Red Blood Cell Triphenotype Classification (RTC): A Functional Framework Integrating Glucose Sink Capacity, Oxygen Kinetics, and Antioxidant Reserve

1. Introduction

The red blood cell is the most numerous cell type in the human body—approximately 2.5 × 10¹³³ cells, comprising 80% of all cellular material by count—yet its clinical characterization remains essentially as it was in the early twentieth century. Routine haematology reports describe a patient’s erythrocytes by haemoglobin concentration, haematocrit, and mean corpuscular volume, with the occasional addition of reticulocyte count and red cell distribution width. This parsimony reflects a long-standing assumption that erythrocytes are functionally homogeneous oxygen carriers whose biological state is fully captured by quantitative measures of mass and size.

This assumption has been progressively undermined over the past decade and decisively challenged by three convergent strands of recent research. First, the 2026 demonstration by Martí-Mateos and colleagues [1] that erythrocytes function as a primary systemic glucose sink during chronic hypoxia—accounting for up to half of glucose disposal in extreme adaptation states—revealed that the red cell is not merely a passive recipient of glucose for its own energy needs but an active and programmable metabolic compartment of the whole organism. Second, the 2021 elucidation by Issaian and colleagues [2] of the band 3 N-terminus as a bidirectional metabolic switch demonstrated that erythrocytes possess an exquisite oxygen-sensing mechanism that toggles glycolytic flux against pentose phosphate flux in response to haemoglobin oxygenation state. Third, the work of Prchal and colleagues on neocytolysis [3,4] has revealed that hypoxia-conditioned erythrocytes carry an inducible self-destruction programme—a miR-21-mediated suppression of catalase activity that renders them selectively susceptible to oxidative clearance upon return to normoxic conditions.

These findings, taken together, describe a fundamentally different biological entity than the passive oxygen carrier of the textbooks. The erythrocyte that emerges from this literature is a programmable, environment-responsive, metabolically active cell whose functional phenotype is determined by its biogenesis environment and modulated dynamically across its lifespan. Different individuals—and different time points within the same individual—harbour erythrocytes with substantially different functional properties.

This biological reality demands a classification framework. Clinical medicine currently lacks any systematic vocabulary for describing the functional state of an individual’s erythrocyte population. As a result, clinicians are unable to articulate distinctions that we now know are biologically real and clinically consequential: between the high-GLUT1 erythrocytes of an acclimatized highlander and the GLUT1-suppressed erythrocytes of a critically ill ICU patient on prolonged hyperoxia; between the 2,3-DPG-enriched erythrocytes of an unrepaired cyanotic heart disease child and the antioxidant-depleted erythrocytes of advanced chronic mountain sickness. These are not abstractions—they are clinically distinct biological states with predictable downstream consequences for systemic glucose disposal, oxygen delivery, and oxidative stress.

In this perspective, we propose the Red Blood Cell Triphenotype Classification (RTC): a three-axis functional framework that organizes erythrocyte phenotypes along axes corresponding to glucose handling, oxygen kinetics, and antioxidant defence. We define ten clinically interpretable phenotype categories (Types α through κ), describe a measurement protocol feasible with existing clinical laboratory infrastructure, and advance four testable predictions for empirical validation. The framework is not intended as a definitive taxonomy—rather, it is intended as a hypothesis-generating heuristic that organizes the rapidly accumulating molecular and clinical literature into a form that can guide further investigation, clinical interpretation, and therapeutic targeting.

2. Theoretical Framework: Three Axes of Erythrocyte Function

The RTC framework rests on three observations from the molecular literature, each corresponding to one axis of the classification.

2.1. Axis 1 — Glucose Sink Capacity (GSC)

Martí-Mateos and colleagues exposed mice to 8% O₂ for four weeks and demonstrated that newly produced erythrocytes display approximately twofold upregulation of GLUT1 protein per cell and threefold increase in glucose uptake rate [1]. Together with hypoxia-induced erythrocytosis, the absolute glucose-handling capacity of the erythrocyte compartment increases four- to sixfold during chronic adaptation. PET/CT imaging confirmed that visceral organs account for only approximately 30% of hypoxia-induced systemic glucose uptake; the remaining 70% is attributable to the erythrocyte compartment. Phlebotomy and transfusion experiments established erythrocytes as both necessary and sufficient for this glycaemic effect.

The Glucose Sink Capacity axis (GSC) operationalizes this property at the individual level. Core indicators include per-cell GLUT1 expression measured by flow cytometric mean fluorescence intensity, ex vivo glucose uptake rate, and total red cell mass. Conceptually, GSC reflects the contribution of the erythrocyte compartment to whole-body glucose disposal—a contribution that ranges from approximately 10% in healthy normoxic adults to 30–50% in chronic hypoxic adaptation and potentially below 5% in conditions where erythrocyte glucose uptake is suppressed.

2.2. Axis 2 — Oxygen Kinetics State (OKS)

Issaian and colleagues mapped the interactome of the band 3 N-terminus and revealed it to function as a bidirectional metabolic switch [2]. Under deoxygenation, deoxyhaemoglobin competitively displaces glycolytic enzymes (notably GAPDH) from the band 3 N-terminus, releasing them into the cytosol and accelerating glycolysis with downstream accumulation of 2,3-bisphosphoglycerate. Under oxygenation, glycolytic enzymes preferentially bind the band 3 N-terminus, suppressing flux through the Embden–Meyerhof pathway and redirecting it through the pentose phosphate pathway with consequent NADPH generation for antioxidant defence. Sun and colleagues independently demonstrated that sphingosine-1-phosphate stabilizes the deoxyhaemoglobin–band 3 interaction in chronic hypoxia [5].

The Oxygen Kinetics State axis (OKS) operationalizes the erythrocyte’s position on the resulting oxygen-release-versus-retention spectrum. Core indicators include 2,3-DPG concentration, P50 of the oxygen dissociation curve, methaemoglobin fraction, and (in research contexts) band 3 phosphorylation status. High OKS reflects rightward-shifted dissociation, prioritizing oxygen release to tissues—characteristic of chronic hypoxic adaptation. Low OKS reflects leftward-shifted dissociation and retention of oxygen by haemoglobin—characteristic of stored erythrocytes, voxelotor-treated sickle cell disease, and the foetal-haemoglobin-dominant infant.

2.3. Axis 3 — Antioxidant Reserve State (ARS)

Prchal and colleagues demonstrated that hypoxia-induced miR-21 suppresses catalase expression in erythrocyte progenitors [3], and that this catalase-poor erythrocyte population becomes selectively vulnerable to oxidative clearance upon return to normoxia—the molecular basis of neocytolysis [4]. The framework that emerges is one of “conditional phenotypic armament with built-in retirement”: newly minted hypoxia-adapted erythrocytes carry both the offensive weapons (enhanced glucose handling, enhanced oxygen release) and the latent vulnerabilities (suppressed antioxidant defence) that enable their selective clearance when no longer needed.

The Antioxidant Reserve State axis (ARS) operationalizes the erythrocyte’s capacity to withstand oxidative stress. Core indicators include catalase enzyme activity, the ratio of pentose phosphate to glycolytic flux, the reduced/oxidized glutathione ratio, and the proportion of erythrocytes expressing eryptotic markers (phosphatidylserine externalization, calcium accumulation). Low ARS marks erythrocytes that are functioning under armament but are vulnerable to clearance; high ARS marks erythrocytes with intact defensive capacity.

2.4. Independence and Interaction of the Axes

Although the three axes are conceptually independent, they are biologically coupled at the molecular level. GSC and OKS share the band 3-GAPDH switch as their upstream regulator—both are coordinately upregulated during chronic hypoxia. ARS is decoupled from GSC and OKS in the sense that it is regulated separately by miR-21 and BNIP3L pathways, but it follows GSC/OKS in the typical hypoxic-adaptation pattern. As a consequence, of the 27 theoretically possible combinations of L/M/H along the three axes, only approximately 8–10 represent commonly observed and clinically meaningful biological states. The remaining combinations either represent rare disequilibrium states (e.g., the pharmacologically-induced phenotype of voxelotor-treated sickle cell disease) or are likely artefactual.

3. The RTC Phenotype Catalog: Types α through κ

Synthesizing biological constraints with clinical observations, we propose ten functional phenotype categories. Each is defined by its position along the three axes and is mapped to specific clinical contexts described in the existing literature.

3.1. Summary Table

Table 1. The Red Blood Cell Triphenotype Classification: ten functional phenotype categories. GSC = Glucose Sink Capacity; OKS = Oxygen Kinetics State; ARS = Antioxidant Reserve State. L = low, M = medium, H = high. Arrows (→) indicate progression during transitional states.

Type	Name	GSC	OKS	ARS	Representative Clinical Context
α	Baseline normoxic	M	M	H	Healthy adults at low altitude
β	Fully adapted hypoxic	H	H	M	Long-term altitude residents; trained altitude athletes (4+ weeks)
γ	Over-adapted hypoxic	H	H	L	Chronic mountain sickness; unrepaired cyanotic CHD (>6 months)
δ	Retiring transitional	M→L	M→L	L	Astronaut return; altitude descent; post-CPAP initiation (0–3 months); post-CCHD repair (0–4 months)
ε	Hyperoxia-suppressed	L	L	M-H	ICU sustained hyperoxia ≥48h; mechanical ventilation FiO2 ≥40%
ζ	Septic collapse	L	L	L	Advanced sepsis; severe ARDS
η	Stored senescent	L	L	L	Stored erythrocytes >21 days; post-transfusion early phase
θ	HbF-dominant infant	M	L	M-H	CCHD infants <3 months; preterm infants
ι	Pharmacological mimetic	M-H	L	L	Sickle cell disease on voxelotor (illustrative; drug withdrawn 2024)
κ	Quantity-quality dissociated	L	M-H	M	CKD-associated renal anaemia

Table 1. The Red Blood Cell Triphenotype Classification: ten functional phenotype categories. GSC = Glucose Sink Capacity; OKS = Oxygen Kinetics State; ARS = Antioxidant Reserve State. L = low, M = medium, H = high. Arrows (→) indicate progression during transitional states.

Type	Name	GSC	OKS	ARS	Representative Clinical Context
α	Baseline normoxic	M	M	H	Healthy adults at low altitude
β	Fully adapted hypoxic	H	H	M	Long-term altitude residents; trained altitude athletes (4+ weeks)
γ	Over-adapted hypoxic	H	H	L	Chronic mountain sickness; unrepaired cyanotic CHD (>6 months)
δ	Retiring transitional	M→L	M→L	L	Astronaut return; altitude descent; post-CPAP initiation (0–3 months); post-CCHD repair (0–4 months)
ε	Hyperoxia-suppressed	L	L	M-H	ICU sustained hyperoxia ≥48h; mechanical ventilation FiO2 ≥40%
ζ	Septic collapse	L	L	L	Advanced sepsis; severe ARDS
η	Stored senescent	L	L	L	Stored erythrocytes >21 days; post-transfusion early phase
θ	HbF-dominant infant	M	L	M-H	CCHD infants <3 months; preterm infants
ι	Pharmacological mimetic	M-H	L	L	Sickle cell disease on voxelotor (illustrative; drug withdrawn 2024)
κ	Quantity-quality dissociated	L	M-H	M	CKD-associated renal anaemia

3.2. Selected Phenotype Profiles

Rather than describing all ten phenotypes in detail (provided in Supplementary Material), we highlight four illustrative profiles that demonstrate the framework’s explanatory power.

Type β — Fully Adapted Hypoxic (RTC-β.HHM)

The prototype of beneficial hypoxic adaptation. Erythrocytes display upregulated GLUT1, increased glycolytic flux, elevated 2,3-DPG, and rightward-shifted oxygen dissociation. Antioxidant reserve is moderately reduced reflecting miR-21 activity, but does not yet manifest as accelerated clearance because the environmental oxidative load is correspondingly low. Long-term highlanders and altitude-trained athletes occupy this phenotype. The reduction in diabetes prevalence at high altitudes [6,7] and the enhanced exercise capacity following altitude training [8] are both explicable as consequences of the Type β phenotype.

Type γ — Over-adapted Hypoxic (RTC-γ.HHL)

When hypoxic adaptation persists at the molecular extreme—either because the hypoxic stimulus is exceptionally chronic and severe (chronic mountain sickness) or because the patient cannot escape it (unrepaired cyanotic congenital heart disease beyond 6 months of age)—the antioxidant reserve becomes critically depleted. Type γ erythrocytes carry the full hypoxic armament but their latent vulnerabilities are now exposed. Clinical manifestations include the cardiovascular complications of chronic mountain sickness [9], the increased fasting hypoglycaemia susceptibility of unrepaired CCHD children originally described by Haymond and colleagues in 1979 [10], and the paradoxically increased adult diabetes risk in repaired CCHD survivors [11,12].

Type ε — Hyperoxia-suppressed (RTC-ε.LLM)

The mirror image of Type β. Sustained ICU hyperoxia (FiO₂ ≥ 40% for ≥48 hours) drives band 3-bound retention of GAPDH, suppressing glycolytic flux and redirecting metabolism through the pentose phosphate pathway. The erythrocyte glucose sink is effectively closed. We have hypothesized elsewhere that this contributes to stress hyperglycaemia in critical illness [13]—a mechanism not currently incorporated into standard models of ICU glucose homeostasis. Reisz and D’Alessandro’s mass spectrometric analysis of stored erythrocytes [14] provides direct experimental evidence for this metabolic shift under hyperoxic conditions.

Type ι — Pharmacological Mimetic (RTC-ι.HLL)

Voxelotor (Oxbryta) pharmacologically left-shifts the haemoglobin oxygen dissociation curve in sickle cell disease, mimicking the hypoxic-adaptation signal while the patient remains in a normoxic environment. The result is a paradoxical phenotype: increased erythrocyte glucose handling under upregulated GLUT1, but suppressed oxygen release (low OKS, opposing the typical hypoxic pattern), with antioxidant reserve depleted by the conditional armament programme. We hypothesize that this mismatch—an erythrocyte armed for hypoxia operating in a normoxic, oxidatively stressful environment—contributed to the increased vaso-occlusive crises and mortality that led to voxelotor’s voluntary market withdrawal in September 2024 [15]. This connection between the pharmacology and the clinical outcomes has not, to our knowledge, been previously articulated.

3.3. Encoding Notation

Each phenotype is denoted by the format RTC-{type}.{GSC}{OKS}{ARS}. For example:

• RTC-α.MMH — healthy normoxic baseline
• RTC-β.HHM — fully adapted hypoxic phenotype
• RTC-ε.LLM — ICU hyperoxia-suppressed phenotype
• RTC-ι.HLL — voxelotor-induced pharmacological mimetic

Transitional states between phenotypes are denoted by arrows (e.g., RTC-β→δ for an altitude-descended individual in the process of neocytolytic clearance). This notation preserves both mechanistic interpretability (the three axes describe what the erythrocyte is doing) and clinical pattern recognition (the type letter directly indicates which clinical population the erythrocyte phenotype most closely resembles).

4. Clinical Applications

4.1. Measurement Protocols at Three Levels

RTC measurement is proposed at three levels of analytical depth, matching different research and clinical contexts.

Level 1: Bedside Approximation (4 markers)

Uses widely available clinical laboratory infrastructure: flow cytometric GLUT1 mean fluorescence intensity, 2,3-DPG concentration by HPLC or enzymatic assay, methaemoglobin fraction by co-oximeter, and reticulocyte count by automated haematology analyser. Resolves GSC and OKS axes precisely; ARS is approximated through methaemoglobin and reticulocyte count. Suitable for clinical observational research (e.g., ICU cohort studies) and rapid bedside characterization. Estimated turnaround: 24–48 hours.

Level 2: Comprehensive Phenotyping (10+ markers)

Adds ex vivo glucose uptake rate, catalase enzyme activity, pentose-phosphate-to-glycolysis flux ratio (requires ¹³³C-glucose metabolomics), reduced/oxidized glutathione ratio, eryptosis quantification (annexin V flow cytometry), and band 3 phosphorylation (research-grade Western blotting with membrane fractionation). Resolves all three axes with full precision. Suitable for mechanistic research and international collaboration (e.g., with metabolomics-equipped laboratories such as the D’Alessandro group at the University of Colorado Anschutz).

Level 3: Longitudinal Trajectory Tracking

Weekly Level 1 measurements in selected patients to track phenotype trajectories through clinical interventions (ICU admission and discharge, CCHD repair, altitude descent, CPAP initiation). The trajectory itself—the sequence of phenotypes traversed and the speed of traversal—is hypothesized to predict clinical outcomes more accurately than any single time-point measurement.

4.2. Integration with Existing Clinical Parameters

RTC does not replace existing haematological assessment; it provides interpretive context. Two illustrations:

Haemoglobin elevation. A haemoglobin of 18 g/dL signifies erythrocytosis. The clinical significance depends on RTC context: in RTC-β.HHM (fully adapted hypoxic) this is a beneficial adaptation; in RTC-γ.HHL (over-adapted) it indicates erythrocyte fragility requiring intervention; in a CKD patient (RTC-κ.LMM) it is paradoxical and suggests EPO supplementation or other modifying factor.

HbA1c interpretation. An HbA1c of 7.0% can correspond to substantially different mean plasma glucose depending on RTC context. In RTC-α.MMH (baseline), HbA1c approximates traditional interpretation. In RTC-γ.HHL (hypoxic over-adaptation), HbA1c is amplified by intracellular erythrocyte glucose exposure beyond the plasma signal—true glycaemic control may be better than the HbA1c suggests. In RTC-ε.LLM (ICU hyperoxia-suppressed), HbA1c may underestimate true mean glucose because intracellular erythrocyte glucose handling has been suppressed. This framework provides a unified mechanistic explanation for the long-recognized HbA1c discordances in high-altitude populations [16], chronic kidney disease [17], and post-ICU patients.

5. Testable Predictions

Four falsifiable predictions follow from the RTC framework:

Prediction 1. In an unselected adult cohort spanning healthy controls and patients with chronic hypoxic conditions, ICU exposure, sleep-disordered breathing, and CKD, unsupervised clustering of the four Level 1 markers will yield discrete groupings that correspond to the ten predicted phenotype categories. Clustering quality—assessed by silhouette coefficients, within-cluster sum of squares, and gap statistics—will be superior to that obtained from CBC parameters alone.

Prediction 2. In ICU patients exposed to sustained hyperoxia (FiO₂ ≥ 40% for ≥48 hours), erythrocyte GLUT1 expression will decrease compared to baseline by at least 20%, accompanied by detectable shifts in 2,3-DPG and methaemoglobin in the directions predicted by Type ε. The magnitude of the GLUT1 reduction will correlate inversely with insulin requirement for glycaemic control, independently of severity scores.

Prediction 3. Following clinical interventions that alter the chronic oxygen environment (altitude descent, CPAP initiation in OSA, surgical correction of CCHD, ICU discharge), erythrocyte phenotype will progress through a predictable trajectory of phenotype types over approximately one red cell turnover cycle (3–4 months). The speed of this transition will correlate with reticulocyte production index and with clinical recovery markers.

Prediction 4. HbA1c will systematically discrepant from continuous glucose monitor-derived mean glucose in a direction predictable from RTC phenotype: amplified in Type β and Type γ, attenuated in Type ε, and concordant in Type α. The magnitude of the discrepancy will correlate quantitatively with the position along the GSC axis.

A planned clustering validation study at Chung Shan Medical University Hospital (CSMUH-RTC-2026-01) will test Prediction 1; complementary observational studies (CSMUH-RHMI-ICU-2026-01 and related protocols) will test Predictions 2–4.

6. Relationship to Prior Frameworks

RTC builds on, and is distinct from, several existing approaches to erythrocyte heterogeneity.

Storage-induced metabolic phenotypes characterized by the D’Alessandro group [14,18,19] established that erythrocytes ex vivo exhibit reproducible metabolic states determined by environmental conditions. RTC extends this insight to the circulating erythrocyte population in vivo.

Donor sex and age effects on erythrocyte function [20] and the recent recognition of clonal haematopoiesis-driven erythrocyte heterogeneity [21] establish that within-individual heterogeneity exists at the level of erythrocyte production. RTC organizes such heterogeneity into a functional taxonomy.

Single-cell erythrocyte glucose uptake heterogeneity [22] established that GLUT1 functional capacity varies substantially even within individuals. RTC provides the macroscopic categories within which this single-cell heterogeneity operates.

The Red Cell Hypoxic Metabolic Index (RHMI), which we previously proposed [23], operates on the same biological substrate but collapses the three axes into a single scalar score. RHMI is suited to rapid clinical bedside use where a unidimensional summary is sufficient; RTC is suited to research and detailed clinical interpretation where axis-level decomposition adds value.

Most importantly, RTC is positioned within the broader programme of functional cellular classification that has transformed immunology over the past three decades (e.g., Th1/Th2/Th17/Treg classification of CD4+ T cells [24], M1/M2 macrophage polarization [25]). These frameworks demonstrate that organizing functionally heterogeneous cell populations into discrete operational categories can substantially advance both mechanistic research and clinical translation. We argue that erythrocyte biology is now at a comparable inflection point.

7. Limitations and Caveats

Several limitations of the present framework should be explicitly acknowledged.

Empirical validation is pending. RTC is presented here as a hypothesis-driven framework. The ten phenotype categories are derived by integrating molecular and clinical literature rather than from primary clustering of unselected patient data. The CSMUH-RTC-2026-01 protocol is designed specifically to test whether these predicted categories emerge from data; it is entirely possible that the empirical landscape will support a smaller or larger number of distinct phenotypes, or a more continuous distribution that is not well captured by discrete categories.

Translation from murine to human. The GSC axis rests in significant part on murine evidence [1]. Although ex vivo human erythrocyte studies are consistent with the framework [2,5], direct demonstration of the magnitude of GLUT1 dynamic range in humans across the proposed phenotypes awaits Level 1 validation studies.

Axis independence is partial. As noted in Section 2.4, the three axes are biologically coupled through the band 3-GAPDH switch (GSC and OKS) and through coordinated hypoxic-adaptation programmes (GSC, OKS, and ARS together). In data-driven analyses, the axes may exhibit substantial correlation rather than the orthogonality that the framework conceptually implies.

Some phenotype categories are speculative. Type ι (pharmacological mimetic) is illustrated through voxelotor, but no patient on voxelotor has been studied with the RTC measurement protocol. The proposed connection between Type ι and voxelotor’s clinical failure remains a hypothesis. Similarly, Type ζ (septic collapse) is inferred from limited published evidence on erythrocyte changes in advanced sepsis [26,27].

The framework is parsimonious by design. Several biological dimensions are deliberately omitted from the three-axis structure: erythrocyte mechanical properties, microvesicle release, ATP and nitric oxide signalling, and clonal genetic background. These may warrant inclusion in expanded versions (e.g., a four- or five-axis framework) as data accumulate.

8. Conclusion

The erythrocyte, long understood as a passive oxygen vehicle, is in fact a dynamic, classifiable, and clinically actionable metabolic system. The Red Blood Cell Triphenotype Classification represents an attempt to construct the first systematic vocabulary for the functional state of an individual’s circulating erythrocyte population—a vocabulary that current clinical haematology lacks despite a fundamental shift in our biological understanding.

The framework’s value will ultimately be determined by three empirical questions: whether the proposed phenotype categories survive unsupervised clustering analysis; whether clinical outcomes can be predicted from RTC phenotype above and beyond conventional clinical markers; and whether therapeutic interventions can be guided by phenotype assessment. These questions are amenable to investigation with existing clinical laboratory infrastructure and modest research resources. The framework presented here is offered as a starting point for that investigation, not as a definitive taxonomy.

We close with an observation. In the words used to describe the erythrocyte in textbooks of the past hundred years—oxygen carrier, biconcave disc, non-nucleated—there is no language for what is now becoming visible: that the circulating erythrocytes of an individual constitute a heterogeneous, environment-responsive, classifiable metabolic system. RTC is an attempt to provide such a language. Whether the specific categories survive empirical scrutiny matters less than that the conceptual shift takes hold. The red blood cell deserves a vocabulary commensurate with the biology we now know it to embody.