A carbon monoxide ‘single breath’ method to measure total haemoglobin mass: a feasibility study

What is the central question of this study? Is it possible to modify the CO‐rebreathing method to acquire reliable measurements of haemoglobin mass in ventilated patients? What is the main finding and its importance? A ‘single breath’ of CO with a subsequent 30 s breath hold provides almost as exact a measure of haemoglobin mass as the established optimized CO‐rebreathing method when applied to healthy subjects. The modified method has now to be checked in ventilated patients before it can be used to quantify the contributions of blood loss and of dilution to the severity of anaemia.


INTRODUCTION
Haemoglobin (Hb) is the oxygen-carrying pigment of the circulation.
Its circulating concentration ([Hb]) is routinely measured in clinical practice, and low values are used to define 'anaemia' (Beutler & Waalen, 2006). However, [Hb] is determined by the total circulating mass of Hb (tHb-mass) and the volume of plasma (PV) in which it is carried. The measurement of such independent variables has distinct advantages given that PV can change substantially with disease. The importance of tHb-mass measurement in the clinical setting and the advantages over [Hb] have been discussed previously (Otto et al., 2017a,b;Plumb et al., 2016). Indeed, [Hb] correlates poorly with tHb-mass in patients with chronic liver disease or heart failure, in whom PV may be expanded (Otto et al., 2017a). Despite this fact, [Hb] is the major trigger for the transfusion of red blood cells.
Likewise, perioperative changes in tHb-mass and PV are common (Iijima et al., 2013;Makaryus et al., 2018) due to blood loss, administration of red blood cells or haemodilution through administration of intravenous fluids or through salt/water retention due to the 'surgical stress response' (Rassam & Counsell, 2005). Fluid distribution between physiological compartments and the impact of hypo/hypervolaemia on the glycocalyx and therefore the functional integrity of the intravascular space also influence PV (Strunden et al., 2011). However, the decision to transfuse blood to a patient in clinical practice in general, and perioperative and critical care settings in particular, hinges on a variety of factors. There is a growing recognition that [Hb] may not be the best clinical indicator to guide such decisions (Plumb et al., 2016;Shander & Ferraris, 2017).
tHb-mass and derived PV can be determined by different dilution methods, in which carbon monoxide (CO) has been found to be the easiest and most precise marker to use (Gore et al., 2005).
Using the inhalation of a known volume of CO (thus labelling Hb as carboxyhaemoglobin, COHb) allows the measurement of tHbmass and, thus, the calculation of PV. In self-ventilating subjects, this is achieved using the so-called 'optimized CO rebreathing method' (oCOR). However, this protocol relies upon the participant being alert, able to follow instructions, and able to control their breathing through a closed circuit, which in turn precludes the use of oCOR in participants who are receiving mandatory ventilation from a mechanical ventilator, either under anaesthesia or when sedated in the intensive care unit.
We hypothesized that delivery of a single CO bolus into the breathing circuit of a participant without the need for rebreathing could be used to reliably measure tHb-mass, so long as exhaled gas could be analysed.
We thus sought to develop such a technique. Here, we describe this development and early data relating to its likely reliability. The primary aim of this study was to develop a new method for measuring tHbmass in healthy participants simulating a procedure that might be used in participants on a mechanical ventilator (Proc new ) and to evaluate the feasibility of this novel method. We aimed to assess reliability compared to the standard oCOR method. We also repeated the oCOR test to quantify the reliability of the standard method within this experiment.

New Findings
• What is the central question of this study?
Is it possible to modify the CO-rebreathing method to acquire reliable measurements of haemoglobin mass in ventilated patients?
• What is the main finding and its importance?
A 'single breath' of CO with a subsequent 30 s breath hold provides almost as exact a measure of haemoglobin mass as the established optimized CO-rebreathing method when applied to healthy subjects. The modified method has now to be checked in ventilated patients before it can be used to quantify the contributions of blood loss and of dilution to the severity of anaemia.

Subjects
This feasibility study took place at the University Hospital Southampton NHS Foundation Trust, UK, and at the University of Bayreuth, Germany. Eleven healthy non-smoking test subjects (five women, six men) with moderate physical training status took part in the study (for anthropometric data of these subjects see Table 1).

Study design
In preliminary tests, we checked whether a single inhalation of a CO bolus could achieve an increase in COHb concentration ([COHb]) that would be sufficient to determine tHb-mass. For this purpose, tHb-mass was measured twice; results from the established CO-rebreathing method (Schmidt & Prommer, 2005), that is, 2 min CO inhalation within a closed spirometry system, were compared to those from a single inhalation with subsequent 10 s breath holding and 15 min collection of expired air. Because the difference in tHb-mass was lower than 50 g CO exhalation was determined from six subjects in a fifth test approach using the same methodology as described above after a conventional 2 min CO-rebreathing procedure (oCOR +20min ).

Established carbon monoxide rebreathing method
tHb-mass was determined using the optimized CO-rebreathing (oCOR) method as described and modified by Schmidt and Prommer (Prommer & Schmidt, 2007;Schmidt & Prommer, 2005). Briefly, a bolus of 99.97% CO (0.8-1.0 ml CO/kg body mass, depending on the training status) was administered to subjects and rebreathed along with 3 litres of 100% O 2 for 2 min. Three arterialized capillary blood samples were taken from a hyperaemic earlobe (Finalgon, Sanofi-Aventis, Frankfurt, Germany) before the rebreathing procedure, and at minutes 6 and 8 after the rebreathing procedure, and each sample was analysed in triplicate using an OSM3 haemoximeter (Radiometer, Brønshøj, Denmark). End-tidal [CO] was assessed before and 2 min after the rebreathing procedure using a portable CO detector (Draeger Pac7000, Lübeck, Germany). tHb-mass was assessed in duplicate (test 1 and test 2) using this method, and the mean of both tests was used for comparison with the results of the modified procedures. The typical error for tHb-mass measurements determined from these duplicate tests was 1.0%.

Modified method
To adapt the method so that it can be used in everyday clinical practice, several modifications were necessary (see Figure 1). Since a patient frequently cannot put the spirometer into their mouth by themselves, the gas supply was replaced by a mask (Hans Rudolph, Inc., Shawnee, KS, USA) with an access port for the CO supply. This access port is designed in such a way that the manually administered CO from a syringe passes the mask via a small tube directly into the back of Until the fifth minute after starting the test, the CO concentration and the volume of the exhaled air were monitored at 30 s intervals and thereafter at 1 min intervals until disconnecting the subject from the equipment after 20 min. In the same way as in test 1 and test 2, three capillary blood samples were taken before the test, one sample each was collected after 1 and 2 min, and then further samples were taken every 2 min until min 20.

tHb-mass calculation
For the established method (tests 1 and 2), tHb-mass was calculated as described previously (Schmidt & Prommer, 2005): where  of oCOR, tHb-mass was calculated for min 7 (tHb-mass min7 ) as well as using data from the whole test, that is, the mean from the plateau between min 6 and 20 (tHb-mass plateau ) and for min 20 using the data from the air collected in the Douglas bag (tHb-mass DouglasBag ).

Statistical analysis
Statistical analysis was performed using IBM SPSS Statistics (version 25 for Windows, IBM Corp., Armonk, NY, USA). Values are presented as the mean ± standard deviation (SD), unless otherwise stated.
Categorical variables are presented as frequencies (%).
This was a feasibility study, and a formal power calculation was therefore not required. Test-retest data (repeated measures from the same patient with different analytical methods) are presented using Bland-Altman plots with limits of agreement (Bland & Altman, 1986).
Additionally, a specific approach to compute reliability statistics to compare test-retest performance expressed as the typical error of measurement (TE) was used (see Hopkins, 2000).
Student's paired t-test was used to compare mean values from both tests at identical time points, and a paired t-test was also used to compare the mean values at different time points of the identical test. All tests were two-sided, and statistical significance was set at P < 0.05. To minimize the risk for type I errors, a correction for multiple measurements according to Benjamini & Hochberg (1995) was performed.

RESULTS
All of the tests were conducted without complications or adverse events. All participants inhaled the identical CO volume during the four tests (63.3 ± 23.1 ml; males 82.1 ± 16.8 ml, females 44.5 ± 6.3 ml).
The exhaled CO volume was highest in the first minute of Proc new 15s (6.3 ± 3.5 ml; Proc new 30s 2.9 ± 1.6 ml). This initial phase was followed by an almost linear and parallel increase in both new procedures, showing an accumulation of 8.7 ± 3.6 and 5.1 ± 2.0 ml in min 7 and 12.0 ± 4.4 and 8.4 ± 2.6 ml in min 20, respectively. When the expired air was collected after oCOR +20min , the values (3.1 ± 0.3 and 5.9 ± 1.1 ml) were clearly below those of Proc new 15s and Proc new 30s ( Figure 2) and not different from the volume exhaled 7 min after oCOR.
[COHb] exhibited well-known time-dependent changes, with a fast increase in the first minute followed by a rapid and then decelerating decrease during the rest of the observation period ( Figure 3). The values of the new procedures were clearly below those of oCOR +20min .
In min 7, the CO volume ligated to Hb was 59.5 ± 22.5 ml (oCOR), 57.2 ± 21.5 ml (Proc new 30s) and 53.9 ± 21.6 ml (Proc new 15s),   very similar values for Proc new 30s over a large range of tHb-mass values between 450 and 1300 g ( Figure 5).

DISCUSSION
We wished to identify a way to measure tHb-mass in mechanically ventilated patients. We thus explored whether it is possible to reliably measure tHb-mass using a single-breath inhalation ( (Burge & Skinner, 1995). The current technique described by Schmidt and Prommer reduced the rebreathing period to only 2 min to improve convenience for participants (Otto et al., 2017a,b;Plumb et al., 2020;Schmidt & Prommer, 2005). The finding that a bolus of CO gas inhaled with a single breath and only rebreathed for 2 min led to valid and reliable results characterized by a typical error between 1% and 2% allowed the method to be used in a variety of different settings. Initially, these were primarily focused on elite sports physiology and performance, but more recently, oCOR has also been used to answer clinical questions (Otto et al., 2017a,b (oCOR +20min ) with the established method (oCOR). As we did not find any difference in the resulting tHb-mass, we conclude that the exhalation protocol does not affect the precision of the new procedures.
When breath was held for 15 s, the initial CO exhalation after 1 min was twice as high as that in Proc new 30s, indicating that ∼10% (Proc new 30s ∼4%) of the inhaled CO did not diffuse into the blood.
After 7 min, that is, when the CO mixing in the blood was completed and therefore used in the established oCOR for blood sampling after the test, the loss of CO was ∼13% in Proc new 15s and only ∼8% in Proc new 30s. Although this loss in CO clearly exceeded the CO volume exhaled after oCOR (∼4%), these data demonstrate the rapid diffusion of CO from the lungs into the blood, which is also a precondition for the determination of the lung diffusion capacity by means of a deep inhalation of a 0.3% CO-containing gas followed by a 10-s breath hold (Modi & Cascella, 2020). We therefore suggest that tHb-mass might be easily and exactly calculated after a single breath when considering sufficient mixing time of the inhaled CO bolus.
These considerations are supported by the calculated tHb-mass over time. In the new procedures, tHb-mass reached a plateau in min 6 or in min 8 indicating complete mixing (Bruce & Bruce, 2003), that is, that any time point beyond can be used for tHb-mass and blood volume determination (Wachsmuth et al., 2019). In this study, we used the plateau value between min 6 and 20 and compared its mean with the time point usually used in the oCOR (min 7) and with the results obtained in min 20 from the exhaled air collected in the Douglas bag.
As shown in Table 2, there is no obvious difference between tHb-mass obtained from the oCOR and that obtained from the new procedures at the time points mentioned above. In the Bland-Altman plot, a small but systematic underestimation by approximately 25 g compared with the reference method becomes obvious in Proc new 15s. As such a deviation does not occur between Proc new 30s and oCOR, we suggest that the smaller CO volume taken up from the blood during Proc new 15s may be the cause. Additionally, the very low limits of agreement in the comparison of both new methods with oCOR indicate that Proc new 15s already presents a promising tool for tHb-mass and blood volume determination, and Proc new 30s seems to be as exact as the established oCOR.
In future studies, the administration of higher CO volumes than those used for the oCOR may be taken into consideration to compensate for the lower CO uptake during the single-breath application. On the one hand, this procedure increases the [COHb], reducing the measurement error of the CO oximeter (Alexander et al., 2011) but also increases the volume of exhaled CO and thereby also been extensive debate about the merits of having a higher [ΔCOHb%] versus the increased toxicity risk (Alexander et al., 2011;Garvican et al., 2010;Turner et al., 2014).
[COHb] of up to 10% has been described without remarkable side effects in healthy subjects , but to our knowledge, it has never been studied in seriously ill patients. Because CO is endogenously produced and is actually considered for the treatment of various diseases (Motterlini & Otterbein, 2010), we are convinced that the increase in COHb by 4-5%, as achieved in our study, represents a reliable compromise balancing sources of error with minimal patient risk.

Practical application
We hypothesize that this method might be used to diagnose and provide more information on the origin of anaemia in intensive care (Magee & Zbrozek, 2013) and the amount of blood loss during surgery (Shoemaker et al., 1996), as well as for distinguishing between dilutional anaemia and genuine anaemia in patients with heart failure (Miller & Mullan, 2015) and liver failure (Plumb et al., 2020). Here, it is of critical importance that the modified method has sufficient accuracy to reveal clinically relevant changes in tHb-mass and their contribution to changes in [Hb]. Although the reliability of the modified method was not explicitly determined in this feasibility study, the methodological error (typical error, TE; Hopkins, 2000) compared to the established method is 3.2% (Proc new 30s) or 3.5% (Proc new 15s).
This is higher than the TE of the established CO rebreathing methods (TE 2.2%) but close to the TE of the gold standard methods using radioactive markers ( 51 Cr, 2.8%; Gore et al., 2005). Since even mild real anaemic states ([Hb] 11.2 g/dl) are associated with a reduction of at least ∼15% tHb-mass (Wachsmuth et al, 2015), the accuracy of the method should be sufficient to distinguish anaemia due to reduced tHb-mass from that due to dilution. This contention is supported by Otto et al. (2017a) who describe two patients with liver disease who had identical tHb-mass (9.2 g/kg body mass), but with a normal [Hb] (16.1 g/dl) in one and dilutional anaemia ([Hb] 10.7 g/dl) in the other.
They also describe two cases of heart failure in which the presence of a severe reduction in tHb-mass (5.2 g/kg) was reflected in a low [Hb] in one (6.9 g/dl), but masked by a contracted PV in another ([Hb] 10.7 g/dl) (Allsop et al., 1998;Otto et al. (2017a). In addition to the frequently occurring dilution anaemia, decompensated heart failure can also be associated with proportional increases in both tHb-mass and PV (Miller, 2016). In all these cases, the determination of the tHb-mass, also when using the modified method, enables a much more precise diagnosis.
The modification described here may permit the CO method to be used in ventilated patients. CO is applied to the inhalation path, and breathing is interrupted in the inhalation position for 15 or 30 s. The In such clinical studies CO mixing time must be considered. It is prolonged in patients with polycythaemia (Wachsmuth et al., 2019) and heart failure (Ahlgrim et al., 2018), and perhaps also in other patient groups. This should, however, not be a major problem as the exhaled CO is collected for 20 min and a significant increase in mixing time can be tolerated if the individual COHb plateau is determined for each patient after inhalation of the CO. The use of the new method in patients with pulmonary diffusion disorders could be more problematic if sufficient CO cannot diffuse from the alveoli into the blood within 30 s. Higher CO doses may have to be used in such circumstances, but this can have an adverse effect on the accuracy of the test.
In healthy subjects, there is no risk of interrupting breathing for 30 s, and the risk can be classified as very low also in ventilated patients.
Since the oxygen consumption during the 30-s breath interruption is only approximately 150 ml, the arterial O 2 saturation does not change during this period (Parkes et al., 2016); but in any case, it must be checked during and after the test. In severely anaemic patients, the test might be used with great caution after extensive validation.

CONCLUSION
Using the single-breath method, tHb-mass and blood volume can be determined with approximately the same accuracy as that with established CO-rebreathing methods. We recommend that this method be developed further for use in ventilated patients, that is, patients in intensive care, patients undergoing major surgery, and patients with heart and liver failure.