Conventional histopathology remains the gold standard for microscopic evaluation of skin histopathology. Tissue excision, processing, sectioning, and staining are required to allow light-microscopic analysis; yet although cellular details are visualized at high resolution, biopsies are painful, expensive, time-consuming, and associated with scar formation. Furthermore, routine histology does not allow repeated evaluations of dynamic processes over time, because the tissue under investigation is removed for evaluation. Finally, given anatomic variations and cellular pleomorphism, sampling errors remain a significant challenge in a variety of skin diseases with advanced cutaneous involvement such as nonmelanoma skin cancer (NMSC)1, 2 or T-cell lymphoma.3
Recent developments in optical imaging techniques may offer new avenues for real-time noninvasive, high-resolution analysis of skin and several of these technologies have become commercially available. Among these, optical coherence tomography (OCT), high-frequency ultrasound (HFUS), magnetic resonance imaging (MRI), multispectral fluorescence polarization imaging, TetraHertz imaging, near infrared and Raman spectroscopy, and reflectance or fluorescence confocal microscopy are the most promising for real-time evaluation of cutaneous tissue. 4, 5, 6, 7, 8, 9, 10 OCT, HFUS, and MRI obtain full thickness vertical optical sections that correlate well with the gross morphology of routine histology, yet the resolution is limited and does not reach cellular levels. On the other hand, in vivo reflectance and fluorescence confocal microscopy (CM) generate horizontal optical sections at a resolution that is comparable to routine histology, but at the present time, these technigues are limited by their penetration depth.11, 12, 13, 14, 15, 16
The underlying principle of fluorescent confocal laser scanning microscopy (FLSM) systems is the excitation and detection of fluorophores by scanning a focal plane within the tissue using a coherent laser light source. A fiber-optic coupler system separates tissue illumination from fluorescence detection, thereby enabling a high resolution of FLSM images of about (Fig. 1 ).14, 15 Contrast and image interpretation is based on the presence of endogenous or exogenous fluorophores. In that respect, two sources of fluorescent molecules can be differentiated: autofluorescence (endogenous fluorophores) or applied fluorescent contrast agents (exogenous fluorophores).12, 13, 14 Autofluorescence from flavins, porphyrins, and collagen has an excitation maximum at the spectrum of UV light, which—due to the shallow penetration of UV radiation—only provides a poor contrast signal. For clinical purposes in FLSM imaging, the application of exogenous “fluorescent dyes” is required to provide sufficient contrast within the tissue.11
To visualize cellular structures, lasers of different wavelengths may be used, yet although the longer wavelengths penetrate deeper into the skin, they also yield lower resolution compared to shorter wavelengths. The fluorescence confocal microscope (FLSM) used in this study employs an argon-ion laser at a wavelength of for tissue illumination, but detection occurs at longer wavelengths. After excitation, the emitted fluorescence then allows visualization of the morphological details of cellular microstructures.13, 14 Image contrast is achieved through distribution and accumulation of exogenously applied fluorophores in the intercellular and intracellular compartments of live skin.13, 14
Other confocal microscopy systems are based on reflectance measurements of structures with different refractive indices.17, 18, 19 Melanin, hemoglobin, and other cellular microstructures with high-refractive indices produce strong backscatter with visible and near infrared wavelengths, and the visualization of epidermal structures is based on differences in different structural reflectivity. In the case of reflectance confocal microscopy (RCM), systems illumination and detection takes place at the same wavelength. In both systems, when correlating CM images to routine histology, the major differences as compared to conventional histology are that the images are oriented horizontally to skin surface (en face) and that they are analyzed in gray scale.18, 19
This study evaluates the clinical applicability of FLSM in the evaluation of normal and diseased skin in vivo. The evaluations were used to (a) describe topographic variations of normal skin and (b) evaluate nonmelanoma skin cancer noninvasively, in reference to normal skin and in correlation with vertical hematoxylin and eosin (H&E) histopathology. Lastly, (c) FLSM was used for noninvasive monitoring of skin disease and therapeutic response to novel treatment modalities over time.
Materials and Methods
Forty individuals aged between 37 and were asked to participate in this study. All patients were recruited from the Department of Dermatology at the University Hospital of the Charité, Berlin. Written consent was obtained prior to investigation. All research was conducted according to the Declaration of Helsinki principle.
Fluorescence Confocal Laser Scanning Microscope
The system is equipped with a single-moded optical fiber incorporated into a handheld scanning device, which is used for both illumination of the tissue and detection of fluorescence signals. A coherent argon-ion laser light source at a wavelength of is used to illuminate the tissue, thereby exciting the fluorescent molecules within the specimen to a singlet state. Upon returning to their respective ground states, the wavelengths emitted by these fluorophores are longer than illumination wavelengths, permitting them to be propagated through the optical fiber. A fiber-optic coupler separates the shorter wavelengths used for illumination from the longer wavelengths returning from the focal point within the tissue, and signals are then propagated to the photomultiplier tube (PMT) for detection. Light coming from out-of-focus planes will not be propagated, thus enabling the high resolution of FLSM images. Similarly, reflected or backscattered laser light is eliminated by a fluorescence barrier filter, only permitting longer wavelengths corresponding to fluorescence signals to pass through to the filter. Tissue emitted fluorescence is then focused onto a PMT for detection, generating images of a lateral resolution of approximately and an axial resolution (section thickness) of . Images are resolved in gray scale and the scanned field of view (FOV) is producing images of . Based on a FOV, when this is displayed on the standard monitor on the Stratum used in the present study, the displayed image is approximately , which equates to approximately magnification. The numerical aperture (NA) of the system was 0.5, illumination power on tissue level was less then . Figure 1 illustrates the optical scheme of FLSM.
Briefly, after intraepidermal injection of 0.2% sodium fluorescein ( total volume), the optical fiber is placed onto the skin using a handheld contact device without fixing the optical system to the patient’s skin. By moving the objective lens in the (vertical) direction with respect to the skin surface, it is possible to image at different horizontal planes within the tissue. During the imaging process, the depth level being imaged was ascertained by the morphological appearance of tissue or by using a micrometer attached to the stage of the objective lens. By zeroing the imaging depth at the level of the most superficial layer before going vertically into the tissue, depth measurements can be obtained. By moving the tip of the scanning device in the and directions, different areas within the specimen are evaluated. To allow an evaluation of corresponding skin sites over time, skin markings were made using permanent marker on the day of the baseline evaluation, whereby skin sites were determined for all followup evaluations. Additionally, a schematic depiction of the lesion and its outline was obtained on transparencies, such that by localizing anatomic hallmarks, relatively precise colocalization was possible over extended periods of time. Furthermore, four to six representative FLSM images of each level are obtained to demonstrate the reproducibility of each parameter and correct for potential variations with the tissue under investigation.
Application of the Fluorescent Dye
A 0.2% solution of sodium fluorescein (Fluorescein SE Thilo, Producer Alcon, France) was used as fluorescent dye and contrast agent for labeling of skin structures. The solution was administered under standardized conditions using a tuberculin syringe fitted with a 32-gauge needle13, 14 for superficial, intraepidermal injection. The injected volume did not exceed a total volume of and a diameter area was infiltrated with fluorescein sodium. In addition, one drop of sodium fluorescein was applied onto the site of FLSM evaluation for labeling of skin surface structures. Exact placement of the syringe within the epidermal compartment was ensured by using a angle with respect to the surface. Once intraepidermally injected, fluorescein is readily distributed within the skin. There is minimal pain associated with the insertion of the needle. In concordance with previous publications,13, 14 the administration of fluorescein was performed at approximately prior to the evaluation. The fluorophore is first detectable in the extracellular space, contrasting the intercellular compartment resulting in a bright white-gray contrast. It then diffuses into the cytoplasm and nuclei of epidermal keratinocytes, to finally be redistributed into the cytoplasm. Following fluorescein injection, a drop of immersion oil was applied to the sites selected for imaging, matching the refraction index of the stratum corneum to reduce backscatter and reflectance from surface structures.
FLSM evaluation of normal skin
Ten individuals with no prior history of skin disease were systematically evaluated using FLSM to define the morphologic features of normal skin in variable anatomic skin sites. Our FLSM evaluation parameters were based on observations previously made by Swindle 13, 14 Sun-protected skin sites were evaluated by obtaining sequential images at matching skin sites of the volar aspect of the right or left forearm. Additional evaluations were performed on sun-exposed skin sites, by examining dorsal aspects of the forearms. For FLSM evaluation of eccrine glands and the influence of dry skin (xerosis) on the structural integrity of the skin, both palms and lower legs of each volunteer were imaged, respectively.
FLSM evaluation of neoplastic skin lesions
A total of 30 patients with a clinically suspected diagnosis of NMSC were evaluated using FLSM to define the histomorphometric features of NMSC in correlation with and reference to routine histology. All lesions were systematically evaluated for the presence or absence of characteristic features of actinic keratoses (AK) or basal cell carcinoma (BCC), based on preliminary experiments and previously published reflectance confocal microscopy criteria of AK and BCC.20, 21, 22 FLSM evaluations were performed in correlation with established criteria of routine histopathology such as nuclear pleomorphism, cellular atypia, tumor nesting, stratum corneum disruption, inflammation, and increased vascularity. Following FLSM evaluation, all lesions were excised for histopathological diagnosis and correlation with routine H&E histology. Fifteen out of 30 were histologically characterized as AK; the remaining 15 were histologically classified as BCC.
Use of FLSM for noninvasive monitoring of therapy
We tested the applicability of FLSM for noninvasive monitoring of skin response to therapy in those 15 patients with a diagnosis of BCCs confirmed by routine histopathology. A topical immune response modifier (Imiquimod) was used for topical treatment for BCC.23, 24 Imiquimod (Aldara, Laboratoires Santé, Cergy Pontoise Cedex, France) has received Food and Drug Administration approval for the treatment of NMSC and numerous investigations have demonstrated its effectiveness in treating BCC.23 Local applications of Imiquimod were initiated at the recommended dosing schemes (i.e., three times weekly for six weeks) and patients were evaluated using FLSM at baseline, into therapy, and 8 and after cessation of therapy, respectively. FLSM images were obtained at well-defined evaluation sites and the respective degree of atypia, cellular pleomorphism, and inflammation were noted at different time points. Observational analysis included those FLSM features of BCC established by the initial FLSM evaluation described above. Their presence or absence was recorded in each skin layer and a minimum of four images were captured and stored for repeated analysis at a later time point.
We obtained punch skin biopsies from selected skin sites suspicious for AK or BCC upon clinical evaluation for correlation with FLSM. After fixation with formalin, the , paraffin-embedded sections were stained with H&E for histological characterization of the tumor. All cases were examined by a board-certified dermatopathologist. No biopsies were obtained from control sites. Histological evaluation criteria for AK included the presence of parakeratosis (presence of nuclei in the stratum corneum) and hyperkeratosis (increased thickening of the stratum corneum), both of which indicate epidermal hyperproliferation and aberrant epidermal differentiation. Nuclear pleomorphism and cellular atypia appear as variable sizes of keratinocyte nuclei, the eccentric placement of the nuclei, and an aberrant nuclei-to-cytoplasm ratio with atypically enlarged nuclei. The loss of normal epidermal architecture is indicative of aberrant keratinocyte differentiation and notably refers to the loss of the characteristic layering of the epidermis due to the replacement by aberrant, neoplastic keratinocytes. Histological evaluation criteria for BCC include the formation of tumor nests consisting of aberrant basal cells. The polarization of cells gives a monomorphic appearance to the tumor nests, and a characteristic cleft formation may be observed surrounding these tumor lobules. The pleomorphism of the epidermal keratinocytes surrounding BCC nests reflects an aberrant maturation and differentiation of keratinocytes and corresponds to the long-standing actinic damage that has occurred in the affected skin sites. Unspecific criteria of both malignancies include the presence of inflammatory cells, spongiosis (inter- and intracellular edema), and increased vascularity in the superficial dermal layer, all of which reflect an unspecific response to the presence of tumor tissue and neovascularization.
Fluorescence Confocal Microscopy (FLSM) Findings of Normal Skin
We have based our analysis on observations previously made by Swindle 13, 14 and our findings were in agreement with those obtained in other experiments.13, 18 When imaging the skin in real time starting from the surface and progressing deeper, the most superficial images obtained are from the stratum corneum. The morphological appearance is that of polygonal cells measuring approximately in size, grouped in “islands” separated by skin folds, which appear very dark [Fig. 2a ]. The stratum granulosum consists of two to four layers of cells, each cell measuring in size. By using FLSM, nuclei can be appreciated as either bright or dark, round to oval structure placed centrally within the cell depending on the distribution of fluorescein as a function of time [Fig. 2b]. The cells of the stratum spinosum are arranged in a tight honeycombed pattern of smaller cells, each cell measuring in size, with well-demarcated cell borders.
The deepest layer of epidermis, the basal layer, is seen as dark clusters of cells, each cell measuring about . When imaging at the suprapapillary epidermal plate, the dermoepidermal junction appears as round to oval rings of dark basal cells surrounding dermal papillae, which appear bright due to the homogenous distribution of fluorescein in this layer. A central area of blood flow corresponds to the presence of superficial dermal capillary loops [Fig. 2c]. Other features that can be observed in normal skin include hair shafts with pilosebaceous units [Figs. 3a and 3b ]; the latter appear as whorled centrally hollow structures with elliptically elongated cells at the circumference and a central hair shaft. Similarly, eccrine ducts, which appear as bright centrally hollow structures that spiral through epidermis and dermis, can be visualized [Figs. 3c and 3d].
When looking at the topographical variations of normal skin, patterns vary depending on the site and skin color being evaluated. Chronically sun-exposed skin also demonstrates a thicker and more fissured or wrinkled stratum corneum and more randomly arranged and irregularly shaped dermal papillae. Furthermore, bright reticulated patterns consistent with collagen and irregular elastin fibers may be visualized. According to our preliminary findings, there may be significant numerical variations of epidermal keratinocytes such that the number of keratinocytes is greater in sun-protected sites. The palms and soles are characterized for having an extremely thick stratum corneum and a greater number of eccrine ducts. Similarly, dry skin exhibits more superficial disruptive features in the stratum corneum compared to healthy skin [Figs. 4a, 4b, 4c, 4d ].
Fluorescence Confocal Microscopy of Neoplastic Skin Lesions
Actinic keratoses (AK)
FLSM features of AK corresponded well to the established key histopathological features of AK [Fig. 5a ]. In comparison with normal skin [Fig. 5b], FLSM was able to visualize the architectural disarray and keratinocyte pleomorphism at the level of the basal, spinous, and granular layers. [Figs. 5c and 5d]. Keratinocyte atypia are seen as cells and nuclei, varying in size and orientation. Irregular fluorescein distribution reflects the loss of clear cell-to-cell demarcation. Prominent superficial disruption associated with aberrant differentiation and hyperproliferation of epidermal keratinocytes resulted in detached corneocytes and visualization of bright superficial scale. Exocytosis and mild-to-moderate spongiosis was frequently observed, but the increased vascularity was not consistently visualized using FLSM. The absence of tumor nesting served as a distinguishing feature against BCC.
Basal cell carcinoma (BCC)
FLSM was able to detect the morphologic characteristics of BCC that were in correlation with routine histology. Tumor nests corresponded to the presence of islands of atypical basal cells that are monomorphic in shape [Fig. 6c ]. Large, elongated nuclei were often oriented along the same axis, resulting in an overall polarized appearance. Areas of clefting demarcated the tumor nests against surrounding epidermal structures [Fig. 6d]. A superficial inflammatory infiltrate was visualized as multiple, round to oval structures of in diameter [Fig. 6d]. Prominent atypia and pleomorphism of epidermal keratinocytes compared well to that observed in AK. At the level of the dermoepidermal junction [Fig. 7a ], the aberrant epidermal cells interfere with the normal honeycomb pattern of normal skin [Fig. 7b] and result in the disruption of dermal papillary architecture [Fig. 7c]. Dermal papillae exhibit abundant blood vessels with prominent tortuosity as well as increased blood vessel dilatation [Fig. 7d, 7e, 7f].
FLSM for Monitoring of Therapy
Upon initial FLSM examination, BCC sites showed characteristic atypia of basal keratinocytes with loss of cell-to-cell demarcation, tumor nesting, and nuclear pleomorphism [Fig. 8a ]. Similar evaluations following localized Imiquimod application showed spongiosis and an inflammatory infiltrate seen as multiple, round to oval structures with bright appearance. At three weeks into treatment, FLSM illustrated an inhomogenous appearance of pleomorphic basal cells and atypical keratinocytes remained. Residual atypia, nuclear pleomorphism, and a scarce inflammatory infiltrate can be observed. [Fig. 8b]. At 8 and after initiation of therapy, a normalization of epidermal architecture can be observed by FLSM in selected treatment sites, and residual atypia and nuclear pleomorphism was still detectable in others [Fig. 8c].
There are several advantages of in vivo laser scanning microscopy (LSM) over conventional histology. The imaging is painless and noninvasive, causing no tissue damage or alteration by processing or staining. Another aspect is the rapidity with which the data can be collected, and the ability of LSM for repeated evaluations of dynamic processes in real time.14, 18, 19, 20, 21, 22 These may include the longitudinal evaluation of physiologic events, disease evolution over time, or response to therapy.24, 25 In that regard, FLSM characterization of neoplastic lesions is an important area for research, with the potential to aid in the noninvasive diagnosis and management. With the advent of newer less invasive or topical therapies, it appears desirable to employ noninvasive diagnostic tools for evaluation of tumor subtypes, tumor margins, and treatment responses.14, 26
In the present study, FLSM was used for diagnosis and therapeutic monitoring of NMSC in reference to normal skin and correlation with routine histology. Several characteristic differences in the skin morphology, well known from histology, could be detected by FLSM and allowed the distinction of diagnostic features of AK and BCC. When evaluating AK, the most characteristic feature was the disruption of the stratum corneum and epidermal pleomorphism, which—in the absence of tumor nesting—was more indicative of AK. It should be mentioned, however, that an accurate assessment of the dermoepidermal junction may be difficult in the presence of significant hyperkeratosis. Interestingly, corresponding FLSM findings of AK were also present in areas where no pathologic findings were noted upon initial clinical examination, indicating the ability of FLSM to detect subclinical keratinocyte atypia. On the other hand, BCC showed comparable epidermal pleomorphism and the characteristic nesting of atypical basal cells was easily visualized using FLSM. BCC did not commonly present with hyperkeratosis, but demonstrated increased blood vessel tortuosity.
To allow the noninvasive monitoring of treatment response to Imiquimod, all corresponding FLSM features for superficial BCC were evaluated at baseline and variable time points during and after therapy. Our preliminary observations showed an initial increase of inflammatory cells relating to local immune response, and these findings subsided with cessation of therapy. The vascular dilatation or increased vascularization of treatment sites was observed after a course of of topical treatment and persisted at the followup, relating to the induction of neovascularization following Imiquimod therapy as described in previous reports.23, 24, 27
Previous studies have used RCM in the monitoring of therapeutic response,24, 25, 26 however, no studies using FLSM have been published to date. Although RCM allows the distinction of AK and BCC based entirely on endogenous contrast and tissue reflectivity,15 FLSM allowed the visualization of structures through intra- and intercellular accumulation of exogenously applied fluorophores. Nuclear morphology is readily identified in both systems, yet when compared to RCM, where reflectance noise is a potential confounding factor in the evaluation, FLSM allows the visualization of cellular and nuclear dimensions based on fluorescein distribution over time.13, 14 Cell-to-cell demarcation in normal skin is clearly distinguishable using FLSM due to the accumulation of fluorescein in the intercellular compartment. In contrast, barrier dysfunction of diseased skin may be associated with increased permeability of cellular membranes such that cell-to-cell demarcations can be less defined. With FLSM timing of the evaluation remains an important aspect of the evaluation, because the distribution of fluorescein follows a characteristic dynamic in live skin. RCM, on the other hand, allows an evaluation without exogenous contrast agents, thus underlining the noninvasive nature of this technology.15 The reproducibility of features appears comparable in both systems, with a systematic evaluation yet to be conducted. Overall, the ability of FLSM to image skin noninvasively and in real time without any discomfort makes it a promising tool in clinical dermatology.
Although there are many benefits of FLSM, there are also some limitations. The current technology is complex and expensive and thus not widely available to researchers and clinicians. Another potential limitation may be that imaging requires the application of exogenous fluorophores and the intraepidermal injection of fluorescein. With respect to image interpretation, the differences are often subtle and difficult to appreciate, such that a reliable interpretation is often based on correlation with normal skin or in reference with routine histology.
FLSM offers numerous potential applications for the advancement of medical research and clinical care by evaluating the status of normal skin, dynamic skin processes such as secretory and eccrine functions, and/or the management of skin disease. Our findings indicate that FLSM may be a feasible tool to monitor therapeutic skin response, thereby evaluating the effectiveness of topical therapies for NMSC noninvasively and in vivo. FLSM may allow a systematic, noninvasive histomorphometric evaluation of AK and BCC, potentially aiding in the detection of subclinical AK and early therapeutic management. The findings presented herein are of observational and preliminary nature, and further studies are warranted to corroborate our findings.
The authors would like to acknowledge OptiScan Ltd., Australia for providing the Stratum microscope for the study and for their help with the preparation of this manuscript. We would like to thank the Foundation “Skin Physiology” of the Donor Association for German Science and Humanities for their financial support.