We report a detailed longitudinal study of the first patient to be treated (in 1973) for paroxysmal nocturnal hemoglobinuria (PNH) with syngeneic bone marrow transplantation (BMT). The patient subsequently relapsed with PNH in 1983, and still has PNH to date. Analysis of thePIG-A gene in a recent blood sample showed in exon 6 an insertion-duplication causing a frameshift. Polymerase chain reaction (PCR) amplification of the PIG-A exon 6 from bone marrow (BM) slides obtained before BMT showed that the duplication was not present; instead, we found several single base pair substitutions in exons 2 and 6. Thus, relapse of PNH in this patient was not due to persistence of the original clones; rather, it was associated with the emergence of a new clone. These findings support the notion that the BM environment may create selective conditions favoring the expansion of PNH clones.

© 1998 by The American Society of Hematology.

PAROXYSMAL NOCTURNAL hemoglobinuria (PNH) is an acquired chronic disorder associated with intravascular hemolysis, increased tendency to venous thrombosis, and cytopenia due to bone marrow (BM) failure.1 PNH is associated with somatic mutations in the X-linked PIG-A gene in an early hematopoietic stem cell.2 The biochemical defect in PNH has been localized to an early step in the glycosyl phosphatidylinositol (GPI) anchor biosynthetic pathway. Consequently, PNH cells are deficient in GPI-anchored proteins, including the Decay Accelerating Factor (DAF or CD55) and the Membrane Inhibitor of Reactive Lysis (CD59), both of which are involved in the regulation of complement activity on the cell surface.3

The only curative therapy available for PNH is BM transplantation (BMT). A first report of the successful application of BMT for PNH associated with severe aplastic anemia (AA) appeared in 1973.4 Subsequently, out of 17 PNH patients who had BMT, 11 received human leukocyte antigen (HLA) identical sibling marrow with conditioning, 1 received HLA-haploidentical marrow with conditioning, and 5 received syngeneic marrow without conditioning. Long-term follow up of these 17 patients showed that only the 5 patients who had been transplanted with syngeneic BM, without immunosuppressive conditioning therapy relapsed, although some were doing well without need of transfusion.5-10 Here, we show that relapse of PNH after BMT can result from the expansion of new PNH clones rather than from the persistence of the original PNH clone.


R.S. (MSK13) was investigated at the age of 18 because of anemia (date of birth, 03/13/54). In September 1972, based on a positive Ham test, a diagnosis of PNH was made. In June 1973, the patient had an infusion of BM from his syngeneic twin, without conditioning.11 The patient had a good clinical and hematological response (see Fig 1), but his white cell count remained rather low, and after 4 years the platelet count started declining. Ten years after BMT (in 1983) the complement lysis test was again positive (10.7% lysis) and the patient could be regarded as having a lab relapse of PNH6; by 1987, with the recurrence of anemia, the patient had to be regarded as having clinical relapse of PNH. The patient also developed pancreatitis and splenic vein thrombosis and underwent splenectomy in 1989. Subsequently, he had gastrointestinal bleeding from esophageal varices requiring sclerotherapy. In 1992, tests for hepatitis C were positive and a liver biopsy showed evidence of hepatitis but no fibrosis or cirrhosis. Currently, the patient is mildly pancytopenic and is being managed conservatively.

Fig. 1.

Long history of PNH in patient MSK13.


Flow cytometric analysis.

Analysis of GPI-anchored protein on red blood cells (RBCs), polymorphonuclear neutrophils (PMN), and mononuclear cells (MNC) was performed by flow cytometry (FACscan; Becton Dickinson, Mountain View, CA) using monoclonal antibodies towards GPI-anchored protein.12

DNA extraction.

DNA was extracted separately from PMN and from MNC by dodecyl sulfate proteinase K-method.13 Archival Wright-stained BM smear slides were processed as follows. After soaking for 3 hours in xylene, the coverslips were removed and the slides were then soaked further overnight to remove residual mounting medium. The xylene was then removed by evaporation. Cells were scraped from slides into a 1.5 mL Eppendorf tube containing lysis buffer (100 mmol/L Tris-HCl, pH 7.4; 5 mmol/L EDTA, pH 8; 0.2% sodium dodecyl sulfate [SDS]; 200 mmol/L NaCl; and 100 mg/mL proteinase K) using a sterile scalpel. After overnight incubation at 55°C, DNA was extracted twice by equal volume of phenol, and twice with an equal volume of chloroform-isoamyl alcohol (24:1).13 The DNA was precipitated with an equal volume of isopropanol, mixed by inverting the tube, and then incubated 1 hour at −70°C. After centrifugation, the pellet was washed once with 70% ethanol and left to dry at room temperature. Invisible pellet was resuspended in 10 μL of TE (TE=10 mmol/L Tris-HCl, pH 7.4; 1 mmol/L EDTA, pH 8).

Polymerase chain reaction (PCR) amplification.

The PIG-A coding region was PCR amplified in 4 fragments from genomic PMN DNA as previously described.14 15 For slide-stripped DNA analysis, the first round of PCR amplification did not give any visible PCR product. Therefore, the primary PCR products (5 μL) were reamplified by nested PCR for exons 2, 3, and 6, and seminested PCR for exons 4 and 5 by using internal oligonucleotides (Table 1).

Table 1.

Details of Oligonucleotide Primers Used in This Study

Characterization of PIG-A gene mutations.

Single-strand conformation analysis (SSCA) and heteroduplex analysis (HA) were performed, as previously described,14 15 on the primary and on the secondary PCR products obtained from the post-transplant and the pre-BMT samples, respectively. Nucleotide (nt) sequencing was performed by Sequenase Version 2.0 DNA sequencing kit (US Biochemical Corp, Cleveland, OH), after cloning the abnormal fragments into phage M13.


Molecular analysis of PNH relapse.

FACscan analysis of a recent blood sample (1995) showed that 2% of the patient’s RBCs were deficient in CD59, and 90% of his PMN were deficient in CD59 (Fig 2), CD24, and CD16 (data not shown), consistent with PNH. HA and SSCA showed no abnormalities in exons 2-5 of the PIG-A gene. Analysis of aBstNI restriction enzyme digest obtained from PCR-amplified exon 6 (474 bp) showed an abnormally large double-stranded fragment (Fig 3). Nt sequence analysis showed (in 19 out of 23 M13 clones) an insertion of 2 nucleotides (AA) at nt position 1355, followed by a duplication of 32 nucleotides encompassing nt 1324-1355, (Fig 4). This insertion-duplication will cause a frameshift resulting in a truncatedPIG-A protein of 462 instead of 484 amino acids (aa), in which aa 453-to-462 are abnormal. The PIG-A protein will be functionally inactivated; therefore, we regarded this mutation as responsible for PNH in this patient at this time. The duplicated DNA element was flanked by a 4 bp TTGA direct repeat (Fig 5A); the same repeat is found four times in the mutant sequence. Therefore, we presume that a nonhomologous recombination event (by sister chromatid exchange) must have occurred between TTG in the normal sequence at nt position 1357 and the duplicated sequence at nt position 1324 (Fig 5A). It appears that this recombinational event was either preceded or followed by the insertion of two A at position nt 1355. For brevity, we will refer to this abnormality simply as an exon 6 duplication.

Fig. 2.

Flow cytometry analysis with anti-CD59 of PMN from a normal control (broken line) and PNH patient MSK13 (full line).

Fig. 3.

Analysis of exon 6 of the PIG-A gene before BMT and after relapse. (A) Heteroduplex Analysis (HA); Two extra bands (235 bp, mutant homoduplex; H, heteroduplex) in patient MSK13 are clearly observed after relapse (lane 4) but not before BMT (lane 2). (B) Single-Strand Conformation Analysis (SSCA). Along with double-stranded fragments DNA, an altered electrophoretic mobility (shift) of the single-stranded DNA (SS) is only present after relapse (lane 4). Normal controls (lanes 1, 3, 5); normal homoduplex, 201 bp.

Fig. 4.

Insertion duplication in exon 6 of the PIG-A gene in patient MSK13. The sequence of 32 nt starting by 5′-TCAA-3 until 5′-CAA-3 in the normal control is replicated in MSK13, but after the insertion of two thymidines. The insertion-duplication introduces a frameshift at codon 452 and leads to the production of truncated PIG-A protein of only 462 AA.

Fig. 5.

High-sensitivity analysis of the 1355insAA-duplication nt 1324-1355 before and after BMT. (A) Schematic representation of the insertion-duplication in exon 6 of the PIG-A gene. At the top, two copies of the normal sequence are aligned to show direct-sequence homology (underlined) that may favor nonhomologous exchange below the sequence in the recent sample of patient MSK13. Identical bases are indicated by a vertical line. The short repeat of 4 nt flanking the duplication are boldface. The 19 mer D has been designed to match completely the mutant sequence, whereas the last 4 nt (5′-… .ttTC-3′) are mismatched to the normal sequence (see arrow bent at the left). The fragment amplified with primers -I5b and D and -I5b and f are expected to be 292 bp and 415 bp in length, respectively. (B) Gel electrophoretic analysis of nested-PCR products of exon 6 of PIG-A gene from PMN in the pretransplant sample. A fragment of 292 bp amplified with the primers -I5b and D was present in the post-transplant sample (lane 2); by contrast no amplification was obtained in the pretransplant sample (lane 1) and in a normal control sample (lane 3). When the 3 primers -I5b, D, and f were used in the same PCR reaction, a fragment of 415 bp only was amplified in the pretransplant sample and in a normal control (lanes 4 and 6), and a fragment of 292 bp only was amplified in the posttransplant sample (lane 5).

Molecular analysis of original PNH cells from archival material.

To determine whether in this patient the relapse of PNH took place because of resurgence of the same PNH clone that had originally caused his disease, we needed to analyze pretransplant DNA. The only available archive material consisted of two BM slides from the time of diagnosis, before BMT (1973). Amplification of PIG-A exon 6 from these slides (performed by the so called nested-PCR approach) was successful, but it failed to produce the double-stranded fragment of abnormal size observed in the recent sample (Fig 3). Nevertheless, 36 M13 clones were sequenced. None of them had the exon 6 duplication; but in 13 of them we found a 1442 C→T point mutation, causing a 481 ser→phe amino acid replacement (Table 2). To test the possibility that the exon 6 duplication might have been present already in 1973 in a very small percentage of cells, we made efforts to increase the sensitivity of our detection method.16 For this purpose, we designed a reverse primer called D that spans the insertion at the 3′ end of the duplication (Table 1 and Fig 5A). By using primer D and primer -I5b, and the primary PCR product as template (because no more DNA extracted from BM slides was left), we amplified the abnormal fragment very efficiently from the post-transplant sample, but not at all from the pre-BMT sample (Fig 5B). Hybridization of the Southern-blot of the same gel with a PIG-A cDNA probe did not show any signal (data not shown). All others exons were also amplified by nested-PCR from the pre-BMT slides and an abnormal fragment in exon 2 was observed by HA and SSCA. Nt sequence analysis of the appropriate DNA fragments showed several single bp substitutions (Table 2). In exon 2, 50% of the M13 clones sequenced had a 211A→C base change, causing a 71 thr→ala amino acid replacement; and 28% of these clones also have 251 C→T, causing 84 thr→ile (suggesting that the latter mutation arose in a cell belonging to the clone that had the former mutation). Still in exon 2, 14% of the M13 clones sequenced have a 16 G→T, causing 6 gly→stop (Fig 6).

Table 2.

Somatic Mutations in the PIG-A Gene of Patient MSK13 Before and After BMT

Fig. 6.

Point mutations in exon 2 of the PIG-A gene in patient MSK13 before BMT. (A) Exon 2 of the PIG-A gene was first amplified by external primers -I1 and I2, and then reamplified with the internal primers (see Table 1). (B) Nt sequence of M13 clones containing exon 2 fragment of patient MSK13. All (G) reactions were loaded adjacent to each other, followed by the (A, T, and C) reactions. Two of the 3 M13 clones with substitution 211 T → C (clones 21, 22, 24) also have the 251 G → A substitution (clones 22 and 24).

The finding of more than one clone in PNH is not unusual.15 17-20 However, we were concerned about the possibility that the two-stage amplification procedure we used might be associated with a higher probability of artifacts. Therefore, we performed control experiments by applying exactly the same nested-PCR amplification protocol to a 24-year-old BM slide from a non-PNH subject and to a relatively recent BM slide from another PNH patient (MSK11). In addition, we similarly analyzed a PMN DNA from a normal subject. In the normal control several different single bp substitutions were found; however, we never found more than one M13 clone with the same nt change (Table 3, line 6). In the regular DNA sample from patient MSK11 we found an abnormal exon 2 fragment, and direct sequencing showed a deletion of cytidine at nt position 259. After cloning in M13 this mutation was confirmed in 11 out of 12 M13 clones (Table 3, line 4). The same mutation was found in 9 out of 10 M13 clones from the nested-PCR product (Table 3, line 5); and in all 12 M13 clones from BM slides (Table 3, line 3). Therefore, we regarded this frameshift mutation as responsible for PNH in patient MSK11. Some of these clones had the PNH-related mutation plus another point mutation (different in each clone, as shown in column 5 of Table3). In the non-PNH 24 year old BM slide we did not find more than two M13 clones with the same nt change (Table 3, line 2). These control experiments indicate that, not surprisingly, Taq I polymerase errors do occur, but they are always different in different M13 clones, and they are not increased by using a high number of PCR cycles. There is no difference in Taq I polymerase errors between primary and secondary PCR (see Table 3, lines 4 and 5) . By contrast, in the pre-BMT sample from patient MSK13 the same nucleotide changes were found consistently in several independently isolated M13 clones, indicating that they reflect true mutations (Table 3, line 1, column 4 and 5).

Table 3.

Extensive Analysis Can Discriminate Pathogenic Mutations From PCR Artifacts

Analysis of the PIG-A gene in the BM donor.

Given that the PIG-A mutation in the relapse sample was completely different from those existing in this patient before BMT, it was possible that the former was in fact of donor origin. HA and SSCA of all exons of the donor’s PIG-A gene failed to show any abnormality. The highly sensitive duplication-specific nested-PCR technique that we developed (see Materials and Methods) failed to amplify exon 6 of the donor’s PIG-A gene. Thus, there is no evidence of this mutation being of donor origin. The donor remains clinically and hematologically normal.


The study of this patient has been informative in two respects. First, duplications in the PIG-A gene must be very rare. Until now, only one has been reported.21 Second, we have found that over the long history of this patient PNH was caused by clones with different PIG-A mutations at different times in his clinical course (Fig 1). Identifying the duplication mutation that currently underlies PNH in this patient was straightforward. For the pre-BMT phase of this disease, only a few BM slides were available, and PCR-amplification of all exons was performed successfully. Because of the very small amount of material available, we went to considerable lengths to adapt our methodology and to avoid being misled by PCR artifacts. Thus, we regarded a mutation as significant only if it was found in at least two independently isolated M13 clones. Of course, if a Taq I polymerase error occurs early during amplification, we might mistake a PCR artifact for a true mutation, especially if the DNA template consists only of very few copies. However, the comparison we have performed between the MSK13 slide and a non-PNH slide from the same year was clear cut; this control makes it unlikely that fixation, staining, and storage time contributed to create artifacts.

To our surprise, numerous mutations were found in the pre-BMT MSK13 sample. Although we cannot say with certainty which one or which ones were responsible for the patient’s original PNH phenotype, their representation amongst the M13 clones we sequenced was well above the threshold we had set. On the other hand, the duplication observed in the relapse sample gives such a characteristic pattern in HA and SSCA (Fig 3), that it could not be missed; and despite developing a highly sensitive customized nested-PCR methodology, we could not detect the exon 6 duplication in the pre-BMT sample. Of course it is impossible to rule out that the clone with the duplication may have existed in the patient’s BM in a site other than the one that was aspirated. With this proviso we feel confident that at the time the patient originally presented, the duplication could not account for clinical PNH (see Results and Table 2). Thus, we have provided proof that relapse of PNH in this patient was not due to failure of eliminating the original PNH clone, but rather to the emergence of a new clone.

Recently, Endo et al10 have reported that, in another patient who had syngeneic BMT for PNH, a PIG-A mutation present in one out of six T-cell clones (but not in peripheral blood leukocytes) before BMT could also be shown in peripheral blood leukocytes after BMT. Although their case is different because the follow up was only 14 months, the findings are not at all incompatible. We suggest that in the patient of Endo et al10 a minor PNH clone simply expanded after BMT (which is not surprising in the absence of myeloablation); in our patient, a new clone altogether has emerged.

The occurrence of several clones with independently arisenPIG-A mutations is well documented in PNH.15 17-20Our longitudinal study of this patient, spanning more than 20 years, provides more compelling evidence that, as we have previously suggested,1 the PNH clone has a conditional selective advantage in a particular BM environment. Endo et al10 have argued that in their patient the clone that expanded must have had an unconditional advantage; however, we note that if the conditions favoring a PNH clone before treatment are not modified by a nonablative BMT, the conditional advantage of this clone will be also unmodified. Indeed, if any of the PNH clones in our patient had an absolute growth advantage, there is no reason why any of them should have disappeared after a transplant procedure was performed without BM ablation. We think that instead the clone disappeared because the infusion of syngeneic BM provided a surplus of normal hematopoietic cells. Unfortunately, the aplastic environment must have persisted in the recipient, and this favored the growth of a clone with a newPIG-A mutation, whereas the old clones had meanwhile become exhausted. Because the donor and the recipient are syngeneic, no genetic marker is available to determine with certainty whether the relapse PIG-A mutation occurred in a donor cell or in a recipient cell, but this does not affect our interpretation. Indeed, if the mutation was in a donor cell, but this cell did not expand into a detectable clone in the donor BM, this fact would confirm once again that the pathological BM environment in the recipient allowed such expansion to take place.


We are very grateful to the patient and to his twin brother for their cooperation; to Dr D.P. Miller, the patient’s physician; and to Dr W.F. Rosse for performing complement lysis studies on the patient’s blood when a diagnosis of PNH was first made.


  • Supported by the NIH grant ROI-HL-56778, the DeWitt Wallace Clinical Research Fund, and the Kleberg Foundation.

  • Address correspondence to Khédoudja Nafa, Department of Human Genetics, Memorial Sloan-Kettering Cancer Center, 1275 York Ave, New York, NY 10021; e-mail: k-nafa{at}

  • The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" is accordance with 18 U.S.C. section 1734 solely to indicate this fact.

  • Submitted April 3, 1997.
  • Accepted June 21, 1998.


View Abstract