Calibration of the HemoCue point-of-care analyser for determining haemoglobin concentration in a lizard and a fish

Hemoglobin concentration is commonly measured by point-of-care systems for many animal studies. The HemoCue Hb 201+ overestimated hemoglobin in blue-tongued skinks and diploid and triploid Atlantic salmon. The overestimation was systematic in both species, and therefore hemoglobin determined by the HemoCue can be corrected using appropriate calibration equations.


Introduction
Point-of-care (POC) devices, such as the HemoCue, are becoming increasingly common for animal studies because of their relative ease of use and low cost (for review, see Stoot et al., 2014). The portable nature of POC devices enables technically difficult data to be obtained in remote or challenging locations. Although POC devices can be used to show relative changes within the context of a study (e.g. Munns et al., 2012), it is important to be cautious about interpreting the absolute values and comparing between techniques if validation studies have not been conducted.
The HemoCue Hb 201 + system (Angelholm, Sweden) is a small, easy-to-use POC device that determines haemoglobin concentration ([Hb]). It is used widely by health-care providers around the world and commonly used in animal research (Srivastava et al., 2014; Table 1 and references within). In the HemoCue cuvettes, sodium deoxycholate lyses the erythrocytes, then sodium nitrite converts the haemoglobin to methaemoglobin, which forms the azidemethaemoglobin complex because of the presence of sodium azide. The HemoCue measures absorbance at two wavelengths (565 and 880 nm), which additionally compensates for turbidity. Importantly, the internal calibrations of the HemoCue system have been developed for human blood, in which the erythrocytes are non-nucleated, and thus measurements of other species may be compromised owing to differences in erythrocyte size and structure.
Haemoglobin concentration provides an indication of oxygen carrying capacity and thus is a useful bioindicator for a diverse range of studies, including basic physiology, immunology, toxicology and veterinary practice studies (Salice et al., 2009;Nevarez et al., 2011;Krams et al., 2013). As a result, HemoCue systems have been used to determine [Hb] in a wide variety of mammals, fish, reptiles, amphibians and birds (e.g. Colombelli-Négrel and Kleindorfer, 2008;Bazar et al., 2009;McFarland et al., 2012;Norman et al., 2012;Rummer et al., 2013). However, HemoCue systems have only been validated in mammals, birds and fish (Table 1). In birds and fish, the HemoCue overestimates [Hb] and there is some evidence that species-dependent validations may be required (Table 1 and references within).
Mean corpuscular volume (MCV), the presence of a nucleus and incomplete erythrocyte lysis have been proposed as potential reasons why [Hb] of some vertebrate groups is overestimated by the HemoCue (Arnold, 2005;Clark et al., 2008;Gustavsson, 2015). Presumably, incomplete lysis would lead to less [Hb] being converted into azidemethaemoglobin; hence, and underestimation of [Hb] by the HemoCue. If the presence of the nucleus alters the [Hb] determined by the HemoCue, we expect that [Hb] would be overestimated in all animals with nucleated erythrocytes, as has been shown in fish (Clark et al., 2008). If changes in MCV are responsible, then we would expect that the [Hb] of two species with a comparable MCV would be determined to be similar by the HemoCue.
The aims of this study aimed were as follows: (i) to determine whether the [Hb] of reptile blood can be determined accurately using the HemoCue; and (ii) to determine whether differences in MCV or nucleation of the erythrocytes affects the values produced by the HemoCue. To address the second aim, we used commonly farmed diploid and triploid salmon because triploids have larger erythrocytes (MCV) with larger nuclei than their diploid counterparts, and the MCV of diploid salmon is similar to that of reptiles (Sandnes et al., 1988;Benfey, 1999;Cogswell et al., 2002;Moller, 2014). We hypothesized that the presence of the nucleus in fish and reptile erythrocytes would result in an overestimation of [Hb] by the HemoCue in all three animal groups because of the HemoCue's internal calibrations being developed for non-nucleated human erythrocytes.

Materials and methods
Blood was collected from eight male blotched blue-tongued skinks (Tiliqua nigrolutea), seven triploid and five diploid Atlantic salmon (Salmo salar). Skinks were approximately 350-450 g and were maintained in a captive colony at the University of Tasmania. Salmon of ∼1.5 kg were obtained from a commercial aquaculture facility at Dover, Tasmania and maintained at 16°C in a recirculating aquarium facility at CSIRO, Hobart. Animals were fasted for at least 24 h before approximately 0.4-1.0 ml blood was collected from tail arteries or veins in heparinized plastic syringes and stored on ice for <2 h before use. Skinks were restrained by hand during blood sampling, and all samples were collected from the caudal artery using a heperanized 1 ml needle and 25 gauge syringe within 2 min. Salmon were anaesthetised with Aqui-S (Aqui-S, Australia), then sampled in air within 30 s. Experiments were conducted under University of Tasmania animal ethics permits (A0013794 and A0014283).
A dilution series was created from each blood sample to establish a range of [Hb] that was wider than the biological range for [Hb], which is approximately 30-140 g l −1 for bluetongued skinks and other Tiliqua spp. and approximately 89-104 g l −1 in Atlantic salmon (Sandnes et al., 1988;Moller,  . Each blood sample was divided into five aliquots (including whole blood). Two aliquots were haemodiluted with different volumes of 0.6 and 0.9% saline, which are approximately isotonic with reptile and fish blood, respectively (Thorarensen et al., 1991;Martinez-Jimenez and Hernandez-Divers, 2007). A further two aliquots were haemoconcentrated by centrifuging the sample, removing plasma and subsequently resuspending the erythrocytes in the reduced volume of plasma. Once the dilution series was created, [Hb] was determined using both the HemoCue and the well-established Drabkin's method (Sari et al., 2001).

HemoCue
Samples were mixed using a vortex, then 10 µl was loaded into HemoCue cuvettes using a pipette. Skink and salmon blood samples were read 1 and 7 min after being loaded into cuvettes, respectively. This reflected the longer time required for the reactions within the cuvettes to stabilize for salmon blood as has been reported previously (Clark et al., 2008). Preliminary experiments measured [Hb] every 30 s for 10 min to determine the appropriate incubation time. Replicate readings were obtained until two concurrent readings varied by <2g l −1 (generally, duplicate readings per sample).

Drabkin's method
In triplicate, 6 µl of blood was added to 600 µl aliquots of Drabkin's reagent (D 5941; Sigma Aldrich, Castle Hill, NSW, Australia). Samples were immediately mixed using a vortex and incubated for at least 1 h at room temperature. Degraded proteins were removed by centrifuging samples at 10 000g for 10 min. Next, 350 µl of the supernatant was transferred into 96-well plates alongside pure Drabkin's reagent as a blank. The absorbance was read on a spectrophotometer at 540 nm (Spectramax 190; Molecular Devices, USA).
The [Hb] (in grams per litre) was calculated according to equation (1): where A 540 is the absorbance of the sample at 540 nm, A Blank is the absorbance of the blank (i.e. Drabkin's reagent) at 540 nm, W Hb is the molecular mass of human Hb tetramer (=64 458), D F is the dilution factor (=101), C E is the extinction coefficient for terameric cyanomethaemaglobin at 540 nm (=44), d is the light path in centimetres (=1), and 1000 converts milligrams to grams (Dacie and Lewis, 1975;Clark et al., 2008).

Results and discussion
The [Hb] determined by the HemoCue Hb 201 + was overestimated in both Atlantic salmon and blue-tongued skinks (Fig. 1). In both species, the systematic nature of the HemoCue overestimation of [Hb] meant that calibration equations to correct data collected by this rapid POC system can be used reliably. Systematic overestimation of [Hb] by the HemoCue has previously been reported in fish and birds (Clark et al., 2008;Harter et al., 2015).
The slopes of the relationships between [Hb] determined by the HemoCue and Drabkin's method were not significantly different between diploid and triploid Atlantic salmon (P > 0.05); therefore, the data were pooled for further analysis (Fig. 1). Equation (4), which describes the relationship between the [Hb] values for Atlantic salmon determined by the two methods, has a slope that is significantly different from the line of equality (P < 0.010), as follows: The slope and elevation of equation (4) were not significantly different from those of the validation equation determined for four fish species, including two salmonids, by Clark et al. (2008; P < 0.001; Table 1). The similarities suggest that a common calibration can be used for salmonids and, potentially, other saltwater fish.
Likewise, the [Hb] of blue-tongued skinks was overestimated by the HemoCue (Fig. 1). Equation (5) ; Fig. 1B). Although this is smaller than the 30% overestimation that would occur in fish at the same [Hb], it is not inconsequential, and best practice would correct all bluetongued skink [Hb] measurements that are determined by the HemoCue.
The presence of the nucleus, the larger size of the erythrocytes (i.e. mean corpuscular volume) and incomplete red blood cell lysis have been proposed as potential causative factors and will be discussed in relationship to the data collected in the present study (Arnold, 2005;Clark et al., 2008;Gustavsson, 2015). No differences were observed between triploid and diploid salmon [Hb] calibrations in spite of triploid salmon having an MCV that is ∼40% greater than diploids (see Benfey, 1999, and references within). This is in contrast to the larger inaccuracies measured in ostrich erythrocytes manipulated to have larger MCV (Gustavsson, 2015). Likewise, if the overestimation was the result of differences in MCV we would expect [Hb] to be overestimated to a similar degree in skinks and Atlantic salmon because they have similar MCV. Mean corpuscular volume is 441-553 × 10 −15 l in Atlantic salmon and 422 × 10 −15 l in Tiliqua spp. (Sandnes et al., 1988;Moller, 2014). Instead, the overestimation was lower in skinks compared with Atlantic salmon.
If the presence of the nucleus is altering the readings of [Hb] in blood from animals with nucleated compared with nonnucleated erythrocytes, then we would expect that a larger volume of nuclear material would magnify the overestimation of [Hb]. The volume of the nucleus is ∼1.7 times larger in triploid (78 µm 3 ) compared with diploid (45 µm 3 ) salmon; however, there are 1.6 times more erythrocytes in diploid blood (1.08 cf. 0.68 million mm −1 ; Benfey and Sutterlin, 1984). These two opposing factors are likely to result in the same amount of nuclear material in both ploidies; hence, we would not expect calibration differences if the nucleus was the cause of [Hb] overestimation. In general, the size of reptile erythrocyte nuclei is similar to those of fish (Hartman and Lessler, 1964). Thus, the larger overestimation of [Hb] in salmon compared with skinks suggests that a factor other than erythrocyte nucleation is responsible for the overestimation. This finding, that the presence of the nucleus is not causing the overestimation in [Hb], may be relevant for measuring [Hb] on fish and reptile blood using other POC devices, although validation for the other POC devices and species would be required. Owing to a lack of data on the species tested in the present study (bluetongued skinks), we cannot completely rule out erythrocyte nuclei size differences between the two species in this study and possible influences on [Hb] determination.
Potentially, the structure and/or composition of the Hb itself differs between groups and produces the variation in [Hb] measured by the HemoCue. The haemoglobin of adult humans is composed predominantly (97-98%) of subtype Hb a , with Hb a2 making up a further 2% and Hb F ∼1% (Maniatis et al., 1980). Fish commonly have a larger number of haemoglobin subtypes (e.g. 8-14 in salmonids), which are thought to facilitate oxygen delivery over their wider operational body temperature range (Weber, 2000). It is possible that the structure or incomplete conversion of one or more of these subtypes to azidemethaemoglobin (HemoCue) or cyanmethaemoglobin (Drabkin's method) results in different [Hb] determined by the two methodologies. We were unable to find data in the literature that examines whether Hb subtypes have different absorbance properties; thus, this remains an avenue for further study.
Although it cannot be ruled out, incomplete lysis of erythrocytes is also unlikely to be causing the overestimation of validation. The mechanisms behind the overestimation of [Hb] by the HemoCue are an avenue for further research.
In summary, although the HemoCue 201 + system overestimated [Hb] in both blue-tongued skinks and Atlantic salmon, the error was systematic and can be corrected using appropriate calibration equations. Owing to some interindividual differences (i.e. variance), care should be taken when interpreting [Hb] calculated for individual animals. However, best practice for determining salmonid [Hb] using the HemoCue would be to correct [Hb] values using equation (4), which is appropriate for both diploid and triploid Atlantic salmon and, potentially, for other salmonid species, and use mean values for several animals. Although the overestimation was considerably lower, equation (5) should be used for best practice in blue-tongued skinks, and we would encourage appropriate validation before using the HemoCue to determine [Hb] in other reptiles.