Peripheral blood samples were collected for routine flow cytometry screening for PNH from 24 patients at Seoul St. Mary's Hospital, Seoul, Korea, between January 2010 and April 2012. Samples remaining after the screening test were aliquoted at 200 µL/tube. Two or three microtubes were stored immediately at −20℃ until molecular analysis. Diagnosis was confirmed and assigned according to the guidelines for the diagnosis and management of PNH [9]. At sampling, no patient was actively infected, and no evidence for hereditary bone marrow (BM) failure syndromes was found. All patients provided written informed consent for clinical and molecular analyses. The study protocol was approved by the Institutional Review Board of The Catholic University of Korea, Seoul, Korea (KC12RISE0422).

The patients were classified into two groups based on the presence of cytopenia at diagnosis of PNH: PNH with concomitant AA (PNH/AA, N=12) and classic PNH (N=12). Patients who met at least two of the three peripheral blood cytopenias (Hb level: ≤100 g/L, absolute neutrophil count [ANC]: 0.5–1.5×10⁹/L, and platelet count [PLT]: 20–100×10⁹/L) were classified as PNH/AA [10]. Patients with clinical evidence of intravascular hemolysis without any evidence of other BM failure were classified as classic PNH. There were significant differences between the groups in ANC and PLT. The number of patients who underwent previous immunosuppressive therapy and/or corticosteroid treatment was significantly higher in the PNH/AA than in the classic PNH group. The percentage of patients who were treated with eculizumab was higher in the classic PNH group, whereas allogeneic hematopoietic stem cell transplantation (HSCT) was performed exclusively in the PNA/AA group (Table 1).

GPI-AP deficient granulocytes, T lymphocytes, and RBCs were analyzed by flow cytometry using specific monoclonal antibody cocktails: fluorescent aerolysin reagent (FLAER)-Ax488/CD24-PE for granulocytes, FLAER-Ax488/CD3-PE for T lymphocytes, and CD55-FITC/CD59-PE for RBCs. The FLAER-based flow-cytometric assay is recommended to screen for GPI-AP deficient white blood cells (WBCs) because it has higher sensitivity than that based on other GPI-APs, including CD55, CD59, CD24, and CD14, and can detect small clone sizes in PNH as well as AA or myelodysplastic syndrome [11^,12].

To analyze GPI-AP deficient granulocytes and T lymphocytes, 5 µL of appropriate monoclonal antibody was added to 0.5×10⁶ WBCs and incubated for 20 minutes at 24℃. After RBCs were lysed using 2 mL of BD fluorescence activated cell sorter (FACS) Lysing Solution (BD Pharmingen, Franklin Lakes, NJ, USA), the cells were washed and analyzed on a FACSCanto II flow cytometer (BD Biosciences, San Jose, CA, USA), according to the manufacturer's recommendation. For each tube, a minimum of 50,000 events for granulocytes or T lymphocytes was routinely collected. To analyze GPI-AP deficient RBCs, 20 µL of whole blood was diluted with 3 mL of phosphate-buffered saline (PBS) followed by the addition of 50 µL to one tube, and 5 µL of CD59-FITC and CD55-PE to the second tube. The samples were incubated in the dark for 60 minutes at 24℃ and washed twice with PBS by centrifugation as required to optimize separation of Type I, II, and III GPI-AP deficient RBCs. After washing, the RBC pellet was re-suspended in 0.2 mL of PBS, and 50,000 RBCs were acquired in the list mode.

Granulocytes and lymphocytes were initially selected by forward scatter and side scatter. Each cell lineage was assessed for combined expression of FLAER/CD24 (granulocytes) and FLAER/CD3 (T lymphocytes). The limit of detection (LOD) using FLAER as the most discriminant marker combined with CD24 or CD3 was determined as 0.1%. However, because FLAER affinity seems restricted to certain types of GPI anchors, making it unsuitable for evaluation of RBCs [13], we used both CD55 and CD59 for expression analysis for GPI-AP-deficient RBCs with an LOD of 3%.

Granulocytes were isolated from EDTA-blood samples via standard density gradient separation using a mixture of sodium metrizoate and dextran 500 [14]. T lymphocytes were isolated from EDTA-blood samples using the EasySep Human T Cell Enrichment Kit (STEMCELL Technologies, Vancouver, Canada), according to the manufacturer's protocol. GPI-AP deficient granulocytes and T lymphocytes were additionally sorted into the FLAER(−)/CD24 and FLAER(−)/CD24 CD3(+) cell population, respectively, by flow cytometry using BD FACSAria II (BD Biosciences). Genomic DNA was extracted from the sorted GPI-AP deficient granulocytes and T lymphocytes using the QIAamp DNA Blood Mini Kit (Qiagen GmbH, Hilden, Germany).

Genomic DNA isolated from granulocytes and T lymphocytes was used as a template for amplifying individual exons of PIGA, including entire exons and the flanking intron sequences, as described previously [15]. PCR amplicons were bi-directionally sequenced using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) on an ABI PRISM 3130XL Genetic Analyzer (Applied Biosystems). The sequence electropherogram was analyzed using Sequencher Software 4.9 (Gene Codes, Ann Arbor, MI, USA). RefSeq ID: NM_002641.3 for PIGA was used for cDNA nucleotide numbering. Mutations were confirmed by sequencing via two or more independent PCRs. Genetic variations were compared with the public sequence databases including a database of single nucleotide polymorphisms (dbSNP, https://www.ncbi.nlm.nih.gov/snp/), Exome Aggregation Consortium (ExAc, http://exac.broadinstitute.org), 1000 Genomes Project Phase 3 database (1000 Genomes, http://phase3browser.1000genomes.org/index.html), and Korean Reference Genome Database (KRGDB, http://152.99.75.168/KRGDB/). Pathogenic mutations were predicted using PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2) [16] and PROVEAN (http://provean.jcvi.org) [17]. We used numerical peak height data of the Sanger sequencing electropherogram to measure semiquantitative PIGA mutation burden [18].

The TCR gene rearrangement assay was used to evaluate the clonality of GPI-AP deficient T lymphocytes. PCR was conducted on genomic DNA isolated from sorted GPI-AP deficient T lymphocytes using IdentiClone TCRB, TCRG, and TCRD Gene Clonality Assay (Invivoscribe Technologies, San Diego, CA, USA). The full set of reactions for TCR gene arrangements included one targeting TCRD (Vδ+Dδ+Jδ), one targeting TCRG 2.0 (Vγ+Jγ), and three targeting TCRB (TCRB Tube A: Vβ+J_β1/2; TCRB Tube B: Vβ+J_β2; TCRB Tube C: Dβ+J_β1/2). The PCR products were analyzed on an ABI 3130XL Genetic Analyzer with GeneMapper software (Applied Biosystems). Criteria for defining a positive peak in TCRD and TCRB were as follows: (1) a single prominent positive peak within the valid size range for individual primer pairs and (2) products within the valid size range and with peaks of at least three times the amplitude of the third largest peak in the polyclonal background. The TCRG interpretation algorithm provided by the manufacturer was implemented in the worksheet to define peaks as positive for clonality in TCRG 2.0: (1) the D(x) value of the suspected clonal peak, as calculated within the locked portion of the worksheet, is ≥0.0419; (2) the relative peak ratio of the suspected clonal peak, compared with the smallest adjacent peak, is ≥4; (3) the relative fluorescence unit (RFU) of the suspected clonal peak is ≥20% of the RFU of the highest peak in that sample. All assays were conducted in duplicate.

As the laboratory findings were not normally distributed as indicated by Kolmogorov-Smirnov and Shapiro-Wilk tests, continuous variables were summarized as median with range and were compared using the Mann–Whitney U test. Pearson's chi-square test was used to compare categorical variables. Spearman's correlation was determined to estimate the strength and direction of an association that existed between two continuous variables, and a line of best fit was drawn through the data of two variables in a scatterplot. All tests were two-tailed, and P<0.05 was considered significant. All statistical analyses were conducted using MedCalc 12.7.2 (MedCalc Software, Ostend, Belgium).

We excluded the patients treated with eculizumab from PNH clone size analysis according to cell lineage because eculizumab promotes the survival of GPI-AP-deficient RBCs compared with that of WBCs or lymphocytes [19]. The GPI-AP deficient cell fraction varied according to patient and cell types. The median GPI-AP deficient cell fraction was the highest in granulocytes followed by RBCs and T lymphocytes. GPI-AP-deficient RBCs were more numerous in classic PNH than in PNH/AA patients (P=0.018). The GPI-AP deficient granulocyte fraction correlated strongly with the clonal size of RBCs (r_s=0.69, P=0.015). The GPI-AP deficient RBC fraction correlated moderately with percentage of reticulocyte (r_s=0.565, P=0.004) (Fig. 1).

Nineteen (79.2%) of the 24 PNH patients carried PIGA mutations. Among them, 17 patients carried a single heterozygous mutation, and two patients harbored two heterozygous mutations (Table 2). Each PIGA mutation was exclusive and commonly occurred in exon 2 (48%, 10/21). The frequencies of deletion, duplication, and base substitution in the PIGA gene were 43% (9/21), 19% (4/21), and 38% (8/21), respectively. These mutations led to deleterious effects, including premature terminations (N=13), nonsense mutations (N=3), splicing errors (N=2), and missense mutations (N=3). Among the three missense mutations, p.Arg119Pro and p.Glu120Val were deemed “probably damaging” by PolyPhen-2, with scores of 0.999 and 1.000, respectively, and were expected to be “deleterious” by PROVEAN, with scores of −6.137 and −6.530, respectively. In contrast, p.Ala405Val was considered “benign” by PolyPhen-2 (score 0) and “neutral” by PROVEAN (−0.871). The frequency of these three variants was searched in public sequence databases. p.Ala405Val was registered as a “likely benign allele” (RS201361742) in dbSNP, and its variant frequency was reported as 0.0003/27 in ExAc and 0.0003/1 in 1000 Genomes, but it was not registered in KRGDB.

The incidence of PIGA mutation did not differ between PNH/AA and classic PNH. There were no significant differences between patients with and without PIGA mutation in blood cell counts, reticulocytes, and GPI-AP deficient cell fractions. Median PIGA mutation burden was significantly higher in granulocytes (44%) than in T lymphocytes (19.4%) (P=0.004). Interestingly, PIGA mutation burden in granulocytes correlated with the GPI-AP deficient fractions of RBCs (P=0.023) and granulocytes (P=0.049). Similarly, the mutation burden of T lymphocytes correlated with the GPI-AP deficient T lymphocyte fraction (P=0.006; Fig. 2).

TCR clonality was detected in 19 (79.2%) out of the 24 PNH patients. TCRB, TCRG, and TCRD clonality was detected in 50%, 50%, and 41.7% of the patients, respectively (Table 3). Interestingly, PIGA mutation burden of T lymphocyte was more frequently detected in patients with clonal rearrangements in any of the TCR genes, compared with those with no TCR rearrangements (P=0.015). The PIGA mutation burden of T lymphocytes was higher in patients with clonal TCRB rearrangement than those with no TCR rearrangements (27.2% vs 8.2%; P=0.048; Fig. 3).