Intra-operative imaging of surgical margins of canine soft tissue sarcoma using optical coherence tomography

Abstract

Optical coherence tomography (OCT) is a rapid non-invasive imaging technique that has shown high sensitivity for intra-operative surgical margin assessment in human breast cancer clinical trials. This promising technology has not been evaluated in veterinary medicine. The objective of this study was to correlate normal and abnormal histological features with OCT images for surgical margins from excised canine soft tissue sarcoma (STS) and to establish image evaluation criteria for identifying positive surgical margins. Fourteen client-owned dogs underwent surgical resection of a STS and OCT imaging of 2 to 4 areas of interest on the resected specimen were performed. Following imaging these areas were marked with surgical ink and trimmed for histopathology evaluation. Results showed that different tissue types had distinct characteristic appearances on OCT imaging. Adipose tissue exhibited a relatively low scattering and a honey-comb texture pattern. Skeletal muscle and sarcoma tissue were both dense and highly scattering. While sarcoma tissue was highly scattering, it did not have organized recognizable structure in contrast to muscle which showed clear fibre alignment patterns. In this investigation, we showed different tissue types had different and characteristic scattering and image texture appearances on OCT, which closely correlate with low-power histology images. Given the differentiation between tissue types the results support that OCT could be used to identify positive surgical margins immediately following resection of STS. Further research is needed to assess the diagnostic accuracy of this method for surgical margin assessment.

1 |. INTRODUCTION

In veterinary medicine, the current standard-of-care for surgical margin evaluation is histologic assessment of surgical margins following formalin fixation. The results of this evaluation often take days to obtain. Typically, the histopathologic margins for small- and moderate-sized tumours are assessed by cross-sectioning (or radial method) of tissue trimming. In this method, the specimen is sectioned through the centre of the tumour on the short axis and then each of these sections is cut again (perpendicular to the original cut) along the long axis.¹ Utilizing this method only a small proportion of the surgical margins is assessed (<1%), which can lead to sampling error due to the limited tissue assessment and the possibility of asymmetric tumour growth.^1–3 This (under-) sampling error can lead to lack of detection of cancer cells extending to the surgical margins (constituting incomplete resection), which is especially possible in the case of invasive tumours that exhibit asymmetric growth like soft tissue sarcomas (STS).^1–3 Failure to recognize incomplete margins can result in inappropriate treatment recommendations and missed opportunities for local disease control.

STSs comprise 15% of all skin tumours in dogs.⁴ Surgical removal is commonly the treatment of choice for local control of these tumours; the success of which is often based on histological margin assessment.^5–10 While histopathology is currently the gold standard for surgical margin evaluation in veterinary medicine, reliable intra-operative assessment of surgical specimens is not possible with current methods. Novel imaging techniques that allow intra-operative margin assessment are critically needed to advance the current standard-of-care for the successful surgical treatment of tumours in companion animals. To date, there have been a few reports of the use of fluorescence imaging techniques in canine patients to assist in identification of residual tumour following resection.^11–15

Optical coherence tomography (OCT) is an emerging diagnostic oncologic imaging modality that uses light waves to generate real-time, high-resolution images of tissue at a microscopic level. It has shown promise for the real-time evaluation of human breast cancer margins studies reporting 91.7% to 100.0% sensitivity (and 82.0%−92.0% specificity) for identifying residual tumour cells along the entire surgical margin.^16,17 Preliminary clinical studies in humans and data in rodents and other tumour models strongly support that this imaging modality will be successful in canine patients.^16–19 The ability to intra-operatively visualize residual microscopic disease in real-time could optimize surgical interventions to allow removal of residual tumour in an attempt to decrease the necessity of second surgeries and potentially other adjunctive therapies. To the authors’ knowledge, this is the first description of OCT being used for imaging features of normal and abnormal canine tissues at surgical margins, and is a crucial step in the assessment and validation of OCT for the real-time imaging of surgical margins of STS in dogs.

The study aim was to correlate normal and abnormal histological features of surgical margins from excised canine STS with OCT images. Furthermore, this study sought to establish image evaluation criteria for identifying positive surgical margins. The hypothesis was that OCT images would have a high correlation with histological features of surgical margins of excised canine STS.

2 |. MATERIALS AND METHODS

Fourteen client-owned dogs underwent resection of one or two STS arising from the skin or subcutaneous tissues as part of clinical study protocols approved by the Institutional Animal Care and Use Committee at the University of Illinois at Urbana-Champaign. Prior to enrolment in the study informed owner consent was obtained, the dogs needed a cytologic or histopathologic diagnosis of a STS or mesenchymal tumour. The tumour may have arisen in the skin or subcutaneous tissues for the first time, or be a recurrent tumour. The tumours were excised by an American College of Veterinary Surgeons boarded small animal surgeon either primarily or supervising a resident in training. The planned surgical margins were categorized as “marginal” if the tumour was excised narrowly with excision of gross disease, “wide” if lateral margins of 3 cm from palpable tumour and one fascial plane deep was used, and “radical” if a body part was removed to excise the tumour within a body compartment.

Immediately following surgical excision, the resected specimen was imaged in an adjacent laboratory. The resected specimen was placed on a Petri dish or flattened board to allow positioning on the mounted micrometre stage of a commercial spectral-domain OCT system (Envisu C2300; Bioptigen Inc., Durham, North Carolina). Areas of interest were identified to be: a possible area where STS might have been present at the surgical margin (based on palpation or visual assessment) or areas containing different tissue types. Generally, 2 to 4 areas of interest were imaged on each specimen depending on the size and availability of different tissue types. The sample beam arm was placed directly over the area and 3-D volumetric scans were acquired at each of the 2 to 4 areas of interest per resected tumour sample. The OCT system had a central wavelength of 1310 nm and an incident illumination power of approximately 5 mW. The image resolution was approximately 8 μm axially and approximately 10 μm laterally, with an imaging depth of approximately 1 to 2 mm depending of the optical properties of the tissues. In the sample arm of the OCT system, lateral beam scanning was performed with mirrors mounted on computer-controlled galvanometers, passing the OCT beam through a fixed focus objective lens to volumetrically scan the area of interest (5 × 5 × 2 mm³) at 7.4 frames/second (6 kHz A-scan rate) via raster collection of B-scans along the x-y plane. The imaged area corresponded to 600 sequentially-acquired and adjacent OCT images, taking about 1 minute to collect the volumetric image data at each site. OCT images were evaluated in real-time to ensure the images acquired were of diagnostic quality.

Following imaging of each area of interest, the boundary of the imaged area was marked in a U shape with surgical ink (The Davidson Marking System; Bradley Products, Bloomington, Minnesota; Figure 1) to identify the area for later histologic sectioning, and to correlate the OCT and histologic images. The surgical ink was applied with a toothpick and allowed to dry for 5 minutes. Specimens were then placed in 10% neutral-buffered formalin. The inked areas were paraffin embedded, sectioned and stained with haematoxylin and eosin (H&E). Prepared slides were digitized with a slide scanner (NanoZoomer 2.0 RS, Hamamatsu C10730, Bridgewater, New Jersey). The images were orientated based on inked areas for correlation with OCT images. The standard tumour histopathology and assessment of the inked sections was performed by an American College of Veterinary Pathology board certified pathologist (E.D.) or pathology resident (J.S.) blinded to the results of the OCT imaging, thus providing an unbiased assessment of the histology of sections for correlation to OCT images.

FIGURE 1.

Photograph illustrating resected soft tissue sarcoma prepared for the optical coherence tomography (OCT) imaging with the U-shaped marking of on area of interest (black and yellow coloured)

The OCT images were acquired with proprietary software (InVivoVue 1.7; Bioptigen Inc., Durham, North Carolina). Processing of the OCT images involved exporting the raw data in another type of propriety software (MatLab; Mathworks, Natick, Massachusetts) to generate 600 OCT frames. Each B-scan had physical dimensions of 5 mm (lateral) × 2 mm (depth), represented by 600 × 257 pixels, respectively. As described in Mesa et al, a mask was generated for every OCT frame using an active contour segmentation method.²⁰ Masks were created by choosing areas of interest, avoiding areas of saturation and background noise, thus rendering a cleaner version of each OCT frame. Correlation of OCT images with histopathology sections of the same areas was performed by an observer experienced in the interpretation of OCT images of surgical margins (K.J.M.) with known histopathologic interpretation of tissue type and results of surgical margin interpretation. Tissues were evaluated for two characteristics, which were assessment of light scattering characteristics and evidence of a tissue-organized microstructure or texture of tissues. The combination of characteristics that were identified and indicative of incomplete surgical margins or different tissue types were defined as image evaluation criteria.

3 |. RESULTS

Fourteen dogs with 15 STS enrolled in this clinical study: one dog had two STS concurrently, which were both surgically excised. The demographics of the study patient population and the location of the tumours are summarized in Table 1. The tumours were resected with different planned surgical margins depending on the area of the body and the size of margins chosen by the primary surgeon. The margins planned at surgery and the number of areas of interest imaged are also summarized in Table 1.

TABLE 1.

Demographics, description of tumour characteristics, and surgical approach

Dog number	Breed	Age at surgery (years)	Tumour location	Longest tumour diameter (cm)	Surgical resection (marginal/wide/radical)	Number of imaged areas
1	Vizsla	9.1	Right elbow	1.0	Marginal	2
2	Boxer	8.1	Ribs	4.5	Marginal	2
3	Rat terrier	10.1	Right hock	3.5	Marginal	2
4	Mixed breed	13.6	Left hip	6.2	Wide	2
5	Siberian husky	11.1	Left hip	5.0	Wide	2
6	Golden retriever	12.8	Mass 1: Right brachium	1.0	Mass 1: Wide	4
			Mass 2: Left triceps	5.0	Mass 2: Wide	4
7	Labrador retriever	13.7	Left lateral thorax	24.0	Marginal	4
8	Labrador retriever	4.8	Right carpus	5.0	Marginal	4
9	Bulldog	9.3	Interscapular	7.0	Wide	4
10	Mixed breed	6.3	Left metacarpus	2.0	Radical	4
11	Labrador retriever	6.8	Right elbow	2.0	Marginal	4
12	Mixed breed	10.8	Left flank	12.0	Marginal	4
13	Labrador retriever	8.5	Left metacarpus	2.0	Marginal	4
14	Mixed breed	16.1	Right metacarpus	7.0	Marginal	4

The imaged areas of interest consisted of different tissue types on histologic evaluation. Four tissue types were predominantly present in the evaluated samples: skin, adipose, muscle and sarcoma. These results are detailed in Table 2. The tissue characteristics based on OCT imaging were determined for the different tissue types represented in the imaged areas. Depending on the tissue type and the optical properties of the tissue, the OCT imaging depth within the tissue varied. Low backscattering adipose tissue was imaged up to 2 mm in depth, however, denser highly-scattering cellular tissues, like sarcoma, was imaged to shallower depths.

TABLE 2.

Description of tissue types present in histologic sections

Dog #	Surgical resection (marginal/wide/radical)	Histopathologic diagnosis	Area 1	Area 2	Area 3	Area 4
1	Marginal	Grade I STS, margins less than 0.1 cm	75% STS 1 mm from margins, other tissue types skin and subcutis	75% STS <1 mm from margins, remaining skin and subcutis	–	–
2	Marginal	Grade I STS, complete, deep margin 1 mm, lateral margins 1.5 to 3 cm	90% STS, <1 mm from deep margin Other tissues adipose and skeletal muscle	Adipose tissue and skeletal muscle	–	–
3	Marginal	Grade I STS, complete margins	70% to 75% STS, tumour extends to deep margin Other tissue skin	75% to 90% STS, <1 mm from deep margins Remaining tissue is skin and fibrous connective tissue	–	–
4	Wide	Grade II STS, complete, lateral margins >0.7 cm and deep margin 1 cm	Adipose tissue with small amount of fibrous connective tissue	<15% STS, 6 mm from deep margins, other tissue skin and subcutis	–	–
5	Wide	Grade II STS, complete excision, >0.3 cm lateral and 0.1 cm deep	85% STS, tumour is 1 mm from deep margin Other tissues subcutis	90% STS, <1 mm from deep margin Other tissues subcutis	–	–
6	Mass 1: Wide Mass 2: Wide	Mass 1: Grade II STS, complete excision, deep margin 0.1 cm and lateral >1.2 cm Mass 2: Grade III STS, complete excision, >0.5 cm lateral and 0.1 cm deep	Mass 1: 5% to 20% STS, 5 to 7.5 mm from margin Other tissues skin and subcutis Mass 2: 5% to 60% STS, 7 mm from margin Other tissues skeletal muscle and subcutis	Mass 1: Skeletal muscle and adipose tissue, other tissues haired skin and subcutis Mass 2: 35% to 70% STS, extends to deep margin	Mass 1: Skeletal muscle and adipose tissue, other tissues haired skin and subcutis Mass 2: 60% to 75% STS, extends to deep margin Other tissues skin, subcutis and skeletal muscle	Mass 1: Skeletal muscle and adipose tissue, other tissues haired skin and subcutis Mass 2: 80% to 85% STS, extends to deep margin Other tissues subcutis and skeletal muscle
7	Marginal	Grade II STS, incomplete excision, neoplastic cells extend to all margins	>90% STS, extends to margin Other tissue subcutis	>95% STS, extends to margin Other tissue subcutis	>95% STS, extends to margin Other tissue subcutis	>95% STS, extends to margin Other tissue subcutis
8	Marginal	Grade I STS, incomplete excision, neoplastic cells extend to multiple margins	>90% STS, <1 mm from margin Other tissues fibrous connective tissue and adipose tissue	85% STS, extends to margin Other tissue haired skin	>95% STS, extends to margin Other tissue fibrous connective tissue	> 85% STS, extends to margin Other tissues haired skin
9	Wide	Grade II STS, complete excision, >1 cm lateral margins and 1 cm deep margins	Skeletal muscle, fibrous connective tissue and adipose tissue	Adipose tissue and skeletal muscle	Subcutis and skin	Adipose and skin
10	Radical	Grade II STS, complete excision, >30 cm	Skin, subcutis and skeletal muscle	Skin and subcutis	Skeletal muscle and adipose tissue	Skeletal muscle and adipose tissue
11	Marginal	Grade I STS, complete excision, >0.1 mm lateral margins and <0.1 mm deep margins	75% STS, <1 mm from margin Other tissue skin and subcutis	35% STS, <1 mm from margin Other tissues skin and subcutis	25% STS, <1 mm from margin Other tissues skin and subcutis	75% STS, <1 mm from margins Other tissues skin and subcutis
12	Marginal	Grade II STS, complete excision, lateral margins >1.2 cm and deep margin 0.4 cm	Skin and subcutis	80% STS, 0.3 cm from margins Other tissues skin and subcutis	80% STS, 0.5 cm from margin Other tissues skin and subcutis	Skin and subcutis
13	Marginal	Grade I STS, complete excision, <0.1 cm from margins	80% to 85% STS, <1 mm from margins Other tissues skin	80% to 85% STS, <1 mm from margins Other tissues skin	80% to 85% STS, <1 mm from margins Other tissues skin	80% to 85% STS, <1 mm from margins Other tissues skin
14	Marginal	Grade III STS, incomplete excision, extends to 3 out of 4 submitted margins	>98% STS, <1 mm from margins Other tissue connective tissue	95% STS, <1 mm from margins Other tissue connective tissue	90% STS, <1 mm from margins Other tissue connective tissue	95% STS, extends to margin Other tissue connective tissue

Abbreviation: STS, soft tissue sarcoma.

The microscopic histologic and OCT appearance of adipose, skin, muscle and sarcoma tissues are shown in Figure 2. Adipose tissue exhibited relatively lower scattering as well as a hole-filled (honey-comb) appearance and texture pattern. Skin appeared to have intermediate scattering often irregular in contour, and to contain adnexal structures which were seen clearly on OCT images. On the other hand, muscle and sarcoma tissue were both dense and highly scattering; thus, more challenging to differentiate with OCT. However, sarcoma tissue did not have a defined structure, while alignment patterns typically present in muscle fibres could be observed. Three examples of where sarcoma was present at surgical margins (incomplete margins) are shown in Figure 3. Incomplete surgical margins were characterized by dense, highly scattering tissues with no evident alignment patterns at the edge of the specimen. Of the three areas that are seen, two are from the same dog and illustrate differences in the outer-most part of the sarcoma. In dog 8 area 1, there are some air bubbles underneath the initial layers of tissue/pseudocapsule giving the appearance of black linear structures. Looking at another area in this dog, area 4, there is the same distinct pseudocapsule but more strongly scattering and deeper tissue imaging is possible with the lack of air.

FIGURE 2.

Correlation of histology properties of A, adipose; B, skeletal muscle; C, skin and D, soft tissue sarcoma with corresponding optical coherence tomography (OCT) images (E-H, respectively). The left-sided row (A-D) shows the haemotoxylin and eosin stained digitized histological images of the resected tissue and the right-sided row (E-H) shows the corresponding OCT images. The black and white scale bars across the bottom of images D and H represent 500 μm to show the scale of all histologic and OCT images shown

FIGURE 3.

A-C, Examples of surgical specimens and corresponding optical coherence tomography (OCT) and histology images with sarcoma at surgical margins. On the left side, is a photograph of gross surgical specimen with incomplete boxes marked with surgical inks highlighting imaged areas. On the right side, the bottom row shows haemotoxylin and eosin stained digitized histological images, and the corresponding OCT images are shown in the top row. The white or black scale bars at the bottom of each OCT or histologic image represent 500 μm. A, Images for case 2, area 1 sample: The OCT and histological sections were taken from the area bounded by the U-shaped yellow surgical ink. The OCT images show layer of skeletal muscle with deeper high scattering area consistent with underlying sarcoma (white arrows). B, Images for case 8, area 1: The OCT and histological sections were taken from the area bounded by the U-shaped orange surgical ink. The OCT image shows a thin layer of tissue with low scattering areas consistent with the pseudocapsule with air bubbles (black linear structures, yellow arrow) overlying a high scattering tissue with no alignment patterns underlying consistent with sarcoma (white arrowheads). The imaging of the deeper tissues is attenuated by the overlying air bubbles. This result is consistent with sarcoma <1 mm from the deep surgical margin. C, Images for case 8, area 4: The OCT and histological sections were taken from the area bounded by the U-shaped yellow surgical ink. The OCT image shows a distinct and condense pseudocapsule immediately under the probe (red arrow) but with high scattering tissue consistent with sarcoma underlying. This result is consistent with sarcoma <1 mm from the deep tissue margins. Deeper tissue imaging of the tumour is seen in comparison to that seen in Figure 3B due to the lack of air attenuating light penetration

4 |. DISCUSSION

This study represents the first descriptive study of the use of OCT for imaging of surgical margins in dogs. We showed it was possible to image and characterize different tissue types that occur at surgical margins of canine STS including skin, adipose, muscle and sarcoma based on their light scattering properties and their texture appearance on OCT. The greatest contrast between tissues was between fat and skin or muscle or STS. The proposed reason why adipose tissue is more distinct when comparing to the other tissue types is likely because of large size and regular arrangement pattern of mature adipose cells that create the honey-comb appearance. Although muscle and sarcoma were both high scattering, there were other visual characteristics that allowed differentiation of these two tissues and recognition of STS at surgical margins, representing incomplete excision of the tumour. We found that if muscle tissue is imaged in a plane perpendicular to the direction of the muscle fibres, then the bundled fibre structure is most clearly observed. Given the similarity of appearance of these tissues, there has been limited investigation into methods to objectively differentiate between these tissues. Mesa et al found that a novel image texture based algorithm could be used to objectively differentiate between OCT images of sarcoma and muscle.²⁰ Wang et al reported on the use of 3-D computational methods of assessment of STS from normal tissue types.²¹ Polarization-sensitive OCT, a variant of OCT that is sensitive to the birefringent optical properties of tissues, such as highly ordered collagen or muscle fibres, has been shown to differentiate between human breast tumours and normal fibrous breast stroma.²² Polarization-sensitive OCT similarly may be helpful to differentiate between muscle tissue with highly ordered fibres and STS that lacks such aligned and organized microstructure. In the future, these and other techniques may be further developed to automate objective differentiation of the tissue types to aid in observer interpretation of the OCT images, even during surgery, and in real time.

Characteristics seen between different tissue types varied but there was also variation in appearance of sarcoma between different dogs and different areas in the same dog. Sarcoma tissue was always highly light scattering on evaluation, but the texture of the tumour in the OCT images was often different between tumours. This represents differences in sarcoma tissue microstructure, particularly the pseudocapsule which is very superficial, which varies between tumour types. There was a predominance of low-grade tumours in this study, which could have led to the structure having a prominent pseudocapsule and being less irregular as the sarcomas were less invasive and heterogeneous.

In this study, different characteristics of tissues and tumour at surgical margins following STS removal in dogs have been identified. Differentiation between normal and cancerous tissues is key for utilization of intra-operative imaging for assessment of surgical margins of the excised sample or more ideally the in vivo tumour bed to assess whether there is residual microscopic STS at the surgical site. There is some evidence that clean resection in canine STS may be associated with prolonged tumour-free intervals and decreased risk of tumour recurrence.^5,7,9,23 It remains underdetermined but possible that removal of residual microscopic disease may be associated with decreased recurrence similar to scar re-excision when primary resection is incomplete.²⁴ Alternatively, that if the areas of residual microscopic disease could be identified that these cases could have targeted treatment of this residual microscopic disease with radiation therapy or topical chemotherapy or similar.

The population of dogs in this study consisted of middle-aged to older, pure and mixed bred dogs, which reflects the population of dogs developing STS as reported in other studies.^24–26 Many dogs had tumours arising on limbs, which may reflect our referral population as referring veterinarians may be removing truncal STSs and referring distal limb STSs to specialists in oncology and surgery. Most tumours were grade I or II, which is typical of most STSs reported in the veterinary literature.^24–26 Given the high proportion of limb tumours and the high number of cases where planned marginal excision was performed, it was not surprising there were several cases where the histologic margins were incomplete.

Good correlations were found between OCT and histological images of different normal and abnormal tissue types at the surgical margins of excised canine STSs. This early promise would suggest continued research is warranted to assess this technique for diagnostic accuracy of detection of incomplete margins during STS resection. The future goals of our group are to develop an accurate tool for assessment of surgical margins and detection of incomplete margins that could improve the efficacy of cancer surgery and reduce patient morbidity and mortality.