Chromosome 6p amplification detected in blood cell-free DNA in advanced intraocular retinoblastoma

Abstract

Background:

In patients with retinoblastoma, gains of chromosome 6p have been associated with less differentiated tumors. In cell-free DNA from the aqueous humor (AH), 6p gain has been associated with an increased risk of enucleation. While the identification of somatic copy number alterations (SCNAs) via the AH has been well established, these alterations are not routinely identified in the blood due to low tumor fraction.

Materials and methods:

SCNAs were considered positive at 20% deflection from the baseline. Somatic RB1 pathogenic variants were identified with targeted sequencing using a panel including all RB1 exons.

Results:

A 24-month-old patient presented with unilateral retinoblastoma (Group D/AJCC Stage cT2B) and was treated with primary enucleation. In the peripheral blood, a heterozygous mutation (c.3920T>A) in the APC gene was reported. Genomic analysis of the tumor and AH revealed two novel somatic RB1 mutations (c.1589_1590del and c.2330dupC). Both also demonstrated highly recurrent RB-related SCNAs. Chromosome 6p gain was detected in the blood with an amplitude suggesting approximately 12% tumor fraction. At a follow-up of 24 months, there has been no evidence of metastatic disease.

Conclusions:

To our knowledge, this is the first time an SCNA has been detected in the blood of an RB patient, suggesting in some advanced eyes there may be a high enough tumor fraction to detect these alterations (>5% needed). It remains unclear whether 6p gain or increased tumor fraction in the blood is indicative of increased risk of metastatic disease or new primary cancer; studies to address this are ongoing.

Introduction

Retinoblastoma (RB) is the most common primary pediatric intraocular cancer, diagnosed in approximately 8,000 children and infants each year worldwide (1,2). In most cases, tumor formation in RB is in response to biallelic inactivation of the retinoblastoma gene (RB1) tumor suppressor gene, typically caused by single-nucleotide variants or small insertions or deletions in the gene (InDels). After these initiating events, highly recurrent chromosomal alterations—called somatic copy number alterations (SCNAs) – specifically gains on 1q, 2p, 6p, and losses on 13q and 16q are thought to contribute to subsequent tumorigenesis and progression (3-7).

Gains of chromosome 6p have been associated with less differentiated tumors and higher rates of invasion (8). It has most recently also been shown to correlate with severe anaplasia and risk of extraocular spread (9). In previous studies that analyzed the cell-free DNA (cfDNA) in the AH of RB patients, we have identified that a 6p gain is associated with a 10-fold increased risk of enucleation and significantly decreased rate of ocular salvage (10). This highlights the potential value of a 6p gain in the AH in serving as a prognostic biomarker for RB.

The clinical utility of blood biopsies has also been studied in RB, with advantages including being less invasive and having fewer potential side effects than AH paracentesis. These studies have shown that circulating RB1 mutant DNA can also be detected in blood (11,12). However, SCNAs have not been identified in the blood of an RB patient (13); identification of SCNAs via low-pass whole-genome sequencing requires at least approximately 5% tumor fraction for detection. While the fraction of tumor DNA is very high in the AH, it is relatively low in the blood. Herein, we present the first RB case to our knowledge with an SCNA (6p gain) detected in the blood.

Materials and methods

This report is a case study of a patient diagnosed with unilateral retinoblastoma. The patient was diagnosed and treated at Children’s Hospital Los Angeles (CHLA), and patient consent was obtained for participation in the study. Relevant clinical information of the patient described in this report was obtained from retrospective chart reviews. AH, blood, and tumor samples were extracted immediately following primary enucleation.

Copy number analysis of cfDNA was performed as previously described (10,14,15). Briefly, isolated cfDNA was constructed into whole-genome libraries followed by shallow sequencing at 0.3x on an Illumina platform. SCNAs were considered positive at 20% deflection from the human baseline genome (hg38). Correlation between SCNA profiles of different samples from the same patient was analyzed using Pearson correlation coefficient. Whole-genome libraries were also used to identify somatic RB1 pathogenic variants using a comprehensive cancer panel that includes all RB1 exons. Tumor fractions of AH and blood were calculated based on the ratio of the amplitude of 6p gain of each biofluid in comparison to the amplitude of the 6p gain in the tumor, with the assumption that the tumor sample contains solely tumor DNA. Tumor fractions obtained were then verified using ichorCNA. The resulting tumor fractions are not derived by measuring circulating tumor cells, rather by estimating the percentage of sampled cfDNA that is tumor derived. ichorCNA is an established tool for measuring cfDNA tumor fraction through using the amplitude of SCNAs as a surrogate, and has become the standard tool for measuring cfDNA tumor fraction from blood-based biopsies in multiple types of cancers (16-18).

Results

The patient is a 24-month-old male diagnosed with unilateral retinoblastoma in the left eye. The tumor was grouped as International Intraocular Retinoblastoma Classification (IIRC) Group D (19)/AJCC Stage cT2B 20. Bscan ultrasonography showed a dome-shaped tumor covering the posterior pole, with intralesional calcification. The tumor mass measured 9.75 mm in height and 15.5 mm in diameter. Ultrasound biomicroscopy showed tumor mass adjacent to the pars plana inferiorly and nasally without ciliary body thickening. MRI results demonstrated left intraocular retinoblastoma with no evidence of optic nerve or brain involvement. This patient was treated with primary enucleation; there were no high-risk histopathologic features and adjuvant chemotherapy was not required. Fundus and histology images are shown in Figure 1.

Figure 1.

(a) Fundus Image: Advanced Group D retinoblastoma with large dome shaped endophytic retinal masses and spherical vitreous seeds studding the ora. (b) Microphotograph of the retinoblastoma (H&E, A&B 20X, C 40X, D 100X, E&F 400X): (a) the tumor showed a combined endophytic (black arrow heads) and exophytic (white arrow heads) growth. The tumor was negative for optic nerve invasion or choroidal involvement. (b) There were foci of necrosis (black arrow heads) that affect approximately 20% of the tumor. (c) the tumor demonstrated extensive subretinal seeds. (d) Foci of dystrophic calcifications were present within the tumor. (e) the tumor was comprised of “small round blue cells” with hyperchromatic nuclei and scant cytoplasm forming occasional Flexner-Wintersteiner rosettes (black arrow heads) and Homer Wright rosettes. (f) Some areas showed mild to moderate anaplasia with enlarged, irregular-shaped nuclei. Mitotic figures were commonly seen.

In the peripheral blood clinically tested at Invitae (San Francisco), a germline RB1 mutation was not detected, however a heterozygous mutation (c.3920T>A; p.Ile1307Lys) in the APC gene was reported. Genomic analysis of the tumor revealed two somatic RB1 mutations that were not present in germline DNA, c.1589_1590del (p.Lys530SerfsTer24) with a variant allele frequency (VAF) of 84.0% and c.2330dupC (p. Thr778TyrfsTer17) with VAF of 40.0%. Genomic analysis of both tumor and AH demonstrated highly recurrent RB SCNAs such as gains on 1q, 2p, and 6p, in addition to gains on 5p, 7q, 9q, 10q, 13q, 15q, 17q, and loss on 3q. The tumor genomic profile is illustrated in Figure 2. As expected (10-21), the AH SCNA profile was highly concordant with the tumor profile (r = 0.987; p < 0.0001). 6p gain, being the highest amplitude alteration in the tumor, was the only SCNA detected in the blood, with an amplitude suggesting approximately 12% tumor fraction in the blood, consistent with data from the widely used ichorCNA software. The amplitude of 6p gain in the AH suggests a tumor fraction of approximately 99%. The genomic profiles from the AH and blood are illustrated in Figure 3. At a follow-up of 24 months, there has been no signs of metastatic disease.

Figure 2.

Tumor genomic profile. (a) Integrative genomics viewer (IGV) displays the somatic RB1 variants (c.1589_1590del and c.2330dupc). The RB1 gene is located on chromosome 13 and has 178,240 base pairs. Each vertical bar represents one base pair and gray color indicates there is no change in base pair compared with the human reference genome (hg19). (a1) Demonstrates two base-pair deletion (c.1589_1590del) with a variant allele frequency (VAF) of 84.0%. (a2) Demonstrates single nucleotide insertion (c.2330dupc) in purple with a VAF of 40.0%. (B) SCNA profile demonstrating highly recurrent RB SCNAs such as gains on 1q, 2p, and 6p, in addition to gains on 5p, 7q, 9q, 10q, 13q, 15q, 17q and loss on 3q.

Figure 3.

AH and blood genomic profiles. (A) Both SCNA profiles from AH cfDNA and tumor demonstrate highly recurrent RB SCNAs, such as gains on 1q, 2p, and 6p, in addition to gains on 5p, 7q, 9q, 10q, 13q, 15q, 17q and loss on 3q. An overlay of both SCNA profiles illustrates high concordance, with a correlation of r = 0.987 (p <0001). (B) SCNA profile from blood cfDNA demonstrates 6p gain with the width of the gain concordant to that of the tumor and with an amplitude suggesting approximately 12% tumor fraction in blood.

Discussion

The tumor genomic SCNA profile was reflected in the AH cfDNA, in concordance with prior studies demonstrating that the AH is a rich source of eye-specific tumoral genomic information (10,14,22). This case is the first time a significant 6p gain was detected in the blood. It is not currently known if there are systemic associations with 6p gain, however the higher tumor fraction in the blood suggests the potential for metastatic cancer. Studies from Memorial Sloan Kettering Cancer Center (MSKCC) suggest that higher tumor fraction of circulating tumor DNA in the blood has been associated with development of metastatic disease (11,12); however, there has been no objective validation of when there is increased risk based on tumor fraction alone. For this patient, MRI results showed no signs of extraocular spread in the optic nerve or brain at the time of diagnosis; this patient has now been followed clinically for 24 months after diagnosis and there is no evidence of metastatic disease. Thus, at this time, we do not know the risk to the patient with circulating tumor DNA in the blood. Further, despite no evidence in the patient’s MRI, it also remains to be seen if 6p gain in the blood also relates to increased risk of a new primary cancer such as pineoblastoma, which may also have a 6p gain (23).

In addition to being the first RB case with SCNA detection in blood, interestingly this patient did not have a germline RB1 mutation but did have a germline adenomatous polyposis coli (APC) gene mutation. APC encodes a tumor-suppressor protein that has a role in processes such as cell migration and adhesion, transcriptional activation, and apoptosis. Mutations in the APC gene are responsible for familial adenomatous polyposis (FAP), which has a role in a majority of sporadic colorectal cancers (24). In the pediatric population, mutations in APC have also been associated with increased risk of hepatoblastoma (25). This specific APC mutation, c.3920T>A, increases the risk for colorectal cancers by leading to a hypermutable region of the gene resulting in increased cancer predisposition (26,27). A study investigating the association between APC c.3920T>A and non-colorectal cancers reported that this variant was detected in 11.8% of cancer patients compared to 4.7% of healthy subjects (28). Males had significantly increased carrier prevalence in lung, urologic, pancreatic, and skin cancers and females had significantly higher carrier prevalence in breast and skin cancer (28). An association between APC and RB1 variants has not been previously described in RB patients. It remains to be determined whether this germline APC mutation had a contributory role in the unusually high tumor fraction detected in the blood in this patient.

Further, in this patient, we have identified a novel somatic RB1 variant, c.1589_1590del, that is located in the RB1 coding region. It is likely we will continue to identify novel RB1 variants as we continue to study and further develop liquid biopsy platforms for this cancer.

This report is the first time, to our knowledge, an SCNA has been detected in the blood of an RB patient, suggesting in rare cases with a large intraocular tumor burden, there may be high enough tumor fraction to detect this in the blood. Our group has previously published that the AH is superior to blood as a liquid biopsy for retinoblastoma 13 and due to the higher tumor fraction in the AH as well as the fact that it is eye-specific, we continue to stand by this statement. This case highlights the potential for blood monitoring for SCNAs especially in advanced cases of RB. It remains to be seen whether 6p gain in the blood, given the association with anaplasia and invasive features, may portend an increased risk of metastatic disease; in addition whether this risk is potentiated by a high tumor fraction in the blood, as suggested in preliminary studies at MSKCC (11,12). Larger studies with long-term follow-up are needed to evaluate these associations.