Remarkable stability in clonal hematopoiesis involving leukemia-driver genes in patients without underlying myeloid neoplasms.

Abstract

Clonal hematopoiesis (CH) of indeterminate potential is defined by the acquisition and subsequent expansion of somatic, leukemia-associated driver mutations in hematopoietic stem cells in apparently healthy individuals, without underlying hematological neoplasms.1 The true prevalence of CH is yet to be defined, with current estimates depending upon sequencing techniques used along with a clear age dependency.2, 3 The development of error-corrected sequencing using unique molecular identifiers has allowed for detection of CH clones up to two orders of magnitude below conventional next generation sequencing techniques (NGS) (1 per 10 000 cells), demonstrating the ubiquitous nature of these mutations.4 CH often involves epigenetic regulator genes (DNMT3A, TET2 and ASXL1–75%), with the presence of CH not only increasing the risk for hematological malignancies (HR 11.1; 95% CI 3.9–32.6) but also associating with an increased all-cause mortality (HR 1.4, 95% CI 1.1–1.8) largely due to cardiovascular disease (HR 2.1, 95% CI 1.2–3.4).2, 3 In a seminal study, 20 nurses, aged 50–70 years, had paired DNA samples assessed for CH at two time points approximately 10 years apart. Sixteen of twenty (80%) were found to have clonal single nucleotide variants; 30 at both time points, 15 at the first time point only, and 60 at the second time point only, with remarkable longitudinal stability (including TP53 and KRAS).4 This study was agnostic to cell intrinsic and extrinsic clonal selection pressures that shape clonal evolution. We carried out a similar assessment in 13 patients who received their care at Mayo Clinic, allowing us a more detailed insight into clonal selection pressures, with assessment for CH being carried out at both time points. The study was approved by the Mayo Clinic Institutional Review Board and consent was obtained prior to enrollment. Thirteen patients had blood DNA samples available at two time points, approximately 7 years apart. These patients had consented to the Mayo Clinic biobank and were selected for this study as they received their ongoing healthcare at Mayo Clinic (annual documented visits in the electronic medical record). DNA was extracted from peripheral blood mononuclear cells and subjected to a customized targeted NGS assay covering 220 CH-related genes, as previously described (median sequence read depth ~ 1000x, including all coding regions and consensus splice sites).5, 6 Plasma samples at diagnosis were obtained and analyzed for cytokines and chemokines as previously described.7 Follow-up plasma samples unfortunately were not available for these patients. Clinical data, exposure to radio- and chemotherapy, and environmental exposures were retrospectively abstracted. Outcomes of interest included the development of cytopenias, progression to hematological malignancies, comorbidities, concurrent neoplasms, and thromboembolic events (deep vein thrombosis/DVT and pulmonary embolism/PE). Continuous variables were compared using the Mann–Whitney or Kruskal-Wallis test. Categorical variables were compared using the Fischer's exact test. All statistical calculations were carried out using JMP statistical software, version 16.0, SAS Institute Inc. Cary, NC, 1989–2019. Thirteen patients were included in the study: 64% female (Table S1). The median age at first sample acquisition was 69 years (range, 51–83 years), while the median age at second sample acquisition was 77 years (range, 57–90 years). The median time interval between the two sample acquisitions was 6.4 years (range, 5.4–11 years). Eight (61%) patients had a detectable CH mutation at the first time point, with 2/8 (25%) having two CH related mutations (Figure 1). The frequency of individual CH mutations seen at time point 1 included DNTM3A n = 4 (40%), with n = 1 (10%) each involving TET2, JAK2V617F, PPM1D, SETD2, SF3B1 and CBL (Supplementary Table 1). At time point 2, all 13 (100%) patients had detectable CH (18 mutations in total), with individual mutational frequencies being DNTM3A n = 5 (27.7%), TET2 n = 4 (22.2%), n = 2 (11%) each for JAK2V617F, SETD2, and SF3B1 and n = 1 (5.5%) each for PPM1D, CBL, and KRASG12D. Five (38%) patients had two mutations at the second time point. Mutations included missense mutations (67%) detected in DNMT3A (C861T, R736H, R882H, F868L), TET2 (R1951W, R1871S), JAK2 (V617F), SF3B1 (K666N, K700E), KRAS (G12D), and CBL (R420Q), nonsense (27%) mutations detected in TET2 (Q976*, Q1527*), SETD2 (R441*, K637*), and PPM1D (C407*), and one splice site mutation in DNMT3A (c.1015-2A > G). At time point 1, the median variant allele frequency (VAF) of mutations seen in eight patients was 5% (range, 2%–21%), with two patients having two mutations and three patients having VAF ≥20% (SETD2 K637*-20%, DNMT3A F868L-21% and CBL R420Q-21%). In spite of these VAFs' being in the clonal cytopenias of undetermined significance (CCUS) range, none of these patients had significant blood count abnormalities. The median VAF at the second time point was 4% (range, 2%–36%), with five patients having two mutations and three patients having VAF ≥20% (SETD2 K637*-23%, DNMT3A F868L-21%, and CBL R420Q-31%). SETD2 is a histone methyltransferase, solely responsible for trimethylation of histone 3 at lysine 36 (H3K36me3), thus closely regulating gene transcription.8 Somatic SETD2 mutations have been described in several malignancies, with SETD2K637* having been reported in lung cancer and SETD2R441* having been reported in early precursor T cell acute lymphoblastic leukemia, respectively.9, 10 CBL on the other hand is a gene that encodes an E3 ubiquitin ligase and a signaling adaptor and is commonly mutated in myeloid neoplasms (15% of patients with chronic myelomonocytic leukemia).6 The CBLR420Q variant seen in patient 13 impacts the RING finger domain, leading to loss of ubiquitin ligase activity and deregulated downstream receptor tyrosine kinase signaling, and has been seen in myeloproliferative neoplasms (MPN).11 Despite having these pathogenic variants, the two above-mentioned patients did not develop hematological neoplasms at last follow up. Within limitations of small numbers, the cytokine comparisons between patients with CH at time point 1 versus those without, revealed no significant differences, apart from an elevated interferon-α alpha level in patients with CH (p = 0.03). At last follow up, all patients were alive with patient 4 demonstrating disease evolution to a JAK2 V617F mutant MPN, consistent with essential thrombocythemia (age at diagnosis 58 years, JAK2V617F VAF- 8%, 11.2 months after time point 2 and 7.3 years after time point 1). The median time from last assessment of CH to last follow up for the cohort was 8.67 months (range, 1.3–11.2 months), while the median time from last clinical assessment to last follow up for the cohort was 0.76 months (range, 0–11.1 months). Among the three patients with high VAF CH, patient 8 with a SETD2 mutation at last follow up was found to have monoclonal B lymphocytosis (MBL) with deletion 13q and trisomy 12, with accompanying mild monocytosis. Patient 13 with TET2 and DNMT3A mutations had an increase in the red cell distribution width (RDW-47.4, normal range 36.4–46.3) without any blood count abnormalities. Patient 13 with TET2 and CBL mutations had mild thrombocytopenia (147 × 10 (9)/L) with a persistently elevated RDW (51.4). For all patients, the median change in VAF between the two time points was 2% (−2% to 33%), indicative of marked stability of most clones. Three of five (60%) patients who did not have CH at time point 1 developed TET2 mutant clones at time point 2 (Figure 1). Patient 1 with 2 DNMT3A CH, first detected in 2013, had chronically elevated C-reactive protein and IL-6 levels and an unprovoked DVT in 2019 (Table S1). Patient 2 with DNMT3A CH, first detected in 2014, was diagnosed to have an IgG kappa monoclonal gammopathy in 2019. Patient 3 with JAK2V617F CH, first detected in 2014, also had a diagnosis of seronegative inflammatory arthritis (2011) managed conservatively. Patient 4 with JAK2V617F and DNMT3A CH, first detected in 2020, had a history of a T1, N2, M0 squamous cell carcinoma of the head and neck treated with cisplatin and radiation therapy in 2013. Patient 5 with PPM1D CHIP, first detected in 2014, had a provoked DVT and PE in 2016. Patient 7 with SETD1 CH, first detected in 2014, was also diagnosed to have a markedly stable, low white count, MBL in 2006. Patient 9 with SF3B1 CH, first detected in 2019, was found to have a nonfunctional adrenal adenoma several years prior to CH diagnosis. Patient 11 with TET2 and KRASG12D CH, first detected in 2019, did have a history of a T1a, N0, M0, clear cell renal cell carcinoma and underwent a nephrectomy in 2010. This patient also had an unprovoked DVT and PE in 2013. Patient 12 with TET2 and DNMT3A CH, first detected in 2014, did have a provoked DVT in 1987. Patient 13 with TET2 and CBL CH, first detected in 2014, was diagnosed to have transitional cell bladder cancer treated with mitomycin in 2006, and then again in 2009. In this study, we demonstrate the spectrum and marked stability of CH in 13 apparently healthy individuals, at two time points, approximately 7 years apart. Sixty-one percent of patients had detectable CH at the first time point, while 100% had detectable CH at the second time point, with DNMT3A mutations being the most common in 40%, followed by TET2 and JAK2V617F.1-3 While ASXL1 mutant CH is seen in 5%–10% of CH patients, with a higher prevalence in tobacco smokers (OR 1.1, p = 0.005), in our study, in spite of 75% of patients being former smokers (48.5 mean pack years; range, 7–120 pack years), we did not encounter any ASXL1 mutations.12 This is most likely due to the smaller sample size. For patients who did not have CH at time point 1, TET2 mutations were the most common mutations acquired at time point 2. In addition, while three patients at time point 2 acquired mutations largely involving signaling pathways (JAK2V67F, KRASG12D, and CBLR420Q), none of these patients developed CCUS or myeloid neoplasms at last follow up, as has been previously described.4 The median change in CH VAF between the two time points was 2% (range, −2% to 33%), reflecting overall stability with variable changes in individual cases, as previously published.4 Measuring time-dependent changes in CH-related mutational VAF is difficult, given inherent variabilities in VAF measurement based on quality of samples, sequencing depth, number of reads, and batch effect, making interpretation of relatively small changes difficult to attribute to true biological differences. We also document three patients in this cohort with VAF >20% without significant cytopenias, contrary to retrospective studies of patients with CCUS, where most patients with CH VAF >20% had clinically significant cytopenias.5, 13, 14 CH VAF >20% has also strongly been associated with progression to myeloid neoplasms such as MDS, MDS/MPN, and AML.5, 13 In our series, patient 4 with a JAK2V617F CH developed essential thrombocythemia with a JAK2V617F VAF of 8%, indicating that CH is not a binary variable and that individual nuances related to gene variants and disease phenotypes need to be taken into consideration. There were two patients in our cohort who developed CH after prior exposure to chemo/radiation therapy (JAK2 + DNMT3A and TET2 + CBL), without typical therapy selected mutations involving TP53 and PPM1D, while the one patient with PPM1D CH had no relevant antecedent chemo/radiation exposures.12 Four (30%) CH patients had venous thromboembolism, two unprovoked, once again highlighting the potential role of CH in endothelial dysfunction that needs further prospective exploration. In conclusion, this study demonstrates the relative longitudinal stability of CH clones and highlights nuances associated with individual variants, difficulties in understanding clonal selection pressures (cell intrinsic and extrinsic) and clonal dynamics, and in measuring biologically significant changes in clone sizes. We would like to acknowledge all patients and families that contributed to the study. We would like to acknowledge the Mayo Clinic Center for Individualized Medicine, Mayo Clinic biobank and the Henry J Predolin Leukemia grant at the Mayo Clinic for funding this study. All authors declare no relevant conflicts of interest and disclosures. All NGS sequencing data has been outlined in supplemental material. Raw data can be obtained by contacting the corresponding author. Table S1. Clinical and molecular characteristics of 13 patients with clonal hematopoiesis and longitudinal follow-up Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.