Genomic profiling of retinoblastoma using aqueous humor liquid biopsy - A rapid review

Literature search and data extraction

The rapid review was carried out following Cochrane guidelines for rapid reviews.^[18] We focused on cohort studies offering longitudinal data on patient populations. This allowed for a thorough evaluation of the diagnostic accuracy and clinical significance of genomic analysis. We searched three databases, “PubMed,” “MEDLINE,” and “Cochrane Library” on May 24, 2024, using the search terms “Genomic analysis,” “Liquid biopsy,” “aqueous humor,” and “retinoblastoma” or their synonyms. We also performed forward and backward citation tracking and identified relevant articles published in the past 6 years based on these search terms. Screening of articles was conducted using the preferred reporting items for systematic reviews and meta-analyses 2020 screening tool. Two authors evaluated titles and abstracts and any articles with ambiguous eligibility were discussed with all authors to reach a consensus. Case studies, case series, editorials, and duplicates were excluded.

The data were extracted using a data extraction table by two authors and accuracy and completeness of the extracted data were assessed by a third. Extracted data included study aim, targeted population, type of intervention, and key findings. The risk of bias and quality of studies were assessed using Cochrane’s ROBINS-I tool and visualized using ROBVIS tool.^[19,20] Meta-analysis was not possible as the heterogeneity of the studies did not allow for the same and we conducted a narrative review of principles involved in the included studies.

Patient and sample characteristics

The details of the patient and sample characteristics involved in the studies and the study periods have been described in Table 1.

AH Liquid cfDNA concentration (sequential samples) and sequencing technique

RB1 mutations detected in cfDNA from AH samples in the three studies have been summarized in Table 2.

RB 1 gene mutation detection in cfDNA from AH (pathogenic variant detection)

RB1 pathogenic variants were detected in almost all types of samples except for the ones that had alternative mechanisms (MYCN amplification, 6p gain). The mean detection rate of RB1 pathogenic variants ranged from 89.9% to 100% in different types of samples [i.e., diagnosis (Dx), primary enucleation (PE)]. Samples that could not provide RB1 mutation were either because of an underlying MYCN amplification or insufficient cfDNA in the sample (poor sequencing quality, i.e., low total reads). RB1 pathogenic variant detection means for different samples are given in Table 2. Samples from enucleated eyes were concordant with tumor DNA and those from patients undergoing conservative treatment, and the samples were matched against clinical diagnostic criteria.

RB1 variant allele frequency (VAF)

Targeted sequencing of the samples indicated that more than 90% of the cfDNA is tumor derived with high VAF. Both heterozygote and homozygote VAF for RB1 variants were detected and ranged from 45% to 100% in AH samples. Tumor DNA had relatively low levels of VAF as compared to AH (e.g., one of the cases had 56.5% VAF in AH vs. 6.7% in tumor).

Table 1: Patient and sample characteristics.

Study	Number of patients	Bilateral ocular involvement	Enucleation done	Stage of tumor	Follow-up period (months)
Study 1 (Xu et al.)[12]	6	1	2	cT2b (5), cT2a (1), cT1b (1)	17.3 (average)
Study 2 (Gerrish et al.)[8]	68	19	34 (from 68 patients)	Not specified	12
Study 3 (Schmidt et al.)[16]	11	2	11 (all eyes)	cT2b (9), cT1b (1), cT3c (1)	95 days

Close

Figure 1:

Preferred reporting items for systematic reviews and meta-analyses flowchart.

Export to PPT

Table 2: RB1 mutation detection in cfDNA from AH.

Study	Category	Results	Details
Study 1 (Xu et al.)[12]	RB1 mutation detection in cfDNA from AH	- RB1 variants identified in 5/7 AH samples. - VAF ranged from 66.67% to 100%.	- 2 samples with no RB1 variants had alternative mechanisms (e.g., MYCN amplification).
Study 2 (Gerrish et al.)[8]	RB1 mutation detection in cfDNA from AH	- PE: 91% detection rate (50/55) in primary enumeration samples. - SE : 29% detection rate (4/14) in secondary enucleation samples. - Dx1 +: 75% detection rate (21/28) in anterior chamber tap samples. - Tx : 46% detection rate (25/54) in IViC samples.	- 5 new somatic RB1 variants were detected - Detection was higher (95%) in Dx ≤2 cases (≤2 chemotherapy cycles).
Study 3 (Schmidt et al.)[16]	RB1 mutation detection in cfDNA from AH	- 54.5% of cases had RB1 mutations. - 88.9% concordance between AH and tumor samples.	- Some variants were detected only in AH or tumor.

RB: Retinoblastoma, cfDNA: Cell-free DNA, AH: Aqueous humor, VAF: Variant allele frequency, PE: Primary enucleation, SE: Secondary enucleation

RB1 SCNA and SNV

SCNAs and SNVs for RB1 gene were successfully detected in most AH samples. Common RB SCNAs identified included gain of 1q, 2p, 6p, loss of 13q, 16q, etc. Both SCNA and MYCN amplification indicated a poor prognosis in patients. In addition to the detection of SCNAs, high concordance was documented between targeted sequencing and WGS (median 96.2%). Further, some AH samples revealed SCNAs which were not present in their respective tumor samples yielding higher detection rate in AH than tumor samples. However, significance of these SCNAs is yet to be explored.

Close

Figure 2:

Risk of bias assessment.

Export to PPT

These SCNA detections from AH samples were diagnostic and also altered the treatment outcomes in patients. Similar to SCNA, high number of SNVs were also detected in the studies with 33 RB1 SNVs in study 1. Both SCNA and SNV levels were very low in blood samples indicating poor sensitivity in blood. However, it is essential to note that cfDNA quantity directly affects the detection rates (93% detection rate with more than 250 pg DNA input). Eventually, in patients with conservative treatment, the concentration of cfDNA decreased in response to therapy (in patients with RB1 mutations) leading to decreased rate of detection of SCNA and SNV in these samples.

AH for RB monitoring

AH liquid biopsies were shown to be particularly effective in detecting genomic alterations with a high VAF, often surpassing that found in tumor DNA, as noted in several studies.^[8] This review elaborates on the growing potential of genomic profiling using AH liquid biopsies in managing RB. The studies under vision consistently show high detection rates of RB1 gene mutations, with rates ranging from 89.9% to 100% in different types of samples (Dx, PE). The techniques used for the detection of genetic mutations leading to the pathogenesis of RB in the concerned studies identify RB1 pathogenic variants excluding other significant alterations such as MYCN amplification and other specific chromosomal gains. Detection was also not possible in case of inadequate cfDNA in the AH sample. In addition, the ability of AH liquid biopsies to identify specific SCNAs and SNVs that were not present in tumor tissue highlights the superior sensitivity of this method and the significance of this study.

A multitude of screening techniques have been employed for the detection of RB from DNA samples which are used in combination and no single technology carries the efficacy and sensitivity to detect all mutations. Small pathogenic variants are identified through sequencing techniques such as Sanger sequencing and next-generation sequencing (NGS). Array-based methods (array comparative genomic hybridization and single-nucleotide polymorphism arrays), quantitative multiplex polymerase chain reaction, and multiplex ligation-dependent probe amplification (MLPA) can all be used to identify large RB1 deletions or duplications at exonic or chromosomal levels. Sanger sequencing and NGS are used first followed by other techniques like MPLA as germline RB1 mutations are commonly found as small-scale mutations.^[21] RB1 copy number variation can be detected by NGS. However, the findings are validated by Sanger sequencing and MLPA.^[22,23]

Unlike other cancers, direct biopsy of RB carries a risk of tumor seeding and dissemination beyond the eye.^[24] The clinical implications of AH liquid biopsies are profound, offering a non-invasive yet highly effective means of diagnosing, monitoring, and predicting the prognosis of RB.^[25] The detection of genomic markers like MYCN amplifications in addition to commonly studied markers like RB1 paves the way to guide tailored treatment strategies including indications for enucleation particularly in MYCN mutations and chromosome 6p gain mutations, leading to improved patient outcomes.^[12,26-28] Moreover, AH liquid biopsies may play a pivotal role in predicting treatment response, as genomic alterations in AH cfDNA have been shown to correlate with therapeutic outcomes, particularly in cases undergoing conservative management including intravitreal chemotherapy (IViC) (focal therapy), systemic chemotherapy, focal consolidation with transpupillary thermotherapy, laser photocoagulation and cryotherapy, and radiation treatment with plaque brachytherapy.^[12,29]

As demonstrated in studies used for the review as well as various other studies, over time through AH samples allows for dynamic monitoring, the ability to track genomic changes of disease progression, and treatment efficacy. This capability is especially valuable in pediatric oncology, where minimizing invasive procedures is paramount.

Advantages and limitations

The key advantages of AH liquid biopsies include their minimally invasive nature and high concordance with tumor DNA.^{[11,12,27,28,30-32]} This method not only preserves the ocular structure but also provides comprehensive genomic data that can inform clinical decisions, thereby reducing the need for enucleation in many cases.^[9] Furthermore, AH liquid biopsies can detect mutations that might be missed in traditional tumor biopsies, particularly in cases with low tumor cell content.^[33,34] The AH liquid biopsy has demonstrated established analytical validity through its capacity to accurately and reliably identify RB1 pathogenic mutations and SCNAs, consistently achieving mean concordances exceeding 95% between genomic profiles derived from AH samples and their corresponding tumor tissue.^{[11,12,27,28]}

However, there are limitations to consider. A minimal risk is always expected during the procedure (clear corneal paracentesis and genomic detection).^[35] The biopsy needles should only enter the anterior chamber and not make contact with the iris or the lens as it may result in iris scarring or cataracts which may pose setbacks in treatment monitoring. There is also a risk of needle penetrating the vitreous chamber or the tumor which may hypothetically increase the risk of tumor seeding and extraocular dissemination of the disease.^[24,36,37] The quantity of cfDNA in AH samples can vary, impacting the sensitivity of mutation detection, given >10 ng of cfDNA is required for detection which is commonly retrieved at the time of diagnosis or primary enucleation.^[12,28] Some studies have reported that certain SCNAs detected in AH samples may not be clinically significant, highlighting the need for further research to validate these findings. An additional limitation is that SCNAs cannot be identified at TFxs (tumor fractions) lower than 5%, which compromises the ability to monitor disease in cases where the tumor burden has significantly decreased; eyes that show response to IViC frequently result in this observation. Objective methods for monitoring the response to IViC include fundus photography under general anesthesia and B-scan ultrasonography.^[38,39] TFx detection softwares fail to recognize genetic material in cases absence of SCNAs in RB tumors.^[28,40-42] Moreover, while AH liquid biopsies have shown promise, standardization of methods across different studies is necessary to ensure consistent and reliable results.

Comparison with other liquid biopsy approaches

Compared to other liquid biopsy methods, such as blood-based cfDNA in analysis, AH liquid biopsies offer distinct advantages in RB management. The identification of SCNAs in the bloodstream is constrained by reduced ctDNA TFx levels and the challenge of establishing a correlation with each eye in 40% of patients who present with bilateral disease.^[28] Studies have consistently shown higher detection rates for RB1 mutations and other critical genomic markers in AH samples compared to blood, likely due to the higher concentration of tumor-derived cfDNA in AH. This makes AH a more suitable medium for genomic profiling in RB, offering greater diagnostic accuracy and prognostic value.

Various other biomarkers studied include lactate dehydrogenase,^[43] neuron-specific enolase,^[44] survivin,^[45,46] TGF-β1,^[46] and cytokines;^[47] other studies also showed elevated protein content in AH as compared to cataracts as controls.^[48]

Trefoil family factor peptide 1, a peptide secreted in AH which is expressed ectopically in more advanced subset of RB tumors has emerged as a new biomarker under study in recent years expression of which may indicate more severity and metastases.^[49]

Current gaps and future directions

Despite the promising findings, several gaps in the research remain. For instance, the clinical relevance of certain SCNAs detected only in AH samples is not fully understood and warrants further investigation. While the clinical validity of the AH liquid biopsy platform for RB has been confirmed, it is presently authorized solely for research purposes.^{[27,28,30-32,50]} Future research should focus on large-scale, multicenter studies to validate the efficacy of AH liquid biopsies in diverse patient populations. In addition, there is a need to explore the potential of integrating AH liquid biopsy results into routine clinical practice, particularly in combination with other diagnostic tools.

Moreover, advancements in sequencing technologies and bioinformatics could enhance the sensitivity and specificity of AH liquid biopsies, making them an even more powerful tool in the management of RB. Research should also explore the potential of AH liquid biopsies in detecting other ocular tumors, thereby broadening their clinical utility.^[51]