Instrument Gauge and Type in Uveal Melanoma Fine Needle Biopsy: Implications for Diagnostic Yield and Molecular Prognostication

Abstract

PURPOSE:

To systematically evaluate and compare the effects of using small-gauge needles and vitrectors on the ability to obtain adequate diagnostic and prognostic uveal melanoma biopsy specimens.

DESIGN:

Comparative evaluation of biopsy instruments.

METHODS:

Survival of uveal melanoma cells was evaluated in vitro following needle aspiration. Five therapeutically enucleated eyes were sampled in triplicate for ex vivo diagnostic biopsy experiments with 25 gauge (25 G) needle, 27 gauge (27 G) needle, and 27 G vitrector. During surgery in 8 patients, paired diagnostic transscleral fine needle aspiration biopsies were performed using both 25 G and 27 G needles. A review of cytologic specimens was performed by a panel of 3 expert cytopathologists. A retrospective chart review was performed to evaluate 100 consecutive tumors undergoing prognostic biopsy for gene expression profiling to assess the relationship between needle gauge and prognostic adequacy.

RESULTS:

No significant cell shearing of uveal melanoma cells occurred in vitro with 25 G, 27 G, or 30 G needles. For ex vivo biopsy samples, diagnostic yield was 100% using 25 G needle (5/5) or 27 G vitrector (5/5) but 60% using a 27 G needle (3/5). For in vivo samples, no difference in diagnostic yield was found between 25 G (75%, 6/8) or 27 G (75%, 6/8) needle sizes. Of 100 molecular prognostic biopsy samples evaluated, 65 were obtained using 27 G needles; for these biopsies, the prognostic yield was 65/65 (100%).

CONCLUSIONS:

For diagnostic biopsy of uveal melanoma, a larger-gauge needle or a 27 G vitrector may have better overall cellularity and diagnostic yield when compared to a 27 G needle. However, for much more common molecular prognostic testing, a 27 G needle provided adequate sample in 100% (65/65) of cases, and a larger needle provided no additional benefit.

UVEAL MELANOMA IS THE MOST COMMON PRIMARY intraocular malignancy, arising from melanocytes that reside in the uveal tract. Diagnosis of uveal melanoma is most often made through a combination of detailed physical examination and ancillary imaging studies. In this manner, the rate of clinical misdiagnosis can be as low as 0.48% for larger melanomas.¹ However, clinical misdiagnosis can range as high as 6%–9%, especially in the cases of smaller tumors that must be differentiated from benign nevi or other disease processes.^2,3 Despite advances in local treatment and tumor control, uveal melanoma continues to have a high rate of metastasis and mortality.^4–6 Most fatalities resulting from uveal melanoma can be attributed to metastatic spread to the liver; indeed, as many as 50% of patients go on to develop metastases to the liver, resulting in death.^4–6 The risk of future metastatic spread can be accurately prognosticated based on genetic characteristics of cells within the primary intraocular tumor.

For uveal melanoma, both diagnosis and molecular prognostication of metastatic risk can be aided by fine needle aspiration biopsy (FNAB) of tumor samples.⁷ Aspiration biopsies involve the use of a small-caliber needle through which a suction force is applied to extract a small amount of the tissue of interest. Specimens for diagnosis or molecular prognostication can be collected by FNAB, typically at the time of radiotherapy surgery. The amount of material required to obtain a sufficient specimen depends on whether the indication for the biopsy is to provide diagnostic or prognostic information. The most common indication for tumor sampling through biopsy is to gain prognostic information, which allows for more informed counseling of patients about the risk of metastasis and life expectancy. The commercially available DecisionDx-UM gene expression profile (GEP) test (Castle Biosciences, Inc, Friendswood, Texas, USA) evaluates mRNA expression of 12 genes that are differentially expressed in class 1 vs class 2 tumors.^8,9 This RNA-based assay requires significantly less tissue than traditional histologic classification of metastatic risk based on epithelioid vs spindle cell type, and less tissue than is required for diagnostic biopsies based on cytopathologic analysis. Thus, the amount of tissue that is sufficient for prognostic biopsies using GEP testing is substantially less than for diagnostic biopsies.

Several techniques for FNAB exist, including trans-scleral biopsy as well as multiple transvitreal approaches, using either a needle or a vitrector.^8–23 However, FNAB has potential risks that must be balanced with the need for obtaining adequate specimen yield. Using a smaller-caliber needle may reduce the risks of complications such as hemorrhage, retinal tear and detachment, and extraocular tumor seeding.^10,11 Accordingly, the lowest rate of complications was published in biopsies obtained with a 27 gauge (27 G) needle.¹² However, there are concerns that the use of smaller-caliber needles may reduce sample yield, and thus may more frequently provide an insufficient specimen. While the use of smaller gauge needles and vitrectors has become more common, techniques are not standardized and the effect of smaller-caliber needles on diagnostic and molecular prognostic yield has not been systemically evaluated.

We describe a series of in vitro, ex vivo, and in vivo studies to evaluate the impact of needle or vitrector size on specimen adequacy for uveal melanoma biopsies, with the goal of guiding the choice of biopsy instrument to achieve sufficient sample yield for FNAB both for diagnosis and for molecular prognostication of metastatic risk.

METHODS

PARTICIPANTS:

This retrospective case series was approved by the Institutional Review Board at Vanderbilt University Medical Center. All research adhered to the tenets of the Declaration of Helsinki and the Health Insurance Portability and Accountability Act. Written informed consent was obtained from all participants for all procedures performed.

IN VITRO EXPERIMENTS:

Uveal melanoma cell lines representing different histologic subtypes were cultured: Mel270 (spindle), OM431 (epithelioid), OCM1 (mixed). All 3 grow as adherent cell lines. Trypsin was added to elevate cells and the concentration of cells was counted. Then 5 × 10⁴ cells were aspirated and expelled through 30 gauge (30 G), 27 G, or 25 gauge (25 G) needles into a well of a 96-well plate, and percentage of viable cells was determined using CellTiter-Blue assay (Promega) according to the manufacturer’s instructions using a spectrophotometer with fluorescence measured at 560/590 nm. The control was the same number of cells plated directly into the plate well without being aspirated and expelled through a needle first, and then counted using the same CellTiter-Blue assay. Absorbance following needle aspiration was compared to the unaspirated control and was expressed as percentage of unaspirated control. Each needle gauge, as well as the control unaspirated cells, were performed as 5 replicates, which were averaged for each gauge. GraphPad Prism, version 8 (GraphPad Software, La Jolla, California, USA), was used for statistical analysis.

EX VIVO BIOPSIES:

Ex vivo biopsies were performed on 5 freshly enucleated eyes. Globes were bisected to expose the tumor and specimens were collected under direct visualization from each eye using 25 G needles, 27 G needles, and 27 G vitrectors on a low cut rate setting. For each instrument/gauge, 3 separate samples were collected from different locations within the tumor, using a completely new needle (and tubing and syringe) for each aspiration sample. Each sample was expelled using air from the syringe, and smears of the specimens were prepared by the surgeon. A panel of 3 experienced cytopathologists (A.C., S.J.S., V.L.W.), masked to needle gauge for each specimen, determined cellularity and diagnostic adequacy by consensus.

IN VIVO DIAGNOSTIC BIOPSIES:

We reviewed specimens obtained from 8 patients who had undergone clinically indicated biopsy with both 25 G and 27 G needles. In all cases, biopsies were performed trans-sclerally using a needle connected with IV tubing to a 10-mL syringe. (Our set-up for aspiration biopsies has been described previously; see Figure 1 in Snyder and Daniels.¹²) Specimens were obtained under manual suction, prepared by a cytotechnologist in the operating room using rapid on-site evaluation (ROSE), and needle washings were collected as a second specimen. A panel of 3 experienced cytopathologists (masked to needle gauge for each specimen) determined cellularity and diagnostic adequacy of the original specimens by consensus.

FIGURE 1.

Cell survival of uveal melanoma cell lines following aspiration and expulsion from various small-gauge needle sizes to simulate fine needle aspiration biopsy. Smaller-gauge needles did not decrease cell viability of any of the 3 cell lines tested: Mel270 (spindle cell morphology), OM431 (epithelioid cell morphology), or OCM1 (mixed spindle and epithelioid cell morphology). Experiments were performed as 5 replicates per cell line, and viability within each cell line following passage through each needle gauge was compared to the unaspirated control and to other needle gauges using 1-way analysis of variance (ANOVA) with Tukey’s multiple comparisons. Each is expressed as a percent of the unaspirated controls, and error bars represent standard deviations. Significance was set at P < .05 and no decreases in viability were found. Mel270: 1-way ANOVA, P = .0865; Tukey’s multiple comparisons: 25 G vs 27 G, P = .9681; 25 G vs 30 G, P = .2462; 27 G vs 30 G, P = .116. OCM1: 1-way ANOVA, P = .9262; Tukey’s multiple comparisons: 25 G vs 27 G, P = .9994; 25 G vs 30 G, P = .9845; 27 G vs 30 G, P = .9953. OM431: 1-way ANOVA, P = .6982; Tukey’s multiple comparisons: 25 G vs 27 G, P = .9832; 25 G vs 30 G, P = .9997; 27 G vs 30 G, P = .9677.

RETROSPECTIVE STUDY OF PROGNOSTIC BIOPSY IN CLINICAL PRACTICE:

We reviewed the last 100 patients who consented to undergo clinically indicated biopsy at the time of treatment for molecular prognostication of metastatic risk. We recorded surgical approach, needle or vitrector gauge, GEP class, discriminant value, tumor location, tumor height, and largest basal diameter. Largest basal diameter was calculated based on funduscopy combined with ultrasonography using the accurate methods described recently. The primary outcomes were the ability of the GEP test to successfully assign a molecular prognostic class and the discriminant value, which is a measure of confidence of the prognostic call being made. Our technique for transvitreal biopsy has been published previously.¹² The external comparator group was those samples from among the 100 patients that had been obtained at the time of enucleation, from which the amount of tissue that can be obtained is essentially limitless.

RESULTS

EFFECT OF NEEDLE GAUGE ON CELL SHEARING AND VIABILITY:

To evaluate the effect of needle gauge on cell membrane integrity, cultured human uveal melanoma cell lines of various morphologies (epithelioid, spindle, mixed) were aspirated and expelled through different gauge needles to mimic biopsy sample collection, and residual intact cells were counted to calculate percent viability, relative to unaspirated cells. Needle passage caused no decrease in cell viability for any needle gauge (Figure 1). Importantly, there was no increase in loss of cellular viability when decreasing the needle gauge from 25 G to 27 G (OCM1, P = .9994; OM431, P = .9832; Mel270, P = .9681), or even from 25 G to 30 G (OCM1, P = .9845; OM431, P = .9997; Mel270, P = .2462).

EX VIVO DIAGNOSTIC BIOPSIES:

We performed biopsies on 5 freshly enucleated eyes using 25 G needle, 27 G needle, and 27 G vitrector (Figure 2A). For each eye, 3 biopsies were obtained with each gauge/instrument. All (5/5) tumors biopsied using a 25 G needle or 27 G vitrector contained adequate cellular material for diagnosis; however, only 3 of 5 tumors biopsied using a 27 G needle had sufficient material for cytologic diagnosis (Table). The 3 pathologists were in complete agreement in all cases.

FIGURE 2.

Cytopathologic evaluation of ocular melanoma aspirates. (A) Diff-Quik preparation of ex vivo fine needle aspirates performed using a 27 gauge needle, 25 gauge needle, and 27 gauge vitrector. For each gauge, 3 separate aspiration samplings were performed from different locations within each tumor (using separate needles/set-ups). (B) Hematoxylin-eosin preparation of in vivo fine needle aspirates performed using 27 gauge and 25 gauge needles. (C) Cell block preparation following 27 gauge vitrector aspiration. All images obtained at 20× magnification, scale bar 50 μm.

TABLE.

Cellular and Diagnostic Yields for Diagnostic Biopsies

	Ex Vivo Diagnostic Biopsiesa
	27 G Vitrector	25 G Needle	27 G Needle
Total	5	5	5
Cellular	5	5	3
Diagnostic	5	5	3
% Diagnostic	100	100	60
	In Vivo Diagnostic Biopsiesb
		25 G Needle	27 G Needle
Total		8	8
Cellular		7	7
Diagnostic		6	6
% Diagnostic		75	75

G = gauge

^aCellularity and diagnostic adequacy were determined by a panel of 3 masked, experienced cytopathologists. There was 100% agreement between the 3 cytopathologists in each case.

^bSamples were expelled by air followed by washings. There was 100% agreement between the 3 cytopathologists in each case.

IN VIVO DIAGNOSTIC BIOPSIES:

The panel of cytopathologists reviewed specimens of 8 patients who had undergone paired 25 G and 27 G trans-scleral biopsies (Figure 2B). Six of 8 samples were cellular and diagnostically adequate, with no difference between biopsies performed with 25 G or 27 G needles (Table). One of 8 in each group had cellular material but was nondiagnostic. One of 8 in each group (different specimens) was acellular. The 3 pathologists were in complete agreement in all cases and so no weighted kappa was calculated.

SPECIMEN ADEQUACY FOR MOLECULAR PROGNOSTICATION IN CLINICAL EXPERIENCE:

We performed a retrospective chart review of our last 100 tumors undergoing prognostic biopsy for GEP. Of 100 tumors, 21 underwent enucleation and specimen was obtained directly. Of the remaining 79 undergoing fine needle biopsies, the needle gauge was reported for 76. The most common needle size was 27 G (65), with 25 G needle (10) and 27 G vitrector (1) being less common. Transvitreal (37) and transscleral (37) approaches were equally common and a transcorneal approach (5, all iris tumors) was less common. The primary outcomes that were evaluated were the ability of the GEP test to successfully assign a molecular prognostic class and the discriminant value of the test, which measures the confidence of the prognostic call.

Only 2 of the 100 biopsies performed had failed amplification. One sample was from an enucleation and 1 sample was from a 25 G trans-scleral biopsy. One fresh tissue sample from an enucleation had reduced confidence (discriminant value = 0.04). Normal confidence, reported as discriminant value >0.1, was reported for all other samples. Specifically, 100% (65/65) of biopsies taken with 27 G needles had adequate yield.

For the 2 largest groups (the 37 patients biopsied in a transvitreal fashion and the 37 patients biopsied in a trans-scleral fashion), we reviewed tumor factors associated with the biopsy approach employed. A transvitreal approach was used for smaller tumors. The average height was 3.67 ± 1.86 mm for transvitreally biopsied tumors vs 6.83 ± 2.34 mm for trans-sclerally biopsied tumors (P < .0001, t test). The average largest basal diameter was 11.84 ± 2.61 mm for transvitreally biopsied tumors vs 15.84 ± 3.57 mm for trans-sclerally biopsied tumors (P < .0001, t test). Similarly, a transvitreal approach was more likely to be used for tumors located posterior to the equator, whereas tumors located anterior to the equator were more likely to be biopsied using a trans-scleral approach (P < .0001, Fisher exact test). Choice of needle gauge correlated closely with surgical biopsy approach, with 27 G needles used more frequently for transvitreal biopsies and 25 G needles used more frequently for trans-scleral biopsies (P = .0426, Fisher exact test).

We also reviewed complications associated with biopsy. No transcorneal or trans-scleral biopsies had complications related to the biopsy, regardless of whether a 25 G or 27 G needle was used. Of the 37 patients in the retrospective study who underwent transvitreal biopsy, 34 used a 27 G needle, employing a technique that we have described previously.¹² Immediate intraoperative complications for 27 G needle transvitreal biopsies included a clot or small trickle of blood at the biopsy site in 28 of 34 (82.3%) patients, small localized subretinal hemorrhage in 9 of 34 (26.5%), and small vitreous hemorrhage in 3 of 34 (8.8%). In all 28 patients with small clot or blood at the biopsy site, the blood was gone by the 3-month postoperative visit. In 7 of 9 (77.8%) patients with localized subretinal hemorrhage, the bleeding occurred into a pre-existing pocket of subretinal fluid overlying/adjacent to the tumor, and all had resolved by the 3-month postoperative visit. Of the 3 patients with vitreous hemorrhage noted intraoperatively, 2 had small amounts of dehemoglobinized blood present at 3 months, and none of these patients required vitrectomy to clear the small amount of residual blood. One patient had a small piece of pigmented tissue become suspended within the vitreous fluid. This tissue was presumed to be tumor tissue and a limited vitrectomy was performed to aspirate this tissue. Of the 3 of 37 transvitreal biopsy patients in whom a 27 G needle was not used, in 2 of the patients a 25 G needle was used; 1 of these patients had a small clot of blood at the biopsy site and 1 had a localized vitreous hemorrhage. The third patient had a transvitreal biopsy with a 27 G vitrector; this patient had a localized vitreous hemorrhage and a localized subretinal hemorrhage into a pre-existing pocket of subretinal fluid. There were no cases of iatrogenic retinal detachment from the biopsy in any patient. There were no cases of epibulbar seeding in any of the patients who underwent trans-scleral (or transvitreal or transcorneal) biopsy. In all 8 patients who underwent diagnostic biopsy, there were no complications, and melanoma was confirmed in all cases. We proceeded with radiotherapy in all 8 cases.

DISCUSSION

CURRENT PRACTICE PATTERNS FOR BIOPSY OF UVEAL MELA-noma vary. The most common approaches to FNAB are a direct trans-scleral approach, in the case of anterior tumors, and a transvitreal approach for more posterior tumors. Other approaches include a transvitreal approach of a needle through a vitrectomy cannula,^12,13 the use of a vitrector to obtain a biopsy specimen,^14–18 and transscleral incisional biopsy using Essen forceps.^19,20 For approaches using needles, calibers range from 23 G to 30 G, with 25 G being most common.²¹ While the use of small-gauge needles may reduce complications, this must be balanced with concerns that their small internal lumen caliber may limit biopsy yield.

There is currently no standardized method to guide instrument selection. Each approach must balance the need for obtaining an adequate tissue sample with the risks for complications. Several authors have suggested that uveal melanoma biopsies can be broken down into categories based on their indication.^22,23 For our study, we focused on the broad categories of diagnostic vs prognostic biopsies and evaluated the ability of different instruments to produce adequate specimen yields for the indicated purpose. We first evaluated the concern that smaller-gauge needles may have the potential to destroy the passaged cells through shearing owing to the decreased intraluminal size of smaller-gauge needles. In our in vitro experiments, we found that needles as small as 30 G do not result in shearing of tumor cells from uveal melanoma cell lines, regardless of whether the cells were epithelioid, spindle, or mixed epithelioid/spindle type in morphology.

Diagnostic biopsies must contain sufficient tissue to allow a cytopathologic diagnosis. Often the diagnosis of uveal melanoma can be made on morphologic appearance alone, but samples must be cellular enough to allow morphologic evaluation.²² In some cases, lack of yield can also be informative, suggesting a more cohesive and thus potentially less malignant tumor.²⁴ In our ex vivo experiments, with biopsies obtained from freshly enucleated eyes, we found a 27 G needle to have decreased diagnostic yield. A specimen adequate for cellular diagnosis was obtained in 3 of 5 eyes with a 27 G needle, compared to 5 of 5 eyes using a 25 G needle and 5 of 5 eyes using a 27 G vitrector. In our more clinically relevant in vivo biopsies (collected with the assistance of an on-site cytotechnologist performing ROSE), we found 25 G and 27 G needles to be similarly adequate for diagnosis. Specimens were diagnostic for 6 of 8 eyes in each condition. We did observe that 27 G needles resulted in generally sparser cellular yields than 25 G, although this did not affect the ability to achieve a diagnostic result. Taken together, for diagnostic biopsies 27 G needles performed very well, but there may be some advantage to using a 25 G needle or a 27 G vitrector, particularly if ROSE is not available.

In the literature, the use of a 25 G vitrector has had good reported success, leading to a tissue diagnosis in 13 of 14 patients in 1 study and 121 of 124 patients in another.^14,25 Recent reports using 27 G vitrectors have demonstrated similar efficacy in sample acquisition, even for small tumors.^26–28 The increased cellular material obtained with a vitrector can also allow for creation of a tissue block (Figure 2C) and immunohistochemical (IHC) characterization. In cases with inconclusive results, access to larger amounts of tissue is especially valuable for performing subsequent IHC and special staining preparations.²² In a study of 33 patients, Faulkner-Jones and associates found that adjunct IHC increased specificity from 67% to 83% and changed the cytologic diagnosis in 3 patients, altering management.²⁹ IHC can also aid in identifying the cell type of origin in tumors found to be not of primary intraocular origin—that is, in identifying the source of a choroidal metastasis.^30,31 The use of a vitrector for diagnostic biopsy therefore has the advantage of providing additional material on which these additional studies can be performed. The use of a 25 G needle does not have the advantage of additional tissue characterization that a vitrector provides and increases the size of the needle track compared to a 27 G needle or a 27 G vitrector. We thus suggest that 25 G needles should perhaps be reserved for trans-scleral diagnostic FNAB, in which a vitrector cannot be easily used.

Increased complications such as retinal detachment have been associated with larger-gauge needles or with the blunt tip of vitrectors when used in a transvitreal (and hence, transretinal) fashion. The most common complication is postoperative vitreous hemorrhage with spontaneous clearance within days, although a small number of patients do require subsequent vitrectomy.^25,26,32 In our study, we had no cases of iatrogenic retinal detachment and no cases of epibulbar seeding. The few cases of postoperative vitreous hemorrhage were largely resolved by postoperative month 3 and none required vitrectomy to clear the small amount of residual, inferior, old blood. Although ocular complications may be associated with melanoma biopsy, and though epibulbar seeding has been reported in rare cases,³³ studies have not supported a link between tumor biopsy and increased risk of hematogenous dissemination or mortality.^34,35 The potential for increased complications, along with the increased cost of using a vitrector pack, must be weighed against the need to obtain adequate tissue when making instrument selection for diagnostic FNAB.

Performing FNAB of uveal melanoma for molecular prognostication of metastatic risk, which aids in patient counseling and guides metastatic surveillance imaging frequency, is a much more common indication. Here, we found no advantage to anything other than a 27 G needle. Our success rate with a 27 G needle of 100% (65/65) is equivalent to that of Correa and Augsburger (158/159), who used a 25 G needle.³⁶ This is likely owing to the minute amount of genetic material required for modern molecular testing, especially for RNA-based tests. For non-RNA-based prognostic tests, tumor size may be more relevant. For example, McCannel and associates found that 30 G needles yielded sufficient material for monosomy 3 testing via fluorescence in situ hybridization for 91% (58/64) of tumors with height >5 mm but only 53% (27/49) of tumors with height <3 mm.³⁴ However, in our experiments using GEP testing, specimen yield was sufficient for almost all tumors regardless of size. For example, 16 of 65 tumors biopsied with a 27 G needle had a height <3 mm, yet 100% (16/16) were diagnostic, in contrast to McCannel and associates’ data using a 30 G needle. Thus, in light of the low complication rate we have reported previously with 27 G needles,¹² there does not seem to be any benefit to a needle smaller than 27 G for GEP testing.

Our results thus suggest that a guiding factor for instrument choice for FNAB of uveal melanoma is biopsy indication. For diagnostic biopsies, the need to obtain larger tissue samples may outweigh the minimally increased risk of complications that comes with a larger needle or a blunt-tip vitrector. Selecting instruments that allow for larger tissue samples may improve yield and allow for further histologic testing. For prognostic biopsies, especially when using RNA-based assays, the amount of tissue needed is very small and thus it is appropriate to select a smaller-caliber needle. In a separate large study of 880 cases performed mostly with a 25 G needle, Correa and Ausburger found that prognostic biopsies were more likely to yield sufficient specimens than diagnostic or confirmatory biopsies (94.7% for prognostic, vs 28.4% for diagnostic and 20.7% for confirmatory).²³ This finding supports our conclusions, namely that the larger amount of tissue needed for diagnostic biopsies warrants a larger-caliber instrument. In contrast, prognostic biopsies are possible with much smaller samples, and therefore it is appropriate to select a smaller needle that minimizes the risk of complications.

Strengths of our study include the use of multiple sampling modalities to directly compare sufficiency of samples obtained using the different instruments. We benefited from collaboration with 3 experienced cytopathologists. Additionally, all biopsies were performed by a single surgeon (A.B.D.), limiting variability owing to individual operative technique. The main limitations of our study were the relatively small sample size for the ex vivo and in vivo biopsies, consisting of 5 and 8 eyes, respectively. Though we recommend the use of a 27 G vitrector for diagnostic biopsies, expense may limit its adoption in some cases. Our study of prognostic biopsies, although benefiting from a relatively large number of patients, was retrospective in nature and thus was not randomized. Additionally, 2 samples from enucleated eyes had amplification problems, 1 with reduced confidence and 1 with multiple gene failure. At the time these 2 samples were collected, our practice was to place a small amount of fresh tissue directly in RNA extraction buffer. We have since moved to sending formalin-fixed paraffin-embedded (FFPE) samples with the rationale that larger fresh tissue samples may not be adequately stabilized by the volume of RNA buffer. Sending FFPE samples also allows for sampling of more regions of the tumor, limiting the potential effects of tumor heterogeneity.^17,37

In conclusion, we propose that uveal melanoma biopsy indication may be used to guide instrument selection. Although a 27 G vitrector for transvitreal biopsies (or a 25 G needle for trans-scleral biopsies) may be helpful to maximize yields for diagnostic biopsies, a 27 G needle achieves excellent yields for molecular prognostication of metastatic risk by GEP, with a potentially more favorable safety profile.