Frequency and perturbations of various peripheral blood cell populations before and after eculizumab treatment in paroxysmal nocturnal hemoglobinuria

Abstract

While red blood cells (RBCs) and granulocytes have been more studied, platelets and reticulocytes are not commonly used in paroxysmal nocturnal hemoglobinuria (PNH) flow-cytometry and less is known about susceptibility to complement-mediated destruction and effects of anti-complement therapy on these populations.

We performed flow-cytometry of RBCs and granulocytes in 90 PNH patients and of platelets and reticulocytes in a subgroup (N=36), to unveil perturbations of these populations during PNH disease course before and after anti-complement treatment.

We found that platelets and reticulocytes were less sensitive to complement-mediated lysis than RBCs but not as resistant as granulocytes, as shown by mean sensitive fraction (difference in a given PNH population vs. PNH granulocyte clone size). In treated patients, reticulocytes, platelets, RBCs (with differences between type II and III) and granulocytes significantly increased post-treatment, confirming the role of PNH hematopoiesis within the context of anti-complement therapy. Moreover, we found that PNH platelet clone size reflects PNH granulocyte clone size. Finally, we established correlations between sensitive fraction of PNH cell-types and thrombosis.

In sum, we applied a flow-cytometry panel for investigation of PNH peripheral blood populations’ perturbations before and after eculizumab treatment to explore complement-sensitivity and kinetics of these cells during the disease course.

BACKGROUND

Somatic phosphatidylinositol N-acetylglucosaminyltransferase subunit A (PIGA) mutations occurring in the hematopoietic stem cells (HSCs) underlie the pathogenesis of paroxysmal nocturnal hemoglobinuria (PNH).[1] PIGA mutations lead to impaired biosynthesis of glycosylphosphatidylinositol (GPI)-anchor with various functional consequences.[2] Mainly, PNH cells are sensitive to complement-mediated hemolysis due to lack of CD55 and CD59, both complement regulatory molecules, on the surface of affected red blood cells (RBCs), while leukocytes are less sensitive due to alternative defense mechanisms.[3] A triad of symptoms can dominate the clinical picture of PNH: intravascular hemolysis, bone marrow failure and thromboembolic complications.

Detection of PNH cells is essentially based on flow cytometry identification of white blood cells (WBC) and RBCs, which are negative for markers of GPI-anchored proteins(APs).[4] The grade of GPI-APs deficiency on the RBCs can distinguish: PNH type I (T-I) with a normal expression pattern; PNH type II (T-II) with a partial reduction; and PNH type III (T-III) with a total absence of GPI-APs.[5]

While PNH RBCs seem to be most sensitive to complement, other blood cell types show different sensitivity patterns. Thus, PNH T-II and T-III erythrocytes are respectively 3–5 and 15–25 times more sensitive to complement lysis compared to normal RBCs,[6] whereas in PNH WBC and platelets, complement-induced lysis is estimated to be 5–10 to 10–32 times higher than normal counterparts, as shown with immune lysis technique in the pre-flow cytometry era.[7]

A recent prospective study on a large, multi-center cohort showed that PNH WBC clone size correlates with the risk of thromboembolic events and, more generally, with disease burden.[8] Clinically, the therapeutic application of eculizumab, a humanized monoclonal antibody that prevents the terminal steps of complement cascade, is effective in preventing intravascular hemolysis and improving anemia and it is also able to reduce the risk of thrombosis.[9, 10] Experimentally, eculizumab responses have provided new opportunities to study the effects of complement therapy on cells from various blood lineages, including platelets and reticulocytes. However, data concerning the effects of anti-complement therapy on different PNH cell subtypes are fragmented. While routine diagnostic PNH testing includes quantitation of the GPI-APs deficient fraction within granulocytes and RBCs, the contribution of PNH platelets and reticulocytes production and the effect of eculizumab exposure on other blood components has been less well studied.

Here, we characterized all PNH-derived populations in a large cohort of PNH patients (N=90), to determine the dynamics of clonal expansion of PNH blood lineages under standardized conditions and upon complement fraction 5 (C5) blockade.

MATERIALS AND METHODS

Patient selection

Among 295 patients diagnosed between 2002 and 2018 with a PNH clone of any size (>0.5% of PNH granulocytes by flow cytometry), we identified 90 patients who fulfilled the diagnostic criteria for PNH, including granulocyte and/or RBC PNH clone size >20%, as shown in Table I. Blood samples were obtained and clinical data were reviewed for our historical cohort of patients in accordance with protocols approved by the Institutional Review Board of the Cleveland Clinic and the Declaration of Helsinki. Among our cohort of 90 patients, flow cytometric data for granulocytes and type II/III RBCs were available for 57 (63%) naïve patients and 43 (48%) patients while on eculizumab. Platelets and reticulocytes data were available for 36 patients (40%), 15 naïve and 21 treated. Serial data of 22 patients were available for granulocytes and RBCs, with 12 of these patients being characterized also for type II/III RBCs. Median age was 49 years (range 18–90) with no significant differences in gender distribution (47 female and 43 male patients, see also Supplemental Table I). Median PNH granulocyte clone size was 76.85% (20 – 99.72%) for the entire PNH cohort, with no differences between primary (p)PNH (N=48) and secondary (s)PNH (e.g., patients who had a history of antecedent AA, N=42) cohorts (70% vs. 58% respectively, p=0.4225).

Table I.

Patient and disease characteristics

Characteristics	Total	pPNH	sPNH
N (%)	90	48 (53)	42 (47)
Age, years, median (range)	49 (18–90)	49 (18–90)	50 (23–86)
Gender, female, n (%)	47 (52)	26 (54)	21 (50)
Eculizumab Treatment, n (%)	55 (61)	29 (60)	26 (62)
Thrombotic event, n (%)	20 (22)	10 (21)	10 (24)
PNH clone size in granulocytes, %, median (range %)	76.85 (22.4 – 99.72)	82.44 (22.4 – 99.72)	73.3 (26.3–99.61)
Patients with Predominant Type II RBC, n (%)	18 (20)	10 (21)	8 (19)
sPNH	Presence or history of bone marrow failure in conjunction with clinical and laboratory evidence of intravascular haemolysis and PNH clone >20%.
pPNH	Clinical and laboratory evidence of intravascular haemolysis and PNH clone >20%, but no evidence of bone marrow failure.

PNH: paroxysmal nocturnal haemoglobinuria; sPNH: seconday; pPNH: primary, RBCs: red blood cells.

PNH clone analysis

A PNH clone was defined as the percentage of GPI-APs deficient cells (granulocytes, RBCs, platelets, reticulocytes) determined by flow-cytometry. A 5-color cocktail to identify granulocytes-monocytes affected by GPI-linked protein deficiency was employed to determine PNH granulocytes clone size (Fig. 1 and Supplemental Fig. 1). The addition of thiazole orange to our established RBCs flow assay allowed for simultaneous detection of the RBCs/reticulocytes PNH type and size (Supplemental Fig. 1). We have also developed a flow cytometric analysis of platelets to reproducibly measure PNH platelet clone size (Fig. 1B and Supplemental Fig. 2). Complete gating strategy for both the RBCs/reticulocytes and platelets test is outlined in Supplemental Fig. 1. Antibody clones and conjugates are listed in Supplemental Table II.

Figure 1. Flow cytometric analysis of PNH patients utilizing new and established color combinations for multiple mature hematopoietic cell lineages.

(A) Three color flow cytometric analysis with CD59, CD235a and thiazole orange on erythrocytes and two color flow cytometric analysis with CD41 and CD55 on platelets for 4 representative patients and 1 control samples. The gating strategy can delineate Type II and Type III reticulocytes and red blood cells (RBCs). The pattern of Type II and Type III reticulocytes vs. RBCs is proportionated to one another per patient. (B) A comprehensive snapshot of multiple cells for PNH flow cytometric analysis (RBCs, Reticulocytes, Platelets, Granulocytes and Monocytes). One representative patient and one control sample are shown. RBC/Reticulocyte and platelet staining strategy are described above. A five-color strategy for analyzing granulocytes and monocytes was adopted from a protocol published by Sutherland et al, 2015.[11]

Terminology and definitions

T-II/III predominant = patients with PNH with a predominant amount of type II/III RBCs.

Mean sensitive fraction (Δ) was defined as difference in a given PNH population vs. PNH granulocyte clone size (see also Supplemental material). In particular:

$$\text{ΔRBC}\text{=}\text{\%PNH}\text{Granulocytes}\text{−}\text{\%PNH}\text{RBCs.}$$

$$\text{Δreticulocytes}\text{=}\text{\%PNH}\text{Granulocytes}\text{−}\text{\%PNH}\text{reticulocytes.}$$

$$\text{Δplatelets}\text{=}\text{\%PNH}\text{Granulocytes}\text{−}\text{\%PNH}\text{platelets.}$$

Detection of PNH cell fractions

Results were collected from experiments performed using BD FACSVerse while NFC500 or XL-MCL (Beckman Coulter) cytometers were used to validate these results. FlowJo software was used to analyze flow cytometry data.

Erythroid lineage.

Three color analysis of CD235a, CD59 and thiazole orange was used to identify both PNH RBCs and reticulocytes. In brief, 5μL of whole blood was added to a solution containing thiazole orange (0.01mg/mL) in 1X PBS and 2% FBS according to manufacturers’ instruction (BD Biosciences). Cells were first suspended and 100μL was transferred to a fresh tube. Previous experiments of antibody titration were performed to optimize the ratio of antibody versus whole blood. After that 2.5μL each of anti-CD235a-BV421 and anti-CD59-APC antibodies was added to the fresh tube and incubated in the dark for 30 minutes at room temperature. Cells were washed twice in 1X PBS and 2% FBS and suspended in the same buffer before acquisition. Acquisition was done on a BD FACSVerse cytometer.

Platelets.

The interval between time of blood draw and analysis was within 24 hours in all patients analyzed. Two-color analysis of CD41 and CD55 was used to identify GPI-deficient platelets; 1 mL of whole blood collected in EDTA tubes was spun at 100 x g for 10 minutes. The platelet-rich plasma (PRP) was collected and placed into a fresh tube. The PRP was spun at 900 x g for 10 minutes to enrich for platelets in the pellet. Cells were suspended in 200μL of staining buffer (1xPBS + 2% FBS) and 2μL each of CD41-FITC and CD55-APC was added to the tube. Incubation at room temperature was carried out for 20 minutes followed by washing prior to being analyzed. Both the incubation at room temperature, as well as minimal agitation/washing and the few pipetting steps, were employed to avoid potential activation, damage and aggregation of the platelets. A pilot testing experiment was performed in duplicate samples from the same patients (N=2 PNH samples and N=2 negative controls) within the same time point to ensure reproducibility and lack of experimental variability. Moreover, we internally validated this method studying 38 patients with detectable PNH clones of all size showing linear correlation with PNH granulocytes and monocytes (Supplemental Fig. 2).

Granulocytes.

Five-color analysis was conducted according to Sutherland et al 2015. (Alternate protocol 2: “Simultaneous High-Sensitivity Detection of PNH Neutrophils and Monocytes Using FLAER and CD157-Based 5-Color Assay”) [11, 12]. Both PNH neutrophils (representing granulocytes) and monocytes were detected using this method, although only the PNH neutrophils are discussed in the current manuscript.

Statistical analysis

Statistical analyses (descriptive statistics, t-test, regressions) were performed by comparisons of means using GraphPad PRISM 7.0 (GraphPad Software, Inc., LaJolla, CA, USA). A two-sided p-value of less than 0.05 was considered statistically significant.

RESULTS

Lineage-specific contribution of PNH clone in naive patients.

We designed a flow cytometry panel for assessment of GPI-APs deficient peripheral blood lineages, including RBCs, granulocytes, reticulocytes and platelets (see materials and methods section and Fig. 1 as an example).

First, we characterized mean population sizes of each lineage for patients who were naïve to eculizumab (Fig. 2A, N=57). The most notable difference was found between the percentages of granulocytes and RBCs. For each PNH cell population, we established a mean sensitive fraction, defined as the difference between granulocytes percentage and RBCs, platelet or reticulocyte percentages (ΔRBC, Δplatelets, Δreticulocytes). This measure was particularly high for RBCs (37±3% Fig. 2B) whereas mean sensitive fractions for reticulocytes and platelets were 8±5% and 11±5%, respectively (Fig. 2B).

Figure 2. Comparison of multiple PNH blood cells pre and post eculizumab.

(A) PNH fraction (%) was analysed in our cohort before eculizumab treatment [N=57 for granulocytes and Red Blood Cells (RBCs); N=15 for reticulocytes and platelets].(B) Proportion of PNH cell types that are sensitive to hemolysis as compared to mean PNH granulocyte (%) pre-eculizumab [N=57 for granulocytes, RBCs (37±3%) and type II (35±5%)and type III (41±3%) RBC; N=15 for reticulocytes (8±5%) and platelets (11±5%)]. (C) Cross-sectional analysis of granulocyte and RBC PNH fraction (%) in eculizumab-naïve (N=57) and eculizumab-treated patients (N=43) (median RBC clone size 20.5 [0.04–88.04] vs. 33.7 [2.7–99.5], respectively; p=.003); median granulocyte clone size 70% [22.4–99.72] vs. 88.9% [22–99.65]; p=.0325. (D) Cross-sectional analysis of reticulocyte and platelet PNH fraction (%) in eculizumab-naïve (N=15) and eculizumab-treated patients (N=21); median reticulocyte clone size 36.84% (26.61–86.65) vs. 91.42% (80.5–97.4), p= 0.0014; median platelet clone sizes 26.85% (7.96–56.6) vs. 59.75% (8.69–59.5), p=0.2571. (E) Cross-sectional analysis of RBC Type II and III dominant PNH fraction (%) in eculizumab-naïve (N=57) and eculizumab-treated patients (N=43) demonstrating a shift up in TIII RBCs in the treated cohort (16±2% vs. 3±5%; p=.0001) while no differences were seen in the TII RBC clone size. (F) Serial analysis (N=22) showing a shift up in the PNH fraction in eculizumab treated patients for RBCs (32±5% naive vs. 44±5% treated, p=.0065) and granulocytes (69±5% naive vs. 82%±4% treated, p=.0145). (G) Serial data of PNH fraction for T-II (6±3% pre vs. 10±3% post, p=0.0303) and T-III (15±3% pre vs. 23±4% post, p=.0145) RBCs in eculizumab-naïve and eculizumab-treated patients (N=22).

Next, we analyzed separately the difference between patients with predominant T-II or T-III RBCs populations (Fig. 2B). The group T-II-dominant had a lower ΔRBC in comparison to the T-III-dominant, although it did not reach any statistical significance. (35.1 vs. 41.01%, p=0.451)

Effects of eculizumab.

In patients who received eculizumab treatment, RBC clone size was significantly higher than in treatment naive patients (Fig. 2C, median RBC clone 33.7 [2.7–99.5, N=43] vs. 20.5 [0.04–88.04, N=57], respectively, p=.003). Similarly, the granulocyte clone size also appeared to be higher in treated patients (median 88.9% [22–99.65, N=43] in treated vs. 70% [22.4–99.72, N=57] in treatment naïve cohort; p=.0325). Additionally, also platelet and reticulocyte clone size was higher in patients receiving anti-complement therapy (Fig. 2D). Indeed, for reticulocytes we detected a median clone size for naive patients of 36.84% (26.61–86.65) vs. 91.42% (80.5–97.4) (p= 0.0014) in the setting of eculizumab therapy. Similarly, PNH platelet median clone sizes was 26.85% (7.96–56.6) vs. 59.75% (8.69–59.5), respectively in untreated and treated patients (p=0.2571). Upon further analysis of T-II and T-III populations, in individuals with predominant T-III RBC clones, we recorded a dramatic shift up in T-III RBCs in the treated cohort (16±2% vs. 3±5%; p=.0001; Fig. 2E). In contrast, the increase of T-II RBCs clone size in patients on eculizumab compared to the untreated cohort was less pronounced.

These data were further confirmed via analyses of samples retrieved from patients pre- and post-eculizumab initiation (N=22). Median time after eculizumab start was 1569 days (346–2956) while median time between sequential time-points was 960 (99– 2975). We registered a shift up in the PNH fraction post-eculizumab treatment for RBCs (28.095% [0.89–86.1] pre- vs. 33.695% [2.562–82.4] post, p=0.0065) and granulocytes (62.8% [22.4–99.61] pre- vs. 85.1% [22–99.5] post-eculizumab initiation, p=0.0145) (Fig. 2F). Likewise, for a subset of patients (N=12) we acquired serial data for T-II and T-III RBCs pre- and post-initiation of eculizumab (Fig. 2G). The findings revealed that RBC T-II clone size significantly increased after treatment (median 7.82% [0.62–23] pre- vs. 10.48% [0.6–32.14] post eculizumab initiation, p=.0303). Similarly, the RBC T-III clone size expanded on eculizumab therapy (median 13.99% [0.62–30.84] pre- vs. 22.48% [2.53–42.64] post eculizumab initiation, p=.0145). Of note, 4 patients with T-III RBC population did not display this increase after therapy.

Relationship between PNH blood lineages.

The cross-sectional analysis of the relationships among the different PNH clone types showed an exponential correlation between PNH RBC and granulocyte clone sizes (Fig. 3A, r2=0.42). In contrast, there was a linear correlation between PNH granulocytes vs. platelets (0.86±.18; r2=.64; N=15, Fig. 3B) and granulocytes vs. reticulocytes (0.6±.15; r2=.56; Fig. 3C).

Figure 3. Correlation of multiple PNH blood cells with the PNH granulocyte fraction.

(A) PNH Red Blood Cells (RBCs) correlation with PNH granulocytes in eculizumab-naïve patients via non-linear regression. r² = 0.421. (N=44) (B) PNH platelets correlation with PNH granulocytes pre-eculizumab via linear regression. Slope = 0.648 +/− 0.2742. r² = 0.3006. (N=15). (C) PNH reticulocytes correlation with PNH granulocytes pre-eculizumab via linear regression. Slope = 0.8627 +/− 0.1782. r² = 0.6432. (N=15). (D) PNH Type II and III RBCs correlation with PNH granulocytes pre-eculizumab. PNH Type II RBCs best fit was found using non-linear regression. r² = 0.2362. PNH Type III RBCs best fit was found using linear regression. r² = 0.01969. (N=44)

A logarithmic relationship also occurred between PNH granulocytes and T-II RBCs (r2=.24), but not with T-III, likely dependent on hemolysis (Fig. 3D). Moreover, in patients receiving treatment, the correlation of total RBCs with PNH granulocyte population was less robust (r2=.193), (Supplemental Fig. 3).

Relationship with thrombosis.

PNH-related thrombosis was noted for 17 patients of which 42% were diagnosed with Budd-Chiari syndrome. Thrombosis incidence was similar in treated patients with predominant T-II vs. T-III PNH RBCs (23% vs. 18%, p=0.2644), with no differences between pPNH vs. sPNH. When compared to non-thrombotic patients (N=54), patients with a history of thrombotic events (TE) had a similar size of PNH granulocyte clones (84.4% [22–99.02] vs. 84.45% [8.11–99.72] for TE and no TE, respectively; p= 0.649) and PNH RBCs (22.87% [1.6–85.04] vs. 34.98% [0.04–99.5] for TE and no TE, p=0.3617) on eculizumab treatment. Platelet clone size was significantly lower in thrombotic patients (31.05% [8.69–61.9, N=5] vs. 75.7% [7.96–99.8, N=18]; p=0.001), but no difference in absolute platelet counts was found (p= 0.8491). Moreover, no differences were found in terms of mean sensitive fraction ΔRBC (30% vs. 40%, p=0.270) or LDH levels (median 588 U/L [248–1971] vs. 847 U/L [199–3885], p=0.401) for TE and no TE patients respectively.

DISCUSSION

In this study, we designed a flow cytometry panel for investigation of GPI-APs deficient peripheral blood RBCs, granulocytes, platelets and reticulocytes perturbations in PNH before and after anti-complement treatment to better understand vulnerabilities of the individual blood cell types to complement-mediated destruction.

We demonstrated that, in addition to PNH RBCs, also reticulocytes and platelets display various degree of resistance to intravascular complement-mediated lysis. To better highlight the sensitivity of each blood cell lineage affected by the PNH defect to complement-mediated destruction, we took advantage of the analysis of patients undergoing treatment with eculizumab. We showed that the gradient between the proportion of PNH granulocytes and T-III RBCs was higher compared to that of T-II PNH RBCs, which was less improved by complement blockade, as shown by Richards et al [13], being TIII RBCs the subtype most sensitive to complement-mediated lysis. Moreover, when we analyzed the effects of C5 blockade on clone dynamics, we found that PNH platelet and reticulocyte sizes increased, indicating that their relative resistance to complement-mediated destruction is more similar to that of PNH granulocytes than to that of PNH RBCs, as shown by in vitro complement-mediated lysis experiments in the pre-eculizumab and pre-cytofluorimetric era.[7]. Of note, complement-mediated destruction is a stochastic phenomenon, thus cells with longer life-span have much higher chance to undergo through this phenomenon.[14]

Unlike PNH RBCs, granulocytes have been used to quantitate the PNH clone size because of their resistance to complement-mediated lysis, attributed to the presence of a non-GPI based complement inhibitor such as CD46.[15] Although we observed a relative increase in PNH granulocyte clone size with the commencement of eculizumab therapy, possibly suggesting that the resistance is not absolute, this finding is more likely to be explicable within the scenario of expansion of the PNH clone, privileged in replacing the bone marrow hematopoiesis because of the contemporary presence of AA in at least half of the cases.

Historically, the analysis of PNH platelets has been challenging and no conclusive data are present in the literature as to the grade of resistance of platelets to hemolysis.[16–19]. No standardized methods are used to analyze this particular PNH population and different studies have reported various techniques trying to overcome common pitfalls of platelets analysis: size of the cells, resolution of singlets/doublets, aspecific binding and activation due to steps required by flow cytometry protocols (vortexing, incubation on ice, centrifugation, pipetting). [19] For instance, using CD59 and CD55 as expression markers, the fragments of lysed RBCs can mimic platelets with the similar physical parameters of RBCs and, to a lesser extent, confound the discrimination of the two cell subsets. Therefore, we adopted an analytic strategy for PNH platelets enumeration by CD41 and CD55 gating to eliminate the bias of eventual contaminating RBCs ghosts, contained in the platelets-rich-plasma. The strategy of gating strictly on singlet platelets helped us to avoid doublets between PNH and normal platelets, a phenomenon which may lead to underestimation of PNH platelet counts in circulation. As indicated in our data, the fraction of PNH platelets correlates with the percentage of PNH granulocytes as previously shown. [16, 19] However, unlike previously reported [8, 20] in our cohort we did not find a correlation between PNH granulocyte clone size and thrombosis occurrence, likely because all patients had large clones (with a loss of linear relationship precluding differentiation) and the analysis was done under eculizumab therapy. Surprisingly, in our study PNH platelets were present at lower percentages in thrombotic patients. This result was in contrast to that reported by Araten et al.[19] who found a similar size of PNH platelets (and of PNH granulocytes and RBCs) in patients with PNH, irrespectively of thrombosis. Indeed, the absolute platelet counts did not show any difference in line with the aforementioned study. We hypothesize that in thrombotic patients, the lower PNH platelet counts are due to their selective destruction.

Similarly to PNH platelets, our data demonstrated that PNH reticulocytes may be useful to monitor PNH clone size to better resolve the nuances of RBC production. Correlation between the size of PNH reticulocyte and PNH granulocyte populations has been previously reported [21, 22] and Hochsmann et al [23] have suggested the superiority of PNH reticulocytes analysis over RBCs due to transfusion and hemolysis independency, especially for patients with small PNH clones. [24] In our data, PNH reticulocytes, similarly to PNH RBCs, increased in patients treated with eculizumab. However, it is well established that eculizumab treatment does not normalize the combined reticulocyte count [9, 25], although hemoglobin levels may normalize. One could theorize that this discrepancy may represent the general response to the iatrogenic extravascular hemolysis related to the C3-opsonization after the chronic C5 blockade.[26] Whether intra- or extravascular destruction is at play, analysis of normal and PNH reticulocytes may augment the yield of traditional PNH RBC and granulocyte flow cytometry in terms of etiology of anemia following eculizumab therapy.[4] In either case, the clonal dynamics in PNH may be influenced by the presence of AA, immunosuppressive therapy and additional somatic mutations. [27]

In conclusion, our phenotypic analysis of principal cellular subsets in PNH shows the complexity of their dynamics during unopposed complement lysis or under eculizumab. Future studies are warranted to explore the effects on PNH cell populations of novel and proximal complement inhibitors.