Germline variants in patients developing second malignant neoplasms after therapy for pediatric acute lymphoblastic leukemia—a case-control study

Le résumé fourni par la source

Today, most children treated for acute lymphoblastic leukemia (ALL) can be cured by the application of intensive combination chemotherapy regimens [ 1 , 2 ]. However, within 20 years, up to 10% develop a second malignant neoplasm (SMN) with cure rates often being dismal [ 3 , 4 ]. Therefore, strategies are needed for an early identification of patients at risk for SMN development, an improved understanding of the underlying pathobiology and, ideally, the development of preventive actions. In this study, we analyzed genetic variation in 159 selected cancer predisposition genes (CPG) to determine their potential association with SMN. Patients included were 1–22 years of age at diagnosis of ALL and enrolled in the German part of the European AIEOP-BFM ALL 2000 multicenter clinical trial for frontline treatment of pediatric ALL [ 5 ]. Median follow-up for the entire patient group was 14.8 (4.6–19.2) years as of November, 2020. Treatment details were described previously [ 6 ]. We identified 75 SMN patients with sufficient DNA available, representing 94% of all 80 SMN cases observed in the entire AIEOP-BFM ALL 2000 study population up to this follow-up date. In a matched case-control approach, we analyzed germline DNA obtained from remission bone marrow samples of 223 individuals (75 cases, 148 controls); see Supplementary Methods section for sequencing details and Supplementary Table 1 for clinical background information. Matching criteria were sex, age at diagnosis of ALL, immunophenotype, and treatment. Candidate genes included genes based on own work and/or genes recommended by Byrjalsen et al. 2021; [ 7 ] variant pathogenicity was assessed according to standard interpretation guidelines (see Supplementary Table 2 for details). Statistical analyses and plotting were conducted with SAS (SAS-PC, Version 9.1, Cary, NC: SAS Institute Inc.), SPSS (IBM Deutschland GmbH, Ehningen, Germany) or R (4.0.5), and Rstudio (1.0.153); see Supplementary Methods section. Overall we observed 31 conspicuous findings in 22 genes: 17 (6 likely pathogenic and 11 pathogenic) in 16 (21%) cases with ALL and subsequent SMN and 14 variants (9 likely pathogenic and 5 pathogenic) were detected in 14 (9%) control patients with ALL; for details see Table 1 and Fig. 1 . While some variants were found in patients from both groups (e.g., rs28909982 in CHEK2 , rs113993993 in SBDS and rs764079291 in NF1 ), certain genes were exclusively altered in SMN patients (e.g., ATM, TP53, A2ML1, MSH6 and PMS2 ). Table 1 (Likely) pathogenic variants determined in patients with acute lymphoblastic leukemia with or without second malignancies. Full size table Fig. 1: Characteristics of second malignant neoplasms (SMN) in the 75 included patients from trial AIEOP-BFM ALL 2000. A Entities of SMN; ( B ) time from diagnosis of the acute lymphoblastic leukemia (ALL) to SMN development; median time to SMN was 3.36 ± 3.98 (0.57–13.37) years for patients with hematologic SMN, 8.61 ± 1.69 (5.12–11.43) years for patients with second brain tumors, and 8.45 ± 4.04 (1.92–15.91) years for patients with other second solid tumors. Black-bordered dots represent patients with a (likely) pathogenic variant (LP/P) in one of the candidate genes; the adjacent numbers correspond to the respective patient IDs (for further details, compare Table 1 and Supplementary Tables 3, 4 ). Abbreviations: MDS myelodysplastic syndrome, AML acute myeloid leukemia, CML chronic myeloid leukemia, PTLD post-transplant lymphoproliferative disorder. Full size image Among the SMN patients, (likely) pathogenic variants in one of the candidate genes were observed in 3 of 11 (27%) cases with brain tumors, 5 of 28 (18%) with other solid tumors, and 8 of 36 (22%) with hematologic SMN. Three individuals with hematologic SMN carried variants in Ras signaling pathway genes ( LZTR1 and KRAS ); for clinical details see Supplementary Table 3 . One of these patients, determined with LZTR1 variant rs869320686 p.(Gly248Arg), developed a myelodysplastic syndrome (MDS) 2.1 years after undergoing hyperdiploid pre B-cell ALL. Revision of the clinical reports revealed that this patient was diagnosed with mild developmental delay, facial anomalies, and a short stature. Suspected clonal hematopoiesis was confirmed for KRAS variant rs121913530, by analyzing hair follicle-derived DNA. Two male patients, identified with astrocytoma at 9.2 and 9.7 years after diagnosis of ALL, harbored deleterious variants in ATM . Two SMN patients carried variants in mismatch repair genes: one patient with a mucoepidermoid carcinoma ( MSH6 ) and one individual with an astrocytoma ( PMS2 ); the later also carried an ATM variant. One male patient with a melanoma 16 years after his pre B-cell ALL carried a variant in A2ML1 , rs773034576. While conducting this study, the clinical significance of A2ML1 variants related to the Noonan syndrome was refuted [ 8 ]. However, for this patient, this likely pathogenic variant was the only one we determined. The TP53 variant, rs121912651, was observed in a female patient developing MDS at 8.8 years after diagnosis of ALL. Two observations, Pro470Valfs*3 ( PMS2 ) and rs587776650 ( NBN ) were homozygous. All other variants were detected with a variant allele fraction of approximately 50% (i.e., heterozygous); for further details see Supplementary Tables 3 and 4 . Excluding the variants KRAS rs121913530 and A2ML1 rs773034576, (likely) pathogenic variants in one of the CPG were 1.97-times more frequent in SMN patients (14/148 vs. 14/75 affected patients; P ( Pearson’s X 2 ) = 0.050). Conditional Cox-Regression analysis revealed that patients with a conspicuous finding in one of the candidate genes were at increased risk for SMN development (OR = 2.21, 95% CI = 0.989–4.944), P = 0.053 compared to the remaining patients. Within the last decade, it became increasingly obvious that germline genetic variation plays an important role in the development of childhood cancers, including pediatric ALL. Moreover, cases of ALL plus subsequent SMN were reported in the context of familial cancer predisposition syndromes (CPS) such as the Li-Fraumeni syndrome or the ETV6 deficiency [ 9 , 10 ]. Previously published large-scale sequencing studies, including pediatric ALL patients, determined 4–5% of the patients with variants in CPG [ 11 ] or an underlying CPS [ 12 ]. In line, here we identified 14 (9%) of the ALL patients without SMN with unfavorable CPG variants, and in 6 of these cases (4% of the control patients) an underlying CPS may be suspected, either because the variant was determined to be homozygous (i.e. NBN) or due to an autosomal dominant inheritance mode of the associated phenotype (see Table 1 for details). Notably, this was also the case in 9 (12%) of the SMN patients in our study. It is important to note that patients with a monoallelic/heterozygous variant in a gene that is associated with an autosomal recessive disease only have a carrier status for this condition. Furthermore, the aim of this study was not to diagnose previously unrecognized CPS, but to describe the mutational landscape of variants in known CPGs and to identify potential indicators of increased SMN risk in pediatric ALL patients. Nevertheless, in addition to a specific variation, we identified some patients with additional characteristics consistent with features of suspected CPS: In particular, the male patient carrying the LZTR1 variant rs869320686 p.(Gly248Arg), previously described to be associated with the Noonan syndrome [ 13 ], presented several additional features [ 14 ]. In contrast, other variants we observed were previously reported in the context of SMN development, e.g., the mutational hotspot TP53 variant, rs121912651, which was previously observed in B-cell ALL patients with SMN [ 9 ]. Using remission bone marrow samples, we cannot completely exclude clonal hematopoiesis here. Hence, besides considering detailed individual clinical and familial information, an adequa

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.