POT1 mutation spectrum in tumor types commonly diagnosed among POT1-associated hereditary cancer syndrome families

Abstract

Background

The shelterin complex is composed of six proteins that protect and regulate telomere length, including protection of telomeres 1 (POT1). Germline POT1 mutations are associated with an autosomal dominant familial cancer syndrome presenting with diverse malignancies including glioma, angiosarcoma, colorectal cancer, and melanoma. Although somatic POT1 mutations promote telomere elongation and genome instability in chronic lymphocytic leukemia, the contribution of POT1 mutations to development of other sporadic cancers is largely unexplored.

Methods

We performed logistic regression, adjusted for tumor mutational burden, to identify associations between POT1 mutation frequency and tumor type in 62,368 tumors undergoing next-generation sequencing.

Results

A total of 1,834 tumors harbored a non-benign mutation of POT1 (2.94%), of which 128 harbored a mutation previously reported to confer familial cancer risk in the setting of germline POT1-deficiency. Angiosarcoma was 11 times more likely than other tumors to harbor a POT1 mutation (P=1.4×10⁻²⁰), and 65% of POT1-mutated angiosarcoma had >1 mutations in POT1. Malignant gliomas were 1.7 times less likely to harbor a POT1 mutation (P=1.2×10⁻³) than other tumor types. Colorectal cancer was 1.2 times less likely to harbor a POT1 mutation (P=0.012), while melanoma showed no differences in POT1 mutation frequency versus other tumors (P=0.67).

Conclusions

These results confirm a role for shelterin dysfunction in angiosarcoma development, but suggest that gliomas arising in the context of germline POT1 deficiency activate a telomere lengthening mechanism that is uncommon in gliomagenesis.

INTRODUCTION

Eukaryotic telomeres, located at the ends of linear chromosomes, function to protect against the loss of genomic content during repeated rounds of cellular replication.¹ The shelterin complex is a protein complex that binds and protects human telomere ends and regulates telomere length.^2,3 Protection of telomeres 1 (POT1) is one of six proteins in the shelterin complex and one of the core components that directly bind to telomeres. POT1 loss causes telomere uncapping and leads to unprotected telomere ends, aberrant telomere lengthening, and chromosomal fusions.^4,5 As such, telomere dysregulation caused by POT1 deficiency is associated with tumorigenesis.⁶

POT1 is frequently mutated in aggressive forms of chronic lymphocytic leukemia (CLL).^7–11 Because POT1 mutations arise early in CLL pathogenesis, they are believed to function as drivers in disease progression.^7–11 Furthermore, germline POT1 mutations have been shown to underlie a number of hereditary familial cancer syndromes involving CLL,¹¹ glioma,¹² melanoma^13–18 and colorectal cancer.¹⁹ Furthermore, a deleterious germline missense variant in POT1 (p. R117C) was identified in three Li-Fraumeni-like (LFL) families with angiosarcomas,²⁰ and additional POT1 variants were subsequently identified in sporadic angiosarcoma patients.²¹

Tumor-based next-generation sequencing (NGS) panels are increasingly common in clinical practice, and disease-associated variants identified from tumor sequencing can provide clinical insights.^22–24 The frequency and spectrum of POT1 mutations in solid tumors, especially those that are associated with POT1 germline deficiency, have not been rigorously evaluated. In the current study, we obtained NGS data from 62,368 tumors and reviewed POT1 mutation frequency in pan-cancer analyses.

METHODS

Next Generation Sequencing

This study involved collection of existing data and publicly available diagnostic specimens, and the information gathering process precludes direct and indirect identification of subjects, which therefore exempts it from requiring institutional review board approval under HHS regulations at 45 CFR 46.101(b). POT1 sequencing data were obtained for 62,368 tumors from Caris Life Sciences. Each tumor was reviewed by board certified pathologists and has been classified into one of 172 tumor categories based on World Health Organization (WHO) classification guidelines.

NGS was performed on genomic DNA isolated from formalin-fixed paraffin-embedded (FFPE) tumor samples using the NextSeq platform (Illumina, Inc., San Diego, CA). Tumor specimens were collected via manual microdissection and harvesting targeted tumor tissue. Matched normal tissue was not sequenced. A custom-designed SureSelect XT assay was used to enrich 592 whole-gene targets, including POT1 (Agilent Technologies, Santa Clara, CA). All variants were detected with >99% confidence based on allele frequency and amplicon coverage with an average sequencing depth across all amplicons >500x and an average depth across POT1 amplicons >750x. DNA sequencing results are validated per Clinical Laboratory Improvement Amendments (CLIA) and Internal Organization for Standardization (ISO) requirements. All analyses were completed in the same testing environment with the same sequencing and analytic pipeline, and in compliance with CLIA and ISO regulations.

Genetic variants were categorized according to the American College of Medical Genetics and Genomics (ACMG) standards (Richards et al., 2015, Tables 3–5) as: “pathogenic”, “presumed pathogenic”, “variant of unknown significance”, “unclassified”, “presumed benign”, or “benign”, according to ACMG guidelines.²⁵ “Presumed benign” and “benign” mutations were excluded from POT1 mutation frequency analyses. The remaining categories are collectively referred to in this study as “non-benign”. A flowchart of variant classification is included in Supplementary Figure 1. A sensitivity analysis was also performed limiting to only the variants classified as “pathogenic” and “presumed pathogenic”.

Tumor Mutation Burden

Tumor mutation burden (TMB) was measured by counting all non-synonymous single-nucleotide variants found per tumor (592 genes). TMB was calculated in accordance with the Friends of Cancer Research TMB harmonization project.²⁶ Nonsynonymous, nonsense, in-frame deletion and frameshift variants were added after filtering out presumed germline variants determined from a cosmopolitan pooled reference database including: the Genome Aggregation Database (release 2.1), the Single Nucleotide Polymorphism Database human build 137, and the Caris in-house benign database. Per tumor sample, a total of 1.4 Mb was sequenced.²⁷ Although comparison of tumors to a reference database rather than to matched germline specimen may inflate estimates of TMB, the similar treatment of all sequenced samples in these analyses and the comparison of these samples to a common reference database suggests that modest inflations of TMB are unlikely to create substantial biases in downstream analyses across samples.

Statistical Analysis

Data were analyzed using multivariate logistic regression in SPSS v25.0 (SPSS Inc., Chicago, IL, USA). POT1 mutation status (mutated/non-mutated) was used as the outcome of interest and tumor classification was used as the predictor of interest, adjusting for TMB to prevent passenger mutations in highly-mutated tumor types (e.g., melanoma) from confounding associations. TMBs of tumors were compared among different POT1 mutation categories using the non-parametric Mann-Whitney U-test.

RESULTS

Each of 62,368 tumors was classified into one of 172 tumor categories, with 83 tumor categories including more than 50 cases (Supplementary Table 1). A total of 1,834 tumors harbored a non-benign mutation of POT1 (2.94%), with all mutations having variant allele frequencies ≥7% (mean 34.8%, median 33%). Of these, 128 tumors harbored a mutation previously reported to confer familial cancer risk in the setting of germline POT1-deficiency (Figure 1, Supplementary Table 2). Logistic regression analyses were conducted in tumor types previously associated with germline POT1 deficiency, including: 2,200 malignant gliomas, 1,735 melanomas, 8,904 colorectal cancers, and 86 angiosarcomas. The database is almost exclusively comprised of solid tumors, so hematologic malignancies (N=306), including CLL (N=10), were not assessed.

Figure 1:

Summary of reported POT1 mutations in POT1-associated hereditary familial cancer syndromes.

After adjustment for TMB, angiosarcoma was 11 times more likely than all other tumors combined to harbor a POT1 mutation (N=20/86, 23%; P=1.4×10⁻²⁰) (Table 1). Deleterious mutations were found on both the DNA oligonucleotide/oligosarccharide and the TPP-1-binding domains of POT1 (Figure 2, Supplementary Table 3). Of these 20 POT1-mutated angiosarcoma, 12 tumors harbored two POT1 mutations and 1 harbored three POT1 mutations. No angiosaroma harbored >1 frameshift or nonsense mutation (Supplementary Table 4). However, the average variant allele frequency difference between POT1 mutations within the same tumor was 13.5%, suggesting a mutational process wherein a “second hit” in POT1 arises later during clonal evolution of the tumor. We did not observe any evidence that second hits in POT1 were more or less likely to affect the same POT1 protein domain (i.e. DNA-binding versus TPP1-binding) as the more clonal mutation.

Table 1:

Association of POT1 mutation frequency with tumor type among cancers known to be associated with germline POT1mutations.

Tumor Type	N	Average Tumor Mutation Burden (mutations/Megabase)	Percent with POT1 mutationa	Odds Ratio (95% Confidence Interval)b	P-value
Angiosarcoma	86	8.91	23.26%	10.99 (6.62–18.18)	1.40 × 10−20
Malignant gliomac	2200	8.03	1.73%	0.58 (0.41–0.80)	1.2×10−3
Colorectal cancer	8904	10.48	2.59%	0.83 (0.72–0.96)	0.012
Melanoma	1735	21.36	3.69%	0.95 (0.73–1.23)	0.67

^aThe following are included in the calculation of percent with POT1 mutation: “mutated, pathogenic”, “mutated, presumed pathogenic”, “mutated, variant of unknown significance”, and “unclassified mutation”. 20 angiosarcomas harbored a POT1mutation, of which 12 harbored a second POT1 mutation and 1 harbored three POT1 mutations. 38 gliomas harbored a POT1 mutation, of which 3 also harbored a second POT1 mutation. 231 colorectal cancers harbored a POT1 mutation, of which 14 also harbored a second POT1 mutation. 64 melanomas harbored a POT1 mutation, of which 5 harbored a second POT1 mutation.

^bAssociation is adjusted for tumor mutation burden.

^cIncludes glioblastoma, anaplastic astrocytoma, diffuse astrocytoma, anaplastic oligodendroglioma, and oligodendroglioma.

Figure 2:

Summary of POT1 mutations in angiosarcoma. Deleterious mutations were found on both the DNA oligonucleotide/oligosarccharide and the TPP-1-binding domains of POT1. Variants are listed with complete HGVS nomenclature in Supplementary Table 3.

Glioblastoma and other malignant gliomas were 1.7 times less likely than all other tumors combined to harbor a POT1 mutation (N=38/2200, 1.73%; P=0.0012). Of the 38 POT1-mutated gliomas, only 3 harbored a second POT1 mutation. Colorectal cancer was 1.2 times less likely to harbor a POT1 mutation than all other tumors combined (N=231/8904, 2.59%; P=0.012), with 14/231 harboring a second POT1 mutation. Melanoma showed no difference from other tumor types in POT1 mutation frequency (N=64/1735, 3.69%; P=0.67), and 5/64 POT1-mutated melanomas harbored a second POT1 mutation. Sensitivity analyses limiting to only those variants classified as “pathogenic” or “presumed pathogenic” were largely consistent with results from our primary analyses, with POT1 associations being strengthened in the angiosarcoma group, attenuated in the glioma group, and remaining approximately the same in the colorectal cancer and melanoma groups (Supplementary Table 5). All POT1 variants classified as “pathogenic” or “presumed pathogenic” across all tumor histologies are listed in Supplementary Table 6.

Among glioma histologies, glioblastoma was 1.56 times less likely than other tumors to harbor a POT1 mutation (N=30/1602, 1.87%; P=0.019) after adjustment for TMB. Neither anaplastic astrocytoma nor diffuse astrocytoma displayed significant differences in the frequency of POT1 mutation (N=6/230, 2.61%; P=0.82 and N=1/125, 0.8%; P=0.23). There were no POT1 mutations identified in oligodendrogliomas or anaplastic oligodendrogliomas (N=0/156) (Table 2).

Table 2:

Association of POT1 mutation frequency with tumor type among glioblastoma and other malignant gliomas.

Tumor Type	N	Percent with POT1 mutationa	Odds Ratio (95% Confidence Interval)b	P-value
Glioblastoma, WHO grade IV	1602	1.87%	0.64 (0.44–0.93)	0.019
Anaplastic Astrocytoma	230	2.61%	0.91 (0.40–2.07)	0.822
Diffuse Astrocytoma	125	0.8%	0.30 (0.04–2.14)	0.229
Oligodendroglioma and anaplastic oligodendroglioma	156	0%	N/A	N/A

^aThe following groups are included in the calculation of Percent with POT1 mutation: “mutated, pathogenic”, “mutated, presumed pathogenic”, “mutated, variant of unknown significance”, and “unclassified mutation”.

^bAssociation is adjusted for tumor mutation burden.

Overall median TMB for 62,368 tumors was 7 mutations/Mb. The median TMB for tumors harboring a POT1 mutation characterized as “non-benign” was 11 mutations/Mb, significantly higher than the median TMB for tumors lacking a POT1 mutation (P<0.0001) and the median for tumors harboring a POT1 mutation characterized as “benign or presumed benign” (median=8.5; P<0.0001) (Figure 3). Because TMB was higher in tumors with non-benign POT1 mutations than those with benign POT1 mutations, the observed association is unlikely to be attributable to more-highly mutated tumors simply harboring more mutations in any queried gene.

Figure 3:

Box plots of average TMB for different categories of POT1 mutation status.

In addition to affecting genomic instability, POT1 mutations have been shown to augment telomere length,²² helping cells escape replicative senescence. Although TERT promoter mutation data were unavailable, ATRX mutations that lead to the alternative lengthening of telomeres (ALT) phenotype were available for analysis. Across all 62,368 tumors, logistic regression revealed that ATRX-mutated tumors were 1.7 times less likely to have a POT1 mutation after adjusting for TMB (P=0.001). However, this association was not observed when analyses were limited to angiosarcoma (P=0.999), glioma (P=0.398), colorectal cancer (P=0.398), or melanoma (P=0.369).

DISCUSSION

Our study is the first to report the prevalence of POT1 mutations in a diverse set of solid tumors, revealing that angiosarcomas are 11 times more likely than other tumor types to harbor a POT1 mutation, often harboring multiple POT1 mutations. This indicates that shelterin complex dysfunction may be a key contributor to angiosarcoma development. Pathogenic mutations were detected in both the oligonucleotide/oligosarccharide and the TPP-1-binding domains of POT1, previously associated with abnormally long telomeres and telomere fragility,^{10,15,28–29} genomic instability, and tumorigenesis.^4,30–32 Further studies are required to reveal which of the reported POT1 mutations are particularly damaging and which POT1-dependent molecular mechanisms are critically involved. Notably, five deleterious POT1 variants have been reported in sporadic angiosarcoma patients and angiosarcoma patients in the context of Li-Fraumeni-Like Syndrome (LFLS).^20–21 These POT1 mutations were distributed across all conserved domains. Our analyses of angiosarcoma tumor genomes revealed a high frequency of POT1 mutations distributed in a similar pattern, complementing other emerging studies that have observed a high frequency of POT1 mutations in angiosarcoma.³³ Because the database lacks family history data, the cases profiled in the current study may contain patients meeting criteria for LFLS.

We previously identified three germline POT1 variants in association with familial glioma.¹² While all affected individuals in these families were found to have oligodendroglioma or mixed oligoastrocytoma, a subsequent molecular analysis revealed that at least one proband had a molecular astrocytoma according to revised 2016 WHO criteria (i.e. IDH-mutated, 1p/19q intact).³¹ Here we have identified POT1 mutations in the tumor genome of astrocytic-lineage gliomas (e.g. glioblastoma, anaplastic astrocytoma) but not oligodendroglioma. The frequency of POT1 mutations was lower in glioma, particularly high-grade malignant glioblastoma, relative to other tumor types, indicating that POT1 mutations are likely not a major contributor to gliomagenesis in non-familial cases. In our previous observations of 20 familial glioma patients, those harboring a germline mutation in the shelterin complex genes POT1 or TERF2 were the only patients not to have a somatic TERT promoter or ATRX mutation.³¹ As the vast majority of gliomas maintain telomere length by reactivating TERT expression through hotspot TERT promoter mutations³⁴ or by activating ALT through loss of ATRX,³⁵ it is possible that, in rare instances, POT1 mutations activate a non-canonical mechanism of ALT via shelterin dysfunction. As pathways of telomere maintenance are associated with clinical features, including age at diagnosis³⁶ and glioma patient survival,³⁵ gliomas caused by shelterin dysfunction may also harbor unique clinical presentations and therapeutic vulnerabilities.

The frequency of POT1 mutations appear lineage specific and colorectal cancer was another cancer type that was less likely to harbor a POT1 mutation. Three putatively pathogenic germline POT1 variants have been identified in colorectal cancer patients.¹⁹ One of these variants, D617Efs was also identified in familial glioma.^12,19 A shared risk allele could confer pleiotropic risk of both colorectal cancer and glioma through a telomere-dependent mechanism. This has precedent, as specific polymorphisms in the telomerase component TERC have been associated with longer telomere length and increased risk of both colorectal cancer and high-grade glioma.^37–38

Although CDKN2A is the primary high-penetrance melanoma predisposition gene,^39–40 germline POT1 variants have been reported in melanoma-prone families by several groups.^13–18 More than a dozen pathogenic germline mutations of POT1 have been identified in CDKN2A-wild-type melanoma kindreds, distributed across the DNA-binding and TPP1-binding domains. Melanoma has among the highest TMB of any cancer due to its association with UV-induced DNA damage.⁴¹ In our dataset, melanoma had a TMB of 21.36 mutations/Mb, substantially higher than the average TMB across all tumors (10.12 mutations/Mb). Associations between POT1 mutation frequency and tumor histology can be confounded by TMB due to the presence of passenger mutations that do not have biological or phenotypic consequences.⁴² However, our analyses adjusted for TMB to reduce the possibility of this confounder and observed no differences in POT1 mutation frequency in melanoma versus other tumor types.

A limitation to our study was that the NGS data were processed in the setting of tumor-only sequencing, which precludes the ability to definitively differentiate rare germline variants from somatically-acquired mutations.²³ As a result, a portion of the mutations identified in our data may indeed be rare or previously-unreported germline variants. It is worth mentioning that many identified POT1 mutations – including in angiosarcoma – were detected in substantially less than 50% of sequencing reads, suggesting a somatic rather than a germline origin and potentially a role for POT1 in disease progression as opposed to simply tumor initiation. Although the inability to definitively distinguish germline from somatic variants does not impact our interpretation that POT1 dysfunction is a hallmark of angiosarcoma development and an uncommon driver of tumorigenesis in sporadic glioma, melanoma and colorectal cancer, it does limit our ability to directly translate findings to familial cancer syndromes or risk counseling. NGS using paired tumor-normal tissue would help to separate germline from somatic mutations and should be a focus of future research efforts, especially in angiosarcoma. Additionally, our POT1 sequencing data included only the coding region of the gene. This approach may miss functionally-relevant mutations in non-coding regions. However, with the exception of splice-site mutations, pathogenic non-coding variants are typically activating in nature (e.g. TERT promoter mutations), whereas relevant POT1 mutations appear to be loss-of-function or hypomorphic variants.

In summary, we evaluate POT1 mutation status in a pan-cancer analysis and report the prevalence of POT1 mutations across tumor types, including those previously associated with germline POT1 mutations. The high prevalence of POT1 mutations in angiosarcomas suggests that loss of POT1 function is a major driver of angiosarcoma development and that angiosarcoma sequencing studies should prioritize POT1 analyses. The observation that glioblastoma and anaplastic astrocytoma, and to a lesser extent colorectal cancers, were less likely to harbor POT1 mutations than other tumor types suggests that when these tumors arise in the context of germline POT1 deficiency they are utilizing non-canonical pathways of telomere maintenance. This is underscored by our observation that POT1-mutated tumors had higher TMB, fewer ATRX mutations, and our previous report that glioma patients with germline POT1 mutations lacked a hallmark telomere-maintenance mutation in either ATRX or the TERT promoter.³¹ Our results provide insight into the various roles that POT1 may play across a diverse group of tumors.

Supplementary Material

Table S5

Supplementary Table 5: Association of POT1 mutation frequency with tumor type among cancers known to be associated with germline POT1 mutations, limiting analyses to only “pathogenic” and “presumed pathogenic” variants.

Table S4

Supplementary Table 4: POT1 mutation data for angiosarcomas harboring >1 POT1 mutation.

Table S6

Supplementary Table 6: List of POT1 mutations classified as “pathogenic” or “presumed pathogenic” in data analyses.

Table S3

Supplementary Table 3: Complete HGVS nomenclature of POT1 variants appearing in Figure 2.

Table S2

Supplementary Table 2: POT1 variants previously reported in hereditary familial cancer syndromes and the frequency at which they were detected among 62,368 tumors undergoing clinical sequencing.

Table S1

Supplementary Table 1: List of the 172 tumor categories that contain more than 50 samples, with number of cases and number of POT1 mutations noted for each category.