Testosterone supplementation increases red blood cell susceptibility to oxidative stress, decreases membrane deformability, and decreases survival after cold storage and transfusion

Abstract

Background:

Blood collection from donors on testosterone therapy (TT) is restricted to red blood cell (RBC) concentrates to avoid patient exposure to supraphysiological testosterone (T). The objective of this study was to identify TT-related changes in RBC characteristics relevant to transfusion effectiveness in patients.

Study design:

This was a two-part study with cohorts of patients and blood donors on TT. In part one, we conducted longitudinal evaluation of RBCs collected before and at 3 time points after initiation of T. RBC assays included storage and oxidative hemolysis, membrane deformability (elongation index), and oximetry. In part two, we evaluated the fate of transfused RBCs from TT donors in immunodeficient mice and by retrospective analyses of NIH’s vein-to-vein databases.

Results:

TT increased oxidative hemolysis (1.45-fold change) and decreased RBC membrane deformability. Plasma free testosterone was positively correlated with oxidative hemolysis (r = 0.552) and negatively correlated with the elongation index (r = −0.472). Stored and gamma-irradiated RBCs from TT donors had lower posttransfusion recovery in mice compared to controls (41.6±12 vs. 55.3±20.5%). Recipients of RBCs from male donors taking T had 25% lower hemoglobin increments compared to recipients of RBCs from to non-TT male donors, and had increased incidence (OR, 1.80) of requiring additional RBC transfusions within 48 hours of the index transfusion event.

Conclusions:

TT is associated with altered RBC characteristics and transfusion effectiveness. These results suggest that clinical utilization of TT RBCs may be less effective in recipients who benefit from longer RBC survival, such as chronically transfused patients.

INTRODUCTION

Testosterone (T) supplementation is commonly prescribed to hypogonadal men and to transmen undergoing gender-affirming hormone therapy. In recent years, the growing awareness of ‘low T’ and its potential impact on men’s health has led to a rise in the popularity of testosterone therapy (TT) and to concerns regarding side effects and potential abuse of this clinical practice.¹ Erythrocytosis, a condition defined as abnormally high concentration of red blood cells (RBCs) in the circulation, is one of the most common side effects of TT^2,3 that typically occurs in patients taking excessive T dosages. This condition can be managed by therapeutic phlebotomies at blood collection centers. The FDA has authorized the use of RBCs from TT-related therapeutic phlebotomies for transfusion purposes, whereas plasma and other blood components are discarded to avoid patient exposure to supraphysiological testosterone.⁴ Despite these regulations, we reported abnormally high concentrations of testosterone in leukocyte-reduced RBC units from TT donors.⁵

Male sex is associated with increased susceptibility to hemolysis in RBC disorders, such as sickle cell disease^6,7, and blood products from male donors hemolyze more than females in routine cold storage.^8,9 Our preclinical studies in mice have demonstrated that the sex dichotomy in RBC biology and hemolysis is principally mediated by exposure to testosterone in males.¹⁰ In these studies, castration was associated with lower rates of oxidative and osmotic stress-induced hemolysis and with improved RBC storage and post-transfusion recovery (PTR; defined as the percentage of donor RBCs in the recipient circulation) compared with intact male RBCs. Further, T administration in castrated mice restored the hemolytic phenotype of male RBCs, whereas ovariectomy had no significant effect on RBC phenotype for stress-induced hemolysis.¹⁰ As mature RBCs (erythrocytes) lack DNA, we suggested that the primary effects of TT on RBC biology take place during erythropoiesis.

In light of these findings, we hypothesized that T supplementation in human produces adverse changes in key RBC functional characteristics including antioxidant capacity, membrane deformability, and susceptibility to hemolysis. This is of clinical concern to patients on TT and recipients of blood transfusion from TT donors, as intravascular hemolysis and cell-free hemoglobin may increase the risks of vasoconstriction, oxidative injury, inflammation, and infections.^11–13 To test this hypothesis, we conducted a two-part study in which we initially determined the effects of longitudinal increases in plasma testosterone concentration on RBC characteristics including deformability, oxidative hemolysis, hematological phenotype, and metabolism. In the second part, we evaluated the transfusion effectiveness of RBCs from TT donors using immunodeficient mice and retrospective analyses of patient hemoglobin increments (a marker of RBC transfusion effectiveness) following a single RBC unit transfusion event from a TT donor using the National Heart, Lung, and Blood Institute Recipient Epidemiology Donor Evaluation Study III (REDS-III) databases.¹⁴

MATERIALS and METHODS

Study cohorts:

This study was conducted under regulations applicable to all human subject research supported by federal agencies including institutional review board approval from the University of Washington in Seattle, University of California in San Francisco, and WIRB-Copernicus Group (WCG). The data were obtained from two cohorts of individuals who consented to participate in this study. Part one was aimed at evaluating longitudinal changes in RBC characteristics and hemolysis before and after initiation of TT. This cohort was managed by Dr. Bradley Anawalt from the University of Washington, who oversaw the recruitment of 8 individuals who received TT for hypogonadism (N=5), as part of gender reassignment treatments (N=2) or treatment for male XX syndrome (N=1). Two of the 8 participants missed one visit due to the COVID 19 pandemic. Part two was aimed at determining the transfusion effectiveness of TT RBCs using a xenotransfusion mouse model and retrospective analyses of REDS-III vein-to-vein databases. TT donors were recruited by Vitalant’s Cell Sourcing & Special Collection group at Vitalant Research Institute in Denver. This cohort included 31 individuals on TT and 29 controls (not on TT), who were matched with TT donors based upon sex, age, self-reported ethnicity/ancestry, and ABO blood group. Among participants on TT, 8 were registered as allogenic blood donors at Vitalant, and 23 were specifically recruited for this study.

Part 1: Longitudinal evaluation of plasma and RBCs from individuals on TT:

Each participant from the University of Washington cohort provided whole blood samples (3–6 mL) before and about 2, 6, and 12 months after TT initiation. Clinical evaluation of whole blood samples included quantification of plasma free and total testosterone, plasma sex hormone binding globulin (SHBG), complete blood count, and iron indices. All tests were conducted at the University of Washington CLIA-certified clinical laboratories. Additionally, RBCs isolated from whole blood samples were evaluated for hemolysis and metabolic assays as detailed below.

Part 2: Blood collection, RBC manufacturing procedures, RBC gamma irradiation, RBC hemolysis and functional assays:

Detailed description of these procedures is available in Supplemental Methods and Materials file. In brief, leukocyte-reduced RBC units (LR-RBCs) in additive solution-3 (AS-3) collected from TT donors and matched controls were stored under blood banking conditions (1–6° C) for six weeks and were tested weekly for RBC functional and hemolysis assays. Gamma-irradiation was conducted on aliquots from 8 TT and 7 control LR-RBC units. Hemolysis was determined using high throughput assays^8,10,15,16, which evaluated RBC responses to spontaneous (cold storage) stress, oxidative stress, and osmotic shock, as described in Supplemental Methods and Materials. Oxidative hemolysis was induced by incubating RBCs in the presence of 2,2’-azobis-2-methyl-propanimidamide, dihydrochloride (AAPH, 150 mmoL). This assay generates reactive oxygen species and lipid radicals via thermal (37°C) decomposition of AAPH.¹⁷ Osmotic hemolysis was induced by incubating RBCs in hypotonic solution (pink test buffer).¹⁸ RBC metrics of membrane deformability were measured by a Laser Optical Rotational Red Cell Analyzer (LORRCA, RI, USA) according to the manufacturer protocol.¹⁹ Relevant to this study, we report the maximal elongation index (EI_max), which represents maximal RBC deformability at the isotonic point (~300 mOsm/kg), and the elongation index at the hypertonic region (~350 to 500 mOsm/kg, EI_hyper), which reflects cellular hydration and cytoplasmatic viscosity.²⁰ Blood oximetry measurements were quantified on a blood gas analyzer (ABL90 FLEX PLUS, Radiometer America, Brea, CA, USA), which provided the concentrations (millimolar) of potassium and lactate, percent methemoglobin, and percent oxygen saturation (sO₂). RBC hematological indices including RBC concentration (M/μL), mean corpuscular volume (MCV), and RBC distribution width (RDW) were determined by blood bank hematology analyzer (Sysmex, IL, USA).

Quantification of RBC posttransfusion recovery in NSG mice:

The Institutional Animal Care and Use Committee at Labcorp Early Developmental Laboratories Inc. (San Carlos, CA) approved research under Animal Welfare Assurance A3367–01. RBC posttransfusion recovery was determined using a xenotransfusion mouse model as we described before.^21,22 In this study, non-irradiated (N = 16 TT and 15 matched controls) and gamma-irradiated (N = 8 TT and 7 matched controls) RBCs were transfused into immunodeficient NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice.²³ Blood was collected from mice pre-injection and at 10 minutes, 2, 4, 8, and 24 hours post infusion. RBC PTR was quantified by flow cytometric analysis using the human-specific marker for erythrocyte membrane glycophorin A (CD235a-phycoerythrin; Biolegend, HI264). Data normalization was generated using the National Centre for the Replacement, Refinement and Reduction of Animals in Research (NC3Rs) method.²⁴ Detailed description of the experimental procedures is available in Supplemental Materials and methods.

Determination of TT-RBC transfusion effectiveness:

RBC transfusion effectiveness was determined by retrospective analyses of donor and recipient (patients) electronic records available from the National Heart, Lung, and Blood Institute REDS-III RBC-Omics¹⁴ and vein-to-vein databases²⁵, as described before.^26–28 The primary outcome was recipient’s hemoglobin increment after a single unit RBC transfusion from a TT donor or a male control (not receiving TT). This is calculated as the difference in patient hemoglobin (Hb) concentration (g/dL) before and after each RBC transfusion episode (reported as delta hemoglobin or ΔHb) and adjusted for other co-factors including donor age, sex, hemoglobin - component apheresis, irradiation, storage duration (days), storage solution, recipient age, sex, body mass index, hemoglobin level, and information on concomitant transfusion of other blood components (platelets, plasma). Only recipient Hb values collected ≤18 h before transfusion and within 24 hours after transfusion were included in this study. Additional outcomes included the need for subsequent RBC transfusion events following a single RBC transfusion from TT donors and controls. This was determined by calculating the odds ratio of subsequent RBC transfusion 48-hours after the index transfusion event. Overall, data from 2429 RBC-Omics donors were linked to recipients, of whom 32 received TT-RBCs and 2,397 received control RBCs.

Statistical analysis:

Data were analyzed using GraphPad Prism 10 (GraphPad Software, Inc, La Jolla, CA). Differences in donor demographics (e.g., age, testosterone, hemoglobin) between TT and control donors were determined by unpaired t-test. Testing for differences in repeated measurements of certain variables were conducted using Prism’s multivariable mixed-effects model and Fisher’s LSD post hoc tests. Correlations between T and RBC metrices of hemolysis or deformability were determined by Person correlation tests. In all analyses, significance was determined by p≤0.05. Specific to RBC transfusion effectiveness, univariable linear regression analyses determined the associations of all a priori selected donor, component, and recipient covariates with the outcomes. Any covariate that was associated with the outcomes was included in a multivariable linear mixed effects model, which accounted for modifiers of clinical outcomes (e.g., repeated transfusions, gamma irradiation, storage duration).

RESULTS

Part 1. Testosterone therapy is associated with changes in RBC susceptibility to oxidative stress, membrane deformability, and blood metabolites:

We characterized RBCs from 8 individuals, who provided blood samples before and at 3 time points (1–2, 5–6, 10–12 months) after initiation of TT. Participant demographic information including age, gender identity, and reasons for TT is summarized in Table 1. Longitudinal evaluation of whole blood samples suggested that the increase in plasma free and total testosterone was concomitant with decreased serum SHBG, and with increased RBC concentration, hemoglobin, hematocrit, RDW, transferrin, total iron-binding capacity (TIBC), and lactate. These changes reached significance (p<0.05) in at least on visit after TT initiation (Table 2). Other noteworthy changes were decreased plasma ferritin and chloride concentrations, and increased plasma potassium concentration (not significant).

Table 1.

Demographic parameters of 8 patients who enrolled in part one the study.

Patient #	Age (years)	Gender identity	Reason for TT	Ethnicity	Testosterone preparation
1	56	Cisgender man	Primary hypogonadism	White	Testosterone gel daily
2	69	Cisgender man	Secondary hypogonadism	White	IM testosterone cypionate weekly
3	37	Cisgender man	Primary hypogonadism due to testicular dysgenesis	Latino/Hispanic	IM testosterone cypionate weekly
4	33	Transgender man	Gender-affirming hormone therapy	White	IM testosterone cypionate weekly
5	21	Transgender man	Gender-affirming hormone therapy	White	IM testosterone cypionate weekly
6	69	Cisgender man	Mixed primary and secondary hypogonadism (mumps orchitis plus obesity and undertreated sleep apnea)	White	Testosterone gel daily
7	31	Cisgender man	Disorder of sexual development (XX male with CAH due to 21-hydroxylase deficiency	Latino/Hispanic	IM testosterone cypionate weekly
8	21	Cisgender man	Congenital hypogonadotropic hypogonadism (Pasqualini syndrome- previously known as fertile eunuch)	White	IM testosterone cypionate weekly

TT, testosterone therapy. CAH, congenital adrenal hyperplasia. IM, intramuscular.

Table 2.

Longitudinal evaluation of plasma testosterone, RBC hematological indices, iron indices, and blood metabolites in blood specimens from individuals before and after testosterone therapy (TT) initiation.

	Pre-TT Mean±SD (Min-Max)	TT-1 Mean±SD (Min-Max)	TT-2 Mean±SD (Min-Max)	TT-3 Mean±SD (Min-Max)
Free testosterone (pg/mL)	17.7±15.6 (3.1–39.9)	153.3±153.9** (41.5–383.8)	129.3±95.9 * (52.2–308.1)	122.8±71.2* (15.8–225.2)
Total testosterone (ng/dL)	16.6±34.6 (0.4–100)	232.1±478.3 (2.6–1286)	291.8±403 (5.4–1178)	567.9±426.1** (4.1–1076)
RBCs (10 6 /μL)	4.6±0.35 (4.1–5.1)	4.9±0.4* (4.3–5.5)	5.1±0.3*** (4.5–5.6)	5.1±0.5** (4.4–5.8)
Hemoglobin (g/dL)	14.1± 0.7 (12.7–14.8)	14.7±0.9 (13.1–15.6)	15.1±0.5** (14.4–15.9)	15.2±1.1** (14.3–16.9)
Hematocrit (%)	41.9± 2.2 (39–45)	44.4±2.5* (41–48)	45.5±1.5** (43–47)	45.7±3.7** (40–50)
MCV (fL)	91.1±4.4 (83.9–100)	91.1± 5.1 (85–99)	89.6±4.9 (84–97)	90.3±3.5 (85–95)
RDW-CV (%)	12.7±0.5 (11.9–13.6)	12.9±0.3 (12.2–13.1)	13.6±1.1* (12.4–15.3)	12.9±0.9 (11.5–14.1)
SHBG (nmol/L)	53.8±21.8 (22–86)	38.1±15.2* (20–60)	37.9±22.7** (17–86)	37.7±17.5** (22–70)
Transferrin (mg/mL)	244.1±35.5 (206–306)	277.7±58.8*** (196–355)	264.4±52.6** (180–327)	257.7±43.7 (183–312)
Ferritin (ng/mL)	81.4±83.3 (16–236)	64.6±107.3 (4–301)	55.8±69.3 (9–214)	53.3±87.9 (12–252)
TIBC (μg/mL)	341.6±49.7 (288–428)	388.7±82.5*** (274–497)	370.1±73.8** (252–458)	360.9±61.3 (256–437)
Lactate (mmol/L)	3.6±0.7 (2.6–5.1)	3.5±0.9 (2.3–4.4)	4.8±1.2* (3.4–7.1)	4.6±1 (2.9–5.8)
Glucose (mmol/L)	3.8±0.4 (3.2–4.3)	4.0±0.6 (3.3–4.7)	4.1±1.3 (2.2–6.7)	3.7±0.7 (2.4–4.4)
Potassium (mmol/L)	26.2±4.3 (21.8–35.8)	27.9±3.9 (24.4–36.8)	28.3±3.8 (24.2–35.3)	29.5±6.0 (21.8–40.6)
Chloride (mmol/L)	100±1.7 (97–102)	99.7±2.2 (97–102)	99.6±2.3 (96–102)	98.6±2.2 (95–102)

TT-1, TT-2, and TT-3, 1–2 months, 5–6 months, and 10–12 months after TT initiation (Pre-TT). N=7–8. RBC, red blood cells; MCV, mean corpuscular volume; RDW, red cell distribution width; SHBG, sex hormone binding globulin; TIBC, total iron-binding capacity.

^*p<0.05,

^**p<0.01,

^***p<0.001 compared with Pre-TT using mixed-effects analysis and Fisher’s LSD test.

Mean oxidative hemolysis significantly increased shortly (TT-1), and 5–6 months (TT-2) after TT initiation (Fig. 1A, ~1.45 and 1.35 fold change, respectively). Oxidative hemolysis was positively correlated with individuals’ plasma free testosterone (Pearson r = 0.552, p=0.002; Fig. 1B). A decrease in whole blood levels of percent methemoglobin was observed 10–12 months after TT initiation (TT-3; Fig. 1C). Methemoglobin was negatively correlated with free testosterone (Pearson r=−0.430, p=0.025; Fig. 1D). TT was also associated with changes in RBC rheological properties (Fig. 2A–B) manifested by decreased RBC membrane deformability (EI_max) and cellular hydration (EI_hyper). EI_max was negatively correlated with plasma free testosterone (Pearson r = −0.472, p=0.015; Fig. 2C).

Figure 1. Testosterone therapy (TT)-related changes in RBC susceptibility to oxidative hemolysis and methemoglobin concentration.

Data derived from 8 individuals before and after testosterone replacement therapy (TT) initiation. TT-1, TT-2, and TT-3, 1–2 months, 5–6 months, and 10–12 months after TT initiation (Pre-TT). N=7–8. A. Percent oxidative hemolysis on each visit. *, p<0.05 compared with Pre-TT. B-C. Pearson correlation analyses between individuals’ plasma free testosterone (B) and percent oxidative hemolysis. r, Pearson correlation coefficient. C. Percent methemoglobin on each visit. ***, p=0.0007 compared with Pre-TT. D. Pearson correlation analyses between individuals’ plasma free testosterone and percent methemoglobin. A and C, before-after plots. Each symbol represents a unique individual.

Figure 2. Testosterone therapy (TT)-related changes in RBC membrane deformability and hydration state (viscosity).

RBC indices of membrane deformability (maximal elongation index, EI_max) and viscosity (elongation index at the hypertonic region, EI_hyper) were collected (LORRCA Osmoscan module) from 7 individuals before and after testosterone replacement therapy (TT) initiation. TT-1, TT-2, and TT-3, 1–2 months, 5–6 months, and 10–12 months after TT initiation (Pre-TT). N=6–7. A. Longitudinal measurements of EI_max. *, p<0.05 compared with Pre-TT. B. Longitudinal measurements of EI_hyper. *, p=0.026 compared with Pre-TT. A and B, before-after plots. Each symbol represents a unique individual. C. Pearson correlation analyses between individuals’ plasma free testosterone and EI_max. r, Pearson correlation coefficient.

Part 2. Testosterone therapy is associated with altered RBC survival in cold storage and after transfusion of gamma-irradiated RBCs into mice:

We verified whether TT modulates RBC survival in storage and after transfusion in a cohort of 21 donors on TT and 20 matched controls. In this cohort, TT was significantly (p<0.05) associated with higher donor BMI and plasma free testosterone (Table 3). Mean percent storage and oxidative hemolysis were higher in stored LR-RBCs from TT donors (Fig. 3A–B); however, these differences did not reach statistical significance in a multivariable analysis, which accounted for storage duration, donor effect, and TT status. The same analyses associated TT with higher storage-related RBC potassium leak and lactate formation, percent RBC oxygen saturation (sO₂), and RDW compared with controls (Fig. 3C–F). These differences were significant (p<0.05) in a multivariable analysis (i.e., TT status for potassium, sO₂, and RDW), and in post hoc multiple comparison analyses of differences in specific time points.

Table 3.

Baseline parameters of testosterone therapy and control blood donors for post-transfusion recovery study in immunodeficient mice.

	TT (N=21)		Controls (N=20)		p (t-test)
Age (years)	50.5±10.3		52.2±12.2		0.638
Sex	Males N=19	Females N=2	Males N=18	Females N=2
Hemoglobin (g/dL) at donation	17.2±1.6		16.2±1.9		0.068
BMI (Kg/m2)	29.9±3.7		27.2±3.5		0.023
Blood group: A,B,O,AB	N= 7,3,10,1		N= 6,3,10,1
Ethnicity: White, African American, Asian American, NA	N=17,1,1,2		N=17/1/1/1
Free testosterone (pg/mL)	266.3±309.3		97.5±101.9		0.034
Total testosterone (ng/dL)	802.4.±697.1		459.4±317		0.063

TT, testosterone therapy. BMI, body mass index. Free testosterone, normal range 47–244 pg/mL. Total testosterone, normal range 300–890 ng/dL. NA, not available

Figure 3. Effect of TT on RBC characteristics and indices of hemolysis in cold storage.

Leukocyte-reduced RBC units collected from 21 donors on testosterone replacement therapy (TT) and 20 controls were stored for six weeks and evaluated weekly for A. Percent storage hemolysis. B. Percent oxidative hemolysis. C. Extracellular potassium (K⁺) concentration (mmol/L). D. Lactate concentration (mmol/L). E. Percent oxygen saturation (sO₂). F. Red cell distribution width-coefficient of variation (RDW-CV). *, p < 0.05; ***, p < 0.001. Mean±SD. p values obtained from Prism’s multivariable mixed-effects model with Fisher’s LSD test. A, B, F, N=21 TT and 20 Controls. C, D, E, N=17–18 TT and 16–17 Controls.

Relevant to RBC transfusion effectiveness, we determined the PTR of naïve (non-irradiated) and gamma-irradiated stored RBCs in immunodominant mice. As shown in Fig. 4A, the PTR of naïve RBCs from TT donors was similar to that of controls. Conversely, gamma-irradiated RBCs from TT donors exhibited lower PTR compared with controls shortly after transfusion (10 min, 41.6±12 % versus 55.3±20.5 %, p=0.0034; Figure 4B). Additionally, an interaction was identified between TT status and the kinetics of RBC clearance over time (TT status x time, p=0.036).

Figure 4. Effect of TT on RBC posttransfusion recovery in immunodeficient mice.

Stored (39–42 days) leukocyte-reduced RBC units collected donors on testosterone replacement therapy (TT) and matched controls were transfused into immunodeficient mice (NSG), as detailed in Materials and Methods. A. Percent RBC posttransfusion recovery in non-irradiated RBCs (N=16 TT and N=15 controls). B. Gamma-irradiated RBCs (N=8 TT and N=7 controls). Mean±SD. **; p=0.0034 from Prism’s multivariable mixed-effects model with Fisher’s LSD test.

To verify whether TT-donor RBCs are more susceptible to gamma irradiation-induced injury, we determined the storage recoveries of naïve and gamma irradiated LR-RBC units from TT donors and matched controls (Supplemental Fig. S1 and supplemental Table S1). TT was associated with increased storage hemolysis, lactate formation, potassium leak, higher percent sO₂, and with lower percent methemoglobin compared with controls, irrespective of gamma irradiation. RBC gamma-irradiation was associated with increased storage hemolysis on storage weeks 4 and 6, and with RBC potassium leak in both donor groups. No significant differences between the donor groups were observed in percent oxidative hemolysis.

TT-donor blood is associated with lower hemoglobin increments in transfusion recipients and with higher odds ratio for additional RBC transfusion:

We evaluated the transfusion effectiveness of RBCs from TT donors by conducting retrospective analyses of hemoglobin increments (REDS-III RBC-Omics and vein-to-vein databases) following single RBC transfusion events from TT donor (N=32) and controls (male donors not on TT, N=2,397). Donor, component and transfusion recipient characteristics are summarized in Supplemental Table S2. As summarized in Table 4, hemoglobin increments were reduced (p=0.048) for RBC units from TT donors (0.75 g/dL) compared to RBC units from non-TT male donors (1.03 g/dL) in multivariable regression models. Further, RBC units from TT donors were associated with an increased incidence (odds ratio, 1.80; p = 0.035) and number of additional RBC transfusions (1.22 versus 0.61 RBC units, TT versus controls, respectively; p = 0.006) within 48 hours of the index transfusion event. Multivariable models also demonstrated that in addition to TT, factors such as donor and recipient hemoglobin levels, RBC unit gamma irradiation, and recipient body mass index were significantly associated with hemoglobin increments following RBC transfusion (Supplemental Table S3).

Table 4:

Hemoglobin increments and subsequent transfusion events after single unit RBC transfusion from donors on testosterone therapy (TT) and controls.

Donor group	Hgb increments g/dL [95% CI]	48-hour RBC transfusion event Odds ratio [95% CI]	# of RBC transfusions within 48 hours [95% CI]
Controls (n=2,397)	1.03 [1.00, 1.06]	[Ref]	0.61 [0.60–0.62]
TT (n=32)	0.75 [0.47,1.03]1	1.80 [1.04,3.12]2	1.22 [0.79–1.66]3

RBC, red blood cell; TT, testosterone therapy;

CI=confidence interval.

¹p=0.048.

²p=0.035.

³p=0.006

DISCUSSION

This study continued our evaluations of genetic, biologic, and lifestyle factors among blood donors that impact RBC function and survival in cold storage and after transfusion.^{21,26,29–32} TT represents an intriguing case, as the intraindividual fluctuations in plasma testosterone provided insights into androgen-mediated changes in RBC biology, with possible clinical implications on transfusion effectiveness. The key findings from our study suggest that TT is coupled with increased RBC oxidative stress, changes in membrane properties and elasticity, altered iron metabolism, and lower transfusion effectiveness. In our view, these observations require further evaluation of the clinical utilization of RBC products from donors on TT.

Testosterone therapy to raise serum testosterone concentrations into the normal cis-man range increases RBC count, hemoglobin, and hematocrit. One mechanism is testosterone-mediated erythropoiesis via hepcidin inhibition and altered iron transportation.³³ Studies on blood donors from REDS-III’s RBC-Omics program have associated TT with decreased ferritin and increased risk of iron deficiency.³⁴ We observed significant changes in iron indices including increased plasma transferrin and TIBC, and decreased ferritin, which could be perceived as symptoms of iron deficiency. However, these changes may reflect the aforementioned effects of TT on iron stores, and one might mistakenly diagnose iron deficiency in individuals who have started TT within the past year. Thus, iron deficiency studies should be interpreted cautiously in people who have recently started TT.

Another noteworthy observation is the significant decrease (~30 %) in serum SHBG shortly after TT initiation and across all follow up visits. SHBG is the major carrier of testosterone in plasma, and mutations in the gene SHBG may result in low serum testosterone.^35,36 The clinical implications of TT-related decline in serum SHBG are not clear; however, low SHBG is associated with various health conditions including obesity and type 2 diabetes.³⁷

The observation of increased RBC susceptibility to oxidative hemolysis after initiation of TT supports our previous study, which demonstrated how manipulations of testosterone signaling in mice modulated hemolysis. Specifically, castration was associated with decreased oxidative and osmotic hemolysis, and with increased RBC transfusion survival compared with transfused RBCs from testosterone-naïve males, whereas testosterone repletion in those mice was associated with increased RBC oxidative hemolysis.¹⁰ The mechanism by which TT promotes RBC oxidative stress is not clear and may be concentration dependent. Supraphysiological plasma testosterone has been associated with increased oxidative stress and interference with nitric oxide synthesis in the vasculature in a mechanism that could induce endothelial dysfunction.³⁸ Our findings of positive association between plasma free testosterone and oxidative hemolysis further suggest that abnormally high concentrations of plasma testosterone may promote RBC dysfunction and increase the risk of hemolysis-driven vascular injury. In the context of blood transfusion, we reported a high prevalence of supraphysiological testosterone (~30%) among TT donors⁵, which may be caused by testosterone overdosing or misuse.¹

Stored RBCs from TT donors were characterized by higher concentrations of extracellular potassium, lactate, RDW, and percent oxygen saturation values (% sO2) compared with matched controls. Potassium leak is a marker of the RBC storage lesion, which was associated with membrane vesiculation, membrane phosphatidylserine exposure, and oxidative injury.³⁹ Increased lactate concentrations were observed in TT RBCs stored under different additive solutions (Fig. 3 and Fig. S1), and in whole blood after initiation of TT (Table 2). Lactate is the end product of glycolysis suggesting that TT is associated with changes in cellular metabolism, as we reported before.¹⁵ Higher oxygen saturation values in RBC units have been associated with increased RBC susceptibility to oxidative stress in storage.⁴⁰ Interestingly, we found a negative association between TT and methemoglobin. Reduction of methemoglobin (heme iron in the ferric state) to hemoglobin (ferrous state) is catalyzed in RBCs by cytochrome b5 reductase.⁴¹ Oxidative stress may increase this enzyme’s activity, and may explain the decreased methemoglobin levels observed in RBCs from patients on TT. Taken together, TT was associated with increased expression of RBC stress markers in storage that could explain the observed differences in transfusion effectiveness.

To our knowledge, RBC products from TT donors have not been associated with transfusion reactions or adverse events. Our preclinical evaluations in mice suggested that gamma irradiation had a negative effect on PTR in stored RBCs from TT donors. Gamma irradiation produces changes in RBC susceptibility to oxidative stress and membrane deformability.⁴² The severity of gamma irradiated-induced cellular injury may be higher in donors with underlying conditions that favor RBCs oxidative stress. For example, gamma irradiation had an additive and negative effect on RBC transfusion effectiveness in blood donors who smoke.⁴³ TT may represent a similar situation given the effects of this therapy on RBC susceptibility to oxidative hemolysis. Of note, the highest values of storage hemolysis were recorded in gamma-irradiated TT RBCs (Fig. S1). Additional evidence for TT effects on RBC transfusion outcomes came from our retrospective analyses of NHLBI’s REDS-III databases, which demonstrated lower transfusion effectiveness and increased odds ratio of issuing additional RBC transfusions following a single RBC unit transfusion from a TT donor. The effect size on RBC transfusion effectiveness (Δ hemoglobin of 0.75 vs 1.03, TT and matched controls, respectively) is comparable or larger than known conditions (e.g., G6PD deficiency, 0.75), and newly identified single nucleotide polymorphisms in the genes SEC14L4, and MYO9B (0.83 and 0.89, respectively).²⁶

This study had several limitations. First, our interpretation of the data and conclusions are based on a small number of individuals on TT. Second, only a quarter of TT donors in our study had supraphysiological plasma testosterone that limited our ability to conduct correlation analyses with our measured outcomes. Third, the rapid clearance of transfused human RBCs in the mouse circulation limited our ability to detect donor differences beyond 24 hours, and there were some inconsistencies between our mouse and human data.

In conclusion, our cumulative findings identified differential responses of RBCs from individuals on TT to oxidative and shear (membrane deformability) stressors, cold storage-induced injury, and transfusion into mice after storage and gamma irradiation. Further, we found RBC transfusion effectiveness of TT RBCs is about 25 % lower than reference controls. Although additional studies are required, our findings have implications for precision transfusion medicine regarding the clinical utilization of blood products from TT donors. Given their lower transfusion effectiveness, RBC concentrates from TT donors might be more suitable in scenarios such as trauma and acute hemorrhage, and less effective in chronically transfused patients or patients with suppressed erythropoiesis, in which longer survival of transfused RBCs is desirable.

Supplementary Material

SUPINFO 2

Supplemental Figure S1. Effect of TT and RBC gamma irradiation on RBC characteristics and indices of hemolysis in cold storage. One week old leukocyte-reduced RBC units collected from 10 donors on testosterone replacement therapy (TT) and 9 controls were split into two mini bags after one week of storage. One bag was gamma irradiated and the other was not (naïve RBCs). Naïve and gamma irradiated RBCs were stored for 5 weeks and tested on weeks 2, 3, 4, and 6. A. Percent storage hemolysis. B. Lactate concentration (mmol/L). C. Extracellular potassium (K⁺) concentration (mmol/L). D. Percent oxygen saturation (sO₂). E. Percent methemoglobin. F. Percent oxidative hemolysis. Mean±SD. Asterisk symbols (*) represent significant differences between TT and controls in non-irradiated (naïve) RBC bags at specific time points. Dollar symbols ($) represent significant differences between TT and controls in gamma-irradiated RBC bags at specific time points. * or $, p < 0.05. ** or $ $, p < 0.01. *** or $ $ $, p < 0.001. p values obtained from Prism’s multivariable mixed-effects model with Fisher’s LSD test.