Vitrectomy-Assisted Biopsy: An in vitro Study on the Impact of Cut Rate and Probe Size

Biopsy of intraocular tumors is increasingly used to secure the diagnosis and provide material for histo- and immunohistochemical, molecular, and genetic analyses. Genetic testing of uveal melanoma tissue provides additional prognostic information on the risk of developing metastatic disease, which is important for patient counseling and to ensure appropriate inclusion in future clinical trials [1, 2]. Patients at high risk for metastatic disease can be monitored more closely and receive treatment at an early stage of the disease [3, 4].

Biopsy with a vitreous cutter has proved to provide sufficient material for histopathological and genetic testing of both anterior and posterior segment tumors [3, 5]. Ocular complications are rare, with intraocular hemorrhages and retinal detachment being the most commonly reported [3, 6, 7]. To reduce the risk of complications, the biopsy should be done with as little manipulation of the vitrectomy probe as possible. Still, it is important to obtain enough tissue for analysis, which may be especially difficult in small tumors and tumors in the anterior eye segment considering the limited time and space available for the procedure. The method of using a vitrectomy probe for tissue sampling is well-known and described [7-9]. However, the optimal settings for vitrectomy-assisted biopsies have not been thoroughly investigated. In particular, there is no consensus on the most appropriate cut rate for tissue sampling. Furthermore, vitrectomy probes ranging from 20 to 27 gauges have all been reported to be safe and effective for biopsy of intraocular tumors [3], but direct comparisons between various probe types are lacking. Here, we report the results of an experimental study evaluating the efficacy of vitrectomy-assisted biopsy and comparing the cytohistological results of biopsies obtained by different vitrectomy probes and cut rates in an in vitro porcine model.

The study was conducted at the Departments of Ophthalmology at Haukeland University Hospital, Bergen, Norway, and at Rigshospitalet, University of Copenhagen, Copenhagen, Denmark. At both centers, vitrectomy-assisted biopsies were obtained from fresh ex vivo porcine livers using an identical sampling procedure. The biopsies obtained at the Haukeland University Hospital were prepared for smear analysis, and those obtained at Rigshospitalet were prepared for microscopy by cytospin. The following variable vitrectomy parameters were compared with respect to the cytohistological results: cut rate (60, 600 and 6,000 cuts per minute [cpm]), vitrectomy probe type (standard and two-dimensional cutting [TDC]), probe diameter (23-gauge and 25-gauge), and whether the probe was filled with fluid or air at the start of the procedure.

The biopsies were performed with the Eva vitrectomy system (DORC International, Zuidland, the Netherlands) using either a 23- or 25-gauge standard probe or a 23- or 25-gauge TDC probe. A large portion of the fresh porcine liver was placed in a container filled with saline. The vitrectomy system was first primed, and for half of the 23-gauge standard probe biopsies, air was aspirated so that the probe and tubing was filled with air at the start of the biopsy procedure. The vitrectomy probe was then inserted through the liver capsule and deep into the hepatic parenchyma. Each sampling was performed in the vacuum mode with the vacuum level set at 300 mm Hg. The probe was held still during an aspiration time of 30 s to reduce unintended variations associated with the sampling procedure. Each vitrectomy probe was used for a maximum of 6 consecutive biopsies and then discarded.

For the biopsies prepared at the Haukeland University Hospital, the cutter portal was placed directly on a glass slide. All the aspirated material was expressed by using an air-filled syringe connected to the vitrectomy aspiration tube to gently blow the material through the probe onto the slide. Once the content was expelled on the slide, a second glass slide was placed lightly on and parallel to the sample slide and pulled lengthways to thinly spread the sample. After air-drying, both slides were stained with hematoxylin-eosin (HE) and further processed for cytohistology.

For the biopsies prepared at Rigshospitalet, the aspirate was withdrawn and flushed via the aspiration tube into an empty syringe. Portions of approximately 0.1 mL aspirate were prepared for cytospin by centrifugation at 1,500 rpm for 5 min (Shandon Cytospin 4; Thermo Electron Corporation, Cheshire, UK) onto glass slides. The slides were subsequently air-dried and stained with HE.

The HE-stained samples were digitized using an automated whole-slide scanner (Scan Scope XT; Aperio Technologies, Vista, CA, USA) with a ×40 objective magnification and a scanning resolution of 0.247 μm/pixel (Fig. 1a, b). Automated whole-slide imaging analyses with respect to the number of hepatocytes, the total area of cell clumps and tissue fragments (mm²), and the total stained area (mm²) of each slide were performed with Aperio algorithms and ImageScope software v.6 (Aperio Technologies, Vista, CA, USA). The automated analyses enable quantification of specific colored tissue types and cellular structures by color thresholding. The Nuclear algorithm detects the nuclear staining of individual cells and classifies them based on their staining intensity. The output is the number of detected cell nuclei in each class (Fig. 1c). The Color Deconvolution algorithm visually highlights stained areas in different colors depending on staining intensity and outputs the respective total area (mm²) of the slide (Fig. 1d, e).

In addition to the automated imaging analyses, the HE-stained samples were examined by conventional light microscopy. For each vitrectomy setting combination, the following cytohistological features were evaluated: (1) the amount of diagnostic cellular material present; (2) the degree of cellular degeneration and trauma; and (3) the retention of cellular architecture [10].

Continuous variables were given as mean ± standard deviation and assessed by nonparametric analyses using the Kruskal-Wallis test for comparison between 3 groups and the Mann-Whitney U test for comparison between 2 groups. In the statistical analyses, each slide was considered as a statistical unit. The data were analyzed using IBM SPSS Statistics v.25 (SPSS Inc., Chicago, IL, USA). Two-tailed p values <0.05 were considered statistically significant.

Based on 48 vitrectomy-assisted biopsies performed at the Haukeland University Hospital, a total of 92 smear slides were available for analyses. The automated imaging analysis of the smear slides showed a statistically significant positive correlation between the number of hepatocytes and the cut rate of the vitrectomy probe. For all specimens obtained by 23- and 25-gauge standard probes (24 and 12 biopsies, respectively), the mean (±standard deviation) number of hepatocytes per slide was 41,322 (±37,681), 132,559 (±126,724), and 130,600 (±115,179) for a cut rate of 60, 600 and 6,000 cpm, respectively (p = 0.001). Similarly, the total area of cell clumps and tissue fragments, as well as the total stained area, was significantly and positively correlated with the cut rate. The mean total area of cell clumps and tissue fragments per slide was 2.0 mm² (±2.6), 6.3 mm² (±8.4), and 4.4 mm² (±4.8) for a cut rate of 60, 600 and 6,000 cpm, respectively (p = 0.013). The mean total stained area per slide was 37.8 mm² (±38.0), 131.4 mm² (±97.9), and 155.9 mm² (±135.2) for a cut rate of 60, 600 and 6,000 cpm, respectively (p < 0.0001) (Fig. 2a–c).

For specimens obtained by 23-gauge TDC probes (12 biopsies), the mean number of hepatocytes per slide was 98,154 (±54,049), 162,344 (±102,500), and 498,799 (±282,638) (p = 0.004); the mean total area of cell clumps and tissue fragments per slide was 7.9 mm² (±3.9), 24.8 mm² (±11.0), and 43.6 mm² (±21.1) (p = 0.001); and the mean total stained area per slide was 120.4 mm² (±97.8), 169.0 mm² (±86.9), and 452.0 mm² (±222.2) (p = 0.013) with a cut rate of 60, 600 and 6,000 cpm, respectively (Fig. 3a–c). The quantity of each of these cytohistological components was significantly higher for specimens obtained by the TDC probes than those obtained by the standard probes, independent of the cut rate.

There were no statistically significant differences between the use of 23- and 25-gauge standard probes regarding the number of hepatocytes, total area of cell clumps and tissue fragments, and total stained area per slide. Neither was there any clear and systematic difference between the quantities of these cytohistological components depending on whether the 23-gauge standard probes were filled with fluid (12 biopsies) or air (12 biopsies) at the beginning of the biopsy procedure.

Based on 24 vitrectomy-assisted biopsies performed at Rigshospitalet, a total of 36 cytospin slides were available for analyses. The automated imaging analysis of the cytospin slides revealed a positive correlation between the cytohistological components and the cut rate of the vitrectomy probe, although this was only statistically significant for the area of cell clumps and tissue fragments. For all specimens obtained by 25-gauge standard and 25-gauge TDC probes (12 and 12 biopsies, respectively), the mean number of hepatocytes per slide was 1,229 (±2,264), 3,287 (±4,401), and 4,193 (±5,371) (p = 0.23); the mean total area of cell clumps and tissue fragments per slide was 0.2 mm² (±0.4), 0.3 mm² (±0.3), and 0.7 mm² (±0.7) (p = 0.02); and the mean total stained area per slide was 3.8 mm² (±4.0), 4.4 mm² (±4.8), and 7.1 mm² (±6.8) (p = 0.27) for a cut rate of 60, 600 and 6,000 cpm, respectively.

Conventional light microscopy revealed that cellular material was present in sufficient quantity and quality to establish the presence of normal hepatic cells and tissue fragments in all the HE-stained samples. An estimate of the amount of cellular material present on each slide corresponded well with the results of the automated whole-slide imaging analysis, showing a positive correlation with the cut rate. No differences were noted between the different vitrectomy settings regarding the degree of cellular degeneration and trauma and retention of cellular architecture.

It is generally assumed that vitrectomy-assisted biopsy provides a larger sample of tissue than fine-needle aspiration biopsy [5-7, 9], but a reduction in cell viability and damage of cellular structures by the vitreous cutter have been considered a problem. Most authors recommend a relatively low cut rate (between 80 and 600 cpm) when performing vitrectomy-assisted biopsies [3, 6, 8, 11]. This is probably due to the belief that a low cut rate reduces the risk of cell damage by decreasing the shear stress on the tissue and also provides larger pieces of tumor tissue. Only a few studies have evaluated the effect of various cut rates on the diagnostic yield and quality of the biopsies. Ratanapojnard et al. [12] examined specimens from bacterial, fungal, and leukocyte suspensions and found that although the leukocyte viability was markedly diminished, the clinical utility of all specimen types was well-preserved with cut rates up to 1,500 cpm. In a rabbit uveitis model, Huang et al. [13] found no difference in loss or damage of lymphocytes sampled at various cut rates up to 600 cpm. Lymphoma cells are very fragile and can easily degenerate during vitreous sampling [14, 15]. By evaluation of cultured B-cell lymphoma cells obtained by a 20-gauge vitrectomy probe and cut rates between zero (aspiration only) and 2,500 cpm, Jiang et al. [16] reported a reduced number of viable cells and an increased damage of cell morphology at cut rates above 600 cpm.

The main finding of the present study was the significant positive correlation between all the cytohistological parameters and the cut rate of the vitrectomy probe. This correlation was even stronger for specimens obtained by the TDC probes, which can be explained by the probe design. The TDC probes facilitate cutting both forward and backward of each stroke, actually doubling the predetermined cut rate. In addition to an increased number of hepatocytes, higher cut rates also led to more cell clumps and tissue fragments. This was a rather unexpected finding in view of the common belief that a low cut rate allows the probe port to be filled with more tissue between each cut, thereby producing larger pieces of tissue. A likely explanation is that due to the elasticity of solid tissue, a low cut rate allows the tissue to retract from the probe port between each cut. We also found a positive correlation between the total stained area of each slide and the cut rate. As all smears were made in the same way by the same person and all the aspirated material was used for the smear preparation, the total stained area may reflect the amount of the aspirate. Because higher cut rates lead to lower aspiration flow, a positive correlation between the aspirate volume and cut rate was unexpected. However, as opposed to ordinary vitrectomies where the material to be aspirated is mostly gelatinous or liquid, a vitrectomy probe placed within solid tissue can only remove the adjacent tissue and limited amounts of the blood and plasma.

Contrary to the standard probes, the port of the TDC probes is always open, with a constant aspiration flow independent of the cut rate so that the tissue is continuously sucked into the probe port during cutting. The combination of a constant aspiration flow and a high actual cut rate may thus explain the significantly higher cytohistological scores obtained by the TDC probes. Also for the cytospin slides, which were prepared by using just a fixed portion and not the entire aspirate, there was a positive correlation between all the cytohistological scores and the cut rate. Although this was statistically significant only for the area of cell clumps and tissue fragments, the results indicate that a high cellular yield is not only due to a large aspirate volume but also to a high cut rate per se.

The success rate of intraocular biopsy is highly dependent on tumor size [3, 17]. For all the vitrectomy probes used in the study, the distance from the distal end of the probe to the cutter port is only 0.22 mm, making it possible to biopsy even small tumors. The strong positive correlation between the diagnostic yield and the cut rate suggests that high-speed cutting is particularly advantageous in small tumors, due to the limited space to orient and maneuver the probe. We found no statistically significant differences in any of the examined cytohistological parameters between samples obtained by 23- and 25-gauge standard probes. Analogous results have been found with conventional fine-needle aspiration biopsy of other organs, where the diagnostic yield has proved to be similar between 22- and 25-gauge needles [18, 19]. There was also no clear difference in outcomes between standard vitrectomy probes filled with fluid or air prior to the biopsy procedure. This may indicate that our fixed aspiration time was sufficient to remove all the air from the system before harvesting the sample.

Evaluation by conventional light microscopy revealed a sufficient quantity and quality of the cellular material to establish the presence of normal hepatic cells and tissue fragments in all the samples, and no major differences were found between those obtained with different vitrectomy probes or cut rates. These results differ from those reported in the previous studies on leukocytes and lymphoma cells, showing a reduction in the number of viable cells and an increase in cell damage with higher cut rates [12, 16]. However, for tissues and solid tumors with less vulnerable cells, we believe that the risk of cell damage is low and outweighed by the benefits of high cut rates to optimize the diagnostic yield of the samples.

Over the last decades, genetic testing has become increasingly important in the management of uveal melanoma. Many studies have shown that small-gauge vitrectomy-assisted biopsy yields sufficient material for various types of cytogenetic analyses and gene expression profiling [5, 7, 20]. A large sample size also reduces the risk of misclassification due to tumor heterogeneity [5]. In the present study, no attempts were made to compare the yield and quality of RNA and DNA from the tissue samples. As we found a high number of intact cells in all the obtained samples and the amount of cellular material is the main limiting factor in molecular and genetic analyses, it seems reasonable to assume that our results also apply to genetic testing.

The present study has several limitations. First and foremost, our in vitro setting does not fully reflect the in vivo clinical situation. Given the different properties and volumes of the tissue to be sampled, the translatability of our results to intraocular tumor biopsies is unsettled. However, compared to an ocular tumor model, our approach makes it easier to control variables and standardize all the vitrectomy procedures. The liver also offers a homogenous tissue with characteristic cell morphology, well-suited for this type of study. Second, in order to reduce the number of variables, the vacuum was set to a fixed value for all the experiments, and other vacuum levels may have provided different results.

In summary, we found that the higher the cut rate, the larger is the amount of aspirated cellular material. Cut rates up to 6,000 cpm for standard probes, equivalent to an effective cut rate of 12,000 cpm for TDC probes, did not adversely affect the cytohistological features or quality of the samples. No differences were found between the use of 23-gauge and 25-gauge vitrectomy probes. Considering the general advantages of small-gauge vitrectomy surgery, our results suggest that the use of a 25-gauge TDC probe with a high cut rate provides the best diagnostic yield in vitrectomy-assisted biopsies of intraocular tumors.

We thank Associate Professor Sabine Leh for sharing resources and advice on automated whole-slide imaging analysis.

The research was conducted ethically in accordance with the World Medical Association Declaration of Helsinki. The study is exempt from Ethical Committee approval because no human subjects or live animals were used in this study.

The authors have no conflicts of interest to declare. Dr. Heegaard is an editorial board member of Ocular Oncology and Pathology.

This study was supported in part by research grants from Fight for Sight, Denmark. DORC Scandinavia AB kindly provided the vitrectomy probes used at the Haukeland University Hospital but had no influence on the study design or implementation.

E.U., J.F.K., and J.K. planned and designed the study. J.K. and E.U. drafted the manuscript. E.U., J.F.K., N.M., D.S., S.H., and J.K. contributed to data acquisition and analysis, revision of the manuscript, and approval of the final version to be published.