1.

Introduction

Anemia is a common complication observed in critically ill patients admitted to the intensive care unit (ICU). In the ICU setting, anemia can result from severe bleeding after trauma, acute blood loss, or hemorrhage during surgery. In addition, it arises in non-bleeding but critically ill patients experiencing a rapid decrease in blood arterial total hemoglobin concentration (Hgb)¹ (details about the terminology and acronyms used in this article in relation to hemoglobin concentration are reported in Appendix A). Typically, anemia is defined by low arterial oxy-hemoglobin concentration (${\mathrm{HgbO}}_2$) in blood or low hematocrit (HCT) percentage ($\sim 8$ to 10 g/dL^2,3 and 30%,⁴ respectively) relative to normal values ($\sim 14$ to 15 g/dL and 35% to 50%, respectively). This is quite common in the general ICU critically ill population within the first three days after admission.

Under anemic conditions, low ${\mathrm{HgbO}}_2$ translates into a reduced supply, release, and diffusion of oxygen into tissues. This affects the blood flow [BF, cerebral blood flow CBF when related to the brain, and peripheral blood flow (PBF) when related to peripheral parts of the body]² due to its relationship with oxygen delivery (${\mathrm{DO}}_2$), which is the product of BF and arterial oxygen content (${\mathrm{CaO}}_2$).^4,5 As such, anemia can have deleterious effects on the patient’s health;^2,5 thus, severe anemia must be avoided by action taken as early as possible.⁶ Furthermore, patients with severe traumatic brain injury (TBI) usually present low CBF at the beginning of hospitalization,^2,4,6,7 which increases the risk of ischemia whenever the oxygen supply cannot meet anymore the demand.²

Anemic patients generally undergo red blood cell transfusion (RBCT),¹ with most clinicians accepting this intervention to restore the ${\mathrm{HgbO}}_2$ content.^1,4 However, RBCT also alters the BF^1,6 and other directly related physiological parameters, such as the Hgb, the tissue oxygen saturation (${\mathrm{StO}}_2$), the metabolic rate of oxygen (${\mathrm{MRO}}_2$), and the oxygen extraction fraction (OEF). In the case of the healthy brain, cerebral autoregulation and other protective mechanisms ensure that these changes do not impair brain function. In patients with severe TBI, however, cerebral autoregulation and the systemic cardiovascular response are frequently compromised.^2,4 In non-critical organs, such as peripheral muscles, which lack BF control mechanisms similar to those in the brain, significant and potentially unsustainable alterations may occur. Consequently, the balance between local hemodynamics and metabolism can be disrupted by RBCT.

The indications for RBCT for critical care patients are the subject of considerable debate among clinicians. For example, the “Transfusion Requirements in Critical Care” study, published in 1999^5,8 and reviewed in 2012,⁹ questioned the risk-benefit ratio of transfusions and whether they are appropriate in ICU patients with acute brain injuries.^2,4–6,10 Several studies have reported side effects of RBCT in TBI patients,⁴ and recent guidelines recommend a more restrictive transfusion strategy in these patients, with a ${\mathrm{HgbO}}_2$ threshold lower or equal to 7g/dL.^{1–4,9,11,12} The identification of the hemodynamic and metabolic alterations induced by RBCT in the brain compared with peripheral muscle could offer insights into the specific impact on the vulnerable brain of neurocritical patients.

Our primary argument for this study is that advanced photonic methods utilizing non-invasive hybrid diffuse optics (DO) employing near-infrared light can facilitate a deeper understanding of these phenomena. Hybrid DO, comprising time-resolved spectroscopy (TRS) and diffuse correlation spectroscopy (DCS), could provide a more complete overview of these physiological alterations for the reasons described below.

DCS quantifies the local, microvascular BF (CBF and PBF), whereas TRS quantifies ${\mathrm{StO}}_2$, microvascular oxy- and deoxy-hemoglobin concentrations (${[\mathrm{HbO}}_2]$ and [Hhb]), total microvascular hemoglobin concentration ([HbT]), and OEF (COEF when cerebral, PEOF when peripheral). Together, they convey information about the ${\mathrm{MRO}}_2$ (${\mathrm{CMRO}}_2$ when cerebral, ${\mathrm{PMRO}}_2$ when peripheral). However, the literature regarding the effects of RBCT on these signals is limited.¹³ Continuous-wave near-infrared spectroscopy (NIRS) is the most commonly used method for measuring blood oxygen saturation^5,13 with few studies conducted on children (i.e., premature babies) ^14–20 and even fewer on adults.^10,21–24 Hence, further studies such as this one are warranted.

We have conducted a pilot study using hybrid DO for both local cerebral and peripheral continuous monitoring during RBCT in critically ill patients. Therefore, as a first objective, we will validate its usability in such a scenario.

This study was exploratory and observational, analyzing cerebral and peripheral hemodynamic and metabolic biomarkers to elucidate the ability of DO spectroscopic techniques to quantify changes due to RBCT, as our second objective.

2.

Materials and Methods

2.2.

Study Protocol

A schematic of the measurement protocol and probe positioning is provided in Fig. 1. One optical probe was placed on the forehead of the recruited subject while at the bedside (cerebral placement). CT scans were checked prior to the measurements whenever available to inspect the tissue and decide the optimal positioning. Accordingly, the hemisphere more easily accessible and/or without underlying trauma was always preferred for the probe placement. A second probe was preferably placed on the quadriceps femoris or, if not accessible, on the brachioradialis muscle (peripheral placement).

Fig. 1

Protocol illustration and optical probe placement. Study protocol: following a baseline pre-RBCT period (*generally $\sim 30\min$ but adjusted to the patient’s needs), the RBC transfusion started (**2 to 3 h duration according to the bag volume and/or transfusion rate decided by the nurse) and was measured until finished, when additional 30 min post-transfusion was monitored. Three blood gas samples were taken during each period of the measurement. As depicted on the right, one probe was placed on the forehead of the subject, whereas the second was placed at the peripheral level, generally on the quadriceps femoris. This figure was created by modifying and merging images from Servier Medical Art,²⁶ part of Laboratoires Servier, licensed under a Creative Commons Attribution 3.0 Unported License.

After the positioning of the optical probes, critical care personnel re-verified the blood bag compatibility with the patient, and the standard transfusion procedure was followed.

The continuous monitoring of the patients was carried out before, during, and after transfusion with a protocol of 30 min of baseline (pre-RBCT), 2- to 3-h-long RBCT, and 30 min of observational post-transfusion monitoring (post-RBCT). The timing of each period was adjusted according to the situation, as illustrated in Fig. 1.

Standard vials of blood samples were extracted from an artery (generally the arteria radialis) through a catheter during the protocol at baseline when half the blood bag was transfused, and during the observational final period, for a total of three times. In the ideal scenario, the initial and final blood samples were taken $\sim 10\min$ after each protocol part began but were adjusted according to the decisions of the medical personnel.

To standardize the protocol and its duration, we have decided to carry out our study only after one single RBCT even in those cases where more blood bags were indicated to restore the patient’s hemoglobin values.

3.

Results

3.2.

Representative Data During Transfusion

A representative set of data (male, 18 y.o. with polytrauma and severe TBI) is presented in Fig. 2 (case 2 in Table 1) where darker curves show data from the peripheral muscle ($\mathrm{rPBF}$, $\mathrm{rPOEF}$, ${\mathrm{rPMRO}}_2$) and the lighter curves correspond to data from the brain ($\mathrm{rCBF}$, $\mathrm{rCOEF}$, ${\mathrm{rCMRO}}_2$). The start and the end of the RBCT are indicated by vertical dashed lines. Additional events, such as blood gas sampling and movements, are depicted by lighter vertical lines. Initial Hgb, HCT, and ${\mathrm{PaCO}}_2$ values were 7.9 g/dL, 26%, and 36 mmHg, respectively.

Fig. 2

Time-traces example. Time traces collected during the RBCT of a patient for an entire protocol, comprising baseline pre-RBCT, blood unit transfusion, and post-transfusion period. The signals are depicted with peripheral and cerebral recordings color-coded, as shown in the legend. The beginning and end of the process are highlighted by thick dashed vertical lines, synchronized in time for all variables. Additional events marked during the protocol are represented by vertical lines.

By eye, it can be seen that there is a general trend for the rCBF to decrease with respect to the pre-RBCT period, whereas the rPBF has an increasing trend. Instead, rOEF shows a decreasing trend for both peripheral and cerebral signals. The response of the metabolic rate of oxygen extraction is more subtle with a small decrease in ${\mathrm{rCMRO}}_2$ by the end of the protocol.

For completeness, the time traces of ${\mathrm{rMRO}}_2$ for all subjects are provided in Appendix B.

3.3.

Pre- and Post-Transfusion Comparisons

As far as blood gas sample-related variables are concerned, the results of the comparison between pre-RBCT and post-RBCT distributions of mean values to check whether a significant change occurred are provided in Table 2, where the p-values and direction of changes are reported. RBCT caused a significant increase in the Hgb by $\sim 0.9\mathrm{g}/\mathrm{dL}$ ($p<0.001$) and HCT measured by blood gas samples by $\sim 3\%$ ($p<0.001$). ${\mathrm{PaCO}}_2$ did not show any significant change ($p=0.22$).

In relation to the tests for pre- and post-transfusion distributions from the hemodynamic time traces, the results are detailed in Table 3 and depicted in Fig. 3. Differences between cerebral baseline pre-RBCT distributions and peripheral baseline pre-RBCT distributions for all variables were not statistically significant ($p>0.05$) and were omitted from Table 3. All post-transfusion test results are provided together with the direction (Dir.) of the change with respect to the baseline. Moreover, the results of the tests between the post-RBCT cerebral values and the peripheral values are reported in the third column.

Table 3

Results of the optically derived variables. Summary of the results of all tests carried out for the optically derived variables (rBF, rOEF, rMRO2, Δ[HbO2], Δ[Hhb], Δ[HbT]). Under the column “cerebral,” the post-transfusion values were tested for difference with respect to baseline. The results concerning the post-transfusion muscle-related values are reported below “peripheral,” again tested with respect to baseline. Post-transfusion values for the cerebral level were tested for significant differences with respect to the post-transfusion values at the peripheral level in the last column (“cerebral versus peripheral”). Significance was defined as p-values lower than 0.05. The direction (Dir.) of the change of the median value for the distribution is represented by an arrow: upward if it increased and downward if decreased.

Variable	Cerebral		Peripheral		Cerebral versus peripheral
	p-value	Dir.	p-value	Dir.	p-value	Dir.
rBF	0.07	—	$<0.001$	$\uparrow$	0.1	—
rOEF	$<0.001$	$\downarrow$	$<0.001$	$\downarrow$	0.2	—
${\mathrm{rMRO}}_2$	0.03	$\uparrow$	$<0.001$	$\uparrow$	0.03	$\uparrow$a
$\mathrm{\Delta }{[\mathrm{HbO}}_2]$	0.008	$\uparrow$	$<0.001$	$\uparrow$	0.9	—
$\mathrm{\Delta }[\mathrm{Hhb}]$	0.2	—	0.1	—	0.2	—
$\mathrm{\Delta }[\mathrm{HbT}]$	0.01	$\uparrow$	$<0.001$	$\uparrow$	0.5	—

^aIn this case, the arrow indicates that the peripheral level was significantly higher than the cerebral level after RBCT.

rBF, relative blood flow; rOEF, relative oxygen extraction fraction; rMRO2, relative metabolic rate of oxygen; Δ[HbT], change in [HbT]; Δ[HbO2], change in [HbO2]; Δ[Hhb], change in [Hhb].

Fig. 3

Group distributions relative to baseline. Graphs representing the post-transfusion data distribution for all optically derived variables and divided into cerebral and peripheral muscle. For each distribution of mean values, its boxplot and violin plot are represented (median, first, and third quartiles). Dashed grey lines are drawn, indicating the baseline pre-RBCT values for the distributions, which they are relative to and compared with. Subjects are color-coded in the same way for all variables. An asterisk on the top of a boxplot indicates that the distribution tested significantly different than the baseline. No significant difference was found between peripheral and cerebral post-transfusion values apart from the metabolic rate of oxygen. rOEF, relative oxygen extraction fraction; rBF, relative blood flow; ${\mathrm{rMRO}}_2$, relative metabolic rate of oxygen; $\mathrm{\Delta }{[\mathrm{HbO}}_2]$, change in microvascular oxy-hemoglobin concentration; $\mathrm{\Delta }[\mathrm{Hhb}]$, change in microvascular deoxy-hemoglobin concentration; $\mathrm{\Delta }[\mathrm{HbT}]$, change in microvascular total hemoglobin concentration.

RBCT provoked a significant increase ($p<0.001$) in rPBF by 68% (by comparing the percent change between the median post- to median pre-values), whereas no significant change in rCBF was observed ($p=0.07$). The rPOEF median value fell significantly by 8% ($p<0.001$) and rCOEF by 7% ($p<0.001$) with respect to their pre-RBCT values. The median ${\mathrm{rPMRO}}_2$ significantly raised by 89% ($p<0.001$) and ${\mathrm{rCMRO}}_2$ by 10% ($p=0.03$), with a significantly greater median increase in the muscle by 79% than in the brain ($p=0.03$). No significant differences were found in the post-RBCT comparison for the other hemodynamic signals ($p=0.1$ for rBF, $p=0.2$ for rOEF).

Both $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ and $\mathrm{\Delta }[\mathrm{HbT}]$ had a significant increase both at cerebral and peripheral levels, with similar median shift with respect to baseline pre-RBCT values (as visible in Fig. 3). $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ increased by 0.0040(0.0053) mMol on the brain and 0.0036(0.0014) mMol on the muscle; $\mathrm{\Delta }[\mathrm{HbT}]$ raised by 0.0041(0.0071) mMol on the brain and 0.0034(0.0016) mMol on the muscle. Instead, $\mathrm{\Delta }[\mathrm{Hhb}]$ had no significant changes ($p=0.2$ and $p=0.1$ for cerebral and peripheral placements, respectively). No statistical difference was found between post-RBCT cerebral versus peripheral placement for any of the hemoglobin concentration-related variables.

4.

Discussion

The study aimed to demonstrate the usefulness of hybrid DO to quantify the physiological changes after RBCT and to provide a preliminary characterization of the response at both the peripheral muscle and the brain in a small, pilot population of critically ill patients. We have obtained a rich array of signals. As the RBCT response for the invasive and standard variables is well documented, we focus on discussing the results of the optical monitoring with respect to the literature, contributing to a thorough characterization.

With respect to our objectives, we have shown that hybrid DO could be validated in the ICU scenario, monitoring critically ill patients undergoing RBCT. Moreover, we were able to quantify and evaluate changes in the measured parameters in correspondence with RBCT (by which we mean “happening simultaneously to,” in relation to cause–effect causality), similarly to the other non-optical parameters.

RBCT provoked a significant increase in rPBF, whereas no significant change in rCBF was observed. However, this is in contrast to the results obtained on the brain in ill but young populations (i.e., sickle cell disease, very preterm infants) where a decrease in the CBF was found after the intake of blood.^13,18,39 For instance, in Ref. 18, the decrease of cerebral blood volume (CBV) was 0.5 mL/100g, whereas, in Ref. 39, CBV decreased by a median of 18.2% and CBF by a median of 21.2%. However, we acknowledge that the populations, the age, the origin of their anemia, and the etiology are not fully relatable and may account for these discrepancies.

As for alternative monitoring methods, for example, a reduction by 18% in the blood flow velocity measured by transcranial Doppler ultrasound was found in stroke patients after 3 h post-RBCT both on infarcted and non-infarcted regions, with significantly lower baseline values for the infarcted areas.⁴⁰ Nonetheless, in another study involving subarachnoid hemorrhage (SAH) patients, post-RBCT global CBF values monitored by ${}^{15}\text{O}$-positron emission tomography (PET), remained unchanged,⁴¹ similar to our findings. Differences in the post-RBCT CBF response between children and adults with sickle cell anemia were also found by magnetic resonance imaging (MRI) and magnetic resonance angiography:⁴² post-RBCT CBF did not significantly change in the adults ($N=16$), whereas it significantly decreased in the children ($N=10$, with $-22.3\mathrm{mL}/100\mathrm{g}/\min$ in 9/10 participants). In Ref. 43, CBF evaluated in children with sickle cell anemia by MRI with arterial spin labeling significantly decreased by 6.5 mL/100g. These findings are mostly in contrast with our results. However, a possible explanation may lie again in the etiology itself and the age of the participants. Moreover, the differences may be due to the fact that we have decided to monitor only the transfusion of a single RBC bag, the first one, even when more units had been assigned to the patient. This was done to be consistent among patients and to compare the amount of change in the physiological signals. According to the literature, greater changes are expected when more bags are transfused.⁴⁴

Another set of parameters that we report is related to blood oxygenation and total hemoglobin concentration. Here, we have used TRS, which is well-known to be more accurate and precise compared with the commercially and clinically available continuous-wave NIRS, which dominates the literature. However, for the purposes of this pilot study, we do not get into a discussion of these differences and discuss the findings together. Moreover, as several NIRS studies characterized the tissue oxygen saturation (${\mathrm{StO}}_2$) rather than OEF, which is generally inversely related to the OEF, we will use it as well as an inverse comparison to our findings. In our study, at the cerebral level, we have encountered a significant decrease in rCOEF, a significant increase in $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ and $\mathrm{\Delta }[\mathrm{HbT}]$, and no change in $\mathrm{\Delta }[\mathrm{Hhb}]$. For rCOEF, this is in agreement with the literature and was mainly confirmed in young populations after RBCT;^5,13,17,45 although not always, a significant change was found.²⁴ A significant decrease of rCOEF by 0.06% by CW-NIRS was reported, for instance.

Studies using other modalities reported a significant decrease in post-RBCT rCOEF, which was observed both in children and adults with sickle cell anemia evaluated by MRI ($-0.1$),⁴² in children with sickle cell anemia assessed by asymmetric spin echo sequence MRI technique ($-2\%$ for the whole brain)⁴³ and in SAH patients by PET ($-0.11$).⁴¹ These results are consistent with ours. Our results are also strengthened by several studies highlighting a significant increase in cerebral ${\mathrm{StO}}_2$.^{15–19,21,23,46,47} In Ref. 15, cerebral regional ${\mathrm{StO}}_2$ measured by CW-NIRS increased by 17% in infants and by 4.2% in patients undergoing transfusion during aortic or spinal surgery.²¹ Only in Ref. 24, McCredie et al. did not detect any significant change in cerebral ${\mathrm{StO}}_2$ by bifrontal NIRS measurements in TBI after RBCT. Only one study used a TRS device in infants¹⁸ and found the same: cerebral ${\mathrm{StO}}_2$ increased by $\sim 2\%$ after transfusion. As for $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ and $\mathrm{\Delta }[\mathrm{HbT}]$, our findings are in line with other studies using NIRS.^14,18,24,46 For instance, in Ref. 14, $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ was 0.026 mMol, and $\mathrm{\Delta }[\mathrm{HbT}]$ was 0.02 mMol, which was in the same direction as our study, although with greater absolute value and measured by a frequency domain tissue oximeter.

In our study, at the peripheral level, we have found a significant decrease in rPOEF, a significant increase in $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ and $\mathrm{\Delta }[\mathrm{HbT}]$, but no significant change in $\mathrm{\Delta }[\mathrm{Hhb}]$. The rPOEF result is in agreement with the literature,^5,13,17,45 where a 0.13% significant decrease in the abdomen was found, for example, comparable to ours. Conversely, several studies highlighted a significant increase in the ${\mathrm{StO}}_2$ peripheral level,^{15–19,21,23,46,47} which is also in accordance with our study. For instance, peripheral ${\mathrm{StO}}_2$ increased by 17% in infants¹⁵ and by 1.6% in patients undergoing transfusion during aortic or spinal surgery.²¹ The peripheral increase in $\mathrm{\Delta }[{\mathrm{HbO}}_2]$ and $\mathrm{\Delta }[\mathrm{HbT}]$ was confirmed by NIRS in the literature.^14,18,24,46 Regarding $\mathrm{\Delta }[\mathrm{Hhb}]$, a similar result was found interscapularly by CW-NIRS.¹⁴ However, $\mathrm{\Delta }[\mathrm{Hhb}]$ was often reported to decrease in the literature.

It is possible to speculate that the variations in [Hhb] were too small and could have probably been more relevant in the case of full restoration to normal hemoglobin concentration and hematocrit values, which were not reached by the end of the measurements due to our protocol choice. Furthermore, some studies highlighted RBCT failure in raising the oxygen uptake except that there is a dependency between ${\mathrm{O}}_2$ uptake and ${\mathrm{DO}}_2$, which is generally true only in severe anemic conditions. Generally, more pronounced changes in the brain were found for groups with lower hemoglobin ($<9.7\mathrm{g}/\mathrm{dL}$).^5,45

Both ${\mathrm{rPMRO}}_2$ and ${\mathrm{rCMRO}}_2$ raised due to RBCT, with a greater median increase in the muscle than in the brain and with a statistically significant difference between the two. This result is in disagreement with what was found in some literature related to infants where, for instance, a decrease¹³ or no change ($p>0.05$)^39,48 had been found in ${\mathrm{CMRO}}_2$ by optical methods. Likewise, no ${\mathrm{rCMRO}}_2$ change was found in SAH patients by PET⁴¹ nor in children with sickle cell anemia by MRI.⁴³ Further research is needed to clarify these discrepancies.

As for the difference between cerebral and peripheral level response after RBCT, significant statistical difference was confirmed only for ${\mathrm{rMRO}}_2$ tests.

Overall, the findings for the cohort highlight an improvement in oxy- and total hemoglobin concentration and the oxygen supply, both by optical and standard methods, as expected after RBCT. Moreover, with the obtained information, it is possible to evaluate on an individual basis whether these changes pose any risks or lack them. Indeed, although anemia and RBCT are crucial for a patient’s recovery, collateral issues may arise, especially in cases of impaired cerebral autoregulation, potentially leading to stemming misery perfusion with consequent risk of cerebral ischemia. In light of the post-RBCT findings for the cohort, for instance, we did not identify any misery perfusion, as defined in Ref. 27, nor a hyperperfusion or hypoperfusion, because the rCBF did not significantly change and the rCOEF significantly decreased.

Cerebral autoregulation is often impaired in critically ill subjects,^49,50 especially in TBI.⁵¹ In such eventuality, systemic changes may directly affect the brain and damage it.^51,52 Therefore, it is reasonable to assume that if a robust monitor of cerebral autoregulation becomes available, it would be recommendable to utilize it frequently. In our study, the significant decrease in rCOEF along with no change in the rCBF and ${\mathrm{rCMRO}}_2$ suggests that the cerebral autoregulation was at play. In fact, the decrease in the rCOEF indicates that the brain was extracting less oxygen from the blood, potentially allowing for better ${\mathrm{DO}}_2$ to meet the metabolic demand. This is further supported by the lack of change in rCBF and ${\mathrm{rCMRO}}_2$, indicating a regulation mechanism to maintain consistent oxygen supply. Conversely, the significant rPBF and ${\mathrm{rPMRO}}_2$ increase in quadriceps along with decreased rPOEF indicates hyperemia or luxury perfusion at the peripheral level. This suggests an increased supply of oxygen to the quadriceps muscle, possibly to meet increased metabolic demands, that does not imply deleterious consequences. These are rather adaptive responses to ensure adequate oxygen supply and brain protection.

The first strong limitation of this study is the small sample size. This study was a good starting point to prove the feasibility of this method and spare the need for larger clinical trials.

Once more, we recognize that our study protocol design based only on the monitoring during the first RBC bag transfusion may have been a limitation in the sense of the lower response associated with it than multiple bags. In fact, greater blood exchange may have led to potentially greater changes in our variables and could be worth further exploration to verify whether the results keep consistent over several bags or lead to significant changes.

Furthermore, we acknowledge that the heterogeneity of the patient’s condition prior to RBCT should be addressed and taken into consideration as it plays an important role in the clinician’s decision-making upon RBCT. However, this would only be possible with a larger dataset.

Another important limitation is the fact that, as mentioned in Sec. 2.5, the correction for the hematocrit value was taken into account in the calculation of the ${\mathrm{MRO}}_2$ in the analysis. However, in Ref. 36, a correction method was outlined for the BFI in phantom experiments. This approach is currently under validation against MRI measurements³⁹ by the same group, and if confirmed in vivo, it would be a further step to implement in the future. Moreover, another approach was very recently outlined in Ref. 53. Therefore, our calculations of the BFI should be considered preliminary.

Further limitation can be considered the placement of the probe on a different muscle for one subject which may introduce some variability due to potential differences in hemodynamic and metabolic responses, we believe that the limited sample size and the nature of the study minimize the impact of this variation on our overall conclusions. To reduce potential confounding effects, future studies should maintain consistent probe placement and/or utilize the muscle type/location as a confounding variable in statistical analysis.

The limited number of female participants is another limitation of our study. However, due to the specific nature of our study and the limitations of our sample size, a detailed analysis of sex differences was not statistically feasible. This study did not take any specific action to select patients based on biological sex, and all eligible subjects were approached for recruitment. Similar considerations apply to race/ethnicity, which were not considered here. We acknowledge the importance of further research in a larger clinical trial in the future.

Despite RBCT’s role in critical care (improving oxygen delivery), optimal hemoglobin thresholds for transfusion in critically ill patients, especially TBI, remain debated.^{2,4–6,8–10} It is the clinicians who carefully weigh risks and benefits to decide on RBCT, considering factors such as illness severity, blood pressure stability, brain function, and underlying health conditions. The clinician also chooses the appropriate blood product, volume, and transfusion rate considering the patient’s blood type, specific needs, and risk of reactions. They then monitor the patient’s response to assess transfusion effectiveness. Recent concerns highlight potential negative effects of RBCT in TBI due to increased blood viscosity and impaired cerebral autoregulation.² This has led to a shift toward more restrictive transfusion strategies.^{1–4,9,11,12} Nonetheless, the optimal threshold is patient-specific. Several studies proposed moving beyond hemoglobin levels and using surrogate markers for oxygen delivery issues.^6,21,54 NIRS-derived parameters, such as cerebral ${\mathrm{StO}}_2$, show promise as additional data points for transfusion decisions, potentially enabling personalized thresholds based on brain oxygen needs and contributing to the algorithm decision-making.^10,20

Finally, it is essential to acknowledge the potential influence of the well-known partial volume effect in our study, which is a recognized limitation when using techniques such as DCS and TRS.^55,56 In fact, they are susceptible to variations in tissue composition within the measurement volume, and the presence of different morphological tissue types and structures can lead to uncertainties in the interpretation of the acquired data, for instance, the underestimation of rCBF and rOEF. Although we made efforts to minimize this effect by carefully selecting measurement locations, the inherent challenge of disentangling signals from multiple tissue types, especially the extracerebral layer, was not entirely eliminated. More advanced analysis could be employed to do so. In our case, the probed tissue composition below the probes also additionally considerably varied between cerebral and peripheral measurement locations.

5.

Conclusion

This work focused on establishing the compatibility of DO measurements and critically ill subjects, with responses to RBCT that are more unpredictable with respect to other populations, and quantified it both at cerebral and peripheral levels. However, the applicability could be extended to other populations where anemia is a trigger to RBCT, such as during some types of surgery.

We have shown the potential usefulness of hybrid DO in RBCT management, ideally aiming at improving RBCT decision-making in the future and paving the way to a widespread implication of hybrid DO monitors in clinical trials about RBCT. The acquired knowledge in this important clinical field could even help in personalizing RBC therapy in critical patients and enhance the quality of transfusion therapies and strategies, avoiding deleterious consequences such as hypoxia, hypoperfusion, hyperperfusion, ischemia, and transfusion-associated circulatory overload.

6.

Appendix A: Terminology and Acronyms

In this section, we clarify the distinction between the parameters derived by blood gas sample analysis from the ones by DO measurements. In fact, these two methods/fields often use the same terminology to define similar parameters although of another origin.

The oxy-hemoglobin concentration that is calculated from a blood gas sample by a co-oximeter refers to the fraction of the amount of hemoglobin present in the arterial blood that is bound to the oxygen: it corresponds to the level of oxygenated hemoglobin in the blood, defining its capability to carry oxygen. In this article, we have used the acronym ${\mathrm{HgbO}}_2$ and Hgbr for the derived oxy- and reduced hemoglobin concentrations. As for the arterial total hemoglobin concentration, it includes the cumulative concentration of various forms of hemoglobin, such as carboxyhemoglobin and methemoglobin, in addition to the previously mentioned ones. We used Hgb to refer to it.

Oxy- and deoxy-hemoglobin concentrations that are derived from TRS measurements provide a non-invasive estimate of the content of such chromophores from deep tissue microvasculature, locally related to a specific tissue region and the underneath volume. This method does not derive these components from the blood only, rather from an average between all tissues regionally interacting with the light source and within its penetration depth. Among the usual acronyms used for optical hemoglobin concentrations, in this article, we have decided to adopt the following ones to avoid confusion: microvascular deoxy-hemoglobin concentration ([Hhb]) and microvascular oxy-hemoglobin concentration (${[\mathrm{HbO}}_2]$). The total microvascular hemoglobin concentration is calculated from the sum of these two latter and acronymized as [HbT].

We summarize this in Table 4.

Table 4

Acronyms of the parameters related to hemoglobin concentration to clarify which one is derived by which method.

7.

Appendix B: rMRO₂ Time-Traces

In this section, we showcase the ${\mathrm{rMRO}}_2$ time-traces for all subjects, including the one in Fig. 2, both at cerebral and peripheral level in Fig. 4, making it easier to compare them.

Fig. 4

${\mathrm{rMRO}}_2$ time-traces for all subjects. ${\mathrm{rMRO}}_2$ time-traces for all subjects showing baseline pre-RBCT, blood unit transfusion and post-transfusion period. The beginning and end of the process are highlighted by thick dashed vertical lines, synchronized in time for all variables. Additional events marked during the protocol are represented by vertical lines. The signals are depicted with peripheral (dark green) and cerebral (light green) recordings color-coded as shown in the legend.