Microvesicles in haemoglobinopathies offer insights into mechanisms of hypercoagulability, haemolysis and the effects of therapy


Maxwell Westerman, Department of Medicine, Mt Sinai Hospital, 15th St and California Ave, Chicago, IL, 60608, USA. E-mail: wesm@sinai.org


Levels of circulating red blood cell (RBC)-derived vesicles are increased in sickle cell anaemia (SCA) and thalassaemia intermedia (TI) but the mechanisms, effects and controlling factors may differ. This study found that levels of vesicles and intravascular haemolysis were linked as shown by the correlation between levels of vesicles and plasma Hb. Vesicle levels were 6-fold greater in SCA and 4-fold greater in TI than in controls. The proportion of plasma Hb within vesicles was increased in SCA and TI with a significantly higher proportion in TI. We examined whether subpopulations of RBC expressing phosphatidylserine (PS) were a source of PS(+) vesicles and observed a significant association. Thrombin generation was promoted by the vesicles in which 40–50% expressed PS. In TI, markers of thrombin generation were significantly related to PS(+) RBC. Splenectomy in TI had significant effects including greater increases in vesicle levels, plasma Hb, PS(+) RBCs and thrombin generation markers than in unsplenectomised patients. In hydroxycarbamide (HC)-treated SCA patients these measures were decreased compared with untreated controls. The relationship between vesicle levels and plasma Hb suggests a mechanism linking vesiculation to haemolysis and consequently nitric oxide (NO) bioavailability and suggests a means by which HC treatment improves NO bioavailability.

The presence of circulating levels of RBC-derived vesicles has long been recognized in sickle cell anaemia (SCA) and was initially described by Allan et al (1982) who observed that vesicle shedding occurred during unsickling of sickle red blood cells (SRBC). The vesicles contained Hb but were almost completely devoid of the membrane cytoskeletal polypeptides, spectrin, ankyrin and band 4·1. We, and others, have reported that circulating vesicle levels are also observed in patients with other haemolytic anaemias (Wagner et al, 1986; Setty et al, 2000; Westerman et al, 2002; Pattanapanyasat et al, 2004). The effects of vesiculation include loss of RBC surface area and microspherocyte formation (Allan et al, 1982), and simulates the vesiculation observed in senescent and apoptotic normal cells (Bosman et al, 2005). The vesicles have strong procoagulant effects caused by the presence of phosphatidylserine (PS) on the outer surface of the vesicle and RBC membranes (Westerman et al, 1984; Franck et al, 1985). Vesiculation of cells also provides a model for the study of the selective removal of membrane proteins (Knowles et al, 1997). Differences in the increased levels of the circulating vesicles during steady states in SCA and TI may provide a better understanding of the vesiculation process.

In the present study we have determined the levels of circulating RBC-derived vesicles during steady states in patients with SCA and thalassaemia intermedia (TI) with moderate anaemia, not requiring regular blood transfusion. The characteristics and origin of vesicles expressing PS were studied in these two groups of patients: we have related the findings to disturbances of haemolysis and have considered the potential effects on nitric oxide (NO) bioavailability. We have examined the in vitro effects of vesicles on thrombin generation and the correlation of the vesicles to markers of haemostasis. The effects of splenectomy and hydroxycarbamide (HC) treatment on vesicle levels in patients with SCA and TI respectively were assessed as they may yield insights into the pathophysiology of vesiculation.


Patient selection and blood collection

Patients (18–56 years) with SCA and TI were recruited from the Haematology Clinic at University College London Hospital. Thirty four patients with SCA (six men, 28 women), 19 patients with TI (eight men, 11 women) and 27 controls (14 men, 13 women) were studied. All patients were in a steady state and had not been transfused for at least 3 months. TI patients were homozygotes or compound heterozygotes for β-thalassemia mutations with the exception of two patients: one was a compound heterozygote for HbE-β thalassaemia and the second was a compound heterozygote for HbE and HbH Constant Spring. Mean Hb values and ranges were similar in the SCA patients (mean 72·9 g/l, range 56–100, n = 31) and TI patients (mean 75·4 g/l, range 46–95, n = 13). Hb values in the patients with HbE-β thalassaemia and the HbE- H Constant spring fell within these ranges. Informed consent was obtained and the study was approved by the hospital Ethical committee.

Venous blood for vesicle and RBC examination (10 ml) was collected using 19 or 21-guage needles or a butterfly into a tube containing CPDA (Citrate-Phosphate-Dextrose-Adenine) to give a ratio of 8:1 (blood:anticoagulant). The samples were immediately transported for staining and analysis. RBC quantitation was obtained on a KX21 analyzer (Sysmex, Milton Keynes, UK). In preparation for flow cytometric analysis, blood samples were spun at 2000 g for 10 min, the supernatant carefully removed and respun at 3000 g for 10 min. The supernatant was used for vesicle analysis. Blood for haemostatic studies was taken by clean venepuncture into 0·105 M trisodium citrate, with a ratio of one part anticoagulant to nine parts whole blood. The samples were centrifuged at 2000 g for 15 min. The upper two thirds of the plasma was removed and then subjected to a second centrifugation of 2000 g for 15 min. Aliquots of plasma were stored in polypropylene cryo-tubes (2 ml cryo-tubes (Sarstedt Ltd. Leicester, UK) at –70°C until required for assay, then thawed at 37°C.

Flow cytometric analysis of vesicles and RBC in whole blood

R-phycoerythrin (RPE)-labeled mouse monoclonal antibody against human glycophorin A (GPA)(anti-glycophorin A RPE) clone JC 159 (Dako, Denmark) was employed to identify RBC and vesicles, while annexin V-fluorescein isothiocyanate (FITC) (Roche Diagnostics GmbH, Germany) was used as a marker for PS positivity. Briefly, 5 μl of CPDA-anticoagulated blood was incubated with 2 μl of anti-glycophorin A RPE antibody for 30 min at 0°C, after which 500 μl of annexin-V buffer (140 mM NaCl, 5 mM CaCl2, 10 mM HEPES pH 7·1) containing 2 ul FITC- AV was added. Samples were stored on ice and flow cytometric analysis was performed within 30 min. All samples were run as duplicates. Analysis was performed on a EPICS ELITE flow cytometer [Beckman Coulter (UK) Ltd. High Wycombe, Bucks, UK].

Acquisition and data analysis were performed by using EPICS ELITE work station software version 4·01. At least 50 000 total events per sample were acquired to minimize sampling errors. The data was collected at a flow rate of less than 500 events per second since it was determined previously that higher sample flow rates resulted in particle coincidence artifacts. Samples were analyzed for 5 min and data collection began after the sample flow rate had been stabilized.

The vesicles and RBC populations were defined on histograms by their GPA and forward light scatter characteristics (Fig 1A). Annexin V binding histograms for vesicle and RBC populations (Fig 1B and C) yielded bimodal distributions i.e. annexin V-negative and annexin V-positive populations. Analysis gates for the annexin V binding histograms were applied using the following rules, a) gate G1 was set with the leftmost edge at the nadir point between the bimodal peaks (Fig 1B), b) gate G2 was set in a similar fashion (Fig 1C). The RBC annexin V-positive population (Fig 1C) formed a minor peak to the right of the RBC annexin V-negative population and the y-scaling was temporarily reduced to enable accurate placing of gates. The diameters of the vesicles were between 100 and 500 nm.

Figure 1.

 (A) Bivariate dot log plot of forward scatter on the x-axis versus Glycophorin A-RPE fluorescence on the y-axis, for a healthy control. Region R1 represents the RBC population and Region R2 represents RBC-derived microvesicles. Annexin-V positivity is displayed on single parameter histograms (B, Region 2) and (C, Region 1). PS is shown on the x-axis (log scale) and event frequency is shown on the y-axis for vesicles (R2) and for the RBC (R1). Gated regions G1, G2 show the PS-positive regions for vesicles and RBC respectively.

Calculation of the total [GPA(+)] vesicles, PS(+) vesicles and PS(+) RBC

Total [GPA(+)] vesicles × 109/lblood = (R2/R1) × (RBC count × 9/8 × 106) where R1 = Total number of RBC [GPA(+)] and R2 = Total number of vesicles [GPA(+)] (Fig 1A).

PS(+) vesicles × 109/lblood = (Gate G1/R1) × (RBC count × 9/8 × 106) where G1 = Number of vesicles [annexin V(+)] (Fig 1A and B).

PS(+) RBC × 1012/lblood = (Gate G2/R1) × (RBC count × 9/8× 106) where G2 = Number of RBC [annexin V(+)] (Fig 1A and C).

Vesicle enrichment

For some experiments, high-purity populations of vesicles were obtained by electronic sorting. CPDA-anticoagulated blood was centrifuged 400 g for 10 min at room temperature and the plasma collected. The plasma was diluted 2:1 in annexin V buffer and further centrifuged at 10 000 g for 30 min at 4°C. Most of the supernatant was removed leaving approximately 5 μl of plasma (vesicles) which was stained with GPA antibody as above. Electronic sorting was carried out on an EPICS ELITE flow cytometer. RBC-derived vesicles were identified on bivariate dot plots as above and electronic sorting carried out on these populations.

Plasma Hb evaluation and contribution of vesicle Hb to total plasma Hb level

Plasma Hb was assayed by a micromethod using Drabkin’s reagent Hb (Moore et al, 1981). Since the encapsulated vesicles contain Hb, which would contribute to total plasma Hb levels, we measured the contribution of vesicle Hb to total plasma Hb levels. An aliquot of blood from each patient was spun as described above for preparing plasma vesicle samples for flow cytometric analysis. An aliquot of this was taken and prepared to obtain vesicle-free plasma samples. This aliquot was centrifuged additionally at 10 000 g for 60 min at 4°C to sediment the vesicles in a pellet after which the supernatant was then removed for testing for plasma Hb. Both samples were stored at −70°C prior to testing. The frozen samples were thawed and the absolute numbers of vesicles in each sample were determined by flow cytometry. Triton X-100 (1% v/v) was added to the samples and plasma Hb was measured in each sample. The level of vesicle Hb in plasma was obtained by subtracting the plasma Hb level in the vesicle-free sample from the plasma Hb level in the solubilized sample which contained both vesicle Hb (tritonized) and free Hb. The fraction of plasma Hb that was contributed by vesicle Hb was then determined.

In separate experiments, the effect of freezing of the plasma samples was determined by comparing the same sample, fresh and freeze-thawed. Results obtained confirmed that vesicle numbers were unaffected by freezing (data not shown).

Haemostasis testing

The samples were assessed for in vivo markers of thrombin generation: prothrombin fragments 1·2 (PF1.2) and thrombin/antithrombin complex (TAT) using commercial enzyme-linked immunosorbent assay methods (Enzygnost kits, Dade Behring, Marburg, Germany) and D-dimer, using a latex immuno-turbidometric procedure (D-dimer Plus reagent; Dade Behring), on a CA-1500 coagulometer (Sysmex).

Thrombin generating potential

Thrombin generation was measured by a modification of a previously described method (Helley et al, 1996). Briefly, 50 μl of vesicle enriched plasma was suspended in HEPES CaCl2 buffer pH 7·5 (100 mM HEPES, 136 mM NaC1, 2·7 mM KCl, 2 mM MgCl2, 5 mM D-Glucose, 4 mM CaCl2 containing 1 mg/ml bovine serum albumin [BSA]) and incubated for 2 min at 37°C with 50 μl of human prothrombin (reaction concentration [rc] 8·17 uM, in HEPES CaCl2 pH 7·5). To this was added 50 μl of factor Xa (rc 4 nM, in HEPES CaCl2 buffer pH 7·5) plus 50 ul of buffered factor Va (rc 0·01 uM, in HEPES CaCl2 pH 7·5) and the reaction mixture was then incubated for 8 min at 37°C followed by the addition of 450 μl of stop buffer (50 mM Tris, 90 mM NaC1, 20 mM Na EDTA, pH 7·5 containing 1 mg/ml BSA). Thrombin activity was subsequently assessed on a CA-7000 automated coagulation analyzer (Sysmex) using the amidolytic substrate S2238 (Chromogenix, Instrumentation Laboratory, Milan, Italy). The purified proteins, Factors X, Va, II, were obtained from Enzyme Research Laboratories (Swansea, UK). All other chemicals were obtained from Sigma-Aldrich (Poole, Dorset, UK).

Statistical analysis

The data generated by the study was analyzed using SAS version 8 for Windows (Cary, NC). All data are shown as mean ± SEM. As the data analysis related to vesicle and PS(+) RBC levels, and to haemostatic markers, failed to show a normal distribution (proved by Kolmogorow-Smirnow normality test), a non-parametric Kruskal-Wallis test was used for an overall comparison between SCA and TI patients and normal controls. For comparisons between SCA and TI patients, the Wilcoxon rank sum test was used. Spearman’s rho test was used to determine bivariate correlations. Correlations between different continuous variables were determined using the Spearman’s rank correlation test. All statistical tests employed an alpha-level of 0·05. Differences were considered significant at P < 0·05 (two-tailed).


Levels of vesicles, PS(+) vesicles and PS(+) RBC in all patients with SCA, TI and controls

Patients with SCA and TI had significantly increased levels of circulating vesicles, being approximately 6-fold and 4-fold more than normal controls, respectively (Table I, column 1). The increase in vesicles was even more marked when expressed in relation to the total number of circulating red cells, being 11-fold and 6-fold greater than controls in SCA and TI, respectively (Table I, column 6). PS(+) vesicles were also significantly increased in SCA and TI, being approximately 9- and 3-fold greater, respectively, than controls (Table I, column 2). The levels of vesicles in the E-β thalassaemia patient and the HbE-HbH Constant Spring patient fell within the range of other TI patients (data not shown). The percentage of vesicles expressing PS was less than 50% in the patient groups (Table I, column 3) but was greater than controls in SCA and less than controls in TI. The number of RBC expressing PS was significantly greater (approximately 4-fold) in SCA and approximately 3-fold greater in TI than controls (Table I, column 4). The percentage of RBC expressing PS in SCA and TI were 7-fold and 5-fold, respectively, greater than controls (Table I, column 5).

Table I.  Vesicle, RBC and plasma Hb levels in patients with SCA, TI and controls.
 Vesicles (×109/l blood)PS(+) vesicles (×109/l blood)%PS(+) vesiclesPS(+) RBC (×109/l blood)%PS(+) RBCVesicle/RBC* (ratio × 100)Plasma Hb (mg/l)
  1. All samples were taken during a steady state in the absence of transfused RBC.

  2. *Mean ± SEM.

  3. †Number of patients in whom measurements were taken.

  4. p11 = SCA vs. controls; p22 = TI vs. controls; p33 = SCA vs. TI.

 13·2 ± 2·8*8·5 ± 2·349·2 ± 4·436 ± 7·01·6 ± 0·40·56 ± 0·1145·2 ± 32·9
Thalassaemia intermedia
 8·9 ± 1·02·9 ± 0·436·7 ± 4·525·3 ± 5·80·9 ± 0·30·32 ± 0·01341·7 ± 62·9
 2·1 ± 0·51 ± 0·244·5 ± 2·88·1 ± 1·60·2 ± 0·0020·05 ± 0·011·6 ± 0·05

The extent of PS positivity in the circulating vesicles in patients with SCA (<50%) (Table I, column 3), differed from previously described findings in which 100% of the vesicles were PS(+)(Setty et al, 2000). The difference may result from patient selection or from technical differences. Our measurements were on functionally asplenic adults with SCA, while the earlier study examined children with SCA who would have functional spleens. A progressive rise in vesicle numbers and in the proportion of vesicles expressing PS occurred if samples were not analyzed within 2 h (data not shown). Our results are compatible with those which indicate that vesiculation of RBC may continue to occur at least up to 24 h (Wagner et al, 1986) and were similar to studies in normal rats in which 67% of RBC shed vesicles were PS(+)(Willekens et al, 2005).

Relationship of PS(+) vesicles to PS(+) RBC

There was a clear positive relationship between the number of circulating PS(+) vesicles and the number of circulating PS(+) RBC in all patients (Fig 2). When the relationship was examined by individual patient groups, the linear relationship between PS(+) vesicles and PS(+) RBC for patients with SCA was r = 0·55, P = 0·001, n = 32, for patients with TI was r = 0·67, P = 0·006, n = 15, for controls was r = 0·71, P < 0·0001, n = 25 and for all patients and controls taken together was r = 0·70, P < 0·0001, n = 72. Although the above analyses included SCA patients, both treated and untreated with HC, when SCA patients treated with HC were excluded from analysis, a linear relationship between PS(+) vesicles and PS(+) RBC was still seen (r = 0·56, P = 0·005).

Figure 2.

 The relationship between PS (+) vesicles and PS (+) RBC are shown on a log scale. SCA patients (•), TI patients (x) and controls (o).

The proportion of vesicles expressing PS increased in patients with SCA as the number of circulating vesicles increased (r = 0·49, P = 0·004) (n = 34), but not in patients with TI (r = 0·038, P = 0·88) or in controls (r = 0·023, P = 0·91) (n = 27).

Plasma Hb and the relationship of vesicle levels to plasma Hb in SCA and TI

Levels of plasma Hb were significantly higher in patients with SCA and TI than in controls (Table I, column 7) being 9- and 20-fold, respectively, greater in SCA and TI than those in controls. The number of vesicles in SCA and TI were significantly correlated to plasma Hb in SCA (r = 0·45, P = 0·006, n = 21) and TI (r = 0·79, P = 0·01, n = 9) (Fig 3).

Figure 3.

 The relationship between plasma Hb and vesicles is shown in healthy controls, TI patients and SCA patients. [(o) healthy controls, (x) TI patients, (•) SCA patients]. The correlations shown are those between plasma Hb and vesicles in TI (A) and plasma Hb and vesicles in SCA (B).

The proportion of plasma Hb that was contained within vesicles in patients with SCA was 15·7 ± 10·1% of the total plasma Hb whereas the proportion was significantly higher (34·6 ± 16·5%) (P = 0·008) in TI patients.

Relationships of vesicles to markers of thrombin generation

Thrombin generation in vitro was quantified in vesicle preparations in four patients with SCA. It can be seen in Fig 4 that thrombin generation increased as the number of vesicles in each patient increased. Markers of thrombin generation (PF1.2, TAT and D-dimer) in vivo are shown in Table II in patients with SCA and TI and in controls. In patients with SCA the levels of PF1.2, TAT and D-dimers were significantly greater than controls while in TI the level of D-dimers was significantly greater than controls. PF1.2 was significantly higher in SCA than in TI.

Figure 4.

 Relationship between number of vesicles and thrombin generation in patients with SCA. Each dot represents a single patient. Vesicles were Glycophorin A(+) (99% pure).

Table II.  Markers of thrombin generation in patients with SCA, TI and controls.
 PF1.2 (nM/l)TAT (ug/l)D-dimer [ng/ml (FEU)]
  1. All samples were taken during a steady state in the absence of transfused RBC.

  2. *Mean ± SEM.

  3. †Number of patients in whom measurements were taken.

  4. p11 = SCA vs. controls; p22 = TI vs. controls; p33 = SCA vs. TI.

 1·68 ± 0·3*15·48 ± 4·08407 ± 55
Thalassaemia intermedia
 1·0 ± 0·18·33 ± 2·41267 ± 61
 0·96 ± 0·075·48 ± 0·5117 ± 38

In patients with TI, correlations between both PF1.2 and TAT and PS(+) RBC were significant [(r = 0·67, P = 0·007, n = 11) and (r = 0·55, P = 0·04, n = 11) respectively]. Correlations between both PF1.2 and TAT and PS(+) vesicles were r = 0·37 and r = 0·39 respectively. D-dimers were not related to PS(+) RBC or to PS(+) vesicles. In SCA relationships between PF1.2, TAT and D-dimers and PS(+) RBC and PS(+) vesicles were not significant.

Relationship of splenectomy to vesicles, PS(+) RBC, plasma Hb and thrombin generation in TI

In splenectomised patients with TI, the number of vesicles and the levels of TAT were significantly higher than in unsplenectomised patients (Table III). The levels of PS(+) vesicles, PS(+) RBC and plasma Hb, as well as PF1.2 and D-dimers were also higher in the splenectomised patients but did not reach significance, most likely due to the small numbers of patients in these subgroups. As the number of vesicles/l of blood may depend on the RBC count, which may differ in splenectomised and unsplenectomised patients, we measured RBC counts in both groups. In splenectomised patients, RBC counts were (mean ± SEM)(N.))s 2·88 ± 0·32 × 1012/l (9) and in unsplenectomised patients 3·84 ± 0·27 × 1012/l (3) P = 0·09). During flow cytometric analysis, the histograms for splenectomised TI patients regularly showed a long “tail” in the R1 cloud which was absent in patients with intact spleens (Fig 5).

Table III.  Effects of splenectomy in patients with thalassaemia intermedia.
 Vesicles (×109/l blood)PS(+) vesicles (×109/l blood)PS(+) RBC (×109/l blood)Plasma Hb (mg/l) PF1.2 (nM/l) TAT (μg/l)D-dimer [ng/ml (FEU)]
  1. All samples were taken during a steady state in the absence of transfused RBC.

  2. *Mean ± SEM.

  3. †Number of patients in whom measurements were taken.

  4. P = No splenectomy vs. splenectomy.

No splenectomy5·62 ± 0·45* (4)†2·26 ± 0·86 (4)14·85 ± 8·11 (4)202·3 ± 41 (3)0·65 ± 0·05 (3)2·53 ± 0·26 (3)149·33 ± 66·66 (3)
Splenectomy11·29 ± 1·12 (9)3·44 ± 0·43 (9)33·55 ± 8·04 (9)486·0 ± 36·6 (5)1·15 ± 0·12 (7)11·57 ± 3·21 (7)300·71 ± 89·06 (7)
P 0·050·34 0·17  0·080·07 0·05  0·33
Figure 5.

 Bivariate dot log plots of unsplenectomized and splenectomized patients with TI.

Relationships of HC treatment to vesicles, PS(+) RBC, plasma Hb and thrombin generation in SCA

The levels of vesicles, PS(+) vesicles, PS(+) RBC and plasma Hb as well as the levels of PF1.2, TAT and D-dimers were lower in patients with SCA treated with HC than in untreated patients (Table IV). Differences in vesicle levels, PS(+) RBC, plasma Hb and D-dimers reached statistical significance. Among the SCA patients analysed (Table I), seven had been receiving HC therapy for at least 3 months at doses of 10–30 mg/kg per day. Patients selected for HC therapy had experienced at least three crises in a 1-year period or an episode of sickle chest syndrome. An unintentional bias in selection of HC-treated patients may have occurred because our study was cross sectional rather than longditudinal. This would seem unlikely, however, since all the HC-treated patients had severe disease and our selection would therefore have underestimated, rather than overestimated, differences.

Table IV.  Effects of HC treatment in patients with sickle cell anaemia.
 Vesicles (×109/l blood)PS(+) vesicles (×109/l blood)PS(+) RBC (×109/l blood)Plasma Hb (mg/l) PF1.2 (nM/l) TAT (ug/l)D-dimer [ng/ml (FEU)]
  1. All samples were taken during a steady state in the absence of transfused RBC.

  2. *Mean ± SEM.

  3. †Number of patients in whom measurements were taken.

  4. P = No HC treatment vs. HC treatment.

No HC treatment15·59 ± 3·07* (27)†10·05 ± 2·69 (26)42·71 ± 8·45 (25)167·4 ± 38·7 (17)1·8 ± 0·39 (12)18·38 ± 5·22 (12)479·95 ± 60·14 (12)
HC treatment3·97 ± 0·83 (7)1·93 ± 0·78 (6)11·83 ± 3·95 (7)51·0 ± 16·0 (4)1·32 ± 0·24 (4)6·8 ± 0·27 (4)189·75 ± 28·69 (4)
P0·04 0·08 0·01  0·050·59  0·26  0·02

Relationship of vesicle numbers to Hb F

An analysis of the relationship between vesicle numbers in SCA and Hb F (g/d) showed that vesicle numbers and Hb F were inversely related (r = 0·4, P = 0·12, n = 17) (Fig 6).

Figure 6.

 The relationship between Hb F and the number of circulating vesicles is shown in patients with SCA.


Vesiculation of RBC in patients with SCA and TI is associated with a number of changes of potential pathological and physiological significance. The findings of this study showed that the level of circulating RBC-derived vesicles were linked with lysis of the parent RBC (Fig 3). In SRBC and TI-RBC premature ageing also contributes to the vesiculation (Bosman, 2004)(Yuan et al, 1992). In both SCA and TI, vesicle levels and intravascular haemolysis, as indicated by plasma Hb levels, were increased (Table I, columns 1,2,7). Of interest is the observation that intravascular haemolysis, which appears to be a biomarker for NO resistance, characterizes clinical subphenotypes in patients with haemolytic anemia (Kato et al, 2006). In SCA the subphenotype includes priapism, leg ulceration, pulmonary hypertension and sudden death and, in TI, may similarly include pulmonary hypertension (Atichartakarn et al, 2003), priapism (Jackson et al, 1986) and leg ulceration (Camaschella & Cappellini, 1995). The linkage between vesicle levels and haemolysis identifies an additional cause for the intravascular haemolysis observed in SCA and TI.

Free Hb is considered to have a significant effect on bioavailability of NO (Reiter et al, 2002), so it could be important to determine the extent of the contribution of vesicle Hb to free plasma Hb. Furthermore, Hb release from vesicles may differ from that which occurs in the parent RBC since the membranes of vesicle Hb are depleted of cytoskeletal polypeptides. Our observations indicate that intravesicular Hb contributes approximately 16% of the total plasma Hb in SCA and 35% in TI, which could have significant effects on the level of plasma Hb and consequently on NO bioavailability. The relatively small size of the vesicles (100–600 nm) may also influence scavenging rates of NO and consequently NO bioavailability, as decreases in surface area of the unstirred diffusional barrier relative to decreases in intracellular Hb concentration would increase reaction rates of NO (Jeffers et al, 2006).

The significantly greater level of plasma Hb in patients with TI as compared to patients with SCA (Table I, column 7) would not be predicted because vesicle levels and plasma Hb were closely linked in SCA and TI (Fig 3). This may in part be explained by the significantly higher proportion of plasma Hb contained within vesicles in TI (35%) than within SCA (16%). A contributing factor may also be the higher rate of ineffective erythropoiesis and intramedullary haemolysis that occurs in TI. Intramedullary lysis of marrow RBC in TI would add increments of Hb to the already increased plasma Hb, while vesicles derived from marrow RBC would be phagocytized in the marrow and not contribute to circulating vesicle levels.

The precise origin of PS(+) vesicles has not previously been examined. The close relationship between the number of RBC expressing PS and the circulating PS(+) vesicles in patients with SCA and TI (Fig 2), as well as the similarly increased PS(+)RBC and PS(+) vesicles in splenectomised patients with TI (Table III), suggest that PS(+) RBC are a significant source of PS(+) vesicles in SCA and TI. It is possible that, in TI, vesicles are shed, both early at the reticulocyte stage or later as senescent cells, as cells are cleared. In principle, the age at which vesicles are shed from red cells could be explored by examining for transferrin receptor expression on the red cell vesicles, their presence suggesting that red cells shed vesicles as early as the reticulocyte stage.

Our findings support and define the link between circulating vesicle levels and activation of thrombin in SCA and in TI. Since a prothrombotic state is known to exist in both SCA and thalassaemia syndromes, the procoagulant PS, which is expressed on vesicles and RBC in SCA and TI, may contribute to such a state (Helley et al, 1996; Wood et al, 1996). We examined mechanisms which may affect the relationship of vesicle levels to markers of thrombin activation in the disorders. A direct effect of vesicles on thrombin generation is shown with the in vitro experiment linking vesicle numbers to thrombin generation in patients with SCA (Fig 4). The observations that increased PS(+) vesicles in SCA and TI (Table I, column 2) are accompanied by increased markers of thrombin generation in SCA and TI (Table II), and that PS is present on 49% of SRBC derived vesicles and on 37% of TI-RBC derived vesicles (Table I, column 3), show a linkage between vesicle levels and activation of thrombin. Furthermore, in TI, there are associations between PS(+) vesicles and PF1.2 and TAT and between PS(+) TI-RBC and PF1.2 and TAT. The latter finding is significant.

These results may contribute to the increased risk of thrombosis in splenectomised patients with TI (Cappellini et al, 2000). The finding that TAT values were significantly higher in TI patients (splenectomised) (Table III), while PF1.2 and D-dimer were not altered, implies that TAT may be the more significant indicator of thrombin generation. Increased total vesicle levels in SCA associated with increased levels of markers of coagulation activity have been previously noted (Franck et al, 1985; Shet et al, 2003).

Sickle RBC may make a greater contribution to thrombin activation than TI-RBC. In SCA the number of PS(+) RBC was 4-fold higher than controls compared to TI, in which TI-RBC were 3-fold higher than controls (Table I, column 4). The factors causing the difference may include different proportions of PS(+) RBC in SCA and TI, differences in the proportion of type 1 and type 2 RBC (Yasin et al, 2003), phenotypic differences in SRBC and TI-RBC or possibly the presence of other procoagulants on SRBC and TI-RBC.

Levels of vesicles, PS(+) RBC, plasma Hb and markers of thrombin generation, which differ in patients with SCA and TI, appear to be affected by a number of factors including diagnosis, splenectomy and treatment with HC. Comparisons between patients with SCA and TI of overall vesicle levels, including PS(+) vesicle levels, the number of PS(+) RBC, the ratio of vesicles to RBC and markers of coagulation activity, revealed greater levels in SCA than in TI (Tables I and II).

In splenectomised patients with TI, greater levels of vesicles, including PS(+) vesicles, PS(+) RBC and markers of coagulation activity, were observed than in unsplenectomised patients (Table III). The role of splenectomy in contributing to the increased levels of circulating vesicles in TI has been unclear. A likely explanation is that splenectomy is associated with increased binding of globin chains to the RBC membrane skeleton (Schrier et al, 1989), which are potential sites of vesicle formation and shedding. Furthermore, in the absence of a functioning spleen, clearance of aberrant RBC, which produce vesicles, would be diminished and would contribute to the increase in vesicle levels. The levels of vesicles and PS(+) RBC in splenectomized TI patients (Table III) approach those of SCA patients, suggesting that vesicle levels in SCA and TI reflect the presence or absence of the spleen (Table I, columns 1,4).

The possibility that the effect of splenectomy on vesicles may be due to an increase in RBC counts with consequent increased vesicle levels would seem less likely since RBC counts in splenectomised patients did not appear to be greater than unsplenectomised patients. The finding of a significant increase in plasma Hb (Table III) in splenectomised TI patients is consistent with the observed increase in circulating vesicles. Levels of vesicles in three patients, who had received splenectomy prior to successful treatment for Hodgkin disease, did not show increased vesicle levels compared to unsplenectomised healthy controls (data not shown). Although vesiculation might be facilitated by the spleen in healthy individuals (Willekens et al, 2003), this would not explain the higher vesicle levels in splenectomised TI patients.

The increase in vesicle levels in splenectomised patients is associated with the presence of a characteristic “tail” observed in the histograms of the splenectomised patients with TI (Fig 5). The tail contains GPA(+) particles, which are smaller RBC than those observed in the main population of TI-RBC. These cells probably represent cells from which vesicles have previously been released with consequent decrease in the size of the parent cells. In the presence of a functioning spleen, the cells contained in the “tail’ would have been cleared more rapidly.

Hydroxycarbamide-treated patients with SCA have lower levels of circulating vesicles, PS(+) RBC, plasma Hb and markers of coagulation activity than untreated patients (Table IV), supporting a possible link between vesicle levels, haemolysis and activation of coagulation. The decreased vesicle levels in HC-treated patients may be related to the increased Hb F concentrations observed within HC-treated SRBC (Fig 6). In HC-treated patients, however, lessened haemolysis and improvement in anaemia can precede increments in HbF levels (Rodgers et al, 1990), suggesting that additional mechanisms may be involved in the response of patients with SCA to HC.

The decrease in haemolysis induced by HC (Table IV) is also of interest in view of the potential effect of HC on the treatment of priapism in patients with SCA. In a recent study in patients with SCA, priapism was related to increased intravascular haemolysis with presumed effects on NO bioavailability (Nolan et al, 2005). Since the effectiveness of HC in the treatment of priapism, as shown in a earlier study, was related to the absence of the spleen (Jackson et al, 1986), the above observations suggest that HC treatment may be useful in the treatment of priapism in asplenic adults, such as those with SCA. The presence of increased HbF levels does not appear to protect patients with SCA against priapism (Nolan et al, 2005).

The lowering effects of HC on haemolysis and thrombin generation may also contribute to the therapeutic effect of HC on the vasocclusive pain episodes observed in SCA since the episodes are associated with haemolytic rates and coagulation activity which are greater than those observed during steady states (Naumann et al, 1971; Francis & Hebbel, 1994; Wood et al, 1996; Shet et al, 2003; Hillery et al, 2004; Ballas & Marcolina, 2006).

The present work is the first to compare levels of vesicles and PS(+) RBC in SCA and TI and to examine factors that determine such levels as well as their consequences in these conditions. This study has included an examination of relationships between levels of circulating vesicles and intravascular haemolysis with its potential consequences to NO bioavailability. The study has also yielded insights into the affects of circulating vesicles on the relationship between vesicle levels and on coagulation activity. The findings, that splenectomy in TI and HC treatment in SCA are associated with modulation of vesicle levels, provide further insight into the consequences of changes in vesicle levels, PS(+) RBC, plasma Hb and markers of thrombin generation. On the basis of these findings, we suggest that prospective studies to examine the effects of HC on vesicle levels and the consequent effects on haemolysis and thrombin generation should be considered.


The authors would like to thank Dr David Allan for his very valuable discussions and suggestions.