- Email: [email protected]
Measurement of blood gases by gas chromatography
Respiration
Physiology
MEASUREMENT
(1966) 1, 398-407;
OF
BLOOD
North-Holland
GASES
BY
Publishing
GAS
Company,
Amsterdam
CHROMATOGRAPHY’
C. LENFANT AND C. AUCUTT Institute
of Respiratory
and of Physiology
Physiology,
and Biophysics,
Firland Sanatorium,
University
and Departments
of Washington,
Seattle,
of Medicine
Washington,
U.S.A.
Abstract. A method using gas chromatography to measure blood gases has been developed. Equipment and procedure are described. Respiratory gases, 0~ and COz, and Nz, can be measured, although these analyses should be performed separately. For all three gases thecalibration is linear and passes through zero. The maximum deviation from linearity was found to be 0.1 vol ‘A for Coz, 0.3 vol ‘A for CCO, and 2.2 mm Hg for PN~. The average reproducibility tested on many samples was within -CO.06 vol ‘A for 02, 5 0.18 vol % for CO2 and j, 2. mm Hg for PN~. Gas chromatography Measurement of respiratory
Measurement
of dissolved gases in blood
gases in blood
The major difficulty in measuring blood gases by gas chromatography is obtaining rapid and complete release of the gases from the blood, including those bound to the respiratory pigment. The methods available to extract gases from blood, or any other liquid, are of three types: (1) the “vacuum extraction” method (RAMSEY, 1959; LUCAS and AYRES, 1961; FARHI, EDWARDS and HOMMA, 1963). In this method the gases are extracted from the blood prior to being introduced into the gas chromatograph, thus fast elution and good resolution of the peaks are easily obtained. It requires, however, in addition to the gas chromatograph, a Van Slyke apparatus, which is itself a source of problems. Also a fraction of the gases always is retained in the liquid remaining in the Van Slyke extraction chamber. (2) The “washout” or “gas flow-through” method (WILSON et al., 1961; GALLA and OTTENSTEIN, 1962; WILSON, 1964). In this method the sample is injected directly into the carrier gas stream. The procedure is very simple and all gases eventually will be released from the blood. However, resolution is poor and elution is slow. Since the peaks cannot be measured accurately in terms of peak height, it is necessary to determine peak surface area by integration or approximation. (3) The “diffusion process” or “equilibration” method (HAMILTON, 1962). This method combines the advantages of the two preceding techniques. Accepted for publication I June 1966. 1 Supported by National Institute of Health Grants H.E. 08465 and H.E. 01892.
398
MEASUREMENT
A version
OF BLOOD
of this latter method
GASES
is described
BY GAS CHROMATOGRAPHY
here and its accuracy
399 is appraised.
Method PRINCIPLE
A blood sample, with gas-free reagents added, is stirred in an extraction chamber isolated from the carrier gas line. If this chamber contains an atmosphere of He, or any other carrier gas different from those to be extracted, a vacuum exists relative to the gases contained in the blood sample. The gas molecules then diffuse from the blood into the surrounding gas phase until equilibrium is reached. At this point the carrier gas is allowed to flow through the extraction chamber, to the chromatograph columns and detector, transporting with it both the extracted gases and all gases which are still retained in the blood at the end of the equilibration period. This method offers the following advantages: all gases are extracted with certainty, even those with a high partition coefficient, and sharp narrow peaks are produced making possible simple quantitative measurements in terms of peak height. EQUIPMENT
A Beckman gas chromatograph Model 2A and a Bristol 1 mV full scale recorder are used. The chromatograph is equipped with two parallel columns, one packed with silica gel and the other with molecular sieve SA. The length of the columns is adjusted to provide good resolution and fast elution. When only N, is measured, it is preferable
IO
inside ,-_-.__-_--
capillary
-
I / I
PRESSURE REGULATOR
columns
G C / 2 A
dessicotor
prrssure gage blood sample inlet ‘\
gas
outlet EXTRACTION
gas l
P r. Larricr
capillary
Fig. 1. Schematic representation
inlet \
gas inlet valve
exhaust
CHAMBER
valve
.-MOTOR
of the blood gas extraction chamber and accessory
apparatus.
400
C. LENFANT
AND C. AUCUTT
to eliminate the silica gel column. Either column can be reactivated by baking it overnight at 210 “C. The blood extraction chamber and accessory apparatus (fig. l), built in our laboratory, consist of the following components: (a) The extraction chamber is made up of two parts. A glass cover (Fisher # 6.390-10) is hermetically sealed on a brass reservoir. The cover has a sample inlet and a gas outlet. The reservoir has a carrier gas inlet and an exhaust for discarding the analyzed blood. The extraction chamber contains a magnetic stirrer activated by a small phonograph motor mounted beneath it. (b) There are 3 valves in the accessory apparatus: the isolating or bypass valve, the carrier gas inlet valve, and the exhaust valve. The isolating or bypass valve (Republic valves 330 series or Beckman #23800) is a four way valve designed to isolate the extraction chamber from the main carrier gas line. The carrier gas inlet valve (toggle valve-Hoke 490 series) is opened only to free the reagents from their dissolved gases and to permit the pressure in the extraction chamber to equilibrate with that of the main carrier gas line after the sample has been introduced. The flow of carrier gas through this valve is regulated by a capillary tube interposed just ahead of the valve. The exhaust valve (toggle valve-Hoke 490 series) is opened to flush out the analyzed blood and to wash out the chamber. (c) The pressure gauge, graduated in 4 pounds up to 30 pounds is used to measure the pressure in the carrier gas line and in the extraction chamber, and to monitor the system for leaks. (d) The dessicator is filled with silica gel and is used only when N, is measured. It is essential that the dessicator be inserted beyond the bypass valve. (e) These various components are connected with Swagelok fittings of brass or nylon when they are in contact with glass. Nylaflow tubing (4” and +” o.d.) is used to convey the carrier gas from the gas cylinder to the various parts. (f) A “gas tight Hamilton syringe” with a Chaney adaptor, and a 2 inch, 24 gauge needle are used to introduce the blood or gas samples into the extraction chamber. Our measurements were made with 0.5 and 1.0 ml syringes.
REAGENTS
The reagents used differ according to the kinds of gas to be measured. To analyze for N, the blood is stirred with twice its own volume of a previously prepared reagent containing the following ingredients : 1 part powder, containing sodium hydrosulfite (10 g) and sodium anthraquinone betasulfonate (1 g), and 5 parts 1 N potassium hydroxide solution. To analyze for 0, and CO,, the blood is stirred with twice its own volume of each of the following solutions: a
Potassium ferricyanide Saponin Water
20 g 2g 100 ml
MEASUREMENT
b
A drop of octanol PREPARATION
OF BLOOD
GASES
Sodium
sulfate
Sulfuric Water
acid
BY GAS CHROMATOGRAPHY
401
30 g 7 ml 100 ml
is added to all reagents.
FOR ANALYSIS
Several successive steps are identical both for calibrating with standard samples and for analyzing unknown samples. (a) Since the gas chromatograph is never turned off, the equipment is made ready simply by turning on the recorder, the stirrer and the detector filament. The carrier gas flow, which is kept low when the instrument is not in use, is increased to 30 lbs equivalent to a flow of approximately 50 ml/min. Stabilization is obtained within half an hour. For our measurements, the following settings were used: temperature 70 “C, current 400 mA. The isolating valve is now opened (flow-through position) and the carrier gas inlet and exhaust valves closed. In our experiments the pressure in the extraction chamber read 17 lbs. In order to simplify the description of the nextsteps, this figure of 17 lbs will be used, although it must be borne in mind that this pressure depends upon the arrangement of the equipment. At this time, the isolating valve is closed again (bypass position), and the carrier gas inlet valve and the extraction chamber sample inlet are opened. (b) The measured reagents and octanol are introduced into the extraction chamber through the sample inlet which is then sealed with a rubber cap. A 2” 24 gauge needle is stuck into the chamber through the rubber cap. The carrier gas, entering the chamber at the bottom, bubbles through the reagents and exits through the needle until the reagents are gas free (about 2 min). (c) The needle is rapidly removed and the carrier gas inlet valve closed. These two movements must be done quickly so that the pressure in the extraction chamber does not rise more than 1 pound. PREPARATION
OF BLOOD
SAMPLES
An anaerobic blood sample is drawn directly from the blood vessel, or from a tonometer, into a Luer-Lok syringe. In the first instance the syringe dead space is filled with heparin. After the sample has been collected, the tip of the syringe is submerged in mercury, a drop of which is drawn into the syringe. The opening is then covered with a cap tightly fitting the lock (B-D # 411 AC) in such a manner that no bubble is forced into the syringe. The syringe is stored at room temperature for N, analysis and in ice for O2 and CO, analysis. At the time of analysis, the blood is transferred into the Hamilton syringe by thrusting its needle through the cap into the Luer-Lok syringe. Then the two syringes are firmly held together and vigorously shaken to mix the blood which is then pushed into the Hamilton syringe. The same sample volume and same needle must be used for each series of measurements, including calibration. For each analysis the Hamilton
C. LENFANT
402
AND C. AUCUTT
syringe is rinsed and freed of bubbles with a small amount of the blood which is then discarded. When the sample is taken, a small excess volume is injected into the Hamilton syringe. Before introducing the blood sample into the extraction chamber, a drop of the excess blood is placed on top of the rubber cap covering the sample inlet to prevent microscopic bubbles from being trapped in the needle tip. If replicate analyses of the same sample are desired, the Hamilton syringe is immediately reinserted in the LuerLok syringe; otherwise it is washed with lukewarm saline.
ANALYSIS
(a) The blood sample prepared in the Hamilton syringe is introduced into the extraction chamber through the rubber cap. A timer is set for a time predetermined to assure equilibration of the extracted gases. The carrier gas inlet valve is opened to reinstate the 17 lbs pressure in the extraction chamber. (b) At the end of this time, the bypass valve is opened to let the carrier gas sweep the extracted gases to the columns where they will be separated. When the first peak appears on the recorder, the bypass valve is closed. (c) The carrier gas inlet and the exhaust valves are opened, the rubber cap is removed and the extraction chamber is cleaned several times with tap water which is flushed out through the exhaust by simply placing a finger tip on the sample inlet. The exhaust valve is then closed, and more reagents are introduced for the next sample. Each analysis requires 5+ to 6 min.
CALIBRATION
OF THE INSTRUMENT
Standard samples, different for each gas, are injected the response is measured in terms of peak height.
into the gas chromatograph
and
(a) For measuring oxygen, the instrument is calibrated‘with gas mixtures. Since the relationship between Fo, and peak height is linear, only two mixtures are necessary. One is room air which contains 20.93 % oxygen, and the other is a mixture of 10 % O2 and atmospheric nitrogen (argon 1.185 % and nitrogen 98.8 15 %. Handbook of Chemistry and Physics). For both gases it is important to moisten the Hamilton syringe in order to saturate the sample with water vapor. The syringe must also be rinsed many times with the standardizing sample before collecting the final sample. Two corrections must be made to arrive at a true calibration. First the volume percentage of 0, must be expressed in STPD conditions, thus the 0, concentration of the standard sample must be multiplied by the following factor: (1)
F, = ‘;6poHzo
in which P = the ambient the ambient temperature,
273 x 273+t barometric pressure, PHzO = the pressure and t = the ambient temperature in “C.
of water vapor at
MEASUREMENT
Second, the contribution subtracted.
If the response,
of gas analyzed
403
OF BLOOD GASESBY GAS CHROMATOGRAPHY
of argon to the peak common D, of the instrument
to both O2 and A must
is directly proportional
be
to the volume
Vo: D = Sensitivity
x Vo
or, since the volume of gas, Vo, equals the volume of the sample, V,, times the fraction of gas, D = Sensitivity x V, x F . Thus, if O2 is completely absorbed into the extraction chamber, the first argon in room air. Since the product depends only on F, which varies as nitrogen. Then F,/FN2 = DA/D,,
= constant
in a sample of room air after it has been injected peak, DA, is proportional only to the volume of (Sensitivity x V,) is the same in all samples, D, atmospheric F,, in any gas mixture containing
.
Once this ratio has been measured (it remains the same during contribution of argon to the 0, and A peak is given by: F, =
DJ! r!??f?z D,, room air
x D,,
successive
days), the
any gas mixture
and true Do, = Do, + A- F, . It is the plot of the corrected concentrations versus the corrected deflections which passes through zero and is used to calculate the 0, concentration in the blood samples. Since argon, as nitrogen, is only dissolved in blood, the argon correction need not be applied to the blood 0, peaks. (b) To calibrate the instrument for CO, bicarbonate solutions in distilled water are used. As for O,, the relationship between peak height and CO, concentration is linear and passes through zero. The range of CO, concentrations used for calibration depends on the expected values in the unknown samples. Simple calculations show that 0.378 g of dry, purified, bicarbonate must be dissolved in 1 liter of distilled water to make up a solution releasing 10 vol % of CO,. Any volume of CO, determined by this method of calibration is expressed in vol % STPD. (c) To calibrate for nitrogen equilibrated blood samples are used. The details pertaining to good equilibration have been described by FARHI et al. (1963). The only difference in our procedure is the introduction of a drop of mercury into the Luer-Lok syringe containing the equilibrated blood. As in Farhi’s experiments, the dead space of this syringe is not filled with heparin, but the syringe is filled and rinsed many times before the sample is collected. Two kinds of tonometers have been used with equal performance (FINLEY et al., 1960; FARHI, 1965) but that described by Farhi greatly simplifies the procedure. Results and discussion The results
presented
in this paper
constitute
an evaluation
of the method.
Two
404
C. LENFANT
AND C. AUCUTT
criteria were used: linearity of the gas chromatograph response to increasing concentration of a gas in a constant volume and reproducibility of a single analysis. RESPIRATORY
GASES (0,
AND c&)
The system’s linearity can be demonstrated by using one known gas sample as a standard to calculate the gas concentration in other known samples. Any deviation from linearity is shown by the difference between the calculated and known values. Typical results are shown in table 1 for O2 and CO,, The values given for 0, are those obtained after correction for STPD conditions and for argon. If the samples with the lowest concentration of 0, and CO, are used as standard, the differences between the actual and measured concentration are higher when expressed in vol % STPD but they would be the same when expressed in percentage of the actual concentrations. The values of 1.1% for 0, and 1.O% for CO, were the highest ever measured in our experience. The mean difference in percentage of actual concentration for a series of 25 calibration curves was 0.54% for 0, (I .lO to 0.12) and 0.60% for CO, (1.00-0.03). Th ese results indicate an extremely small deviation from perfect linearity. The reproducibility of 0, and CO, measurements in a given sample was assessed by determining the coefficient of variation (C.V. = standard deviation divided by the mean) in numerous gas, bicarbonates solutions and blood samples, all of which were analyzed from 3 to 6 times. As shown in table 2, the deviation from the mean is consistently less than 1% of the absolute value. The mean coefficient of variation TABLE 1 Evaluation Standard VOi
*A STPD
of linearity, with one sample used as a standard’ .--__ Measured Actual Difference Conceniration Concentration VOl ‘A STPD VOl
7; STPD
VOl
od STPD
~..._.. Oxygen 18.79 18.65 18.79 18.84 23.20 23.20
---
Difference in percentage of actual concentration _~__~__. ~ .
9.04 8.97 9.04 9.18 13.15 13.15
9.097 8.990 9.124 9.079 13.066 13.192
+ .057 + .ozo + .084 -.I01 - ,084 ,f .042
63 .22 .92 1.10 .64 .32
40.00 40.00 20.00 30.00 30.00 15.00
39.83 40.27 19.97 29.99 30.21 14.85
-.17 + .27 - .03 - .Ol +.21 -.15
.42 67 .15 .03 .70 1.00
Carbon Dioxide 60.00 60.00 60.00 45.00 45.00 45.00
__.--
---__1 All values are the mean of a duplicate or triplicate determination. was the same (0.4 ml).
In
all cases
the
volume
of sample
MEASUREMENT
OF BLOOD
405
GASESBY GAS CHROMATOGRAPHY TABLE 2
Reproducibility of 02 and CO2 measurements Nature of sample
Gas Gas Blood Blood Bicarbonates Blood
Measurement Range of concentrations vol 0%
02 02
02 02
CO2 co2
23.8-18.5 11.7- 8.8 22.5-15.7 8.0- 0.2 60.0-15.0 70.0-10.0
Number of samples
Volume of sample
Mean coefficient of variation percentage
Range of coefficient of variation percentage
20 20 20 20 40 40
04 0.4 0.4 0.2-0.4 0.4 0.2-0.4
0.22 0.31 0.28 0.34 0.27 0.35
0.71-0.09 0.78-0.12 0.83-0.08 0.92-0.07 0.67-0.11 0.87-0.16
indicates a good reproducibility. The less satisfactory results were obtained with the smaller blood samples. These were samples of blood from fishes and amphibians which have large red cells that precipitate rapidly. INERT GAS
(NJ
The linearity of the nitrogen calibration was assessed in the same manner as for 0,. The difference between the known and the estimated nitrogen partial pressure was always within +2.2 mm Hg in blood samples equilibrated with a gas mixture containing from 8.2 to 91.73 % nitrogen. In all cases air was used as the standard. The reproducibility in a single sample was established by repeating the analysis from 3 to 8 times for 120 samples equilibrated with standard gas mixtures. Fig. 2 shows the coefficient of variation (C.V.) for these samples as a function of the mean deflection. It can be seen that C.V. is inversely related to the mean indicating that the error must be larger in small blood samples and/or in samples equilibrated with a low in table 3 in which our P,,. The relationship between P,, and C.V. is demonstrated results have been grouped according to a relatively narrow range of F,, in the standard gas. Although the coefficient of variation increases, the absolute error in terms of mm Hg remains the same over the complete range of P,,. No significant relationship was established between the coefficient of variation and blood samples of various sizes (1 ml to 0.6 ml) equilibrated with the same gas mixture. Since small blood samples are more difficult to handle, we believe it important to use as large a sample as possible; the lower the expected P,, the more mandatory this rule becomes. Absorption of the respiratory gases is absolutely necessary since by this method the extracted gases are diluted in a large volume of carrier gas. Failure to absorb the respiratory gases leads to overlapping peaks whose shape may be distorted (FARHI et al., 1963). Some details of calibration and analysis were found to be extremely critical in obtaining satisfactory results for 02, CO, and Nz measurements, (a) All gas mixtures used for the 0, calibration, especially the compressed gas mixtures, must really be at ATPS.
C. LENFANT
406 COEFFICIENT
.50
I
of VARIATION
0
l
0
AND C. AUCUTT
(%)
: .
5
e .
.
8
0. .
. .
:
: I
0 .
.
.
.
. ** e
0
ok
.
too
50
150
200 DEFLECTION
.
250 (mm)
Fig. 2. Coefficient of variation as a function of the mean deflection of 3 to 8 replicates. The deflection depends on one or a combination of the following factors: P N* of the sample, size of sample, age and condition of the column and setting of the gas chromatograph (flow of carrier gas, temperature, current through the detector filament and attenuation). TABLE 3 Reproducibility Fraction NZ in standard gas Number of samples Range of coefficient of variation % Mean C.V. % Absolute error1 mm Hg
of PN~ measurements
0.9330.90
0.82-0.75
0.60-0.50
0.401
9 0.45-O. 16
56 0.5990.11
5 0.66-0.40
4 1.34-0.37
0.54 2.08
0.73 2.00
0.32 2.02
0.33 1.78
r Calculated for the mean FN, in each group, and in the case of P -
0.26-0.20
0.16-0.08
20 1.55-0.54
26 1S2-0.33
0.92 I .59
0.96 0.9
Pn,o = 700 mm Hg.
(b) When filled with standard gases the Hamilton syringe barrel must not be held too long in the hands or volume may vary due to heat transfer. (c) Precise weighing and dilution of bicarbonates are very important. (d) The temperature of equilibration as compared to that of the unknown samples is essential to calibrate the instrument for N, measurements (FARHI et al., 1963). (e) Pressure in the extraction chamber must be low when the samples are injected, and must remain the same for all the gas samples, otherwise unequal volumes of gas are injected.
MEASUREMENT OF BLOOD GASES BY GAS CHROMATOGRAPHY
(f) Pressure
in the extraction
chamber
during
407
the bypass must be the same for all
samples, gaseous or liquid, since the peak shape depends on the volume of carrier gas in which the sample is diluted. (g) Duration of bypass must be the same for all samples and must not be shorter than a predetermined value (3 min in our experiments). (h) Contamination of the gas samples and microscopic bubbles in the blood samples are by far the most important causes of error and the easiest to make. (i) The blood samples must be well mixed in the Luer-Lok syringe before the Hamilton syringe is filled which, in turn, should be emptied without delay into the extraction chamber. For O2 and CO, measurements the reason for this is evident, for measurement of N, it is necessary because of the difference in N, coefficient of solubility between red cells and plasma (VAN SLYKE, DILLON and MARGARIA, 1934; FARHI et al., 1963). (j) Leakage through the rubber cap on the sample inlet was observed following repeated puncture. It is advisable to change the cap after every other analysis. In conclusion, this method provides an easy means of measuring 0, and CO, concentration and the nitrogen tension in small blood samples. The procedure can be mastered with reasonable training and does not exceed in difficulty or time required that demanded for any other technique. Calibration of the instrument is simple; it reveals in samples of same volume a linear relationship between gas concentration and the peak height. The calibration line passes through zero for all gases. The reproducibility of a single analysis in the same sample is very satisfactory and is not affected by the concentration of gas in the sample.
References L. E. (1965). Continuous duty tonometer system. J. Appl. Physiol. 20: 1098-l 101. FARHI, L. E., A. W. T. EDWARDSand T. HOMMA(1963). Determination of dissolved Nz in blood by gas chromatography and (a-A) N2 difference. J. Appl. Physiol. 18 : 97-106. FINLEY,T. N., C. LENFANT,P. HAAB, J. PIIPERand H. RAHN (1960). Venous admixture in the pulmonary circulation of anesthetized dogs. J. Appl. Physiol. 1.5: 418-424. GALLA, S. J. and D. M. OI-~ENSTEIN(1962). Measurement of inert gases in blood by gas chromatography. Ann. N. Y. Acad. Sci. 102 : 4-14. HAMILTON,L. H. (1962). Gas chromatography for respiratory and blood gas analysis. Ann. N. Y. FARHI,
Acad.
Sci. 102: 15-28.
Handbook of Chemistry and Physics (42nd edition) 1960. Cleveland, Ohio, Chemical Rubber Publ. IKELS, K. G. (1964). Determination of the solubility of nitrogen in water and extracted human fat. J. Gas Chromatogr.
2: 374-379.
LUKAS,D. S. and S. M. AYRES(1961). Determination J. Appl. Physiol.
of blood oxygencontent
by gaschromatography.
16: 371-374.
RAMSEY,L. H. (1959). Analysis of gas in biological fluids by gas chromatography. Science 129: 900-901. VAN SLYKE,D. D., R. T. DILLONand R. MARGARIA(1934). Studies of gas and electrolyte equilibria in blood. XVIII. Solubility and physical state of atmospheric nitrogen in blood cells and plasma. J. Biol. Chem. 105: 571-596. WILSON,R. H. (1964). Biologic application of gas chromatography. J. Gas Chromatogr. 2: 365-373. WILSON,R. H., B. JAY, V. DOTY, H. PINGREEand E. HIGGINS(1961). Analysis of blood gases with gas adsorption chromatographic technique. J. Appl. Phvsiol. 16: 374-377.
Physiology
MEASUREMENT
(1966) 1, 398-407;
OF
BLOOD
North-Holland
GASES
BY
Publishing
GAS
Company,
Amsterdam
CHROMATOGRAPHY’
C. LENFANT AND C. AUCUTT Institute
of Respiratory
and of Physiology
Physiology,
and Biophysics,
Firland Sanatorium,
University
and Departments
of Washington,
Seattle,
of Medicine
Washington,
U.S.A.
Abstract. A method using gas chromatography to measure blood gases has been developed. Equipment and procedure are described. Respiratory gases, 0~ and COz, and Nz, can be measured, although these analyses should be performed separately. For all three gases thecalibration is linear and passes through zero. The maximum deviation from linearity was found to be 0.1 vol ‘A for Coz, 0.3 vol ‘A for CCO, and 2.2 mm Hg for PN~. The average reproducibility tested on many samples was within -CO.06 vol ‘A for 02, 5 0.18 vol % for CO2 and j, 2. mm Hg for PN~. Gas chromatography Measurement of respiratory
Measurement
of dissolved gases in blood
gases in blood
The major difficulty in measuring blood gases by gas chromatography is obtaining rapid and complete release of the gases from the blood, including those bound to the respiratory pigment. The methods available to extract gases from blood, or any other liquid, are of three types: (1) the “vacuum extraction” method (RAMSEY, 1959; LUCAS and AYRES, 1961; FARHI, EDWARDS and HOMMA, 1963). In this method the gases are extracted from the blood prior to being introduced into the gas chromatograph, thus fast elution and good resolution of the peaks are easily obtained. It requires, however, in addition to the gas chromatograph, a Van Slyke apparatus, which is itself a source of problems. Also a fraction of the gases always is retained in the liquid remaining in the Van Slyke extraction chamber. (2) The “washout” or “gas flow-through” method (WILSON et al., 1961; GALLA and OTTENSTEIN, 1962; WILSON, 1964). In this method the sample is injected directly into the carrier gas stream. The procedure is very simple and all gases eventually will be released from the blood. However, resolution is poor and elution is slow. Since the peaks cannot be measured accurately in terms of peak height, it is necessary to determine peak surface area by integration or approximation. (3) The “diffusion process” or “equilibration” method (HAMILTON, 1962). This method combines the advantages of the two preceding techniques. Accepted for publication I June 1966. 1 Supported by National Institute of Health Grants H.E. 08465 and H.E. 01892.
398
MEASUREMENT
A version
OF BLOOD
of this latter method
GASES
is described
BY GAS CHROMATOGRAPHY
here and its accuracy
399 is appraised.
Method PRINCIPLE
A blood sample, with gas-free reagents added, is stirred in an extraction chamber isolated from the carrier gas line. If this chamber contains an atmosphere of He, or any other carrier gas different from those to be extracted, a vacuum exists relative to the gases contained in the blood sample. The gas molecules then diffuse from the blood into the surrounding gas phase until equilibrium is reached. At this point the carrier gas is allowed to flow through the extraction chamber, to the chromatograph columns and detector, transporting with it both the extracted gases and all gases which are still retained in the blood at the end of the equilibration period. This method offers the following advantages: all gases are extracted with certainty, even those with a high partition coefficient, and sharp narrow peaks are produced making possible simple quantitative measurements in terms of peak height. EQUIPMENT
A Beckman gas chromatograph Model 2A and a Bristol 1 mV full scale recorder are used. The chromatograph is equipped with two parallel columns, one packed with silica gel and the other with molecular sieve SA. The length of the columns is adjusted to provide good resolution and fast elution. When only N, is measured, it is preferable
IO
inside ,-_-.__-_--
capillary
-
I / I
PRESSURE REGULATOR
columns
G C / 2 A
dessicotor
prrssure gage blood sample inlet ‘\
gas
outlet EXTRACTION
gas l
P r. Larricr
capillary
Fig. 1. Schematic representation
inlet \
gas inlet valve
exhaust
CHAMBER
valve
.-MOTOR
of the blood gas extraction chamber and accessory
apparatus.
400
C. LENFANT
AND C. AUCUTT
to eliminate the silica gel column. Either column can be reactivated by baking it overnight at 210 “C. The blood extraction chamber and accessory apparatus (fig. l), built in our laboratory, consist of the following components: (a) The extraction chamber is made up of two parts. A glass cover (Fisher # 6.390-10) is hermetically sealed on a brass reservoir. The cover has a sample inlet and a gas outlet. The reservoir has a carrier gas inlet and an exhaust for discarding the analyzed blood. The extraction chamber contains a magnetic stirrer activated by a small phonograph motor mounted beneath it. (b) There are 3 valves in the accessory apparatus: the isolating or bypass valve, the carrier gas inlet valve, and the exhaust valve. The isolating or bypass valve (Republic valves 330 series or Beckman #23800) is a four way valve designed to isolate the extraction chamber from the main carrier gas line. The carrier gas inlet valve (toggle valve-Hoke 490 series) is opened only to free the reagents from their dissolved gases and to permit the pressure in the extraction chamber to equilibrate with that of the main carrier gas line after the sample has been introduced. The flow of carrier gas through this valve is regulated by a capillary tube interposed just ahead of the valve. The exhaust valve (toggle valve-Hoke 490 series) is opened to flush out the analyzed blood and to wash out the chamber. (c) The pressure gauge, graduated in 4 pounds up to 30 pounds is used to measure the pressure in the carrier gas line and in the extraction chamber, and to monitor the system for leaks. (d) The dessicator is filled with silica gel and is used only when N, is measured. It is essential that the dessicator be inserted beyond the bypass valve. (e) These various components are connected with Swagelok fittings of brass or nylon when they are in contact with glass. Nylaflow tubing (4” and +” o.d.) is used to convey the carrier gas from the gas cylinder to the various parts. (f) A “gas tight Hamilton syringe” with a Chaney adaptor, and a 2 inch, 24 gauge needle are used to introduce the blood or gas samples into the extraction chamber. Our measurements were made with 0.5 and 1.0 ml syringes.
REAGENTS
The reagents used differ according to the kinds of gas to be measured. To analyze for N, the blood is stirred with twice its own volume of a previously prepared reagent containing the following ingredients : 1 part powder, containing sodium hydrosulfite (10 g) and sodium anthraquinone betasulfonate (1 g), and 5 parts 1 N potassium hydroxide solution. To analyze for 0, and CO,, the blood is stirred with twice its own volume of each of the following solutions: a
Potassium ferricyanide Saponin Water
20 g 2g 100 ml
MEASUREMENT
b
A drop of octanol PREPARATION
OF BLOOD
GASES
Sodium
sulfate
Sulfuric Water
acid
BY GAS CHROMATOGRAPHY
401
30 g 7 ml 100 ml
is added to all reagents.
FOR ANALYSIS
Several successive steps are identical both for calibrating with standard samples and for analyzing unknown samples. (a) Since the gas chromatograph is never turned off, the equipment is made ready simply by turning on the recorder, the stirrer and the detector filament. The carrier gas flow, which is kept low when the instrument is not in use, is increased to 30 lbs equivalent to a flow of approximately 50 ml/min. Stabilization is obtained within half an hour. For our measurements, the following settings were used: temperature 70 “C, current 400 mA. The isolating valve is now opened (flow-through position) and the carrier gas inlet and exhaust valves closed. In our experiments the pressure in the extraction chamber read 17 lbs. In order to simplify the description of the nextsteps, this figure of 17 lbs will be used, although it must be borne in mind that this pressure depends upon the arrangement of the equipment. At this time, the isolating valve is closed again (bypass position), and the carrier gas inlet valve and the extraction chamber sample inlet are opened. (b) The measured reagents and octanol are introduced into the extraction chamber through the sample inlet which is then sealed with a rubber cap. A 2” 24 gauge needle is stuck into the chamber through the rubber cap. The carrier gas, entering the chamber at the bottom, bubbles through the reagents and exits through the needle until the reagents are gas free (about 2 min). (c) The needle is rapidly removed and the carrier gas inlet valve closed. These two movements must be done quickly so that the pressure in the extraction chamber does not rise more than 1 pound. PREPARATION
OF BLOOD
SAMPLES
An anaerobic blood sample is drawn directly from the blood vessel, or from a tonometer, into a Luer-Lok syringe. In the first instance the syringe dead space is filled with heparin. After the sample has been collected, the tip of the syringe is submerged in mercury, a drop of which is drawn into the syringe. The opening is then covered with a cap tightly fitting the lock (B-D # 411 AC) in such a manner that no bubble is forced into the syringe. The syringe is stored at room temperature for N, analysis and in ice for O2 and CO, analysis. At the time of analysis, the blood is transferred into the Hamilton syringe by thrusting its needle through the cap into the Luer-Lok syringe. Then the two syringes are firmly held together and vigorously shaken to mix the blood which is then pushed into the Hamilton syringe. The same sample volume and same needle must be used for each series of measurements, including calibration. For each analysis the Hamilton
C. LENFANT
402
AND C. AUCUTT
syringe is rinsed and freed of bubbles with a small amount of the blood which is then discarded. When the sample is taken, a small excess volume is injected into the Hamilton syringe. Before introducing the blood sample into the extraction chamber, a drop of the excess blood is placed on top of the rubber cap covering the sample inlet to prevent microscopic bubbles from being trapped in the needle tip. If replicate analyses of the same sample are desired, the Hamilton syringe is immediately reinserted in the LuerLok syringe; otherwise it is washed with lukewarm saline.
ANALYSIS
(a) The blood sample prepared in the Hamilton syringe is introduced into the extraction chamber through the rubber cap. A timer is set for a time predetermined to assure equilibration of the extracted gases. The carrier gas inlet valve is opened to reinstate the 17 lbs pressure in the extraction chamber. (b) At the end of this time, the bypass valve is opened to let the carrier gas sweep the extracted gases to the columns where they will be separated. When the first peak appears on the recorder, the bypass valve is closed. (c) The carrier gas inlet and the exhaust valves are opened, the rubber cap is removed and the extraction chamber is cleaned several times with tap water which is flushed out through the exhaust by simply placing a finger tip on the sample inlet. The exhaust valve is then closed, and more reagents are introduced for the next sample. Each analysis requires 5+ to 6 min.
CALIBRATION
OF THE INSTRUMENT
Standard samples, different for each gas, are injected the response is measured in terms of peak height.
into the gas chromatograph
and
(a) For measuring oxygen, the instrument is calibrated‘with gas mixtures. Since the relationship between Fo, and peak height is linear, only two mixtures are necessary. One is room air which contains 20.93 % oxygen, and the other is a mixture of 10 % O2 and atmospheric nitrogen (argon 1.185 % and nitrogen 98.8 15 %. Handbook of Chemistry and Physics). For both gases it is important to moisten the Hamilton syringe in order to saturate the sample with water vapor. The syringe must also be rinsed many times with the standardizing sample before collecting the final sample. Two corrections must be made to arrive at a true calibration. First the volume percentage of 0, must be expressed in STPD conditions, thus the 0, concentration of the standard sample must be multiplied by the following factor: (1)
F, = ‘;6poHzo
in which P = the ambient the ambient temperature,
273 x 273+t barometric pressure, PHzO = the pressure and t = the ambient temperature in “C.
of water vapor at
MEASUREMENT
Second, the contribution subtracted.
If the response,
of gas analyzed
403
OF BLOOD GASESBY GAS CHROMATOGRAPHY
of argon to the peak common D, of the instrument
to both O2 and A must
is directly proportional
be
to the volume
Vo: D = Sensitivity
x Vo
or, since the volume of gas, Vo, equals the volume of the sample, V,, times the fraction of gas, D = Sensitivity x V, x F . Thus, if O2 is completely absorbed into the extraction chamber, the first argon in room air. Since the product depends only on F, which varies as nitrogen. Then F,/FN2 = DA/D,,
= constant
in a sample of room air after it has been injected peak, DA, is proportional only to the volume of (Sensitivity x V,) is the same in all samples, D, atmospheric F,, in any gas mixture containing
.
Once this ratio has been measured (it remains the same during contribution of argon to the 0, and A peak is given by: F, =
DJ! r!??f?z D,, room air
x D,,
successive
days), the
any gas mixture
and true Do, = Do, + A- F, . It is the plot of the corrected concentrations versus the corrected deflections which passes through zero and is used to calculate the 0, concentration in the blood samples. Since argon, as nitrogen, is only dissolved in blood, the argon correction need not be applied to the blood 0, peaks. (b) To calibrate the instrument for CO, bicarbonate solutions in distilled water are used. As for O,, the relationship between peak height and CO, concentration is linear and passes through zero. The range of CO, concentrations used for calibration depends on the expected values in the unknown samples. Simple calculations show that 0.378 g of dry, purified, bicarbonate must be dissolved in 1 liter of distilled water to make up a solution releasing 10 vol % of CO,. Any volume of CO, determined by this method of calibration is expressed in vol % STPD. (c) To calibrate for nitrogen equilibrated blood samples are used. The details pertaining to good equilibration have been described by FARHI et al. (1963). The only difference in our procedure is the introduction of a drop of mercury into the Luer-Lok syringe containing the equilibrated blood. As in Farhi’s experiments, the dead space of this syringe is not filled with heparin, but the syringe is filled and rinsed many times before the sample is collected. Two kinds of tonometers have been used with equal performance (FINLEY et al., 1960; FARHI, 1965) but that described by Farhi greatly simplifies the procedure. Results and discussion The results
presented
in this paper
constitute
an evaluation
of the method.
Two
404
C. LENFANT
AND C. AUCUTT
criteria were used: linearity of the gas chromatograph response to increasing concentration of a gas in a constant volume and reproducibility of a single analysis. RESPIRATORY
GASES (0,
AND c&)
The system’s linearity can be demonstrated by using one known gas sample as a standard to calculate the gas concentration in other known samples. Any deviation from linearity is shown by the difference between the calculated and known values. Typical results are shown in table 1 for O2 and CO,, The values given for 0, are those obtained after correction for STPD conditions and for argon. If the samples with the lowest concentration of 0, and CO, are used as standard, the differences between the actual and measured concentration are higher when expressed in vol % STPD but they would be the same when expressed in percentage of the actual concentrations. The values of 1.1% for 0, and 1.O% for CO, were the highest ever measured in our experience. The mean difference in percentage of actual concentration for a series of 25 calibration curves was 0.54% for 0, (I .lO to 0.12) and 0.60% for CO, (1.00-0.03). Th ese results indicate an extremely small deviation from perfect linearity. The reproducibility of 0, and CO, measurements in a given sample was assessed by determining the coefficient of variation (C.V. = standard deviation divided by the mean) in numerous gas, bicarbonates solutions and blood samples, all of which were analyzed from 3 to 6 times. As shown in table 2, the deviation from the mean is consistently less than 1% of the absolute value. The mean coefficient of variation TABLE 1 Evaluation Standard VOi
*A STPD
of linearity, with one sample used as a standard’ .--__ Measured Actual Difference Conceniration Concentration VOl ‘A STPD VOl
7; STPD
VOl
od STPD
~..._.. Oxygen 18.79 18.65 18.79 18.84 23.20 23.20
---
Difference in percentage of actual concentration _~__~__. ~ .
9.04 8.97 9.04 9.18 13.15 13.15
9.097 8.990 9.124 9.079 13.066 13.192
+ .057 + .ozo + .084 -.I01 - ,084 ,f .042
63 .22 .92 1.10 .64 .32
40.00 40.00 20.00 30.00 30.00 15.00
39.83 40.27 19.97 29.99 30.21 14.85
-.17 + .27 - .03 - .Ol +.21 -.15
.42 67 .15 .03 .70 1.00
Carbon Dioxide 60.00 60.00 60.00 45.00 45.00 45.00
__.--
---__1 All values are the mean of a duplicate or triplicate determination. was the same (0.4 ml).
In
all cases
the
volume
of sample
MEASUREMENT
OF BLOOD
405
GASESBY GAS CHROMATOGRAPHY TABLE 2
Reproducibility of 02 and CO2 measurements Nature of sample
Gas Gas Blood Blood Bicarbonates Blood
Measurement Range of concentrations vol 0%
02 02
02 02
CO2 co2
23.8-18.5 11.7- 8.8 22.5-15.7 8.0- 0.2 60.0-15.0 70.0-10.0
Number of samples
Volume of sample
Mean coefficient of variation percentage
Range of coefficient of variation percentage
20 20 20 20 40 40
04 0.4 0.4 0.2-0.4 0.4 0.2-0.4
0.22 0.31 0.28 0.34 0.27 0.35
0.71-0.09 0.78-0.12 0.83-0.08 0.92-0.07 0.67-0.11 0.87-0.16
indicates a good reproducibility. The less satisfactory results were obtained with the smaller blood samples. These were samples of blood from fishes and amphibians which have large red cells that precipitate rapidly. INERT GAS
(NJ
The linearity of the nitrogen calibration was assessed in the same manner as for 0,. The difference between the known and the estimated nitrogen partial pressure was always within +2.2 mm Hg in blood samples equilibrated with a gas mixture containing from 8.2 to 91.73 % nitrogen. In all cases air was used as the standard. The reproducibility in a single sample was established by repeating the analysis from 3 to 8 times for 120 samples equilibrated with standard gas mixtures. Fig. 2 shows the coefficient of variation (C.V.) for these samples as a function of the mean deflection. It can be seen that C.V. is inversely related to the mean indicating that the error must be larger in small blood samples and/or in samples equilibrated with a low in table 3 in which our P,,. The relationship between P,, and C.V. is demonstrated results have been grouped according to a relatively narrow range of F,, in the standard gas. Although the coefficient of variation increases, the absolute error in terms of mm Hg remains the same over the complete range of P,,. No significant relationship was established between the coefficient of variation and blood samples of various sizes (1 ml to 0.6 ml) equilibrated with the same gas mixture. Since small blood samples are more difficult to handle, we believe it important to use as large a sample as possible; the lower the expected P,, the more mandatory this rule becomes. Absorption of the respiratory gases is absolutely necessary since by this method the extracted gases are diluted in a large volume of carrier gas. Failure to absorb the respiratory gases leads to overlapping peaks whose shape may be distorted (FARHI et al., 1963). Some details of calibration and analysis were found to be extremely critical in obtaining satisfactory results for 02, CO, and Nz measurements, (a) All gas mixtures used for the 0, calibration, especially the compressed gas mixtures, must really be at ATPS.
C. LENFANT
406 COEFFICIENT
.50
I
of VARIATION
0
l
0
AND C. AUCUTT
(%)
: .
5
e .
.
8
0. .
. .
:
: I
0 .
.
.
.
. ** e
0
ok
.
too
50
150
200 DEFLECTION
.
250 (mm)
Fig. 2. Coefficient of variation as a function of the mean deflection of 3 to 8 replicates. The deflection depends on one or a combination of the following factors: P N* of the sample, size of sample, age and condition of the column and setting of the gas chromatograph (flow of carrier gas, temperature, current through the detector filament and attenuation). TABLE 3 Reproducibility Fraction NZ in standard gas Number of samples Range of coefficient of variation % Mean C.V. % Absolute error1 mm Hg
of PN~ measurements
0.9330.90
0.82-0.75
0.60-0.50
0.401
9 0.45-O. 16
56 0.5990.11
5 0.66-0.40
4 1.34-0.37
0.54 2.08
0.73 2.00
0.32 2.02
0.33 1.78
r Calculated for the mean FN, in each group, and in the case of P -
0.26-0.20
0.16-0.08
20 1.55-0.54
26 1S2-0.33
0.92 I .59
0.96 0.9
Pn,o = 700 mm Hg.
(b) When filled with standard gases the Hamilton syringe barrel must not be held too long in the hands or volume may vary due to heat transfer. (c) Precise weighing and dilution of bicarbonates are very important. (d) The temperature of equilibration as compared to that of the unknown samples is essential to calibrate the instrument for N, measurements (FARHI et al., 1963). (e) Pressure in the extraction chamber must be low when the samples are injected, and must remain the same for all the gas samples, otherwise unequal volumes of gas are injected.
MEASUREMENT OF BLOOD GASES BY GAS CHROMATOGRAPHY
(f) Pressure
in the extraction
chamber
during
407
the bypass must be the same for all
samples, gaseous or liquid, since the peak shape depends on the volume of carrier gas in which the sample is diluted. (g) Duration of bypass must be the same for all samples and must not be shorter than a predetermined value (3 min in our experiments). (h) Contamination of the gas samples and microscopic bubbles in the blood samples are by far the most important causes of error and the easiest to make. (i) The blood samples must be well mixed in the Luer-Lok syringe before the Hamilton syringe is filled which, in turn, should be emptied without delay into the extraction chamber. For O2 and CO, measurements the reason for this is evident, for measurement of N, it is necessary because of the difference in N, coefficient of solubility between red cells and plasma (VAN SLYKE, DILLON and MARGARIA, 1934; FARHI et al., 1963). (j) Leakage through the rubber cap on the sample inlet was observed following repeated puncture. It is advisable to change the cap after every other analysis. In conclusion, this method provides an easy means of measuring 0, and CO, concentration and the nitrogen tension in small blood samples. The procedure can be mastered with reasonable training and does not exceed in difficulty or time required that demanded for any other technique. Calibration of the instrument is simple; it reveals in samples of same volume a linear relationship between gas concentration and the peak height. The calibration line passes through zero for all gases. The reproducibility of a single analysis in the same sample is very satisfactory and is not affected by the concentration of gas in the sample.
References L. E. (1965). Continuous duty tonometer system. J. Appl. Physiol. 20: 1098-l 101. FARHI, L. E., A. W. T. EDWARDSand T. HOMMA(1963). Determination of dissolved Nz in blood by gas chromatography and (a-A) N2 difference. J. Appl. Physiol. 18 : 97-106. FINLEY,T. N., C. LENFANT,P. HAAB, J. PIIPERand H. RAHN (1960). Venous admixture in the pulmonary circulation of anesthetized dogs. J. Appl. Physiol. 1.5: 418-424. GALLA, S. J. and D. M. OI-~ENSTEIN(1962). Measurement of inert gases in blood by gas chromatography. Ann. N. Y. Acad. Sci. 102 : 4-14. HAMILTON,L. H. (1962). Gas chromatography for respiratory and blood gas analysis. Ann. N. Y. FARHI,
Acad.
Sci. 102: 15-28.
Handbook of Chemistry and Physics (42nd edition) 1960. Cleveland, Ohio, Chemical Rubber Publ. IKELS, K. G. (1964). Determination of the solubility of nitrogen in water and extracted human fat. J. Gas Chromatogr.
2: 374-379.
LUKAS,D. S. and S. M. AYRES(1961). Determination J. Appl. Physiol.
of blood oxygencontent
by gaschromatography.
16: 371-374.
RAMSEY,L. H. (1959). Analysis of gas in biological fluids by gas chromatography. Science 129: 900-901. VAN SLYKE,D. D., R. T. DILLONand R. MARGARIA(1934). Studies of gas and electrolyte equilibria in blood. XVIII. Solubility and physical state of atmospheric nitrogen in blood cells and plasma. J. Biol. Chem. 105: 571-596. WILSON,R. H. (1964). Biologic application of gas chromatography. J. Gas Chromatogr. 2: 365-373. WILSON,R. H., B. JAY, V. DOTY, H. PINGREEand E. HIGGINS(1961). Analysis of blood gases with gas adsorption chromatographic technique. J. Appl. Phvsiol. 16: 374-377.