Single-breath transfer factor in young healthy adults: 21% or 17.5% inspired oxygen?

Transcription

1 Eur Respir J 2004; 23: DOI: / Printed in UK a rights reserved Copyright #ERS Journas Ltd 2004 European Respiratory Journa ISSN Singe-breath transfer factor in young heathy aduts: 21% or 17.5% inspired oxygen? H. Normand*, C. Marie #, A. Mouadi # Singe-breath transfer factor in young heathy aduts: 21% or 17.5% inspired oxygen? H. Normand, C. Marie, A. Mouadi. #ERS Journas Ltd ABSTRACT: For the measurement of the singe-breath transfer factor of the ung for carbon monoxide (TL,CO,sb), the American Thoracic Society (ATS) recommends using a test gas with a 21% inspired fraction of oxygen (FI,O 2 ) whereas the European Respiratory Society (ERS) expressy recommends 17 18% FI,O 2. The ERS committee argues that with a higher concentration (e.g. 21%) the aveoar fraction of oxygen (and accordingy TL,CO,sb) "varies with the voume of the test gas which is inspired", that is, presumaby, in proportion to the voume of the test gas that is diuted in the aveoar voume. The current study measured TL,CO,sb and the transfer coefficient (KCO,sb) in dupicate in 67 heathy aduts (age yrs) using, in random order, an inspired gas containing either 17.5% or 21% oxygen. A correction was appied for carboxyhaemogobin, in ine with ATS recommendations. As expected, TL,CO,sb was higher with 17.5% FI,O 2 test gas compared with 21% FI,O 2 test gas ( versus mmo?min -1?kPa -1, respectivey) as we as KCO,sb ( versus mmo?min -1?kPa -1, respectivey). The ratio of TL,CO,sb measurements was stricty independent of the residua voume/tota ung capacity ratio measured with pethysmography. Hence, the rationae used by the European Respiratory Society, which utiises a 17 18% inspired fraction of oxygen test gas for singe-breath transfer of the ung for carbon monoxide measurements, woud appear to be unwarranted in young heathy aduts. Eur Respir J 2004; 23: *Laboratoire de Physioogie, Facuté demédecine de Caen, and # Service des Exporations Fonctionnees, Centre Hospitaier Universitaire de Caen, Caen, France. Correspondence: H. Normand Laboratoire de Physioogie Facuté de Médecine de Caen Avenue de a Côte de Nacre CAEN Cedex France Fax: E-mai: normand-h@chu-caen.fr Keywords: Methods pumonary diffusing capacity recommendation scientific societies Received: Juy Accepted after revision: January Measurement of the transfer factor for carbon monoxide with the singe-breath technique (TL,CO,sb) is infuenced by severa factors reating to: the equipment, how the operation is performed, the methods of cacuation and the subject9s persona characteristics. Updated recommendations for a standard technique have been issued by the European Respiratory Society (ERS) [1] and by the American Thoracic Society (ATS) [2]. However, athough the recommendations are simiar, there is sti controversy over the inspired fraction of oxygen (FI,O 2 ) that shoud be used for the test gas. The partia pressure of oxygen in erythrocytes contributes to determining the reaction rate of carbon monoxide with haemogobin and, as a resut, affects the transfer factor. As the capiary oxygen partia pressure is determined mainy by the aveoar oxygen tension (PA,O 2 ), this quantity has to be standardised. The ATS committee recommends using a test gas FI,O 2 of 21%, based on standard practice in the USA, arguing that no data are avaiabe to suggest a preference for European or American methods [2]. The ERS committee expressy recommends against using a test gas FI,O 2 of 21%, "a higher concentration (e.g. 21%) is not recommended because the PA,O 2 then varies with the voume of the test gas which is inspired" [1]. Presumaby the committee means that, when the test gas FI,O 2 is different from the aveoar fraction of oxygen (FA,O 2 ), the PA,O 2 varies in proportion to the test gas that is diuted in the aveoar voume. As the inspired voume (vita capacity (VC)) is diuted in the residua voume (RV), the PA,O 2 shoud then be correated to the ratio of the RV/tota ung capacity (TLC). Therefore, the recommendation of the ERS committee woud ceary be justified if the RV/TLC ratio affected TL,CO,sb with FI,O 2 at 21%, whist not doing so with a vaue of FI,O 2 at 17 18%. The aims of the study were to measure TL,CO,sb in a group of heathy subjects using two FI,O 2 test gases and to determine whether, in this popuation, the RV/TLC ratio affected TL,CO,sb differenty, depending on the FI,O 2. Subjects and study protoco Methods Sixty-seven young aduts (31 femaes, 36 maes; age yrs), with no history of cardiac or respiratory disease, gave their informed consent and served as heathy vounteers. A subjects were students in the Facuty of Medicine, which faciitated data coection during practica teaching sessions, foowing a procedure approved by the University Counci. The whoe cass (77 students) was enroed in the teaching session but ony 67 heathy students were incuded in the study. The tests were conducted between 10:00 and 13:30 h. A measurements were competed within 2 months. Linearity

2 928 H. NORMAND ET AL. of the carbon monoxide (CO) and heium (He) anaysers was checked before the session. Error accounted for v0.5% fu scae for the CO anayser and v1% fu scae for the He anayser. Subjects underwent whoe body pethysmography, foowed by four measurements of TL,CO,sb. Two measurements were taken with a 17.5% FI,O 2 test gas and two with a 21% FI,O 2 test gas. The order of the gas was reversed every two students. Despite secondary excusion of the non-heathy students, the randomisation scheme was respected. Thirty-four vounteers (20 maes) began with the 17.5% FI,O 2 test gas and 33 vounteers (16 maes) began with the 21% FI,O 2 test gas. Before the tests, the subjects breathed norma room air, as TL,CO,sb is considered to be independent of the pre-test FI,O 2 condition [3]. Haemogobin (Hb) and carboxyhaemogobin (HbCO) were measured in dupicate on bood capiary microsampes before and after the TL,CO,sb measurements, with two different oximeters (Radiometer OSM3; Radiometer, Copenhagen, Denmark and AVL OMNI6; AVL, Graz, Austria). The mean pre-test and post-test vaues are reported in this study. Whoe body pethysmography Intra-thoracic gas voume was measured near functiona residua capacity with a barometric whoe body pethysmograph (CompactLab; Jaeger, Wuerzburg, Germany) during ow frequency panting (typicay 1 Hz), then, whie sti connected to the pneumotachograph, the subject inspired to TLC then expired to RV, aowing measurement of these voumes. Transfer factor Subjects stayed seated, at rest for 5 min before the TL,CO,sb measurement began, with a 5-min break between trias. To minimise the number of repetitions, the procedure was carefuy expained to the subjects. For each test gas, TL,CO,sb was cacuated as the mean of the first two technicay acceptabe trias (rapid inspiration to w90% of the VC, rapid expiration immediatey foowing the shutter opening). The 17.5% FI,O 2 test gas contained 17.5% oxygen (O 2 ); % CO; 9.05% He; and baance nitrogen (N 2 ), whist the 21% FI,O 2 test gas contained 21.05% O 2 ; % CO; 9.15% He; and baance N 2 (a 2% reative); gases were suppied by Air Liquid Santé, Paris, France. The 17.5% FI,O 2 test gas (which was not the standard gas used in the aboratory) was provided free of charge by the suppier (Air Liquid Santé). A tests were carried out on the same commercia apparatus (MasterLab transfer; Jaeger), with identica settings, incuding a 9-s occusion time (breath-hod time: s), a 0.75-L washout voume and a 0.75-L sampe voume. These settings were chosen because they are common to the ERS and ATS. Anatomica dead space was cacuated as 2.26body weight (kg), corrections were appied for instrumenta dead spaces, carbon dioxide (CO 2 ) and water (H 2 O) absorption according to the ATS, assuming the aveoar fraction of carbon dioxide (FA,CO 2 ) to be The CO (infra-red) and He (therma conductivity) anaysers were caibrated with air and the corresponding inspired gas before each set of trias, to minimise the effect of varying oxygen concentration on the He anayser [4]. Subjects were instructed to reax against the shutter during the apnoea. TL,CO,sb corrected for HbCO, according to the ATS, with HbCO-adjusted TL,CO,sb cacuated as foows, assuming a inear increase during successive tests: HbCO{adjusted T L,CO,sb ~ measured T L,CO,sb 1z % HbCO 100 This assumption appeared reasonabe, the mean cacuated increase in HbCO per test ( %, absoute vaue, mean SD) being simiar whether the subject performed four, five or six tests. Corrections were appied to a tests, incuding the first one, because the ATS does not recommend an adjustment for the soe increase in HbCO above the baseine vaue. No correction was made for Hb. The singe-breath transfer coefficient (KCO,sb) was cacuated for each tria as the ratio of TL,CO,sb and the effective aveoar voume (VA,eff) measured during the tria. Data anaysis Comparisons were made with two-taied paired t-tests. Linear regression anaysis was used to assess the reationships between TL,CO,sb and RV/TLC ratio [5]. A vaues are given as mean SD. Resuts Fifty-nine subjects performed the required four TL,CO,sb tests. Seven had to repeat one test owing to a technica error and ony one subject had to repeat two tests. Additiona trias were performed in some subjects to meet the ERS reproducibiity criteria for TL,CO,sb. The resuts of these trias are not reported. However, the tota number of tests was used for the cacuation of the HbCO correction, as the fina bood samping was taken at the end of the session. There was no difference in TL,CO,sb between the first and the second tria for either test gas after correction for HbCO (21% FI,O 2 : second tria/first tria ratio: %, not different from 100%; 17.5% FI,O 2 : second tria/first tria ratio: %, not different from 100%). TL,CO,sb and KCO,sb measured with the 21% FI,O 2 test gas were significanty ower than when measured with the 17.5% FI,O 2. The intra-individua coefficients of variation (CV) for TL,CO,sb and KCO,sb (first and second measurement) were simiar for both gases, as was VA,eff (tabe 1). The inter-individua CV for TL,CO,sb was amost identica for both test gases (21% FI,O 2 :CV=22.5%; 17.5% FI,O 2 :CV=22.4 %). The sopes of the regression ines of the TL,CO,sb versus RV/ TLC ratio were neary identica for both test gases and not significanty different from 0 (fig. 1), even when TL,CO,sb was expressed in absoute (21% FI,O 2 :p=0.54; 17.5% FI,O 2 :p=0.48) or as a percentage of the ERS predicted vaue. The ratio of TL,CO,sb measured with the 17.5% FI,O 2 test gas on TL,CO,sb measured with the 21% FI,O 2 test gas woud, therefore, appear to be quite independent of the RV/TLC ratio in heathy subjects (fig. 2): the sope of the regression ine was not different from 0 (p=0.97). Discussion The resuts show that, TL,CO,sb is higher when measured with a 17.5% FI,O 2 test gas than with a 21% FI,O 2 test gas. The intra-individua CV for TL,CO,sb and KCO,sb are very cose to those previousy pubished [6] and there is no difference between the two test gases. The difference between the two measurements ( % of the mean or % per kpa ð1þ

3 TL,CO,SB AND FI,O 2 ; ERS VERSUS ATS 929 Tabe 1. Resuts of parameters measured using two different test gases, 17.5% inspired oxygen fraction (FI,O 2 ) compared with 21% FI,O 2 FI,O 2 Mean difference 17.5% 21% TL,CO,sb mmo?min -1?kPa *** CVi TL,CO,sb % # TL,CO,sb,%ERS % *** KCO,sb mmo?min -1?kPa -1? *** CVi KCO,sb % # KCO,sb,%ERS % *** VA,eff,sb, # VA,eff,sb/TLCpeth % VC%ERS Data are presented as mean SD. TL,CO,sb: singe-breath transfer factor; CVi TL,CO,sb: intra-individua coefficient of variation of TL,CO,sb; TL,CO,sb,%ERS: singe-breath transfer factor as a percentage of ERS predicted; KCO,sb: singe-breath transfer coefficient; CVi KCO,sb: intraindividua coefficient of variation of KCO,sb; KCO,sb,%ERS: singe-breath transfer coefficient as a percentage of ERS predicted; VA,eff,sb: effective aveoar voume, measured during singe-breath; VA,eff,sb/TLCpeth: effective aveoar voume, measured during singe-breath, as a percentage of tota ung capacity measured during pethysmography; VC%ERS: vita capacity as a percentage of ERS predicted. # nonsignificant; *** pv0.001 from 21% FI,O 2. inspired oxygen pressure (PI,O 2 ) appears sighty ower than data previousy reported by GRAY et a. [7] who found a vaue of 2.25% per kpa PI,O 2. However, in that study by GRAY et a. [7] the cacuated range of PI,O 2 (obtained from subjects breathing a 21% FI,O 2 test gas at sea eve or in a decompression chamber) was higher ( kpa) than in the present study ( kpa), and there is no evidence that the reationship between TL,CO,sb and PI,O 2 is inear. Furthermore, decompression might affect TL,CO,sb independenty of PI,O 2. In addition, GRAY et a. [7] did not correct TL,CO,sb for HbCO. Data coected at sea eve by KREUZER and VAN LOOKEREN CAMPAGNE [8] were used to cacuate a vaue of 1.42% per kpa PI,O 2, which is very cose to the resut presented in the current study. PA,O 2 at the beginning of the apnoea can be cacuated assuming a fixed vaue for PA,O 2 (e.g. 14 kpa) before inhaation and taking into account the individua inspiratory and residua voumes measured during the test; in this case, the increase in TL,CO,sb (expressed as a percentage of the mean) corresponding to the decrease in the cacuated PA,O 2 is per kpa PA,O 2, which is very cose to the 1.95% per kpa PA,O 2 predicted by OGILVIE et a. [9] for the corresponding variation in PA,O 2. TL,CO,sb was % of the ERS predicted vaue with the 17.5% FI,O 2 test gas, whist the corresponding figure with the 21% FI,O 2 test gas was %. The reference vaues for TL,CO,sb in young persons aged yrs have been poory documented and presumaby not adequatey represented by the ERS data set, at east in femaes [10]. Therefore, it is not surprising that the data in the present study came coser to the ERS predicted vaues under ATS conditions, than under ERS conditions. KANNER and CRAPO [11] showed that PA,O 2, measured in the aveoar sampe from a singe-breath tria conducted with a 25% FI,O 2 test gas at an atitude of 1,520 m, was inversey correated with the RV/TLC ratio measured during the test. KANNER and CRAPO [11] expected that the use of a test gas with a PI,O 2 coser to the resident PA,O 2 woud reduce interindividua variabiity because of the differences in the RV/ TLC ratio among norma subjects. A subsequent report [6] faied to confirm this hypothesis. The data in the current -1 kpa -1 TL,CO,sb mmo min RV/TLC % Fig. 1. Singe breath transfer factor of the ung for carbon monoxide (TL,CO,sb) versus residua voume/tota ung capacity (RV/TLC) ratio measured with pethysmography. The ines show regression and 95% confidence intervas for 21% inspired oxygen fraction (FI,O 2 ) test gas: #---; and 17.5% FI,O 2 test gas: $. Regression ines: TL,CO,sb at 17.5% FI,O2= RV/TLCz12.98; TL,CO,sb at 21% FI,O2= RV/ TLCz TL,CO,sb ratio 17.5% FI,O 2 /21% FI,O 2 % RV/TLC % Fig. 2. Individua ratio of singe breath transfer factor of the ung for carbon monoxide (TL,CO,sb) measured with the 17.5% inspired oxygen fraction (FI,O 2 ) test gas measured with the 21% FI,O 2 test gas versus residua voume/tota ung capacity (RV/TLC) ratio measured with pethysmography....: 95% confidence interva. Regression ine: TL,CO,sb at 17.5% FI,O 2 /21% FI,O 2 (%)= RV/TLC (%)z

4 930 H. NORMAND ET AL. study, obtained under ERS and ATS conditions from a arger group of subjects, does not support this hypothesis either, because the inter-individua CV for TL,CO,sb is amost identica for both test gases. The main resut of the present study is that the RV/TLC ratio does not affect TL,CO,sb differenty, whether the ERS or ATS recommendation for FI,O 2 (17.5 and 21% respectivey) is foowed, i.e. the ratio of TL,CO,sb measured with the 17.5% FI,O 2 test gas on TL,CO,sb measured with the 21% FI,O 2 test gas is not affected by the RV/TLC ratio. If the primary measurements are considered, TL,CO,sb is the product of KCO,sb and VA,eff. Therefore, TL,CO,sb is not independent of VA,eff and thus not independent of RV/VA,eff. The measurement of RV/TLC using a different technique guarantees that the correation deas with independent variabes. However, using the RV/VA,eff ratio measured with He diution during the singe-breath manoeuvre eads to simiar concusions. Furthermore, the decrease in TL,CO,sb as the RV/TLC ratio increased was not significant with either the 21% FI,O 2 or the 17.5% FI,O 2 test gas, and regardess of whether the data were expressed in absoute or as a percentage of the ERS predicted vaue. One may question the reativey sma span of the RV/TLC ratio and whether it was sufficient to detect a significant effect of RV/TLC ratio, when TL,CO,sb itsef has arge variabiity. Using a mass baance approach to tota O 2 in the ung immediatey after inhaation of the test gas, and the RV/VA,eff ratio was measured in each test, it was then possibe to cacuate the difference in PA9,O 2 (PA,O 2 immediatey after inhaation of the test gas) in each subject between inhaation of 21 or 17.5% FI,O 2 test gas, assuming a fixed vaue for PA,O 2 before inhaation (e.g. 14 kpa). Such a cacuation shows that the difference in PA9,O 2 woud run from 2.1 kpa at the highest RV/TLC ratios to 2.9 kpa at the owest (fig. 3). Based on a 2.1% change in TL,CO,sb per kpa PA,O 2, this span in PA,O 2 woud produce a maximum change in TL,CO,sb of 1.7% that woud probaby be difficut to detect, considering the intraindividua CV of TL,CO,sb. It is possibe that using a fixed voume of dead-space washout resuts in the contamination of the aveoar sampe in a certain number of subjects [12, 13] and thus adds noise to the data, reducing the chance of DPa,O 2 kpa RV/VA,eff % Fig. 3. Difference in cacuated aveoar oxygen tension (PA9,O2) immediatey after inhaation of the test gas, according to the inspired oxygen fraction (FI,O2) of the test gas as a function of the mean residua voume/effective aveoar voume (RV/VAeff) ratio. PA9,O 2 was cacuated for each test and each subject with the corresponding inspiratory and residua voume measured during the test, assuming that pre-test PA9,O 2 was 14 kpa. finding differences. However, determining the correct washout voume requires using a continuous anaysis of the expired gas, which is not the standardised technique. Furthermore, using a continuous anaysis and excuding the dead space visuay in a group of 62 respiratory patients (most of them with abnorma ung function), HUANG and MACINTYRE [12] showed a sma but nonsignificant (p=0.09) decrease in the proportion of nonreproducibe measurements [12]. HUANG and MACINTYRE [12] aso showed that the average percentage difference between two consecutive tests tended to be smaer when the washout voume was visuay excuded (p=0.13), but this decrease was more prominent in patients with obstruction or restriction. To increase the power of the present study, either the span of RV/TLC ratio or the number of subjects coud be increased. However, increasing the span woud require pathoogica subjects with a much higher variabiity in TL,CO,sb to be incuded. Increasing the number of subjects might make it any more effective. Indeed, the resuts show that, with both test gases, TL,CO,sb decreases with increasing RV/TLC ratio. If the diution effect was of substantia importance, a positive correation between TL,CO,sb and the RV/TLC ratio shoud be observed, at east when measuring using the 21% FI,O 2 gas, because an increase in the RV/TLC ratio decreases PA,O 2. Using an FI,O 2 test gas coser to the pre-test FA,O 2 was expected to reduce the effect of diution of the test gas in the RV. This was aso the rationae put forward by the ERS recommending the use of a test gas with 17 18% FI,O 2 rather than 21% FI,O 2. The presented data indicate that the ERS and ATS conditions in respect of the FI,O 2 test gas affect the reationship between TL,CO,sb and RV/TLC ratio in exacty the same way and, therefore, the ERS rationae is not compusory. This may be because the RV/TLC ratio makes itte contribution to the intersubject variabiity of TL,CO,sb in this heathy popuation. This is confirmed by the ow (and very simiar) coefficients of regression of TL,CO,sb versus RV/ TLC ratio with both test gases. The resuts do not suggest that one recommendation is more justified over the other. However, it may hep to choose the same recommendation for both societies, using considerations other than the diution effect, e.g. methane can be used as a tracer gas instead of He. As the concentration needed is very ow, the inspired gas can be made from air, reducing the cost of making a 21% FI,O 2 test gas. It might aso be of interest to appraise the prediction equations, taking into account the FI,O 2 of the test gas. Data cannot be extrapoated to patients with pumonary diseases, especiay those with airway diseases and arge RV. However, it might be difficut to demonstrate this diution effect in a pathoogica popuation, e.g. with chronic obstructive pumonary disease, as an increase in the RV/TLC ratio is usuay associated with a decrease in TL,CO,sb. In concusion, the presented data indicate that, in a heathy popuation of young aduts, the singe-breath transfer factor of the ung for carbon monoxide measured, under the European Respiratory Society conditions in respect of the inspired fraction of oxygen test gas, is 5.2% higher than under the American Thoracic Society conditions. Since the ratio of the singe-breath transfer factor of the ung for carbon monoxide measured with the 17.5% inspired fraction of oxygen test gas on the singe-breath transfer factor of the ung for carbon monoxide measured with the 21% inspired fraction of oxygen test gas did not vary with residua voume/tota ung capacity, the European Respiratory Society rationae for using a 17 18% inspired fraction of oxygen test gas for the singebreath transfer factor of the ung for carbon monoxide measurements appears to be unwarranted in young heathy aduts.

5 TL,CO,SB AND FI,O 2 ; ERS VERSUS ATS 931 Acknowedgements. The authors woud ike to thank the students invoved in this study for their cheerfu cooperation and are aso indebted to the technica staff of the Service des Exporations Fonctionnees for outstanding support. The authors are most gratefu to P. Denise for hepfu comments. The authors aso gratefuy acknowedge the work of J. Lee in correcting the Engish text. References 1. Quanjer PH, Tammering GJ, Cotes JE, Pedersen OF, Pesin R, Yernaut J-L. Standardized ung function testing. Officia statement of the European Respiratory Society. Eur Respir J 1993; 6: Supp. 16, ATS. American Thoracic Society. Singe-breath carbon monoxide diffusing capacity (transfer factor). Recommendations for a standard technique update. Am J Respir Crit Care Med 1995; 152: Forster R. Exchange of gases between aveoar air and pumonary bood fow: pumonary diffusing capacity. Physio Rev 1957; 37: Morgan P. Physica gas anaysers. In: Laszo G, Sudow MF, eds. Measurement in cinica respiratory physioogy. London, New York, Academic Press 1983; pp Zar JH. Biostatistica anaysis. 4th Edn. Upper Sadde River, NJ, Prentice Ha Internationa, Crapo RO, Kanner RE, Jensen RL, Eiott CG. Variabiity of the singe-breath carbon monoxide transfer factor as a function of inspired oxygen pressure. Eur Respir J 1988; 1: Gray G, Zame N, Crapo RO. Effect of a simuated 3,048 meter atitude on the singe-breath transfer factor. Bu Eur Physiopatho Respir 1986; 22: Kreuzer F, van Lookeren Campagne P. Resting pumonary diffusion capacity for CO and O 2 at high atitude. J App Physio 1965; 20: Ogivie C, Forster R, Bakemore W, Morton J. A standardized breath hoding technique for the cinica measurement of the diffusing capacity of the ung for carbon monoxide. J Cin Invest 1957; 36: Zanen P, van der Lee L, van der Mark T, van den Bosch JM. Reference vaues for aveoar membrane diffusion capacity and pumonary capiary bood voume. Eur Respir J 2001; 18: Kanner RE, Crapo RO. The reationship between aveoar oxygen tension and the singe-breath carbon monoxide diffusing capacity. Am Rev Respir Dis 1986; 133: Huang YC, Macintyre NR. Rea-time gas anaysis improves the measurement of singe-breath diffusing capacity. Am Rev Respir Dis 1992; 146: Graham BL, Mink JT, Cotton DJ. Overestimation of the singe-breath carbon monoxide diffusing capacity in patients with air-fow obstruction. Am Rev Respir Dis 1984; 129: