A common pattern of somatic mutations in <scp>t‐MDS</scp>/<scp>AML</scp> of patients treated with <scp>PARP</scp> inhibitors for metastatic ovarian cancer

Abstract

Poly ADP-ribose polymerase inhibitors (PARPi) are emerging agents used as maintenance therapy in platinum-sensitive recurrent ovarian cancer (OC), increasing progression-free survival. PARP activity, together with high fidelity double-strand break homologous recombination repair and nonhomologous end-joining repair, is crucial in cell survival by fixing DNA breaks. BRCA1 and BRCA2 mutant neoplastic cells largely depend on PARP-mediated DNA repair mechanism to survive: PARPi (Olaparib, Rucaparib, Niraparib, and Talazoparib) act through synthetic lethality mechanism, producing simultaneous defects in both enzymes. PARPi were first employed in the BCRA-mutated tumors, inhibiting the only still functional DNA repair pathway and sensitizing tumoral cells to DNA breaks induced by other chemotherapy agents or radiation. Therefore, they have the unique ability to selectively kill those cells that have lost the homologous recombination repair pathway. PARPi share a similar toxicity profile. In particular, thrombocytopenia (3.3%), neutropenia and/or leukopenia (1.9% each), anemia (1.4%), and pancytopenia (0.8%) are the most common causes of dose modification or treatment interruption, but the precise mechanisms of hematological toxicity remain largely unknown. Finally, the most life-threatening adverse effect of PARPi is the onset of therapy-related myelodysplastic syndrome/acute myeloid leukemia (t-MDS/AML) still remaining this point under investigation. In this article, we reported 13 consecutive patients affected by OC and treated with PARPi who developed t-MDS/AML. From June 2019, we systematically offered a hematological consultancy for patients treated with PARPi for OC who developed grade 2–4 hematological toxicity. The overall cohort consisted of 300 patients; we investigated those developing complete blood count (CBC) alterations (anemia, leucopenia, or thrombocytopenia), searching for possible causes of cytopenia including vitamin decreased levels, autoimmune diseases, and use of concomitant drugs possibly responsible of hematological toxicity. In those patients with anemia and/or thrombocytopenia G2-4 (according to NCI CTCAE version 5.0) and/or leucopenia G3-4 without an identified reason for persistent cytopenia after temporary discontinuation of PARPi treatment for at least 4 weeks, we performed bone marrow evaluation including morphology, cytogenetics, and NGS for somatic mutations after patients signed informed consent (Table S1). Diagnosis of therapy-related AML/MDS (t-AML/MDS) was performed according to the WHO 2016 criteria. Between June 2019 and January 2022, 34 consecutive patients developed unexplained cytopenia during treatment with PARPi (11.3% of the patient cohort), and were submitted to bone marrow evaluation. In this subgroup, the median duration of PARPi was 15 months (range 1–27) and among them, we identified 13 patients with t-AML/MDS. At the time of PARPi therapy, all patients had a diagnosis of metastatic OC BRCA-mutated. The patients' characteristics are described in Table 1. Median age at OC diagnosis was 53 years (range 42–67), the median number of chemotherapy lines before PARPi treatment was 2 (range 1–4), and the median time between PARPi treatment and the onset of t-AML/MDS was 12 months (range 3–25). Eight out of thirteen patients had been exposed to Olaparib and five to Niraparib. Diagnosis was t-MDS in eight patients, and t-AML in five. According to conventional karyotype we found a complex karyotype in five of them, a hypodiploid pattern in three, a hyperdiploid chromosomal pattern in two patients, del5q in one, and 5q-/-7 in one. We did not find any recurrent molecular alterations including NPM1 mutations or FLT3-ITD in t-AML patients. We found pathogenic mutations by NGS in all patients. In particular, NGS revealed TP53 mutations in all patients. More frequently, we found the following TP53 mutations: c.376-1G>A (splice acceptor), c.78delT and c.524G>A. The median variant allelic frequency was 25% (15–55). Moreover, in two of them, we studied the ovarian tissue collected at cancer diagnosis and the TP53 mutations were absent. Interestingly we found mutations of the following genes: TET2 (in five patients), DNMT3A (in four patients), EZH2 and RUNX1 (in two patients), IKZF1, CSF3R, IDH1, and KIT (in one patient each). Considering TET2 and DNMT3A together, we identified one mutation in 9 out of 13 patients (69%) (Figure S1). At the time of diagnosis of myeloid malignancy, 12 out of 13 patients were in remission for OC; they underwent treatment with demethylating agents azacytidine or decitabine (n = 6), standard chemotherapy (n = 3), upfront allogeneic hematopoietic stem cell transplant (alloSCT) (n = 1), or hematological support (n = 2). After treatment with demethylating agents, one patient was subsequently submitted to alloSCT. One patient instead was considered only for palliation due to OC progression. At the end of the follow-up, the two patients who had received alloSCT were alive but both of them relapsed shortly after the transplant. Eleven patients died for progression of sAML/MDS (n = 7), OC progression (n = 2), or complications related to the treatment of the hematological malignancy (n = 2) after a median of 5 months (range 0–6) from diagnosis of hematological malignancy. The recent demonstration that hematological malignancies may occur in patients treated with PARPi has been raising a lot of interest and concerns. Recently Morice et al.1 reported that PARPi therapy significantly increased the risk of MDS and AML, being in the randomized trials the incidence 0.47% across placebo arms and 1.21% across non-placebo arms. However, patients with recurrent, platinum-sensitive, OC treated with PARPi, had a gain in overall survival, so it cannot be excluded that the higher incidence of t-MDS/AML is related to the prolongation of overall survival giving time for development of second malignancy after exposure to multiple lines of chemotherapy. Oliveira et al. in a period of 6 years identified nine patients who developed a myeloid malignancy over a cohort of 583 PARPi-treated patients. They found cytogenetic abnormalities including complex karyotype and monosomal karyotype with most common abnormalities in chromosome 5 and 7 (5q- and -7). Moreover, in a part of patients they performed also mutation profiling by NGS, finding that TP53 was the most common involved gene, suggesting a role in the development of PARPi-associated myeloid malignancies.2 Navitski et al. described three OC patients who had possible baseline mutations related to the risk of hematological malignancies. The patients had mutations in DNMT3A and TET2, the most commonly mutated genes in clonal hematopoiesis (CH).3 Martin et al. studied 12 OC patients treated with a combination of chemotherapy and PARPi, who developed cytopenia but not myeloid neoplasm, finding CH in 77% of them. Moreover, they studied 17 patients with PARPi-related hematological malignancies finding an unfavorable cytogenetic pattern in 94% of them and a high prevalence of mutations in TP53 gene (75%).4 In a recent meta-analysis on randomized trials in the front-line setting, PARPi therapy was associated with developing MDS/AML (IRR 5.43, 95% CI 1.51–19.60). On the other hand, in patients treated for recurrence, the risk of MDS/AML was similar in either patients randomized to PARPi or control treatment.5 In this article, we reported 13 consecutive patients with OC BRCA-mutated developing hematological myeloid malignancies, over a cohort of 300 patients with OC treated with PARPi (4.3%). According to our data, we can draw several considerations. The incidence of this complication was higher as compared to data already published, including the meta-analysis from Morice et al. This could be explained by the large size of our series of patients treated with PARPi. Moreover, in our cohort under strict hematological follow-up, patients were characterized by having some common features: unfavorable karyotype, the presence of a TP53 mutation (100%), and mutations of TET2 and/or DNMT3A in 9 out of 13 (69%). DNMT3A and TET2 are epigenetic modifiers and, according to literature, the oncogenicity of their mutations has been demonstrated to be pre-leukemic mutations while TP53 mutations may are associated to an increased incidence of subsequent hematologic cancers. The association of several chemotherapy lines and PARPi maintenance may have induced their onset; however, we are not able to establish the exact role of chemotherapy and PARPi. Despite this, the presence of TP53 mutations confers not only a homogenous molecular pattern in these patients, in agreement with Todisco et al.,6 but also a worse prognosis in patients surviving OC. In our series, in fact, the majority of patients were refractory to therapy and only two patients achieved complete remission after alloSCT. Unfortunately, both of them relapsed after transplant at a median time of 10 months. In conclusion, we confirm that there is a relation between the use of PARPi after chemotherapy and the development of t-AML; the incidence of secondary hematological malignancies could be underestimated in this setting in the absence of a careful investigation of hematological toxicity. On the other hand, the exact role of PARPi in the subsequent development of these malignancies is difficult to dissect considering the prior exposure to antineoplastic agents and the prolongation of survival as a result of PARPi efficacy. As described, we found a common pattern of molecular features in these patients sharing TP53 mutations and a high rate of mutations in DNMT3A and TET2. According to the current literature including our series, patients with OC treated with PARPi and who experience unexplained hematological toxicity should be followed with watchful hematological monitoring to recognize earlier signs of clonal hematological evolution. This matter remains stringent considering that the use of PARPi is going to be anticipated in the clinical practice in OC setting associated to chemo or radiotherapy and extended to other cancers including breast and prostate. Patrizia Chiusolo, Claudia Marchetti, Giovanni Scambia, and Simona Sica designed research, Monica Rossi, Gessica Minnella, Angelo Minucci, Maria Colangelo, and Daniela Orteschi performed research and analyzed data, Valerio De Stefano, Mariagrazia Distefano, Sabrina Giammarco, Elisabetta Metafuni, Filippo Frioni, Cristiana Gasbarrino, Anna Fagotti, Domenica Lorusso, Livio Pagano, and Valerio De Stefano treated the patients, Patrizia Chiusolo, Claudia Marchetti, Valerio De Stefano, and Simona Sica wrote the article and all co-authors participated in the critical revision of the manuscript. This work was supported by the “Centro di Ricerca sulle cellule staminali emopoietiche e le terapie cellulari” -Università Cattolica del Sacro Cuore, Roma. The authors have no conflict of interest. Figure S1 Somatic mutations detected at the time of tMDS/AML in ovarian cancer patients treated with PARPi Table S1 List of genes analyzed 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.

Topics & Concepts