Analysis of saliva samples and cluster of differentiation 3 (CD3)+ lymphocytes as a source of germline DNA in myeloproliferative neoplasms.

The widespread use of genetic studies in patients with myeloproliferative neoplasms (MPN) by next-generation sequencing (NGS) has led to the identification of new genetic variants. To establish the potential pathogenic role of these variants with scarce or inexistent literature, it is important to assess whether their origin is somatic or germline,1 as germline variants in myeloid-related genes have been associated with hereditary disorders2, 3or germline predisposition, while somatic variants are associated with clonal expansion of haematopoietic stem cells not only in MPNs, but also in most types of myeloid malignancies. In this context, germline DNA is required to perform these analyses, which is frequently obtained from several sources, e.g. saliva, buccal mucosa, cluster of differentiation 3 (CD3)+ lymphocytes, fibroblasts or hair. The aim of the present study was to ascertain if DNA obtained from saliva samples and CD3+ lymphocytes from peripheral blood is a suitable source of germline DNA for molecular studies of patients with Philadelphia chromosome (Ph)-negative MPNs. Paired samples of saliva and granulocytes from 191 patients with MPNs [76 polycythaemia vera (PV), 109 thrombocythaemia (ET), five primary myelofibrosis (PMF), one unclassifiable MPN] harbouring mutations in Janus kinase 2 (JAK2; n = 159), calreticulin (CALR; n = 26) and MPL proto-oncogene, thrombopoietin receptor (MPL; n = 6) were analysed. Saliva DNA was obtained using the Oragene-DNA kit (DNA Genotek). JAK2V617F was assessed by allele-specific real-time polymerase chain reaction (PCR), CALR mutations were determined by fragment analysis electrophoresis and MPL mutations by digital PCR. Methods and materials are detailed in Data S1. Analysis of DNA from saliva showed that in 69/191 (36·1%) cases, neither the wild type nor the mutated form of the JAK2, CALR or MPL genes were detected, although a large amount of DNA was present in the sample. The absence of amplifiable human DNA suggested that it could come from the microorganisms present in the oral mucosa, and was confirmed by the amplification of the 16s ribosomal gene. Thus, these 69 samples were considered as uninformative. In the remaining 122 (63·9%) saliva DNA samples, driver gene mutations were detected in 89/122 cases (73%) and a strong correlation between the variant allele frequency (VAF) of the mutation in granulocytes and saliva was observed (r = 0·706, P < 0·001, Spearman) (Fig 1). Of note, in 50 of these 122 saliva samples the VAF detected was >30%, meaning that in these cases (41% of the cohort) the use of saliva as a control for germline mutations, would result in considering acquired variants as germline. To get further insight into the origin of this mutated DNA present in saliva, we hypothesised that it might be cell-free tumoral DNA (ctDNA), as it has been described that saliva is a possible source of ctDNA. We assessed the size of the DNA fragments by capillary electrophoresis and found that most DNA fragments corresponded to high molecular weight DNA with no evidence of fragments around 166 base pairs, indicating that the isolated DNA was not ctDNA. Next, we analysed the relation with the leucocyte count. Patients with positive saliva had higher leucocyte counts in peripheral blood [median (range) 8·48 (2·71–34·8) × 109 cells/l in positive saliva vs 6·25 (2·59–21·75) × 109 cells/l in negative saliva; P = 0·001, Mann–Whitney), supporting the hypothesis that the positivity for the driver mutation in saliva samples may be due to leucocyte presence in the oral mucosa. Previous studies have described that saliva DNA from patients with MPN can be positive for JAK2V617F mutation.4, 5Moreover, a strong correlation has also been described between the VAF of JAK2V617F in granulocytes and saliva in a limited number of patients.6 Our present study confirms these observations in a larger cohort and includes for the first time cases with driver mutations in CALR and MPL, indicating that this effect would be affecting all patients with MPN. This is of utmost importance as broad gene panels are being performed in triple-negative patients, in which the detection of new variants should be carefully evaluated in germline DNA. Searching for an alternative sample type for germline studies, we isolated CD3+ lymphocytes by immunomagnetic selection in 64 JAK2V617F-positive patients. In all cases JAK2 was amplifiable, and only one sample showed a high JAK2V617F VAF of 27·3% [median (range) VAF 2·39 (0·01–27·3)%]. Previous studies described that CD3+ lymphocytes are negative or present a very low VAF for JAK2V617F,5 but in some exceptional cases JAK2V617F is detected with a high VAF, suggesting homozygosity in the lymphoid compartment. It is possible that in these positive cases, the JAK2 mutation appears at an earlier stage of the haematopoiesis, affecting several blood lineages.7 Additionally, CD3+ cells from 12 CALR- and five MPL-mutated cases were isolated. For the CALR-mutated cases, the VAF in CD3+ was higher than that observed in the JAK2V617F-positive cases [median (range) VAF 11·2 (3·66–28·4)%], whereas for the MPL-mutated cases the VAF in CD3+ was lower [median (range) VAF 1·45 (0·43–3·54)%] (Fig 2). The higher VAF in the CD3+ fraction of CALR-mutated cases than in JAK2- or MPL-mutated cases could be indicative of an earlier origin of these mutations in the haematopoietic lineage. CALR mutations have been described to originate in the haematopoietic stem cell8 and it has been previously postulated that CALR mutations could be an early event in the development of the disease. Mutations in CALR are generally present in the whole granulocytic cell compartment with a VAF of around 50%, suggesting a greater proliferative advantage of the mutated progenitor CALR clone compared to the JAK2-mutated patients.9 In favour of this hypothesis, CALR mutations are acquired before additional mutations in other non-driver genes, while JAK2 mutations can be preceded by first hits in other genes such as ten–eleven translocation 2 (TET2), DNA methyltransferase 3 alpha (DNMT3A) or additional sex combs like-1 (ASXL1).8, 10 As a whole, although we detected some positive cases for the somatic driver mutation in the CD3+ fraction, the presence of the mutated clone was significantly lower [median (range) VAF 2·83 (0·00–28·37)%] than in the saliva samples [median (range) VAF 21·7 (0·00–94·64)%; P < 0·001, Wilcoxon]. In conclusion, in patients with MPN with somatic mutations, saliva samples are not a reliable source of germline control DNA. Moreover, in an important set of cases in our present cohort (36%), no human DNA was detected in the saliva samples. Therefore, the use of CD3+ lymphocytes is a better option than saliva for the study of germline DNA variants in Ph-negative MPNs. This study was supported in part by grants from El Instituto de Salud Carlos III (ISCIII) and Spanish Ministry of Health, PI16/0153, 2017SGR205, PT17/0015/0011. Beca Gilead 2016 and Xarxa de Banc de Tumors de Catalunya. We thank Roger Anglada and Xavier Calvo for technical support. Nieves Garcia-Gisbert, Laura Camacho, Lierni Fernández-Ibarrondo and Raquel Longarón performed the research. Concepcion Fernández-Rodriguez, Carlos Besses and Beatriz Bellosillo designed the research study. Joan Gibert, Anna Angona and Marcio Andrade-Campos contributed essential reagents or tools. Nieves Garcia-Gisbert, Laura Camacho and Joan Gibert analysed the data. Nieves Garcia-Gisbert, Concepcion Fernández-Rodriguez, Antonio Salar, Carlos Besses and Beatriz Bellosillo wrote the paper. Antonio Salar received honoraria for speaker, consultancy or advisory role from Roche, Janssen Pharmaceuticals, Gilead and Celgene. Carlos Besses received honoraria for speaker, consultancy or advisory role from Gilead. Beatriz Bellosillo received honoraria for speaker, consultancy or advisory role from Astra-Zeneca, Biocartis, Merck-Serono, Novartis, Qiagen, Hoffman–La Roche, ThermoFisher, Pfizer and BMS. 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.