venous oxygen saturation (SvO2) (B), the differences in CO comparedto the values at the beginning of exercise (d-CO) (C), and the differences in SvO2 compared to the values at the beginning of exercise(d-SvO2) (D) during 5 treadmill tests in a normal calf. The area graph expresses the speed of the treadmill.

FIG. 2. The relationship betweenmixed venous oxygen saturation(SvO2) and cardiac output (CO)(A) and the relationship betweenthe differences in SvO2 comparedto the values at the beginning ofexercise (d-SvO2) and the differ-ences in CO compared to the val-ues at the beginning of exercise(d-CO) (B) during 5 treadmill testsin a normal calf are shown.

DEVELOPMENT OF A TAH CONTROL METHOD USING SvO2 237

Artif Organs, Vol. 23, No. 3, 1999

SvO2 and CO was 0.734 in absolute values whereasthat between d-SvO2 and d-CO was 0.827 in differ-ences of values.

The development of the SvO2 based driving mode(SvO2 mode)

To develop an algorithm for TAH control, therelationship between d-SvO2 and d-CO was utilized(8). The correlation equation was

d-CO 4 −0.23 ? d-SvO2 − 0.01 (1)

The value of −0.01 in this equation was ignored as anegligible error. The targeted CO was calculatedwith Eq. 2, in which pre-CO represents the CO at thestart of each exercise test.

Targeted CO 4 pre-CO −0.23 ? d-SvO2 (2)

This targeted CO was regarded as the targeted leftpump output. Considering the right-left output bal-ance and the prevention of overperfusion for thepulmonary circuit, the targeted right pump outputwas adjusted to approximately 90% of the targetedleft pump output. The drive parameters of eachpump, including pump rate and positive/negativepressures, were adjusted to yield a respective targetoutput according to the data from in vitro tests inmock circulation. SvO2 was measured every secondand averaged every 5 s. The drive parameters wereadjusted every 5 s in response to the SvO2 deviation.All regulatory processes were automatically per-formed with a control computer.

Experiment 2The treadmill tests were performed on postopera-

tive Day 27 in the SvO2 mode and on Day 36 in thefixed mode in the TAH implanted calf, respectively.The results of each treadmill test are shown in Table1. Although the TAH implanted calf, either in thefixed mode or the SvO2 mode, could complete thetests to Stage 6 and the endurance time was onlyslightly longer in the SvO2 mode, the peak oxygen

consumption in the SvO2 mode was 1.5 times largerthan that in the fixed mode.

The changes of CO and SvO2 during the treadmilltest are shown in Fig. 3A and B. The CO did notmarkedly increase in the fixed mode during exercisewhereas it markedly increased from 7.3 to 13.0 L/minin the SvO2 mode. The minimum SvO2 values duringexercise were 28 and 32% in the fixed mode and theSvO2 mode, respectively.

The relationship between SvO2 and CO is shownin Fig. 4A. The CO increased only slightly even atlow SvO2 levels in the fixed mode. However, in theSvO2 mode, the CO demonstrated a clear negativecorrelation with SvO2 and increased significantly inresponse to the decrease in SvO2. The difference be-tween the measured CO and the targeted pump out-put in the SvO2 mode is shown in Fig. 4B. It wasrecognized that the CO change was closely corre-lated to the targeted pump output.

The changes in the AoP and the SVR are shown inFig. 5. The SVR was decreased during exercise inboth the fixed mode and the SvO2 mode. In the fixedmode, the changes in SVR were accompanied byAoP changes, and a slight decrease in the AoP wasobserved during exercise. In the SvO2 mode, how-ever, no such decrease due to the decrease in SVRwas observed.

The changes of RAP, LAP, and PAP are shown inFig. 6. Some increases were observed in these pres-sures in both modes, and the range of elevation waswithin acceptable levels. There were some fluctua-tions in the LAP in the fixed mode and in the SvO2

mode, which were considered to have resulted fromnegative pressure during the diastolic phase.

The changes of blood lactate levels are shown inFig. 7. There was no marked elevation in lactate lev-els in either the fixed mode or the SvO2 mode.

DISCUSSION

From a physiological point of view, the applicationof SvO2 as an input parameter for TAH control is a

TABLE 1. Results of treadmill exercise tests in a total artificial heartimplanted calf

SvO2 mode Fixed mode

Postoperative day (day) 27 36Body weight (kg) 89.6 99.6Hemoglobin (g/dl) 10.2 10.4Pre-SvO2 (%) 52 63

Endurance time 16 min 01 s 15 min 32 sMaximal stage Stage 6 (6 km/h) Stage 6 (6 km/h)Peak oxygen consumption (ml/min) 1,049 769

(ml/min/kg) 11.7 7.7

SvO2 mode: SvO2 based control mode; fixed mode: fixed drive control mode.

M. NAKAMURA ET AL.238

Artif Organs, Vol. 23, No. 3, 1999

promising approach because SvO2 is a sensitive indi-cator of the recipient’s oxygen metabolic condition.The concept of applying the oxygen metabolic pa-rameters to the TAH control was first reported byStanley et al. (9). They demonstrated that TAH im-planted calves maintained better aerobic conditionin the control via mixed venous oxygen tension thanin the control via atrial pressure alone. Takatani etal. developed an oxygen saturation sensor for theTAH or biventricular bypass and demonstrated thatthe SvO2 responded well to changes in the recipients’oxygen consumptions in animal experiments by con-tinuous monitoring of SvO2 (10,11). Schima et al. re-ported that the oxygen extraction ratio (EO2), which

was calculated with the SvO2, was a useful parameterfor control of the TAH driving condition to treatTAH recipients in clinical human cases (12). Asmentioned, whether used directly or indirectly, SvO2

is considered to have a substantial potential as aphysiological parameter for TAH control.

From a technological viewpoint, the practical tech-nology of SvO2 measurement has already been es-tablished in the oxymetric catheter, rate responsivepacemaker, and blood gas monitoring system (4–7,10,11,13). The main problems to overcome for TAHcontrol use are sensor stability, reliability, and dura-bility (14). In the rate response pacemaker usingSvO2, the normal sensor function has been report-edly confirmed for up to 58 months follow-up insidethe body (7). Although we used an oxymetric cath-eter to measure SvO2 in this study, our final goalincludes the development of a built-in type oxymet-ric sensor which is placed on the right pump housingor the outlet graft of the TAH. Therefore, SvO2

could be measured with red and near infrared raysthrough the transparent blood pump housing. If thesensor is equipped at an adequate washout portionof the pump, the possibility of measuring error dueto the floating sensor position or the deposit of bloodelements on the sensor surface is reduced, and con-tinuous monitoring for a long period may be pos-sible. Moreover, employing 3 wavelength lights mayresult in the measurement being more accurate andstable and frequent calibration unnecessary. Furtherdevelopment is needed in the sensing device so thatit can be used practically; nevertheless, the potentialfor SvO2 monitoring is great.

Although some people may believe that the re-

FIG. 4. Shown are the relation-ship between mixed venous oxy-gen saturation (SvO2) and cardiacoutput (CO) in the fixed drivingcontrol mode (fixed mode) and inthe SvO2 based control mode(SvO2 mode) (A) and the relation-ship between SvO2 and CO in theSvO2 mode and the targeted leftand right pump outputs in theSvO2 mode (B).

FIG. 3. Shown are the changes in cardiac output (CO) and mixedvenous oxygen saturation (SvO2) during the treadmill test in atotal artificial heart implanted calf in the fixed driving control mode(fixed mode) (A) and in the SvO2 based control mode (SvO2

mode) (B). The area graph expresses the speed of the treadmill.

DEVELOPMENT OF A TAH CONTROL METHOD USING SvO2 239

Artif Organs, Vol. 23, No. 3, 1999

sponse of SvO2 to the changes in metabolic demandmay be slow for physiological control, the fact that aclose relationship between SvO2 and CO was recog-nized both in a normal calf in Experiment 1 and inthe TAH implanted calf under the SvO2 mode inExperiment 2 indicated that the response speed ofSvO2 was similar to that of CO in normal subjectsand demonstrated that the possible delay was notproblematic for TAH output control even during ex-ercise when rapid control was required.

To summarize the results of the treadmill exercisetest in the SvO2 mode compared to the fixed mode,the capacity for exercise was better than in the fixedmode; CO changed sensitively in response to SvO2

change; regarding hemodynamics, no decrease wasobserved in the AoP accompanied by a decrease inthe SVR during exercise; and the blood lactate leveldid not increase during exercise. The feasibility ofthe SvO2 mode was revealed in the in vivo experi-ment. Moreover, the decrease in SvO2 indicated thatthe increase in oxygen consumption and the increasein CO represented an increase in oxygen supply tothe whole body. Because blood pressure rose signifi-cantly during exercise in normal subjects, it is con-sidered that oxygen is likely to be delivered to theperipheral tissues by maintaining the AoP at highlevels during exercise. These findings suggest thatthe SvO2 mode was advantageous for the recipient’smetabolic needs during exercise.

In this study, a negative pressure was observed inthe LAP during exercise due to insufficient pulmo-nary venous return to the left pump output, suggest-ing that the drive setting balance in the right pump,which was approximately 90% of the left, shouldhave been changed during the exercise. Therefore,an appropriate dynamical balancing mechanism or amechanism to prevent the LAP from rising is re-quired to accommodate the variable left-right outputimbalance and to make this control method morepractical and safe.

CONCLUSIONS

A control method for the TAH using the SvO2 wasdeveloped based on a normal relationship betweenthe SvO2 and CO from a study with a normal calf.

FIG. 5. Shown are the changes in aortic pressure (AoP) andsystemic vascular resistance (SVR) during the treadmill test in atotal artificial heart implanted calf in the fixed driving control mode(fixed mode) (A) and in the SvO2 based control mode (SvO2

mode) (B). The area graph expresses the speed of the treadmill.

FIG. 6. Shown are the changes in right atrial pressure (RAP), leftatrial pressure (LAP), and pulmonary arterial pressure (PAP) dur-ing the treadmill test in a total artificial heart implanted calf in thefixed driving control mode (fixed mode) (A) and in the SvO2 basedcontrol mode (SvO2 mode) (B). The area graph expresses thespeed of the treadmill.

FIG. 7. Shown are the changes in blood lactate levels during thetreadmill test in a total artificial heart implanted calf in the fixeddriving control mode (fixed mode) and in the SvO2 based controlmode (SvO2 mode). The area graph expresses the speed of thetreadmill.

M. NAKAMURA ET AL.240

Artif Organs, Vol. 23, No. 3, 1999

Applying this method to a TAH implanted calf, theCO was effectively controlled during exercise justlike it was in the normal calves, and, as a result, thehemodynamics were stably maintained and the ca-pacity for exercise was augmented. In conclusion, anSvO2 based control method of the TAH was revealedto be feasible, and the possibility and potential ofthis control method were successfully clarified.

Acknowledgments: These experiments were supportedin part by a Research Grant for Cardiovascular Diseases(7A-1) from the Ministry of Health and Welfare of Japanand a Grant-in-Aid for Scientific Research from the Min-istry of Education of Japan (No. 07507009) and the NewEnergy and Industrial Technology Development Organi-zation (NEDO).

REFERENCES

1. Deleuze PH, Riebman JB, De Paulis R, Olsen DB, LoisanceDY. Regulation of pneumatic total artificial heart function; Areview. Int J Artif Organs 1988;11:147–52.

2. Nakamura M, Tatsumi E, Taenaka Y, Masuzawa T, Sohn Y,Nakatani T, Nishimura T, Endo S, Ohno T, Takiura K,Takewa Y, Takano H. Early changes in circulating blood vol-ume and volume-regulating humoral factors after implanta-tion of an electrohydraulic total artificial heart. ASAIO J1997;43:M663–8.

3. Ohno T, Masuzawa T, Nakamura M, Tatsumi E, Taenaka Y,Nakatani T, Wakisaka Y, Nishimura T, Takewa Y, Takano H.Characteristics of mixed venous oxygen saturation and physi-cal activity as parameters for artificial heart control. ASAIO J1997;43:M677–81.

4. Gettinger A, DeTraglia MC, Glass DD. In vivo comparison oftwo mixed venous saturation catheters. Anesthesiology 1987;66:373–5.

5. Wirtzfeld A, Heinze R, Liess HD, Stangl K, Alt E. An activeoptical sensor for monitoring mixed venous oxygen-saturationfor an implantable rate-regulating pacing system. Pace1983;6(Part-II):494–7.

6. Seifert GP, Moore AA, Graves KL, Lahtinen SP. In vivo andin vitro studies of a chronic oxygen saturation sensor. Pace1991;14(10):1514–27.

7. Faerestrand S, Ohm OJ, Stangeland L, Heynen H, Moore A.Long-term clinical performance of a central venous oxygensaturation sensor for rate adaptive cardiac pacing. Pace 1994;17:1355–72.

8. Ohno T, Fikui Y, Masuzawa T, Nakamura M, Tatsumi E,Taenaka Y, Takano H. Development of total artificial heartcontrol with mixed venous oxygen saturation and physicalactivity. JJME 1998;36(Suppl.):572.

9. Stanley TH, Volder J, Kolff WJ. Extrinsic artificial heart con-trol via mixed venous blood gas tension analysis. Trans AmSoc Artif Intern Organs 1973;19:258–63.

10. Takatani S, Noda H, Takano H, Akutsu T. Optical sensor forhemoglobin content and oxygen saturation: Application toartificial heart research. Trans Am Soc Intern Organs 1988;348:808–12.

11. Takatani S, Noda H, Takano H, Akutsu T. Continuous in-linemonitoring of oxygen delivery to control artificial heart out-put. Artif Organs 1990;14(6):458–65.

12. Schima H, Trubel W, Coraim F, Huber L, Muller MR, RedlG, Losert U, Thoma H, Wolner E. Control of the total arti-ficial heart: New aspects in human versus animal experience.Artif Organs 1989;13(6):545–52.

13. Takatani S, Tanaka T, Nakatani T, Takano H, Akutsu T,Kohno H, Nudeshima H, Takahashi A. Development of he-moglobin oxygen optical sensors for automatic control of ar-tificial heart output. Trans Am Soc Artif Intern Organs 1985;31:45–9.

14. Allen GS, Murray KD, Olsen DB. Control of the artificialheart. ASAIO J 1996;42:932–7.

DEVELOPMENT OF A TAH CONTROL METHOD USING SvO2 241

Artif Organs, Vol. 23, No. 3, 1999