Rate Adaptive Cardiac Pacing Using Right Ventricular Venous Oxygen Saturation: Quantification of Chronotropic Behavior During Daily Activities and Maximal Exercise CHU-PAK LAU, YAU-TING TAI, WING-HUNG LEUNG, SUM-KIN LEUNG/ JOHN P.-S. LI, GHEUK-KIT WONG, IRIS S.-F. LEE, GHARLES YERIGH,** and MARKERIGKSON** From Division of Cardiology, Department of Medicine, University of Hong Kong, Queen Mary Hospital, and the *Princess Margaret Hospital, Hong Kong; and * "Medtronic. Inc., Minneapolis, Minnesota LAU, C.-P., ET AL.: Rate Adaptive Cardiac Pacing Using Right Ventricular Venous Oxygen Saturation: Quantification of Chronotropic Behavior During Daily Activities and Maximal Exercise. Central venous oxygen saturation (SyOz) closely reflects cardiac output and tissue oxygen consumption. In tbe absence of an adequate chronotropic response during exercise, SyO2 will decrease and the extent of desaturation may be used as a parameter for rate adaptive cardiac pacing. Eight patients with sinoatrial disease received a DDDR pacemaker capable of DDDR pacing by sensing either S^O2 or piezoelectric detected body move- ment. Both sensors were programmed to attain a rate of about 100 beats/min during walking, and with the lower and upper rates set at 50% and 90% of age predicted maximum, respectively. Chronotropic behavior of the two sensors were compared in tbe DDD mode with measurement of sensor responses, during everyday activities (walking, stair climbing, postural changes, and physiological stresses) and at each quartile of workload during a continuous treadmill exercise test. During walking at 2.5 mph, both sensors showed no significant difference in delay time (both react within 15 sees) or half-time (S^O-i = 36 ± 12 sec and activity 24 ± 3 sec; P = NS), although SVO2 driven pacing achieved 90% target rate response slower than activity sensing (124 ± 16secvs77 ± 10 sec; P < 0.02). S^O-i pacing was associated with a more physiological rate response during walking upslope (68 ± 12 beats/min vs 57 ± 10 beats/ min; P < 0.05), ascending stairs (59 ± 10 beats/min vs 31 ±6 beats/min; P < 0.05), and standing (34 ± 7 beats/min vs 9 ± 2 beats/min; P < 0.05). The SvO2 sensor significantly overpaced in the first quartile of exercise (51.8 ± 25.6% in excess of heart rate expected from workload), but the rate was within 20% of expected for the remainder of exercise. "Underpacing" was observed with the activity sensor at the higher workload. In conclusion, the SvO2 sensor demonstrated a more physiological response to activities of daily living compared with the activity sensor. Using a quantitative method, the speed of onset of rate response of tbe SyO2 sensor was comparable to activity sensing, and was more proportional in rate re- sponse. Significant overpacing occurs at the beginning of exercise during S^Oz driven pacing, which may be improved with the use of a curvilinear algorithm. (PACE 1994; 17[Pt. I]:2236-2246) central venous oxygen saturation, activity sensing, rate adaptive pacing, sensors Introduction Address for reprints; Chu-Pak Lau, M.D., Reader in Cardiology, Physical activities result in an increase in car- Dept. of Medicine, University of Hong Kong, Queen Marv Hos- - • t 1 L 1 j pital. Hong Kong. Fax: (852)" 2855-1143. ' diac output and oxygen extraction from the blood. Received October 6. 1993; revision April 7, 1994; accepted If cardiac output does not match the requirements April 11.1994. of increased tissue oxygen consumption, a widen- 2236 December 1994, Part I PACE, Vol. 17
Transcript
Rate Adaptive Cardiac Pacing Using RightVentricular Venous Oxygen Saturation:Quantification of Chronotropic Behavior DuringDaily Activities and Maximal Exercise
From Division of Cardiology, Department of Medicine, University of Hong Kong, Queen MaryHospital, and the *Princess Margaret Hospital, Hong Kong; and * "Medtronic. Inc., Minneapolis,Minnesota
LAU, C.-P., ET AL.: Rate Adaptive Cardiac Pacing Using Right Ventricular Venous Oxygen Saturation:Quantification of Chronotropic Behavior During Daily Activities and Maximal Exercise. Central venousoxygen saturation (SyOz) closely reflects cardiac output and tissue oxygen consumption. In tbe absenceof an adequate chronotropic response during exercise, SyO2 will decrease and the extent of desaturationmay be used as a parameter for rate adaptive cardiac pacing. Eight patients with sinoatrial disease receiveda DDDR pacemaker capable of DDDR pacing by sensing either S^O2 or piezoelectric detected body move-ment. Both sensors were programmed to attain a rate of about 100 beats/min during walking, and withthe lower and upper rates set at 50% and 90% of age predicted maximum, respectively. Chronotropicbehavior of the two sensors were compared in tbe DDD mode with measurement of sensor responses,during everyday activities (walking, stair climbing, postural changes, and physiological stresses) and ateach quartile of workload during a continuous treadmill exercise test. During walking at 2.5 mph, bothsensors showed no significant difference in delay time (both react within 15 sees) or half-time (S^O-i =36 ± 12 sec and activity 24 ± 3 sec; P = NS), although SVO2 driven pacing achieved 90% target rateresponse slower than activity sensing (124 ± 16secvs77 ± 10 sec; P < 0.02). S^O-i pacing was associatedwith a more physiological rate response during walking upslope (68 ± 12 beats/min vs 57 ± 10 beats/min; P < 0.05), ascending stairs (59 ± 10 beats/min vs 31 ±6 beats/min; P < 0.05), and standing (34± 7 beats/min vs 9 ± 2 beats/min; P < 0.05). The SvO2 sensor significantly overpaced in the first quartileof exercise (51.8 ± 25.6% in excess of heart rate expected from workload), but the rate was within 20%of expected for the remainder of exercise. "Underpacing" was observed with the activity sensor at thehigher workload. In conclusion, the SvO2 sensor demonstrated a more physiological response to activitiesof daily living compared with the activity sensor. Using a quantitative method, the speed of onset of rateresponse of tbe SyO2 sensor was comparable to activity sensing, and was more proportional in rate re-sponse. Significant overpacing occurs at the beginning of exercise during S^Oz driven pacing, which maybe improved with the use of a curvilinear algorithm. (PACE 1994; 17[Pt. I]:2236-2246)
central venous oxygen saturation, activity sensing, rate adaptive pacing, sensors
IntroductionAddress for reprints; Chu-Pak Lau, M.D., Reader in Cardiology, Physical activities result in an increase in car-Dept. of Medicine, University of Hong Kong, Queen Marv Hos- - • t 1 L 1 jpital. Hong Kong. Fax: (852)" 2855-1143. ' diac output and oxygen extraction from the blood.Received October 6. 1993; revision April 7, 1994; accepted If cardiac output does not match the requirementsApril 11.1994. of increased tissue oxygen consumption, a widen-
2236 December 1994, Part I PACE, Vol. 17
RATE ADAPTIVE PACING USING RV VENOUS OXYGEN SATURATION
ing of the tissue arteriovenous oxygen differenceoccurs. As a result, mixed venous oxygen satura-tion (SvO ) decreases and the extent of the desatu-ration depends on the level of activity and on thecardiac performance.^"' There is a marked de-crease in SyjO2 at the onset of exercise, with thevalue of SvO^ reaching a new platean within 60seconds. At the end of exercise, SvO^ returns rap-idly to the preexercise level. As there is no appre-ciable pool of oxygen in the hody, changes in SvO2rapidly reflects the metabolic need in response toexorcise.
The idea of using SvO2 for rate adaptive pacinghas been conceived for sometime, although techni-cal difficulties in achieving chronic stable SvO2sensing have limited extensive application of thisprinciple. As a result, only preliminary clinical ex-perience on implanted SvO^ sensors have heen re-ported." ' There is little data available that quan-tify the response of an SvOz sensor during physicalactivities. We report our clinical evaluation of thissensor in comparison with a standard activity sen-sor during daily activities and graded maximumexercise.
Oxygen Sensing Lead andPacemaker
Oxygen Sensing Lead
The DDDR pacemaker (model 8007. OxyElite,Medtronic, Inc., Minneapolis, MN, USA) utilizesa unipolar polyurethane steroid type ventricularlead for oxygen sensing (model 4327, Medtronic,
Inc.). At 25 mm from the ventricular pacing elec-trode, an oxygen sensor is incorporated that usesthe principle of reflectometry to detect SvO2 (Fig.1). The technical details of the SvO2 sensor hasbeen reported.^' In hrief, it consists of red (660 nm)and infrared (880 nm) light emitting diode, her-metically sealed in a sapphire capsule. The reflec-tance of the light from each is received and mea-sured by a photo-detector to generate a relative re-flectance ratio. Tbis ratio varies in proportion tochanges in right ventricular SvO2. The use of a re-flectance ratio instead of a single wavelength hasbeen shown to rednce the variation in O2 satura-tion determination as affected by fibrin coating,intrasystolic variation, velocity of blood flow, ven-tricular wall proximity, and changes in bemato-crit. Si,02 sampling is synchronized to an R waveonce every 4 seconds. The lead is connected tothe pacemaker via an IS-1 in-line "bipolar-like"(unipolar pacing and sensing and SvO2 param-eters) connection.
Dual Sensor DDDR Pacemaker
The pacemaker is based on an activity sensingDDDR pacemaker from the same manufacturer(model 7086, Elite II, Medtronic, Inc.). It can beprogrammed to provide rate response in either oneof the two ways: through the use of aright ventricu-lar SvO2 sensor, or through the use of an activitysensor. The pacemaker features both activity andSvOii telemetry (in real time), and it will also col-lect data for programmer display of an activity orSvO2 rate histogram or trend. On-line sensor deter-
Figure 1. Photograph of the oxygen sensor during S^Oj_ sampling.
PACE, Vol. 17 December 1994, Part I 2237
LAU, ET AL.
mined rate can be obtained by telemetry even inthe DDD mode with the sensor programmed pas-sive.
Algorithm for Rate Response
Activity Sensing
The algorithm of the activity sensing compo-nent of the dual sensor DDDR is identical to thesingle sensor activity DDDR pacemaker (Elite 11).In brief, activity signals measured from a piezoe-lectric crystal above a programmable threshold aretreated as counts, and the number of counts occur-ring per unit time is used to determine a rate re-sponse by means of a set of ten programmable lin-ear rate response curves.
SvO2 Sensing
Because patients have different pacing rateneeds as well as differing levels of SvO2 at rest, twoprogram settings are used to establish the optimalSvO2 driven rate response for each patient. The"resting rate offset" is used to establish the pointfrom which oxygen rate response can begin, aswell as the desired pacing rate at rest, and repre-sents the setting up of a "dynamic" lower rate (Fig.2). To ensure proper pacing for the patient, theresting pacing rate should be slightly ahove theprogrammed lower rate. An increase in the restingrate offset (1-31) results in an increase in the rest-ing rate (Fig. 2). The resting rate offset can bechanged manually or automatically. Thus pro-vided that the resting SvO2 driven pacing rate atthe SvO2 is above the programmed minimum rate,the lower rate can fall to the minimum value when-ever SvO2 increases as, for example, wben the met-abolic demand decreases during sleep.
Once the SvO2 exceeds the programmed rateresponse offset, rate response will he determinedby the programmed "oxygen rate response" (Fig.3). The oxygen rate response setting is chosen toestablish the relationship between changes in SyO^and changes in the pacing rate (beats/min per02%). The relationship is an inverse linear one;a decrease in right ventricular SvOz results in anincrease in pacing rate. One of 16 settings may bechosen. Setting 1 provides the smallest change inrate for a given change in SvOz saturation, whilesetting 16 provides the greatest change in rate.
The SvO2 sensor takes a sample every 4 see-
SvOi: Dynamic Lower Rate
80
70
60
50
40
30 L
tRRO =20
t
Resting
->• Dec SvO:
Figure 2. Schematic representation of tbe setting up ofa "dynamic" lower rate of an oxygen sensing DDDBpacemaker. A resting rate offset (HRO) defines tbe SyOzat which a rate response occurs. Wben tbe RHO is setbelow tbe resting SyO2 (RRO — 15), the latter results ina pacing rate at rest of 55 hpm despite a programmedminimum rate of 50 bpm since tbe resting S^,O2 is lowerthan the value defined by RRO ^ 15 (point A). If tbeRRO is increased to 20, rate response will start at ahigher S^O^ level despite an identical rate responseslope programmed, resulting in a higher rate at tbe rest-ing SyO2 (point B, 65 bpm). In eitber case, as tbe restingrate is above the programmed minimum rate, tbe lowerrate can decrease to this level wben tbe metabolic de-mand decreases. SvO2 — mixed venous oxygen satura-tion.
onds, synchronous with a ventricular pace orsense event, and the telemetered SvO2 value isheld until the next sensor sample. An internal av-eraging algorithm using a 10-second moving aver-age is operative to avoid fluctuation in SvO2 val-ues. The changes in SvO2 encountered were slowenough that rate acceleration/deceleration pro-grammed would probably not have been a factorin the SvO2 determined rate, as a rate accelerationlimit of 0.25 minute (see Methods) would allow achange for 50 beats/min to 90% target beart ratein 15 seconds.
Patients
Eight female patients with symptomatic sino-atrial disease received the dual sensor DDDR pace-
2238 December 1994, Part I PACE, Vol. 17
RATE ADAPTIVE PACING USING RV VENOUS OXYGEN SATURATION
190
170
150
I 130uS 110aia 90S.
70
50
30
RRWl
10 15 20
Figure 3. Oxygen rate response curves (1-16). The pac-ing rate changes as a function of the percentage ofS^Ozdecrease from the programmed rate response offsetlevel. RR — rate response curve (only representativecurves sbown); SyO2 = mixed venous oxygen satura-tion.
maker. All gave informed written consent to par-ticipate in this study, which was approved hy thelocal ethics committee. The clinical data and pace-maker settings are shown in Table I. All patientshad normal left ventricular systolic function as
assessed by transthoracic echocardiography. Pa-tient 6 had hypertension and left ventricular hy-pertrophy involving the septnm. Two patientshad atrial tachyarrhythmias rendering their atrialrate unreliable for assessment of chronotropic re-sponse.
Methods
Sensor Programming
Both SvO2 and activity sensors were pro-grammed to achieve a change in rate of 50 beats/min over the lower rate during a walking test ona treadmill at 2.5 mph. The programmed lower ratewas set at 50 beats/min and the upper rate at 90%of the age predicted maximum, i.e., 0.9 x (220 -age in years). A standard rate acceleration of 0.25minute was used for hoth sensors. The settings foreach sensor are summarized in Table I. To obviatethe effects of different chronotropic response onthe exercise performance, study was performed inthe DDD mode, with sensor determined rate col-lected once every 15 seconds by telemetry.
Exercise Protocol
The following protocol was performed in ran-domized order with the patient blinded to the sen-sor type programmed:
1. Brief treadmill exercise tests. This con-sisted of four exercise tests lasting for 3 minuteseach with the following treadmill settings: 1.2mph, 0%: 1.2 mph, 15%: 2.5 mph, 0%; and 2.5mph, 15%.
2. Going up and down four flights of stairs.3. Passive tilt from supine to 90° upright for
1 minute, and then from upright to supine for 1minute.
4. Active standing from supine position for 1minute, followed by lying down.
5. Valsalva maneuver performed to maintainan airway pressure of 30 mmHg for 15 seconds.Airway pressure was measured with an adaptedmercury sphygmomanometer.
6. Handgrip by the arm opposite to the sideof the pacemaker for 3 minutes at one-third of themaximum grip strength.
7. Hyperventilation for 1 minute.8. Mental stress as induced by serial subtrac-
tion of 7 from 100 for 2 minutes.
Maxima! Exercise
Graded treadmill exercise test in the DDDmode was performed twice in each patient, witbeither tbe sensor programmed to collect SvO^ oractivity determined rate. Tbe two tests were per-formed in random order and separated by at leastone hour apart. Tbe Cbronotropic Assessment Ex-ercise Protocol*' was used, wbich utilizes a grad-ual, linear increase in workload during exercise.
Data Analysis
The maximum changes in pacing rate duringeveryday activities for eacb sensor were used forcomparison.
To assess tbe kinetics of rate response of thetwo sensors, the speed of rate response was quanti-fied during walking at 2.5 mpb according to pub-lished methods." Tbe delay time (DT) refers to thetime for the pacemaker rate to increase hy at least5 beats/min from the onset of exercise. T and T90refer to the time required for the pacing rate toreach balf and 90% of rate response, respectively.
Proportionality of rate response is quantifiedduring maximum treadmill exercise.'" Botb themaximum rate and workload (in METS) are nor-
malized into 100%. A linear relationship betweenpercentage change in pacing rate and percentageworkload cbanges is assumed. Tbe expected heartrate percentage at the mid-point and end-point ofeacb quartile of exercise are 12.5, 25, 37.5,50,62.5,75, 87.5, and 100%, respectively. The actual pac-ing rate recorded are similarly normalized and ex-pressed as a percentage of the expected beart rate.
Wbere appropriate, paired comparisons weremade using the Student's paired ^test. Results areexpressed as mean ± 1 standard error of tbe mean.A P value < 0.05 between two variables is consid-ered to indicate a statistically significant differ-
ence.
Results
Tbe implantation of the ventricular oxygensensing lead was similar to that of an ordinary ven-tricular lead, altbougb positioning of tbe lead wasslightly more difficult as tbe stylet guide wirecould not reach tbe tip because of tbe sensor. Tbeimplantation was accomplisbed in all patients bymaking an acute bend in the distal part of tbe styletguidewire.
The SvOa sensor showed minimal response inone patient at 1 month after implantation (maxi-mum SvO^ change 2%), and tbis patient was ex-cluded from the exercise study. Anotber patientcould not perform maximum exercise test becauseof logistic reasons (patient 8), after completing tbeprotocol for everyday activities.
Tbe relative changes in telemetered derivedSvO2 for all activities are shown in Table IL
Everyday Activities
Figure 4 shows the rate cbanges during walk-ing and stair climbing. The SvO^ sensor showedan increase in rate when the patients walked at afaster speed or a higher gradient, and the pacingrate was bigber on ascending compared with de-scending stairs. On tbe otber band, tbe activitysensor responded to walking at a faster speed, andthe rate was paradoxically lower during ascendingstairs compared with descending stairs.
During passive head-up tilt test, tbe SyO idriven rate increased, and a reduction in rate wasobserved wbon tbo patients were returned to su-pine position (Fig. 5). A larger cbange, but direc-
tionally similar, was observed when the posturalchanges were actively performed. Tbe activity sen-sor showed minimal change dnring tilt tests, andthe pacing rate increased to a similar extent duringstanding and lying down.
The rate changes during otber pbysiologicalstresses such as the Valsalva maneuver, hand grip,hyperventilation, or mental stresses were small(Fig. 6). With tbe exception of hyperventilation,tbe rate changes were greater witb tbe SvO^ sensorcompared witb tbe activity sensor.
<| 30
20
2 5 mph 2.5 mpn Downslans Upslairs
1 I Ac l i v ' l v • B < O.OS. " P C 1 &HR 3( lower worUlDaa
Figure 4. Changes in beart rate (J HR) during walkingand stair climbing. The SyO2 driven rate was higberwben tbe patients walked at a faster speed (P < 0.01 for1.2 and 2.5 mpb) and at a higber gradient, and tbe beartrate was bigber on ascending .stairs. Tbe activity drivenrate only responded to an increase in speed of tbe tread-mill (P < 0.01 for 1.2 and 2.5 mpb) but not to an increasein gradient, and tbe pacing rate was slower on ascendingstairs.
Quantification of Rate Response
In the four patients in whom the sinns ratecould be evaluated during maximal exercise, themean heart rates in tbe SvO , and activity sensing
POSTURAL CHANGES
_ 15B% 10
Head-up Mil I TIHioaiptne | Slanding | Lying clown
Figure 5. Feasibility of S^O-^ and activity sensors to de-tect postural change. Tbe SyOz sbowed an increase inrate on upright posture (both passively or actively per-formed) compared with tbe rates during supine posture.An indiscriminate increase in rate was observed witbtbe activity sensor especially during active posturalchanges.
PACE. VoL 17 December 1994, Part I 2241
LAU, ET AL.
PHYSIOLOGICAL STRESSES
— 7
5^
Figure 6. Rate cbanges during pbysiological stresses.
DDD modes were 120 ± 12 beats/min and 118 ±10 beats/min, respectively (P - NS).
Speed of Response
Both sensors reacted promptly without asignificant delay time during walking, and thetime to achieve half of the response was similar(Fig. 7). The SvOa sensor showed a delay in reach-ing T90 of response compared with the activitysensor.
QUANTIFICATION OF SPEED |
140
120
100
60
60
40
20
0
T1/2 T90
Figure 7. Speed of response of activity and oxygen sen-sor during walking at a constant treadmill speed (2.5mpb). DT ^ delay time; Tl/2 = time to achieve 50%of target response; T90 ^ time to achieve 90% of targetresponse.
Proportionality of Response
The relation hetween the percentage of ratechange and percentage of workload during maxi-mum exercise for a typical patient is shown indi-vidually in Figure 8. Figure 9 shows the heart rateexpressed as a percentage of heart rate expectedfrom the workload for each quartile of exercise foreach sensor. SvO2 sensor significantly overpacedfor the first quartile of exercise, although the re-sponse was within 20% of expected heart rate forthe higher workload. As programmed, the activitysensor showed a response within 20% of expectedheart rate up to the middle range of exercise, butthereafter significantly underpaced comparedwith the workload.
Discussion
Main Findings
The chronotropic response of a SvO2 sensorwas compared to a conventional piezoelectric ac-tivity sensor during activities of everyday livingand maximum exercise. The SvO2 sensor showeda more proportional rate response during walkingand stair climbing, postural changes, and physio-logical stresses compared with activity sensing.Quantification of rate response showed an accept-able speed of rate response for both sensors, al-though the maximum rate tended to be attainedlater with the SvO^ sensor. Compared with theheart rate expected from the workload, the SvO2sensor overpaced at the lower workload, whereasthe activity sensor underpaced at the higher work-load.
Everyday Activities
Isotonic Exercise
Although maximal exercise is often used toassess the responses of sensors, few pacemaker re-cipients perform snch exercise during everydaylife. Thus an attempt was made in this study toevaluate these sensors through structured every-day activities. The results demonstrate a more pro-portional rate response of SvO2 sensor, with in-creased sensor rate on walking both at a fasterspeed and at a higher gradient. Oxygen extractionat the tissue increases as a result of a higher muscu-lar work, which reduces SvO2 proportionately,
2242 December 1994, Part I PACE, Vol. 17
RATE ADAPTIVE PACING USING RV VENOUS OXYGEN SATURATION
OxyElite Exercise Test; SvO2, Patient 903
0.75 -
0.50
0.25 -
O.GO I ' ——1\ -^
D.00 0.2S 0.50
Nofmoliied METS
1.00
l.C»
oxn 0 25 0.60 0.75
B
Figure 8. Typical rate responses during maximum exercise using the Cbronotropic AssessmentExercise Protocol in a patient witb OxyElite pacemaker, in either tbe S^Oz sensing (panel A) oractivity sensing mode (panel B). Botb tbe workload and beart rate are expressed as a percentageof the maximum. "Overpacing" and "underpacing" are identified if the resultant rate responseare above and below tbe line of identity, respectively. More variation in activity sensing deter-mined rate was observed during some levels of exercise, wbicb are shown as additional datapoints in tbe grapb.
PACE, Vol. 17 December 1994, Part I 2243
LAU, ET AL.
QUANTIFICATION OF PROPORTIONALITY \
Figure 9. Tbe mean beart rate response expressed as apercentage of the predicted beart rate during tbe midand end of each quartile of a progressive exercise test. Apositive value suggests overpacing and a negative valuesuggests underpacing from the ideal. The heart rate re-sponse was considered acceptable if it occurred witb20% of the expected heart rate. The S^O^ sensor signifi-cantly overpaced at tbe lower quartiles of exercise,wbereas tbe activity sensor significantly underpaced atthe bigher end of the exercise. Q1-Q4 refer to the firstto the fourth quartiles of exercise.
leading to a higher rate response. On the otherhand, the response of an activity sensor is depen-dent on the manner in which an exercise is per-formed rather than on the workload. ^ Hence, therate observed u'as higher when the patientswalked at a faster speed and remained unchangedwhen they walked at a higher gradient. For similarreasons, rate response was clearly more appropri-ate with SvO2 sensor during ascending and de-scending stairs compared with the activity sensor.
Sensor for Postural Detection
During passive head-up tilt followed by re-turning to the supine position, the SvO2 changedby -6.7 ± 1.2 and 7.1 ± 1.6, respectively. Thiswas likely due to a reduction in cardiac outputin the upright posture, resulting in an increasedoxygen extraction and fall in SvO2. SvO2 returnedto the resting level promptly when the patients re-sumed the supine posture. The change was largerin magnitude but directionally similar when thepostural change was effected voluntarily. On theother hand, minimal change in rate for the activity
sensor was observed during passive posturalchange, and the rate changes on standing and lyingdown were identical because the activity sensordetects only body movement associated with theposture changes rather than the posture itself. Areduction of SvOv by 9 ± 2% was also observedwhen subjects changed from supine to sitting posi-tion in a recent study.^^
A recent study^^ has addressed the usefulnessof rate adaptation in the prevention of postural hy-potension. This report suggests that in patientswith a more pronounced decrease in blood pres-sure on standing {> 10 mmHg reduction), an ap-propriate rate adaptation may reduce the severityof a hypotensive episode. In a case report,^'* anincrease in rate achieved by a preejection intervalsensor ameliorated the consequence of a patientwith severe orthostatic hypotension in a tilt tabletest. It is also possible that the ability of a sensorto detect posture may be clinically significant innot only avoiding postural hypotension but alsoas a sensor to detect and intervene the onset ofmalignant vasovagal syndrome. Few existing sen-sors for pacing have this capability and some sen-sors, such as the detection of the ventricular depo-larization gradient, responded paradoxically topostural change.'•''
Physiological Stresses
Compared with the activity sensor, SvOz sen-sor demonstrated a higher rate response to a vari-ety of physiological stresses. Although the physio-logical basis of such responses remains to be stud-ied, the SvO2 sensor showed a higher level ofsensitivity to nonexercise related physiologicalchanges.
Quantification of Rate Adaptation
The ideal characteristics of a sensor includean appropriate speed of response, proportionalityto workload, sensitivity to physiological changes,and specificity of rate response.^^ The everydayactivity protocol has addressed, semi-quantita-tively, the proportionality and sensitivity of theSvO2 and activity sensors. Using published meth-ods,^"' the speed and portionality were also quan-tified in this study.
2244 December 1994, Part I PACE, Vol. 17
RATE ADAPTIVE PACING USING RV VENOUS OXYGEN SATURATION
Speed of Onset of Rate Response
During a steady walking exercise, it has beenreported that a normal individual showed no sig-nificant DT, and a Ti and T90 of about 30 secondsand 60-90 seconds, respectively.^ This behaviorwas closely simulated by the activity sensor,whereas a slower speed of response was observedwith other sensors such as the sensing of minuteventilation, respiratory rate, QT interval, and thedetection of first derivative of right ventricularpressure. We have shown that the SvO2 sensor gavean appropriate speed of response during everydayactivities, although the time to attain tbe maxi-mum rate was somewhat slower compared withthe activity sensor. However, the maximum ratewas reached within the 3 minutes of the walkingexercise. The results of tbis test confirm tbe previ-ous observation by Wirtzfeld et al.,'' who showedthat a marked fall of SvO2 occurred at the onsetof exercise, reaching a plateau within a period ofapproximately 60 seconds. Averaging of SvO2 andactivity signals are operative in the present device,which might have influenced the speed of re-sponse. Thus the speed of response is referable tothe current implementation of the sensors.
Prop ortionality
By assuming a linear rate response to all levelsof workload, an expected heart rate can be com-puted for each quartile of a gradual exercise testwith linear increase in work rate.^" This enablesan "ideal" rate to be computed for an individualpatient. Using this theoretical assumption and a±20% acceptable limit, it was shown tbat SvO2sensor significantly "overpaced" at the start of ex-ercise. This behavior is probably contributed to thebehavior of SvO^ to workload, in whicb tbe de-crease in SvO2 at the start of exercise is muchhigher compared to changes during the higherwork levels in a progressive exercise.^^ The reasonfor the abrupt decrease in SvO^ at exercise onsetis unclear, but is presumed to be due to a surge ofvenous blood from the periphery as skeletal mus-cles contract, as well as the increasing oxygen ex-traction at the onset of exercise.^" A linear aigo-rithm was used in the present SvO2 driven system.An exponential relationship to reduce this "over-pacing" behavior at the onset of exercise would bemore appropriate in future versions of this device.
An acceptable rate response was observed
witb the activity sensor up to mid-third quartileof exercise. Because of failure of further increasesin rate with progression of exercise workload, theactivity sensor "underpaced" at the higher worklevels. We have not attempted to readjust the rateadaptive setting of the activity sensor, and argu-ably the upper rate response could be improvedby reducing the threshold for detection of activity.However, most clinicians and investigators havenow used a simple walking test to optimize therate response of rate adaptive pacemakers, as wasused in this study.^^ In addition, the method ofprogramming was identical for both sensors, ren-dering the rate response reasonably comparable.
Limitations of Study
Using a sensor that requires a specialized lead,the long-term stability of this sensor in humansremains a crucial issue for chronic SvO2 pacing. Ina mean follow-up of 35 months (16-48) of a similarSvO2 sensor, Faerestrand et al. ^ reported 1 of 14patients bad sensor-driven tachycardia at rest,which required removal of the device within 2weeks of implantation. The remaining deviceswere functioning, and invasive SvO2 sampling hadconfirmed a satisfactory SvO2 sensor response.Battery depletion was observed in six patientsafter 33-47 months. In animal studies,^• '* long-term stability of this and another SvO2 sensor over2 years were reported. We have encountered anearly sensor failure in one patient. The reason forthis failure was not clear. This patient had grossventricular hypertrophy on echocardiography,and it was possible for the sensor to be buriedwithin the myocardial mass between the thick-ened septum and right ventricular wall and thusexcluded from contact with the blood pool, result-ing in failure of response. In the remaining pa-tients, the sensor response was adequate withinthe follow-up period during repeated stresstesting.
The percentage changes in SvO2 during var-ious activities were only relative, as the SvO2 sen-sor was not calibrated according to in vivo mea-surement in this study. Such calibration would benecessary if SvO2 is used to reflect cardiovascularhemodynamics in addition to being used as a sen-sor for rate response. We have used a sensor pas-sive mode in this study, as most patients were notpacemaker dependent during the tests. While bav-
PACE, Vol. 17 December 1994, Part I 2245
LAU, ET AL,
ing the limitations in that the system was not freerunning, the actual rate responses between SvO2and activity sampling tests were identical, makingthe comparison unbiased. As reported in an earlierstudy,^^ exercise induced reduction in SvO^ mighthave been accentuated in the DDD mode becauseof chronotropic incompetence compared to DDDRpacing.
Clinical Implications
The SvO2 sensor showed a physiological rateresponse to everyday activities, with an acceptablespeed of response. The sensor may have the poten-tial of postural detection, both for the managementof postural hypotension and malignant vasovagalsyndrome. The use of a nonlinear SvO2 rate algo-
References
1. Wirtzfeld W, Goedl-Meinen L, Bock T, et aL Gen-tral venous oxygen saturation for the control of au-tomatic rate-responsive pacing. PAGE 1982- 5-829-835.
2. Wirtzfeld A. Heinze R, Liess HD, et al. An activeoptical sensor for monitoring mixed venous oxy-gen-saturation for an impiantabie rate- regulatingpacing system. PACE 1983; 6:494-497.
3. Stangl K, Wirtzfeld A, Gobe G. et al. Rate controlwith an external SO2 closed loop system. PAGE198B; 9:992-996.
4. Stangl K, Wirtzfeld A, Heinze R, et al. First clinicalexperience with an oxygen saturation controlledpacemaker in man. PAGE 1988; 11:1882-1887.
5. Faerestrand S, Ohm 0-J. Long-term follow-up of arate-variable pacemaker controlled by central ve-nous oxygen saturation, (abstract) J Am Goll Gar-diol 1991: 17:289A.
6. Seifert GP, Moore AA, Graves KL, et al. In vivoand in vitro studies of a chronic oxygen saturationsensor. PAGE 1991; 14:1514-1527.
8. Wilkoff BL, Govey J, Blackburn G. A mathematicalmodel of the cardiac chronotropic response to ex-ercise. J Electrophysiol 1989; 3:176-180.
9. Lau GP, Butrous GS, Ward DE, et al. Gomparativeassessment of exercise performance of six differentrate adaptive right ventricular cardiac pacemakers.Am J Gardiol 1989; 63:833-839.
10. Kay N. Quantification of chronotropic response:Gomparison of methods for rate-modulating per-manent pacemakers. J Am Goll Gardiol 1992; 20-1533-1541.
n . Lau GP, Mehta D, Toff W, et al. Limitations of rateresponse of activity sensing rate responsive pacing
rithm. calibration of sensor to in vivo SvOa leveland demonstration of long-term reliability repre-sent future challenges of the use of long-term SvO2sensor for rate adaptation and hemodynamic mon-itoring.
Conclusion
The sensing of SvO2 was shown to exhibitphysiological rate response to exercise and nonex-ercise related physiological changes. Quantifica-tion of rate adaptation suggested that the SvO2 sen-sor had a nearly comparable speed of response toan activity sensor, with a better proportionality ofrate response during everyday activities and pos-tural stresses.
to different forms of activity. PAGE 1988; 11-141-150.
12. Faerestrand S, Ohm OJ, Stangleland L. Glinicalperformance of a central venous oxygen saturationsensor for rate adaptive pacing, (abstract) PAGE1993; 16:894.
13. Leitch JW, Arnold JM, Klein GJ, et al. Should aVVIR pacemaker increase the rate on standing?PAGE 1992; l.=):288-294.
14. Gruhh BP. Wolfe DA, Samoil D, et al. Adaptive ratepacing controlled hy right ventricular preejectioninterval for severe orthostatic hypotension. PAGE1993; 16:801-805.
15. Paul V, Garrett G, Ward DE, et al. Glosed loop con-trol of rate adaptive pacing: Glinical assessment ofa system analyzing the ventricular depolarizationgradient. PAGE 1989; 12:1896-1902.
16. Lau GP. The range of sensors and algorithms nsedin rate adaptive cardiac pacing. PAGE 1992; 15"334-343.
17. Wirtzfeld A, Heinz R, Stangl K, et al. Regulation ofpacing rate hy variations of mixed venous oxygensaturation. PAGE 1984; 7:1257-1262.
18. McElroy PA, Janicki JS, Wcjber KT. Physiologic cor-relates of the heart rate response to upright isotonicexercise: Relevance to rate-responsive pacemak-ers. J Am Goll Gardiol 19H8; 11:94-99.
19. Lau GP. Rate Adaptive Gardiac Pacing: Single andDual Ghamhers. Futura Publishing Go., Inc.,Mount Kisco, NY, 1993, p. 415.
20. Snell J, Gohen D. Hedberg SE. In vivo performanceof a hemo-reflective type oxygen sensor for rateresponsive pacing, (abstract) PAGE 1988; 11:504.
21. Faerestrand S, Ohm OJ, Gentral venous oxygen sat-uration at rest and exercise during hradycardia andrate responsive pacing, (abstract) PAGE 1990; 13:529.
