Page 60 - 2021_03-Haematologica-web

S. Vantyghem et al.
Supplementary Table S2). The most frequent ones were observed in JAK2 (p.Asn1108Ser, p.Arg1063His, p.Ile35Thr), MPL (c.1565+5C>T, p.Pro70Leu, p.Gln433Arg, p.Ala622Pro), ETV6 (p.Ala329Thr, p.Pro223Leu, p.Ala329Thr) and BCORL1 (p.Arg21His, p.Ala612Thr, p.Ile1022Thr, p.Arg1183Gln). Among patients with VUS, 14 had no clear somatic mutations, no proof of clonal hematopoiesis and no significant associa- tion with a given suspected context. Of note, identifica- tion of a poor prognostic mutation led to intensification of therapy in 3/12 patients in group A.
Prognostic/theranostic impact of mutations detected by next-generation sequencing (group B)
The median number of mutations in group B was two (range for the whole cohort, 0-9) and 75% of these patients had at least one mutation (Figure 1B, Online Supplementary Table S2). NGS identified prognostic muta- tions in 31/95 (33%) patients at different proportions according to the pathology, i.e., 14/40 (35%) in MDS, 11/29 (38%) in MPN/MDS, 5/17 (29%) in myelofibrosis and 1/6 (17%) in AA. Among the 24 patients with poor prognostics mutations, the conclusion was to proceed to an allogeneic stem cell transplantation for 12 of them and to start a hypomethylating agent in another one (Table 4). Among these patients, ASXL1 was the most frequent poor prognostic marker with therapeutic impact (n=11). Conversely, SF3B1, a mutation associated with a good prognosis, was identified in six cases of MDS and one of MPN/MDS. In five patients, this information led to a de- escalation of treatment intensity, i.e., allogeneic stem cell transplantation was postponed because of the presence of a good prognostic mutation (SF3B1) or absence of poor prognostic mutations.
Of note, VUS were observed in 18/95 cases, three being detected in RUNX1, thereby potentially conveying a poor prognosis.
Discussion
The interest of high-throughput NGS has been widely demonstrated in onco-hematology, mostly in clinical tri- als, but evaluation of these new molecular data outside clinical research is lacking. In the same way as for the usage of a drug, it appears essential to evaluate the medico-economic impact on clinical decisions of the use of these new tools, which are modifying the workload and costs in diagnostic laboratories.24 Nevertheless, the use of high-throughput sequencing to aid diagnosis and treat- ment decisions in chronic myeloid malignancies has sel- dom been evaluated in “real life”. Here, we retrospectively examined the impact of NGS assays on a series of 177 patients from ten centers. The impact on diagnosis and therapeutic decisions were assessed separately by dividing the patients into two groups.
The main objective of NGS for diagnostic purposes is the search for somatic mutations to provide or not proof of clonal hematopoiesis in favor of myeloid malignancies when standard diagnostic criteria for the disease have not been reached with classical tests (cytology, cytogenetics, pathology). ASXL1, DNMT3A and TET2 mutations are the most frequent but can be age-related25 and observed in the hematopoietic cells of apparently healthy older individu- als without MDS i.e., clonal hematopoiesis of indetermi-
nate potential (CHIP).26 Consequently, the interpretation of such molecular abnormalities in subjects over 70 years of age requires a multidisciplinary discussion. Clonal hematopoiesis was identified here in 33% of the patients and a firm diagnosis of myeloid malignancy was retained in 18%. The diagnosis of CCUS was reached in 9% of the patients and had an impact on the continuation of clinical follow-up because there was a higher risk that these patients would develop a myeloid neoplasm.5,27 Conversely, because more than 85% patients with MDS have one or more somatic mutations that can be detected using a minimal panel of recurrently mutated genes,1 the absence of mutation provided significant help to exclude a diagnosis of MDS in the vast majority of cases. In the con- text of aplasia, the absence of clonal hematopoiesis and the presence of a PIGA mutation (mostly associated with AA) helped to exclude a diagnosis of hMDS.16,23
Overall, in this series, the search for clonal hematopoiesis by NGS was useful for 83% of the patients, allowing a diag- nosis to be confirmed (33%) or excluded (50%). Nevertheless, it is essential to integrate these results with clear diagnostic criteria.28 The absence of detection of clonal hematopoiesis could interestingly help to exclude MDS, hMDS or mixed MPN/MDS but not MPN.
Molecular studies are also interesting to guide treatment decisions, especially through prognostic evaluation. In group B, 33% of the patients had a molecular anomaly with a reported prognostic impact. Treatment was modi- fied for 19%, in most cases with a reinforcement of ther- apy and/or a decision to perform allografting.
Integration of a search for molecular abnormalities by NGS is thus shown here to have a clear impact on patients’ therapeutic management provided that it com- plements a thorough diagnostic algorithm and multidisci- plinary indication. Currently, decisions on therapeutic intensification are rarely based on the integration of molecular data given the lack of international consensus.
These new tools also reveal the presence of VUS for which the somatic or constitutional origin as well as the pathogenicity are still unknown. This may be problematic when such VUS are observed in genes whose mutations carry a prognostic impact. Our data show that in the cases of VUS in RUNX1, patients were not ultimately consid- ered as being at high risk. Non-hematopoietic DNA test- ing and updating of international databases for the inter- pretation of these VUS are required. Among patients for whom treatment has not been modified, a minority was found to harbor a high-risk mutation and a majority of cases lacked risk-conferring mutations. Again, NGS results cannot be considered individually but should be integrat- ed with the usual IPSS-R markers.29
In summary, we report a benefit in terms of diagnosis and therapeutic impact for, respectively, 83% and 19% of the patients in a large real-life cohort. This outcome is interesting given that decisions to run these tests were made in a critically explored context and not in a system- atic fashion. The medico-economic implications of such integrative and multidisciplinary prescription and analysis deserve specific, thorough investigations. The latter would involve a calculation of the savings generated by promptly stopping investigations and alleviating treatment after rul- ing out a neoplasm, including the reduced costs generated by the decision of not performing an allograft. This should be balanced by the cost of adapted therapy resulting from the positive identification of a neoplastic disease. Such a
706
haematologica | 2021; 106(3)