Implementation of and systems-level barriers to guideline-driven germline genetic evaluation in the care of patients with myelodysplastic syndrome and acute myeloid leukemia

Abstract

Purpose:

Knowledge of an inherited predisposition to myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) has important clinical implications for treatment decisions, surveillance, and care of at-risk relatives. National Comprehensive Cancer Network (NCCN) guidelines recently incorporated recommendations for germline genetic evaluation of patients with MDS/AML based on personal and family history features, but the practicality of implementing these recommendations has not been studied.

Methods:

A hereditary hematology quality improvement (QI) committee was formed to implement these guidelines in a prospective cohort of patients diagnosed with MDS/AML. Referral for germline genetic testing was recommended for patients meeting NCCN guideline criteria. Referral patterns and genetic evaluation outcomes were compared to a historical cohort of MDS/AML patients. Barriers to evaluation were identified.

Results:

Of 90 patients with MDS/AML evaluated by the QI committee, 59 (66%) met criteria for germline evaluation. Implementation of the QI committee led to more referrals for germline evaluation in accordance with NCCN guidelines (31% versus 14%, p = 0.03). However, the majority of those meeting criteria were never referred due to high medical acuity or being deceased or in hospice at the time of QI committee recommendations. Despite this, two of 12 (17%) undergoing genetic testing were diagnosed with a hereditary myeloid malignancy syndrome.

Conclusion:

Current NCCN guidelines resulted in two thirds of patients with MDS/AML meeting criteria for germline evaluation. A hereditary hematology-focused QI committee aided initial implementation and modestly improved NCCN guideline adherence. However, the high morbidity and mortality and prolonged inpatient stays associated with MDS/AML challenged traditional outpatient genetic counseling models. Further improvements in guideline adherence require innovating new models of genetic counseling and testing for this patient population.

Introduction

Myeloid malignancies such as acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS) are traditionally thought of as sporadic disorders. However, genetic predispositions to MDS/AML, also known as hereditary myeloid malignancy syndromes (HMMS), are becoming increasingly recognized. Several constitutional syndromes have long been known to predispose to myeloid malignancy, including Fanconi anemia, telomere biology disorders, Shwachman-Diamond syndrome, and Diamond-Blackfan anemia^1–4. More recently, a growing number of HMMS without syndromic features caused by pathogenic or likely pathogenic (P/LP) variants in genes including RUNX1, DDX41, CEBPA, ETV6, and ANKRD26 are relevant to MDS/AML care^5–9. Despite these advancements, only 30% of families clustering MDS/AML have a P/LP variant identified, suggesting that additional causative genes remain to be discovered¹⁰.

HMMS are underrecognized due to limited provider awareness, insufficient recording of family history,¹¹ and lack of HMMS-trained genetics providers in many areas. In addition, subtle clinical presentations, anticipation, variable expressivity, and phenotypic heterogeneity make establishing the diagnosis of HMMS challenging. Though the exact prevalence is unknown, HMMS are likely more common than previously recognized. Approximately 15% of patients with AML have a family history of hematologic malignancy¹². Of those evaluated in a hereditary clinic, 12–18% are ultimately diagnosed with a known HMMS^13,14. Furthermore, in unselected MDS/AML populations, clinically actionable P/LP germline cancer susceptibility gene variants are identified in at least 7% of cases regardless of age^15,16.

Identification of HMMS is critical for optimizing care for patients and their families. Knowledge of HMMS may influence clinical management of the patient’s underlying MDS/AML, such as the decision to pursue hematopoietic stem cell transplant (HSCT), choice of conditioning regimen, and donor selection,^17–21 or have implications for other organ and cancer surveillance^22,23. Despite the importance of HMMS detection among patients with MDS/AML, there is no formal consensus as to which patients should undergo germline genetic evaluation, and expert-opinion recommendations for whom to test have evolved over time, making it difficult for providers to know who to refer. In 2021, the National Comprehensive Cancer Network (NCCN) added recommendations for germline genetic evaluation for patients meeting specific clinical and/or molecular criteria to the MDS and AML management guidelines^24,25. However, data are lacking on the proportions of patients that will then need germline assessment as well as the practicality of implementing these guidelines in the care of patients with MDS/AML.

Here, we report our single institution experience aiming to improve detection of HMMS by implementing NCCN MDS/AML germline genetic testing guidelines through a quality improvement (QI) committee. At the onset of our intervention, our institution had no standardized workflows for identifying those needing HMMS evaluation, for accessing HMMS evaluation at the appropriate level of urgency, or for reporting and reviewing somatic sequencing data for potential germline variants in patients with MDS/AML. Thus, we created a multidisciplinary QI committee to implement the systems-level changes required to facilitate germline genetic testing for patients with MDS/AML meeting guideline criteria. We report frequencies of patients meeting NCCN criteria, describe the proportions actually evaluated and tested and overall yield, and identify areas needing optimization to maximize detection of HMMS while minimizing burden on patient care and the healthcare system.

Methods

Patient Cohorts and Data Collection

The post-intervention cohort consisted of patients with either newly diagnosed MDS/AML (including chronic myelomonocytic leukemia (CMML) and MDS/myeloproliferative neoplasm (MPN) overlap) or those with new molecular data (e.g., relapsed AML with no prior multigene panel) between 9/1/2021 and 8/31/2022 who were prospectively reviewed by a hereditary hematology QI committee as described below. In order to assess the effectiveness of implementing this QI committee, data from patients newly diagnosed with MDS/AML between 1/1/2019 and 9/1/2021 with next-generation sequencing (NGS) data available were retrospectively reviewed (hereafter referred to as the pre-intervention cohort). For both cohorts, clinical, molecular, and family history data and germline genetic testing results (if applicable) were abstracted from the electronic medical record (EMR). This study was approved by the University of Wisconsin-Madison Institutional Review Board.

Hereditary Hematology QI Committee Intervention

On 7/1/2021, a hereditary hematology QI committee was formed to prospectively evaluate patients diagnosed with MDS/AML at our center for possible HMMS (Supplemental Appendix 1). The committee first identified steps needed to implement NCCN guidelines within our institution including: 1) creating concise consensus criteria for HMMS evaluation encompassing both NCCN AML (version 1.2022) and MDS (3.2022) guidelines^24,25; 2) educating providers regarding key HMMS personal and family history details; 3) creating a system to capture all somatic NGS test results for centralized review; 4) adding a HMMS referral order to the EMR; 5) creating a concise referral recommendation template note for the EMR to be used to document reason for referral recommendation, degree of urgency, and how to refer the patient; and 6) developing a process for addressing urgent referrals.

After finding solutions to the above steps, beginning on 9/1/2021, the committee met monthly to review each patient’s history and molecular data to determine if HMMS evaluation criteria were met (Figure 1). These criteria consisted of clinical and molecular components. Clinical criteria were defined as 1) a personal or family history suggestive of an HMMS or other cancer predisposition syndrome (Supplemental Appendix 2); 2) a family history of a hematologic cancer in a first-, second-, or third-degree relative; OR 3) MDS/AML diagnosed age ≤50 years. Molecular criteria were defined as the presence of a P/LP variant in ANKRD26, CEBPA, DDX41, ETV6, GATA2, RUNX1, TERT, and/or TP53 at a variant allele frequency (VAF) of ≥30% detected on molecular genetic testing of MDS/AML cells using either a 57-gene (Myeloid Malignancies Mutation Panel; ARUP Laboratories, Salt Lake City, UT) or 42-gene (OncoHeme; Mayo Clinic Laboratories, Rochester, MN) panel. Variant interpretations were reassessed per American College of Medical Genetics and ClinGen Committee standards to identify those meeting P/LP status if confirmed in the germline^26,27. For patients meeting criteria, a referral recommendation, documenting which criteria were met as well as the urgency of referral, was entered in the EMR (Supplemental Appendix 3) and routed to the patient’s primary hematologist for review. If the patient was deceased or unable to pursue germline genetic evaluation for any reason at the time of QI committee review and suspicion for HMMS was strong (e.g. a known pathogenic recurrent germline DDX41 variant present on the patient’s somatic panel at a heterozygous VAF), our concerns, including risks, implications, and recommendations for family members, were communicated to the treating hematologist, who could then facilitate Genetics referral for at-risk relatives at his or her discretion. Urgent cases requiring expedited review were triaged within 24–48 hours by one or more committee members outside of regularly scheduled meetings when brought to the committee’s attention by the treating hematologist based on family history or by the molecular pathologist interpreting somatic sequencing results. If urgency was confirmed, a germline genetic evaluation was performed within one to two weeks. In all other cases, timing was determined on a case-by-case basis, occurring on average two to six weeks from initial referral.

Figure 1.

Workflow of the hereditary hematology quality improvement committee to facilitate evaluation for and diagnosis of HMMS. Referral to Genetics and germline genetic testing was recommended if the patient met criteria developed by the committee based on NCCN AML (version 1.2022) and MDS (version 3.2022) guidelines. AML, acute myeloid leukemia; HMMS, hereditary myeloid malignancy syndrome; MDS, myelodysplastic syndrome; NGS, next generation sequencing.

Genetic Evaluation

Patients referred to Genetics underwent a comprehensive history and physical examination and family pedigree analysis by a certified genetic counselor (KSS) and cancer geneticist/hematologist/oncologist (JEC). All consenting patients with a clinical history or molecular data confirmed to meet the above criteria were offered germline genetic testing on cultured skin fibroblasts via a single or multigene panel personalized according to the individual patient scenario.

Statistical Analysis

Descriptive statistics including two sample tests of proportions, Fisher exact or Chi squared tests for comparing categorical variables, and Wilcoxon rank sum or Student t tests for comparing continuous variables were generated using Stata v17 (StataCorp, USA).

Results

Post-Intervention Cohort QI Committee Review

Between 9/1/2021 and 8/31/2022, the QI committee reviewed 90 patients (Table 1, Supplemental Table 1). Forty one (46%) were diagnosed with AML, 32 (36%) with MDS, 10 (11%) with CMML, six (7%) with MDS/MPN overlap, and one (1%) with isolated myeloid sarcoma. Fifteen (17%) patients had a family history of hematologic malignancy.

Table 1.

Pre- and post-hereditary hematology quality improvement committee intervention patient cohort characteristics

Characteristic	Pre-Intervention Cohort (n, %)	Post-Intervention Cohort (n, %)	p-value
Total	103 (100)	90 (100)
Diagnosis
AML	50 (49)	41 (46)	0.96
MDS	35 (34)	32 (36)
CMML	11 (11)	10 (11)
MDS/MPN	7 (7)	6 (7)
Myeloid sarcoma	0 (0)	1 (1)
Age at diagnosis in years (median, range)	66 (1–90)	71 (5–89)	0.06
Male sex assigned at birth	62 (60)	49 (54)	0.47
Race/ethnicity
White	95 (92)	85 (94)	0.93
Black or African American	2 (2)	1 (1)
Asian	1 (1)	1 (1)
Hispanic/Latino	3 (3)	1 (1)
American Indian or Alaska Native	1 (1)	1 (1)
More than one race	0 (0)	1 (1)
Declined to answer	1 (1)	0 (0)
Personal history of another cancer	23 (22)	24 (27)	0.51
Family history of hematologic malignancy	26 (25)	15 (17)	0.16
HMMS-associated variants *
ANKRD26	1 (1)	0 (0)	0.05
CEBPA	2 (2)	7 (8)
DDX41	0 (0)	4 (4)
ETV6	1 (1)	0 (0)
GATA2	0 (0)	3 (3)
RUNX1	9 (9)	6 (7)
TERT	0 (0)	0 (0)
TP53	8 (8)	10 (11)
Total	21 in 19 patients (18%)#	30 in 26 patients (29%)

^*Variant allele frequency ≥30% and determined to be pathogenic/likely pathogenic if confirmed in germline by QI Committee review. Six individuals had more than one variant.

^#Gene panel type was at the discretion of the treating physician. All gene panels obtained during the post-intervention time period included the eight HMMS-associated genes. However, not all gene panels clinically available during the pre-intervention time period were comprehensive, and 22 (21%) patients in the pre-intervention cohort had incomplete sequencing of the eight HMMS-associated genes.

See Supplemental Table 2 for more details.

AML, acute myeloid leukemia; CMML, chronic myelomonocytic leukemia; HMMS, hereditary myeloid malignancy syndrome; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm.

Overall, 59 (66%) patients met one or more criterion, resulting in a Genetics referral recommendation (Figure 2). Of these, 33 (37%) met clinical only, 9 (10%) met molecular only, and 17 (19%) patients met both criteria. Among the 50 patients meeting clinical criteria, 24 (48%) had a personal history of another cancer and/or presumed therapy-related disease, 15 (30%) had a family history of hematologic malignancy or pattern suspicious for a cancer predisposition syndrome, 12 (24%) were diagnosed at age 50 or younger, and two (4%) had clinical features suggestive of a specific HMMS. Eight patients (16%) met more than one clinical criterion. Among the 26 (29%) patients meeting molecular criteria, a total of 30 P/LP HMMS-associated variants with a VAF ≥30% were detected on somatic sequencing, distributed among TP53 (10 variants), CEBPA (7 variants), RUNX1 (6 variants), DDX41 (4 variants), and GATA2 (3 variants). Overall, 26 patients (29% of entire cohort, 44% of those meeting referral criteria) were recommended to undergo Genetics evaluation urgently, predominantly for transplant planning purposes. Most patients referred urgently were seen in clinic within one to two weeks.

Figure 2.

Referral patterns, uptake of genetic counseling, and germline genetic sequencing outcomes of 90 patients reviewed by the hereditary hematology quality improvement committee. Green symbols refer to detection of a pathogenic or likely pathogenic variant in an HMMS-associated gene and clinical diagnosis of HMMS. Red symbols refer to negative genetic sequencing or detection of a variant of uncertain significance in an HMMS-associated gene. AML, acute myeloid leukemia; HMMS, hereditary myeloid malignancy syndrome; MDS, myelodysplastic syndrome; NCCN, National Comprehensive Cancer Network.

Referral and Germline Genetic Testing Outcomes

In total, 18 patients meeting NCCN criteria were ultimately referred to Genetics, corresponding to 20% of the entire cohort and 31% of those meeting referral criteria. Of the 41 not referred despite committee recommendations, reasons were available for 23 (56%) patients and included 13 (32%) being deceased or enrolled in hospice, 5 (12%) unable to attend an outpatient Genetics appointment due to medical acuity requiring frequent or prolonged hospitalizations, and 5 (12%) declining referral.

Once referred, the majority (n = 14 of 18; 78%) completed evaluation in the Genetics Clinic and all but two (n = 12 of 14; 86%) consented to proceed with germline genetic testing (Table 2). Among these, 17% (2/12) had a positive result, 67% (8/12) had a negative result, and 17% (2/12) had a variant of uncertain significance (VUS) in one or more HMMS-associated genes (Figure 2; Table 2).

Table 2.

Clinical data and germline genetic evaluation results among patients in the post-intervention cohort who underwent genetic counseling and were offered germline genetic sequencing (n = 14 of 90 (16%))

Patient ID	Age and Sex	Diagnosis	Personal History	Family History	Cytogenetics	NGS Panel Results	Criteria for Referral	Germline Sequencing Results
3	5M	AML	Hepatoblastoma treated with chemotherapy andliver transplant	Prostate cancer (paternal grandfather)	46,XY,t(10;16)(q22;p13.3)[19];46,XY[1]	BCOR, p.Ser1642* (63%); CEBPA, p.Arg333Pro (30%); PHF6, p.Cys326Phe (10%); WT1, p.Val380Leu (5%)	Clinical	CEBPA wild type; 139 other genes negative
17	77M	MDS	Inflammatory bowel disease	No significant family history	47,XY,+8[1]; 46,XY[19]	DDX41, p.Gln41* (49%)	Molecular	Pathogenic variant in DDX41 (c.121C>T, p.Gln41*, heterozygous); 47 other genes negative
18	67M	MDS	Pulmonary fibrosis, cryptogenic cirrhosis, polyneuropathy	Unspecified pulmonary disease (two maternal uncles)	46,XY,del(20)(q11.2q13.1)[14]; 46,XY[1]	U2AF1, p.Ser34Phe (34%)	Clinical	VUS in ANKRD26 (c.1628A>T, p.Gln543Leu, heterozygous); VUS in TP53 (c.642T>G, p.His214Gln, heterozygous); 23 other genes negative
33	57M	MDS	Stroke	Pancreatic cancer (sister), breast cancer (maternal aunt)	46~48,XXY,add(1)(q32),−5,+6,add(6)(q25),add(7)(q11.2),add(7)(q32),der(7),add(7)(p22)add(7)(q11.2),+8,add(9)(p22),add(9)(p12),der( 10)t(10;13)(p15;q14),−13,add(13)(q34),−16,add(17)(p11.2),add(20)(q11.2),+1~2,+mar[cp17]; 47,XXY (constitutional)	TP53, p.Met160Trpfs*10 (44%)	Both	TP53 wild type; 267 other genes negative
43	62F	AML	Sarcoidosis	Lymphoma (father), acute leukemia (maternal uncle), liposarcoma (maternal uncle), melanoma (brother)	46,XX[20]	IDH2, p.Arg140Gln (45%); NPM1,p.Trp288Cysfs*12 (45%)	Clinical	233 genes negative
54	76F	CMML	Chronic macrocytosis and thrombocytopenia	Autoimmune hemolytic anemia (son), chronic macrocytosis(daughter)	47,XX,+1,der(1;18)(p10;q10),+8,del(20)(q11.2q13.1) [20]	RUNX1, p.Leu94Glyfs*18 (25%)	Clinical	RUNX1 wild type; 232 other genes negative
56	75F	MDS	Atrial fibrillation	Aplastic anemia (mother)	46,XX[20]	ASXL1, p.Glu635Argfs*15 (52%); DDX41, c.3G>A, p.M1? (50%); ETNK1, p.Asn244Ser (47%); EZH2, p.Trp15* (45%); EZH2, p.Leu620* (41%); FLT3,p.Asp835Glu (27%); RUNX1,p.Pro330Argfs*327 (35%)	Both	Pathogenic variant in DDX41 (c.3G>A, p.Met1?, heterozygous); RUNX1 wild type; 231 other genes negative
61	18F	AML	None	Breast cancer (maternal aunt), prostate cancer (maternal grandfather), pancreatic cancer (paternal grandfather)	46,XX,inv(3)(q21q26.2)[20]	NRAS, p.Gly12Val (50%); WT1, p.Arg302Leufs*6 (28%); WT1, p.Val303Cysfs*14 (37%)	Clinical	140 genes negative
63	23F	Relapsed AML	AML diagnosed at age 19 treated with daunorubicin, cytarabine, etoposide	No significant family history	46,XX,inv(16)(p13.1q22)[11];47,sl,+8[3]; 46,XX[6]	NRAS, p.Gly13Asp (33%)	Clinical	Declined genetic testing
64	78M	AML	None	Low blood counts (sister), colon cancer (father)	91,XXYY,−16[3]; 46,XY[6]	ASXL1, p.Gly646Trpfs*12 (34%); EZH2,p.Gln61Glu (9%); RUNX1, p.Arg139Gln (61%); ZRSR2, p.Arg295* (75%)	Both	RUNX1 wild type; 82 other genes negative
70	69F	MDS/MPN	Mantle cell lymphoma treated with HyperCVAD and autologous PBSCT	Brain tumor (father), CLL (father and brother)	Not obtainable	ASXL1, p.Glu728* (35%); EZH2, p.Glu346Aspfs*3 (33%); EZH2, c.1240+1G>A (splice site) (35%); GATA2, p.Arg396Trp (11%)	Clinical	Whole-exome sequencing negative
78	42M	AML	None	No significant family history	46,XY,del(9)(q13q32)[20]	CEBPA, p.Pro45Hisfs*115 (46%); CEBPA, pArg306_Gln311dup (39%); NRAS, p.Gly13Arg (41%); WT1, p.Ala47Valfs*81 (38%)	Both	CEBPA wild type
80	39F	MDS	Pancytopenia for >10 years	No significant family history	46,XX[20]	None	Clinical	VUS in FANCD2 (gain of exons 33–43, copy number 3); VUS in PTPN11 (c.1124A>G, p.Tyr375Cys, heterozygous); 157 other genes negative
88	11M	Relapsed AML	AML treated with allogeneic matched sibling PBSCT at age 9	Ovarian cancer (maternal grandmother), testicular cancer (maternal grandfather)	46,XY,t(6;11)(q27;q23)[20]	CEBPA, p.49fs (44%); NRAS, p.Gly12Val (10%); NRAS, p.Gly12Ala (25%); WT1, p.Ala282fs (42%)	Both	Declined genetic testing

AML, acute myeloid leukemia; CMML, chronic myelomonocytic leukemia; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm; PBSCT, peripheral blood stem cell transplant; VUS, variant of uncertain significance.

Both patients with positive genetic testing were found to have a heterozygous pathogenic variant in DDX41 on a somatic NGS panel which was confirmed to be in the germline, consistent with the previously reported high germline confirmation rates of DDX41 variants in patients with MDS/AML²⁸. Patient 17 (DDX41, c.121C>T, p.Gln41*) had nine siblings yet no family history of hematologic malignancy; he was diagnosed with MDS at age 74. Patient 56 (DDX41, c.3G>A, p.Met1?) had a family history of aplastic anemia in her mother and was diagnosed with MDS at age 75.

Two patients were found to have a VUS in one or more HMMS-associated genes. Patient 18 was referred for evaluation of potential telomere biology disorder (TBD) due to personal history of MDS, pulmonary fibrosis, and cryptogenic cirrhosis. Germline genetic sequencing of 15 TBD-associated genes was negative, and telomere length was >10^th percentile for age in all six lymphocyte subsets by flow fluorescent in situ hybridization (FlowFISH). He underwent additional genetic testing which revealed a heterozygous VUS in ANKRD26 (c.1628A>T, p.Gln543Leu) and a heterozygous VUS in TP53 (c.642T>G, p.His214Gln). Patient 80 was referred due to MDS diagnosed at a young age. Germline genetic sequencing revealed a heterozygous VUS in FANCD2 (gain of exons 33–43, copy number 3) and a heterozygous VUS in PTPN11 (c.1124A>G, p.Tyr375Cys). Eight patients referred had negative germline genetic testing.

Comparison of Guideline Adherence Pre- and Post-Implementation of the QI Committee

In order to assess the effectiveness of the QI committee in increasing adherence to NCCN HMMS evaluation guidelines, Genetics referral and germline genetic testing outcomes were compared between patients with MDS/AML prospectively discussed by the QI committee (post-intervention cohort) and retrospective review of equivalent patients diagnosed prior to initiation of the QI committee (pre-intervention cohort) (Table 1; Supplemental Figure 1).

The pre-intervention cohort consisted of 103 patients with MDS/AML (Table 1). Compared to the post-intervention cohort, there were no significant differences in diagnoses, age at onset, sex, personal history of another cancer, or family history of hematologic malignancy. Although not significantly different, a lower total number of P/LP HMMS-associated genetic variants were identified in the pre-intervention cohort. This was likely due to the later addition of ANKRD26, DDX41, GATA2, and TERT to some somatic NGS panels utilized during the pre-intervention study period, accounting for the lower than expected number of variants in these genes. In total, 22 of the 103 (21%) patients in the pre-intervention cohort had incomplete sequencing of the eight HMMS-associated genes (Supplemental Table 2).

Overall, a similar proportion of pre-intervention patients met one or more criterion for Genetics referral (n = 58, 56% versus n = 59, 66%; p = 0.25) (Supplemental Figure 1). The distribution of specific criteria met were also similar (p = 0.35), including 39 patients (38%) meeting clinical criteria only, 7 (7%) molecular criteria only, and 12 (12%) both clinical and molecular criteria.

A significantly smaller proportion of patients in the pre-intervention versus the post-intervention cohort overall and of those meeting referral criteria were referred for genetic evaluation (n = 8, 8% versus n = 8, 20%, p = 0.02; and n = 8, 14% versus n = 18, 31%, p = 0.03, respectively). Despite this, among those who had germline genetic testing, the proportion with a positive genetic test was similar (n = 1 of 7, 14% versus n = 2 of 12, 17%; p = 0.86). The single pre-intervention patient with a positive test was found to have a pathogenic variant in ANKRD26 in the context of a personal and family history of thrombocytopenia and leukemia.

Discussion

In this study, we report our single institution experience using a hereditary hematology QI committee to implement current NCCN germline genetic testing criteria for patients with MDS/AML. In a cohort of 90 patients over a 12-month period, a surprisingly high proportion, 66%, met referral criteria for germline genetic evaluation. In addition, nearly half of referrals were considered urgent based on implications for treatment or transplant decision-making. Monthly review of patient history and molecular data as well as a formalized referral recommendation note in the EMR significantly increased guideline-concordant genetic referrals (14% versus 31%, p = 0.03). However, the majority of eligible patients were not referred at all, often due to critical illness, prolonged hospital stays, and early deaths. To our knowledge, this is the first study to prospectively test the clinical impact of implementing NCCN-based germline genetic testing referral criteria in an unselected MDS/AML patient population. This experience demonstrates that implementing current NCCN guidelines will significantly increase the proportion of patients referred and potentially diagnosed with HMMS. However, it also demonstrates the need for MDS/AML-specific models of genetic counseling and germline genetic testing, likely in the inpatient setting, to maximize benefit for patients and their families.

With the yield of germline genetic testing for known HMMS in patients with MDS/AML being similar to germline genetic testing yields in other cancers for whom testing is recommended for all,^29–31 a universal screening strategy has been proposed, in which all patients diagnosed with MDS/AML undergo germline evaluation regardless of personal or family history^15,16,32. Our data, showing that the majority of patients with newly diagnosed MDS/AML will meet NCCN guidelines for germline genetic testing, and that nearly all who are evaluated and counseled will elect to proceed, support such a universal genetic testing approach. However, despite our QI committee increasing awareness of HMMS, providing formalized recommendations for Genetics referral, and having accessible, trained HMMS-focused genetic counselors and cancer geneticists, only ~30% of patients meeting criteria were ultimately referred. This was most often due to factors related to the high morbidity and mortality associated with an MDS/AML diagnosis. Specifically, nearly a third of eligible patients were deceased or enrolled in hospice at the time of our referral recommendation. Others were critically ill or were hospitalized for prolonged periods, precluding the typical outpatient genetic counseling and germline genetic evaluation workflow. Thus, our data highlight a critical need for urgent inpatient models of germline genetic evaluation for patients with MDS/AML, similar to those utilized for neonatal rapid whole exome sequencing,³³ that will allow access close to the time of diagnosis when disease- and treatment-related morbidity is high, but which arguably is the optimal time for genetic evaluation to occur to assist in treatment planning.

While implementation of a hereditary hematology QI committee was a significant time commitment for the members involved, we believe such a committee was essential for establishing a workflow for germline genetic evaluation for patients with MDS/AML. Having multiple myeloid malignancy providers as well as hematopathologists, molecular pathologists, cancer geneticists, and genetic counselors working together to ensure feasibility, clinical relevance, appropriate level of urgency, and buy-in from the broader hematology faculty at our institution was critical. As our data show, we identified many systems-level barriers to guideline-concordant genetic evaluation in this patient population, which our QI committee is now addressing. For example, we are training our hematology providers to ask HMMS questions during the initial history gathered for newly diagnosed patients, to recognize when patients need urgent referrals, and to perform skin biopsies for germline genetic testing purposes. When skin is collected at the time of diagnostic bone marrow biopsy, fibroblast culture and germline genetic testing results are expedited, which is especially important when outpatient Genetics evaluation will be delayed. Our genetic counseling team is also addressing barriers to inpatient genetic counseling services by enacting telephone/video visits for inpatients well enough to have these conversations. Thus, although a hereditary hematology QI committee is resource-intensive, it aided initial implementation by developing a consensus workflow for HMMS evaluation and is now able to address the systems-level barriers to germline genetic evaluation for this patient population. Future studies will examine the effectiveness of our systems-level interventions at increasing guideline concordance.

In summary, despite HMMS evaluation being indicated for the majority of patients with MDS/AML, uptake of Genetics services was low, often due to the high clinical acuity of this patient population and lack of inpatient genetic counseling services, both limiting the accessibility of germline genetic evaluation resources. Implementation of a hereditary hematology QI committee improved adherence to NCCN germline genetic testing guidelines in patients with MDS/AML and is continuing to help resolve systems-level barriers to germline genetic testing for this unique patient population.

Data Sharing Statement

All individual genetic variant data are included in Table 1 and Supplemental Table 1. Additional details are available upon request to the corresponding authors.