Intra- and interlaboratory variability of paroxysmal nocturnal hemoglobinuria testing by flow cytometry following the 2012 Practical Guidelines for high-sensitivity paroxysmal nocturnal hemoglobinuria testing

Authors


Correspondence to: Iuri Marinov, Institute of Hematology and Blood Transfusion, U nemocnice 1, 128 20, Prague 2, Czech Republic. E-mail: Iuri.Marinov@uhkt.cz

Abstract

Background

Sutherland et al. recently published the Practical Guidelines for high-sensitivity detection of paroxysmal nocturnal hemoglobinuria (PNH) clones by flow cytometry (FCM), containing concise protocols for PNH testing.

Methods

Using this approach, we studied the intra- and interlaboratory variability observed in a multicenter study in which fresh blood samples containing three clinically relevant PNH clone sizes within the granulocytic, monocytic, and red blood cell (RBC) populations were shipped to each participating center.

Results

Coefficients of variation (CVs) for precision/reproducibility analysis ranged from 0.01%/0.02% to 0.48%/0.45% (big clone), from 0.69%/1.52% to 4.24%/5.80% (small-intermediate clone), from 1.47%/3.91% to 15.01% /17.83% (minor clone) for PNH white blood cells (WBCs) and from 0.24%/0.48% to 1.76%/1.83% (big clone), from 0.80%/1.14% to 2.39%/4.45% (small-intermediate clone), from 1.09%/3.36% to 10.54%/10.23% (minor clone) for PNH RBCs, respectively. Linear regression analysis showed excellent performance correlation between centers (r > 0.99), Wilcoxon rank test revealed no statistically significant differences for PNH granulocytes, monocytes, and RBCs (P > 0.05%), Bland–Altman analysis demonstrated good performance agreement for all target PNH clones (mean bias ranging from −1.47 to 0.71).

Conclusion

Our results demonstrate very good intra- and interlaboratory performance characteristics for both precision and reproducibility analyses and excellent correlation and agreement between centers for all target PNH clone sizes. Our data confirm the reliability and robustness of the recently published Practical Guidelines approach for high sensitivity PNH testing by flow cytometry and suggest that such an approach represents an excellent basis for standardization of PNH testing by flow cytometry. © 2013 International Clinical Cytometry Society

Paroxysmal nocturnal hemoglobinuria (PNH) is a rare acquired clonal stem cell disorder, which is characterized by proliferation of stem cells lacking proteins linked to the cell membrane via glycophosphatidylinositol (GPI) anchors due to a somatic mutation of the phosphatidylinositolglycan class A (PIG-A) gene [1-4]. The disease is associated with fatigue, abdominal pain, dysphagia, poor physical functioning due to haemolysis, and elevated risk of life threatening complications related to chronic renal failure, pulmonary hypertension, venous/arterial thrombosis at unusual sites, and/or bone marrow aplasia [5, 6].

Characterization of PNH cells by flow cytometry (FCM) was first introduced in 1990 [7] and since 1996 FCM has become the method of choice for identifying GPI deficient cells [8-10]. PNH testing is becoming increasingly important for disease monitoring in terms of progression, regression, remission, or response to therapy with humanized monoclonal antibody (MoAb) against the C5 complement protein [11-13]. The implementation of high sensitivity assays enables additional screening for minor PNH clones (<1.0%) and/or cells with PNH phenotype (<0.1%) in patients with aplastic anemia, myelodysplastic syndrome, and other bone marrow failures [14-17].

In 2010, the ICCS published the first consensus guidelines for the diagnosis and monitoring of PNH and related disorders [18]. The document underlines the clinical significance of PNH testing by FCM, determines the clinical indications for PNH analysis, and provides suggested approaches for routine and high-resolution analysis of PNH white blood cells (WBCs) (granulocytes and monocytes) and red blood cells (RBCs), including antibody and panel recommendations for 3, 4, 5, and 6-color instruments. However, standardized procedures utilizing specific cocktails and gating strategies were not identified and laboratories continued to experience problems with this assay. Recently, Sutherland et al. published a supplementary document to the ICCS Guidelines for PNH testing containing concise practical protocols for high sensitivity detection and monitoring of PNH clones by FCM [19]. This supplementary document provides specific additional information concerning selection of optimal antibody conjugates for both GPI-specific and lineage-gating MoAbs, addresses specific reagent cocktails, provides concise practical protocols and detailed analytic strategies for high sensitivity detection of PNH RBCs and WBCs. The extensive comparison of a large number of gating and GPI-specific reagents as well as most appropriate conjugates applicable for both Becton Dickinson (BD) and Beckman Coulter (BC) platforms provides a great deal of additional information not present in the original ICCS Guidelines.

At present, no single specific combination of markers has been designated to rule out the others, despite the tremendous progress in this direction, PNH testing by FCM is still not fully standardized and interlaboratory variability reported from quality proficiency assessment schemes is considerable. During the last few months, we implemented a harmonized approach for PNH testing based on the recent Practical Guidelines [19] in several clinical laboratories within central Europe, one of the main rationales being the study of intra- and interlaboratory variability of this approach. The study was preceded by an on-site practical workshop and consecutive validation steps on normal samples within a local project for harmonization of PNH testing following the Practical Guidelines [19], organized and sponsored by Alexion Pharma International. Participating centers were invited to use a common standardized protocol comprising MoAb clones, conjugates, reagent titration, instrument set up, acquisition, analysis, and data reporting. Fresh samples representing different, clinically relevant WBC and RBC PNH clone sizes were shipped at room temperature to participating centers. Sample preparation, acquisition, and analysis for all participants started within 24 h of blood collection. Results were sent for statistical proceeding in the reference center.

MATERIALS AND METHODS

PNH Clone Sources

Following informed consent, peripheral blood (PB) from a patient with PNH was appropriately diluted with compatible blood of a healthy donor to obtain three target PNH clone sizes for granulocytes, monocytes, and RBCs: big (>20%), small to intermediate (1–20%), and minor (<1%) PNH clone, respectively.

2-Tube/4 Color WBC Analysis

The main features of the approach are listed in Table 1. Briefly, 100 μL of well mixed PB was incubated with the appropriate amount of pre-titrated MoAbs for 30 min in the dark at 4–8°C. RBCs were lysed according to local validated procedures, cells were washed twice in phosphate buffered saline with bovine serum albumin (PBS/PBA) and resuspended in 0.5–1.0 mL of isotonic solution prior to acquisition. WBCs were gated on forward scatter channel (FSC)/side scatter channel (SSC)/CD45/CD15 for granulocytes and FSC/SSC/CD45/CD64 for monocytes, 50,000 granulocytes were then acquired and PNH clone assessment was determined by FLAER/CD24 deficiency for granulocytes and FLAER/CD14 deficiency for monocytes.

Table 1. Main Characteristics of Approaches and Reagents Used in the Study
MethodTarget populationGating strategy reagentInformative reagent
1-tube/2 colorRed blood cellslogFSC/logSSC/CD235aCD59
2-tubes/4 colorGranulocytesFSC/SSC/CD15/CD45FLAER/CD24
MonocytesFSC/SSC/CD64/CD45FLAER/CD14

1-Tube/2 Color RBC Analysis

The main features of the approach are listed in Table 1. Briefly, 50 μL of 1:100 diluted PB (10 μL PB + 1 mL PBS) was incubated with the appropriate amount of pre-titrated MoAb for 30 min in the dark at room temperature. RBCs were washed twice with PBS and resuspended in 0.5–1.0 mL of PBS. Prior to acquisition the cells were “racked” and 50,000 RBCs gated on logFSC/logSSC/CD235a were acquired within 15 min. PNH clones were determined based on CD59 negativity for Type III RBCs or partial positivity for Type II RBCs.

Monoclonal Antibodies

The specific reagents and MoAbs used in the study are listed in Table 2.

Table 2. Specific Reagents and Fluorochromes Used in the Study
MoabClone-fluorochrome
Becton Dickinson userBeckman Coulter user
FLAER
CD14Mϕ P9 (PE)RMO52 (PE)
CD15HI98 (APC)80H5 (PC5)
CD24ML5 (PE)ALB9 (PE)
CD452D1 (PerCP-Cy5.5)J.33 (PC7)
CD59MEM43 (PE)MEM43 (PE)
CD6410.1 (APC)22 (PC5)
CD235aKC16 (FITC)KC16 (FITC)

Flow Cytometry

Acquisition and analysis was performed on a BD FACSCanto™ equipped with BD FACSDiva™ 5.0 software and BC FC500™ equipped with CXP2.2™ software.

Statistical Analysis

Precision and reproducibility analysis (performance characteristics) were performed by replicate analysis, results were reported as mean, standard deviation (SD), and coefficients of variation (CV, %). Correlation of results was performed by linear regression analysis and Pearson's correlation coefficient (r) significant at level 0.01. Comparison between methods was determined by the Wilcoxon signed rank test for paired samples significant at level 0.05. Performance agreement was evaluated by Bland and Altman analysis of the relationship between the differences and the mean of differences reported as mean bias (mean of the differences ± 2 SD). Mean bias equal to zero shows absolute performance agreement.

RESULTS

For precision analysis, each center performed three replicate determinations of the big (>20%), small to intermediate (1–20%), and minor (<1%) PNH clones within the granulocyte, monocytes, and RBC populations. For the big PNH clone, intralaboratory CVs/means varied from 0.01% to 0.18%/98.51% to 99.73% (interlaboratory CV = 0.28%) for granulocytes, from 0.02% to 0.48%/95.82% to 99.67% (interlaboratory CV = 1.12%) for monocytes, from 0.32% to 1.76%/46.76% to 48.88% (interlaboratory CV = 2.10%) for Type III RBCs, and from 0.24% to 1.69%/47.62% to 49.57% (interlaboratory CV = 2.0%) for Type II+III RBCs. For the small to intermediate PNH clone, intralaboratory CVs/means varied from 0.69% to 1.79%/6.27% to 7.95% (interlaboratory CV = 6.42%) for granulocytes, from 0.09% to 4.24%/8.52% to 9.85% (interlaboratory CV = 4.25%) for monocytes, from 0.80% to 2.39%/10.10% to 11.55% (interlaboratory CV = 4.57%) for Type III RBCs, and from 0.81% to 2.13%/10.40% to 11.64% (interlaboratory CV = 3.67%) for Type II+III RBCs. For the minor PNH clone, intralaboratory CVs/means varied from 2.43% to 15.1%/0.35% to 0.62% (interlaboratory CV = 14.35%) for granulocytes, from 1.57% to 14.27%/0.69% to 1.20% (interlaboratory CV = 15.76%) for monocytes, from 1.99% to 10.54%/0.40% to 0.47% (interlaboratory CV = 4.55%) for Type III RBCs, and from 1.09% to 9.67%/0.42% to 0.49% (interlaboratory CV = 4.68%) for Type II+III RBCs (Table 3). For reproducibility analysis, each center performed five consecutive determinations for all target PNH clone sizes within 24 h. For the big PNH clone, intralaboratory CVs/means varied from 0.02% to 0.29%/99.73% to 98.83% (interlaboratory CV = 0.28%) for granulocytes, from 0.05% to 0.45%/95.91% to 99.58% (interlaboratory CV = 1.12%) for monocytes, from 0.48% to 1.83%/47.23% to 50.62% (interlaboratory CV = 2.10%) for Type III RBCs, and from 0.50% to 1.91%/48.12% to 51.40% (interlaboratory CV = 2.0%) for Type II+III RBCs. For the small to intermediate PNH clone, intralaboratory CVs/means varied from 1.52% to 5.34%/5.68% to 8.05% (interlaboratory CV = 8.52%) for granulocytes, from 2.37% to 5.80%/8.32% to 10.12% (interlaboratory CV = 5.49%) for monocytes, from 1.23% to 4.41%/10.34% to 11.28% (interlaboratory CV = 3.27%) for the Type III RBCs, and from 1.14% to 4.45%/10.66% to 11.33% (interlaboratory CV = 2.65%) for Type II+III RBCs. For the minor PNH clone, intralaboratory CVs/means varied from 3.91% to 10.33%/0.33% to 0.63% (interlaboratory CV = 16.47%) for granulocytes, from 5.47% to 17.83%/0.85% to 1.23% (interlaboratory CV = 12.41%) for monocytes, from 3.36% to 9.04%/0.42% to 0.46% (interlaboratory CV = 2.27%) for Type III RBCs, and from 3.46% to 10.23%/0.44% to 0.46% (interlaboratory CV = 2.25%) for Type II+III RBCs (Table 3). For correlation study, linear regression analysis of results obtained from the replicate determinations in time for WBCs (both granulocytes and monocytes) and RBCs (both Type III and Type II+III) showed: r = 0.99, slope 1.02, intercept −0.15 (WBCs) and r = 0.99, slope 0.98, intercept −0.27 (RBCs) for reference center vs. Center 1, r = 0.99, slope 1.00, intercept −0.37 (WBCs) and r = 0.99, slope 0.98, intercept −0.11 (RBCs) for reference center vs. Center 2, r = 0.99, slope 1.00, intercept 0.20 (WBCs), and r = 0.99, slope 0.93, intercept −0.13 (RBCs) for reference center vs. Center 3, finally r = 0.99, slope 1.00, intercept −0.47 (WBCs) and r = 0.99, slope 0.96, intercept 0.04 (RBCs) for reference center vs. Center 4 (Figs. 1 and 2). The Wilcoxon rank test for comparative analysis of median results from the precision analysis between centers showed: P = 0.5 for granulocytes, P = 1.0 for monocytes, P = 0.25 for Type III RBCs, and P = 0.25 for Type II+III RBCs (reference center vs. Center 1), P = 1.0 for granulocytes, P = 0.5 for monocytes, P = 0.5 for Type III RBCs, and P = 0.5 for Type II+III RBCs (reference center vs. Center 2), P = 0.5 for granulocytes, P = 0.5 for monocytes, P = 0.25 for Type III RBCs, and P = 0.25 for Type II+III RBCs (reference center vs. Center 3) and P = 1.0 for granulocytes, P = 0.75 for monocytes, P = 0.5 for Type III RBCs, and P = 0.5 for Type II+III RBCs (reference center vs. Center 4). Bland and Altman analysis of the relationship between the differences and the mean of differences of results from the replicate assays in time for WBCs (both granulocytes and monocytes) and RBCs (both Type III and Type II+III) from each center showed: mean bias 0.71 for WBCs and −0.55 for RBCs (reference center vs. Center 1), mean bias −0.18 for WBCs and −0.34 for RBCs (reference center vs. Center 2), mean bias 0.40 for WBCs and −1.47 for RBCs (reference center vs. Center 3), and mean bias −0.30 for WBCs and −0.72 for RBCs (reference center vs. Center 4), (Figs. 3 and 4).3

Figure 1.

Correlation analysis of WBCs (granulocytes and monocytes) between centers: reference center vs. Center 1 (a), reference center vs. Center 2 (b), reference center vs. Center 3 (c), reference center vs. Center 4 (d).

Figure 2.

Correlation analysis of RBCs (Type II and Type II+III) between centers: reference center vs. Center 1 (a), reference center vs. Center 2 (b), reference center vs. Center 3 (c), reference center vs. Center 4 (d).

Figure 3.

Bland–Altman agreement analysis for WBCs (granulocytes and monocytes): reference center vs. Center 1 (a), reference center vs. Center 2 (b), reference center vs. Center 3 (c), reference center vs. Center 4 (d).

Figure 4.

Bland–Altman agreement analysis for RBCs (Type II and II+III): reference center vs. Center 1 (a), reference center vs. Center 2 (b), reference center vs. Center 3 (c), reference center vs. Center 4 (d).

Table 3. Results from Precision/Reproducibility Analysis from Individual Centers
Clone sizePNH cloneCenterNMeanSDCV (%) intralaboratory
>20%GrRef. center3/599.65/99.720.01/0.030.01/0.03
Center 13/598.51/98.830.14/0.290.14/0.29
Center 23/599.73/99.730.01/0.040.01/0.04
Center 33/599.7/99.440.03/0.140.03/0.14
Center 43/599.49/99.680.18/0.020.18/0.02
MoRef. center3/599.67/99.580.02/0.050.02/0.05
Center 13/595.82/95.910.46/0.430.48/0.45
Center 23/599.18/99.190.09/0.140.09/0.14
Center 33/598.49/98.390.33/0.250.33/0.26
Center 43/599.38/99.560.11/0.070.11/0.07
RBC IIIRef. center3/546.76/47.230.27/0.710.58/1.51
Center 13/548.12/47.230.21/0.710.44/1.51
Center 23/547.6/47.820.15/0.870.32/1.83
Center 33/548.88/50.620.33/0.440.34/0.87
Center 43/548.73/49.110.86/0.241.76/0.48
RBC II+IIIRef. center3/547.62/48.120.25/0.680.53/1.42
Center 13/548.83/48.930.22/0.310.45/0.63
Center 23/548.52/48.750.11/0.930.24/1.91
Center 33/549.74/51.40.3/0.50.6/0.96
Center 43/549.57/49.930.84/0.251.69/0.5
1–20%GrRef. center3/56.78/6.480.05/0.10.69/1.52
Center 13/56.69/6.630.12/0.231.79/3.43
Center 23/56.56/6.340.09/0.191.41/3.03
Center 33/56.27/5.680.05/0.30.74/5.34
Center 43/57.95/8.050.13/0.181.59/2.2
MoRef. center3/58.9/8.450.01/0.490.09/5.8
Center 13/58.82/8.80.37/0.464.24/5.27
Center 23/59.85/10.120.19/0.241.64/2.37
Center 33/58.52/8.320.06/0.340.71/4.09
Center 43/59.25/8.80.37/0.263.95/2.99
RBC IIIRef. center3/510.1/10.340.11/0.291.13/2.82
Center 13/510.93/11.280.17/0.181.53/1.63
Center 23/510.48/10.840.08/0.180.08/1.7
Center 33/511.55/11.540.12/0.511.05/4.41
Center 43/510.1/10.810.24/0.132.39/1.23
RBC II+IIIRef. center3/510.45/10.660.14/0.271.35/2.54
Center 13/511.03/11.330.12/0.171.06/1.53
Center 23/510.67/11.010.09/0.190.81/1.73
Center 33/511.64/11.60.12/0.521.05/4.45
Center 43/510.4/10.890.22/0.122.13/1.14
< 1%GrRef. center3/50.62/0.630.02/0.053.78/7.31
Center 13/50.51/0.480.01/,.032.43/6.99
Center 23/50.6/0.580.09/0.0415.01/7.08
Center 33/50.54/0.530.03/0.025.45/3.91
Center 43/50.25/0.230.02/0.027.64/10.33
MoRef. center3/50.62/0.850.02/0.053.78/5.47
Center 13/50.79/0.80.01/0.141.57/17.83
Center 23/51.04/0.90.15/0.1114.27/12.6
Center 33/50.94/0.960.02/0.12.0/10.63
Center 43/51.2/1.230.05/0.194.54/15.4
RBC IIIRef. center3/50.42/0.850.02/0.054.93/5.47
Center 13/50.41/0.420.04/0.0210.54/5.22
Center 23/50.41/0.440.01/0.011.99/3.36
Center 33/50.47/0.460.02/0.034.6/5.55
Center 43/50.4/0.420.02/0.044.08/9.04
RBC II+IIIRef. center3/50.45/0.460.02/0.033.63/6.18
Center 13/50.43/0.470.04/0.029.67/3.65
Center 23/50.43/0.460.0/0.021.09/3.46
Center 33/50.49/0.480.02/0.023.45/4.82
Center 43/50.42/0.440.02/0.054.01/10.23

DISCUSSION

The ICCS Guidelines [18] identified a number of reagent combinations that could be used for the high-sensitivity detection of PNH on a variety of instrument platforms equipped with 3, 4, 5, and 6 photomultiplier tube detectors (PMTs). During the last year, we have successively validated most of these approaches for PNH WBC and RBC analysis as described in the ICCS Guidelines [18, 20]. The consecutive study proved good correlation and agreement between individual protocols (submitted for publication). Identification of PNH granulocytes is of critical importance because of its diagnostic significance [5, 18, 21]. A relative disadvantage of the 2-tube/3 color approach for PNH granulocyte and monocyte analysis is the lack of CD45 gating reagent, which in some cases hampers debris exclusion for optimal live-gate setting. The one-tube/4 color combination of FLAER/CD14/CD15/CD33 or CD64 is missing a second GPI-specific marker for granulocytes as well as CD45 and optimal debris exclusion is insufficient for bona fide high sensitivity testing. The one-tube/5 color approach (FLAER/CD14/CD15/CD24/CD45) is missing a lineage-specific gating reagent for monocytes. The single-tube/6 color approach enables optimal gating of target populations and can include all necessary GPI-specific markers, is less time consuming and more cost effective (1-tube, 6 MoAbs). However, it is more demanding with respect to optimizing instrumentation set-up, optimal compensation setting, protocol set-up, and is more limited regarding the choice of commercially available bright fluorochrome conjugates for both gating and GPI-specific reagents. A further disadvantage of the single-tube/6 color approach is that not all clinical laboratories have instruments equipped with at least six PMTs. All these relative disadvantages represent the rationale for the 2-tube/4 color protocol for high sensitivity detection and monitoring of PNH clones by FCM as described in the recently published Practical Guidelines [19]. In the present study, we studied the intra- and interlaboratory variability of this approach analyzing the performance characteristics for three clinically relevant PNH clone sizes within granulocytic, monocytic, and red blood cell populations. Most of participating centers had limited or no previous experience in PNH testing. Following a practical workshop and consecutive validation steps on normal PB, we performed a single ring study using a harmonized approach adapted to local instrumentation platforms. The results of the study proved very good intra- and interlaboratory performance characteristics for both precision and reproducibility analysis for all clinically relevant WBC and RBC PNH clone sizes, excellent performance correlation, statistically no significant differences at level 0.05, and very good agreement (mean bias) between centers. Besides proper validation of PNH assays, interlaboratory comparison and proficiency surveys play an essential role for systematic improvement of confidence levels in the detection of PNH clones [17]. In this context, our ring study was the starting point of a regional central European Quality Assessment scheme, consisting of the exchange of normal and pathological samples at regular intervals using fresh blood samples analyzed 24 h following collection and distribution at room temperature.

CONCLUSION

Up to present, no specific combination of markers for gating and analysis has been designated to rule out the others, each approach described in the ICCS Guidelines [18] has relative advantages and disadvantages. The results from our multicenter study confirmed the reliability and robustness of the recently published Practical Guidelines approach for high sensitivity PNH testing by FCM [19], which in our hands demonstrated low intra- and interlaboratory variability thus providing strong arguments for standardization.

Acknowledgments

The authors would like to thank Petra Palickova, Jana Pauerova and Pavla Pilar-Steinerova for assistance with the ring study organization and Alexion Pharma International for the financial and logistic support. Andrea Illingworth and Iuri Marinov have taken part in Alexion Pharmaceuticals PNH Diagnostic Advisory Board meetings.

Ancillary