Validation of Guidelines for Genetic Investigation of Myeloid Neoplasms with Germline Predisposition: Results from a Prospective Cohort Study

Abstract

Purpose:

In a multicenter prospective cohort study, we assessed the diagnostic yield of the Nordic guidelines for germline investigation in myeloid neoplasms and mapped the spectrum of inherited and somatic variants.

Experimental Design:

Eighty-five patients (acute myeloid leukemia, n = 38; myelodysplastic syndromes, n = 26; thrombocytopenia, n = 14; and other, n = 7) fulfilling the Nordic criteria for germline investigation, based on (i) medical history or family history suggestive of a germline condition and (ii) relevant findings from the somatic diagnostic work-up (CytoMol), were recruited. The genetic analysis included enhanced whole-exome sequencing (n = 69) or sequencing of specific variants of interest (n = 16).

Results:

Pathogenic or likely pathogenic (P/LP) germline variants were identified in 35% of patients (30/85). The diagnostic yield varied from 6% (1/16) in the family history group to 52% (17/33) in the CytoMol group. Germline DDX41 P/LP variants were the most frequent finding (13/30, 43% of all positive cases) almost exclusively found within the CytoMol group (12/13). Seven variants of unknown significance were also detected (TERT n = 2 and DDX41, RTEL1, ETV6, PARN, and SAMD9 n = 1). Five patients carried a P/LP variant in genes associated with another hereditary cancer syndrome (BRCA1 n = 3; PALB2 n = 1; and CHEK2; n = 1). Survival analysis showed a trend for longer survival among patients with acute myeloid leukemia and confirmed or suspected germline predisposition that underwent allogeneic stem cell transplantation.

Conclusions:

The implementation of the Nordic guidelines in a prospective Swedish cohort results in a high overall diagnostic yield (35%), proving the feasibility and utility of these or similar guidelines in a clinical setting.

Translational Relevance.

To the best of our knowledge, this study is the first prospective validation of guidelines for the management of patients with suspected predisposition for myeloid neoplasms, mainly acute myeloid leukemia and myelodysplastic syndrome, performed in parallel with the clinical praxis. The Nordic guidelines for myeloid neoplasms with germline predisposition demonstrate high diagnostic yield, reaching up to 53% depending on the applied criterion. Germline investigation requires comprehensive genetic analysis and close collaboration of clinicians and genetic counselors. The spectrum of germline variants in hereditary acute myeloid leukemia and myelodysplastic syndrome is heterogeneous and varies depending on the criterion that motivates the germline investigation. Our findings suggest that germline investigation based on concrete guidelines should be widely offered for patients with myeloid neoplasms.

Introduction

Germline genetic variants are increasingly recognized as contributing factors to the development of myeloid neoplasms (MN) such as myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML; refs. 1, 2). Recent studies suggest that up to 13% of patients with MDS and AML harbor a germline predisposition to MN (3, 4). However, the majority of cohort studies published thus far are retrospective in nature and therefore limited by sample availability or lack of uniformity in genetic investigation protocols (3–6).

Since the recognition of MN linked to germline predisposition as distinct entities in the 2016 World Health Organization classification of hematologic malignancies (7), it has become important to accurately identify and diagnose these patients within standard clinical care. To facilitate this process, the Nordic MDS working group on MN with germline predisposition published criteria for germline testing in patients with MDS/AML and related conditions in 2019 (8). Similar guidelines have been proposed by the European Leukemia Network, the European Society for Blood and Marrow Transplantation, the UK Cancer Genetics Group, and the National Comprehensive Cancer Network (refs. 9–12). Results from the implementation of the National Comprehensive Cancer Network guidelines in a single center were recently presented (13). However, real-world validation of any of these guidelines in multi-institutional substantial cohorts is still lacking.

We report the results from a prospective Swedish multicenter study in a cohort of patients with MN, predominantly MDS and AML, recruited based on the criteria for germline testing proposed by the Nordic MDS working group on MN with germline predisposition.

Materials and Methods

Study design and participants

A multi-institutional ethical approval to conduct the study was obtained by the Swedish Ethical Board (Dnr. 2019-04188 and Dnr. 2022-02571-02) in accordance with the Declaration of Helsinki. All patients included in the study provided their written informed consent. Patients were identified by treating hematologists at both university and regional hospitals in Sweden during October 2019 to January 2023 based on the published Nordic guidelines (8). All included patients were referred to the Department of Clinical Genetics to receive information about the study, pre-test genetic counseling, and collection of a three-generation family history (FH). The diagnostic yield was expected to be up to 20% based on the current experience from the field of solid tumors (14). Patients were divided into two main study groups (Fig. 1A): The first group, called “MH/FH,” refers to patients with medical history (MH) or FH suggestive of germline predisposition. For inclusion in this group, patients had to fulfill one of the following criteria: (i) for “MH,” patients with MDS/AML or bone marrow failure with other symptoms suggesting an underlying syndrome and being younger than 50 years of age at diagnosis; (ii) for “FH,” (a) patients with MDS/AML or other related hematologic phenotype (e.g., cytopenia/s) with a first-degree (FDR) or second-degree relative (SDR) with MDS/AML or other related hematologic phenotype and (b) patient with MDS/AML or other related hematologic phenotype with two FDR or SDR with cancer. In both cases, at least one of the affected members in the family (patient or relative) had to be younger than 50 years of age at diagnosis; and (iii) strong FH (S-FH): three close relatives with MDS/AML or other related hematologic phenotype, regardless of age at diagnosis. The second group, called “CytoMol,” included patients with MDS/AML in which (i) the tumor-only somatic mutation panel analysis performed as part of the diagnostic work-up detected variants suspected to be germline based on variant allele frequency (VAF) above 40% in known genes for hereditary hematologic malignancies (HHM) regardless of age and (ii) aberrations of chromosome 7 in patients with MDS/AML younger than 50 years of age at diagnosis. Patients fulfilling more than one inclusion criterion were assigned to a main study group according to the following prioritization: FH > MH > CytoMol. This prioritization was applied as the inclusion criteria were decided at the time of inclusion, which for many of the patients co-occurred with the time of diagnosis when the results of the somatic profiling were not yet known. Genetic findings from the study were reported back to all patients and their treating physicians.

Figure 1.

Overview of the study cohort. A, Patient enrollment and subcohorts. B, Sankey diagram showing the referral paths of the 85 patients from eight Departments (Dept) of Hematology to five Departments of Clinical Genetics. The number of patients from each department is indicated. CytoMol, inclusion due to findings in the somatic profiling.

Statistical analysis

All statistical analyses were performed using R Project for Statistical Computing (RRID: SCR_001905; v4.3.3; R Core Team 2024). A Wilcoxon rank-sum test was performed for pairwise comparison of age at inclusion. Survival time was calculated from the day of diagnosis to death or the last day of follow-up. Overall survival (OS) probability was estimated using the Kaplan–Meier method and comparisons between subgroups were conducted using the log-rank test. HR with 95% confidence intervals (CI) were calculated using Cox proportional hazards regression. For patients with AML, a landmark analysis was conducted with a landmark time set at 5 months after diagnosis to study the impact of allogeneic stem cell transplantation (allo-HSCT) on survival (15). This time-point was chosen since, by that time, three-fourths of the transplanted patients had undergone allo-HSCT. Two patients were excluded because of death before the landmark, whereas two patients transplanted after the landmark were regarded as not transplanted.

Procedures

Germline DNA was isolated from the following sample types according to standard procedures with the automated QIAsymphony SP or EZ1 systems from Qiagen (RRID: SCR_008539): cultured fibroblasts from skin biopsy (n = 58), blood samples in complete remission (n = 4) or biobanked before study inclusion because of other genetic investigations (n = 3), and T cells from peripheral blood (n = 9). In patients with nonmalignant disease and without clonal events in the somatic gene panel and/or cytogenetic analysis, peripheral blood was used as the source of germline DNA (n = 11; Supplementary Fig. S1).

Enhanced whole-exome sequencing (WES) included WES with copy-number analysis [copy-number variants (CNV)] and targeted sequencing of selected noncoding regions of interest. WES libraries were prepared from 50-ng genomic DNA using the Twist Comprehensive Exome Panel from TWIST Bioscience (RRID: SCR_025817). The libraries were sequenced with High Output kits on NextSeq 500/550 sequencers from Illumina (RRID: SCR_010233), producing 150-bp paired-end reads. Demultiplexing was performed using bcl2fastq (RRID: SCR_015058), and alignment of sequence reads to the human reference GRCh37.75 was performed using BWA (RRID: SCR_010910) v0.7.17. The alignments were deduplicated using Picard (RRID: SCR_006525) MarkDuplicates, followed by single-nucleotide variant (SNV) and indel calling using GATK (RRID: SCR_001876) Haplotypecaller as implemented in bcbio-nextgen (https://github.com/chapmanb/bcbio-nextgen). CNV were called using CoNIFER (RRID: SCR_013213) and ExomeDepth (RRID: SCR_002663; refs. 16, 17). For SNV analysis, OMIM morbid genes were reviewed (18). For CNV analysis, an in silico panel of 217 genes (Supplementary Table S1) was used. At the end of the study, all samples were reanalyzed for the occurrence of SNV and CNV in novel candidate genes published during the study period (Supplementary Table S2). Variant annotation and analysis were performed using the Alissa Interpret software (v5.2.10; Agilent Technologies RRID:SCR_013575). Alignments around candidate variants were visualized using the Integrative Genomics Viewer (RRID: SCR_011793). Exonic variants and variants within 10 bp from exon boundaries were included in the analysis. Variant classification was performed using the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) variant interpretation sequence guidelines as well as gene-specific guidelines proposed by the Clinical Genome Resource (19–21).

Noncoding genetic regions of interest were included in a TWIST Bioscience (RRID: SCR_025817) custom-designed myeloid sequencing panel covering genetic regions across 191 genes used for diagnostics of myeloid disorders. The noncoding targets included TERC (chr3:169482182-169483654), ANKRD26 (chr10:27389007-27389433), and GATA2 (chr3:128201827-128202419). The same panel was used for following up specific variants of interest identified in patients (n = 16) from the CytoMol group. qPCR was used to measure relative telomere length in selected patients (22).

Results from genetic investigations performed on diagnostic samples to detect somatic aberrations were retrieved from the medical records. For patients who underwent routine somatic investigation before June 2020 (n = 26), the somatic investigation was performed with an Illumina (RRID: SCR_010233) TruSight Myeloid panel, which did not include DDX41, whereas patients included afterward were tested with a national panel that included DDX41 (23). As a tumor/normal approach is followed at one university hospital in Sweden, patients with at least one somatic variant in DDX41 detected through this approach were included in the study in the CytoMol group.

Cancer diagnoses in relatives were confirmed through the Swedish Cancer Registry or medical reports whenever possible. Relevant clinical information was retrieved from the medical records.

Data availability

Genetic data for this study were generated and deposited at Clinical Genomics Uppsala, Science for Life Laboratory, Department of Immunology, Genetics, and Pathology, Uppsala University, Sweden (project number CGU_2018_16). Both genetic and clinical data are not publicly available because of the Swedish legislation but are available from the corresponding author (panagiotis.baliakas@igp.uu.se) upon request. Requests should reference the project number “CGU_2018_16.”

Results

Patient characteristics

Between October 2019 and January 2023, 85 sequential unrelated patients were included in the study (Fig. 1A). The patients were treated in eight different hematologic centers throughout Sweden, whereas genetic counseling was performed at five university hospitals (Fig. 1B). The majority of patients (n = 72, 85%) were recruited at hematologic centers from the Stockholm and Uppsala region. The median age at inclusion was 47 years (IQR, 37–58; with a male-to-female ratio of 1:1.1). Most patients were diagnosed with either AML (38/85, 45%) or MDS (26/85, 31%), whereas the remaining patients had persistent thrombocytopenia (14/85, 16%) or other diagnoses (7/85, 8%), such as clonal cytopenia of uncertain significance or aplastic anemia (Fig. 2A). Fifty-two patients (52/85, 61%) were included in the MH/FH groups, whereas the rest (33/85, 39%) were included in the CytoMol group (Fig. 2B). Fifteen patients (15/85, 18%) fulfilled more than one inclusion criterion. The median age at inclusion was highest for the CytoMol group (Fig. 2C). Patients with AML and MDS were included in all four groups (Fig. 2C) and represented the majority (31/33, 94%) of patients in the CytoMol group. In contrast, most patients with thrombocytopenia were included in the S-FH group (11/14, 78%; Fig. 2C). Within the S-FH group, nine patients (9/18, 50%) belonged to families with isolated thrombocytopenia, whereas the rest had FDR and/or SDR with myeloid malignancies. The most common reason for inclusion in the CytoMol group was the detection of a DDX41 variant on somatic gene panel testing (14/33, 42%), followed by a variant in RUNX1 (10/33, 30%). Three patients (3/33, 10%) were included because of the presence of monosomy 7 or deletion of 7q and <50 years of age at diagnosis.

Figure 2.

Characteristics of included patients. A, Hematologic disease at inclusion. B, Fulfilled criteria for germline testing. C, Age at inclusion for each inclusion criterion. Individual patients are depicted according to their hematologic disease at diagnosis using different colors and symbols. TP, thrombocytopenia.

Germline findings

Sixty-nine (69/85, 81%) patients, including all but one patient from the MH/FH group, and 18 patients from the CytoMol group, were analyzed by WES, whereas the remaining patients (16/85, 19%) underwent direct testing for a specific variant of interest detected during somatic genetic investigation (Supplementary Fig. S1).

Germline findings in established genes associated with hereditary blood disorders

A pathogenic or likely pathogenic (P/LP) germline variant in an established gene for HHM, inherited bone marrow failure syndromes, or hereditary thrombocytopenia was identified in 30 patients (30/85, 35%; Fig. 3A; Supplementary Table S3). The diagnostic yield varied depending on the applied criterion, from 6% (1/16) in the FH group to 52% (17/33) in the CytoMol group (Fig. 3B). Among patients fulfilling ≥2 criteria for germline testing (n = 15), the diagnostic yield was 53% (8/15; Fig. 3C). Twelve of these 15 patients fulfilled the CytoMol criterion, which was considered the main inclusion criterion in six of them according to the study prioritization. Reassignment of all these 12 patients to the CytoMol group did not significantly alter the diagnostic yield per criterion with the exception of the MH criterion (diagnostic yield of 12.5% instead of 22%; Supplementary Fig. S2). In the FH group, the only germline P/LP finding was an LP variant in the TERT gene (GH01_P27). In the MH group, a germline P/LP finding was identified in four of 18 patients (4/18, 22%) and included variants in RUNX1 (n = 1), PARN (n = 1), SBDS (n = 2), and DNAJC21 (n = 2; Fig. 3D). The variants in SBDS and DNAJC21 were biallelic (two heterozygous in trans variants and one homozygous, respectively) in accordance with a recessive hereditary pattern (Supplementary Table S3). In the S-FH group, a P/LP finding was identified in eight of 18 patients (8/18, 44%) and in six of nine patients (6/9, 67%) with family history for thrombocytopenia without hematologic malignancies. In the CytoMol group, 17 of the 33 patients (17/33, 42%) were confirmed to carry a germline P/LP variant. All 17 patients had a suspected germline finding detected using a somatic next-generation sequencing panel affecting either DDX41 (n = 12), CEBPA (n = 2), RUNX1 (n = 1), or GATA2 (n = 1), whereas no germline findings were confirmed among the patients included because of cytogenetic aberrations of chromosome 7 and younger than 50 years at inclusion.

Figure 3.

Germline findings in well-established genes for hereditary blood disorders. A, Diagnostic yield in the whole cohort and (B) criterion-based (C) UpSet plot summarizing fulfilled criteria and germline findings. The bar charts on top display the number of patients who fulfilled a single criterion (single filled-in dot below the X-axis) or a combination of criteria (filled-in dots connected by lines below the X-axis). D, Tile plot in which each column represents a unique patient with a germline variant identified. The rows present, from top to bottom, the mutated gene, the main inclusion criteria, the hematologic phenotype at inclusion, and the variant status, P/LP or VUS. Patients GH01_P51 and GH01_P73 carried homozygous variants in DNAJC21 and compound heterozygous variants in SBDS, respectively.

In total, 34 unique P/LP variants were identified, of which two [NM_016222.4(DDX41):c.936-1G>T and NM_001987.5(ETV6): c.1044G>C] were reported for the first time. The most frequently mutated gene was DDX41, which accounted for 43% (13/30) of P/LP findings across all criteria and for 70% (12/17) of findings in the CytoMol group (Fig. 3D). The detected DDX41 mutations consisted of 10 unique variants (premature protein termination, n = 4; in-frame deletions, n = 2; missense variants, n = 2; and splicing, n = 2; Supplementary Table S3). Five patients with germline DDX41 variants were diagnosed with AML, seven with MDS, and one with cytopenia that later evolved to MDS (Fig. 2D). Their median age at inclusion was 73 years (range, 23–85; Fig. 4A), with a male-to-female ratio of 2.3. All but one patient (GH01_P20) was included based on the CytoMol criterion. Interestingly, patient GH01_P13 developed AML at the age of 23 (Fig. 4A). P/LP variants in RUNX1 were observed in four patients, all of whom were diagnosed with AML (age range, 33–72 years). Two patients were included because of extensive family history fulfilling the S-FH criteria and typical for familial platelet disorder with associated myeloid malignancies (RUNX1-FDPMM), whereas the other two (GH01_P50 and GH01_P74) were included based on fulfillment of the MH or CytoMol criterion, respectively. The lack of a family history suggestive of RUNX1-FDPMM in GH01_P50 (AML at 33 years of age with preceding thrombocytopenia) could later be explained by a de novo origin of the variant. Preceding thrombocytopenia was documented in three of four patients with germline RUNX1 variants. Among patients included based on family history for thrombocytopenia, three P/LP variants were identified at the 5’ untranslated region of ANKRD26 (Figs. 3D and 4B), whereas two additional patients harbored P/LP variants in ACTN1. No cases of hematologic malignancies were reported in these families (Fig. 4B), but one patient with a germline ANKRD26 variant (GH01_P45) had clonal hematopoiesis detected at age 48 due to a somatic variant in PTPN11 with a VAF of 4%. Germline P/LP variants in CEBPA were detected in two patients from the CytoMol group. Both patients lacked family history for MN and developed AML at 55 years of age.

Figure 4.

Clinical characteristics and selected pedigrees. A, Boxplot of age at inclusion for the three diagnostic outcomes with regard to findings in established genes for hereditary blood disorders. “None” refers to patients without any findings. Actual data points are overlaid, their shape illustrating the hematologic phenotype at inclusion. Patients with germline findings in DDX41 genes are presented in red. A Wilcoxon rank-sum test was performed for pairwise comparison of age at inclusion. Adjusted P values are indicated above the boxplot. Ns, not significant. B, Pedigree of three patients belonging to families with inherited thrombocytopenia without hematologic malignancies harboring germline pathogenic variants in ANKRD26 (NM_014915.2). Affected individuals are represented by filled squares (males) and circles (females). The index case is denoted by an arrow. Number or “n” within the symbols represents the known or unknown number of relatives, respectively. P, Pathogenic; TP, thrombocytopenia.

Seven patients (7/85, 8%) carried a variant of unknown significance (VUS) in established genes for HHM and inherited bone marrow failure syndrome (Fig. 3A–D). Four of these (4/85, 4%) were found in genes associated with known telomere biology disorders (TERT n = 2; RTEL1 n = 1; and PARN n = 1). Further investigation of telomere length by qPCR on DNA from peripheral blood samples showed reduced relative telomere length compared with age-matched controls in two patients, thus suggesting a telomere biology disorder in these patients (Supplementary Fig. S3A and S3B). However, segregation analysis was neither possible nor informative enough to enable a reclassification of these two variants to LP/P. GH01_P3, a 65-year-old man with MDS whose sister developed MDS at age 53, carried a germline VUS in DDX41 as well as a somatic mutation in the same gene. If these three variants (TERT n = 2 and DDX41 n = 1) were to be reclassified as LP, the overall diagnostic yield in our cohort would increase to 39% (33/85).

Germline findings in candidate genes

In addition to established genes linked to HHM and related disorders, P/LP findings in genes associated to a predisposition for solid tumors were detected in five patients (BRCA1 n = 3; PALB2 n = 1; and CHEK2 n = 1). One of the patients (GH01_P50, AML at 33 years of age and preceding thrombocytopenia) also carried a pathogenic variant in RUNX1. A pathogenic variant in PALB2 was detected in a 40-year-old man with AML and family history of lipomatosis (GH01_91_S41). Unfortunately, segregation analysis was not possible, and therefore, we do not know whether the variant segregated with lipomatosis as previously reported in another family with lipomatosis and a germline pathogenic variant in PALB2 (24).

The remaining patients from the MH/FH group without P/LP findings were also assessed for the presence of rare variants in recently described candidate genes for HHM (Supplementary Table S2) but with no further findings.

Somatic findings

Results from the routine somatic gene panel were available for 62 of 64 included patients with AML or MDS (Supplementary Fig. S4). RUNX1 was the most frequently somatically mutated gene (19/62 patients, 31%), with at least one somatic variant detected in each of the 19 patients. This high frequency of somatic RUNX1 variants is partly explained by the inclusion criteria for the CytoMol group, as 10 patients were included because of RUNX1 variants with high VAF detected on somatic next-generation sequencing panels, of which only one (GH01_P74; 1/10, 10%) turned out to be germline. In two of the four patients with germline RUNX1 findings, a somatic second hit in RUNX1 was observed. Somatic DDX41 variants were observed in 11 patients (11/62, 18%), of which 10 had an additional germline P/LP or VUS in DDX41, whereas one patient did not carry any germline variants in DDX41. The DDX41 p.R525H was the most common second hit (6/10). Only two patients with germline DDX41 variants lacked a second hit in the same gene; patient GH01_P13, who developed AML at 23 years of age, carried a somatic TP53 variant, whereas no somatic variants were identified in the second.

Survival analysis

Complete follow-up data were available for 57 of the 64 included patients with MDS or AML. OS in patients with MDS (n = 23) and AML (n = 34) was 69% (at 32 months) and 73% (at 30 months), respectively (Fig. 5A). No statistically significant difference in OS was observed between patients with confirmed germline findings or suspected predisposition because of fulfilled MH/S-FH/FH criteria compared with patients without germline finding who only fulfilled the CytoMol criterion (Fig. 5B and C; Supplementary Table S4). Similarly, survival rates showed no significant difference when comparing patients with MDS and AML with confirmed germline findings to those with either a VUS or negative finding (Supplementary Fig. S5A and S5B; Supplementary Table S4). Among patients with AML with confirmed or suspected predisposition due to fulfilled MH/S-FH/FH criteria (n = 27), we observed better survival for patients who underwent allo-HSCT (HR, 0.22; 95% CI, 0.05–0.95; log-rank P = 0.027; Fig. 5D). Similarly, a statistical difference in survival was observed among patients with AML with confirmed germline predisposition who underwent allo-HSCT (log-rank P = 0.0042; Fig. 5E). However, a specific benefit of allo-HSCT for patients with germline predisposition could not be confirmed by either Cox proportional hazards regression (HR, 0.22; 95% CI, 0.01–4.46) or by landmark analysis with the landmark set at 5 months after diagnosis when the majority of patients with AML had received a transplant (Supplementary Table S4; Supplementary Fig. S6). Of the 18 patients with AML and confirmed or suspected predisposition who underwent allo-HSCT (median age at inclusion 51 years; IQR, 39–60), 17 received a graft from a matched unrelated donor, whereas one received a graft from a related haplo-identical donor. Previous data have suggested a higher incidence of acute GVHD in patients with hereditary DDX41 variants (25). Focusing on patients with germline DDX41 pathogenic variants, no increased incidence of GVHD was observed among the six patients with available data, with only two developing mild GVHD of the skin.

Figure 5.

Survival analysis. A, Kaplan–Meier OS curve for patients with MDS (blue) and AML (red) with available data (not censored for allo-HSCT). B, Kaplan–Meier OS curve for patients with MDS and AML (C) with suspected (MH/FH/S-FH patients without findings) or confirmed germline predisposition compared with patients only fulfilling the CytoMol criteria in which germline predisposition was excluded. D, Kaplan–Meier OS curve for patients with AML with suspected (MH/FH/S-FH patients without findings) or confirmed germline predisposition grouped on whether they received allo-HSCT. E, Kaplan–Meier OS curve for patients with AML grouped on whether they had germline predisposition confirmed through the study and allo-HSCT status. Survival probability on the y-axis and time in months after landmark on the x-axis. Comparison among subgroups was performed using the log-rank test, with the resulting P value presented within the plot.

Discussion

Recognition of germline predisposition in hematologic malignancies, and especially in MDS and AML, is one of the most significant recent advancements in the field of hemato-oncology. Yet, many clinical challenges remain. For instance, it is still unclear which patients should undergo germline testing, as for the time being, it is not feasible to perform germline investigations in all patients with MN. Several consensus statements and expert opinions have therefore been published during the last 5 years to provide guidance on patients that should undergo germline testing (8–12). We present the results of a Swedish prospective observational study aiming to validate the Nordic guidelines in patients with suspected germline predisposition for MN.

The diagnostic yield of the Nordic guidelines reached up to 35%, which was significantly higher than the expected 20% based on prior data from patients with solid tumors. This high detection rate could be partially attributed to the comprehensive somatic profiling that is performed in Sweden in all patients at the time of diagnosis of MDS/AML (23). For example, the CytoMol criterion was extremely effective in identifying patients with germline DDX41 variants, who often lacked FH for hematologic malignancies. Furthermore, all patients with germline RUNX1 variants included in the study, regardless of their inclusion criteria, would have been identified even if only the CytoMol criteria were used. Another explanation for the observed high diagnostic yield is our comprehensive approach, which combined WES with the analysis of CNV and targeted sequencing of known noncoding regions of interest as well as the high number of patients fulfilling more than one criterion (Fig. 2C). Finally, the high diagnostic yield in our study could indicate that the Nordic Guidelines are rather strict and could argue in favor of a less stringent approach similar to the European Leukemia Network recommendations (9). That being said, one may also claim that recommendations should have high diagnostic yield to spare on resources.

Applying arbitrary age cutoffs in germline investigation, such as our FH criterion, is associated with under-diagnosis in solid tumors (26). On the contrary, the high number of affected family members, independent of age, increases the likelihood of identifying a germline causative variant. Our study confirms the importance of shared phenotypes in the same family as indicated by the large difference in the diagnostic yield between the FH (6%) criterion and S-FH (44%) criterion, in which only the number of affected individuals is considered.

The spectrum of genes implicated in germline predisposition also varied depending on the applied criterion for germline testing. Overall, germline P/LP variants in the DDX41 gene were the most common finding, although predominantly in the CytoMol group, in line with previous studies (4, 27). As nationwide analysis of DDX41 in all patients undergoing a somatic panel for myeloid disorders was introduced only 6 months after study start, it is reasonable to argue that the frequency of germline DDX41 mutations may be higher. This probability is, however, rather tenuous. The great majority of the patients recruited within this period (n = 26) were stratified to MH/FH groups (20/26, 77%) and therefore analyzed with WES, with one of them being a carrier of a pathogenic DDX41 variant. Among the remaining six patients, four were analyzed with WES at a later time-point, with no DDX41 variants detected. Genetic heterogeneity for the underlying hereditary conditions was broader in the MH/FH group encompassing genes that are not commonly included in somatic panels (Fig. 3D), thus highlighting the importance of pursuing comprehensive germline testing in patients with normal results from somatic panels but with positive family history or clinical features suggestive of an underlying predisposing condition.

Despite the high diagnostic yield, our study failed to identify any germline variant in well-established genes for HHM in more than 60% of the patients included in the MH/FH group. In this group, no causative variants were identified among known HHM genes, whereas neither P/LP variants nor candidate VUS were identified in genes recently proposed as novel candidates for this patient group (3, 6, 9). However, a handful of patients were instead carriers of LP/P variants in genes that are mainly associated with higher risk for solid tumors than for MN (BRCA1, PALB2, and CHEK2). Recent reports support an association between mutations in these genes and the development of hematologic malignancies (9); however, further studies are needed. Considering that many of the patients with MN undergo allo-HSCT (42/60, 70% in our cohort), it is reasonable to argue that the knowledge of the mutational status for genes traditionally linked to solid tumor predisposition may be clinically valuable.

The diagnostic failure in patients with strong suspicion of an underlying hereditary condition that we and others (5, 6) experience suggests that not all genes predisposing to MN have yet been identified (5, 6). It also highlights the need for even more comprehensive diagnostic methodologies based on different types of omics. However, we cannot exclude that some familial cases may have a polygenic/multifactorial explanation. Prompted by these results, we provide at our institutions the opportunity of participation in research projects to the families fulfilling the Nordic criteria, for whom no P/LP variant can be identified with the standard diagnostic pipeline.

Survival analysis in patients diagnosed with AML revealed a slight advantage for those that were treated with allo-HSCT, with statistical significance being reached when limiting the analysis to patients with AML and confirmed germline predisposition. This, however, could be a general effect of allo-HSCT, not necessarily limited to patients with MN with germline predisposition, as suggested by landmark analysis. Although this finding should be evaluated with caution because of the low numbers, the limited follow-up time, as well as the heterogeneity of this group, one could argue that allo-HSCT could be beneficial for these patients. Our observation highlights the need for further studies in order to investigate whether allo-HSCT is associated with prolonged survival in patients with MN with germline predisposition.

Our findings further underline the importance of germline investigation in patients treated with allo-HSCT, as it can guide the treatment decisions related to the choice of donor or the type of conditioning regimen. Moreover, no increase in the incidence of GVHD could be observed in patients with pathogenic DDX41 variants who underwent allo-HSCT (25). This observation should also be considered with caution because of the low numbers, as well as the various clinicobiological parameters that may affect the development of GVHD within the allo-HSCT context.

By study design, genetic counseling by a trained geneticist was mandatory before and after germline testing. As a result, we observed an inclusion bias across Sweden, favoring hospitals with expertise in the field of HHM in the respective Department of Clinical Genetics. This bias could also indicate that, despite their inclusion in international guidelines, part of the hematologic community does not prioritize germline investigations. Another relevant question that arose is whether the role of hematologists and geneticists in germline diagnostics is well defined or whether there is uncertainty about their actual repsonsibilities (28).

In conclusion, we prospectively validated the Nordic guidelines about germline investigation in patients with suspected germline predisposition for MN and showed that their implementation in the clinical routine is feasible and results in a high diagnostic yield. The validation of the Nordic criteria-based guidelines therefore provides an important framework, enabling the establishment of solid clinical routines.

Supplementary Material

Supplementary Table 1

Supplementary Table 1. List of genes assessed by copy number analysis on WES data.

Supplementary Table 2

Supplementary Table 2. List of genes that have been proposed to be associated with HHM during the study time.

Supplementary Table 3

Supplementary Table 3. Overview of included patients and genetic findings.HGVS=Human Genome Variation Society.

Supplementary Table 4

Supplementary Table 4. Pairwise Comparisons of Survival Outcomes using cox proportional hazards regression. The figure in main manuscript or supplementary materials showing the Kaplan-Meier analysis for the same comparison is also indicated. * indicates when HR could not be calculated due to lack of events in one of the groups being compared.

Supplementary Figure 1

Supplementary Figure 1. Source of germline DNA used for WES and direct test of specific variants of interest, respectively. PB=peripheral blood.

Supplementary Figure 2

Supplementary Figure 2. Diagnostic yield criterion-based when all patients fulfilling CytoMol criterion (n=39) are assigned to CytoMol regardless of whether they were included in the study based on another criteria. MH: medical history; FH: family history; S-FH: strong family history; CytoMol: inclusion due to findings in the somatic profiling; P: Pathogenic; LP: likely pathogenic; VUS: variant of unknown significance.

Supplementary Figure 3

Supplementary Figure 3. Pedigrees and Relative telomere length (TL) in two patients with a missense VUS in TERT measured by qPCR in peripheral blood leukocytes as previously described.1 The reference percentiles were determined from telomere length analysis of blood leukocytes from 283 healthy subjects (0–78 years of age). The curves shown depict the first, 10th, 50th, 90th, and 99th normal percentiles at each age, where the 50th percentile represents the mean. A) Patient GH01_P11 was included due to cytopenias, premature graying of hair, infections, skin hypomelanosis. The variant segregated in the mother, who did not show clinical signs of telomere biology disorders but displayed shorter TL. B) Patient GH01_P33 was included due to MDS, lung fibrosis, premature graying of hair. The family history was positive for lung fibrosis on paternal side and for leukemia and premature graying of hair on the maternal side. Segregation analysis in parental samples was not possible. Reference: 1. Norberg A, Rosén A, Raaschou-Jensen K, et al: Novel variants in Nordic patients referred for genetic testing of telomere-related disorders. Eur J Hum Genet 26:858–867, 2018

Supplementary Figure 4

Supplementary Figure 4. Oncoplot showing the most frequently mutated genes among 62 patients with malignant disease included in the study where results from the diagnostic somatic panel performed during routine workup were available. Although the panel was performed as tumor-only in 53 patients, the data is presented as tumor-normal for all, since confirmed germline variants were not included in the analysis. Each column represents an individual sample, while each row corresponds to a specific gene. The presence of mutations is indicated by colored blocks, where light blue indicates single-hit, while red a multi-hit in the same gene. The barchart to the right of the oncoplot provides a summary count of mutations in specific genes. Patients are sorted according to the affected gene according to results from germline investigations.

Supplementary Figure 5

Supplementary Figure 5. Overall survival in patients with AML (A) or MDS (B) based on whether a P/LP germline variant in a well-established gene for hereditary blood disorders was found (“Yes”, blue line) or not (“No”, grey line). Comparison among subgroups was performed using the Log-rank test, with the resulting p-value presented within the plot.

Supplementary Figure 6

Supplementary Figure 6. Kaplan–Meier curve representing OS by germline and allo-HSCT status at landmark 5 months. Survival probability on the y-axis and time in months after landmark on the x-axis.

Authors’ Disclosures

B. Tesi reports grants from Region Stockholm (FoUI-985957), Swedish Society of Medicine (SLS-973171), and CIMED (FoUI-989107) during the conduct of the study. G. Barbany reports grants from Swedish Childhood Cancer Foundation (Barncancerfonden) during the conduct of the study. M. Jädersten reports other support from AbbVie, CanCell Therapeutics, AstraZeneca, and Pfizer outside the submitted work. No disclosures were reported by the other authors.

Authors’ Contributions

B. Tesi: Conceptualization, data curation, supervision, validation, methodology, visualization, project management, formal analysis, writing–original draft, writing–review and editing. A. Robelius: Conceptualization, data curation, formal analysis, investigation, writing–review and editing. B. Baskin: Data curation, formal analysis, investigation, methodology, writing–review and editing. V. Lazarevic: Investigation, methodology, writing–review and editing. S. Deneberg: Investigation, writing–review and editing. M. Höglund: Conceptualization, writing–review and editing. L. Fogelstrand: Investigation, writing–review and editing. J. Ungerstedt: Data curation writing–review and editing. T. Pandzic: Data curation, investigation, writing–review and editing. M. Tobiasson: Data curation, investigation, writing–review and editing. H.G. Garelius: Investigation, writing–review and editing. E. Kuchinskaya: Investigation, writing–review and editing. F. Persson: Investigation, writing–review and editing. H. Ågerstam: Investigation, writing–review and editing. H. Hallböök: Investigation, writing–review and editing. T. Fioretos: Supervision, investigation, methodology, writing–review and editing. J. Nordin: Data curation, formal analysis, writing–review and editing. A. Norberg: Investigation, writing–review and editing. A.-C. Thuresson: Investigation, writing–review and editing. S. Lehmann: Supervision, investigation, writing–review and editing. C. Ladenvall: Data curation, supervision, investigation, writing–review and editing. G. Barbany: Investigation, writing–review and editing. L. Vennström: Investigation, writing–review and editing. E. Ejerblad: Investigation, writing–review and editing. L. Cavelier: Formal analysis, investigation, writing–review and editing. J. Cammenga: Conceptualization, supervision, investigation, writing–review and editing. M. Jädersten: Investigation, writing–review and editing. E. Hellström-Lindberg: Supervision, validation, investigation, writing–review and editing. P. Baliakas: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.