Variability of oxygen affinity of normal an automated method of measurement
blood:
ROBERT M. WINSLOW, JANICE M. MORRISSEY, ROBERT L. BERGER, PAUL D. SMITH, AND CARTER C. GIBSON Clinical Hematulugy Branch, Laboratory of Technical Development, National Heart, Lung and Blood Institute, and Biomedical Engineering and Instrumentation Branch, Division of Research Services, Nutionul Institutes of Health, Bethesda, Maryland 20014
WINSLOW, ROBERT M., JANICE M, MORRISSEY, ROBERT L. BERGER, PAUL Il. SMITH,AND CARTERC. GIBSON. Variability of oxygen affinity of vzormal blood: an automated method of measurement. J. Appl. Physiol. : Respirat Environ. Exercise Physiol, 45(Z): 289-297, 1978. -Oxygen equilibrium curves of 48 healthy adult subjects have been measured by the method of Rossi-Bernardi et al. (Chin. Chem.. 21: 1747, 1975), in which I&O, is gradually added to a sample of deoxygenated blood that contains an excess of catalase. The mean P,, for nonsmokers was 26.9 Torr and the distribution of values was slightly skewed to the right (range 24.2-29.9 Torr). The method differs from others previously available in that pH, Pcoz, and HCO,are constant during oxygenation. The system for control of the experiment and data collection and processing has been automated by the use of a microprocessor so that the equilibrium curve can be obtained quickly, reproducibly, and relatively simply. With the aid of a digital computer, the parameters of the generalized Adair equation can also be estimated.
hemoglobin; processor
oxygen dissociation
curve; Adair
equation;
micro-
METHODS FOR THE MEASUREMENT of the hemoglobin oxygen equilibrium curve (OEC) have evolved recently allowing the systematic investigation of properties of purified hem.oglobin (8). However, analogous methods for measuring the OEC of whole blood have been slow to appear because the red blood cell is far more complex than purified hemoglobin. Blood OEC’s obtained by van Slyke manometric analysis have served as standards for clinicians, physiologists, and biochemists for many years. For example a “standard curve” (16) was obtained using such analysis on samples of blood from several different subjects and were assumed to have the same oxygen half-saturation pressure of hemoglobin (P& and Bohr effect. Several recent methods have been introduced for the continuous measurement of the whole blood OEC by use of small samples from single individuals. All of the advantages an.d disadvantages of these various methods will not be discussed in detail here. Rossi-Bernardi et al. (14) described a method that employs the gradual oxygenation of a sample of completely deoxygenated blood by the controlled addition of H,O, in the presence of catalase. The accompanying released hydrogen ions are titrated with NaOH. This
method has the inherent advantage that fresh whole blood can be used, the entire curve can be measured in as little as 30 min from venipuncture, measurements of carboxyhemoglobin (HbCO), methemoglobin (MetHb), and 2,3-diphosphoglycerate (2,3-DPG) can be made on the sample before and after the run, and carbon dioxide partial pressure (Pcoz) and pH are constant over the entire oxygenation range. No other system that we are aware of combines these advantages, Recently we have described a normal whole blood OEC obtained with the H202 method with extensions at the ext.reme ends of the curve by a simple mixing method (20). Although this technique is relatively simple, the mixing experiments require rearrangement of the experimental apparatus and additional time. The results of those experiments were analyzed according to the Adair stepwise oxygenation scheme, an analysis that is independent of stereochemical models of hemoglobin function. We report experiments in which improvement of the H,O, method has been achieved by automation of the data acquisition and computation system. Adair parameters, which fit the experimental dat.a very well and are therefore useful in simulating the OEC for various physiological computations, can be estimated by this method. METHODS
Apparatus The cuvette is a modification of that described by Rossi-Bernardi et al. (14). It consists of a Lucite chamber containing a water bath that is kept at 37°C. Paz and Pco2 electrodes (Instrumentation Laboratory model 17026 and 16050, respectively) are mounted in the apparatus directly in contact with the blood sample. Stirring is achieved by a small magnetic mixer and a motor mounted beneath the chamber. H,O, (0.4 M) and NaOH (0.4 M) are added to the sample from 50-~1 Hamilton syringes through polyethylene tubing. The syringes are driven by motorized pumps (Raze11 Scientific Instruments, Stamford, Conn.) under computer control. The cuvette was modified by the addition of a thermistor probe (Thermonetics), which protrudes slightly into the chamber. 289
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
290
WINSLOW
Calculations. The symbols used in calculating the oxygenation data are defined in Table 1. Hemoglobin saturation. The fractional saturation of Hb with oxygen is given by
of oxygen bound at a given point is
The amount
of oxygen dissolved
The initial
(4) of an
the addition
0 i = (Oi-1 + AO) x [I - AVl(V
+ AV)]
(5)
is
A0 = H x R x At x ; x lo-”
Symbol
Bi G CO
A0 At AV Q Do E cal Ei
F H Hb HbCO MetHb NaOH Oi Ql
P cal pi
R T cal T”K Ti V VB yi
xl
used in oxygenation
(6)
equations
Quantity
Coefficient of solubility of oxygen in blood Oxygen bound to Hb at ith point Oxygen capacity at ith point Initial oxygen capacity Oxygen content increment Time interval for a sampling Total volume increment in a sample interval Oxygen dissolved in blood at ith point Initial dissolved oxygen Oxygen electrode voltage with standard gas Oxygen electrode voltage at ith point Electrode factor H@, concentration in syringe Hemoglobin concentration Carboxyhemoglobin concentration Methemoglobin concentration NaOH concentration Oxygen content at ith point Initial oxygen content Paz of standard gas Pop at ith point Rate of H,02 addition Temperature of cuvette with standard gas Absolute temperature Temperature of blood at ith point Cuvette volume Volume of NaUH Hemoglobin saturation at ith point Initial hemoglobin saturation
Units
pmol/mm
4 pm01 pm01 pm01 pm01 S
pm01 pm01 V V Dimensionless a PM, heme % % PM pm01 pm01 Torr Torr plls
OC “K “C 4 4 Fraction Fraction
x
Pi = where
(8)
is
pressure
(3)
T”K
after
O2 capacity
oxygen partial
273
AV = VB + R x At
a
is not zero, the oxygen is
C 0 = Hb x [l - (HbCO
Since the cuvette volume is fixed, when each increment of H,O, and NaOH is added, an equal volume of mixed blood is expelled. This requires the correction of the 0, content by a dilution factor. The volume added to the cuvette during each sample interval is
1. Symbols
(7)
0 0 = (Y, x co> + Do
in the blood is
D i =pixvxa-
TABLE
-t- AV)]
(1) content at the start of the experiment
The amount
The 0, increment
Hb saturation
AL.
corrected . for dilution
of the system,
Ci = Ci-1 x [l - AV/(V When the initial
Bi Yi = Ca
Therefore, the 0, content in .crement of oxygen (AO) is
The oxygen capacity is
ET
xv
+ MetHb)]
(9)
(Po,) is given by Ei
F, the “electrode
x
F
x
factor,”
(10)
PcallEcal
is defined below.
Curve Fitting 1
l
Estimation of the parameters of the Aaaw oxygenation equation was done as previously described (20) afier transfer of data to the PDP-10 digital computer. For the final fit in the procedure, a scaling factor was used as a multiplier for all saturation values. This factor was included as a parameter of the fit and reduced somewhat the final sum of squared residuals. Its use rests on the assumption that the greatest errors in the data are in saturation, not Paz, and its value was always within the range 0.98-1.02. Titration
of Pco2
As blood is oxygenated the released protons cause a decrease in pH and an increase in Pcoz. The protons are titrated by the addition of NaOH under computer control. Sufficient NaOH is added to maintain Pcoz constant. Preliminary experiments revealed that approximately 2 pmol NaOH were required per pmol of hemoglobin tetramer during oxygenation. Therefore the total NaOH added was first calculated from the quantity of hemoglobin present 2xHbxV total NaOH = (11) ccl NaOH The total time of oxygenation depends upon the amount of hemoglobin present and the rate of addition of H,02 total time = l/2
The estimated then becomes
rate of NaOH
K1 This is only adjusted by tional to the ith point and
Hb x V seconds x R x H addition
from the syringe
RxH =
NaOH
(12)
PUS
(13)
an approximation, and the rate is further inclusion of a term (Q), which is propordifference between the Pcoz voltage at the the initial value NaOH
rate = K, + (Kz x Q)
K, is an empirical constant+ To damp the fluctuation in Pcoz, a third term, proportional to the rate of change of Pcoz voltage, is added
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY
OF
BLOOD
OXYGEN
NaOHrate=K,+(&xQ)+(K,x$) where K, is also empirically
291
AFFINITY
(15) determined.
Data Collection and Analysis The data collection and analysis system is controlled by an Intel 8080 microprocessor. It receives inputs from the experiment, controls the NaOH and H20, pumps and outputs results to a self-scan display, a plotter (Zeta Research, Inc.) and flexible disc (model 4821, Tektronix, Inc.). The program (Real Time Systems, Fairfax, Va.) is written in assembly language and uses 14 K of readonly memory (ROM) and 17 K of random accessmemory (RAM) for storage of data. A block diagram of the hardware is shown in Fig. 1. The flow of an experiment is given in Fig. 2. At user-specified intervals the PO, and PCO~electrodes are sampled and Paz and saturation are calculated (EQS. l-1 0). The H,O, pump is driven at Calculate a constant rate, and the NaOH pump is driven at intervals for a time which depends on the Pcoz (EQ, 15). Display Paz and Hb saturation are plotted in real time and at (self scan) the end of the run the required constants, Po2, and NaOH volume for each point are transmitted to the [Aatpointj flexible disc. A set of data can be recalled to memory FIG. 2. Experiment flow diagram for the automatic OEC apparafrom the disc for off-line calculation and plotting. A set tus. In addition to flow shown, a data set can be recalled to memory of data can also be transmitted to the PDP-10 digital from flexible disc unit for recalculation and off-line plotting. computer for further analysis as desired. This step is controlled by a separate Fortran program on the PDP- veins of normal volunteers and immediately placed into 10. EDTA anticoagulant. Determinations of Hb, HbCO, and MetHb were done on IO-~1 samples using the Blood Samples and Reagents Microblood Analyzer (kindly loaned by Carlo Erba, Whole blood samples were obtained from antecubital Strumentazione, Milan) (15). 2,3-DPG concentrations were measured by the method of Nygaard and Rsrth 100Ins (12) using kits from Calbiochem. All reagents were Clock reagent grade. Three percent H,O, (USP) was obtained from Parke Davis. All gas mixtures were obtained from 32 CHARD&P. 8080A c Lif-o-Gen, Cambridge, Md. Their compositions were CPU certified to a nominal -t- 0.1%.
I-
Key Board r
Experimental
Mode Sel ’
POWER SUPPlIES fl FIG. 1. Hardware block diagram for automatic OEC apparatus. Abbreviations are: mode sel, mode selector; DPM, digital panel meter; A/D, analog-to-digital converter; char disp, character display; CPU, central processing unit,
Procedure
With the stirrer on, calibrating gas was passed into the cuvette after humidification at 37”C, and the temperature in the cuvette during this period was carefully noted (T,& The 0, and CO, electrodes were calibrated at two settings, 0 and 12% 02, and 4 and 10% Cob,. A small amount of catalase (Calbiochem) was added tithe cuvette before the run to assure complete conversion of H,O, to molecular oxygen without significant oxidation of hemoglobin. With normal blood, the catalase can be omitted without effect on the shape or position of the OEC. A 2-ml sample of blood was deoxygenated by equilibration with 5.6% CO, (balance N,) in an IL model 237 tonometer for 15-20 min. Then by use of a gas-tight Hamilton syringe the sample was transferred anaerobically into the reaction cuvette, which contained 10 ~1 of catalase. When the temperature was stable (T;) a small aliquot of the blood was removed for measurement of pH, Hb, HbCO, and MetHb. The tubes from the NaOH and H,O, pumps were placed into the cuvette
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
292 and ment from lO+l ment
WINSLOW
after entering the required constants the experiwas begun by giving an appropriate command the keyboard. At the end of the oxygenation run a sample of the blood was removed for the measureof HbCO and MetHb concentrations.
RESULTS
Temperature Strict control of temperature during the experiment is very important: in addition to its effect on the position of the OEC, the output from the oxygen electrode is strongly temperature dependent. Figure 3 describes an experiment in which the temperature of the cuvette was gradually increased while exposing the electrode to 5% O2 at a constant flow rate. A linear relationship was found with the equation E Cal = Teal X (0.018) + 0.024
(16)
where Ecal is the amplifier output of the 0, electrode in volts, and Teal the temperature at which it is measured. The slope of the line can be used to determine the PO, voltage (Ei) at the temperature of the blood during the experiment (Ti). A difference between Tear and Ti of about 2°C was commonly observed, even though the gas was bubbled through water at 37°C. The error in PO,
ET
AL.
introduced from this source would be 3.6% for a 2°C temperature difference. In practice, the electrode factor (see below) corrects for this discrepancy. If the temperature of the blood during the oxygenation run is not constant the result is virtually uninterpretable because in addition to the calibration error the position of the OEC itself will be altered. The temperature of the blood was monitored during each run and was always constant, Electrode
Factor
The “electrode factor” is the ratio of output observed from the oxygen electrode with blood in the cuvette to that with calibration gas (5). This value, F (Ea. IO), is normally less than 1 because of the formation of a thin layer of stagnant blood at the surface of the PO:! electrode membrane. Therefore, it is not surprising that its actual value would vary with the rate of stirring, geometry of the cuvette, and the rheological properties of the blood itself, In the experiments with normal blood described here, hematocrit values were distributed over a narrow range and the stirring rate was the same for all runs. The electrode factor was checked prior to a set of experiments (i.e*, each working day) by introducing into the cuvette blood that had been equilibrated with gas of the same composition. The actual value of F varied between 0.95 and 1.00 but was invariant over a given day and with a given 0, electrode membrane. It must be noted that F varies with the rate of stirring, the gas-liquid difference, and the difference between the temperature in the cuvette during calibration and when filled with blood. The Data
TEMPERATURE,degree C
3. Calibration of cuvette temperature. With 5% O2 flowing into cuvette, temperature of water bath was gradually increased. Relation between temperature and voltage output of the O2 amplifier was linear, Regression line had slope 0.018 V/“C and intercept 0.024 FIG.
During the experiment PO, and Pco2 are recorded directly on a strip recorder (Fig. 4). This recording, though not essential, is useful in monitoring the NaOH titration. Concomitantly the computed curve is plotted in real time (Fig. 5). The Pcoz remains constant in a typical run to approximately 20.3 Torr. Figure 4 is a direct tracing of a typical experiment and Fig. 5 is a photograph of an on-line plot. Since several independently measured values are used in the calculation of saturation (see Eqs. I-10) it is not unusual to observe at 150 Torr a final saturation
FIG. 4. Direct tracing of the Paz and Pco2 during an experiment. Short line at top represents a change in x-axis sensitivity so that full scale is 200 Torr. H,Oz was started as indicated by arrow on ordinate.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY
OF 100
BLOOD
OXYGEN
293
AFFINITY
-
FILE
NFIHE
1207
SUBJECT UNIT
Flu
ORTE OEQXY
r
T
r
r
T
I
I
I
I
plot of calculated
data.
7.414
PC02
39.13
TEtIPERRTURE P50
026.10
I
5. On-line
I-JUtiPH
I
that varies by -+2% Therefore a small aliquot of the sample is removed it the end of the run to measure saturation independently with the microblood analyzer. This value can then be used to normalize all of the saturation values by multiplication by the ratio of saturation expected to saturation measured. A second method of estimating the final saturation employs the estimation of a scaling factor from the best fit of the data to the Adair equation. However, this cannot be done by the microprocessor at present. NaOH
FIG.
NUIIBER
37.0
4
1
I
1
2. Effect of NaOH addition on change in pH and Pm2 during oxygenation TABLE
PH
PC%, Torr
No titration Beginning End
7.458 7.337
36 46
With titration Beginning End
7,364 7.362
40 40
Additiun
During oxygenation of blood in the reaction cuvette, the release of protons and decrease of pH will be accompanied by a rise in CO, (Table 2). The addition of NaOH ensures that pH and Pcop are held constant to within the limits of the errors of measurement. Figure 6 is the amount of NaOH added as a function of the total, plotted against hemoglobin saturation. As has been observed by others for pure hemoglobin (19) this is not a linear relationship: relatively more NaOH is required at the early stages of saturation. This could have been a result of the slow response time of CO, electrode (54 s, 90% response). We therefore performed an oxygenation experiment in which the H,O, pump was stopped at l-min intervals until the Pco2 had stabilized. This procedure gave a result identical with the continuous method (Fig. 6). An important advantage of this method over the system described by Duvelleroy et al. (3) is that no correction need be applied for the nonuniformity of the protons released during oxygenation, since the Bohr protons are titrated very closely during the run, Thus to interpret the full OEC measured by the latter method one needs to know the quantitative magnitude of the Bohr effect as a function of saturation in order to correct the data to uniform pH (6). In contrast our method can be used to study the Bohr effect. Adair’s Equation Figure 7 shows the fit of a set of data obtained by the automatic method to the Adair stepwise oxygenation formula (1)
I .2
1
.4 HEMOGLOBIN
I
I
1
.6 48 SATURATION
FIG. 6. Nonlinearity of NaOH addition with increasing hemoglobin saturation. Rate of NaOH addition diminishes as saturation increases. Automatic (0) and manual (e) titrations, see text for details, Degree of departure from linearity depends on experimental conditions (especially pH, unpublished observation),
Y
a,P + 2a2P2 + 3a,P3 + 4aqP4 = 4(1 + alP + a2P2 + a,P3 + a4P4)
(17)
The curve-fitting routine, which was described previously, was applied to data sets using the PDP-10 digital computer (20). Since the errors in the calculation are greater in saturation than Po2, the routine was modified slightly to incorporate a scaling factor to be used with the saturation values to obtain the least value of the residual sum of squares. By this method one assumes that the data can be described by the Adair scheme and that a departure of the data from the fit can be accounted for by errors in the concentration of H2021 Hb, HbCO, and MetHb concentration.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
WINSL0W
ET
AL.
FIG. 7. Fit of a typical data set to Adair equation (Eq. 17). For c1arity, every 10th experimental point is plotted with curve described by Adair simulation (continuous Line).
75 PO2
TABLE 3. Errors* in Adair parameters for present method and extended continuous method (Ref- 20) Continuous a,
x
a2 x
0.157 -36 to +65%
10-l
lo-” -45
u3
x 1o-5 -100
u4 x 1o-5 * 95% confidence
0.117 to +ll% 0 to +lOO%
0.245 -6 to +7%
I
I
CONTINUOUS
I
I
EXTENDED
Extended
-29
0.151 to +38%
-28
0.097 to -17%
-100
I1 :.: :I I .: ‘, 1\ ;i I \ iir \ iir \\ :‘I:ir \\’iit1 .-.---+~..3.-.-*-.-.-. I
0,170 to +lOO%
0.167 -5 to +5%
limits.
Analysis of the errors in the Adair parameters (Table 3) is shown in Fig. 8. This analysis, as described previously (20) was done by varying the value of one of the a’s systematically, while determining the value of the sum of squared residuals after fitting the other three parameters to the data. The resultant errors are significantly larger than those determined previously be extending the OEC at its extremes (20)+ Furthermore, inspection of Fig. 8 indicates that the error distributions are not symmetrical. Therefore, the standard errors given by the application of a linear curvefitting routine to such nonlinear data could be extremely misleading.
I -50
I 0
I 50
1 -50
.---I
0
I
50
% ERROR IN a
FPG, 8. Analysis of the errors in Adair parameters. Value of each of parameters (see Eq. 17) was systematically varied during determination of residual sum of squares after fitting the other three to the data. (-, al; - - -, a2; - -, a3; . . . , a+) Analysis for a single curve (continuous) was compared to a curve which had been extended at its extremes (extended, see Ref. 20) Horizontul lines repres,ent 95% confidence limits (f test). In both instances, 60 experimental points were used for analysis, l
TABLE File
1204 1205 1206 1207 1208
4. Reproducibility PH
7.412 7.412 7.418 7.414 7.421
l
of continuous a, x 10L2
OEC
$$
nmax*
a, X 10-l
26.5 26.8 26.5 26.4 26.2
2.68 2-70 2.70 2.73 2.69
0.142 0.123 0.144 0,159 0.103
0.126 0.121 0,123 0.116 0.131
0.241 0.227 0.242 0.247 0.241
0.134
0.123
0.240
a3
a4 X 10v5
Reproducibility Mean
Table 4 presents the results of five consecutive measurements on blood from a single venipuncture. These determinations were completed within 3 h and the blood sample was kept on ice during this period. Survey
of Normal
OEC’s
Forty-eight normal individuals of both sexes, both smokers and nonsmokers but of a narrow age distribu-
* Maximum
26.5 slope
2.70 of the Hill
plot
(see Ref. 8).
tion, were used to study the variability of the normal PSO(Table 5). Figure 9 shows the results of these studies. Normal P,, ranged from 22.5 to 29.9 Torr and appeared to lie in a normal distribution but slightly skewed to the right. The mean PSOwas 26.5 Torr. This value is the same as the value which has been accepted as a stan-
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY TABLE
variables
OF
BLOOD
OXYGEN
295
AFFINITY
5. Statistical analysis of hematologic in nonsmoking subjects n
Range
Hb, g/100 ml
Male Female
Sex
15.2 & 1.2 13.7 2 0.8
23 15
12.9-18.1 11.6-15.3
2,3-DPG, mall mol Hb
Male Female
0.928 k 0.129 0,885 t 0.194
23 15
0.739-l-340 0.396-1.274
HbCO,
Male Female
2.3 * 0.8 2.7 & 1.2
23 15
1.2-4.0 1.0-4.5
Age, yr
Male Female
32.4 k 5.7 31.6 k 9.2
16 14
20-40 19-48
PSO, Torr
Male Female
26.7 2 1.7 27.2 k 1.2
23 15
24.2-29.9 25.7-29.6
%
Mean
5 SD
(for a 2°C difference between Teal and Ti). In all of our experiments, temperature was continuously measured and did not vary by more than O.l”C. Furthermore, the electrode factor (F) could vary with the electrode used, the condition of its membrane, mixing, and blood viscosity. This could be the source of another 5% error. Therefore, monitoring the temperature of the cuvette and frequent checking of the electrode factor is of critical importance. Reproducibility of our data is excellent (Table 4) and we therefore believe that the distribution of P,, values shown in Fig. 9 is due to biological, not experimental, factors. Comparison
-
with Previous
“Standard
Curve”
A typical OEC is compared to the “standard” OEC of Severinghaus (17) in Table 6. The PO, values at 50% saturation are in very close agreement, but a discrepancy of as much as 1.7% occurs at high Paz* Table 5 also illustrates that our OEC can be represented quite well by the Adair formula (Eq. 17). The differences between our OEC and the Severinghaus standard curve must be considered in the light of the differences in the methods used. In the latter instance, data from several individuals, indeed from several investigators, were compiled to assemble an average OEC. We believe that biological differences between normal individuals (cf. Fig. 9, Table 6) can be detected with our method since reproducibility is good (Table 4), Therefore such averaging will obscure important differences between subjects. DISCUSSION
22
24
26
20
30
~50, mmHg FIG, 9. Blood oxygen affinity of 33 healthy adult subjects. Paz at half saturation (P,,) was corrected to whole blood pH of 7.4 assuming a linear Bohr factor (Alog PJA pH) of - 0.40). This value is based on unpublished data obtained with apparatus described here,
dard from previous work (17). Many of the subjects whose blood PsO lies on the left tail of the normal distribution were cigarette smokers and had increased HbCO. Correlation between PsO and the level of 2,3DPG was rather poor (r = 0.018). Likewise no significant differences could be detected between sexes when smokers and nonsmokers were examined separately. The wide range of normal values obtained in our studies was rather unexpected. Modulation of the whole blood oxygen affinity by 2,3-DPG and pH are well known, but the variation we observe cannot be accounted for by variations in these effecters. We assume that the structure of the hemoglobin molecule itself does not vary among normal individuals and therefore the factors responsible for the variation are either conditions that exist within the red blood cell or variations in the experimental conditions. An example of the latter would include fluctuation in temperature, not only of the whole blood sample during the run but of the cuvette during calibration of the oxygen electrode. This discrepancv could cause errors in PO, of as much as 3.6%
Other methods for the continuous recording of the OEC are currently available (3, 9, 10). Those using the optical measurement of saturation have several inherent disadvantages. First, since the optical absorbance of whole blood is very high some methods require that TABLE 6. Comparison of measured and computed saturation Pas, Torr
Standard Curve Saturation,* %
Measured Saturation,t %
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
3.39 9.61 19.13 32.22 46.00 57.60 66.72 74.43 80.47 85.12 88.25 91.00 93.00 94.12 95.13 95.86 96.40 96.86 97.21 97.50
3.02 9.24 19.82 32.42 45.86 58.82 67.70 75.55 80.48 84.81 87.60 89.66 91.21 92.54 93.53 94.28 94.94 95*45 95.91 96.31
* See Ref. 17. Table 3 (continuous
t Curve curve).
1207,
Table
Adair
Saturation,S %
3.23 9.43 19.41 32.44 46.15 58.33 68.05 75.38 80.78 84.76 87.72 89.94 91.64 92.96 94.00 94.83 95.51 96.00 95.52 96.90 4.
$ Constants
from
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
296
WINSLOW
cells be diluted into buffer before the experiment is run. This is therefore not desirable when physiologically relevant data are required. An exception to this is the method that employs a very thin layer of blood in a controlled atmosphere (Ref. 10, and the Hem-O-Scan, American Instrument Co.) Second, optical methods require the defmition of 100% saturation at some absorbance. This is usually done by exposing a sample to air at the end of the run. At ambient PO, (about 150 Torr) normal blood is 98% saturated (21) and pathological blood with reduced oxygen affinity, especially sicklecell anemia blood, can be even less saturated (unpublished observations). Finally optical methods cannot deal with changes in the absorbance spectrum which could occur by denaturation of hemoglobin during the experiment (2). The present system, being closed, does not have a liquid interface for equilibration. The fact that oxygen is added as H,Oz means that the time required for the experiment is limited only by mixing of the two liquids, H,O, and blood, and is therefore rapid. The major advantage of the closed system, however, is that pH, Pco~, and HCO,- can be held constant throughout the run and, as opposed to many other systems, no correction need be applied to the data for the Bohr effect. This is very important since the Bohr effect is not constant with saturation (19), The use of the microprocessor means that the cost of automation can be kept at a minimum and the reliability and simplicity of operation to a maximum. The personnel operating the equipment need not have any specialized training in computer technology. Mistakes are minimized and as many as lo-12 complete curves can be performed in a single working day. Immediate feedback of the OEC to the operator has the advantage that curves can be repeated when desired while blood samples are still fresh, As can be seen from Fig. 7 and Table 6, the fit of the data to the Adair equation is excellent. Therefore, the Adair parameters can be used to calculate saturation at any PO,, or to perform a variety of other calculations of physiological and physicochemical interest, For example, the arterial-venous oxygen difference calculated from these parameters is a true reflection of the OEC, whereas the OEC plotted according to the logarithmic form of the Hill equation (7) is not linear Y 1% 1y = n log P + k
(18)
over the entire range of oxygenation. Since the Hill equation does not fit the experimen tal data at its extremes, the parameter n, a measure ofs ‘ubuni t cooperativity, is not fixed but reaches a maximum with minima at high and low saturation (18). The maximum value of n, and its position with respect to the saturation, is of physicochemical interest and can be easily computed from the Adair parameters. Theoretically, it should be possible to calculate the
ET
AL.
values of the equilibrium constants for the four successive oxygenation steps of the Hb from the Adair parameters. This can be done within reasonable error for pure Hb solutions (18), but the errors are larger for whole blood (Fig. 8), even with extensions at the top and bottom of the OEC (20). Whether this large error is due to experimental inaccuracy (see Ref. 20 for a discussion) or to the fact that whole blood is an extremely heterogeneous system compared to hemoglobin solutions is not yet clear. The magnitude of these errors means that we cannot at present compute unique values of the equilibrium constants for the four oxygenation steps of whole blood, Calculations of oxygen delivery in peripheral tissues and uptake in the lung are based on a ctstandard” OEC. The overall regulation of oxygen transport in any living organism is extraordinarily complex because it is dependent not only on whole blood oxygen affinity and all the determinants thereof but also cardiac output, pulmonary function, regional blood flow, hemoglobin concentration, and tissue oxygen demand. A desirable goal of current research is the quantitation of the oxygen delivery system so that perturbations of various components of it can be systematically studied. It is therefore important that the oxygen equilibrium data obtained for such modeling should be of the highest precision and reflect physiological conditions. Our results indicate that a wide range of normal values of PsOexists (Fig. 9). The mechanisms underlying this variation are not clear: in our sample of normal subjects, 2,3-DPG concentration, Hb concentration, and age did not correlate with PsO,but HbCO concentration did (r = -0.524, P c 0.001). Excluding smokers, we find PsO values slightly higher for females than males. Because of the lower Hb concentrations in the females (Table 6) thi s might reflect a compensatory mechanism. However, 2,3-DPG was slightly lower in the females and we must conclude that the sample size is too small for a definitive analysis. Thus, the regulation of oxygen affmity in normal subjects is probably mediated by multiple variables. The biological importance of the position of the blood OEC is a subject of current debate (see Ref. 13 for a review). Genetic variability in the whole blood OEC could affect survival at high altitude (11). Furthermore, in some patients with coronary artery disease the OEC might be shifted to the left (4), Important areas such as these can only be investigated with very precise methods since the required accuracy of measurement of the OEC is very high and it is essential to control pH, Pcoz, HC03-, and temperature during the measurement. We are grateful to Richard Shrager and Lawrence Thibault for help with various aspects of this work, to Tektronix, Inc. for help with the computer programs for the control of the flexible disc unit, and to Carlo-Erba Strumentazione, Milan, for use of the micro-blood analyzer. We are particularly indebted to our many colleagues, too numerous to list, who contributed blood samples, and to Exa Murray for typing the manuscript, Received
6 September
1977; accepted
in final
form
24 March
1978.
REFERENCES 1. ADAIR, G. S The hemoglobin system. VI. The oxygen dissociation curve of hemoglobin. J. BioZ. Chem. 63: 529-545, 1925. 2. ASAKURA, T., K. MINAKATA, K, ADACHI, M, 0. RUSSELL, AND E.
SCHWARTZ. Chin. Invest. 3. DUVELLEROY,
Denatured hemoglobin 59: 633-640, 1977. M. A., R. G. BUCKLES,
in
sickle
erythrocytes.
S. ROSENKAIMER,
J.
C. TUNG,
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY
4.
5,
6.
7.
8.
OF
BLOOD
OXYGEN
297
AFFINITY
AND M. B. LAVER. An oxyhemoglobin dissociation analyzer. J. Appl. PhysioZ. 28: 227-233, 1970, ELIOT, R. S., J. M. SALHANY, AND EI. MXZUKAMI. Angina and infarction occurring with patent coronary arteries and decreased rate of oxygen release. In: Hypoxia, High AZtitude, and the Heart. First Conference on Cardiovascular Disease, Aspen, Colorado. New York: Karger, 1970. FATT, I. Polarogruphic Oxygen Sensor. Its Theory of Operation and Its AppZication in Biology, Medicine and Technology. Cleveland, Ohio: CRC Press, 1976. HAHN, C. E. W., P. FOEX, AND C. M. RAYNOR. A development of the oxyhemoglobin dissociation analyzer. J. AppE. PhysioZ. 41: 259-267, 1976, HILL, A. V, The possible effects of the aggregation of the molecules of hemoglobin on its dissociation curves. J. PhysioZ. London 40: IV, 1910. IMAI, K., H. MORIMOTO, M. KOTANI, H. WATARI, W. HIRATA, AND M, KUR~DA. Studies on the function of abnormal hemoglobins. I. An improved method for automatic measurement of the oxygen equilibrium curve of hemoglobin. Biochim Biophys. Actu 200: 189-196,197O. MAY, A., AND E, R. HUEHNS. The control of oxygen affinity of red cells with Hb Sheperds Bush. &it, J. Haematol. 22: 599, 1972. MIZUKAMI, H., A. G, BEAUDOIN, E. BARTNICKI, AND B. ADAMS. Hysteresis-like behavior of oxygen association-dissociation equilibrium curves of sickle cells determined by a new method. Proc. Sot. Exptl. BioZ. Med. 154: 304-309, 1977. MORPURGO, G., P, ARESE, A. BOSIA, G, P, DESCARMONA, M, LUZZANA, G. MODIANO, AND S, K. RANJIT. Sherpas living permanently at high altitude: a new pattern of adaptation. Proc. Natl. Acad, Sci. US 73: 747-751, 1976. NYGAARD, S. F,, AND M. RORTH. An enzymatic assay of 2,3diphosphoglycerate in blood, Stand. J. CEin. Lab. Invest. 24: l
9.
10.
11.
12.
399-403, 1969. 13. RAND, P. Is hemoglobin-oxygen affinity relevant? J. Maine Med. sot. 66: 5-7, 1975. 14. ROSSI-BERNARDI, L., M. LUZZANA, M. SAMAJA, M, DAVI, II. DARIVA-RICCI, J. MINOLI, B. SEATON, AND R. L. BERCER. Continuous determination of the oxygen dissociation curve for whole blood. CZin. Chem. 21: 1747-1753, 1975. 15. ROSSI-BERNARDI, La, M. PERELLA, M. LUZZANA, Me SAMAJA, AND I. RAFFAELE. Simultaneous determination of hemoglobin derivatives, oxygen content, oxygen capacity, and oxygen saturation in 10 ?I whole blood. Clin. Chem. 23: 1215-1225, 1977. 16. ROUGHTON, F. J. W., E. C. &LAND, J, C. KERNOHAN, AND J. W. SEVERINGHAUS. Some recent studies of the oxyhaemoglobin dissociation curve of human blood under physiological conditions and the fitting of the Adair equation to the standard curve. In: Oxygen Affinity of Hemoglobin and Red Cell Acid Base Status, edited by P. Astrup and M. Rsrth. New York: Academic, 1972, p. 73-83. 17. SEVERINGHAUS, J. W. Blood gas calculator. J. Appl. Physiol. 21: 1108-1116, 1966. 18. TYUMA, I., I. IMAI, AND K. SCHIMIZU. Organic phosphates and the oxygen equilibrium function of some human hemoglobins. In: Oxygen Affinity of Hemoglobin and Red Cell Acid Base Status, edited by P. Astrup and M. Rsrth. New York: Academic, 1972, p. 131-146. 19. TYUMA, I., AND Y. UEDA. Non-linear relationship between oxygen saturation and proton release, and equivalence of the Bohr and Haldane coefficients in human hemoglobin. Biochem. Biophys. Res. Commun. 65: 1278-1283,1975e 20. WINSLOW, R. M., M. SWENBERG, R. L. BERGER, R. I. SHRAGER, M. LUZZANA, M. SAMAJA, AND L. ROSSI-BERNARDI. Oxygen equilibrium curve of normal human blood and its evaluation by Adair’s equation. J. BioZ. Chem. 252: 2331-2337, 1977.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
blood:
ROBERT M. WINSLOW, JANICE M. MORRISSEY, ROBERT L. BERGER, PAUL D. SMITH, AND CARTER C. GIBSON Clinical Hematulugy Branch, Laboratory of Technical Development, National Heart, Lung and Blood Institute, and Biomedical Engineering and Instrumentation Branch, Division of Research Services, Nutionul Institutes of Health, Bethesda, Maryland 20014
WINSLOW, ROBERT M., JANICE M, MORRISSEY, ROBERT L. BERGER, PAUL Il. SMITH,AND CARTERC. GIBSON. Variability of oxygen affinity of vzormal blood: an automated method of measurement. J. Appl. Physiol. : Respirat Environ. Exercise Physiol, 45(Z): 289-297, 1978. -Oxygen equilibrium curves of 48 healthy adult subjects have been measured by the method of Rossi-Bernardi et al. (Chin. Chem.. 21: 1747, 1975), in which I&O, is gradually added to a sample of deoxygenated blood that contains an excess of catalase. The mean P,, for nonsmokers was 26.9 Torr and the distribution of values was slightly skewed to the right (range 24.2-29.9 Torr). The method differs from others previously available in that pH, Pcoz, and HCO,are constant during oxygenation. The system for control of the experiment and data collection and processing has been automated by the use of a microprocessor so that the equilibrium curve can be obtained quickly, reproducibly, and relatively simply. With the aid of a digital computer, the parameters of the generalized Adair equation can also be estimated.
hemoglobin; processor
oxygen dissociation
curve; Adair
equation;
micro-
METHODS FOR THE MEASUREMENT of the hemoglobin oxygen equilibrium curve (OEC) have evolved recently allowing the systematic investigation of properties of purified hem.oglobin (8). However, analogous methods for measuring the OEC of whole blood have been slow to appear because the red blood cell is far more complex than purified hemoglobin. Blood OEC’s obtained by van Slyke manometric analysis have served as standards for clinicians, physiologists, and biochemists for many years. For example a “standard curve” (16) was obtained using such analysis on samples of blood from several different subjects and were assumed to have the same oxygen half-saturation pressure of hemoglobin (P& and Bohr effect. Several recent methods have been introduced for the continuous measurement of the whole blood OEC by use of small samples from single individuals. All of the advantages an.d disadvantages of these various methods will not be discussed in detail here. Rossi-Bernardi et al. (14) described a method that employs the gradual oxygenation of a sample of completely deoxygenated blood by the controlled addition of H,O, in the presence of catalase. The accompanying released hydrogen ions are titrated with NaOH. This
method has the inherent advantage that fresh whole blood can be used, the entire curve can be measured in as little as 30 min from venipuncture, measurements of carboxyhemoglobin (HbCO), methemoglobin (MetHb), and 2,3-diphosphoglycerate (2,3-DPG) can be made on the sample before and after the run, and carbon dioxide partial pressure (Pcoz) and pH are constant over the entire oxygenation range. No other system that we are aware of combines these advantages, Recently we have described a normal whole blood OEC obtained with the H202 method with extensions at the ext.reme ends of the curve by a simple mixing method (20). Although this technique is relatively simple, the mixing experiments require rearrangement of the experimental apparatus and additional time. The results of those experiments were analyzed according to the Adair stepwise oxygenation scheme, an analysis that is independent of stereochemical models of hemoglobin function. We report experiments in which improvement of the H,O, method has been achieved by automation of the data acquisition and computation system. Adair parameters, which fit the experimental dat.a very well and are therefore useful in simulating the OEC for various physiological computations, can be estimated by this method. METHODS
Apparatus The cuvette is a modification of that described by Rossi-Bernardi et al. (14). It consists of a Lucite chamber containing a water bath that is kept at 37°C. Paz and Pco2 electrodes (Instrumentation Laboratory model 17026 and 16050, respectively) are mounted in the apparatus directly in contact with the blood sample. Stirring is achieved by a small magnetic mixer and a motor mounted beneath the chamber. H,O, (0.4 M) and NaOH (0.4 M) are added to the sample from 50-~1 Hamilton syringes through polyethylene tubing. The syringes are driven by motorized pumps (Raze11 Scientific Instruments, Stamford, Conn.) under computer control. The cuvette was modified by the addition of a thermistor probe (Thermonetics), which protrudes slightly into the chamber. 289
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
290
WINSLOW
Calculations. The symbols used in calculating the oxygenation data are defined in Table 1. Hemoglobin saturation. The fractional saturation of Hb with oxygen is given by
of oxygen bound at a given point is
The amount
of oxygen dissolved
The initial
(4) of an
the addition
0 i = (Oi-1 + AO) x [I - AVl(V
+ AV)]
(5)
is
A0 = H x R x At x ; x lo-”
Symbol
Bi G CO
A0 At AV Q Do E cal Ei
F H Hb HbCO MetHb NaOH Oi Ql
P cal pi
R T cal T”K Ti V VB yi
xl
used in oxygenation
(6)
equations
Quantity
Coefficient of solubility of oxygen in blood Oxygen bound to Hb at ith point Oxygen capacity at ith point Initial oxygen capacity Oxygen content increment Time interval for a sampling Total volume increment in a sample interval Oxygen dissolved in blood at ith point Initial dissolved oxygen Oxygen electrode voltage with standard gas Oxygen electrode voltage at ith point Electrode factor H@, concentration in syringe Hemoglobin concentration Carboxyhemoglobin concentration Methemoglobin concentration NaOH concentration Oxygen content at ith point Initial oxygen content Paz of standard gas Pop at ith point Rate of H,02 addition Temperature of cuvette with standard gas Absolute temperature Temperature of blood at ith point Cuvette volume Volume of NaUH Hemoglobin saturation at ith point Initial hemoglobin saturation
Units
pmol/mm
4 pm01 pm01 pm01 pm01 S
pm01 pm01 V V Dimensionless a PM, heme % % PM pm01 pm01 Torr Torr plls
OC “K “C 4 4 Fraction Fraction
x
Pi = where
(8)
is
pressure
(3)
T”K
after
O2 capacity
oxygen partial
273
AV = VB + R x At
a
is not zero, the oxygen is
C 0 = Hb x [l - (HbCO
Since the cuvette volume is fixed, when each increment of H,O, and NaOH is added, an equal volume of mixed blood is expelled. This requires the correction of the 0, content by a dilution factor. The volume added to the cuvette during each sample interval is
1. Symbols
(7)
0 0 = (Y, x co> + Do
in the blood is
D i =pixvxa-
TABLE
-t- AV)]
(1) content at the start of the experiment
The amount
The 0, increment
Hb saturation
AL.
corrected . for dilution
of the system,
Ci = Ci-1 x [l - AV/(V When the initial
Bi Yi = Ca
Therefore, the 0, content in .crement of oxygen (AO) is
The oxygen capacity is
ET
xv
+ MetHb)]
(9)
(Po,) is given by Ei
F, the “electrode
x
F
x
factor,”
(10)
PcallEcal
is defined below.
Curve Fitting 1
l
Estimation of the parameters of the Aaaw oxygenation equation was done as previously described (20) afier transfer of data to the PDP-10 digital computer. For the final fit in the procedure, a scaling factor was used as a multiplier for all saturation values. This factor was included as a parameter of the fit and reduced somewhat the final sum of squared residuals. Its use rests on the assumption that the greatest errors in the data are in saturation, not Paz, and its value was always within the range 0.98-1.02. Titration
of Pco2
As blood is oxygenated the released protons cause a decrease in pH and an increase in Pcoz. The protons are titrated by the addition of NaOH under computer control. Sufficient NaOH is added to maintain Pcoz constant. Preliminary experiments revealed that approximately 2 pmol NaOH were required per pmol of hemoglobin tetramer during oxygenation. Therefore the total NaOH added was first calculated from the quantity of hemoglobin present 2xHbxV total NaOH = (11) ccl NaOH The total time of oxygenation depends upon the amount of hemoglobin present and the rate of addition of H,02 total time = l/2
The estimated then becomes
rate of NaOH
K1 This is only adjusted by tional to the ith point and
Hb x V seconds x R x H addition
from the syringe
RxH =
NaOH
(12)
PUS
(13)
an approximation, and the rate is further inclusion of a term (Q), which is propordifference between the Pcoz voltage at the the initial value NaOH
rate = K, + (Kz x Q)
K, is an empirical constant+ To damp the fluctuation in Pcoz, a third term, proportional to the rate of change of Pcoz voltage, is added
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY
OF
BLOOD
OXYGEN
NaOHrate=K,+(&xQ)+(K,x$) where K, is also empirically
291
AFFINITY
(15) determined.
Data Collection and Analysis The data collection and analysis system is controlled by an Intel 8080 microprocessor. It receives inputs from the experiment, controls the NaOH and H20, pumps and outputs results to a self-scan display, a plotter (Zeta Research, Inc.) and flexible disc (model 4821, Tektronix, Inc.). The program (Real Time Systems, Fairfax, Va.) is written in assembly language and uses 14 K of readonly memory (ROM) and 17 K of random accessmemory (RAM) for storage of data. A block diagram of the hardware is shown in Fig. 1. The flow of an experiment is given in Fig. 2. At user-specified intervals the PO, and PCO~electrodes are sampled and Paz and saturation are calculated (EQS. l-1 0). The H,O, pump is driven at Calculate a constant rate, and the NaOH pump is driven at intervals for a time which depends on the Pcoz (EQ, 15). Display Paz and Hb saturation are plotted in real time and at (self scan) the end of the run the required constants, Po2, and NaOH volume for each point are transmitted to the [Aatpointj flexible disc. A set of data can be recalled to memory FIG. 2. Experiment flow diagram for the automatic OEC apparafrom the disc for off-line calculation and plotting. A set tus. In addition to flow shown, a data set can be recalled to memory of data can also be transmitted to the PDP-10 digital from flexible disc unit for recalculation and off-line plotting. computer for further analysis as desired. This step is controlled by a separate Fortran program on the PDP- veins of normal volunteers and immediately placed into 10. EDTA anticoagulant. Determinations of Hb, HbCO, and MetHb were done on IO-~1 samples using the Blood Samples and Reagents Microblood Analyzer (kindly loaned by Carlo Erba, Whole blood samples were obtained from antecubital Strumentazione, Milan) (15). 2,3-DPG concentrations were measured by the method of Nygaard and Rsrth 100Ins (12) using kits from Calbiochem. All reagents were Clock reagent grade. Three percent H,O, (USP) was obtained from Parke Davis. All gas mixtures were obtained from 32 CHARD&P. 8080A c Lif-o-Gen, Cambridge, Md. Their compositions were CPU certified to a nominal -t- 0.1%.
I-
Key Board r
Experimental
Mode Sel ’
POWER SUPPlIES fl FIG. 1. Hardware block diagram for automatic OEC apparatus. Abbreviations are: mode sel, mode selector; DPM, digital panel meter; A/D, analog-to-digital converter; char disp, character display; CPU, central processing unit,
Procedure
With the stirrer on, calibrating gas was passed into the cuvette after humidification at 37”C, and the temperature in the cuvette during this period was carefully noted (T,& The 0, and CO, electrodes were calibrated at two settings, 0 and 12% 02, and 4 and 10% Cob,. A small amount of catalase (Calbiochem) was added tithe cuvette before the run to assure complete conversion of H,O, to molecular oxygen without significant oxidation of hemoglobin. With normal blood, the catalase can be omitted without effect on the shape or position of the OEC. A 2-ml sample of blood was deoxygenated by equilibration with 5.6% CO, (balance N,) in an IL model 237 tonometer for 15-20 min. Then by use of a gas-tight Hamilton syringe the sample was transferred anaerobically into the reaction cuvette, which contained 10 ~1 of catalase. When the temperature was stable (T;) a small aliquot of the blood was removed for measurement of pH, Hb, HbCO, and MetHb. The tubes from the NaOH and H,O, pumps were placed into the cuvette
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
292 and ment from lO+l ment
WINSLOW
after entering the required constants the experiwas begun by giving an appropriate command the keyboard. At the end of the oxygenation run a sample of the blood was removed for the measureof HbCO and MetHb concentrations.
RESULTS
Temperature Strict control of temperature during the experiment is very important: in addition to its effect on the position of the OEC, the output from the oxygen electrode is strongly temperature dependent. Figure 3 describes an experiment in which the temperature of the cuvette was gradually increased while exposing the electrode to 5% O2 at a constant flow rate. A linear relationship was found with the equation E Cal = Teal X (0.018) + 0.024
(16)
where Ecal is the amplifier output of the 0, electrode in volts, and Teal the temperature at which it is measured. The slope of the line can be used to determine the PO, voltage (Ei) at the temperature of the blood during the experiment (Ti). A difference between Tear and Ti of about 2°C was commonly observed, even though the gas was bubbled through water at 37°C. The error in PO,
ET
AL.
introduced from this source would be 3.6% for a 2°C temperature difference. In practice, the electrode factor (see below) corrects for this discrepancy. If the temperature of the blood during the oxygenation run is not constant the result is virtually uninterpretable because in addition to the calibration error the position of the OEC itself will be altered. The temperature of the blood was monitored during each run and was always constant, Electrode
Factor
The “electrode factor” is the ratio of output observed from the oxygen electrode with blood in the cuvette to that with calibration gas (5). This value, F (Ea. IO), is normally less than 1 because of the formation of a thin layer of stagnant blood at the surface of the PO:! electrode membrane. Therefore, it is not surprising that its actual value would vary with the rate of stirring, geometry of the cuvette, and the rheological properties of the blood itself, In the experiments with normal blood described here, hematocrit values were distributed over a narrow range and the stirring rate was the same for all runs. The electrode factor was checked prior to a set of experiments (i.e*, each working day) by introducing into the cuvette blood that had been equilibrated with gas of the same composition. The actual value of F varied between 0.95 and 1.00 but was invariant over a given day and with a given 0, electrode membrane. It must be noted that F varies with the rate of stirring, the gas-liquid difference, and the difference between the temperature in the cuvette during calibration and when filled with blood. The Data
TEMPERATURE,degree C
3. Calibration of cuvette temperature. With 5% O2 flowing into cuvette, temperature of water bath was gradually increased. Relation between temperature and voltage output of the O2 amplifier was linear, Regression line had slope 0.018 V/“C and intercept 0.024 FIG.
During the experiment PO, and Pco2 are recorded directly on a strip recorder (Fig. 4). This recording, though not essential, is useful in monitoring the NaOH titration. Concomitantly the computed curve is plotted in real time (Fig. 5). The Pcoz remains constant in a typical run to approximately 20.3 Torr. Figure 4 is a direct tracing of a typical experiment and Fig. 5 is a photograph of an on-line plot. Since several independently measured values are used in the calculation of saturation (see Eqs. I-10) it is not unusual to observe at 150 Torr a final saturation
FIG. 4. Direct tracing of the Paz and Pco2 during an experiment. Short line at top represents a change in x-axis sensitivity so that full scale is 200 Torr. H,Oz was started as indicated by arrow on ordinate.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY
OF 100
BLOOD
OXYGEN
293
AFFINITY
-
FILE
NFIHE
1207
SUBJECT UNIT
Flu
ORTE OEQXY
r
T
r
r
T
I
I
I
I
plot of calculated
data.
7.414
PC02
39.13
TEtIPERRTURE P50
026.10
I
5. On-line
I-JUtiPH
I
that varies by -+2% Therefore a small aliquot of the sample is removed it the end of the run to measure saturation independently with the microblood analyzer. This value can then be used to normalize all of the saturation values by multiplication by the ratio of saturation expected to saturation measured. A second method of estimating the final saturation employs the estimation of a scaling factor from the best fit of the data to the Adair equation. However, this cannot be done by the microprocessor at present. NaOH
FIG.
NUIIBER
37.0
4
1
I
1
2. Effect of NaOH addition on change in pH and Pm2 during oxygenation TABLE
PH
PC%, Torr
No titration Beginning End
7.458 7.337
36 46
With titration Beginning End
7,364 7.362
40 40
Additiun
During oxygenation of blood in the reaction cuvette, the release of protons and decrease of pH will be accompanied by a rise in CO, (Table 2). The addition of NaOH ensures that pH and Pcop are held constant to within the limits of the errors of measurement. Figure 6 is the amount of NaOH added as a function of the total, plotted against hemoglobin saturation. As has been observed by others for pure hemoglobin (19) this is not a linear relationship: relatively more NaOH is required at the early stages of saturation. This could have been a result of the slow response time of CO, electrode (54 s, 90% response). We therefore performed an oxygenation experiment in which the H,O, pump was stopped at l-min intervals until the Pco2 had stabilized. This procedure gave a result identical with the continuous method (Fig. 6). An important advantage of this method over the system described by Duvelleroy et al. (3) is that no correction need be applied for the nonuniformity of the protons released during oxygenation, since the Bohr protons are titrated very closely during the run, Thus to interpret the full OEC measured by the latter method one needs to know the quantitative magnitude of the Bohr effect as a function of saturation in order to correct the data to uniform pH (6). In contrast our method can be used to study the Bohr effect. Adair’s Equation Figure 7 shows the fit of a set of data obtained by the automatic method to the Adair stepwise oxygenation formula (1)
I .2
1
.4 HEMOGLOBIN
I
I
1
.6 48 SATURATION
FIG. 6. Nonlinearity of NaOH addition with increasing hemoglobin saturation. Rate of NaOH addition diminishes as saturation increases. Automatic (0) and manual (e) titrations, see text for details, Degree of departure from linearity depends on experimental conditions (especially pH, unpublished observation),
Y
a,P + 2a2P2 + 3a,P3 + 4aqP4 = 4(1 + alP + a2P2 + a,P3 + a4P4)
(17)
The curve-fitting routine, which was described previously, was applied to data sets using the PDP-10 digital computer (20). Since the errors in the calculation are greater in saturation than Po2, the routine was modified slightly to incorporate a scaling factor to be used with the saturation values to obtain the least value of the residual sum of squares. By this method one assumes that the data can be described by the Adair scheme and that a departure of the data from the fit can be accounted for by errors in the concentration of H2021 Hb, HbCO, and MetHb concentration.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
WINSL0W
ET
AL.
FIG. 7. Fit of a typical data set to Adair equation (Eq. 17). For c1arity, every 10th experimental point is plotted with curve described by Adair simulation (continuous Line).
75 PO2
TABLE 3. Errors* in Adair parameters for present method and extended continuous method (Ref- 20) Continuous a,
x
a2 x
0.157 -36 to +65%
10-l
lo-” -45
u3
x 1o-5 -100
u4 x 1o-5 * 95% confidence
0.117 to +ll% 0 to +lOO%
0.245 -6 to +7%
I
I
CONTINUOUS
I
I
EXTENDED
Extended
-29
0.151 to +38%
-28
0.097 to -17%
-100
I1 :.: :I I .: ‘, 1\ ;i I \ iir \ iir \\ :‘I:ir \\’iit1 .-.---+~..3.-.-*-.-.-. I
0,170 to +lOO%
0.167 -5 to +5%
limits.
Analysis of the errors in the Adair parameters (Table 3) is shown in Fig. 8. This analysis, as described previously (20) was done by varying the value of one of the a’s systematically, while determining the value of the sum of squared residuals after fitting the other three parameters to the data. The resultant errors are significantly larger than those determined previously be extending the OEC at its extremes (20)+ Furthermore, inspection of Fig. 8 indicates that the error distributions are not symmetrical. Therefore, the standard errors given by the application of a linear curvefitting routine to such nonlinear data could be extremely misleading.
I -50
I 0
I 50
1 -50
.---I
0
I
50
% ERROR IN a
FPG, 8. Analysis of the errors in Adair parameters. Value of each of parameters (see Eq. 17) was systematically varied during determination of residual sum of squares after fitting the other three to the data. (-, al; - - -, a2; - -, a3; . . . , a+) Analysis for a single curve (continuous) was compared to a curve which had been extended at its extremes (extended, see Ref. 20) Horizontul lines repres,ent 95% confidence limits (f test). In both instances, 60 experimental points were used for analysis, l
TABLE File
1204 1205 1206 1207 1208
4. Reproducibility PH
7.412 7.412 7.418 7.414 7.421
l
of continuous a, x 10L2
OEC
$$
nmax*
a, X 10-l
26.5 26.8 26.5 26.4 26.2
2.68 2-70 2.70 2.73 2.69
0.142 0.123 0.144 0,159 0.103
0.126 0.121 0,123 0.116 0.131
0.241 0.227 0.242 0.247 0.241
0.134
0.123
0.240
a3
a4 X 10v5
Reproducibility Mean
Table 4 presents the results of five consecutive measurements on blood from a single venipuncture. These determinations were completed within 3 h and the blood sample was kept on ice during this period. Survey
of Normal
OEC’s
Forty-eight normal individuals of both sexes, both smokers and nonsmokers but of a narrow age distribu-
* Maximum
26.5 slope
2.70 of the Hill
plot
(see Ref. 8).
tion, were used to study the variability of the normal PSO(Table 5). Figure 9 shows the results of these studies. Normal P,, ranged from 22.5 to 29.9 Torr and appeared to lie in a normal distribution but slightly skewed to the right. The mean PSOwas 26.5 Torr. This value is the same as the value which has been accepted as a stan-
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY TABLE
variables
OF
BLOOD
OXYGEN
295
AFFINITY
5. Statistical analysis of hematologic in nonsmoking subjects n
Range
Hb, g/100 ml
Male Female
Sex
15.2 & 1.2 13.7 2 0.8
23 15
12.9-18.1 11.6-15.3
2,3-DPG, mall mol Hb
Male Female
0.928 k 0.129 0,885 t 0.194
23 15
0.739-l-340 0.396-1.274
HbCO,
Male Female
2.3 * 0.8 2.7 & 1.2
23 15
1.2-4.0 1.0-4.5
Age, yr
Male Female
32.4 k 5.7 31.6 k 9.2
16 14
20-40 19-48
PSO, Torr
Male Female
26.7 2 1.7 27.2 k 1.2
23 15
24.2-29.9 25.7-29.6
%
Mean
5 SD
(for a 2°C difference between Teal and Ti). In all of our experiments, temperature was continuously measured and did not vary by more than O.l”C. Furthermore, the electrode factor (F) could vary with the electrode used, the condition of its membrane, mixing, and blood viscosity. This could be the source of another 5% error. Therefore, monitoring the temperature of the cuvette and frequent checking of the electrode factor is of critical importance. Reproducibility of our data is excellent (Table 4) and we therefore believe that the distribution of P,, values shown in Fig. 9 is due to biological, not experimental, factors. Comparison
-
with Previous
“Standard
Curve”
A typical OEC is compared to the “standard” OEC of Severinghaus (17) in Table 6. The PO, values at 50% saturation are in very close agreement, but a discrepancy of as much as 1.7% occurs at high Paz* Table 5 also illustrates that our OEC can be represented quite well by the Adair formula (Eq. 17). The differences between our OEC and the Severinghaus standard curve must be considered in the light of the differences in the methods used. In the latter instance, data from several individuals, indeed from several investigators, were compiled to assemble an average OEC. We believe that biological differences between normal individuals (cf. Fig. 9, Table 6) can be detected with our method since reproducibility is good (Table 4), Therefore such averaging will obscure important differences between subjects. DISCUSSION
22
24
26
20
30
~50, mmHg FIG, 9. Blood oxygen affinity of 33 healthy adult subjects. Paz at half saturation (P,,) was corrected to whole blood pH of 7.4 assuming a linear Bohr factor (Alog PJA pH) of - 0.40). This value is based on unpublished data obtained with apparatus described here,
dard from previous work (17). Many of the subjects whose blood PsO lies on the left tail of the normal distribution were cigarette smokers and had increased HbCO. Correlation between PsO and the level of 2,3DPG was rather poor (r = 0.018). Likewise no significant differences could be detected between sexes when smokers and nonsmokers were examined separately. The wide range of normal values obtained in our studies was rather unexpected. Modulation of the whole blood oxygen affinity by 2,3-DPG and pH are well known, but the variation we observe cannot be accounted for by variations in these effecters. We assume that the structure of the hemoglobin molecule itself does not vary among normal individuals and therefore the factors responsible for the variation are either conditions that exist within the red blood cell or variations in the experimental conditions. An example of the latter would include fluctuation in temperature, not only of the whole blood sample during the run but of the cuvette during calibration of the oxygen electrode. This discrepancv could cause errors in PO, of as much as 3.6%
Other methods for the continuous recording of the OEC are currently available (3, 9, 10). Those using the optical measurement of saturation have several inherent disadvantages. First, since the optical absorbance of whole blood is very high some methods require that TABLE 6. Comparison of measured and computed saturation Pas, Torr
Standard Curve Saturation,* %
Measured Saturation,t %
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
3.39 9.61 19.13 32.22 46.00 57.60 66.72 74.43 80.47 85.12 88.25 91.00 93.00 94.12 95.13 95.86 96.40 96.86 97.21 97.50
3.02 9.24 19.82 32.42 45.86 58.82 67.70 75.55 80.48 84.81 87.60 89.66 91.21 92.54 93.53 94.28 94.94 95*45 95.91 96.31
* See Ref. 17. Table 3 (continuous
t Curve curve).
1207,
Table
Adair
Saturation,S %
3.23 9.43 19.41 32.44 46.15 58.33 68.05 75.38 80.78 84.76 87.72 89.94 91.64 92.96 94.00 94.83 95.51 96.00 95.52 96.90 4.
$ Constants
from
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
296
WINSLOW
cells be diluted into buffer before the experiment is run. This is therefore not desirable when physiologically relevant data are required. An exception to this is the method that employs a very thin layer of blood in a controlled atmosphere (Ref. 10, and the Hem-O-Scan, American Instrument Co.) Second, optical methods require the defmition of 100% saturation at some absorbance. This is usually done by exposing a sample to air at the end of the run. At ambient PO, (about 150 Torr) normal blood is 98% saturated (21) and pathological blood with reduced oxygen affinity, especially sicklecell anemia blood, can be even less saturated (unpublished observations). Finally optical methods cannot deal with changes in the absorbance spectrum which could occur by denaturation of hemoglobin during the experiment (2). The present system, being closed, does not have a liquid interface for equilibration. The fact that oxygen is added as H,Oz means that the time required for the experiment is limited only by mixing of the two liquids, H,O, and blood, and is therefore rapid. The major advantage of the closed system, however, is that pH, Pco~, and HCO,- can be held constant throughout the run and, as opposed to many other systems, no correction need be applied to the data for the Bohr effect. This is very important since the Bohr effect is not constant with saturation (19), The use of the microprocessor means that the cost of automation can be kept at a minimum and the reliability and simplicity of operation to a maximum. The personnel operating the equipment need not have any specialized training in computer technology. Mistakes are minimized and as many as lo-12 complete curves can be performed in a single working day. Immediate feedback of the OEC to the operator has the advantage that curves can be repeated when desired while blood samples are still fresh, As can be seen from Fig. 7 and Table 6, the fit of the data to the Adair equation is excellent. Therefore, the Adair parameters can be used to calculate saturation at any PO,, or to perform a variety of other calculations of physiological and physicochemical interest, For example, the arterial-venous oxygen difference calculated from these parameters is a true reflection of the OEC, whereas the OEC plotted according to the logarithmic form of the Hill equation (7) is not linear Y 1% 1y = n log P + k
(18)
over the entire range of oxygenation. Since the Hill equation does not fit the experimen tal data at its extremes, the parameter n, a measure ofs ‘ubuni t cooperativity, is not fixed but reaches a maximum with minima at high and low saturation (18). The maximum value of n, and its position with respect to the saturation, is of physicochemical interest and can be easily computed from the Adair parameters. Theoretically, it should be possible to calculate the
ET
AL.
values of the equilibrium constants for the four successive oxygenation steps of the Hb from the Adair parameters. This can be done within reasonable error for pure Hb solutions (18), but the errors are larger for whole blood (Fig. 8), even with extensions at the top and bottom of the OEC (20). Whether this large error is due to experimental inaccuracy (see Ref. 20 for a discussion) or to the fact that whole blood is an extremely heterogeneous system compared to hemoglobin solutions is not yet clear. The magnitude of these errors means that we cannot at present compute unique values of the equilibrium constants for the four oxygenation steps of whole blood, Calculations of oxygen delivery in peripheral tissues and uptake in the lung are based on a ctstandard” OEC. The overall regulation of oxygen transport in any living organism is extraordinarily complex because it is dependent not only on whole blood oxygen affinity and all the determinants thereof but also cardiac output, pulmonary function, regional blood flow, hemoglobin concentration, and tissue oxygen demand. A desirable goal of current research is the quantitation of the oxygen delivery system so that perturbations of various components of it can be systematically studied. It is therefore important that the oxygen equilibrium data obtained for such modeling should be of the highest precision and reflect physiological conditions. Our results indicate that a wide range of normal values of PsOexists (Fig. 9). The mechanisms underlying this variation are not clear: in our sample of normal subjects, 2,3-DPG concentration, Hb concentration, and age did not correlate with PsO,but HbCO concentration did (r = -0.524, P c 0.001). Excluding smokers, we find PsO values slightly higher for females than males. Because of the lower Hb concentrations in the females (Table 6) thi s might reflect a compensatory mechanism. However, 2,3-DPG was slightly lower in the females and we must conclude that the sample size is too small for a definitive analysis. Thus, the regulation of oxygen affmity in normal subjects is probably mediated by multiple variables. The biological importance of the position of the blood OEC is a subject of current debate (see Ref. 13 for a review). Genetic variability in the whole blood OEC could affect survival at high altitude (11). Furthermore, in some patients with coronary artery disease the OEC might be shifted to the left (4), Important areas such as these can only be investigated with very precise methods since the required accuracy of measurement of the OEC is very high and it is essential to control pH, Pcoz, HC03-, and temperature during the measurement. We are grateful to Richard Shrager and Lawrence Thibault for help with various aspects of this work, to Tektronix, Inc. for help with the computer programs for the control of the flexible disc unit, and to Carlo-Erba Strumentazione, Milan, for use of the micro-blood analyzer. We are particularly indebted to our many colleagues, too numerous to list, who contributed blood samples, and to Exa Murray for typing the manuscript, Received
6 September
1977; accepted
in final
form
24 March
1978.
REFERENCES 1. ADAIR, G. S The hemoglobin system. VI. The oxygen dissociation curve of hemoglobin. J. BioZ. Chem. 63: 529-545, 1925. 2. ASAKURA, T., K. MINAKATA, K, ADACHI, M, 0. RUSSELL, AND E.
SCHWARTZ. Chin. Invest. 3. DUVELLEROY,
Denatured hemoglobin 59: 633-640, 1977. M. A., R. G. BUCKLES,
in
sickle
erythrocytes.
S. ROSENKAIMER,
J.
C. TUNG,
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
VARIABILITY
4.
5,
6.
7.
8.
OF
BLOOD
OXYGEN
297
AFFINITY
AND M. B. LAVER. An oxyhemoglobin dissociation analyzer. J. Appl. PhysioZ. 28: 227-233, 1970, ELIOT, R. S., J. M. SALHANY, AND EI. MXZUKAMI. Angina and infarction occurring with patent coronary arteries and decreased rate of oxygen release. In: Hypoxia, High AZtitude, and the Heart. First Conference on Cardiovascular Disease, Aspen, Colorado. New York: Karger, 1970. FATT, I. Polarogruphic Oxygen Sensor. Its Theory of Operation and Its AppZication in Biology, Medicine and Technology. Cleveland, Ohio: CRC Press, 1976. HAHN, C. E. W., P. FOEX, AND C. M. RAYNOR. A development of the oxyhemoglobin dissociation analyzer. J. AppE. PhysioZ. 41: 259-267, 1976, HILL, A. V, The possible effects of the aggregation of the molecules of hemoglobin on its dissociation curves. J. PhysioZ. London 40: IV, 1910. IMAI, K., H. MORIMOTO, M. KOTANI, H. WATARI, W. HIRATA, AND M, KUR~DA. Studies on the function of abnormal hemoglobins. I. An improved method for automatic measurement of the oxygen equilibrium curve of hemoglobin. Biochim Biophys. Actu 200: 189-196,197O. MAY, A., AND E, R. HUEHNS. The control of oxygen affinity of red cells with Hb Sheperds Bush. &it, J. Haematol. 22: 599, 1972. MIZUKAMI, H., A. G, BEAUDOIN, E. BARTNICKI, AND B. ADAMS. Hysteresis-like behavior of oxygen association-dissociation equilibrium curves of sickle cells determined by a new method. Proc. Sot. Exptl. BioZ. Med. 154: 304-309, 1977. MORPURGO, G., P, ARESE, A. BOSIA, G, P, DESCARMONA, M, LUZZANA, G. MODIANO, AND S, K. RANJIT. Sherpas living permanently at high altitude: a new pattern of adaptation. Proc. Natl. Acad, Sci. US 73: 747-751, 1976. NYGAARD, S. F,, AND M. RORTH. An enzymatic assay of 2,3diphosphoglycerate in blood, Stand. J. CEin. Lab. Invest. 24: l
9.
10.
11.
12.
399-403, 1969. 13. RAND, P. Is hemoglobin-oxygen affinity relevant? J. Maine Med. sot. 66: 5-7, 1975. 14. ROSSI-BERNARDI, L., M. LUZZANA, M. SAMAJA, M, DAVI, II. DARIVA-RICCI, J. MINOLI, B. SEATON, AND R. L. BERCER. Continuous determination of the oxygen dissociation curve for whole blood. CZin. Chem. 21: 1747-1753, 1975. 15. ROSSI-BERNARDI, La, M. PERELLA, M. LUZZANA, Me SAMAJA, AND I. RAFFAELE. Simultaneous determination of hemoglobin derivatives, oxygen content, oxygen capacity, and oxygen saturation in 10 ?I whole blood. Clin. Chem. 23: 1215-1225, 1977. 16. ROUGHTON, F. J. W., E. C. &LAND, J, C. KERNOHAN, AND J. W. SEVERINGHAUS. Some recent studies of the oxyhaemoglobin dissociation curve of human blood under physiological conditions and the fitting of the Adair equation to the standard curve. In: Oxygen Affinity of Hemoglobin and Red Cell Acid Base Status, edited by P. Astrup and M. Rsrth. New York: Academic, 1972, p. 73-83. 17. SEVERINGHAUS, J. W. Blood gas calculator. J. Appl. Physiol. 21: 1108-1116, 1966. 18. TYUMA, I., I. IMAI, AND K. SCHIMIZU. Organic phosphates and the oxygen equilibrium function of some human hemoglobins. In: Oxygen Affinity of Hemoglobin and Red Cell Acid Base Status, edited by P. Astrup and M. Rsrth. New York: Academic, 1972, p. 131-146. 19. TYUMA, I., AND Y. UEDA. Non-linear relationship between oxygen saturation and proton release, and equivalence of the Bohr and Haldane coefficients in human hemoglobin. Biochem. Biophys. Res. Commun. 65: 1278-1283,1975e 20. WINSLOW, R. M., M. SWENBERG, R. L. BERGER, R. I. SHRAGER, M. LUZZANA, M. SAMAJA, AND L. ROSSI-BERNARDI. Oxygen equilibrium curve of normal human blood and its evaluation by Adair’s equation. J. BioZ. Chem. 252: 2331-2337, 1977.
Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (129.237.035.237) on January 1, 2019.
