MeMAGEN: A Phase IIa/IIb open‐label trial of memantine testing safety and tolerability in sickle cell patients

ABSTRACT

Administration of memantine, an antagonist of the N‐methyl‐ d‐aspartate receptor, prevents Ca²⁺ overload and dehydration of red blood cells (RBCs) in patients with sickle cell disease (SCD). The objectives of the 1‐year dose‐escalation Phase IIa/IIb Memantine trial (MeMAGEN – NCT 03247218) with 17 SCD patients who were under stable hydroxycarbamide therapy were to test the drug's safety and tolerability. Daily memantine doses ranged from 5 to 15 mg for children/adolescents and from 5 to 20 mg for adults. Clinical and laboratory analysis showed that memantine was well tolerated. In children, a decrease in days spent in the hospital was observed. Safety was confirmed by laboratory tests, which were not, or were only minimally, altered during memantine therapy. In a subgroup of six patients whose RBCs presented with elevated K⁺ leakage before treatment, memantine therapy at its lowest dosage reduced this K⁺ loss and increased hemoglobin concentration. This study shows that memantine is safe and well tolerated by SCD patients, including children. Memantine has the potential to become a supportive and low‐cost therapy in conjunction with hydroxycarbamide.

INTRODUCTION

Sickle cell disease (SCD) is the most common monogenic hereditary disease and is the main cause of hemolytic anemia worldwide, with more than 300,000 new patients every year. ¹ The highest prevalence of SCD patients is reported in sub‐Saharan Africa and India. Migration flow carried SCD into the Middle East and Mediterranean region ² , ³ , ⁴ , ⁵ and later into the Americas and Europe. ¹ , ⁶

A homozygous point mutation in the β‐globin gene, resulting in the replacement of glutamic acid by valine at the sixth position in the β‐globin chain, gives rise to the sickle hemoglobin (HbS) variant. During sickling, red blood cell (RBC) membrane damage occurs mechanically due to repeating cycles of aggregation/disaggregation of deoxygenated HbS that is further promoted by dehydration (reviewed in Ref. ⁷ ). Extensive loss of K⁺ and cytosolic water due to the activation of Gardos channels ⁸ is induced by excessive Ca²⁺ uptake ⁷ despite sequestration of some of the Ca²⁺ in the inside‐out vesicles. ⁹ Furthermore, Ca²⁺ overload also enhances production of free radicals by NADPH oxidases, ¹⁰ resulting in persisting oxidative stress and RBC membrane damage in SCD patients (reviewed in Ref. ⁷ ). Oxidized HbS forms Heinz bodies, generating aggregates at the membrane, thereby delaying membrane maturation and promoting clusterization of Band 3 protein. Hyperphosphorylated Band 3 proteins form clusters harboring some of the glycophorin and ankyrin molecules that easily detach from the spectrin cytoskeleton, compromising red cell membrane integrity, ¹¹ and attract naturally occurring autologous IgGs. ⁹ Loss of transmembrane phospholipid asymmetry and externalization of phosphatidyl serine is observed in a fraction of RBCs of SCD patients. ⁷ , ¹¹

Mechanical damage to the RBC membrane, dehydration due to Ca²⁺ overload, and the following extensive leakage of K⁺ promote premature hemolysis and RBC clearance. Furthermore, RBCs of SCD patients show advanced adhesion to the vasculature, presumably due to chronic inflammation. ¹² , ¹³ , ¹⁴ The clinical features of SCD include chronic hemolytic anemia and recurrent acute events, such as vaso‐occlusive crises (VOCs), acute chest syndrome, hyperhemolytic and aplastic crises, hepatic, splenic, and central nervous system insults, and infections and chronic complications. Compound heterozygosity for HbS and other β‐globin abnormalities, such as β‐thalassemia, causes similar clinical manifestations. Hemolysis, VOC, and other manifestations of this disease ¹ , ¹⁵ , ¹⁶ result in significant patient suffering and pose a marked global socioeconomic burden, especially in Africa, where its prevalence reaches ~10.7 per 1000 persons. ¹⁷ , ¹⁸ Life expectancy of SCD patients is reduced by more than 20 years compared to healthy controls. ¹⁹ Most regrettably, the average mortality of children under 5 years of age in African countries reaches at least 36%, as reported in 2022. ²⁰

Stem cell transplantation and gene therapy are the only curative treatments for SCD patients. However, donors for the allogeneic stem cell transplantation are rarely available, and the procedures are expensive and sometimes associated with side effects. ²¹ , ²² Gene therapies that use autologous stem cells may overcome most of the complications associated with SCD, ²³ , ²⁴ , ²⁵ but the high therapeutical cost poses a serious limitation. ²⁶ Therefore, treatment with hydroxycarbamide, which aims to increase fetal hemoglobin (HbF) levels in RBCs and improve cellular deformability and longevity, remains the most effective treatment so far. ²⁷ , ²⁸ , ²⁹ Unfortunately, most of the “next‐generation” drugs to treat SCD symptoms, among them rivipansel (GMI1070) and crizanlizumab (SelG1), both targeting cell adhesion, canakinumab and l‐glutamine, which reduce inflammation, and voxelotor (GBT440) that inhibits HbS polymerization, ³⁰ , ³¹ did not reach the expectations. While rivipansel was safe and well tolerated in SCD patients hospitalized for VOC, application of the drug did not meet primary or secondary endpoints. ³² Moreover, the European Medicines Agency (EMA) has recommended revoking the marketing authorization for crizanlizumab ³³ and voxelotor, ³⁴ and also the company decided to stop all trials and to withdraw voxelotor due to ineffectiveness and even increased risk of death. ³⁴ , ³⁵ On the other hand, symptomatic care of SCD includes blood transfusions, iron chelation, vaccinations, antibiotic/analgesic treatment, and psychosocial support. ³⁶ , ³⁷ , ³⁸ Considering the socioeconomic situation of countries in which SCD is common, ²⁶ the development of a well‐tolerated, effective, life‐long, and inexpensive treatment to reduce the symptoms and organ damage of SCD is of major significance.

The safe and low‐cost drug memantine, an N‐methyl‐d‐aspartate receptor (NMDAR) antagonist, ³⁹ is commonly used to mitigate Alzheimer's disease. ⁴⁰ , ⁴¹ We showed that NMDAR is present in RBC and that the number of NMDAR copies per RBC is substantially increased in circulating RBCs originating from SCD patients, ⁴² leading to facilitated Ca²⁺ uptake. The resulting Ca²⁺ overload then causes activation of Gardos channels and leads to uncompensated K⁺ loss from the cells, their dehydration, an increase in corpuscular HbS concentration, and formation of HbS aggregates in RBCs of patients with the HbSS phenotype. ⁴² Our in vitro experiments revealed that memantine‐induced inhibition of Ca²⁺ uptake via NMDARs allows RBC rehydration and largely prevents the hypoxia‐induced sickling of RBCs. ⁴²

In a recent small pilot Phase IIa/IIb clinical trial (MemSID, NCT02615847), the safety and tolerability of memantine were tested in four adult SCD patients. This 1‐year study suggested that memantine administration is safe for adults and may cause rehydration of RBCs and a quantitative reduction in the probability of their terminal sickling. ⁴³ , ⁴⁴ Here, we describe the outcome of a Phase IIa/IIb trial of memantine treatment in a cohort of SCD patients. The primary objective of this 1‐year‐long dose‐escalation clinical study was to test the drug's safety and tolerability in children, adolescents, and adults. During the study, we examined the effects of escalating memantine doses on the clinical severity of SCD manifestation and on blood parameters.

METHODS

Study registration

The MeMAGEN study was registered as a Phase IIa/IIb, open‐label, single‐center trial to study the safety and tolerability of memantine as supportive long‐term treatment in symptomatic SCD. The study received approval from the local Helsinki committee (EMC‐0071‐17) and was registered by the Israel Ministry of Health (MoH 001900) and the clinical trials registry (NCT03247218). Informed consent was obtained from all patients or their parents.

Recruitment strategy

Key inclusion criteria for participation in the MeMAGEN study were (i) confirmed diagnosis of SCD of any genotype, (ii) age 10 years or older, and (iii) agreement to comply with the study protocol (signed informed consent). Adult females agreed to use two contraception methods during and 6 months poststudy. Exclusion criteria included history of regular blood transfusions or transfusion within the last 3 months before screening. Excluded were patients with active infection(s) requiring systemic treatment, those prediagnosed with HIV or chronic active hepatitis (HCV or HBV). Patients with moderate to severe renal or liver dysfunctions, as well as pregnant or breastfeeding women or those receiving any investigational product within 30 days prior to this trial, were not eligible. Out of the 45 SCD patients under treatment at the Pediatric Hematology Unit, Emek Medical Center (EMC, Afula, Israel), 22 were not enrolled; the rest were screened, and 17 were included (Figure 1A). Demographic data are shown in Figure 1B. Note that all subjects were taking stable doses of hydroxycarbamide for at least 3 months before screening (mean daily dosage: 20.9 ± 7.7 mg/kg; range: 500–2000 mg/day). Of note, most patients enrolled in the present MeMAGEN study were also tested for memantine's effect on cognitive function and neural processing speed as described elsewhere. ⁴⁵

Figure 1.

MeMAGEN study participants: flow chart, demographic data, treatment protocol, and memantine plasma concentration. (A) Flow chart of sickle cell patients' enrollment in the study. (B) Demographic data and brief clinical characteristics of all screened patients. Sβ(0), sickle cell β0 thalassemia; Sβ(+), sickle cell β+ thalassemia; SCD, sickle cell disease; Splx, splenectomy; SS, sickle cell homozygous. Note that data from patient #1001 were excluded from the overall calculations, as this patient did not intake memantine (as confirmed by minimal blood drug concentration, see methods). (C) Memantine treatment protocol. (D) Memantine concentration in plasma.

Study design

The dose‐escalation protocol (Figure 1C) included two screening visits: one a month before the first memantine administration and the other on Day 1 of the study. On Day 1, all patients started on a daily memantine dose of 5 mg, independent of body weight or body surface. During the trial, participants had monthly check‐ups, but outside the trial, they were seen only every three months or less. During the visits, complaints were recorded, followed by a physical examination, drug accountability assessment, and blood and urine collection. The monthly provisions of the study drug were renewed at each visit. A memantine dose‐escalation step of 5 mg/day was performed every 3 months, ending with a maximal dose of 15 mg/day for children and adolescents. Adult patients reached the end of the escalation protocol at a memantine dose of 20 mg/day starting at Month 9 of the study. Compliance was controlled at the end of the study by measuring memantine levels in plasma samples collected at each visit during the study (Figure 1D). Lack of compliance was detected for patient 1001, and this patient's data were excluded from the final analysis (Figure 1B).

Safety and tolerability

Safety was assessed as the frequency of outcomes of adverse events not related to SCD or pathological laboratory parameters not associated with SCD.

Clinical records

Basic demographic data collected at screening (Figure 1B) included age, gender, body weight, SCD diagnosis, and hydroxyurea intake. Cardiac, liver, and spleen status, previous central nervous system episodes, and avascular hip necrosis were recorded for the 2‐year period before and the 1‐year period of the study. SCD‐related clinical history was collected for the 2 years prior to and the year during the study onset and included emergency room visits, hospital admissions, days in hospital, blood transfusions, infections, VOC, and “acute pain.” The reported “acute pain” episodes did not require medical facility visits and were recorded at follow‐up visits as a subjective perception. All collected data were used as individual baselines for each participant and compared to their records during the 12 months of the memantine treatment. Vital signs were measured at screening and at all of the trial's routine visits. Spleen and liver size were also recorded at the monthly visits.

Laboratory testing

Apart from standard routine analyses, RBC exploratory parameters, namely K⁺ leakage, fractionation on Percoll density gradient, detection of the number of cells exposing phosphatidylserine to the outer membrane leaflet, intracellular bulk reduced thiols and free Ca²⁺ levels, and forward and side scatter as a marker of morpho‐rheological changes, were conducted as described previously, ⁴⁶ and intracellular Na⁺ and K⁺ contents were measured as described earlier. ⁴⁷

Blood serum memantine levels and neurocognitive tests

Plasma memantine concentration was determined using a combination of high‐performance liquid chromatography (HPLC) and tandem mass spectrometry (MS/MS) as described. ⁴⁸

Cognitive performance and neural processing were analyzed as part of the screening tests and after 1 year of memantine treatment. A detailed description of the neurocognitive tests is presented separately in a very recently published manuscript. ⁴⁵

Statistics

We summarized and compared the averaged effect of each 6‐month period of treatments: 5 and 10 mg/day (5/10 period) and 15 and 20 mg/day (15/20 period) to the pretreatment data. For children and adolescents, the 6‐month maximal treatment of 15 mg/day was considered the second part of the 1‐year study. Statistical analysis was conducted in R (version 4.3.2, R Foundation for Statistical Computing, Vienna, Austria). Crude incidence rates of events for demographic groups (children, female, and male) at memantine doses of 5/10 and 15/20 mg/day (children only 15 mg) were computed by dividing the number of events or hospitalization days by the number of person‐years at risk (events/1000py). P‐values were computed by comparing the differences in crude incidence rates between treatment and pretreatment periods. To get 95% confidence intervals (95% CI) for the incidence rates, function epi.conf from R‐library epiR was used. To assess the effect of memantine on laboratory parameters, between‐group differences were analyzed using repeated‐measures analysis of variance (ANOVA) or, when assumptions were not met, the Friedman test. Post hoc analyses were performed with Tukey‐adjusted t‐tests (parametric) or Wilcoxon signed‐rank tests with Bonferroni correction (nonparametric). A P‐value < 0.05 was considered statistically significant.

RESULTS

Study cohort information and compliance

Of a total of nine females and eight males aged 23.9 ± 10.1 years, nine had the HbSS genotype, five had the HbS/β0 genotype, and three had the HbS/β+ genotype. Age, gender, body weight, genotype of the β‐globin mutations, spleen status, hydroxycarbamide treatment, and other relevant complications are shown in Figure 1B. Living far from the Pediatric Hematology Unit, and due to the COVID‐19 pandemic, precluding the families from coming to the hospital (see Figure 2), they were followed up by regular phone calls. During the pandemic, memantine was regularly home‐delivered by the hospital's delivery service. As expected, daily memantine intake was associated with a gradual increase in plasma drug levels as the trial progressed (Figure 1D).

Figure 2.

Clinical events before and during memantine treatment. Green lines: pretreatment period of −108 to 0 weeks; blue lines: treatment period of 0–60 weeks. Narrow blue lines represent 5–10 mg/day memantine for all participants, and wide blue lines represent 15 mg/day memantine in children or 15–20 mg/day memantine in adults. SCD, sickle cell disease.

Safety and tolerability

Overall, we observed no adverse events resulting from memantine administration that could not be related to or explained by SCD (Figure 2; Table 1). Occurrence of the most harmful events—VOC, pain, and blood transfusion—was reported at each monthly visit. The frequencies of these events were compared to those in the 2‐year pretreatment period (Table 1). Two patients complained of mild headaches during the memantine treatment, but neither interrupted the treatment, and they reported amelioration of headaches after a few weeks. The same patients also complained of temporary constipation. None of the study participants ceased treatment due to drug‐related side effects. After completion of the 1‐year trial, 7 out of 17 patients voluntarily decided to continue memantine intake. For the others, the main reason not to continue with the drug intake after the trial was intended pregnancy.

Table 1.

Clinical events in treatment groups.

		Memantine dose 0 mg/day					Memantine dose 5/10 mg/day					Memantine dose 15/20 mg/day
Patient	N Patients	N events	Person‐years (py)	Events per 1000 py	95% CI	P‐value	N events	Person‐years (py)	Events per 1000 py	95% CI	P‐value	N events	Person‐years (py)	Events per 1000 py	95% CI	P‐valuea
Hospitalization
Total	16	14	32.0	438	(240–735)	‐	1	7.9	127	(3–706)	0.071	5	9.8	512	(166–1194)	0.774, 0.374
Child	7	8	14.0	572	(247–1127)	‐	1	3.5	288	(7–1606)	0.420	2	4.7	425	(52–1537)	0.686, 0.766
Female	5	2	10.0	200	(24–723)	‐	0	2.5	0	(0–1480)	0.157	1	2.9	340	(9–1895)	0.704, 0.317
Male	4	4	8.0	501	(136–1282)	‐	0	1.9	0	(0–1907)	0.046	2	2.1	939	(114–3393)	0.537, 0.157
Days in hospital
Total	16	48	32.0	1502	(1108–1992)	‐	5	7.9	380	(78–1110)	<0.001	15	9.8	1535	(859–2532)	0.941, 0.064
Child	7	25	14.0	1788	(1157–2640)	‐	5	3.5	865	(178–2527)	0.133	2	4.7	425	(52–1537)	0.004, 0.153
Female	5	8	10.0	801	(346–1578)	‐	0	2.5	0	(0–1480)	0.005	5	2.9	1701	(552–3969)	0.268, 0.025
Male	4	15	8.0	1878	(1051–3097)	‐	0	1.9	0	(0–1907)	0	8	2.1	3757	(1622–7404)	0.184, 0.005
Emergency room visit
Total	16	4	32.0	125	(34–320)	‐	0	7.9	0	(0–467)	0.046	1	9.8	102	(3–570)	0.849, 0.317
Child	7	0	14.0	0	(0–264)	‐	0	3.5	0	(0–1063)	‐	0	4.7	0	(0–785)	‐, ‐
Female	5	1	10.0	100	(3–558)	‐	0	2.5	0	(0–1480)	0.317	1	2.9	340	(9–1895)	0.498, 0.317
Male	4	3	8.0	376	(77–1097)	‐	0	1.9	0	(0–1907)	0.083	0	2.1	0	(0–1733)	0.083, –
Vaso‐occlusive crisis
Total	16	11	32.0	344	(172–616)	‐	3	7.9	380	(78–1110)	0.883	5	9.8	512	(166–1194)	0.505, 0.677
Child	7	5	14.0	358	(116–835)	‐	0	3.5	0	(0–1063)	0.025	1	4.7	213	(5–1185)	0.586, 0.317
Female	5	1	10.0	100	(3–558)	‐	1	2.5	401	(10–2236)	0.467	3	2.9	1021	(210–2983)	0.124, 0.385
Male	4	5	8.0	626	(203–1461)	‐	2	1.9	1034	(125–3735)	0.602	1	2.1	470	(12–2617)	0.775, 0.516
Acute chest syndrome
Total	16	1	32.0	31	(1–174)	‐	0	7.9	0	(0–467)	0.317	0	9.8	0	(0–378)	0.317, ‐
Child	7	0	14.0	0	(0–264)	‐	0	3.5	0	(0–1063)	‐	0	4.7	0	(0–785)	‐, ‐
Female	5	0	10.0	0	(0–369)	‐	0	2.5	0	(0–1480)	‐	0	2.9	0	(0–1255)	‐, ‐
Male	4	1	8.0	125	(3–697)	‐	0	1.9	0	(0–1907)	0.317	0	2.1	0	(0–1733)	0.317, ‐
Blood transfusion
Total	16	6	32.0	188	(69–409)	‐	1	7.9	127	(3–706)	0.680	2	9.8	205	(25–740)	0.918, 0.833
Child	7	4	14.0	286	(78–733)	‐	1	3.5	288	(7–1606)	0.995	2	4.7	425	(52–1537)	0.676, 0.766
Female	5	1	10.0	100	(3–558)	‐	0	2.5	0	(0–1480)	0.317	0	2.9	0	(0–1255)	0.317, ‐
Male	4	1	8.0	125	(3–697)	‐	0	1.9	0	(0–1907)	0.317	0	2.1	0	(0–1733)	0.317, ‐
Pain
Total	16	14	32.0	438	(240–735)	‐	17	7.9	2153	(1254–3447)	0.001	15	9.8	1535	(859–2532)	0.008, 0.346
Child	7	5	14.0	358	(116–835)	‐	7	3.5	2017	(811–4157)	0.033	1	4.7	213	(5–1185)	0.586, 0.023
Female	5	1	10.0	100	(3–558)	‐	6	2.5	2408	(884–5241)	0.020	5	2.9	1701	(552–3969)	0.037, 0.569
Male	4	8	8.0	1001	(432–1973)	‐	4	1.9	2068	(564–5295)	0.329	9	2.1	4227	(1933–8024)	0.026, 0.217

Note: 0, screening or pretreatment period; 5/10, 5–10 mg memantine dose/day; 15/20, 15–20 mg memantine dose/day. Vaso‐occlusive crises (VOCs) are defined as documented events that required hospital admission, while pain refers to events reported either by phone call or retrospectively at the monthly visits. Data are presented as incidence rates (median of the cases per 1000 py and 95% CI) for the whole group (total) and subdivided for children, females, and males. Significant values are marked in bold.

^a First P‐value compares dose 15/20 mg/day with 0 mg/day, while the second P‐value compares dose 15/20 mg/day with 5/10 mg/day.

Clinical observations

When comparing the pretreatment with the treatment period, we observed a tendency in children that the VOC frequency decreased from 5 (14 py) to 1 (8.173 py, P = 0.243) (Table 1). The number of hospitalization days during the pretreatment period with 1502 events/1000 py (95% CI: 1108–1992) was higher compared to the memantine 5/10 mg/day period with 380 events/1000 py (95% CI: 78–1110; P < 0.001), whereas dose escalation from 5/10 to 15/20 mg/day did not result in further improvement. In children, moreover, a continuous reduction in the number of days in hospital was observed from 1788 events/1000 py (95% CI: 1157–2640) to 865 events/1000 py (95% CI: 178–2527; P = 0.133) and to 425 events/1000 py (95% CI: 52–1537; P = 0.004) for pretreatment, 5/10 mg/day, and 15 mg/day memantine, respectively. We have recorded the acute pain episodes reported by the patients during the inspection visits (Figure 2). At the 5/10 mg dose, the incidence rate was 2153 events/1000 py (95% CI: 1254–3447), which was significantly higher than the 438 events/1000 py reported in the control group (95% CI: 240–735; P = 0.001). The 15/20 mg group also had a higher rate of 1535 evens/1000 py (95% CI: 859–2532; P = 0.008). However, the difference between the two dose groups was not statistically significant (P = 0.346). The events such as emergency room visits, acute chest syndromes, and blood transfusions were infrequent (the absolute events were as low as 5, 1, and 9 cases, respectively) over the entire pretreatment and treatment periods so that no conclusions can be drawn.

Laboratory parameters

Memantine administration was not associated with renal, hepatic, or coagulation test abnormalities, hemolytic activity, or suppression of erythropoiesis (Figures 3 and 4; Table S1).

Figure 3.

Effect of daily doses of memantine on inflammation‐related blood markers. (A) Hemoglobin (Hb). (B) Fetal hemoglobin (HbF). (C) Reticulocytes. (D) White blood cells (WBC). (E) Ferritin. (F) C‐reactive protein (CRP). Data for the whole cohort are presented as median ± CI. A two‐tailed analysis of variance (ANOVA) test was performed as described in the Methods section, and P < 0.05 was considered statistically significant. 0, screening or pretreatment studies; 5/10, 5–10 mg/day memantine dose; 15/20, 15–20 mg/day memantine dose.

Figure 4.

Impact of daily memantine on markers related to red blood cell (RBC) heterogeneity, morphology, and membrane permeability. (A) Hyperchromic RBC. (B) Side scatter. (C) Redistribution of RBCs in Percoll density gradient representative for a child and two adult patients on treatment with the following memantine daily doses: 0, 10, and 15 mg/day for the child and 0, 10, and 20 mg/day for adults. (D) Analysis of the localization of the main fraction within the Percoll gradient. The distance between the zero (0, solid line) point and the top (dotted line) and bottom (dashed line) borders of the main fraction of RBCs was measured using Image J software. Results of numeric analysis are presented as “distance to the top” and “distance to the bottom” panels. (E) Intracellular K⁺. (F) K⁺ loss. Data for the whole cohort are presented as median ± CI. 0, screening or pretreatment studies; 5/10, 5–10 mg/day memantine dose; 15/20, 15–20 mg/day memantine dose. AU, arbitrary units.

RBC turnover markers, bone marrow, and inflammatory effects

Memantine treatment did not affect Hb or erythrocyte counts (Figure 3; Table S1), or reticulocyte counts. A decrease in ferritin levels was observed: the changes in ferritin concentrations showed a strong positive association with blood iron and a negative association with serum transferrin. No correlations between ferritin and inflammatory markers were detected. While no significant increase in HbF was detected in response to memantine therapy, a small but significant reduction in HbA2 was observed (Table S1), suggesting responses of the erythroid precursors to the therapy.

RBC hydration and K⁺ leakage

Memantine administration was accompanied by a decrease in the hyperchromic RBCs (Figure 4A) along with the changes in side scatter (Figure 4B). Treatment with memantine resulted in a dose‐dependent decrease in density and/or aggregability ⁴⁹ of the majority of RBC, as follows from the upward shift of the main fraction of cells (between the top and bottom labels in Figure 4C,D) in the Percoll density gradient. A significant increase in mean intracellular RBC K⁺ content, with no alteration of intracellular Na⁺ levels, was detected in response to memantine treatment (Figure 4E; Table S1). Furthermore, memantine caused a near‐significant decrease in K⁺ loss (P = 0.061), which was more pronounced in patients with initially high K⁺ leakage (Figure 4F). Considering the strong heterogeneity in K⁺ permeability observed during screening, we stratified the patients into two groups that showed different responses in K⁺ leakage intensity to memantine therapy (Figure 4F, right; Table S2). Six study participants, initially showing high K⁺ leakage (above 5 µmole/g Hb*h), responded to memantine administration by suppression of this excessive K⁺ loss. The decline in K⁺ loss was already observed at the lower memantine doses (5/10 mg daily), and the higher doses did not further enhance this effect. The decline in K⁺ leakage in this cohort was associated with an increase in Hb concentration and a decrease in serum ferritin and iron levels during the treatment (Table S2). On the other hand, in the group of 10 patients initially presenting with low K⁺ leakage from their RBCs, memantine did not show any effect. Interestingly, the number of VOCs was reduced to zero at 5/10 mg memantine in patients who presented with initially low K⁺ leakage at screening and remained zero at 15/20 mg memantine. Interestingly, the number of hospitalizations and days in hospital in the group with high K⁺ leakage at screening decreased at 5/10 mg memantine but not at 15/20 mg (Table S3). Note that neither high nor low K⁺ leakage affected the incidence of other clinical events during memantine therapy (Table S3).

DISCUSSION

Our data indicate that memantine treatment is safe and well tolerated by children, adolescents, and adult SCD patients. This observation aligns with an earlier report on safe memantine administration to a small group of adult SCD patients. ⁴³ At present, memantine is a standard therapy for elderly patients suffering from Alzheimer's disease and should be avoided during pregnancy as it may impair brain development. ⁵⁰ In our study, as part of the safety assessment of memantine treatment in SCD patients, a large variety of laboratory parameters were tested to assess liver and kidney function, metabolic state, inflammatory state, hemolysis, and platelet function. All of the laboratory results during the 1‐year trial were in the normal range or in the same range as the screening values associated with the underlying disease (Table S1).

Clinical findings

The number of participants in the MeMAGEN trial was relatively low, and a placebo group was not included in the study, enabling us to draw only partial conclusions on the clinical efficacy of memantine therapy. Nevertheless, compared to their 2‐year medical history prior to trial enrollment, we observed a significant decrease in the number of days spent in hospital for children on memantine (Table 1; Figure 2). Interestingly, in adults, there was first a decrease in days spent in hospital at 5/10 mg/day memantine and at 15/20 mg/day memantine, an elevation to a similar level as in the pretreatment period.

During the present MeMAGEN trial, the patients remained on hydroxycarbamide therapy, which per se reduces VOC frequency and the number of hospitalization days. ²⁷ , ²⁸ Therefore, we propose that the observed improvement with memantine intake represents a complementary benefit. For example, we observed fewer VOCs in children during memantine therapy. Due to the low overall number of events, we abstain from making conclusions. Similar to our study, most of the SCD patients participating in the l‐glutamine and canakinumab trials were on continuous hydroxycarbamide treatment during both pretrial and trial periods. ⁵¹ Canakinumab administration also resulted in a reduction in the number of overall hospitalizations and hospitalization days compared to the pretreatment period and when compared with a placebo group. ⁵² Taken together, these observations strongly support the notion that a combination of drugs having different targets is more effective than monotherapy. This approach also provides flexibility, enabling the selection of the optimal therapy for an individual patient. However, the absolute efficacy of a given single compound is masked. In theory, Phase III trials on hydroxycarbamide‐naive patients could reveal memantine's efficacy, but such a trial goes against the current guidelines stating that SCD patients should start hydroxycarbamide treatment as early as possible, optimally from the age of 2 years on.

Along with the improvement in acute clinical SCD manifestations, we found that patients still reported pain events during the trial (Figure 2). In contrast to other SCD‐related symptoms that can be accurately documented, pain perception cannot be objectively verified by the clinical staff and shall not be interpreted as a stand‐alone symptom due to the complexity of causes of pain induction and its perception by patients. In SCD patients treated with other drugs such as the specific Gardos channel inhibitor senicapoc, no significant improvement in the rate of painful sickle cell crises had been reported. ⁵³ The absence of a significant reduction in pain sensation was also reported for canakinumab, ⁵² whereas crizanlizumab treatment reduced pain crises by 45%. ⁵⁴ Taken together, these observations, including ours, imply the urgent need for a standardized pain‐assessment strategy in a longer follow‐up trial harboring a larger number of participants that would enable establishing a possible association of drug administration and pain sensation.

Targeting SCD pathways and possible molecular mechanisms

New therapies target different SCD pathways. While hydroxycarbamide increases HbF, interferes with HbS aggregation, and ultimately prevents RBC sickling, the P‐selectin inhibitor crizanlizumab suppresses vascular adhesion of RBCs. ⁴ The inhibitor of Gardos channels, senicapoc, improves RBC hydration. l‐Glutamine and canakinumab decrease oxidative damage and inflammation. ⁵² , ⁵³ , ⁵⁵ Glycolytic activity induced by mitapivat was recently shown to improve rehydration of RBCs in SCD patients. ⁵⁶ , ⁵⁷ Differing from these drugs' mechanisms, memantine targets NMDARs in multiple tissues and, in RBC, ⁵⁸ targets several hematopoietic cell lines and RBC myeloid lineage cells and platelets. ⁵⁹

We previously reported that in RBCs originating from SCD patients, Ca²⁺ overload caused by hyperactivation of the NMDARs induces oxidation, dehydration, and proteolysis. ⁴² , ⁶⁰ Accordingly, we suggested that memantine generates a complex systemic response at the level of circulating blood cells and in the bone marrow. In our previous pilot clinical trial, MemSID, memantine increased HbF levels in a patient who was refractory to hydroxycarbamide treatment. ⁴³ , ⁴⁴ In the present study, the participants already presented with high intraerythrocytic HbF levels before memantine treatment, ⁶¹ and most of them did not show any further increase in HbF. Moreover, there were no evident signs of inflammation reduction.

In a previous study, a decrease in platelet count was reported in a mouse model with platelet lineage‐specific deletion of GRIN1. ⁶² The mouse line, which was deficient for the GRIN1 subunit of the NMDAR in megakaryocytes (Pf4‐Grin1^{−/ −}), showed lower platelet counts and prolonged bleeding time. In the present study, we did not see any significant reduction in platelet count or any abnormalities in platelet function in thromboelastogram analysis (data not shown).

During the former MemSID trial, a reduction in serum ferritin concentration during memantine treatment was observed in three out of four patients. ⁴³ Similarly, lower serum ferritin concentration was observed in the present study (Figure 3E). This effect was not associated with changes in inflammatory markers (Figure 3; Table S1). We hypothesize that the changes in ferritin level may be associated with suppression of iron turnover and/or hemolytic activity, but this hypothesis may only be tested in a larger cohort of patients during a future Phase III study. ⁶³ , ⁶⁴

The acute effect of inhibiting one of the Ca²⁺‐uptake pathways, as reported in the MemSID trial, was suppression of excessive K⁺ leakage, most likely mediated by hyperactivated Gardos channels. ⁶⁵ This hypothesis is supported by an increase in the intracellular K⁺ content (Figure 4E), a decrease in the density aggregability ⁶⁵ of the main fraction of RBCs (Figure 4C,D), and a reduction in hyperchromic RBCs (Figure 4A). These findings are consistent with the ones we have previously reported for the MemSID trial. ⁴⁴ Within the present study's cohort, we identified a group of patients with high K⁺ leakage from RBCs. Interestingly, this group responded to memantine treatment with a reduction in passive K⁺ leakage that was associated with an increase in mean corpuscular Hb and a decrease in RBC counts (Table S2). These observations suggest that in some patients, memantine inhibits K⁺ leakage via the Gardos channels, thereby interfering with Ca²⁺ overload and causing activation of these K⁺ channels in RBCs of SCD patients. Regarding the incidence of clinical events, patients with low pretreatment K⁺ leakage showed less VOC events under memantine treatment (Table S3). The differences in clinical responses between those groups warrant further analysis in larger studies.

We could not compare the results of our study with those using the Gardos channel blocker senicapoc ⁵³ , ⁶⁶ because memantine targets NMDARs in multiple tissues, including the brain, bones, and kidneys. ⁵⁸ , ⁵⁹ Considering that memantine has been used to treat patients with dementia and neuronal damage for decades, ⁶⁷ we expect that long‐term administration of this drug to children will also reduce SCD‐associated brain damage. As such, our very recently published results describing pilot cognitive tests, which were part of the present study, support this notion. ⁴⁵ Briefly, notable improvements in processing speed, working memory, and attention suggested memantine's potential to alleviate cognitive deficits associated with SCD. Enhanced executive functions and reduced cognitive load during task‐switching, evidenced by lower error rates and changes in brain event‐related potentials, suggest more efficient neural processing in tasks requiring neurocognitive flexibility. These improvements may contribute to enhanced daily functioning, as well as overall quality of life.

In summary, knowing that memantine is safe and well tolerated, reduces hospitalization days, and improves RBC rehydration, we anticipate that this drug should be tested for clinical efficacy in a larger Phase III, double‐blinded, placebo‐controlled clinical trial, ideally encompassing both treatment‐naïve patients and those already receiving hydroxycarbamide in an affected low‐income country. If found effective, memantine may improve the condition of patients who are proven refractory to other therapies or may be used in combination with other “small molecules” for optimal symptomatic treatment of patients. Given that memantine is a safe, well‐tolerated, off‐patent, inexpensive, and easily stored drug with convenient once‐daily administration, it represents a particularly promising option for patients in low‐income regions, where SCD remains highly prevalent and often insufficiently treated.

AUTHOR CONTRIBUTIONS

Ariel Koren: Conceptualization; investigation; writing—original draft; writing—review and editing; methodology; formal analysis; data curation; supervision. Carina Levin: Data curation; conceptualization; investigation; writing—original draft; writing—review and editing; methodology; formal analysis. Leonid Livshits: Investigation; writing—original draft; writing—review and editing; formal analysis. Fabio Valeri: Writing—original draft; formal analysis; writing—review and editing; validation. Sari Peretz: Writing—review and editing; data curation. Sivan Raz: Data curation; investigation. Anna Yu Bogdanova: Conceptualization; methodology; writing—review and editing; writing—original draft; supervision. Max Gassmann: Conceptualization; funding acquisition; writing—original draft; writing—review and editing; visualization; methodology; supervision; investigation; validation; project administration.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

FUNDING

This work was supported by Ernst Göhner Stiftung, René und Susanne Braginsky Stiftung, Israeli Ministry for Development of the Negev and the Galilee, Baugarten Stiftung, and Fondation Botnar.

Supporting information

Supporting Information.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available in the Supporting Information of this article.