Main

Achieving a stable circulation in the first hours of life is fundamental to the provision of successful neonatal intensive care. Many clinical studies describe the administration of boluses of fluid to expand the total circulating volume (TCV) in response to clinical signs of poor skin perfusion, hypotension or acidosis and yet studies both in adults and the newborn show the prediction of hypovolemia by clinical assessment to be poor (1, 2). The availability of a simple cotside method to estimate TCV that could be used shortly after birth would be a useful tool both for the neonatal clinician and the researcher. Methods used previously involving the injection of autologous red cells labeled with biotin (3) or repeated blood sampling following injection of a marker such as indocyanine green (4) are not suitable for routine clinical practice.

The technique of Near Infrared (NIR) Spectroscopy is attractive to the neonatologist because of its noninvasive nature. The attenuation signals from a NIRS monitor reflect changes of Hb concentrations in any of the vascular compartments. Examples of this type of instrumentation include the NIRO-300 and 500 (Hamamatsu KK, Japan). These NIR monitors have been applied in the clinical measurement of a number of hemodynamic parameters in the newborn and adult including tissue oxygen saturation (5, 6), cerebral blood flow (7, 8) and absolute cerebral blood volume (9, 10).

To calculate the absolute cerebral blood volume (CBV), the existing techniques require a small change in concentration of an intravascular tracer. The newborn studies have used a change of oxy-Hb (9) and in a study in adults, indocyanine green has been used (10). The methodology is dependent on knowing the concentration of tracer in the blood while the NIR monitor measures the change in cerebral tissue: in the neonatal studies oxygen saturation measured noninvasively via a pulse oximeter was used while the adult study obtained the ICG elimination curve from repeated venous sampling.

In this paper, a new technique called tissue dye densitometry (TDD) is described which estimates both CBV and TCV from measurements with a NIRS monitor having both difference spectroscopy (11) and Spatially Resolved Spectroscopic (SRS) capabilities (NIRO-300, Hamamatsu KK, Hamamatsu, Japan) (5, 6) following a single injection of ICG. This method does not require the use of any direct invasive monitoring of blood tracer levels. Estimates of TCV made using the new method have been compared with simultaneous measurements using the technique proposed by Phillips et al.(12) using the percentage of fetal Hb before and after a blood transfusion.

METHODS

The tissue dye densitometry technique.

Using SRS, it has been shown that a scaled absorption coefficient can be calculated from the attenuation gradient measurements from multiple, incrementally spaced, detectors (5, 6). Subsequently the scaled absorption coefficient can be converted to a scaled value of the chromophore concentration. If absorption due to water and background absorbers is initially ignored (see later), the scaled concentration of three major chromophores, deoxy-Hb (k ·CHHb), oxy-Hb (k ·CHbO2) and ICG (k ·CICG) can be calculated (where k is the unknown scaling factor accounting for scattering in tissues and has the same value for CHHb, CHbO2 and CICG). An example of the scaled concentration of the three major chromophores is shown in Fig. 1. A new index known as the ICG Hb Index (IHI) can also be defined as follows:

Figure 1
figure 1

An example recording of the scaled concentrations of the three major chromophores. The high concentration of ICG during the 1st circulation results in some cross-talk between ICG and the Hb species as marked.

Full size image

ICG Haemoglobin Index

It can be seen in the definition of IHI that the unknown scaling factor k appears as a factor of both the numerator and denominator and thus cancels out. The IHI can be regarded as the ICG concentration per unit concentration of Hb. Using this simple approach, the resulting IHI signal has a negative offset (as shown in Fig. 2a) because the effects of water and background absorbers have been ignored. The total Hb concentration in tissue (tcHb′) can be calculated by the linear regression between IHI (Fig. 2a) and changes in ΔCICG,tis (Fig. 2b) as shown in Fig. 3:

Figure 2
figure 2

(a) ICG Hb Index (IHI) calculated based on a TDD measurement; (b) Changes of ICG concentration in tissue (ΔCICG,tis) calculated based on NIR difference spectroscopy; dotted parts (defined from 5 s before to 25 s after the peak) was excluded from the linear regression shown in Fig. 3.

Full size image
Figure 3
figure 3

Plotting ΔCICG,tis against IHI: the slope of the linearly regressed straight line =tcHb'.

Full size image

where ΔCICG,tis is the absolute change in ICG concentration calculated using conventional NIR difference spectroscopy. A recent study in rats (13) showed that the extent to which the Hb signals are affected by cross-talk is related to the ICG dosage. The larger the ICG dosage, the longer the effect of cross-talk in Hb. A low dosage of ICG was used in our study (0.2 mg/kg) which was even lower than that used in a previous clinical study in humans (10). We observed a small degree of cross-talk between ICG and Hb species in some subjects during the 1st circulation when ICG concentration in tissues was relatively high (see Fig. 1 for an example). However, the cross-talk effect on the Hb signals disappeared approximately 20–30 s after the initial injection of ICG. This can be seen from Fig. 1 which shows that the cross-talk is only significant between t = 30 second and t = 45 second (injection at t = 22 second), after which the levels of the two Hb species, i.e. CHHb and CHbO2 return to their original values. To avoid the effect of any cross-talk, the data during the first circulation has been excluded in the linear regression. In the calculation of ΔCICG,tis, a constant known as the differential pathlength factor (DPF) has to be taken into account for the additional distance photons have to travel in scattering media (14). The DPF has been previously measured for a range of tissues (15). In this study for the neonatal head, the DPF was set to 5.13. The above equation for tcHb' is similar to the one used in the oxygen desaturation method of assessing CBV (9) except that here ICG instead of oxy-Hb is being used as the tracer.

The background absorption represents a large and constant value arising from all other chromophores in tissues such as melanin and lipids (but excluding water) etc. Previous studies suggest it accounts for approximately 30% of the overall absorption coefficient (μa) (16, 17). The proportion of water in cerebral tissue is in the range of 80% to 90% by volume (18, 19). Despite its low optical absorption property in the NIR range, this large tissue water content results in its absorption also affecting the measurement of tcHb. The background absorption is considered to account for 22% of the total absorption in this paper and is subtracted from the original absorption coefficient before the conversion to chromophore concentration is performed. To minimize the effect of water, the initially biased tcHb' is corrected to true tcHb by the subtraction of 15.15 μmolar (see Appendix for details of this correction):

Once true tcHb is obtained, the CBV can be calculated using the following formula (9):

where tHb (g/100 mL) is the Hb concentration in blood obtained from a venous sample, MWHb=64500 g/mol the molecular weight of Hb, D t =1.05 g/mL the density of tissue and CLVHR = 0.69 the cerebral-to-large vessel hematocrit ratio (20). Using conventional NIR difference spectroscopy, the changes in ICG concentration in tissue, i.e. ΔCICG,tis, can be measured in units of μ Mole of ICG/100 g of tissue. Changes in ICG concentration in tissue blood can therefore be calculated by (note the units):

where MWICG=775 g/mol is the molecular weight of ICG. The dye concentration in tissue (mg/100 g) is now converted into dye concentration in tissue blood (mg/mL) and we call this technique tissue dye densitometry (TDD).

Since the ICG is continuously being eliminated by the liver, it is necessary to identify from the clearance curve the point in time when the ICG is fully mixed with blood but not yet eliminated by the liver. It has previously been suggested when using arterial ICG monitoring that a line can be back-extrapolated from the exponentially decaying part of the clearance curve to either the time of injection or the mean transit time (21, 22) and the corresponding ICG concentration is regarded as the initial fully circulated ICG concentration, i.e. C0ICG,bl. The mean transit time is calculated over the first circulation curve starting from the time of injection and is considered as the time when the dye first arrives in the field of view.

In this paper, the time when C0ICG,bl occurs has been chosen as the time when ICG concentration reaches the maximum, i.e. the peak time as shown in Fig. 3. The rationale behind this is that the tissue monitor used here measures ICG concentrations in tissue blood (in contrast to the arterial blood). The peak time is therefore the time at which the largest amount of tracer is in the field of view. To perform the estimate, the clearance curve is first converted into logarithmic scale and then a linear regression is carried out on the section between 1 min and 4 minute after the peak time. By back extrapolating this line to peak time, as shown in Fig. 4, the initial ICG concentration (C0ICG,bl) can be identified. The total circulating blood volume (TCV) can then be found

Figure 4
figure 4

ICG concentration in tissue blood at peak time derived by back-extrapolation.

Full size image

where ICG dose is the amount of ICG injected in mg.

The fetal Hb dilution (FHD) technique.

After a blood transfusion, the fetal Hb is diluted. Knowing the volume of red blood cells being transfused, the TCVFHD pretransfusion can be calculated as follows (12):

where V is the volume of red cells transfused in ml, HbFpre% is the percentage of fetal Hb before transfusion, HbFpost% is the percentage of fetal Hb after transfusion and Hctpre is the pretransfusion hematocrit. Fetal Hb was measured by HPLC (23) (Varient, Bio-Rad, U.S.A.) and hematocrit was measured by an automated full blood count analyser (Sysmex, SE9500, Japan).

Subjects.

The study was approved by the Local Research Ethics Committee of East London and the City Health Authority and infants were studied with informed written consent from the parents. A total of 25 babies were studied. Data from 7 were excluded because of problems with the NIR method: 5 because of poor signal quality secondary to body movement and probe dislodgement and 2 because the first circulation of the tracer could not be seen on the clearance curve. Results from another infant were excluded because of a laboratory accident with the estimation of fetal Hb. Successful paired measurements were made from 17 infants. The median gestational age of these 17 infants was 28 completed weeks, their weight was 1030 g and the age at testing was 6 days (Table 1). Twelve infants (70.6%) were receiving either mechanical ventilation or nasal continuous positive airway pressure and the other five (29.4%) needed supplementary oxygen at the time of the measurement. All the infants studied had normal pH and arterial blood pressure. No complication was noted related to either the equipment or ICG injection.

Table 1 Results of the study
Full size table

Measurement procedures.

Only babies needing blood transfusion for clinical indications were studied and those included in the analysis were all receiving their first transfusion. The NIRO-300 optical probe was attached to the infant's forehead and the initialization of the monitor was performed. A rapid bolus of 0.2 mg/kg ICG (ICG-Pulsion, Pulsion Medical System AG, Munich, Germany) was injected through an existing umbilical or peripheral venous cannula. The continuous change in ICG concentration was measured noninvasively for 20–30 min using the NIRO-300. The blood transfusion was started soon after completing the noninvasive measurement of ICG concentration and the donor blood transfused over 3–4 h. Three blood samples were collected, a small donor blood sample for Hct, and two infant samples: 0.5 mL shortly before the transfusion for Hb, Hctpre and HbFpre% and 0.3 mL ten to fifteen minutes afterward for HbFpost%.

RESULTS

The mean (±SD) CBV measured by the TDD technique was 1.72 ± 0.76 mL/100g (median: 1.49 mL/100g, range: 0.34–3.32 mL/100g). The median and mean TCVFHD were 67.23 mL/kg and 70.19 ± 13.73 mL/kg, respectively (range: 51.16–97.76 mL/kg), and the median and mean TCVTDD were 61.73 mL/kg and 70.80 ± 32.54 mL/kg (range: 16.36–140.55 mL/kg). A Bland Altman analysis (24) is shown in Fig. 5. The bias between the two methods was 0.61 ± 34.34 mL/kg (the 95% confidence interval: −13.93–15.16 mL/kg) and the limits of agreement (2 S.D.) were −68.07 mL/kg and 69.30 mL/kg, respectively.

Figure 5
figure 5

The Bland Altman plot: Bias of TCV obtained using the FHD and TDD methods.

Full size image

DISCUSSION

Our aim was to develop a method that could measure TCV soon after birth. This precluded the use of fetal Hb dilution, which can only be done at the time of blood transfusion and biotin labeled autologous red cell dilution, which takes several hours to obtain a result. In contrast the NIR monitor is simple to apply and the developed TDD method can be automated. As soon as venous access is established, the ICG can be injected and the result should be available within 20 min.

The use of ICG.

ICG has been used for clinical investigations such as measuring hepatic blood flow, testing liver function and measuring cardiac output for more than 40 years (25, 26). It is firmly bound to plasma proteins, excreted in bile in unconjugated form, and rapidly eliminated by the liver within 10–15 min. ICG has been shown to be safe in adults (27) and when used in newborn and preterm infants (4, 8).

Validity of the TDD method.

The Bland Altman plot in Fig. 5 shows a slight positive slope between the differences and averages of the TCVTDD and TCVFHD, which suggests that the TDD tends to overestimate for larger TCV (with respect to the overall average, ∼70 mL/kg in Table 1) and underestimates for smaller TCV. While the bias was small, the TDD method has limited agreements with the FHD method as suggested by the limits of agreement. Although this could partly be due to inaccuracies inherent in the FHD method, the TDD method is also subject to its own limitations, some of which are applicable to NIR based methods in general (14). Firstly, the calculation of CBV involves the use of a predefined constant known as the DPF as mentioned earlier. Depending on the location of the measurement site and the age of the subject, DPF can vary from 5.13 in infants to 6.26 in adults (15). It has been reported that DPF has a SD of approximately 15% among the same group of subjects (14, 15), accordingly, the accuracy of the measurements of CBV and TCV may also be affected. Secondly, the measured NIR data, whether from conventional difference spectroscopy or SRS, are all calculated assuming the tissues to be homogenous whereas in reality the probing area includes skin, skull, cerebral tissues, and cerebrospinal fluid. Thirdly, in the conversion from tcHb to CBV, another scaling factor, the Cerebral to Large Vessel Hematocrit Ratio is used to compensate for the reduced red cell content of blood in cerebral vessels compared with that in the large vessels (due to the Fahraeus effect) from which the blood sample was taken. The exact value of this scaling factor is still a subject of controversy. While a value of 0.69 was used here (20), another study has suggested a different value of 0.75 (28).

In this study the tissue water content and background absorption were assumed to be 90% and 22%. Although these assumptions are supported by previously published estimates (16, 19), the actual values may be different and again are likely to vary between subjects, causing inaccuracies in the measurement of CBV and TCV. One future development of the TDD technique is to include the estimations of DPF and background absorption using a multi-wavelength spectrometer (e.g. a charge coupled device) and the second differential spectroscopy (16) technique.

The dose of ICG given to the infants was very small (0.13–0.7 mg or 0.13–0.7 mL). The accuracy of the estimate of TCV is dependent upon the accuracy with which this small dose is drawn up and administered. Despite these potential pitfalls, NIR based methods are an attractive clinical option given the inexpensive, noninvasive and portable nature of the systems (5–11, 14, 16, 19). The NIR methods in measuring TCV and CBV in the newborn have not however previously been directly compared with another method.

Data from 7 of the 25 babies recruited had to be excluded because of problems with the NIR based method. The NIR signal is easily affected by body movements that led to loss of data from 5 babies. The failure to see clearly the first circulation of the tracer that occurred in two studies was probably due to slow injection of the ICG. We think that these problems were accentuated because the technique was being learned and that the exclusion rate would be unlikely to be so high again.

It can be seen from Table 1 that TCVTDD results are less close to their TCVFHD counterparts when the correlation coefficients between IHI and ΔCICG,tis are lower than 0.9, most notably in infant 10. These circumstances may reflect “noisy” signals caused by movement artifacts or invalid assumptions being made about the measurement site. This correlation coefficient may therefore have the potential of assessing the reliability of the results obtained using the TDD method.

The infants included in this study were all clinically stable with normal pH although all needed some form of ventilatory support. Indocyanine green is bound to albumin and thus the method should be more accurately regarded as measuring the “albumin space.” In a clinical situation where there is leakage of albumin from the vasculature this will lead to overestimation of TCV. It is hoped that by using only the first four minutes of the dilution curve (Fig. 3) in the calculation of C0ICG,bl) that this effect will be minimized, but the possibility that it will contribute error will need to be considered as the technique is extended to the study of less stable infants.

Comparison with existing techniques.

In choosing a standard method against which to compare the new technique, we rejected the biotin labeled autologous red cell dilution method, which was unlikely to be acceptable to the parents of our babies. Direct comparison with classic ICG dilution would have limited potential recruits to those with an arterial line and introduced the complication of precisely timed repeated sampling to construct a wash out curve (4). Hudson et al.(3) have previously successfully used red cell volume measured by FHD as the standard method in their evaluation of the biotin method in the newborn. It is a safe method that should yield reliable results and was acceptable to parents and staff. Although the SD of the measurements of TCV using the new method was particularly wide it was nonetheless also wide at 70.19 ± 13.73 mL/kg using FHD.

Studies of TCV of newborn infants are limited and previous measurements have all reported a wide range of results. Mollison et al. studied 28 full term healthy babies within 24hrs of birth using radio labeled albumen and reported that the mean TCV was 84.7 mL/kg (range 68.5–100.3 mL/kg) (29). Leipala et al. used FHD and studied 23 preterm infants (26–31 wks gestation) needing blood transfusion and reported that the mean TCV was 70 ± 9 mL in appropriately grown and 70 ± 16 mL in small for gestation babies (30). Anthony et al. measured the plasma volume in 18 preterm and full term infants (median gestation 30 wks) by dye dilution technique using ICG and reported the median plasma volume of 40.9 mL/kg (range: 21.4–106.3 mL/kg) (4). Hudson et al. measured the pretransfusion red cell volume in 20 infants by FHD (25–34 wk gestation) and reported the median red cell volume of 26.3 mL/kg (range: 13.1–41.5 mL/kg) (3). They also measured the paired red cell volume by FHD and biotin labeling red cells and found strong correlation between the two methods (n = 13, r = 0.989) (3). The Bland Altman analysis of our comparison of TCVTDD with TCVFHD measurements showed a small overall bias but the range of the results using TDD was wider than using other methods.

The CBV obtained in this study is similar to that previously reported in infants using another NIR technique with oxygen tracers, i.e. 1.90 ± 0.34 mL/100g (corrected for the DPF(780 nm) = 5.13 used here) (9) but higher than that obtained in adults using NIR, blood sampling and ICG tracers, i.e. 1.1 ± 0.39 mL/100g (10). The CBV obtained in this study is comparatively lower than that obtained from a SPECT study, i.e. 4.81 ± 0.37 mL/100g (28). The discrepancy could be due to the fact that NIR techniques measure in the volume of tissue through which the photons have traveled, including the scalp, bone, cerebrospinal fluid, gray and white matter. The skin and scalp, which have less blood content than the gray matter, may account for about 10% of the total absorption while the cerebrospinal fluid is mainly transparent and does not absorb much light. It has been reported that the regional CBV is lower in white matter than in gray matter (31). Our result agrees with theoretical predictions (32) that photons travel deeper into the neonatal brain and spend more time in the white matter, resulting in a lower average CBV.

Comparison with the pulse dye densitometry.

Aoyagi et al. introduced a NIR based technique known as Pulse Dye-Densitometry (PDD) to measure TCV (21) in adults and the method has been implemented in the dye densitogram analyser (DDG-2001, Nihon Kohden Co., Tokyo, Japan). This technique involves the injection of ICG and the measurement of its concentration in arterial blood using an adaptation of the pulse oximetry technique with either a nose or finger probe. In adults, the technique has been shown to correlate well with results obtained from the radioisotope-based techniques (33). Comparing our TDD method with the PDD method, the former method can provide both CBV and TCV with one injection of ICG while PDD can only measure TCV. Both techniques of course provide data on liver function via measurement of the ICG clearance rate. The major difference between the PDD and TDD methods is that the PDD is based on measurements in the arterial blood while the TDD method is based on tissue blood measurements. The PDD technique requires first of all the ability to detect a pulsatile arterial signal, and as a result, cannot be applied to patients with poor peripheral circulation giving rise to weak arterial signals (33). Also, the temporal resolution of the PDD system is limited by the heart rate whereas the current NIRO-300 can sample at 6 Hz.

In conclusion, the TDD method introduced here has the potential to be a useful cotside tool for studying TCV and CBV. As presently described it could be used in combination with other noninvasive methodologies in the study of circulatory adaptation in the period shortly after birth. Further improvement and studies to determine the limits of its absolute accuracy and repeatability are needed before its role in the clinical setting can be assessed.