Materials and methods

The relevant institutional committee for ethics in animal experimentation approved all experiments.

Cell culture

Human melanoma cell lines NMD, HY, and MW (obtained from the Ludwig Institute, Heidelberg, Victoria, Australia) and normal cultured skin cells (obtained from cadavers) were grown in culture using Dulbecco's modified Eagle's medium (DMEM) supplemented with heat-inactivated 10% fetal bovine serum (FBS). Streptomycin, penicillin, and Fungizone (amphotericin B) were added to the culture medium. Confluent monolayers of cells were prepared for implantation by detaching the cells from the flask using a solution of Hanks' balanced salt solution (HBSS) and trypsin ethylenediamine tetraacetic acid. The cell solution was then centrifuged for 10 min at 450g. The resulting cell pellet was resuspended in 2 ml of DMEM for implantation.

Imaging of melanoma cells in vitro

The binding of the antimelanoma antibodies with human melanoma cells was examined using live cells grown in culture on glass coverslips in six-well culture plates (DMEM +10% FBS). Antimelanoma antibodies were diluted in phosphate-buffered saline (PBS) supplemented with 10% FBS for all in vitro experiments. Culture medium was removed from each well, and then 100 l of the appropriate diluted antibody solution was added to each coverslip for 20 min. The coverslips were then washed gently with HBSS for 20 min, after which the coverslip was inverted over a well of HBSS to facilitate FOCI.

Antimelanoma antibodies

CD 51/61-FITC (mouse monoclonal IgG₁, ) recognizes a human _v3 integrin complex (vitronectin receptor), which is expressed in high levels on melanoma cells. The 14.G2a-FITC (mouse monoclonal IgG_2a) antibody selectively binds to the GD2 ganglioside antigen on the surface of melanoma cells, which is upregulated in melanoma tumor formation (^{Pichla et al, 1997;Mukerjee et al, 1998}). The 9.2.27-FITC antibody (mouse monoclonal IgG_2a) recognizes a chondroitin sulfate group expressed on the cell surface of melanoma cells (^{Bumol and Reisfeld, 1982}). According to information provided by the suppliers, the three antibodies used do not cross react with human naevi. Their ability to distinguish human naevi from melanoma cells, however, will have to be confirmed by extensive testing in future studies. All antibodies were obtained from Pharmingen, Beckton Dickinson, Australia, conjugated to fluorescein isothiocyanate (FITC), except 9.2.27. Conjugating the 9.2.27 antibody to FITC was performed by dissolving 1 mg of FITC (Sigma, Australia) in 1 ml of a Na₂CO₃/NaHCO₃ buffer (1.325 g of Na₂CO₃, 1.050 g of NaHCO₃ in 50 ml water; titrated to pH 9.5). The FITC buffer solution was then added to the antibody solution (50 l of solution per 0.2 ml of antibody) and incubated at 25°C for 30 min. Excess unbound FITC was removed by dialysis of the antibody/FITC buffer solution with 100 ml PBS for 4 h using dialysis tubing. The PBS solution was changed and dialysis was continued overnight. The FITC-conjugated antibody was removed from the dialysis tubing with a pipette and 0.01% (vol/vol) methiolate was added as a preserving agent.

Human melanoma cell implantation

Eight-week-old (20 g) athymic BALB/C mice (Walter and Eliza Hall Institute for Medical Research, Kew, Victoria, Australia) were anesthetized with Nembutal (pentobarbitone sodium, 1.2 mg per 25 g, intraperitoneally). Melanoma cell implantation was performed by injecting the cell solution intradermally (1 10⁶-10⁹ cells, up to 0.2 ml per injection site, up to six sites per animal) into the skin of the haunch of the anesthetized animal. Care was taken to inject the cells as close to the skin surface as possible so as to more closely mimic the clinical situation, which would be targeting early detection, in vivo. Melanoma growth was monitored over an 8 wk period and tumor growth was not allowed to exceed 2 mm in diameter.

In total, some 67 mice were implanted with one of the three melanoma cell types. Of the 67 mice, 35 developed visible tumors, indicating a moderate success rate (52%) for implanted cells to grow in vivo Table I.

Table I - Relationship between melanoma tumor depth and confocal detection in tumors imaged using FOCI.

Full table

Normal human skin cells were harvested and implanted in the same manner as the melanoma cells (see above). After implantation of the normal human skin cells (1 10⁶-10⁹ cells) into the haunch of the anesthetized animal, implant sites were monitored. The implanted cells formed a raised pale white dermal papule.

Imaging of melanoma tumors in vivo using FOCI

The skin surrounding the tumor site was cleaned with a gauze pad soaked in 70% ethanol, followed by Hibiclens (ICI Pharmaceuticals) skin cleanser. The tumor site was imaged prior to antibody administration to assess any potential autofluorescence from the tumor cells. Antimelanoma antibodies were diluted in PBS (with 10% FBS) for all in vivo experiments. Diluted CD 51/61-FITC, 14.G2a-FITC, or 9.2.27-FITC were administered in volumes of 0.2 ml, by intravenous injection into the tail vein of anesthetized (pentobarbitone sodium, 1.2 mg per 25 g, intraperitoneally) mice. Imaging of the tumor site commenced 15 min after the administration of the antibody solution. The mouse was placed on a modified microscope stage and immersion oil (Olympus) was placed on the area of skin to be imaged (to reduce refractive index changes at the skin surface), with a coverslip held in place with an adapted micropositioner. After approximately 60 min of imaging, if no fluorescent emission was detected a second, identical bolus of diluted antibody was administered and imaging was recommenced. In cases where the melanoma tumor was not detected from the skin surface, tumor sites were imaged ex vivo.

Use of an isotype control antibody

To ascertain whether the increased vascularity of melanomas predisposed to nonspecific binding of the tumor-specific fluorescent antibodies, experiments were performed to exclude this possibility by administration of a nonspecific antibody from the same species and isotype into mice with melanomas. Anesthetized nude mice (pentobarbitone sodium, 1.2 mg per 25 g, intraperitoneally) with implanted melanoma cells (n = 3 for each cell line) were administered intravenously (tail vein) the nontumor-specific isotype control antibody mouse IgG₁, FITC-conjugated (1:5 dilution in PBS, 10% FBS, 0.2 ml; BD Pharmingen). FOCI was then performed from 15 min to 4 h after injection. This was followed by the intravenous administration of 0.2 ml of antimelanoma antibody CD 51/61-FITC diluted in PBS (with 10% FBS). Imaging of the tumor site was recommenced 15 min after the administration of the tumor-specific antibody.

Imaging of melanoma tumors ex vivo using FOCI

After in vivo imaging of the melanoma tumor sites from the skin surface, the dermal surface of the melanoma implantation sites was surgically exposed for FOCI. The tumor site was exposed as a partially excised tissue flap using fine scissors whilst the mice remained under anesthesia. Immersion oil was used on the dermal surface of the skin and a glass coverslip was held in place with a modified micropositioner. This procedure was performed to determine the presence of the melanoma tumor and to confirm that the failure to detect a tumor from the epidermal surface was due to excessive tumor depth, rather than lack of labeling by the fluorescent antibody.

Vascular labeling with fluorescently labeled dextrans

The dermal microvasculature was visualized by injecting melanoma-implanted mice with high molecular weight FITC-labeled dextran (260,000 MW; Sigma, Australia) diluted in sterile saline (10 mg per ml; 0.3 ml, intravenously) to label plasma. Single channel confocal images of melanoma-associated vasculature were obtained 10 min after the administration of the FITC-dextran. Athymic BALB/C mice without melanoma implants were also imaged to compare normal dermal microvasculature.

Simultaneous dual channel confocal images were obtained after melanoma-implanted mice were administered CD51/CD61-FITC (1:5 dilution, 0.2 ml, intravenously) to label melanoma cells, followed 30 min later by Texas Red-dextran dissolved in sterile saline (70,000 MW; 0.2 ml of 10 mg per ml, intravenously; Molecular Probes) to label plasma. Imaging was commenced 10 min after Texas Red-dextran administration.

FOCI

FOCI was performed using an Optiscan F900e personal confocal system (Optiscan, Melbourne, Australia) fitted to an Olympus BH-2 microscope. In vivo single channel imaging and in vitro imaging used an argon-ion laser to excite the FITC-labeled antibodies at 488 nm and the fluorescent signal produced was detected above 505 nm. Objectives used were Olympus SPlan Apo 10 0.4 NA and Olympus SPlan Apo 20 0.7 NA.

Simultaneous dual channel imaging of melanoma (FITC-labeled antibodies) and surrounding vasculature (Texas Red-dextran; Molecular Probes) in vivo was performed using an Optiscan F900e rigid endomicroscope attachment (Optiscan) fitted to the F900e personal confocal system. The FITC-labeled antibodies and Texas Red-dextran were excited simultaneously using the 488 nm and 568 nm lines of a krypton-argon laser and their fluorescent emissions were detected at 505–550 nm and above 590 nm, respectively.

Mice were humanely killed using anesthetic overdose after completion of imaging procedures.

Conventional histology

The tumor site and surrounding tissue was excised following FOCI and fixed in 2% ethanol paraformaldehyde for 12–24 h. These tissue samples were then embedded in paraffin wax and 5 m sections were cut with a microtome and mounted on gelatin-coated glass slides. Tissue sections were then subjected to immunoperoxidase staining using HMB 45 (Dako, Australia) or S100 (Dako) primary antibodies to target the implanted melanoma cells (^{Leong and Millos, 1989;Blessing et al, 1998}). The secondary antibody (swine antirabbit Ig horseradish peroxidase; Dako) was removed by washing with PBS after 20 min incubation. The sections were counterstained with hematoxylin. Histologic sections were examined using light microscopy with an Olympus BX60 microscope and Olympus UPlan Apo objectives, to confirm the presence of melanoma cells at the tumor site, and to assess their depth from the skin surface.

Results

Imaging of the melanoma cells in vitro using FOCI

Preliminary experiments were performed using FOCI to observe the binding of antimelanoma antibodies to live melanoma cells grown on coverslips, using CD51/61-FITC Figure 1a, 14.G2a-FITC Figure 1b, and 9.2.27-FITC Figure 1c. This enabled confocal imaging of the cell boundaries. In the absence of antibody, no fluorescence was observed from either normal human skin cells Figure 1d or melanoma cells Figure 1e, nor was any fluorescence detected from normal human skin cells when incubated with antimelanoma antibodies Figure 2f. A range of antibody dilutions was tested in vitro on live melanoma cells and normal human skin cells (data not shown). A 1:5 dilution of each of the antibodies was found to be optimal for labeling all of the melanoma cell types in vitro.

Figure 1.

FOCI of living melanoma cells in vitro. Plasma membrane labeling following incubation with (a) CD51/61-FITC 1:5 dilution; HY cells; (b) 14.G2a-FITC (1:5 dilution; NMD cells); (c) 9.2.27-FITC (1:5 dilution; NMD cells). In the absence of antibody, normal human keratinocytes (d) and MW melanoma cells (e) showed little or no fluorescence. Scale bars: (a, c–e) 60 m; (b) 45 m.

Full figure and legend (59K)

Figure 2.

Subsurface confocal images of implanted melanoma tumors in athymic mice in vivo, after administration of 0.2 ml of 1:5 diluted antimelanoma antibodies. (a) MW melanoma tumor site in an athymic mouse before administration of the antibodies. (b) Normal mouse skin image following CD51/61-FITC administration intravenously into an athymic mouse with no implanted tumor. (c) NMD melanoma after CD51/61-FITC administration. (d) NMD melanoma following administration of 9.2.27-FITC. (e) NMD melanoma tumor following administration of 14.G2a-FITC. (f) Implanted normal human skin cells in an athymic mouse after the administration of CD51/61-FITC showing little or no fluorescence. (g) NMD melanoma tumor following administration of FITC-conjugated mouse IgG₁, monoclonal Ig isotype control showing little, scattered fluorescence. (h) NMD melanoma tumor after the administration of FITC-conjugated mouse IgG₁, monoclonal Ig isotype control followed by the administration of CD52/61-FITC. (i) MW melanoma tumor following administration of FITC-conjugated mouse IgG₁, monoclonal Ig isotype control showing little, scattered fluorescence. (j) MW melanoma tumor after the administration of FITC-conjugated mouse IgG₁, monoclonal Ig isotype control followed by the administration of CD52/61-FITC. (k) HY melanoma tumor following administration of FITC-conjugated mouse IgG₁, monoclonal Ig isotype control. (l) HY melanoma tumor after the administration of FITC-conjugated mouse IgG₁, monoclonal Ig isotype control followed by the administration of CD52/61-FITC. Scale bars: (a, b, g, k) 30 m; (c) 60 m; (d, l) 50 m; (e, f, h–j) 75 m.

Full figure and legend (61K)

Imaging of the melanoma tumor sites in vivo using FOCI

Prior to the administration of fluorescent antibodies in vivo, the tumor site was imaged to determine the extent of autofluorescence present below the surface of the skin. Little or no fluorescent signal was observed below the skin surface prior to antibody administration Figure 2a. Further controls were performed by injecting the fluorescent antibodies (0.2 ml of CD 51/61-FITC, 14.G2a-FITC, or 9.2.27-FITC at 1:5 dilution) into normal athymic BALB/C mice without a melanoma implant. Figure 2(b) shows the minimal fluorescent signal detected using FOCI in these mice, 15 min after initial antibody administration (CD 51/61-FITC shown).

A range of antibody concentrations was tested and optimal imaging was obtained following 0.2 ml of 1:5 dilution in the case of each of the three antibodies (data not shown). After intravenously injecting the fluorescently labeled antibody CD51/61-FITC (0.2 ml, 1:5 dilution), melanoma tumor cells could be observed in the animals in vivo, within 30 min of antibody administration. The intense fluorescent signal showing some cellular clustering at the tumor site is shown in Figure 2(c). Administration of 9.2.27-FITC (0.2 ml, 1:5 dilution, intravenously; Figure 2d) or 14.G2a-FITC (0.2 ml, 1:5 dilution, intravenously; Figure 2e) yielded similar images to those obtained with CD51/61, with strong fluorescence at the tumor site. Cultured normal human skin cells were also injected intradermally and imaged using FOCI. Figure 2(f) shows that no fluorescent signal was detected from these cells when they were injected into the mice and subjected to the same antibody regime in vivo.

To ascertain whether the increased vascularity of melanomas predisposed them to nonspecific binding of the tumor-specific fluorescent antibodies, experiments were performed to exclude this possibility by intravenous administration of the isotype control FITC-conjugated mouse IgG₁, monoclonal antibody into three nude mice for each of the three melanoma tumors (total of nine mice). Following this antibody administration, no tumor-associated fluorescence was detected by FOCI Figure 2g, i, k. Subsequent administration after 4 h in the same mice of the tumor-specific antibody CD51/61 intravenously, however, resulted in fluorescent labeling of the tumor sites (Figure 2h, j, l).

As shown in Table I, for tumors within 0.2 mm of the skin surface, when imaged from the epidermal surface, the CD 51/61-FITC antibody enabled detection of 12 out of 14 (86%) melanoma tumors in implanted mice in vivo; the 9.2.27-FITC antibody enabled detection of seven out of nine (78%) melanoma tumors; and 14G2a-FITC allowed detection of eight out of nine (89%) melanoma tumors. Overall, 84% (27 out of 32 tumors) of those implanted melanoma tumors that were less than 0.2 mm deep could be detected from the epidermal surface using FOCI. It should be noted that some mice had multiple tumor sites and some tumors were detected from only the dermal surface Table I. NMD cells, and to a lesser extent HY cells, formed tumors in implanted mice more readily than the MW melanoma cell line. Two MW tumors were detected from the epidermal surface using CD 51/61-FITC and 14G2a-FITC, respectively. Tumours that were more superficial (less than 120 m) were detected with a strong signal compared to those deeper (120 m-190 m) when imaged from the epidermal surface. Table I indicates that tumors at a greater depth than 200 m were not detected from the skin surface using FOCI, although many were detected from the dermal surface, indicating that antibody was binding to tumors.

Imaging of partially excised melanoma tumors

After epidermal subsurface imaging, FOCI was used to obtain confocal images from the dermal surface (as opposed to epidermal) of a partially excised inverted tissue flap containing the implanted melanoma tumor in anesthetized mice. Intense fluorescent signal was detected at the tumor site after in vivo administration of CD 51/61-FITC Figure 3a, 14.G2a-FITC Figure 3b, and 9.2.27-FITC Figure 3c. In some cases tumor cells appeared more dispersed than those imaged from the epidermal surface using FOCI (e.g., Figure 3a, c). In other cases (e.g., Figure 3b), however, a tighter clustering pattern of the cells was also seen from the dermal surface. No fluorescent signal was detected from the dermal surface of the melanoma implant when no fluorescent antibody was administered Figure 3d.

Figure 3.

FOCI of excised melanoma tumor sites from the dermal surface. (a) NMD melanoma imaged from the dermal surface after epidermal in vivo imaging, intravenous administration of CD51/61-FITC (0.2 ml, 1:5 dilution). (b) Imaged from the dermal surface NMD melanoma tumor after in vivo intravenous administration of 14.G2a-FITC (0.2 ml, 1:5 dilution). (c) HY melanoma imaged from the dermal surface after in vivo intravenous administration of 9.2.27-FITC (0.2 ml, 1:5 dilution). (d) MW melanoma imaged from the dermal surface with no fluorescent antibody administered, showing no fluorescence. Scale bars: (a, c, d) 45 m; (b) 75 m.

Full figure and legend (77K)

Confirmation of melanoma using light microscopy

Immunoperoxidase staining with either HMB 45 (data not shown) or S100 antibody was performed to confirm the presence of melanoma cells in the tumor site biopsies taken from implanted mice with visible tumors. Positively stained melanoma cells appear brown in Figure 4(a) (arrows), as do nerves and skeletal muscle cells when stained with S100. Negative control sections Figure 4b where HMB45 and S100 were omitted from the staining procedure showed no brown staining of cells. The conventional histologic sections enabled melanoma cells within the different skin layers to be visualized relative to other skin structures and depth below the skin surface to be assessed. The 52 tumors (35 NMD, 15 HY, two MW) (in 35 mice) detected using FOCI from the epidermal and/or dermal surfaces were confirmed to be melanomas by conventional histology. For eight mice (with 12 tumors), either they died under anesthesia before the imaging procedure or conventional histology could not confirm the presence of melanoma at the tumor site (see Table I).

Figure 4.

Conventional immunoperoxidase-stained cross-section of mouse skin. (a) S100-stained melanoma cells () below the epidermis (Epi). Hair follicles (HF), collagen fibers (C). Nerves cells (N) are also positively stained. (b) Negative control (no S100) of cross-section through athymic mouse skin showing no positively stained cells below the epidermis (Epi). Scale bars: (a, b) 30 m.

Full figure and legend (300K)

Imaging of vasculature associated with melanoma

Figure 5(a) shows an image following intravenous FITC-dextran administration into an athymic mouse without a melanoma implant. In contrast, Figure 5(b) shows a dense and tortuous pattern of dermal microvasculature in an athymic mouse with a melanoma tumor. The use of Texas Red-dextran and FITC-labeled antimelanoma antibodies enabled dual channel imaging to be performed of both the melanoma tumor and the surrounding vasculature in vivo. The dual channel confocal image shown in Figure 5(c) shows blood vessels in red, whereas the melanoma tumor cells appear green. The close proximity of blood vessels to the melanoma tumor indicates the vascular nature of the tumor.

Figure 5.

Changes in melanoma-associated vasculature as imaged by FOCI, in vivo. (a) Confocal image of normal murine dermal vasculature following intravenous administration of FITC-dextran (0.3 ml, 10 mg per ml) in an athymic mouse without a melanoma implant. (b) FOCI image of the dermal vasculature in and around an implanted melanoma tumor of an athymic mouse, following intravenous administration of (0.3 ml, 10 mg per ml) FITC-dextran. (c) Dual channel confocal image showing melanoma-associated dermal microvasculature labeled red (Texas Red-dextran; 0.2 ml, 10 mg per ml, intravenous) and the green labeled melanoma tumor (CD51/61-FITC; 0.2 ml, 1:5 dilution, intravenous). Scale bars: (a) 200 m; (b) 250 m; (c) 150 m.

Full figure and legend (107K)

Discussion

There are currently a number of techniques in addition to laser scanning fluorescence confocal microscopy that may be suitable for noninvasive imaging of skin tumors. For example, tandem scanning or laser scanning reflectance confocal microscopy can provide images of skin cells at depths similar to that of laser scanning fluorescence confocal microscopy (e.g.,^{Masters et al, 1997;Corcuff and Pierard, 1998;Gonzalez et al, 1999;Rajadhyaksha et al, 1999;Aghassi et al, 2000}).

Other techniques that may be suitable for noninvasive skin tumor imaging include high definition ultrasound and optical coherence tomography (OCT). High definition ultrasound may help differentiate melanomas from common benign pigmented lesions; however, it cannot differentiate them from nonkeratotic acanthotic basal cell papillomas (^{Lassau et al, 1999;Harland et al, 2000}). OCT has a spatial resolution of 10–15 m and a skin penetration depth of 0.5–1.5 mm, but little reported value in the differentiation of choroidal tumors and melanomas (^{Welzel et al, 1997;Schaudig et al, 1998}). Novel ultrahigh resolution (1.1 m) OCT has been evaluated in vitro in gastrointestinal imaging, however, and improves structural tissue recognition (^{Jackle et al, 2000;Li et al, 2000}). Thus, high definition ultrasound and OCT can delimit the size of skin tumors and image deeper structures than fluorescence confocal microscopy, but they have a lower resolution (^{Tadrous, 2000a}) and have not been reported to provide a reliable diagnosis. Indeed, without the use of specific contrast agents, differentiation of melanomas from other nonpigmented lesions using techniques such as tandem scanning, laser scanning reflectance confocal microscopy, and OCT may be difficult. A combination of fluorescence confocal microscopy (to provide high resolution and tumor diagnosis) with high definition ultrasound and/or OCT (to delimit deep tumor margins) may be advantageous for noninvasive skin tumor imaging.

In this study, bright fluorescent staining of melanoma tumors, derived from three human melanoma cell lines, was produced in vivo using the three FITC-labeled antimelanoma antibodies. In contrast, neither implanted normal human skin cells nor normal mouse skin surrounding the tumor site produced fluorescent signal, suggesting specific labeling of the melanoma cells in vivo. In addition, FITC-conjugated isotype control antibodies did not label tumors. Groups of cells were evident when the tumor site was imaged from the epidermal surface in vivo; however, when imaging excised tissues from the dermal surface, a more scattered pattern of fluorescent labeling was sometimes detected. This pattern of fluorescent staining observed in the deeper layers of the tumor mass may be indicative of a greater and more diffuse spread of tumor cells in deeper layers of the skin, or the variable growth patterns of the three different human melanoma cell lines used in this study.

FOCI was performed parallel to the skin surface. In general, those tumors that were located close to the skin surface were detected with a strong fluorescent signal. The intensity of the fluorescent signal detected decreased with increasing tumor depth, however. At greater depths of 120–150 m fluorescent signal strength was found to be diminished.

The depth of imaging below the surface of the tissue using FOCI is limited by the wavelength of light required to excite the fluorophores used in this study (FITC, 488 nm; Texas Red, 568 nm). It is also possible that the relative maximum imaging depth in human skin by FOCI may be less than that in mouse skin, as mouse epidermis is only a few cells thick and may be more transparent than human skin at certain wavelengths. Although FOCI has been performed to approximately 200 m below the skin surface following high pressure jet administration of fluorescein and acridine orange (^{Bussau et al, 1998}), increasing this imaging depth would require the use of dyes excited by longer wavelengths. Laser-tissue interaction prevents imaging to greater depths, with the combination of 488 nm laser line and FITC being limited by epidermal optical properties such as light scattering and refractive index changes caused by different cell types (^{Vargas et al, 1999;Tadrous, 2000b}). The apparent loss of resolution in some tumors imaged from the epidermal surface, compared to those imaged from the dermal surface, could be caused by the heterogeneous nature of skin and increased scattering at greater depths from the epidermal surface. Refractive index changes cause loss of signal and resolution at these imaging depths. Nevertheless, other studies have shown that in vivo imaging of the human dermis (e.g., 200 m depth) is possible using confocal microscopy with visible light and near-infrared wavelengths (e.g.,^{Masters et al, 1997;Bussau et al, 1998;Corcuff and Pierard, 1998;Gonzalez et al, 1999}). The imaging depths achieved in this and other studies, however, allows imaging of the dermal layers of skin in both mice and humans, and as melanoma tumor development begins in the epidermis of humans, most sites would potentially be within the range of in vivo FOCI.

Subsurface imaging of the dermal blood vessels in melanoma-implanted mice enabled resolution of the vascular supply near the tumor site. In comparison with dermal blood vessels in mice without implanted tumors, the density and tortuosity of the blood vessels at the tumor site was increased. This includes changes in blood vessel diameter and increased branching compared with normal blood vessels. These results are consistent with scanning electron microscopy studies showing significant changes in blood vessel morphology with tumor progression (^{Kondering et al, 1999}). The modified pattern of vasculature in the dermis of tumor-implanted mice in this study is consistent with the notion that factors are released from the melanoma tumor cells to initiate angiogenesis and increase overall vascularity around the tumor site (^{Fidler and Ellis, 1994}). As FOCI allows detection and imaging of dermal blood vessels, the angiogenic properties of melanoma tumors may provide prognostic information (^{Jain, 1988}), and with further development FOCI could be used to quantify tumor angiogenesis in vivo and provide a better understanding of the vascular nature of tumors and potential treatment of tumors with antineoplastic therapies.

The heterogeneous expression of melanoma-associated antigens with the progression of melanoma may limit the detection sensitivity achieved using a single antibody for in vivo imaging. For example, CD51/CD61 targets the _v3 integrin expressed exclusively during the vertical growth phase of primary melanomas and metastases (^{Danen et al, 1995}). Consequently, the use of multiple antibodies (to target multiple antigens) in combination is likely to be required to enable in vivo imaging of all types of human melanomas in future studies. Indeed, a combination of antibodies is currently used for conventional diagnosis of melanoma, with S100 antibody being applied if a negative result is first obtained with HMB45 (^{Blessing et al, 1998}).

One limiting factor encountered in this study was the difficulty in reliably obtaining tumors of some cell types in the nude mice. The human melanoma cell lines HY and MW were found to be difficult to grow either in culture or when implanted into the mice, and this is reflected in the relatively small numbers of these tumors examined in this study.

Reliable administration of the melanoma cells into the superficial dermal layers (< 200 m deep) of athymic mouse skin was found to be a significant shortcoming of this animal model of melanoma. It was attempted to implant the melanoma cells as close to the skin surface as possible in order to best simulate the clinical situation where melanoma originates in the epidermal (superficial) layers of the skin. Using the techniques described in this study it was difficult to consistently implant cells at a given superficial depth less than 200 m deep. The ability of FOCI to detect lesions, however, was dependent upon the depth of the implanted cells, rather than the imaging technique. The 200 m limit given by this study is sufficient to detect and discriminate early melanoma in the animal model used. Also, the FOCI technique (encompassing the use of fluorescently labeled antibodies) showed selectivity, in that cancerous cells were detected in preference to surrounding normal mouse cells and implanted normal human skin cells. The ex vivo tissue flap imaging was employed to determine whether the nondetection of a known tumor was due to cell implantation into deeper skin structures.

In conclusion, this study successfully demonstrates that fluorescently labeled antibodies can be administered systemically into an athymic (nude) mouse model of human melanoma to enable detection of some tumors in vivo with FOCI. Potentially, FOCI could offer several advantages over conventional biopsy techniques currently being used clinically, including the ability to detect melanoma cells without anesthesia or surgery whilst allowing cellular level diagnosis. Further studies are required with quantification of tumor size and depth, signal strength, the sensitivity and specificity associated with various antibody combinations, and human cell lines, however, if FOCI is to be used as a tool for melanoma diagnosis. Nevertheless, this study has demonstrated the feasibility of using a novel optical imaging technique and fluorescent antibodies to detect small melanomas in vivo for the first time.

References

1. Aghassi, D, Gonzalez, E, Anderson, RR, Rajadhyaksha, M, Gonzalez, S: Elucidating the pulsed-dye laser treatment of sebaceous hyperplasia in vivo with real time confocal scanning laser microscopy. J Am Acad Derm 2000 43:49–53,  | PubMed | ISI | ChemPort |
2. Blessing, K, Sanders, DSA, Grant, JJH: Comparison of immunohistochemical staining of the novel antibody melan-A with S100 protein and HMB-45 in malignant melanoma and melanoma variants. Histopathology 1998 32:139–146,  | Article | PubMed | ISI | ChemPort |
3. Bumol, TF & Reisfeld, RA: Unique glycoprotein-proteoglycan complex defined by monoclonal antibody on human melanoma cells. Proc Natl Acad Sci USA 1982 79:1245–1249,  | PubMed | ChemPort |
4. Bussau, LJ, Vo, LT, Delaney, PM, Papworth, GD, Barkla, DH, King, RG: Fibre optic confocal imaging (FOCI) of keratinocytes, blood vessels and nerves in hairless mouse skin in vivo. J Anat 1998 192:187–194,  | Article | PubMed | ISI |
5. Corcuff, P & Pierard, GE: Skin imaging: state of the art at the dawn of the year 2000. Cur Prob Derm 1998 26:1–11,  | ISI | ChemPort |
6. Danen, EH, Jansen, KF, Van Kraats, AA, Cornielisen, IM, Ruiter, DJ, Van Muijen, GN: Alpha v-integrins in human melanoma: gain of alpha v and beta 3 and loss of alpha v beta 5 are related to tumour progression in situ but not to metastatic capacity of cell lines in nude mice. Int J Cancer 1995 61:491–496,  | PubMed | ISI | ChemPort |
7. Duman, Y, Burak, Z, Ercan, MT, Dirlik, A, Bilkay, BC, Akin, Y, Taner, M, Bekdik, CF: Clinical evaluation of metastases of malignant melanoma imaging with 99Tcm-glutathione and 99Tcm-anti-melanoma antibody: a comparative study. Nucl Med Commun 1995 16:927–935,  | PubMed | ISI | ChemPort |
8. Fidler, IJ & Ellis, LM: The implications of angiogenesis for the biology and therapy of cancer metastasis. Cell 1994 79:185–188,  | Article | PubMed | ISI | ChemPort |
9. Goldstein, AM & Tucker, MA: Etiology epidemiology, risk factors and public health issues of melanoma. Curr Opin Oncol 1993 5:358–363,  | PubMed | ChemPort |
10. Gonzalez, S, Rajadhyaksha, M, Rubinstein, G, Anderson, RR: Characterization of psoriasis in vivo by reflectance confocal microscopy. J Med 1999 30:337–356,  | PubMed | ChemPort |
11. Harland, CC, Kale, SG, Jackson, P, Mortimer, PS, Bamber, JC: Differentiation of common benign pigmented skin lesions from melanoma by high definition ultrasound. Br J Derm 2000 143:281–289,  | ISI | ChemPort |
12. Jackle, S, Gladkova, N, Feldchtein, F, et al: In vivo endoscopic optical coherence tomography of the human gastrointestinal tract – toward optical biopsy. Endoscopy 2000 32:743–749,  | Article | PubMed | ISI | ChemPort |
13. Jain, RK: Determinants of tumour blood flow: a review. Can Res 1988 48:2641–2658,  | ISI | ChemPort |
14. Kondering, MA, Malkusch, W, Klapthor, B, et al: Evidence for characteristic vascular patterns in solid tumours: quantitative studies using corrosion casts. Br J Can 1999 80:724–732,
15. Kopf, AW, Salopek, TG, Slade, J, Marghoob, AA, Bart, RS: Techniques of cutaneous examination for the detection of skin cancer. Cancer 1995 75:684–690,  | PubMed | ISI | ChemPort |
16. Lane, AM, Egan, KM, Yang, J, Saornil, MA, Alroy, J, Albert, D, Gragoudas, ES: An evaluation of tumour vascularity as a prognostic indicator in uveal melanoma. Melanoma Res 1997 7:237–242,  | PubMed | ISI | ChemPort |
17. Lassau, N, Mercier, S, Koscielny, S, et al: Prognostic value of high-frequency sonography and color Doppler sonography for the preoperative assessment of melanomas. Am J Roentgenol 1999 172:457–461,  | ISI | ChemPort |
18. Leong, AS & Millos, J: An assessment of a melanoma specific antibody (HMB45) and other markers of malignant melanoma in paraffin embedded tissues. Surg Pathol 1989 2: 137
19. Li, XD, Boppart, SA, Van Dam, J, et al: Optical coherence tomography: advanced technology for endoscopic imaging of Barrett's esophagus. Endoscopy 2000 32:921–930,  | Article | PubMed | ISI | ChemPort |
20. Marincola, FM, Hijazi, YM, Fetsch, P, et al: Analysis of expression of the melanoma-associated antigens MART-1 and gp100 in metastatic melanoma cell lines and in situ lesions. J Immunother Emphasis Tumour Immunol 1996 19:192–205,  | ChemPort |
21. Masters, BR, Gonnord, G, Corcuff, P: Three dimensional microscopic biopsy of in vivo human skin: a new technique based on a flexible confocal microscope. J Microscopy 1997 185:329–338,  | ISI |
22. Mauceri, HJ, Hanna, NN, Beckett, MA, et al: Combined effects of angiostatin and ionizing radiation in antitumour therapy. Nature 1998 394:287–291,  | Article | PubMed | ISI | ChemPort |
23. Mukerjee, S, Nasoff, M, McKnight, M, Glassy, M: Characterisation of human IgG1 monoclonal antibody against gangliosides expressed on tumour cells. Hybridoma 1998 17:133–142,  | PubMed | ISI | ChemPort |
24. Pichla, SL, Murali, R, Burnett, RM: The crystal structure of a Fab fragment to the melanoma-associated GD2 ganglioside. J Struct Biol 1997 119:6–16,  | Article | PubMed | ISI | ChemPort |
25. Rajadhyaksha, M, Gonzalez, S, Zavislan, JM, Anderson, RR, Webb, RH: In vivo confocal scanning laser microscopy of human skin II: advances in instrumentation and comparison with histology. J Invest Derm 1999 113:293–303,  | PubMed | ISI | ChemPort |
26. Schaudig, U, Hassenstein, A, Bernd, A, Walter, A, Richard, G: Limitations of imaging choroidal tumors in vivo by optical coherence tomography. Graefes Arch Clin Exp Ophthal 1998 236:588–592,  | ISI | ChemPort |
27. Tadrous, PJ: Methods for imaging the structure and function of living tissues and cells: 1. Optical coherence tomography. J Path 2000a 191:115–119,  | ISI | ChemPort |
28. Tadrous, PJ: Methods for imaging the structure and function of living tissues and cells: 3. Confocal microscopy and micro-radiology. J Path 2000b 191:345–354,  | ISI | ChemPort |
29. Van Der Laak, JAWM, Westphal, JR, Schalkwijk, LJM, Pahlplatz, MMM, Ruiter, DJ, De Waal, RMW, De Wilde, PCM: An improved procedure to quantify tumour vascularity using true color analysis. Comparison with the manual hot-spot procedure in a human melanoma xenograft model. J Path 1998 184:136–143,  | PubMed | ISI | ChemPort |
30. Vargas, G, Chan, EK, Barton, JK, Rylander, HG, Welch, AJ: Use of an agent to reduce scattering in skin. Laser Surg Med 1999 24:133–141,  | Article | ISI | ChemPort |
31. Welzel, J, Lankenau, E, Birngruber, R, Engelhardt, R: Optical coherence tomography of the human skin. J Am Acad Derm 1997 37:958–963,  | PubMed | ISI | ChemPort |

Acknowledgments

We gratefully acknowledge the Anticancer Council of Victoria for providing a grant for this project.