In vivo reflectance confocal microscopy of shave biopsy wounds: feasibility of intraoperative mapping of cancer margins.

AlCl 3 staining was optimized, in terms of concentration versus immersion time, on excised tissue ex-vivo . RCM imaging protocol was tested in patients undergoing shave biopsies. The RCM images were retrospectively analyzed and compared to the corresponding histopathology.

Materials and methods

Pre-clinical study of AlCl 3 as contrast stain Freshly excised specimens were obtained from MMS performed at Memorial Sloan-Kettering Cancer Center (MSKCC). During MMS, excised specimens are frozen and Haematoxylin-and-Eosin (H&E)–stained sections are prepared and examined. Remaining tissue, which is routinely discarded, was collected for this study, under Institution Review Board approval. Each specimen was thawed, rinsed in normal saline and imaged ex-vivo using a previously described bench-top RCM (VivaScope 2000, Lucid Inc., Rochester, NY). 17 , 19 Subsequently, specimens were rinsed with saline, immersed in AlCl 3 and re-imaged. Concentrations of AlCl 3 and immersion times were varied to determine optimal imaging conditions. Concentrations tested were 20% in anhydrous ethyl alcohol, 35% in purified water, 50% in purified water and 50% in equal volumes of isopropanol and purified water. Immersion times were 5, 10, 30 and 45 seconds and 1, 2, 3 and 5 minutes. Minimum immersion time was defined as time required for all nuclei in RCM mosaic to appear consistently brightened. Each condition was retested on minimum of 5 specimens.

RCM imaging study on shave biopsy wounds Patients

Participants were recruited from patients undergoing shave biopsy for diagnosis of suspicious skin lesions at MSKCC. All patients were 18 years or older. Written consent was obtained prior to enrolment. The research protocol was approved by MSKCC Institutional Review Board.

Instrumentation

For imaging patients, a commercially-available, previously described RCM (Vivascope 1500, Lucid Inc., Rochester, NY) was used. 4 , 20 Briefly, RCM uses near-infrared laser at 830nm. A 30X objective lens allows imaging with optical sectioning of 3μm and lateral resolution of 1μm. Contact between the objective lens and skin is achieved with a tissue ring. The RCM acquires images of en face optical sections with 500 ×500μm 2 field-of-view (equivalent to 30× magnification). An automated stepper was used to acquire up to 12×12 contiguous images into a “mosaic” which displays a 6×6mm 2 field-of-view (equivalent to 3× magnification). RCM images can be acquired to depth of approximately 300μm.

RCM imaging protocol

A pilot was conducted during the first 8 cases to qualitatively assess the best imaging conditions and construct a protocol. For immersion medium in the wound cavity, sterile Surgilube gel (Fougera, Melville, NY) was used. The gel’s viscosity was found to be advantageous compared to sterile saline; the gel was better retained in the wound when imaging patients in recumbent position. Sterile conditions were ensured by draping the wound with transparent dressing (Tegaderm, 3M, St. Paul, Minnesota, USA). Imaging was tested through a 1 mm-thick disposable optical window made of polycarbonate disk (General Electric Company, Fairfield, CT); image quality with the polycarbonate disk was qualitatively better than that obtained with a glass window. Following the results of our pre-clinical study, we used AlCl 3 solution as contrast agent in RCM images. The final protocol was as follows: The wound was swabbed with AlCl 3 using sterile applicators. The cavity was filled with sterile gel ( Fig. 1 ). The wound was sealed with sterile transparent adhesive dressing. A drop of Crodamol STS oil (Croda Inc., Edison, NJ) was applied over the dressing. The tissue ring with polycarbonate window was attached to the surface of the dressing. The ring covered part of the wound cavity and part of the wound edge. Ultrasound gel was used as immersion medium between the tissue ring and the objective lens. RCM images and mosaics were acquired at a minimum of three levels ( Fig. 1 ): level of intact epidermis surrounding the wound (“epidermal margin”), level of superficial dermis in the wound (“peripheral dermal margin”) and base of the wound (“deep dermal margin”). Using the final protocol, 39 additional lesions undergoing shave biopsy were included. Histopathological diagnoses for these lesions were BCC (n=10), squamous cell carcinoma (SCC, n=10), actinic keratosis (n=1), irritated seborrheic keratosis/lichen planus-like keratosis (n=11), melanoma (n=3), naevus (n=1) irritated verrucae (n=2) and neurofibroma (n=1).

Assessment of RCM images and histopathologic correlation

Images were jointly assessed by 2 dermatologists (AS and KN), one of whom (KN) is a MMS surgeon. Image quality was assessed as acceptable or poor. A mosaic image was considered acceptable if at least 75% of images that show wound margins displayed adequate resolution and contrast as previously defined. 21 For every RCM mosaic, the observers assessed the presence of the following structures – surrounding epidermis; surrounding dermis; bright keratinocytes, bright adnexal epithelium, collagen bundles and inflammatory cells within the wound; and tumour aggregates. All biopsy specimens were routinely processed with formalin fixation and paraffin embedding, followed by vertical sectioning and H&E staining. Diagnoses were retrieved from the hospital information system. Slides were also examined (by A.S.) for findings which appeared to best correlate with RCM structures under analysis.

RCM imaging protocol

A pilot was conducted during the first 8 cases to qualitatively assess the best imaging conditions and construct a protocol. For immersion medium in the wound cavity, sterile Surgilube gel (Fougera, Melville, NY) was used. The gel’s viscosity was found to be advantageous compared to sterile saline; the gel was better retained in the wound when imaging patients in recumbent position. Sterile conditions were ensured by draping the wound with transparent dressing (Tegaderm, 3M, St. Paul, Minnesota, USA). Imaging was tested through a 1 mm-thick disposable optical window made of polycarbonate disk (General Electric Company, Fairfield, CT); image quality with the polycarbonate disk was qualitatively better than that obtained with a glass window. Following the results of our pre-clinical study, we used AlCl 3 solution as contrast agent in RCM images. The final protocol was as follows: The wound was swabbed with AlCl 3 using sterile applicators. The cavity was filled with sterile gel ( Fig. 1 ). The wound was sealed with sterile transparent adhesive dressing. A drop of Crodamol STS oil (Croda Inc., Edison, NJ) was applied over the dressing. The tissue ring with polycarbonate window was attached to the surface of the dressing. The ring covered part of the wound cavity and part of the wound edge. Ultrasound gel was used as immersion medium between the tissue ring and the objective lens. RCM images and mosaics were acquired at a minimum of three levels ( Fig. 1 ): level of intact epidermis surrounding the wound (“epidermal margin”), level of superficial dermis in the wound (“peripheral dermal margin”) and base of the wound (“deep dermal margin”). Using the final protocol, 39 additional lesions undergoing shave biopsy were included. Histopathological diagnoses for these lesions were BCC (n=10), squamous cell carcinoma (SCC, n=10), actinic keratosis (n=1), irritated seborrheic keratosis/lichen planus-like keratosis (n=11), melanoma (n=3), naevus (n=1) irritated verrucae (n=2) and neurofibroma (n=1).

Pre-clinical study of AlCl 3 as contrast stain RCM imaging of normal epidermis from excised tissue is shown ( Fig. 2 ). Under unstained conditions, nuclei of keratinocytes appear dark and cytoplasm and intercellular borders between keratinocytes appear as bright polygonal outlines resulting in a honeycomb pattern at spinous and granular layers of the epidermis ( Fig. 2A and Table 1 ). After immersion in AlCl 3 , nuclei of keratinocytes appear bright ( Figs 2B-D ) with enhanced nuclear-to-cytoplasm contrast, resulting in a cobblestone pattern ( Table 1 ). The minimum immersion time for various AlCl 3 concentrations is also shown ( Table 2 ). The results of the pre-clinical study were empirically translated to the protocol for imaging shave biopsy wounds in patients. The routinely used concentration of AlCl 3 in our clinic is 35% for which the minimum immersion time was determined to be 1 minute ( Fig. 2C ). This was translated to swabbing the wound with sterile applicator 4 times with AlCl 3 (each swab was used for about 15 seconds), resulting in immersion time of 1 minute.

RCM imaging study on shave biopsy wounds Correlation of confocal and histopathologic features of normal skin Using the final protocol, 39 lesions undergoing shave biopsy were imaged with RCM and analyzed. Bright nuclei of keratinocytes within the wound cavity were seen in 21 lesions (54%). These bright nuclei were uniformly spaced, forming a regular cobblestone pattern ( Table 1 , Fig. 3A ). Since the epidermis overlying the surgical wound lacks a stratum corneum, the keratinocytes within the wound cavity were exposed to AlCl 3 application. In contrast, the epidermis surrounding skin showed a honeycomb pattern ( Table 1 , Fig. 3B ), similar to appearance of unstained epidermis in the preclinical study and similar to in-vivo imaging of epidermis of normal skin. 21 The regular cobblestone and honeycomb patterns correlate with normal pattern of keratinocytes at the spinous and granular layers ( Fig. 3C ). The transition between honeycomb and cobblestone patterns was often notable and demarcated the margins of the wound cavity at the level of the epidermis ( Fig. 3D ). Bright adnexal structures were seen in only 2 lesions (5%). These were round structures in the dermis composed of uniformly spaced bright nuclei of adnexal epithelium. Bright linear or curved structures were seen in the dermis in 25 (64%) lesions. The bright structures were arranged in a retiform arrangement or in parallel (bundles). These structures correlated with dermal collagen. In some cases, these linear structures were apparent in the background of amorphous brightness ( Fig. 3E ), which correlated with solar elastosis ( Fig. 3F ). Bright stellate cells and small bright dots in the dermis were seen in 25 (64%) lesions ( Fig. 3G ), correlating with inflammatory cells (histiocytes and lymphocytes, Fig. 3H ).

Evaluation of shave biopsy wound margins with RCM

Evaluating the RCM mosaics of the surgical wounds, we identified 3 margin levels: “Epidermal margin” - mosaics acquired at the level of surrounding epidermis ( Fig. 4A ). The central portion of the wound cavity appeared dark. Peripheral to it, bright nuclei forming a cobblestone pattern were seen; these were compatible with nuclei of epidermal keratinocytes in the outer, most superficial perimeter of the wound cavity. Peripheral to wound, the epidermis in the surrounding skin showed a honeycomb pattern ( Fig. 3D ). “Peripheral dermal margin” – mosaics acquired at the level of the superficial dermis ( Fig. 4B ). The centre of the wound still appeared dark. Around it, within the wound cavity area, a bright area of tissue was seen. Depending on the level of the mosaic, this bright tissue area displayed either a cobblestone pattern of bright nuclei of keratinocytes (in the more superficial mosaics); or refractile structures in retiform or parallel arrangement (in the deeper mosaics) compatible with dermal collagen. Peripheral to the bright area, the surrounding dermis appeared less refractile, showing either dermal papillae of the dermal-epidermal junction or dermal collagen of the papillary dermis. Collagen in the surrounding dermis appeared more blurred and less refractile than that within the exposed wound area. In addition, in some cases, with progressive imaging depth from the level of “peripheral dermal margins” to that of “deep dermal margins”, mosaics showed only a bright strip of wound margin at the focal plane of imaging. Central to this bright strip was the dark wound cavity, and peripheral to the bright strip, the surrounding skin appeared minimally to non-refractile due to the imaging depth. “Deep dermal margins” – mosaics acquired at the level of the base of the wound ( Fig. 4C ). The wound cavity that appeared dark at the level of “peripheral dermal margin” now appeared bright, showing mostly highly refractile collagen bundles. The surrounding skin and superficial edges of the wound appeared dark. For mosaics with acceptable imaging quality, epidermal margin was visible in 23 of 39 lesions (59%), peripheral dermal margin was visible in 23 lesions (59%), and deep dermal margin was visible in 23 lesions (59%). In 13 lesions (33%), all 3 margins were visible. Reasons for unacceptable quality of mosaic images included air bubbles obscuring images, over-saturation of image brightness compromising resolution, and inaccurate software stitching of images in the mosaic. Correlation of confocal and histopathologic features of skin neoplasms In 4 lesions (10%), bright tumour islands were seen at deep dermal margins (n=1) and peripheral dermal margins (n=3). When tumour islands were observed in exposed wound margins, they were composed of bright closely aggregated nuclei ( Figs 5A&C , Fig. 6D ). When tumour islands were located deeper in tissue, under the exposed surface or in the surrounding dermis, they appeared less refractile and individual nuclei were not discernible ( Fig. 6E ). Tumour islands correlated on histopathology with aggregates of basaloid cells showing peripheral palisading of nuclei ( Figs 5B&D , Fig. 6F ). In the surrounding dermis, a stroma composed of bright collagen bundles and bright stellate cells and small bright dots was observed ( Fig. 6D ); these bright cells correlated on histopathology with histiocytes and lymphocytes, respectively ( Fig. 6F ). In these 4 cases, histopathological diagnosis proved to be BCC, nodular in 2 lesions and superficial in 2 lesions. In 3 lesions (8%), atypical honeycomb was observed in epidermal margins ( Fig. 7G ); histopathological diagnosis proved to be SCC. In 1 SCC lesion, an atypical cobblestone pattern was seen in peripheral and deep dermal margins ( Figs 7E&F ); this atypical cobblestone pattern displayed crowding of nuclei, variability in size and brightness of nuclei and the presence of abnormally large nuclei ( Fig. 7H ). On histopathology, crowding and pleomorphism of nuclei, hyperchromatic nuclei and disordered maturation of the epidermis were seen; the proliferation of atypical keratinocytes extended to the base and peripheral margins of the biopsy ( Fig. 7I ). Of note, in another case, while an atypical honeycomb was observed in epidermal margins, the histopathological diagnosis proved to be lichen planus-like keratosis; in this case, mild atypia of keratinocytes, interpreted to be reactive to the lichenoid inflammation, was observed on histopathology.