Clonal Replacement Underlies Spontaneous Remission in Paroxysmal Nocturnal Haemoglobinuria

Spontaneous clinical remission of paroxysmal nocturnal haemoglobinuria (PNH), characterized by the resolution of hemolysis, cessation of haemoglobinuria, and normalization of the acidified serum test(Dacie and Lewis 1972), has been a biological curiosity for over 5 decades, and has fascinated PNH investigators and patients alike. Seen in up to 15% of PNH patients after 10–20 years of disease(Hillmen, et al 1995), spontaneous remission is still considered the only cure for PNH aside from bone marrow transplantation. No predictors of future remission have been identified, but even patients with severe transfusion-dependent haemolysis accompanied by thrombotic complications can undergo spontaneous remission(Hillmen, et al 1995). Published cases frequently detail a gradual diminution of disease severity, sometimes over decades, culminating in full resolution of symptoms(Charache 1969); laboratory findings may reveal normalization of glycosylphosphatidinositol (GPI)-linked proteins on granulocytes and erythrocytes(Hillmen, et al 1995), although some laboratory abnormalities may persist long-term (Charache 1969, Dacie and Lewis 1972).

Historically, PNH remission have been ascribed to a gradual repopulation of the bone marrow by normal haematopoietic elements(Charache 1969), either by virtue of a finite life-span of PNH clones(Hillmen, et al 1995), neutral stochastic drift(Dingli, et al 2008), or changes in the microenvironment(Pulini, et al 2011). Yet, aside from published case descriptions, there are no studies of patients who achieved spontaneous remission, hindered by the low incidence of PNH and the need for very long-term follow up. Here, we present the first comprehensive genetic analysis of a patient in spontaneous remission, which, together with recent data on clonal hematopoiesis in PNH, demonstrates that spontaneous remission of PNH does not imply restoration of normal haematopoiesis, but instead mirrors the clonal dynamics, where the PNH clone is replaced by another independent haematopoietic clone carrying multiple somatic mutations.

A 46-year-old male was diagnosed with classical haemolytic PNH at the age of 27 years, when he presented with syncope and was found to have severe haemolytic anaemia with a haemoglobin of 40 grams/liter. His first available flow cytometric evaluation revealed a PNH granulocyte clone of 85%. For the first ten years of disease he was managed with bi-weekly transfusions, and, once clinically available, was started on C5 complement inhibitor eculizumab. Over the following 8 years, his PNH clone size slowly decreased from 72% at eculizumab initiation to ~15% (Figure 1A), when he was able to discontinue eculizumab with no recurrence of haemolysis over the 18 months of subsequent follow-up. Bone marrow biopsy at the time of eculizumab discontinuation revealed some erythroid hyperplasia (Figure 1B).

With recent studies showing that the majority of PNH patients carry somatic mutations either ancestral to or subsequent to the PNH-driver mutation in PIG-A(Shen, et al 2014), and because leukemic transformation of PNH has been reported to frequently lead to the disappearance of the PNH clone(Cornelis, et al 1996), we hypothesized that clonal replacement is not limited to leukemic transformation in PNH, but can also underlie PNH remission. To evaluate for clonal replacement in our patient with spontaneous remission of PNH, we performed comparative whole exome sequencing (WES) of the bone marrow and skin fibroblast DNA, looking for emergence of somatic mutations in the patient’s bone marrow. WES and bioinformatics analysis were performed as previously described(Babushok, et al 2015); all putative somatic mutations were validated by bi-directional Sanger sequencing. Structural chromosomal abnormalities were analyzed using metaphase cytogenetics and single nucleotide polymorphism array (SNP-A) analysis, with no aberrant genomic rearrangements detected.

We found 11 somatic mutations in the bone marrow. Of these, two mutations were non-synonymous coding: a p.Pro1038Leu mutation in Serine/Threonine Kinase 36 (Fused homolog, STK36) and a p.Arg993Gln mutation in Mannosidase Alpha Class 2A Member 2 (MAN2A2), and three involved untranslated regulatory regions in Mannosyl(alpha-1,3)-Glycoprotein Beta-1,2-N-Acetylglucosaminyltransferase (MGAT1), Protein Tyrosine Phosphatase, Receptor Type, U (PTPRU), and UDP Glycosyltransferase 2 Family, Polypeptide A1 (UGT2A1) genes (Figure 1C, Figure 2A). Manual review identified a low, 3% allele frequency frameshift mutation in exon 4 of PIG-A (Figure 2B). Five additional somatic mutations were presumed to be passengers as they were either synonymous or intronic (data not shown).

Clonal architecture analysis performed on the patient’s peripheral blood 12 months after WES revealed the presence of a single dominant clone marked by the STK36, MAN2A2 and MGAT1 mutations, and lacking the PIG-A mutation (Figure 2C). Cell lineage mapping using the immunomagnetically-sorted CD3-depleted, CD19-depleted myeloid and CD3-selected, CD19-selected lymphoid peripheral blood cells confirmed persistence of STK36, MAN2A2, MGAT1 mutations 18 months post-WES, with mutations present solely in the myeloid fraction (Figure 2A). An independent, very low frequency clone carrying the PIG-A mutation was also detected, both in the myeloid and in the lymphoid cells (Figure 2B, and data not shown). Intriguingly, a splice site-altering somatic mutation in the Mannosidase, Alpha, Class 1A, Member 2 (MAN1A2), another member of the mannosidase family involved in N-glycan maturation in mammalian cells, was reported in another PNH patient(Shen, et al 2014), hinting at the role of altered N-glycosylation as an immune escape mechanism in PNH, perhaps through the function of N-glycosylation in the peptide-MHC and T-cell receptor interactions(Zhou, et al 2014).

In summary, our study is the first genetic analysis of the haematopoietic compartment in a patient with clinical remission of PNH. Our results show that PNH remission was not brought on by restoration of normal haematopoiesis, but instead coincided with emergence of a new dominant clone carrying multiple somatic mutations. The dominant clone was detected at 18 months of study follow-up, confirming clonal longevity. Our findings expand the understanding of PNH and illustrate longitudinal clonal dynamics in what is traditionally considered a non-malignant disease. Further studies with longitudinal follow-up will be needed to determine the prevalence and prognostic significance of clonal replacement in PNH, and whether these represent competing immune escape pathways or proliferative drivers. Importantly, our study demonstrates that spontaneous remission in PNH does not imply restoration of normal haematopoiesis. Instead, our results underline the importance of clonal haematopoietic composition in determining one’s clinical outcome, and highlight the need for ongoing clinical follow-up of this rare population due to potential implications for prognosis and the risk of malignant transformation.