Germline variants discovered in lymphoma patients undergoing tumor profiling: a case series

Abstract

Clinical tumor sequencing protocols often depend on obtaining germline DNA from patients to aid in the identification of de novo variants in the tumor, and therefore come with the possibility for the incidental discovery of germline variants. Ninety-one adult patients with lymphoma were consented and enrolled in MIONCOSEQ, an IRB-approved tumor profiling protocol that utilizes an exome sequencing platform. Charts were retrospectively reviewed for germline variants from sequencing results, personal and/or family history of cancer and genetic counseling referral. After review of the 91 lymphoma cases, seven (8%) cases revealed germline variants. Only one of these, CHEK2 p.I157T, has been previously recovered as a germline variant in lymphoma. Two of the seven patients received genetic counseling, two died before genetic counseling could be arranged and three did not follow-up with a genetics provider. None of the patients had a personal or family history that would have otherwise suggested an indication for cancer genetics referral, especially notable as lymphoma is not traditionally associated with inherited cancer syndromes. Importantly, as only two of seven patients had appropriate genetic counseling for their variant, timely genetic counseling should be a critical part of all tumor profiling platforms that use non-tumor DNA.

Introduction

In order to identify novel targeted therapies for advanced or otherwise untreatable cancers, tumor genetic sequencing is often used in a clinical context, especially for relapsed or refractory lymphoma [1]. Some of these sequencing protocols depend on obtaining germline DNA from patients to aid in the identification of de novo variants in the tumor. As such, the increasing use of tumor genetic profiling comes with the possibility for the incidental discovery of germline variants. Although studies have placed the frequency of these events at anywhere from 1 to 18% (depending on the sequencing modality [2], [3], [4],) these unanticipated findings have important implications for personal and family risk for cancers.

A few studies have demonstrated an increased risk for lymphoma with an affected first-degree relative, largely based on case-control populations [5], [6]. However, unlike for prototypical hereditary cancer disorders, genetic studies of lymphoma have not uncovered specific causative loci that reliably raise one’s risk for the disease to meet the threshold for genetic testing. Accordingly, lymphoma is not a common component of traditional inherited cancer syndromes.

Nevertheless, certain germline pathogenic variants have been linked with lymphoma development. Much of the work has centered around examining genes involved in the DNA damage response pathway [7]. Lymphocyte maturation involves mechanisms such as somatic hypermutation and recombination that, if dysregulated, can lead to malignant transformation, underscoring the role that DNA repair proteins can specifically play in lymphoma development [7], [8]. Specific genes that regulate DNA damage response include TP53 and CHEK2, which are well-associated with familial cancer risk diseases.

Methods

Ninety-one adult patients with lymphoma treated at University of Michigan Health Systems were consented and enrolled in MI-ONCOSEQ, an institutional board-approved tumor profiling protocol. Patients’ tumor and normal (either blood or saliva) samples under paired exome sequencing as well as tumor-only RNA-seq. Initially, this platform utilized whole exome sequencing, but was transitioned to a more targeted panel of 1700+ cancer-associated genes This protocol and list of sequenced genes are described in greater detail previously [9, 10].

Results

The demographics and lymphoma subtypes of the 91 cases are described in Table 1. After review of these lymphoma cases, seven (8%) cases revealed germline variants, summarized in Table 2. These variants are described here as it relates to their previous association with lymphoma and/or other cancers.

Table 1.

Demographic and disease characteristics of the 91 lymphoma cases sequenced as part of this study.

Characteristics	Value
Number of patients	91
Average age at diagnosis (range)	57.3 (25–88)
Sex
Male	57 (63%)
Female	34 (37%)
Characteristics at Diagnosis
Age >60	39 (43%)
Advanced Stage (III or IV)	62 (68%)
Malignancy types
Aggressive B-cell Lymphomas	39 (43%)
Diffuse Large B-Cell Lymphoma	21 (23%)
High Grade B-Cell Lymphoma	6 (7%)
Other*	12 (13%)
Indolent B-Cell Lymphomas	34 (37%)
Follicular Lymphoma	25 (27%)
Other**	9 (10%)
T-Cell Lymphomas	18 (20%)
Peripheral T-Cell Lymphoma-Not Otherwise Specified	4 (4%)
Cutaneous T-Cell Lymphoma	6 (7%)
Angioinununoblastic T-Cell Lymphoma	5 (6%)
Anaplastic Large Cell Lymphoma	3 (3%)

^*Mantle Cell Lymphoma (aggressive sub-type), Hodgkin Lymphoma, Transformed Follicular Lymphoma, Gray Zone Lymphoma, T-cell Prolymphocytic Letikemia

^**Marginal Zone Lymphoma, Mantle Cell (indolent sub-type)

Table 2.

Seven germline mutations reported as part of MI-ONCOSEQ sequencing, including tumor loss of heterozygosity (LOH) as well as non-tumor (‘normal”) and tumor variant allele fraction (VAF). Refseq accession number included.

Diagnosis Age/Gender	Type of Lymphoma	Mutation (Refseq)	Normal VAF/ Tumor VAF	Tumor LOH	Mutation clinical significance	Personal Hx Cancer	Family Hx Cancer^	Genetics follow-up?
80/M	DLBCL	TP53_p.R175H+ (NG_017013)	13%/3%	Yes	Pathogenic	Lung, BCC	CRC	Died
63/F	FL	TP53_p.R267Q (NG_017013)	50%/97%	Yes	VUS	Melanoma, BCC	Prostate	None
26/F	CTCL & CBCL	WTl_p.R413* (NG_009272)	46%/52%	No	Pathogenic	Wilms		Yes
33/M	HL	PALB2_p.L200fs (NG_007406)	50%/52%	No	Pathogenic			Yes
68/M	CTCL	CSF3R_p.M696T (NG_016270)	53%/77%	Yes	Unknown			None
81/M	DLBCL	ATR_p.I774fs (NG_008951)	12%/7%	No	VUS		Kidney, liver, CRC, breast	Died
62/M	FL	CHEK2_p.I157T (NG_008150)	47%/47%	No	Conflicting		CRC	None

-Clinical significance according to either Clinvar, COSMIC or genetic counseling assessment. (Lymphoma abbreviations- DLBCL: Diffuse Large B-Cell Lymphoma. FL: Follicular Lymphoma. CTCL: Cutaneous T-Cell Lymphoma. CBCL: Cutaneous B-Cell Lymphoma. HL: Hodgkin Lymphoma)

⁺-mosaic vs. early somatic mutation

^{^}-cancer in first degree relative

TP53

TP53 is well-known as a chief member of the DNA damage response pathway, influencing DNA repair mechanisms and then potentiating cell death if the damage becomes overwhelming. Pathogenic germline variants in this gene lead to Li-Fraumeni syndrome, which results in a variety of solid and hematological tumors. TP53 has also been previously identified as having both germline and somatic variants in lymphoma, resulting in a high tumor mutation burden due to impairment of the DNA damage response [11], [12], [13]. Notably, neither of the two variants – R175H and R267Q – identified in our patients have been explicitly associated with lymphoma. The germline R175H variant has been designated as pathogenic and previously associated with Li-Fraumeni syndrome. Genetic analysis of TP53 is complicated by a phenomenon known as aberrant clonal expansion (ACE), whereby the variants recovered from blood and/or saliva do not represent true germline sequence but instead are derived from somatic variation [14]. Since this variant was found in only 13% of this patient’s germline reads, this result likely represents a hematologic-specific version of ACE, clonal hematopoiesis of indeterminate potential; this process can provide a small but measurable risk of hematologic malignancy [14]. The individual carrying this variant had a history of late-onset lung cancer and basal cell carcinoma, as well as one first-degree relative with colorectal cancer, but the ages of onset and low allele frequency of the pathogenic variant make Li-Fraumeni less likely.

The second TP53 variant in this cohort, R267Q, has been previously identified and is listed in ClinVar, however has been classified as a variant of uncertain significance. This patient too had a personal history of cancer (late-onset melanoma and basal cell carcinoma) and a family member with prostate cancer, but not at ages suggestive of a hereditary cancer syndrome. Of note, the patient had TP53 loss of heterozygosity in the tumor.

PALB2

PALB2 helps to target the homologous recombination repair machinery to DNA damage sites and is one of the causative genes in the development of Fanconi Anemia. One small study uncovered bi-allelic inherited variants in two cases of related individuals with early-onset lymphoma [15]. They did not exhibit many of the traditional developmental phenotypes of Fanconi Anemia, potentially because one of the alleles proved hypomorphic based on in vitro testing, making it difficult to correlate PALB2 loss with lymphoma risk [15]. Conversely, germline pathogenic variants are best described in the causation of both breast and pancreatic cancers [16], including the variant described here, L200fs [17]. Notably, the patient with this variant had lymphoma at a relatively young age; family history notable for a maternal grandmother with breast cancer diagnosed in her late 50’s. He ultimately followed up with genetic counseling, who recommended testing family members for this finding. No recommendation for increased cancer screening was made for the patient given his active cancer diagnosis.

ATR

ATR is involved in the cell cycle checkpoint process in response to replication stress, impeding cellular division in the event of DNA damage. From a somatic perspective, a series of three alterations in ATR were recovered in Hodgkin Lymphoma [18]; the truncation described here, I774fs, was previously identified as a somatic variant in endometrial carcinoma [19], but has not been reported in ClinVar as a germline variant. Most germline variants of ATR on ClinVar are likely benign or are of uncertain significance, however certain inherited variants result in a rare autosomal recessive developmental disorder known as Seckel Syndrome or an autosomal dominant telangiectasia and oropharyngeal disorder [20].

CHEK2

CHEK2 acts upstream of TP53 in the DNA damage response pathway, phosphorylating key parts of the cell cycle and cell death machinery in response to DNA damage. Accordingly, germline variants in this gene are mainly associated with breast, colorectal and prostate cancer risk. There are mechanistic data that belie a possible role in tumorigenesis for this patient’s specific variant, I157T [21]. Additionally, a study of 340 Non-Hodgkin lymphoma patients from the Czech Republic recovered multiple germline variants in CHEK2, including the variant described here [7], [22]. However, conflicting data exist as to whether this variant is pathogenic or a variant of uncertain significance, as it may be of low penetrance given its inclusion in normal population databases [23] and presence in a control population [22].

CSF3R

A receptor for granulocyte colony stimulating factor, CSF3R variants are often associated with neutropenia if loss of function or leukemias in instances of gain of function. While the variant described here, M696T, was recovered as a somatic variant in two putative cases of Chronic Neutrophilic Leukemia [24] and in three patients with Chronic Myelomonocytic Leukemia [25]. Otherwise, it is not well-described, not a known germline pathogenic variant and present in normal population databases [23]. The patient had loss of heterozygosity in the tumor.

WT1

WT1 recruits TET2 to specific genomic loci in order to reverse epigenetic inactivation by DNA methylation; inactivating variants in WT1 will therefore negatively impact TET2 activity. Of note, TET2 germline variants have been previously identified in rare cases of lymphoma, but generally TET2 loss is a somatic process in hematologic malignancies [26]. Although WT1 was initially discovered as the gene product involved in Wilms’ Tumor, a pediatric kidney tumor, it is commonly mutated in other neoplasms, namely AML [27]; the patient’s specific truncation has been described previously in Wilms’ Tumor, R413* [28]. Accordingly, the patient described here was diagnosed with Wilms’ Tumor as a child, before being diagnosed with lymphoma as an adult. There was no loss of heterozygosity noted in this case. Almost all cases of hereditary Wilms’ tumor are de novo, and the patient did not report a family history of any other individuals with Wilms’ tumor.

Discussion

Although intended for treatment decision making based on somatic tumor alterations, many tumor profiling protocols will capture germline variants, evidenced by the seven cases presented here.

To review, this cohort includes only one variant that had been previously described as a germline variant in lymphoma (CHEK2). Taking into account the tumor sequencing results, two variants of uncertain significance from this report – CSF3R p.M696T and TP53 p.R267Q – had loss of heterozygosity in their tumors, suggesting that they may ultimately be pathogenic in nature.

Notably, five of the seven patients had germline variants in the DNA damage response pathway, reinforcing the importance of this mechanism in lymphoma development. Variants in these genes, especially TP53, PALB2 and CHEK2, have special clinical significance as they lead to well-described hereditary cancer syndromes. Additionally, identification of TP53 in non-tumor patient samples should prompt discussion of a skin biopsy to truly diagnose these individuals with Li-Fraumeni syndrome, especially in the setting of low “germline” variant allele fraction or equivocal family history.

Moreover, this cohort demonstrates the additional implications of tumor sequencing and the need for accessibility to genetic counseling at the centers conducting these tests. Even in a disease not traditionally associated with known cancer syndromes, pathogenic germline variants were recovered in genes that significantly affect the risk of non-lymphoid cancers. None of the seven patients had an obvious personal and/or family history of cancer that would have prompted a referral to genetic counseling. Two patients ultimately had genetic counseling after being informed of the results of their germline variants. Conversely, two died prior to obtaining a counseling referral. The long term significance of this remains unclear, but demonstrates the importance and providers’ responsibility for prompt genetic counseling referral on all germline variants that result from tumor sequencing protocols.