We performed a study in which we assessed mitoPO2 and mitoVO2 in critically ill patients with anemia before and after RBC transfusion. MitoPO2 values in critically ill patients with anemia were not substantially lower than values previously observed in other critically ill patients and did not significantly change during the first 24 h after RBC transfusion. MitoPO2 and mitoVO2 values were not notably associated with Hb concentrations, parameters of severity of illness, and markers of tissue perfusion or cellular oxygenation in our study population. In patients with a pre-transfusion concentration > 7 g/dL we saw a dissociation between mitoPO2 and mitoVO2 with respect to the effect of RBC transfusion. In patients with a pre-transfusion Hb concentration ≤ 7 g/dL, both mitoPO2 and mitoVO2 did not increase after RBC transfusion but rather both decreased over time (24 h).

Our study showed relatively normal mitoPO2 values, in critically ill patients with anemia, that did not increase after a RBC transfusion. An absence of effect of RBC transfusion on mitoPO2 is in line with a previous trial showing no benefit of guiding RBC transfusion according to a marker of tissue oxygenation, although some observational studies have suggested benefit of this approach [27, 36, 37]. Theoretically one would expect an increase in tissue oxygenation after RBC transfusion in patients who benefit from transfusion, especially in the critically ill patient with markers of low tissue perfusion and low Hb concentration before RBC transfusion [11, 27]. However, most of our study participants had a transfusion trigger above 7 g/dL, as well as normal mitoPO2, lactate, MAP, ScvO2, and pCO2 gap values before RBC transfusion. These ‘normal’ indices of tissue perfusion before RBC transfusion, suggest that a too liberal transfusion trigger was used, which may possibly explain the absence of an increase in mitoPO2 values after RBC transfusion. This lack of effect caused by a too liberal transfusion trigger is supported by the fact that low mitoPO2 values are expected with hematocrit values of 0.14 L/L or lower [23], which none of our study participants had. Just one of our study participants received more than one RBC transfusion units before mitoPO2 measurements, thereby limiting the interpretation of our results to only critically ill patients receiving RBC transfusion. Furthermore, it might be suggested that the critical illness has not (yet) led to low mitoPO2 values. An alternative explanation may be that RBC transfusion is ineffective at improving tissue oxygenation in this cohort of critically ill patients which are not actively resuscitated. Pre-transfusion baseline values have previously been shown important in predicting the response to transfusion on a microcirculatory level. A recent study suggested that a critically low mitoPO2 value of 30 mmHg or lower would be indicative of tissue hypoxia [23]. Of note, only 2 of the 63 critically ill patients with anemia pre-transfusion had a mitoPO2 < 30 mmHg. All the above considerations raise the question if RBC transfusions were needed in most of our critically ill patients with anemia [11, 12]. A final explanation of an absence of effect of a single RBC transfusion on mitoPO2 might be that the effect of RBC transfusion might be too small to increase mitoPO2 value [21]. Being a new monitoring technique, multiple studies have already been performed with the COMET measurement device. Studies with the COMET measurement device in healthy volunteers and critically ill patients have shown a normal mitoPO2 ranging between 40 and 70 mmHg [24, 31,32,33,34]. A recent study into the mitochondrial oxygen measurement with the COMET measurement device has determined that a normal range of mitoPO2 in physiological steady state is between 40 and 60 mmHg in the skin [23]. Indeed, our median mitoPO2 values correspond with normal mitoPO2 values.

The mitoVO2 in our population ranged from 2.8 to 3.7 mmHg/s corresponding with mitoVO2 values between 3.3 and 4.6 mmHg/s in critically ill patients that have been described in other studies [31, 32]. This is lower than the mitoVO2 values found in healthy volunteers ranging from 5.8 to 6.7 mmHg/s [14, 33], suggesting a decreased cellular respiration in critically ill patients with anemia. Since mitoVO2 is not directly measured by the COMET measurement device, the mitoVO2 needs to be calculated from the mitoPO2 values during application of pressure on the COMET probe. Different mechanisms have been described to calculate the mitoVO2, using the Michaelis–Menten kinetics [14], fitting a sigmoid curve [32], or using a linear function [31]. These different approaches could lead to different results, therefore mitoVO2 comparison should be done cautiously.

Interestingly, when looking in patients with lower (≤ 7 g/dL) versus a higher (> 7 g/dL) pre-transfusion Hb concentration, we observed different baseline values of mitoPO2 and mitoVO2 and different effects of RBC on mitoPO2 and mitoVO2, that were unexpected. Contrary to our expectations, the patients with a pre-transfusion Hb concentration < 7 g/dL had somewhat higher mitoPO2 values compared to patients with Hb ≥ 7 g/dL. The higher mitoPO2 value could have been the result of mitochondrial adaptation for an optimal mitochondrial energy metabolism [9, 28, 29]. It has been described that oxygen consumption in the mitochondria can be reduced in response to mitochondrial hypoxia, leading to excess oxygen in the mitochondria. Inflammatory mediators, e.g., nitric oxide, in sepsis and shock have been described causing this mitochondrial adaptation [9, 28, 30]. However, this is contradicted by our finding that the calculated mitochondrial oxygen consumption was higher in in critically ill patients with a pre-transfusion Hb concentration < 7 g/dL compared to patients with Hb ≥ 7 g/dL. It would be interesting to study the activity of mitochondrial adaptation mechanisms and their influence on the mitoPO2 in future studies.

In the patients with a higher pre-transfusion Hb concentration, a dissociation between the effect of RBC on mitoPO2 and mitoVO2 was observed, i.e., mitoPO2 increased and mitoVO2 did not change after RBC. Concomitantly, we observed a decrease both in mitoPO2 and mitoVO2 after RBC in patients with a low pre-transfusion Hb concentration. A possible explanation that has been offered before may be the nitric oxide-dependent vasodilatation effect of RBC transfusions due to plasma-free Hb [27]. Furthermore, this may suggest an inability of cells to use oxygen, previously referred to as cellular dysoxia, as shown before in sepsis patients [28].

A relatively large part of the mitoPO2 values had a low signal quality, which persisted until one hour after the end of RBC transfusion. This has been reported in other studies using the COMET measurement device [31, 32, 35]. Importantly, this is one of the first studies describing the characteristics of the patients with missing mitoPO2 values due to signal quality below a protocol-set threshold. It seems that overall, these patients were more critically ill compared to the critically ill patients with valid mitoPO2 measurements, which may have led to an overestimation of mitoPO2. The critical illness might have influenced the ALA-plaster absorption or PpIX formation, resulting in a sub-par signal quality after four hours ALA plaster induction. Our data suggest that more than 4 h ALA plaster induction may be needed to guarantee adequate upregulation of PpIX in critically ill patients for a qualitative mitoPO2 measurement with the COMET measurement device.

Strengths of our study entailed the prospective nature of our study in multiple study sites, as well as the gathering of data at multiple timepoints. The data gathering was made as complete as possible to gain as much insight into the critically ill patients with anemia. Furthermore, the study design mimics clinical practice, making it more applicable to the daily practice.

Despite the high protocol adherence, missing data could not be prevented. Overall, most missing data were due to logistical issues, i.e., measurement in a weekend day or night time when no one of the study team was available. Therefore, missing not at random could not be ruled out, hence missing data could not be handled with imputation methods. We therefore interpreted our data cautiously, keeping in mind the large confidence intervals of mitoPO2 values, while looking into the mean and median mitoPO2 values.

This study is one of the first studies looking into bedside cellular oxygenation in patients receiving RBC transfusion and the effect of this RBC transfusion on the cellular oxygenation. It shows that in critically ill patients, overall mitoPO2 values are normal, and that when administered based on an Hb trigger, RBC transfusion does not result in an increase in mitoPO2 or mitoVO2. Findings are in line with other studies trying to determine the efficacy of RBC transfusion on the level of tissue oxygenation. Whether results are due to a too liberal RBC transfusion policy, or to an inability to utilize oxygen, or to a decrease in perfusion, or to another cause, cannot be dissected from our findings. In follow-up studies on the utility of mitoPO2 to guide interventions to improve tissue oxygenation, it should be noted that signal quality is impaired in the most severely ill patients.