Multiple Thrombosis in a Patient with Gardos Channelopathy and a New KCNN4 Mutation

Dehydrated hereditary stomatocytosis (DHS) is an autosomal dominant hemolytic disease characterized by red blood cell dehydration due to increased membrane permeability to cations. This default in permeability leads to cell volume dysregulation and morphologic abnormalities. Note, DHS is characterized by heterogeneous phenotypic presentation, varying from hydrops foetalis to mild anemia or asymptomatic well-compensated hemolysis.1 Two subtypes have been molecularly defined, DHS1 due to PIEZO1 gain-of-function mutations (representing 90% of all cases) and DHS2 (also named Gardos channelopathy) due to KCNN4 gain-of-function mutations. Gardos channelopathy is a very rare subtype with less than a dozen families reported so far. So, KCNN4 encodes the Potassium Calcium-Activated Channel Subfamily N Member 4 in which gain of function mutations result in increased sensitivity to calcium influx, leading to KCl loss and cell dehydration.1-4 Previous reports highlighted differences between the two forms of DHS. Although iron overload can be observed in both entities, perinatal edema and post-splenectomy thrombosis have mostly been reported in DHS1 so far, while severe anemia appears to be a frequent manifestation of DHS2. Only a few KCNN4 mutations have been identified to date: V282M, V282E, S314P, R352H and V369_Lys373del, whereas more than 40 different mutations have been reported in PIEZO1. Here, we report on a patient with Gardos channelopathy syndrome, which harbored a novel KCNN4 mutation and suffered from several episodes of venous and arterial thrombosis after splenectomy. The proband is a 49-year-old man referred to our center for undiagnosed hemolytic anemia. He had no family history of anemia or hemolysis. Hemolysis was first diagnosed at 1 month of age, in a context of infectious disease. The patient had a severe anemia (Hb = 3.7 g/dl) requiring transfusion. He then suffered from several hemolytic crises during childhood, generally related to infectious episodes. An abdominal painful crisis at age 10 led to a splenectomy. After splenectomy, hemoglobin stabilized between 8 and 9 g/dl; hemolysis persisted with undetectable haptoglobin levels and high levels of plasma hemoglobin and free heme, and platelet count increased to 700 000–800 000/μl. At age 30, he was investigated for hyperferritinemia (3193 μg/L). Liver biopsy showed iron overload and portal and periportal fibrosis. The HFE gene sequencing was negative. Phlebotomy was then initiated. During the same year, he underwent cholecystectomy for gallstones, which was complicated by catheter-related thrombosis in the left arm. Fluindione (vitamin K antagonist used in Europe, same family as warfarin) was introduced for a period of 3 months. Shortly after the withdrawal of fluindione, he developed facial hemiparesia, revealing an ischemic stroke due to middle cerebral artery thrombosis. Extensive thrombosis investigations showed normal results (absence of FII G20210A and FV Leiden mutations; normal levels of antithrombin III, homocysteinemia, proteins S and C; absence of anticardiolipin antibodies) and there was no family history of thrombosis. Fluindione was reintroduced. Two months later, he was hospitalized for erysipelas complicated by left peroneal vein thrombosis. The INR at admission was 1.92. Despite treatment with Low Molecular Weight Heparin and increased doses of fluindione, thrombosis extended to the left femoral vein. Within 3 days, platelet count increased from 700 000/μl to 2 000 000/μl. Hydroxyurea was introduced but rapidly replaced by anagrelide because it was ineffective. Bone marrow study, performed while the patient was under hydroxyurea treatment, showed a rich marrow with marked signs of dyserythropoiesis, as already described in Gardos patients3: there were numerous binucleated erythroblasts, very rare intercytoplasmic bridges and internuclear bridges, but no evidence of dysgranulopoiesis or dysmegacaryopoiesis. Under anagrelide treatment, platelet count decreased to 800 000/μl. Since then, under a combined treatment with fluindione and anagrelide and despite the fact that he had several episodes of erysipelas during the last 19 years, he had no subsequent thrombotic event. Platelet count remained between 300 000 and 650 000/μl except during two episodes of acute inflammation due to knee trauma and erysipelas (platelet count >1 000 000/μl). At age 47, studies for myeloproliferative neoplasm were again negative (absence of CALR, MPL or JAK2 mutations). At 48, whole-body MRI showed ectopic thoracic (juxta-vertebral T9/T10) hematopoiesis. Red blood cell membrane phenotype investigation with ektacytometry at age 31 showed a profile suggestive of “atypical hereditary dehydrated stomatocytosis” (raw data not available). At age 48, blood smear evaluation showed macrocytosis, poikilocytosis with acanthocytes, target cells, abnormal hemoglobin repartition in RBC, rare spherocytes, rare Howell-Jolly bodies and basophilic stippling (Figure 1(A) control, Figure 1(B) patient's blood smear). Cell blood count and biochemical tests showed regenerative macrocytic anemia with high total and conjugated bilirubin, high LDH, undetectable haptoglobin, high plasma hemoglobin and high free heme, suggestive of a high level of hemolysis (Figure 1(C)). LORRCA (Mechanotronics) profile with osmoscan analysis (performed within 24 h after sampling and following manufacturer recommendations) showed dehydration, increased osmotic resistance and decreased red blood cells (RBC) deformability (Figure 1(D)) similar to those of KCNN4 R352H patients investigated in our lab. Next generation sequencing (NGS) on blood DNA was performed as previously described.5 Genetic analysis revealed a variant of uncertain significance (VUS) in the KCNN4 gene (NM_002250): c.965C > T (p.Ala322Val), absent from all genomic databases (including >125 000 subjects from the gnomAD database). The predicted amino acid change is located within the calmodulin binding site of the Gardos channel. Polyphen-2 and Mutation taster prediction tools (but not ALIGN GVGD or Sift) were suggestive of a deleterious effect. We did not find any other pathogenic or probably pathogenic variation in other genes involved in congenital hemolytic anemia. Functional studies were performed on fresh RBC from patients and controls. Water cell content was determined by calculating the weight difference between wet cells and dried cells. Ionic content was measured using a flame photometer (PFP7, Jenway). The amounts of Na + or K+ measured were reported as mmol/Liter cell water (mmol/Lcw).). Cell morphology analysis was performed on blood smears prepared from patient and control blood shipped within 24 h at 4°C, fixed with methanol and stained with freshly prepared Giemsa (10%, 15 min) (Sigma-Aldrich). Pictures were taken at ×100magnification and morphology analyses were performed using ImageJ software to determine cell circularity and roundness (n = 1301 cells for A322V patient and 1371 for control).6 Circularity was obtained with the formula: and roundness was obtained with the formula: . Results showed that despite a cell water content (1.76) and a cell content of Na + (33 mmol/Lcw) and K+ (106 mmol/Lcw) close to normal values, the patient's RBC had a significantly larger projected surface area than wild type red blood cells (52.4 ± 10.1 μm2 vs 35.4 ± 5.3 μm2, p < 0.001, unpaired t test). The circularity of the A322V cells was significantly altered (0.81+/−0.09 vs WT 0.84+/−0.08, p < 0.001, unpaired t test), confirming the poikilocytosis observed on blood smear. Because of the localization of the mutation within the calmodulin binding domain, the susceptibility of Gardos channel activity to calcium was tested by measuring membrane potential changes in the presence of CCCP (carbonylcyanide-m-chloro-phenyl-hydrazone)4 on fresh blood sampled on heparin and shipped at 4°C within 24 h. Channel activity was assessed using NS309 (100 μM, Alomone), a compound which lowers the threshold for Gardos channel activation by calcium, and thus allows to test calcium sensitivity of Gardos channel at nominal intracellular calcium concentration. In WT RBC, NS309 at a concentration of 100 μM only induced a slight hyperpolarization whereas in A322V and in R352H cells it resulted in a marked hyperpolarization (Figure 1(E)). The same experiments were also performed at a lower concentration of NS309 (50 μM) and showed similar results (data not shown). These results clearly indicate that, similarly to other mutations identified in the calmodulin binding domain, the A322V mutation results in a hypersensitivity of the Gardos channel to Ca2+ and likely allows more frequent transient activations of the channel, which may lead to cell dehydration. This patient harbors a novel KCNN4 variation (p.A322V) identified by NGS and classified as pathogenic thanks to functional studies. To date, only a limited number of Gardos patients have been reported, and no thrombotic events have yet been reported in splenectomized patients.1 This is quite surprising, since KCNN4 mutations result in intravascular hemolysis, which is not improved by splenectomy. The persistence of high levels of hemolysis after splenectomy is a known risk factor for thrombotic event.7, 8 Several PIEZO1-mutated patients developed thrombosis after splenectomy. Our observation shows that Gardos channelopathy also can be complicated by episodes of venous and arterial thrombosis. The exact role of thrombocytosis in the occurrence of thrombosis in our patient remains unclear, but it has to be noticed that platelet count increased at 10 years of age, just after splenectomy, whereas the first thrombotic event occurred 20 years later. We cannot completely exclude that the patient could have essential thrombocythemia (ET) in addition to Gardos channelopathy (bone marrow biopsy was not performed to definitively exclude ET), but this seems highly unlikely: platelet count increased at 10 years of age, 39 years ago, and all relevant tests performed to date (bone marrow study, NGS myeloid results) did not support a diagnosis of myeloproliferative neoplasm. Thrombocytosis investigation did not find any evidence for another cause for thrombosis in this patient. In the literature, platelet count is not correlated to the risk of thrombosis in patients with essential thrombocythemia or with post-splenectomy thrombocytosis.7, 8 Conversely, the involvement of the Gardos channel in platelet activation is supported by experimental data.9, 10 In our patient, the beneficial effect of anagrelide may be mediated by its intrinsic platelet-inhibiting properties and by the modulation of intracellular calcium concentrations,11 resulting in a decreased activity of the mutated Gardos channel. More observations are needed to accurately assess the risk of thrombotic events in KCNN4-related channelopathies. However, precautions (including careful follow-up with extensive platelet and coagulation testing) already seem necessary in case of splenectomy in Gardos patients. We thank Mrs Emmanuelle Faubert and Mrs Christine Gameiro for the realization of NGS and Mr Guillaume Gricourt for bioinformatic analysis. The authors declare no conflict of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request.